{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999530215004933,
  "eval_steps": 500,
  "global_step": 15964,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 6.263799934230101e-05,
      "grad_norm": 3.858246462582563,
      "learning_rate": 2.0876826722338203e-08,
      "loss": 0.6556,
      "step": 1
    },
    {
      "epoch": 0.00012527599868460203,
      "grad_norm": 2.8119121815342027,
      "learning_rate": 4.1753653444676405e-08,
      "loss": 0.6526,
      "step": 2
    },
    {
      "epoch": 0.000187913998026903,
      "grad_norm": 4.37314039414459,
      "learning_rate": 6.263048016701463e-08,
      "loss": 0.6576,
      "step": 3
    },
    {
      "epoch": 0.00025055199736920405,
      "grad_norm": 3.3691798271712194,
      "learning_rate": 8.350730688935281e-08,
      "loss": 0.6632,
      "step": 4
    },
    {
      "epoch": 0.000313189996711505,
      "grad_norm": 3.8850758063243602,
      "learning_rate": 1.0438413361169103e-07,
      "loss": 0.6251,
      "step": 5
    },
    {
      "epoch": 0.000375827996053806,
      "grad_norm": 4.282646414197021,
      "learning_rate": 1.2526096033402926e-07,
      "loss": 0.6821,
      "step": 6
    },
    {
      "epoch": 0.00043846599539610704,
      "grad_norm": 3.255403937244552,
      "learning_rate": 1.4613778705636743e-07,
      "loss": 0.7077,
      "step": 7
    },
    {
      "epoch": 0.0005011039947384081,
      "grad_norm": 3.804064909558229,
      "learning_rate": 1.6701461377870562e-07,
      "loss": 0.6412,
      "step": 8
    },
    {
      "epoch": 0.0005637419940807091,
      "grad_norm": 4.310035823534526,
      "learning_rate": 1.8789144050104384e-07,
      "loss": 0.6659,
      "step": 9
    },
    {
      "epoch": 0.00062637999342301,
      "grad_norm": 3.8627229422269194,
      "learning_rate": 2.0876826722338207e-07,
      "loss": 0.6959,
      "step": 10
    },
    {
      "epoch": 0.000689017992765311,
      "grad_norm": 3.2331488181219994,
      "learning_rate": 2.2964509394572026e-07,
      "loss": 0.6538,
      "step": 11
    },
    {
      "epoch": 0.000751655992107612,
      "grad_norm": 3.348925964280946,
      "learning_rate": 2.505219206680585e-07,
      "loss": 0.6417,
      "step": 12
    },
    {
      "epoch": 0.0008142939914499131,
      "grad_norm": 4.1015297664563795,
      "learning_rate": 2.713987473903967e-07,
      "loss": 0.6999,
      "step": 13
    },
    {
      "epoch": 0.0008769319907922141,
      "grad_norm": 3.8277600827741574,
      "learning_rate": 2.9227557411273485e-07,
      "loss": 0.7035,
      "step": 14
    },
    {
      "epoch": 0.0009395699901345151,
      "grad_norm": 3.491662087714972,
      "learning_rate": 3.131524008350731e-07,
      "loss": 0.6543,
      "step": 15
    },
    {
      "epoch": 0.0010022079894768162,
      "grad_norm": 3.0109277959531355,
      "learning_rate": 3.3402922755741124e-07,
      "loss": 0.5837,
      "step": 16
    },
    {
      "epoch": 0.0010648459888191171,
      "grad_norm": 3.2620424982792846,
      "learning_rate": 3.549060542797495e-07,
      "loss": 0.6755,
      "step": 17
    },
    {
      "epoch": 0.0011274839881614182,
      "grad_norm": 3.1272000178049226,
      "learning_rate": 3.757828810020877e-07,
      "loss": 0.5891,
      "step": 18
    },
    {
      "epoch": 0.0011901219875037191,
      "grad_norm": 3.1006815712132973,
      "learning_rate": 3.9665970772442594e-07,
      "loss": 0.6114,
      "step": 19
    },
    {
      "epoch": 0.00125275998684602,
      "grad_norm": 2.9217370821323105,
      "learning_rate": 4.1753653444676413e-07,
      "loss": 0.6507,
      "step": 20
    },
    {
      "epoch": 0.0013153979861883212,
      "grad_norm": 0.7513542919000871,
      "learning_rate": 4.384133611691024e-07,
      "loss": 0.4267,
      "step": 21
    },
    {
      "epoch": 0.001378035985530622,
      "grad_norm": 2.2397548636664775,
      "learning_rate": 4.592901878914405e-07,
      "loss": 0.574,
      "step": 22
    },
    {
      "epoch": 0.0014406739848729232,
      "grad_norm": 2.8393767549719575,
      "learning_rate": 4.801670146137788e-07,
      "loss": 0.5616,
      "step": 23
    },
    {
      "epoch": 0.001503311984215224,
      "grad_norm": 2.61776792866924,
      "learning_rate": 5.01043841336117e-07,
      "loss": 0.5732,
      "step": 24
    },
    {
      "epoch": 0.0015659499835575252,
      "grad_norm": 2.836607102546222,
      "learning_rate": 5.219206680584552e-07,
      "loss": 0.6257,
      "step": 25
    },
    {
      "epoch": 0.0016285879828998261,
      "grad_norm": 2.5856733948576514,
      "learning_rate": 5.427974947807934e-07,
      "loss": 0.5937,
      "step": 26
    },
    {
      "epoch": 0.0016912259822421272,
      "grad_norm": 1.9897448861760498,
      "learning_rate": 5.636743215031316e-07,
      "loss": 0.5951,
      "step": 27
    },
    {
      "epoch": 0.0017538639815844281,
      "grad_norm": 2.592384218210653,
      "learning_rate": 5.845511482254697e-07,
      "loss": 0.5927,
      "step": 28
    },
    {
      "epoch": 0.0018165019809267293,
      "grad_norm": 2.346491595611199,
      "learning_rate": 6.05427974947808e-07,
      "loss": 0.599,
      "step": 29
    },
    {
      "epoch": 0.0018791399802690302,
      "grad_norm": 1.9743054351340714,
      "learning_rate": 6.263048016701462e-07,
      "loss": 0.6316,
      "step": 30
    },
    {
      "epoch": 0.0019417779796113313,
      "grad_norm": 1.7460246039341718,
      "learning_rate": 6.471816283924843e-07,
      "loss": 0.5596,
      "step": 31
    },
    {
      "epoch": 0.0020044159789536324,
      "grad_norm": 1.8949712382002544,
      "learning_rate": 6.680584551148225e-07,
      "loss": 0.5927,
      "step": 32
    },
    {
      "epoch": 0.002067053978295933,
      "grad_norm": 1.70591562542313,
      "learning_rate": 6.889352818371608e-07,
      "loss": 0.5877,
      "step": 33
    },
    {
      "epoch": 0.0021296919776382342,
      "grad_norm": 1.5314484579364067,
      "learning_rate": 7.09812108559499e-07,
      "loss": 0.6348,
      "step": 34
    },
    {
      "epoch": 0.0021923299769805353,
      "grad_norm": 1.5378350081382752,
      "learning_rate": 7.306889352818372e-07,
      "loss": 0.5205,
      "step": 35
    },
    {
      "epoch": 0.0022549679763228365,
      "grad_norm": 1.6154955839392315,
      "learning_rate": 7.515657620041754e-07,
      "loss": 0.5898,
      "step": 36
    },
    {
      "epoch": 0.002317605975665137,
      "grad_norm": 1.7852969381289066,
      "learning_rate": 7.724425887265135e-07,
      "loss": 0.6176,
      "step": 37
    },
    {
      "epoch": 0.0023802439750074383,
      "grad_norm": 1.4697421973459197,
      "learning_rate": 7.933194154488519e-07,
      "loss": 0.5358,
      "step": 38
    },
    {
      "epoch": 0.0024428819743497394,
      "grad_norm": 1.4585534357045287,
      "learning_rate": 8.1419624217119e-07,
      "loss": 0.6065,
      "step": 39
    },
    {
      "epoch": 0.00250551997369204,
      "grad_norm": 1.2129466315801778,
      "learning_rate": 8.350730688935283e-07,
      "loss": 0.5301,
      "step": 40
    },
    {
      "epoch": 0.002568157973034341,
      "grad_norm": 1.3484335839163204,
      "learning_rate": 8.559498956158664e-07,
      "loss": 0.5794,
      "step": 41
    },
    {
      "epoch": 0.0026307959723766423,
      "grad_norm": 1.2904046929597606,
      "learning_rate": 8.768267223382048e-07,
      "loss": 0.5232,
      "step": 42
    },
    {
      "epoch": 0.0026934339717189434,
      "grad_norm": 1.424258702134985,
      "learning_rate": 8.977035490605429e-07,
      "loss": 0.5588,
      "step": 43
    },
    {
      "epoch": 0.002756071971061244,
      "grad_norm": 1.2645675381197703,
      "learning_rate": 9.18580375782881e-07,
      "loss": 0.4987,
      "step": 44
    },
    {
      "epoch": 0.0028187099704035453,
      "grad_norm": 1.420063779488067,
      "learning_rate": 9.394572025052193e-07,
      "loss": 0.5568,
      "step": 45
    },
    {
      "epoch": 0.0028813479697458464,
      "grad_norm": 1.4144535548841484,
      "learning_rate": 9.603340292275575e-07,
      "loss": 0.609,
      "step": 46
    },
    {
      "epoch": 0.0029439859690881475,
      "grad_norm": 1.303609161047398,
      "learning_rate": 9.812108559498957e-07,
      "loss": 0.5267,
      "step": 47
    },
    {
      "epoch": 0.003006623968430448,
      "grad_norm": 1.33649767185354,
      "learning_rate": 1.002087682672234e-06,
      "loss": 0.5555,
      "step": 48
    },
    {
      "epoch": 0.0030692619677727493,
      "grad_norm": 0.9550151992620151,
      "learning_rate": 1.0229645093945722e-06,
      "loss": 0.457,
      "step": 49
    },
    {
      "epoch": 0.0031318999671150504,
      "grad_norm": 1.2737717691864179,
      "learning_rate": 1.0438413361169103e-06,
      "loss": 0.5109,
      "step": 50
    },
    {
      "epoch": 0.0031945379664573515,
      "grad_norm": 1.0950004816688281,
      "learning_rate": 1.0647181628392485e-06,
      "loss": 0.52,
      "step": 51
    },
    {
      "epoch": 0.0032571759657996522,
      "grad_norm": 1.343756537925798,
      "learning_rate": 1.0855949895615868e-06,
      "loss": 0.5379,
      "step": 52
    },
    {
      "epoch": 0.0033198139651419534,
      "grad_norm": 1.2812735912510242,
      "learning_rate": 1.106471816283925e-06,
      "loss": 0.606,
      "step": 53
    },
    {
      "epoch": 0.0033824519644842545,
      "grad_norm": 1.2786892396300105,
      "learning_rate": 1.1273486430062631e-06,
      "loss": 0.6238,
      "step": 54
    },
    {
      "epoch": 0.0034450899638265556,
      "grad_norm": 1.2363156753261304,
      "learning_rate": 1.1482254697286013e-06,
      "loss": 0.5229,
      "step": 55
    },
    {
      "epoch": 0.0035077279631688563,
      "grad_norm": 1.283827338191788,
      "learning_rate": 1.1691022964509394e-06,
      "loss": 0.5884,
      "step": 56
    },
    {
      "epoch": 0.0035703659625111574,
      "grad_norm": 1.2157241813865975,
      "learning_rate": 1.1899791231732778e-06,
      "loss": 0.568,
      "step": 57
    },
    {
      "epoch": 0.0036330039618534585,
      "grad_norm": 1.0672728629706603,
      "learning_rate": 1.210855949895616e-06,
      "loss": 0.5057,
      "step": 58
    },
    {
      "epoch": 0.003695641961195759,
      "grad_norm": 1.097644549879363,
      "learning_rate": 1.2317327766179543e-06,
      "loss": 0.5258,
      "step": 59
    },
    {
      "epoch": 0.0037582799605380603,
      "grad_norm": 1.1658496382624082,
      "learning_rate": 1.2526096033402924e-06,
      "loss": 0.4955,
      "step": 60
    },
    {
      "epoch": 0.0038209179598803615,
      "grad_norm": 0.6906370577819059,
      "learning_rate": 1.2734864300626308e-06,
      "loss": 0.4645,
      "step": 61
    },
    {
      "epoch": 0.0038835559592226626,
      "grad_norm": 1.149519635390912,
      "learning_rate": 1.2943632567849687e-06,
      "loss": 0.5185,
      "step": 62
    },
    {
      "epoch": 0.003946193958564963,
      "grad_norm": 1.09158686448251,
      "learning_rate": 1.315240083507307e-06,
      "loss": 0.5245,
      "step": 63
    },
    {
      "epoch": 0.004008831957907265,
      "grad_norm": 1.2265208264731924,
      "learning_rate": 1.336116910229645e-06,
      "loss": 0.5647,
      "step": 64
    },
    {
      "epoch": 0.0040714699572495655,
      "grad_norm": 1.1873778162492044,
      "learning_rate": 1.3569937369519833e-06,
      "loss": 0.5121,
      "step": 65
    },
    {
      "epoch": 0.004134107956591866,
      "grad_norm": 1.0719873274622074,
      "learning_rate": 1.3778705636743217e-06,
      "loss": 0.5123,
      "step": 66
    },
    {
      "epoch": 0.004196745955934168,
      "grad_norm": 1.0698863903712088,
      "learning_rate": 1.3987473903966598e-06,
      "loss": 0.5288,
      "step": 67
    },
    {
      "epoch": 0.0042593839552764684,
      "grad_norm": 1.139092848507996,
      "learning_rate": 1.419624217118998e-06,
      "loss": 0.5314,
      "step": 68
    },
    {
      "epoch": 0.004322021954618769,
      "grad_norm": 1.1485930477303963,
      "learning_rate": 1.4405010438413361e-06,
      "loss": 0.4948,
      "step": 69
    },
    {
      "epoch": 0.004384659953961071,
      "grad_norm": 1.1071073623227474,
      "learning_rate": 1.4613778705636745e-06,
      "loss": 0.5292,
      "step": 70
    },
    {
      "epoch": 0.004447297953303371,
      "grad_norm": 1.1328930083981115,
      "learning_rate": 1.4822546972860128e-06,
      "loss": 0.5836,
      "step": 71
    },
    {
      "epoch": 0.004509935952645673,
      "grad_norm": 0.643986939147151,
      "learning_rate": 1.5031315240083507e-06,
      "loss": 0.4517,
      "step": 72
    },
    {
      "epoch": 0.004572573951987974,
      "grad_norm": 1.313623730850765,
      "learning_rate": 1.524008350730689e-06,
      "loss": 0.578,
      "step": 73
    },
    {
      "epoch": 0.004635211951330274,
      "grad_norm": 1.088069049930434,
      "learning_rate": 1.544885177453027e-06,
      "loss": 0.4735,
      "step": 74
    },
    {
      "epoch": 0.004697849950672576,
      "grad_norm": 1.0588882839597165,
      "learning_rate": 1.5657620041753654e-06,
      "loss": 0.5297,
      "step": 75
    },
    {
      "epoch": 0.0047604879500148765,
      "grad_norm": 1.190794928783865,
      "learning_rate": 1.5866388308977037e-06,
      "loss": 0.5103,
      "step": 76
    },
    {
      "epoch": 0.004823125949357177,
      "grad_norm": 1.189382189840441,
      "learning_rate": 1.6075156576200419e-06,
      "loss": 0.5664,
      "step": 77
    },
    {
      "epoch": 0.004885763948699479,
      "grad_norm": 1.093919638250216,
      "learning_rate": 1.62839248434238e-06,
      "loss": 0.5365,
      "step": 78
    },
    {
      "epoch": 0.0049484019480417795,
      "grad_norm": 1.0546717208154948,
      "learning_rate": 1.6492693110647182e-06,
      "loss": 0.513,
      "step": 79
    },
    {
      "epoch": 0.00501103994738408,
      "grad_norm": 1.0111072255357463,
      "learning_rate": 1.6701461377870565e-06,
      "loss": 0.4488,
      "step": 80
    },
    {
      "epoch": 0.005073677946726382,
      "grad_norm": 1.1199797827857312,
      "learning_rate": 1.6910229645093949e-06,
      "loss": 0.5234,
      "step": 81
    },
    {
      "epoch": 0.005136315946068682,
      "grad_norm": 1.0238803544330657,
      "learning_rate": 1.7118997912317328e-06,
      "loss": 0.5295,
      "step": 82
    },
    {
      "epoch": 0.005198953945410984,
      "grad_norm": 1.140501133074839,
      "learning_rate": 1.7327766179540712e-06,
      "loss": 0.513,
      "step": 83
    },
    {
      "epoch": 0.005261591944753285,
      "grad_norm": 1.116573825073439,
      "learning_rate": 1.7536534446764095e-06,
      "loss": 0.4995,
      "step": 84
    },
    {
      "epoch": 0.005324229944095585,
      "grad_norm": 0.9404777653188781,
      "learning_rate": 1.7745302713987475e-06,
      "loss": 0.5193,
      "step": 85
    },
    {
      "epoch": 0.005386867943437887,
      "grad_norm": 1.1415096511283775,
      "learning_rate": 1.7954070981210858e-06,
      "loss": 0.564,
      "step": 86
    },
    {
      "epoch": 0.005449505942780188,
      "grad_norm": 1.0667730622010472,
      "learning_rate": 1.816283924843424e-06,
      "loss": 0.507,
      "step": 87
    },
    {
      "epoch": 0.005512143942122488,
      "grad_norm": 0.9856125126706705,
      "learning_rate": 1.837160751565762e-06,
      "loss": 0.5057,
      "step": 88
    },
    {
      "epoch": 0.00557478194146479,
      "grad_norm": 1.1089144753383688,
      "learning_rate": 1.8580375782881005e-06,
      "loss": 0.4994,
      "step": 89
    },
    {
      "epoch": 0.0056374199408070905,
      "grad_norm": 0.7913516821595887,
      "learning_rate": 1.8789144050104386e-06,
      "loss": 0.4285,
      "step": 90
    },
    {
      "epoch": 0.005700057940149392,
      "grad_norm": 0.9949016658406422,
      "learning_rate": 1.899791231732777e-06,
      "loss": 0.4574,
      "step": 91
    },
    {
      "epoch": 0.005762695939491693,
      "grad_norm": 1.0172594220652533,
      "learning_rate": 1.920668058455115e-06,
      "loss": 0.4798,
      "step": 92
    },
    {
      "epoch": 0.005825333938833993,
      "grad_norm": 1.0235201684444175,
      "learning_rate": 1.941544885177453e-06,
      "loss": 0.4932,
      "step": 93
    },
    {
      "epoch": 0.005887971938176295,
      "grad_norm": 1.1681954155230496,
      "learning_rate": 1.9624217118997914e-06,
      "loss": 0.4854,
      "step": 94
    },
    {
      "epoch": 0.005950609937518596,
      "grad_norm": 1.1234057763473662,
      "learning_rate": 1.9832985386221297e-06,
      "loss": 0.4924,
      "step": 95
    },
    {
      "epoch": 0.006013247936860896,
      "grad_norm": 1.0170230053571934,
      "learning_rate": 2.004175365344468e-06,
      "loss": 0.496,
      "step": 96
    },
    {
      "epoch": 0.006075885936203198,
      "grad_norm": 1.1423557307019825,
      "learning_rate": 2.025052192066806e-06,
      "loss": 0.488,
      "step": 97
    },
    {
      "epoch": 0.006138523935545499,
      "grad_norm": 1.045937875225304,
      "learning_rate": 2.0459290187891444e-06,
      "loss": 0.549,
      "step": 98
    },
    {
      "epoch": 0.006201161934887799,
      "grad_norm": 1.0921519402561854,
      "learning_rate": 2.0668058455114827e-06,
      "loss": 0.5141,
      "step": 99
    },
    {
      "epoch": 0.006263799934230101,
      "grad_norm": 1.1121118003367094,
      "learning_rate": 2.0876826722338207e-06,
      "loss": 0.5147,
      "step": 100
    },
    {
      "epoch": 0.0063264379335724015,
      "grad_norm": 1.076945062360127,
      "learning_rate": 2.108559498956159e-06,
      "loss": 0.5592,
      "step": 101
    },
    {
      "epoch": 0.006389075932914703,
      "grad_norm": 0.9224163935797927,
      "learning_rate": 2.129436325678497e-06,
      "loss": 0.4746,
      "step": 102
    },
    {
      "epoch": 0.006451713932257004,
      "grad_norm": 1.093744856443408,
      "learning_rate": 2.1503131524008353e-06,
      "loss": 0.5046,
      "step": 103
    },
    {
      "epoch": 0.0065143519315993045,
      "grad_norm": 1.0544106225205734,
      "learning_rate": 2.1711899791231737e-06,
      "loss": 0.4844,
      "step": 104
    },
    {
      "epoch": 0.006576989930941606,
      "grad_norm": 1.001594521955387,
      "learning_rate": 2.1920668058455116e-06,
      "loss": 0.4719,
      "step": 105
    },
    {
      "epoch": 0.006639627930283907,
      "grad_norm": 1.0678206659208174,
      "learning_rate": 2.21294363256785e-06,
      "loss": 0.5393,
      "step": 106
    },
    {
      "epoch": 0.006702265929626207,
      "grad_norm": 1.086739825505249,
      "learning_rate": 2.233820459290188e-06,
      "loss": 0.4939,
      "step": 107
    },
    {
      "epoch": 0.006764903928968509,
      "grad_norm": 1.0078869385132547,
      "learning_rate": 2.2546972860125262e-06,
      "loss": 0.4751,
      "step": 108
    },
    {
      "epoch": 0.00682754192831081,
      "grad_norm": 0.7356807226316928,
      "learning_rate": 2.2755741127348646e-06,
      "loss": 0.4496,
      "step": 109
    },
    {
      "epoch": 0.006890179927653111,
      "grad_norm": 1.050764047209418,
      "learning_rate": 2.2964509394572025e-06,
      "loss": 0.4603,
      "step": 110
    },
    {
      "epoch": 0.006952817926995412,
      "grad_norm": 1.0394967666623012,
      "learning_rate": 2.317327766179541e-06,
      "loss": 0.5315,
      "step": 111
    },
    {
      "epoch": 0.007015455926337713,
      "grad_norm": 1.0602497947938774,
      "learning_rate": 2.338204592901879e-06,
      "loss": 0.4901,
      "step": 112
    },
    {
      "epoch": 0.007078093925680014,
      "grad_norm": 1.0257070634927614,
      "learning_rate": 2.359081419624217e-06,
      "loss": 0.4984,
      "step": 113
    },
    {
      "epoch": 0.007140731925022315,
      "grad_norm": 1.0760008010073892,
      "learning_rate": 2.3799582463465555e-06,
      "loss": 0.5352,
      "step": 114
    },
    {
      "epoch": 0.0072033699243646155,
      "grad_norm": 0.9764566829688709,
      "learning_rate": 2.400835073068894e-06,
      "loss": 0.466,
      "step": 115
    },
    {
      "epoch": 0.007266007923706917,
      "grad_norm": 1.0211556850587806,
      "learning_rate": 2.421711899791232e-06,
      "loss": 0.4691,
      "step": 116
    },
    {
      "epoch": 0.007328645923049218,
      "grad_norm": 0.962038893509024,
      "learning_rate": 2.44258872651357e-06,
      "loss": 0.4704,
      "step": 117
    },
    {
      "epoch": 0.007391283922391518,
      "grad_norm": 1.0246135137326515,
      "learning_rate": 2.4634655532359085e-06,
      "loss": 0.5183,
      "step": 118
    },
    {
      "epoch": 0.00745392192173382,
      "grad_norm": 0.9762498734515195,
      "learning_rate": 2.484342379958247e-06,
      "loss": 0.4542,
      "step": 119
    },
    {
      "epoch": 0.007516559921076121,
      "grad_norm": 0.9531644787025032,
      "learning_rate": 2.505219206680585e-06,
      "loss": 0.4976,
      "step": 120
    },
    {
      "epoch": 0.007579197920418422,
      "grad_norm": 1.0088685912113438,
      "learning_rate": 2.5260960334029227e-06,
      "loss": 0.4795,
      "step": 121
    },
    {
      "epoch": 0.007641835919760723,
      "grad_norm": 1.0407768835043478,
      "learning_rate": 2.5469728601252615e-06,
      "loss": 0.4531,
      "step": 122
    },
    {
      "epoch": 0.007704473919103024,
      "grad_norm": 1.0531424574218735,
      "learning_rate": 2.5678496868475994e-06,
      "loss": 0.5124,
      "step": 123
    },
    {
      "epoch": 0.007767111918445325,
      "grad_norm": 1.0592814918464573,
      "learning_rate": 2.5887265135699374e-06,
      "loss": 0.4739,
      "step": 124
    },
    {
      "epoch": 0.007829749917787626,
      "grad_norm": 0.7704512808980731,
      "learning_rate": 2.609603340292276e-06,
      "loss": 0.4555,
      "step": 125
    },
    {
      "epoch": 0.007892387917129927,
      "grad_norm": 0.9619284652004104,
      "learning_rate": 2.630480167014614e-06,
      "loss": 0.4299,
      "step": 126
    },
    {
      "epoch": 0.007955025916472227,
      "grad_norm": 1.0598976965787394,
      "learning_rate": 2.651356993736952e-06,
      "loss": 0.4904,
      "step": 127
    },
    {
      "epoch": 0.00801766391581453,
      "grad_norm": 1.0881172096160947,
      "learning_rate": 2.67223382045929e-06,
      "loss": 0.468,
      "step": 128
    },
    {
      "epoch": 0.00808030191515683,
      "grad_norm": 1.098405838065741,
      "learning_rate": 2.6931106471816287e-06,
      "loss": 0.5234,
      "step": 129
    },
    {
      "epoch": 0.008142939914499131,
      "grad_norm": 0.9619111243894403,
      "learning_rate": 2.7139874739039666e-06,
      "loss": 0.4677,
      "step": 130
    },
    {
      "epoch": 0.008205577913841432,
      "grad_norm": 1.1488886855077485,
      "learning_rate": 2.734864300626305e-06,
      "loss": 0.4932,
      "step": 131
    },
    {
      "epoch": 0.008268215913183732,
      "grad_norm": 1.0453855110191648,
      "learning_rate": 2.7557411273486434e-06,
      "loss": 0.465,
      "step": 132
    },
    {
      "epoch": 0.008330853912526033,
      "grad_norm": 0.9985773191691992,
      "learning_rate": 2.7766179540709813e-06,
      "loss": 0.4622,
      "step": 133
    },
    {
      "epoch": 0.008393491911868336,
      "grad_norm": 1.1367074352126274,
      "learning_rate": 2.7974947807933196e-06,
      "loss": 0.4806,
      "step": 134
    },
    {
      "epoch": 0.008456129911210636,
      "grad_norm": 1.0587405507050538,
      "learning_rate": 2.818371607515658e-06,
      "loss": 0.5026,
      "step": 135
    },
    {
      "epoch": 0.008518767910552937,
      "grad_norm": 1.0763655377426442,
      "learning_rate": 2.839248434237996e-06,
      "loss": 0.5284,
      "step": 136
    },
    {
      "epoch": 0.008581405909895238,
      "grad_norm": 1.0218345924762122,
      "learning_rate": 2.8601252609603343e-06,
      "loss": 0.48,
      "step": 137
    },
    {
      "epoch": 0.008644043909237538,
      "grad_norm": 0.7448265515180501,
      "learning_rate": 2.8810020876826722e-06,
      "loss": 0.4456,
      "step": 138
    },
    {
      "epoch": 0.00870668190857984,
      "grad_norm": 0.9921108695414547,
      "learning_rate": 2.901878914405011e-06,
      "loss": 0.4483,
      "step": 139
    },
    {
      "epoch": 0.008769319907922141,
      "grad_norm": 1.0620431780420618,
      "learning_rate": 2.922755741127349e-06,
      "loss": 0.5347,
      "step": 140
    },
    {
      "epoch": 0.008831957907264442,
      "grad_norm": 1.0126834686491157,
      "learning_rate": 2.943632567849687e-06,
      "loss": 0.4786,
      "step": 141
    },
    {
      "epoch": 0.008894595906606743,
      "grad_norm": 1.038905561380666,
      "learning_rate": 2.9645093945720256e-06,
      "loss": 0.4727,
      "step": 142
    },
    {
      "epoch": 0.008957233905949043,
      "grad_norm": 0.7620131311160847,
      "learning_rate": 2.9853862212943636e-06,
      "loss": 0.4033,
      "step": 143
    },
    {
      "epoch": 0.009019871905291346,
      "grad_norm": 1.0652460741176468,
      "learning_rate": 3.0062630480167015e-06,
      "loss": 0.4968,
      "step": 144
    },
    {
      "epoch": 0.009082509904633647,
      "grad_norm": 1.0328029376865766,
      "learning_rate": 3.0271398747390403e-06,
      "loss": 0.4824,
      "step": 145
    },
    {
      "epoch": 0.009145147903975947,
      "grad_norm": 0.9919628673284245,
      "learning_rate": 3.048016701461378e-06,
      "loss": 0.4649,
      "step": 146
    },
    {
      "epoch": 0.009207785903318248,
      "grad_norm": 0.9737211345001473,
      "learning_rate": 3.068893528183716e-06,
      "loss": 0.4715,
      "step": 147
    },
    {
      "epoch": 0.009270423902660549,
      "grad_norm": 1.076543346506063,
      "learning_rate": 3.089770354906054e-06,
      "loss": 0.4938,
      "step": 148
    },
    {
      "epoch": 0.00933306190200285,
      "grad_norm": 1.0915990957003543,
      "learning_rate": 3.110647181628393e-06,
      "loss": 0.4543,
      "step": 149
    },
    {
      "epoch": 0.009395699901345152,
      "grad_norm": 1.0927755550760756,
      "learning_rate": 3.1315240083507308e-06,
      "loss": 0.4948,
      "step": 150
    },
    {
      "epoch": 0.009458337900687452,
      "grad_norm": 1.0079496628204152,
      "learning_rate": 3.152400835073069e-06,
      "loss": 0.4684,
      "step": 151
    },
    {
      "epoch": 0.009520975900029753,
      "grad_norm": 1.0592697805564688,
      "learning_rate": 3.1732776617954075e-06,
      "loss": 0.4772,
      "step": 152
    },
    {
      "epoch": 0.009583613899372054,
      "grad_norm": 1.0720240701225303,
      "learning_rate": 3.1941544885177454e-06,
      "loss": 0.5035,
      "step": 153
    },
    {
      "epoch": 0.009646251898714354,
      "grad_norm": 1.0294206211900345,
      "learning_rate": 3.2150313152400838e-06,
      "loss": 0.4689,
      "step": 154
    },
    {
      "epoch": 0.009708889898056657,
      "grad_norm": 1.1395191077812998,
      "learning_rate": 3.235908141962422e-06,
      "loss": 0.515,
      "step": 155
    },
    {
      "epoch": 0.009771527897398958,
      "grad_norm": 1.023581244241528,
      "learning_rate": 3.25678496868476e-06,
      "loss": 0.4957,
      "step": 156
    },
    {
      "epoch": 0.009834165896741258,
      "grad_norm": 1.1054189648835393,
      "learning_rate": 3.2776617954070984e-06,
      "loss": 0.4947,
      "step": 157
    },
    {
      "epoch": 0.009896803896083559,
      "grad_norm": 1.2034941276502231,
      "learning_rate": 3.2985386221294363e-06,
      "loss": 0.4652,
      "step": 158
    },
    {
      "epoch": 0.00995944189542586,
      "grad_norm": 1.033781018890459,
      "learning_rate": 3.3194154488517747e-06,
      "loss": 0.4807,
      "step": 159
    },
    {
      "epoch": 0.01002207989476816,
      "grad_norm": 1.0468799975085483,
      "learning_rate": 3.340292275574113e-06,
      "loss": 0.4619,
      "step": 160
    },
    {
      "epoch": 0.010084717894110463,
      "grad_norm": 1.1275812080268477,
      "learning_rate": 3.361169102296451e-06,
      "loss": 0.5194,
      "step": 161
    },
    {
      "epoch": 0.010147355893452763,
      "grad_norm": 1.0224551987015758,
      "learning_rate": 3.3820459290187898e-06,
      "loss": 0.5319,
      "step": 162
    },
    {
      "epoch": 0.010209993892795064,
      "grad_norm": 1.028785125398643,
      "learning_rate": 3.4029227557411277e-06,
      "loss": 0.466,
      "step": 163
    },
    {
      "epoch": 0.010272631892137365,
      "grad_norm": 1.0549527110026886,
      "learning_rate": 3.4237995824634656e-06,
      "loss": 0.4706,
      "step": 164
    },
    {
      "epoch": 0.010335269891479665,
      "grad_norm": 1.0925069832539478,
      "learning_rate": 3.4446764091858044e-06,
      "loss": 0.4652,
      "step": 165
    },
    {
      "epoch": 0.010397907890821968,
      "grad_norm": 1.0217216587012772,
      "learning_rate": 3.4655532359081423e-06,
      "loss": 0.4979,
      "step": 166
    },
    {
      "epoch": 0.010460545890164269,
      "grad_norm": 1.0633692728915884,
      "learning_rate": 3.4864300626304803e-06,
      "loss": 0.4994,
      "step": 167
    },
    {
      "epoch": 0.01052318388950657,
      "grad_norm": 0.9799541517896768,
      "learning_rate": 3.507306889352819e-06,
      "loss": 0.4484,
      "step": 168
    },
    {
      "epoch": 0.01058582188884887,
      "grad_norm": 1.0241758059586266,
      "learning_rate": 3.528183716075157e-06,
      "loss": 0.53,
      "step": 169
    },
    {
      "epoch": 0.01064845988819117,
      "grad_norm": 1.0730025392115772,
      "learning_rate": 3.549060542797495e-06,
      "loss": 0.4777,
      "step": 170
    },
    {
      "epoch": 0.010711097887533471,
      "grad_norm": 1.0851197827633212,
      "learning_rate": 3.569937369519833e-06,
      "loss": 0.4798,
      "step": 171
    },
    {
      "epoch": 0.010773735886875774,
      "grad_norm": 0.9858276946348559,
      "learning_rate": 3.5908141962421716e-06,
      "loss": 0.4618,
      "step": 172
    },
    {
      "epoch": 0.010836373886218074,
      "grad_norm": 0.923299798253563,
      "learning_rate": 3.6116910229645096e-06,
      "loss": 0.4611,
      "step": 173
    },
    {
      "epoch": 0.010899011885560375,
      "grad_norm": 1.034134739823511,
      "learning_rate": 3.632567849686848e-06,
      "loss": 0.5058,
      "step": 174
    },
    {
      "epoch": 0.010961649884902676,
      "grad_norm": 0.7398001387044641,
      "learning_rate": 3.6534446764091863e-06,
      "loss": 0.4597,
      "step": 175
    },
    {
      "epoch": 0.011024287884244977,
      "grad_norm": 0.9368124731619292,
      "learning_rate": 3.674321503131524e-06,
      "loss": 0.495,
      "step": 176
    },
    {
      "epoch": 0.011086925883587279,
      "grad_norm": 1.0924136049020585,
      "learning_rate": 3.6951983298538625e-06,
      "loss": 0.4231,
      "step": 177
    },
    {
      "epoch": 0.01114956388292958,
      "grad_norm": 1.0969746891242427,
      "learning_rate": 3.716075156576201e-06,
      "loss": 0.5264,
      "step": 178
    },
    {
      "epoch": 0.01121220188227188,
      "grad_norm": 0.980930696280059,
      "learning_rate": 3.736951983298539e-06,
      "loss": 0.4286,
      "step": 179
    },
    {
      "epoch": 0.011274839881614181,
      "grad_norm": 1.0404716326236667,
      "learning_rate": 3.757828810020877e-06,
      "loss": 0.4803,
      "step": 180
    },
    {
      "epoch": 0.011337477880956482,
      "grad_norm": 1.050052718634254,
      "learning_rate": 3.778705636743215e-06,
      "loss": 0.4789,
      "step": 181
    },
    {
      "epoch": 0.011400115880298784,
      "grad_norm": 1.0947033180482668,
      "learning_rate": 3.799582463465554e-06,
      "loss": 0.4995,
      "step": 182
    },
    {
      "epoch": 0.011462753879641085,
      "grad_norm": 1.0559025368136914,
      "learning_rate": 3.820459290187892e-06,
      "loss": 0.4069,
      "step": 183
    },
    {
      "epoch": 0.011525391878983385,
      "grad_norm": 1.0043055688658917,
      "learning_rate": 3.84133611691023e-06,
      "loss": 0.4521,
      "step": 184
    },
    {
      "epoch": 0.011588029878325686,
      "grad_norm": 1.1338024311268484,
      "learning_rate": 3.8622129436325685e-06,
      "loss": 0.4891,
      "step": 185
    },
    {
      "epoch": 0.011650667877667987,
      "grad_norm": 0.9601268987873947,
      "learning_rate": 3.883089770354906e-06,
      "loss": 0.4802,
      "step": 186
    },
    {
      "epoch": 0.011713305877010288,
      "grad_norm": 0.9529681904012965,
      "learning_rate": 3.903966597077244e-06,
      "loss": 0.4939,
      "step": 187
    },
    {
      "epoch": 0.01177594387635259,
      "grad_norm": 0.9866849998776137,
      "learning_rate": 3.924843423799583e-06,
      "loss": 0.4364,
      "step": 188
    },
    {
      "epoch": 0.01183858187569489,
      "grad_norm": 1.031337705595775,
      "learning_rate": 3.945720250521921e-06,
      "loss": 0.5092,
      "step": 189
    },
    {
      "epoch": 0.011901219875037191,
      "grad_norm": 0.8743091787107913,
      "learning_rate": 3.9665970772442595e-06,
      "loss": 0.4575,
      "step": 190
    },
    {
      "epoch": 0.011963857874379492,
      "grad_norm": 0.975697332586742,
      "learning_rate": 3.987473903966597e-06,
      "loss": 0.4648,
      "step": 191
    },
    {
      "epoch": 0.012026495873721793,
      "grad_norm": 1.126906857623645,
      "learning_rate": 4.008350730688936e-06,
      "loss": 0.52,
      "step": 192
    },
    {
      "epoch": 0.012089133873064095,
      "grad_norm": 0.9989527488750611,
      "learning_rate": 4.029227557411274e-06,
      "loss": 0.4691,
      "step": 193
    },
    {
      "epoch": 0.012151771872406396,
      "grad_norm": 1.1050127624680133,
      "learning_rate": 4.050104384133612e-06,
      "loss": 0.4697,
      "step": 194
    },
    {
      "epoch": 0.012214409871748697,
      "grad_norm": 1.1104494143040955,
      "learning_rate": 4.07098121085595e-06,
      "loss": 0.4984,
      "step": 195
    },
    {
      "epoch": 0.012277047871090997,
      "grad_norm": 1.0041829747409963,
      "learning_rate": 4.091858037578289e-06,
      "loss": 0.4786,
      "step": 196
    },
    {
      "epoch": 0.012339685870433298,
      "grad_norm": 0.9592971161986891,
      "learning_rate": 4.112734864300626e-06,
      "loss": 0.4733,
      "step": 197
    },
    {
      "epoch": 0.012402323869775599,
      "grad_norm": 1.0826668523833276,
      "learning_rate": 4.1336116910229655e-06,
      "loss": 0.4441,
      "step": 198
    },
    {
      "epoch": 0.012464961869117901,
      "grad_norm": 0.9944836197780231,
      "learning_rate": 4.154488517745303e-06,
      "loss": 0.5008,
      "step": 199
    },
    {
      "epoch": 0.012527599868460202,
      "grad_norm": 1.0539514839279558,
      "learning_rate": 4.175365344467641e-06,
      "loss": 0.459,
      "step": 200
    },
    {
      "epoch": 0.012590237867802502,
      "grad_norm": 1.053781464605634,
      "learning_rate": 4.196242171189979e-06,
      "loss": 0.4788,
      "step": 201
    },
    {
      "epoch": 0.012652875867144803,
      "grad_norm": 1.0211614699426739,
      "learning_rate": 4.217118997912318e-06,
      "loss": 0.4503,
      "step": 202
    },
    {
      "epoch": 0.012715513866487104,
      "grad_norm": 1.0427517617292166,
      "learning_rate": 4.2379958246346555e-06,
      "loss": 0.4924,
      "step": 203
    },
    {
      "epoch": 0.012778151865829406,
      "grad_norm": 1.0502804896506799,
      "learning_rate": 4.258872651356994e-06,
      "loss": 0.4832,
      "step": 204
    },
    {
      "epoch": 0.012840789865171707,
      "grad_norm": 0.9587067322996741,
      "learning_rate": 4.279749478079332e-06,
      "loss": 0.4718,
      "step": 205
    },
    {
      "epoch": 0.012903427864514008,
      "grad_norm": 1.029138473291132,
      "learning_rate": 4.300626304801671e-06,
      "loss": 0.465,
      "step": 206
    },
    {
      "epoch": 0.012966065863856308,
      "grad_norm": 1.0081575809405636,
      "learning_rate": 4.321503131524009e-06,
      "loss": 0.4482,
      "step": 207
    },
    {
      "epoch": 0.013028703863198609,
      "grad_norm": 1.0829791811044427,
      "learning_rate": 4.342379958246347e-06,
      "loss": 0.4797,
      "step": 208
    },
    {
      "epoch": 0.01309134186254091,
      "grad_norm": 1.0996774869825248,
      "learning_rate": 4.363256784968685e-06,
      "loss": 0.4574,
      "step": 209
    },
    {
      "epoch": 0.013153979861883212,
      "grad_norm": 1.092451142863826,
      "learning_rate": 4.384133611691023e-06,
      "loss": 0.4613,
      "step": 210
    },
    {
      "epoch": 0.013216617861225513,
      "grad_norm": 1.0926739930862424,
      "learning_rate": 4.4050104384133615e-06,
      "loss": 0.5453,
      "step": 211
    },
    {
      "epoch": 0.013279255860567813,
      "grad_norm": 0.9401689358719709,
      "learning_rate": 4.4258872651357e-06,
      "loss": 0.4912,
      "step": 212
    },
    {
      "epoch": 0.013341893859910114,
      "grad_norm": 1.020876926317057,
      "learning_rate": 4.446764091858038e-06,
      "loss": 0.437,
      "step": 213
    },
    {
      "epoch": 0.013404531859252415,
      "grad_norm": 0.9842030498288622,
      "learning_rate": 4.467640918580376e-06,
      "loss": 0.4654,
      "step": 214
    },
    {
      "epoch": 0.013467169858594717,
      "grad_norm": 1.0174471998772452,
      "learning_rate": 4.488517745302715e-06,
      "loss": 0.4453,
      "step": 215
    },
    {
      "epoch": 0.013529807857937018,
      "grad_norm": 1.0487063050004148,
      "learning_rate": 4.5093945720250525e-06,
      "loss": 0.4665,
      "step": 216
    },
    {
      "epoch": 0.013592445857279319,
      "grad_norm": 1.1077749997986306,
      "learning_rate": 4.530271398747391e-06,
      "loss": 0.4611,
      "step": 217
    },
    {
      "epoch": 0.01365508385662162,
      "grad_norm": 0.9738701078308191,
      "learning_rate": 4.551148225469729e-06,
      "loss": 0.4538,
      "step": 218
    },
    {
      "epoch": 0.01371772185596392,
      "grad_norm": 1.028132684002268,
      "learning_rate": 4.5720250521920675e-06,
      "loss": 0.4791,
      "step": 219
    },
    {
      "epoch": 0.013780359855306222,
      "grad_norm": 1.0299871015594273,
      "learning_rate": 4.592901878914405e-06,
      "loss": 0.4507,
      "step": 220
    },
    {
      "epoch": 0.013842997854648523,
      "grad_norm": 0.8168442302528722,
      "learning_rate": 4.613778705636743e-06,
      "loss": 0.4681,
      "step": 221
    },
    {
      "epoch": 0.013905635853990824,
      "grad_norm": 1.078313143025839,
      "learning_rate": 4.634655532359082e-06,
      "loss": 0.4852,
      "step": 222
    },
    {
      "epoch": 0.013968273853333124,
      "grad_norm": 1.0961526819838456,
      "learning_rate": 4.65553235908142e-06,
      "loss": 0.4442,
      "step": 223
    },
    {
      "epoch": 0.014030911852675425,
      "grad_norm": 0.9563744096049571,
      "learning_rate": 4.676409185803758e-06,
      "loss": 0.459,
      "step": 224
    },
    {
      "epoch": 0.014093549852017726,
      "grad_norm": 1.092222764494014,
      "learning_rate": 4.697286012526097e-06,
      "loss": 0.4911,
      "step": 225
    },
    {
      "epoch": 0.014156187851360028,
      "grad_norm": 1.0095658512782304,
      "learning_rate": 4.718162839248434e-06,
      "loss": 0.4814,
      "step": 226
    },
    {
      "epoch": 0.014218825850702329,
      "grad_norm": 1.0450061681710097,
      "learning_rate": 4.739039665970773e-06,
      "loss": 0.4972,
      "step": 227
    },
    {
      "epoch": 0.01428146385004463,
      "grad_norm": 0.9572559475918908,
      "learning_rate": 4.759916492693111e-06,
      "loss": 0.4452,
      "step": 228
    },
    {
      "epoch": 0.01434410184938693,
      "grad_norm": 0.9827017531097282,
      "learning_rate": 4.780793319415449e-06,
      "loss": 0.4687,
      "step": 229
    },
    {
      "epoch": 0.014406739848729231,
      "grad_norm": 1.124894650077553,
      "learning_rate": 4.801670146137788e-06,
      "loss": 0.4823,
      "step": 230
    },
    {
      "epoch": 0.014469377848071533,
      "grad_norm": 0.9979297806384481,
      "learning_rate": 4.822546972860125e-06,
      "loss": 0.4256,
      "step": 231
    },
    {
      "epoch": 0.014532015847413834,
      "grad_norm": 0.9390284419190013,
      "learning_rate": 4.843423799582464e-06,
      "loss": 0.4633,
      "step": 232
    },
    {
      "epoch": 0.014594653846756135,
      "grad_norm": 1.0807868026451068,
      "learning_rate": 4.864300626304802e-06,
      "loss": 0.4666,
      "step": 233
    },
    {
      "epoch": 0.014657291846098435,
      "grad_norm": 1.0250484260298092,
      "learning_rate": 4.88517745302714e-06,
      "loss": 0.4637,
      "step": 234
    },
    {
      "epoch": 0.014719929845440736,
      "grad_norm": 1.0496425969341063,
      "learning_rate": 4.906054279749479e-06,
      "loss": 0.445,
      "step": 235
    },
    {
      "epoch": 0.014782567844783037,
      "grad_norm": 1.1014218572219718,
      "learning_rate": 4.926931106471817e-06,
      "loss": 0.4995,
      "step": 236
    },
    {
      "epoch": 0.01484520584412534,
      "grad_norm": 1.0436644342582813,
      "learning_rate": 4.9478079331941545e-06,
      "loss": 0.4479,
      "step": 237
    },
    {
      "epoch": 0.01490784384346764,
      "grad_norm": 1.025364509020331,
      "learning_rate": 4.968684759916494e-06,
      "loss": 0.4943,
      "step": 238
    },
    {
      "epoch": 0.01497048184280994,
      "grad_norm": 1.0115353991579583,
      "learning_rate": 4.989561586638831e-06,
      "loss": 0.4563,
      "step": 239
    },
    {
      "epoch": 0.015033119842152241,
      "grad_norm": 1.1112352476579102,
      "learning_rate": 5.01043841336117e-06,
      "loss": 0.4971,
      "step": 240
    },
    {
      "epoch": 0.015095757841494542,
      "grad_norm": 0.9930497304035641,
      "learning_rate": 5.031315240083507e-06,
      "loss": 0.4736,
      "step": 241
    },
    {
      "epoch": 0.015158395840836844,
      "grad_norm": 1.1491353417688646,
      "learning_rate": 5.0521920668058454e-06,
      "loss": 0.5277,
      "step": 242
    },
    {
      "epoch": 0.015221033840179145,
      "grad_norm": 0.9693151423138642,
      "learning_rate": 5.073068893528185e-06,
      "loss": 0.4509,
      "step": 243
    },
    {
      "epoch": 0.015283671839521446,
      "grad_norm": 1.003465901462217,
      "learning_rate": 5.093945720250523e-06,
      "loss": 0.4503,
      "step": 244
    },
    {
      "epoch": 0.015346309838863747,
      "grad_norm": 1.0782380380736178,
      "learning_rate": 5.1148225469728605e-06,
      "loss": 0.4671,
      "step": 245
    },
    {
      "epoch": 0.015408947838206047,
      "grad_norm": 1.0348471612655046,
      "learning_rate": 5.135699373695199e-06,
      "loss": 0.482,
      "step": 246
    },
    {
      "epoch": 0.015471585837548348,
      "grad_norm": 1.0801449483388263,
      "learning_rate": 5.156576200417537e-06,
      "loss": 0.4937,
      "step": 247
    },
    {
      "epoch": 0.01553422383689065,
      "grad_norm": 1.1709948799670535,
      "learning_rate": 5.177453027139875e-06,
      "loss": 0.4665,
      "step": 248
    },
    {
      "epoch": 0.015596861836232951,
      "grad_norm": 1.1696299371705117,
      "learning_rate": 5.198329853862213e-06,
      "loss": 0.4685,
      "step": 249
    },
    {
      "epoch": 0.01565949983557525,
      "grad_norm": 1.015865638859858,
      "learning_rate": 5.219206680584552e-06,
      "loss": 0.4601,
      "step": 250
    },
    {
      "epoch": 0.015722137834917552,
      "grad_norm": 1.0898355356334632,
      "learning_rate": 5.24008350730689e-06,
      "loss": 0.489,
      "step": 251
    },
    {
      "epoch": 0.015784775834259853,
      "grad_norm": 1.2561835047068353,
      "learning_rate": 5.260960334029228e-06,
      "loss": 0.457,
      "step": 252
    },
    {
      "epoch": 0.015847413833602154,
      "grad_norm": 1.1268966625675128,
      "learning_rate": 5.2818371607515665e-06,
      "loss": 0.4358,
      "step": 253
    },
    {
      "epoch": 0.015910051832944454,
      "grad_norm": 1.0205344319262977,
      "learning_rate": 5.302713987473904e-06,
      "loss": 0.4615,
      "step": 254
    },
    {
      "epoch": 0.015972689832286755,
      "grad_norm": 1.260319243673335,
      "learning_rate": 5.323590814196242e-06,
      "loss": 0.4771,
      "step": 255
    },
    {
      "epoch": 0.01603532783162906,
      "grad_norm": 1.0125043246026921,
      "learning_rate": 5.34446764091858e-06,
      "loss": 0.4568,
      "step": 256
    },
    {
      "epoch": 0.01609796583097136,
      "grad_norm": 5.783045109394937,
      "learning_rate": 5.365344467640919e-06,
      "loss": 0.4561,
      "step": 257
    },
    {
      "epoch": 0.01616060383031366,
      "grad_norm": 0.9627982924180688,
      "learning_rate": 5.3862212943632574e-06,
      "loss": 0.4527,
      "step": 258
    },
    {
      "epoch": 0.01622324182965596,
      "grad_norm": 1.2279708942831664,
      "learning_rate": 5.407098121085596e-06,
      "loss": 0.4674,
      "step": 259
    },
    {
      "epoch": 0.016285879828998262,
      "grad_norm": 0.9597868260368048,
      "learning_rate": 5.427974947807933e-06,
      "loss": 0.4673,
      "step": 260
    },
    {
      "epoch": 0.016348517828340563,
      "grad_norm": 1.006917661253672,
      "learning_rate": 5.448851774530272e-06,
      "loss": 0.4924,
      "step": 261
    },
    {
      "epoch": 0.016411155827682863,
      "grad_norm": 0.9424211662503136,
      "learning_rate": 5.46972860125261e-06,
      "loss": 0.4403,
      "step": 262
    },
    {
      "epoch": 0.016473793827025164,
      "grad_norm": 1.3849835440518103,
      "learning_rate": 5.490605427974948e-06,
      "loss": 0.49,
      "step": 263
    },
    {
      "epoch": 0.016536431826367465,
      "grad_norm": 1.0593460404904704,
      "learning_rate": 5.511482254697287e-06,
      "loss": 0.4197,
      "step": 264
    },
    {
      "epoch": 0.016599069825709765,
      "grad_norm": 1.1157084062962488,
      "learning_rate": 5.532359081419625e-06,
      "loss": 0.4865,
      "step": 265
    },
    {
      "epoch": 0.016661707825052066,
      "grad_norm": 1.1000867158785046,
      "learning_rate": 5.553235908141963e-06,
      "loss": 0.4675,
      "step": 266
    },
    {
      "epoch": 0.01672434582439437,
      "grad_norm": 1.0770863171528489,
      "learning_rate": 5.574112734864301e-06,
      "loss": 0.4957,
      "step": 267
    },
    {
      "epoch": 0.01678698382373667,
      "grad_norm": 0.9714024490747837,
      "learning_rate": 5.594989561586639e-06,
      "loss": 0.4538,
      "step": 268
    },
    {
      "epoch": 0.01684962182307897,
      "grad_norm": 1.165988927555262,
      "learning_rate": 5.615866388308977e-06,
      "loss": 0.4951,
      "step": 269
    },
    {
      "epoch": 0.016912259822421272,
      "grad_norm": 0.9730974309557999,
      "learning_rate": 5.636743215031316e-06,
      "loss": 0.5038,
      "step": 270
    },
    {
      "epoch": 0.016974897821763573,
      "grad_norm": 0.9606155679232153,
      "learning_rate": 5.657620041753654e-06,
      "loss": 0.4488,
      "step": 271
    },
    {
      "epoch": 0.017037535821105874,
      "grad_norm": 0.804596948948002,
      "learning_rate": 5.678496868475992e-06,
      "loss": 0.4338,
      "step": 272
    },
    {
      "epoch": 0.017100173820448174,
      "grad_norm": 0.82495372513188,
      "learning_rate": 5.69937369519833e-06,
      "loss": 0.4594,
      "step": 273
    },
    {
      "epoch": 0.017162811819790475,
      "grad_norm": 1.2140370923133463,
      "learning_rate": 5.7202505219206686e-06,
      "loss": 0.4423,
      "step": 274
    },
    {
      "epoch": 0.017225449819132776,
      "grad_norm": 1.0746583875902784,
      "learning_rate": 5.741127348643006e-06,
      "loss": 0.4638,
      "step": 275
    },
    {
      "epoch": 0.017288087818475077,
      "grad_norm": 1.1261341351323662,
      "learning_rate": 5.7620041753653444e-06,
      "loss": 0.4986,
      "step": 276
    },
    {
      "epoch": 0.017350725817817377,
      "grad_norm": 1.0389684736927427,
      "learning_rate": 5.782881002087684e-06,
      "loss": 0.4936,
      "step": 277
    },
    {
      "epoch": 0.01741336381715968,
      "grad_norm": 0.948839026108661,
      "learning_rate": 5.803757828810022e-06,
      "loss": 0.4362,
      "step": 278
    },
    {
      "epoch": 0.017476001816501982,
      "grad_norm": 0.9552582088939211,
      "learning_rate": 5.8246346555323595e-06,
      "loss": 0.4629,
      "step": 279
    },
    {
      "epoch": 0.017538639815844283,
      "grad_norm": 0.9962394884434483,
      "learning_rate": 5.845511482254698e-06,
      "loss": 0.4312,
      "step": 280
    },
    {
      "epoch": 0.017601277815186583,
      "grad_norm": 1.0011099403035508,
      "learning_rate": 5.866388308977035e-06,
      "loss": 0.4656,
      "step": 281
    },
    {
      "epoch": 0.017663915814528884,
      "grad_norm": 0.908268892804034,
      "learning_rate": 5.887265135699374e-06,
      "loss": 0.4397,
      "step": 282
    },
    {
      "epoch": 0.017726553813871185,
      "grad_norm": 1.0817853192596105,
      "learning_rate": 5.908141962421713e-06,
      "loss": 0.4829,
      "step": 283
    },
    {
      "epoch": 0.017789191813213485,
      "grad_norm": 1.0326547122871144,
      "learning_rate": 5.929018789144051e-06,
      "loss": 0.4826,
      "step": 284
    },
    {
      "epoch": 0.017851829812555786,
      "grad_norm": 1.0256726962101932,
      "learning_rate": 5.949895615866389e-06,
      "loss": 0.4502,
      "step": 285
    },
    {
      "epoch": 0.017914467811898087,
      "grad_norm": 0.9322448247089512,
      "learning_rate": 5.970772442588727e-06,
      "loss": 0.4498,
      "step": 286
    },
    {
      "epoch": 0.017977105811240388,
      "grad_norm": 1.0891047466866584,
      "learning_rate": 5.991649269311065e-06,
      "loss": 0.4523,
      "step": 287
    },
    {
      "epoch": 0.01803974381058269,
      "grad_norm": 1.0227559762408969,
      "learning_rate": 6.012526096033403e-06,
      "loss": 0.457,
      "step": 288
    },
    {
      "epoch": 0.018102381809924992,
      "grad_norm": 1.072883506924392,
      "learning_rate": 6.033402922755741e-06,
      "loss": 0.4818,
      "step": 289
    },
    {
      "epoch": 0.018165019809267293,
      "grad_norm": 0.9846795894752909,
      "learning_rate": 6.0542797494780806e-06,
      "loss": 0.4944,
      "step": 290
    },
    {
      "epoch": 0.018227657808609594,
      "grad_norm": 1.02540808295984,
      "learning_rate": 6.075156576200418e-06,
      "loss": 0.479,
      "step": 291
    },
    {
      "epoch": 0.018290295807951894,
      "grad_norm": 1.0847320806696585,
      "learning_rate": 6.096033402922756e-06,
      "loss": 0.4698,
      "step": 292
    },
    {
      "epoch": 0.018352933807294195,
      "grad_norm": 0.9891368008387524,
      "learning_rate": 6.116910229645095e-06,
      "loss": 0.4667,
      "step": 293
    },
    {
      "epoch": 0.018415571806636496,
      "grad_norm": 1.0530699761424416,
      "learning_rate": 6.137787056367432e-06,
      "loss": 0.4979,
      "step": 294
    },
    {
      "epoch": 0.018478209805978797,
      "grad_norm": 1.0643990774249072,
      "learning_rate": 6.158663883089771e-06,
      "loss": 0.4731,
      "step": 295
    },
    {
      "epoch": 0.018540847805321097,
      "grad_norm": 1.0328454767590418,
      "learning_rate": 6.179540709812108e-06,
      "loss": 0.4376,
      "step": 296
    },
    {
      "epoch": 0.018603485804663398,
      "grad_norm": 1.0017426844796349,
      "learning_rate": 6.200417536534447e-06,
      "loss": 0.4748,
      "step": 297
    },
    {
      "epoch": 0.0186661238040057,
      "grad_norm": 0.8399297779633434,
      "learning_rate": 6.221294363256786e-06,
      "loss": 0.4697,
      "step": 298
    },
    {
      "epoch": 0.018728761803348003,
      "grad_norm": 0.9658580400707074,
      "learning_rate": 6.242171189979124e-06,
      "loss": 0.421,
      "step": 299
    },
    {
      "epoch": 0.018791399802690303,
      "grad_norm": 1.0710821607185472,
      "learning_rate": 6.2630480167014616e-06,
      "loss": 0.4711,
      "step": 300
    },
    {
      "epoch": 0.018854037802032604,
      "grad_norm": 1.0924650153830093,
      "learning_rate": 6.2839248434238e-06,
      "loss": 0.4777,
      "step": 301
    },
    {
      "epoch": 0.018916675801374905,
      "grad_norm": 1.1367377343788119,
      "learning_rate": 6.304801670146138e-06,
      "loss": 0.4882,
      "step": 302
    },
    {
      "epoch": 0.018979313800717205,
      "grad_norm": 0.9941537493168215,
      "learning_rate": 6.325678496868477e-06,
      "loss": 0.4388,
      "step": 303
    },
    {
      "epoch": 0.019041951800059506,
      "grad_norm": 0.7827888700809601,
      "learning_rate": 6.346555323590815e-06,
      "loss": 0.4575,
      "step": 304
    },
    {
      "epoch": 0.019104589799401807,
      "grad_norm": 0.9769272407633175,
      "learning_rate": 6.367432150313153e-06,
      "loss": 0.4957,
      "step": 305
    },
    {
      "epoch": 0.019167227798744108,
      "grad_norm": 1.2174281887361098,
      "learning_rate": 6.388308977035491e-06,
      "loss": 0.4948,
      "step": 306
    },
    {
      "epoch": 0.019229865798086408,
      "grad_norm": 0.8228105578333935,
      "learning_rate": 6.409185803757829e-06,
      "loss": 0.4717,
      "step": 307
    },
    {
      "epoch": 0.01929250379742871,
      "grad_norm": 1.0140585185309565,
      "learning_rate": 6.4300626304801676e-06,
      "loss": 0.4698,
      "step": 308
    },
    {
      "epoch": 0.01935514179677101,
      "grad_norm": 1.0156574282778859,
      "learning_rate": 6.450939457202505e-06,
      "loss": 0.521,
      "step": 309
    },
    {
      "epoch": 0.019417779796113314,
      "grad_norm": 1.0524019675196545,
      "learning_rate": 6.471816283924844e-06,
      "loss": 0.4367,
      "step": 310
    },
    {
      "epoch": 0.019480417795455614,
      "grad_norm": 0.9262597784208584,
      "learning_rate": 6.492693110647183e-06,
      "loss": 0.4538,
      "step": 311
    },
    {
      "epoch": 0.019543055794797915,
      "grad_norm": 0.9426327634140615,
      "learning_rate": 6.51356993736952e-06,
      "loss": 0.3656,
      "step": 312
    },
    {
      "epoch": 0.019605693794140216,
      "grad_norm": 1.0964103544154795,
      "learning_rate": 6.5344467640918585e-06,
      "loss": 0.4674,
      "step": 313
    },
    {
      "epoch": 0.019668331793482517,
      "grad_norm": 0.994367132604669,
      "learning_rate": 6.555323590814197e-06,
      "loss": 0.494,
      "step": 314
    },
    {
      "epoch": 0.019730969792824817,
      "grad_norm": 1.1428090786125302,
      "learning_rate": 6.576200417536534e-06,
      "loss": 0.4932,
      "step": 315
    },
    {
      "epoch": 0.019793607792167118,
      "grad_norm": 1.0107334886798172,
      "learning_rate": 6.597077244258873e-06,
      "loss": 0.5024,
      "step": 316
    },
    {
      "epoch": 0.01985624579150942,
      "grad_norm": 0.9526133922658374,
      "learning_rate": 6.617954070981212e-06,
      "loss": 0.4813,
      "step": 317
    },
    {
      "epoch": 0.01991888379085172,
      "grad_norm": 1.030159601731955,
      "learning_rate": 6.638830897703549e-06,
      "loss": 0.4851,
      "step": 318
    },
    {
      "epoch": 0.01998152179019402,
      "grad_norm": 0.9932425525343781,
      "learning_rate": 6.659707724425888e-06,
      "loss": 0.4296,
      "step": 319
    },
    {
      "epoch": 0.02004415978953632,
      "grad_norm": 1.0137186265445013,
      "learning_rate": 6.680584551148226e-06,
      "loss": 0.4551,
      "step": 320
    },
    {
      "epoch": 0.020106797788878625,
      "grad_norm": 0.9362823546186946,
      "learning_rate": 6.701461377870564e-06,
      "loss": 0.4011,
      "step": 321
    },
    {
      "epoch": 0.020169435788220925,
      "grad_norm": 1.0016938508387971,
      "learning_rate": 6.722338204592902e-06,
      "loss": 0.4513,
      "step": 322
    },
    {
      "epoch": 0.020232073787563226,
      "grad_norm": 0.9830197209703241,
      "learning_rate": 6.743215031315241e-06,
      "loss": 0.4194,
      "step": 323
    },
    {
      "epoch": 0.020294711786905527,
      "grad_norm": 1.0832431016339947,
      "learning_rate": 6.7640918580375795e-06,
      "loss": 0.4661,
      "step": 324
    },
    {
      "epoch": 0.020357349786247828,
      "grad_norm": 1.0961203160099944,
      "learning_rate": 6.784968684759917e-06,
      "loss": 0.4592,
      "step": 325
    },
    {
      "epoch": 0.020419987785590128,
      "grad_norm": 1.0853213801865442,
      "learning_rate": 6.805845511482255e-06,
      "loss": 0.4669,
      "step": 326
    },
    {
      "epoch": 0.02048262578493243,
      "grad_norm": 0.9817851639344812,
      "learning_rate": 6.826722338204593e-06,
      "loss": 0.455,
      "step": 327
    },
    {
      "epoch": 0.02054526378427473,
      "grad_norm": 0.9745276162862645,
      "learning_rate": 6.847599164926931e-06,
      "loss": 0.4672,
      "step": 328
    },
    {
      "epoch": 0.02060790178361703,
      "grad_norm": 0.9485597310391845,
      "learning_rate": 6.86847599164927e-06,
      "loss": 0.4692,
      "step": 329
    },
    {
      "epoch": 0.02067053978295933,
      "grad_norm": 0.9728458860153021,
      "learning_rate": 6.889352818371609e-06,
      "loss": 0.4455,
      "step": 330
    },
    {
      "epoch": 0.02073317778230163,
      "grad_norm": 1.0720947528693843,
      "learning_rate": 6.910229645093946e-06,
      "loss": 0.5045,
      "step": 331
    },
    {
      "epoch": 0.020795815781643936,
      "grad_norm": 1.042613020058983,
      "learning_rate": 6.931106471816285e-06,
      "loss": 0.4667,
      "step": 332
    },
    {
      "epoch": 0.020858453780986237,
      "grad_norm": 1.0063932841881253,
      "learning_rate": 6.951983298538623e-06,
      "loss": 0.4681,
      "step": 333
    },
    {
      "epoch": 0.020921091780328537,
      "grad_norm": 1.0059503424200802,
      "learning_rate": 6.9728601252609605e-06,
      "loss": 0.4821,
      "step": 334
    },
    {
      "epoch": 0.020983729779670838,
      "grad_norm": 0.9824906716228554,
      "learning_rate": 6.993736951983299e-06,
      "loss": 0.4453,
      "step": 335
    },
    {
      "epoch": 0.02104636777901314,
      "grad_norm": 0.9233078581058215,
      "learning_rate": 7.014613778705638e-06,
      "loss": 0.4423,
      "step": 336
    },
    {
      "epoch": 0.02110900577835544,
      "grad_norm": 1.0265948474360445,
      "learning_rate": 7.035490605427976e-06,
      "loss": 0.4581,
      "step": 337
    },
    {
      "epoch": 0.02117164377769774,
      "grad_norm": 1.0193498228285711,
      "learning_rate": 7.056367432150314e-06,
      "loss": 0.454,
      "step": 338
    },
    {
      "epoch": 0.02123428177704004,
      "grad_norm": 0.9715118948039183,
      "learning_rate": 7.077244258872652e-06,
      "loss": 0.4881,
      "step": 339
    },
    {
      "epoch": 0.02129691977638234,
      "grad_norm": 0.9476429877580987,
      "learning_rate": 7.09812108559499e-06,
      "loss": 0.484,
      "step": 340
    },
    {
      "epoch": 0.021359557775724642,
      "grad_norm": 1.092216360482922,
      "learning_rate": 7.118997912317328e-06,
      "loss": 0.4814,
      "step": 341
    },
    {
      "epoch": 0.021422195775066943,
      "grad_norm": 1.0432762013730783,
      "learning_rate": 7.139874739039666e-06,
      "loss": 0.4688,
      "step": 342
    },
    {
      "epoch": 0.021484833774409247,
      "grad_norm": 1.1255772116234146,
      "learning_rate": 7.160751565762005e-06,
      "loss": 0.4873,
      "step": 343
    },
    {
      "epoch": 0.021547471773751548,
      "grad_norm": 1.0350980392828582,
      "learning_rate": 7.181628392484343e-06,
      "loss": 0.4634,
      "step": 344
    },
    {
      "epoch": 0.021610109773093848,
      "grad_norm": 1.0601640408558772,
      "learning_rate": 7.202505219206682e-06,
      "loss": 0.4744,
      "step": 345
    },
    {
      "epoch": 0.02167274777243615,
      "grad_norm": 1.0061908449669457,
      "learning_rate": 7.223382045929019e-06,
      "loss": 0.4412,
      "step": 346
    },
    {
      "epoch": 0.02173538577177845,
      "grad_norm": 1.0508727300316898,
      "learning_rate": 7.2442588726513575e-06,
      "loss": 0.4838,
      "step": 347
    },
    {
      "epoch": 0.02179802377112075,
      "grad_norm": 1.0903114376290823,
      "learning_rate": 7.265135699373696e-06,
      "loss": 0.5477,
      "step": 348
    },
    {
      "epoch": 0.02186066177046305,
      "grad_norm": 1.1430748532007675,
      "learning_rate": 7.286012526096033e-06,
      "loss": 0.5006,
      "step": 349
    },
    {
      "epoch": 0.02192329976980535,
      "grad_norm": 1.1784332576331287,
      "learning_rate": 7.3068893528183725e-06,
      "loss": 0.4369,
      "step": 350
    },
    {
      "epoch": 0.021985937769147652,
      "grad_norm": 0.9108783379481976,
      "learning_rate": 7.327766179540711e-06,
      "loss": 0.4981,
      "step": 351
    },
    {
      "epoch": 0.022048575768489953,
      "grad_norm": 0.9892401630581342,
      "learning_rate": 7.348643006263048e-06,
      "loss": 0.4387,
      "step": 352
    },
    {
      "epoch": 0.022111213767832254,
      "grad_norm": 0.892853676376306,
      "learning_rate": 7.369519832985387e-06,
      "loss": 0.4402,
      "step": 353
    },
    {
      "epoch": 0.022173851767174558,
      "grad_norm": 1.0114216292799079,
      "learning_rate": 7.390396659707725e-06,
      "loss": 0.4749,
      "step": 354
    },
    {
      "epoch": 0.02223648976651686,
      "grad_norm": 0.9922434476456627,
      "learning_rate": 7.411273486430063e-06,
      "loss": 0.4171,
      "step": 355
    },
    {
      "epoch": 0.02229912776585916,
      "grad_norm": 0.9772069829255501,
      "learning_rate": 7.432150313152402e-06,
      "loss": 0.4822,
      "step": 356
    },
    {
      "epoch": 0.02236176576520146,
      "grad_norm": 0.9399639872125047,
      "learning_rate": 7.45302713987474e-06,
      "loss": 0.4378,
      "step": 357
    },
    {
      "epoch": 0.02242440376454376,
      "grad_norm": 1.009429971197964,
      "learning_rate": 7.473903966597078e-06,
      "loss": 0.4977,
      "step": 358
    },
    {
      "epoch": 0.02248704176388606,
      "grad_norm": 0.9942475303092013,
      "learning_rate": 7.494780793319416e-06,
      "loss": 0.5264,
      "step": 359
    },
    {
      "epoch": 0.022549679763228362,
      "grad_norm": 0.9505252607304049,
      "learning_rate": 7.515657620041754e-06,
      "loss": 0.4614,
      "step": 360
    },
    {
      "epoch": 0.022612317762570663,
      "grad_norm": 1.0349102788798663,
      "learning_rate": 7.536534446764092e-06,
      "loss": 0.4749,
      "step": 361
    },
    {
      "epoch": 0.022674955761912963,
      "grad_norm": 0.9967059255639978,
      "learning_rate": 7.55741127348643e-06,
      "loss": 0.4495,
      "step": 362
    },
    {
      "epoch": 0.022737593761255264,
      "grad_norm": 0.9691161651484356,
      "learning_rate": 7.5782881002087694e-06,
      "loss": 0.4231,
      "step": 363
    },
    {
      "epoch": 0.022800231760597568,
      "grad_norm": 1.0471392917543423,
      "learning_rate": 7.599164926931108e-06,
      "loss": 0.4804,
      "step": 364
    },
    {
      "epoch": 0.02286286975993987,
      "grad_norm": 1.1466097172133765,
      "learning_rate": 7.620041753653445e-06,
      "loss": 0.4777,
      "step": 365
    },
    {
      "epoch": 0.02292550775928217,
      "grad_norm": 1.0810933896664272,
      "learning_rate": 7.640918580375784e-06,
      "loss": 0.4708,
      "step": 366
    },
    {
      "epoch": 0.02298814575862447,
      "grad_norm": 0.9477066851146424,
      "learning_rate": 7.661795407098122e-06,
      "loss": 0.4618,
      "step": 367
    },
    {
      "epoch": 0.02305078375796677,
      "grad_norm": 1.0526356549997242,
      "learning_rate": 7.68267223382046e-06,
      "loss": 0.4857,
      "step": 368
    },
    {
      "epoch": 0.02311342175730907,
      "grad_norm": 1.0137781803367414,
      "learning_rate": 7.703549060542797e-06,
      "loss": 0.4601,
      "step": 369
    },
    {
      "epoch": 0.023176059756651372,
      "grad_norm": 0.9035822104579848,
      "learning_rate": 7.724425887265137e-06,
      "loss": 0.4423,
      "step": 370
    },
    {
      "epoch": 0.023238697755993673,
      "grad_norm": 0.9890150970382936,
      "learning_rate": 7.745302713987475e-06,
      "loss": 0.4892,
      "step": 371
    },
    {
      "epoch": 0.023301335755335974,
      "grad_norm": 0.9752754153103337,
      "learning_rate": 7.766179540709812e-06,
      "loss": 0.4647,
      "step": 372
    },
    {
      "epoch": 0.023363973754678274,
      "grad_norm": 1.0294150072952917,
      "learning_rate": 7.78705636743215e-06,
      "loss": 0.4585,
      "step": 373
    },
    {
      "epoch": 0.023426611754020575,
      "grad_norm": 1.0368743151752813,
      "learning_rate": 7.807933194154489e-06,
      "loss": 0.4911,
      "step": 374
    },
    {
      "epoch": 0.02348924975336288,
      "grad_norm": 0.9236772394686724,
      "learning_rate": 7.828810020876827e-06,
      "loss": 0.4694,
      "step": 375
    },
    {
      "epoch": 0.02355188775270518,
      "grad_norm": 0.9764081240870652,
      "learning_rate": 7.849686847599166e-06,
      "loss": 0.4672,
      "step": 376
    },
    {
      "epoch": 0.02361452575204748,
      "grad_norm": 1.025613826615659,
      "learning_rate": 7.870563674321504e-06,
      "loss": 0.4618,
      "step": 377
    },
    {
      "epoch": 0.02367716375138978,
      "grad_norm": 1.0120143895301574,
      "learning_rate": 7.891440501043842e-06,
      "loss": 0.4473,
      "step": 378
    },
    {
      "epoch": 0.023739801750732082,
      "grad_norm": 0.8105118221530545,
      "learning_rate": 7.91231732776618e-06,
      "loss": 0.4704,
      "step": 379
    },
    {
      "epoch": 0.023802439750074383,
      "grad_norm": 1.0360834970242987,
      "learning_rate": 7.933194154488519e-06,
      "loss": 0.4632,
      "step": 380
    },
    {
      "epoch": 0.023865077749416683,
      "grad_norm": 0.9131812803758628,
      "learning_rate": 7.954070981210856e-06,
      "loss": 0.3959,
      "step": 381
    },
    {
      "epoch": 0.023927715748758984,
      "grad_norm": 1.0862417475805177,
      "learning_rate": 7.974947807933194e-06,
      "loss": 0.4341,
      "step": 382
    },
    {
      "epoch": 0.023990353748101285,
      "grad_norm": 1.1857066699203984,
      "learning_rate": 7.995824634655534e-06,
      "loss": 0.3944,
      "step": 383
    },
    {
      "epoch": 0.024052991747443585,
      "grad_norm": 1.0318652268164963,
      "learning_rate": 8.016701461377872e-06,
      "loss": 0.4907,
      "step": 384
    },
    {
      "epoch": 0.024115629746785886,
      "grad_norm": 1.150685862822361,
      "learning_rate": 8.037578288100209e-06,
      "loss": 0.483,
      "step": 385
    },
    {
      "epoch": 0.02417826774612819,
      "grad_norm": 1.1380555353977218,
      "learning_rate": 8.058455114822547e-06,
      "loss": 0.4554,
      "step": 386
    },
    {
      "epoch": 0.02424090574547049,
      "grad_norm": 0.9755624010943118,
      "learning_rate": 8.079331941544886e-06,
      "loss": 0.4541,
      "step": 387
    },
    {
      "epoch": 0.02430354374481279,
      "grad_norm": 1.039565472122167,
      "learning_rate": 8.100208768267224e-06,
      "loss": 0.4325,
      "step": 388
    },
    {
      "epoch": 0.024366181744155092,
      "grad_norm": 0.8816276176909026,
      "learning_rate": 8.121085594989562e-06,
      "loss": 0.4616,
      "step": 389
    },
    {
      "epoch": 0.024428819743497393,
      "grad_norm": 1.037750704492878,
      "learning_rate": 8.1419624217119e-06,
      "loss": 0.5099,
      "step": 390
    },
    {
      "epoch": 0.024491457742839694,
      "grad_norm": 1.0139945790367009,
      "learning_rate": 8.162839248434239e-06,
      "loss": 0.5087,
      "step": 391
    },
    {
      "epoch": 0.024554095742181994,
      "grad_norm": 0.8376043073774213,
      "learning_rate": 8.183716075156577e-06,
      "loss": 0.4529,
      "step": 392
    },
    {
      "epoch": 0.024616733741524295,
      "grad_norm": 1.0786410074917778,
      "learning_rate": 8.204592901878914e-06,
      "loss": 0.4523,
      "step": 393
    },
    {
      "epoch": 0.024679371740866596,
      "grad_norm": 1.0480518416738163,
      "learning_rate": 8.225469728601253e-06,
      "loss": 0.4616,
      "step": 394
    },
    {
      "epoch": 0.024742009740208896,
      "grad_norm": 1.0260583482672316,
      "learning_rate": 8.246346555323591e-06,
      "loss": 0.427,
      "step": 395
    },
    {
      "epoch": 0.024804647739551197,
      "grad_norm": 1.1264105332455978,
      "learning_rate": 8.267223382045931e-06,
      "loss": 0.4976,
      "step": 396
    },
    {
      "epoch": 0.0248672857388935,
      "grad_norm": 1.032063917510832,
      "learning_rate": 8.288100208768268e-06,
      "loss": 0.5019,
      "step": 397
    },
    {
      "epoch": 0.024929923738235802,
      "grad_norm": 0.9715164719205722,
      "learning_rate": 8.308977035490606e-06,
      "loss": 0.4455,
      "step": 398
    },
    {
      "epoch": 0.024992561737578103,
      "grad_norm": 0.9814933514219619,
      "learning_rate": 8.329853862212944e-06,
      "loss": 0.4456,
      "step": 399
    },
    {
      "epoch": 0.025055199736920403,
      "grad_norm": 1.0230326730865775,
      "learning_rate": 8.350730688935283e-06,
      "loss": 0.4471,
      "step": 400
    },
    {
      "epoch": 0.025117837736262704,
      "grad_norm": 1.0595310751493696,
      "learning_rate": 8.371607515657621e-06,
      "loss": 0.4952,
      "step": 401
    },
    {
      "epoch": 0.025180475735605005,
      "grad_norm": 0.9772124143996185,
      "learning_rate": 8.392484342379958e-06,
      "loss": 0.4424,
      "step": 402
    },
    {
      "epoch": 0.025243113734947305,
      "grad_norm": 1.0524528330565162,
      "learning_rate": 8.413361169102298e-06,
      "loss": 0.4521,
      "step": 403
    },
    {
      "epoch": 0.025305751734289606,
      "grad_norm": 0.9482437461981857,
      "learning_rate": 8.434237995824636e-06,
      "loss": 0.4439,
      "step": 404
    },
    {
      "epoch": 0.025368389733631907,
      "grad_norm": 0.9526838767880685,
      "learning_rate": 8.455114822546974e-06,
      "loss": 0.4119,
      "step": 405
    },
    {
      "epoch": 0.025431027732974208,
      "grad_norm": 1.0310039595839549,
      "learning_rate": 8.475991649269311e-06,
      "loss": 0.4911,
      "step": 406
    },
    {
      "epoch": 0.025493665732316508,
      "grad_norm": 1.2591414480657233,
      "learning_rate": 8.49686847599165e-06,
      "loss": 0.503,
      "step": 407
    },
    {
      "epoch": 0.025556303731658812,
      "grad_norm": 1.1191338580026253,
      "learning_rate": 8.517745302713988e-06,
      "loss": 0.4951,
      "step": 408
    },
    {
      "epoch": 0.025618941731001113,
      "grad_norm": 1.02116195368937,
      "learning_rate": 8.538622129436326e-06,
      "loss": 0.4613,
      "step": 409
    },
    {
      "epoch": 0.025681579730343414,
      "grad_norm": 1.0671134830736575,
      "learning_rate": 8.559498956158664e-06,
      "loss": 0.4676,
      "step": 410
    },
    {
      "epoch": 0.025744217729685714,
      "grad_norm": 1.071283682343229,
      "learning_rate": 8.580375782881003e-06,
      "loss": 0.4814,
      "step": 411
    },
    {
      "epoch": 0.025806855729028015,
      "grad_norm": 0.9395657309429192,
      "learning_rate": 8.601252609603341e-06,
      "loss": 0.4845,
      "step": 412
    },
    {
      "epoch": 0.025869493728370316,
      "grad_norm": 1.1414152937755306,
      "learning_rate": 8.62212943632568e-06,
      "loss": 0.4562,
      "step": 413
    },
    {
      "epoch": 0.025932131727712616,
      "grad_norm": 0.9606318805371331,
      "learning_rate": 8.643006263048018e-06,
      "loss": 0.382,
      "step": 414
    },
    {
      "epoch": 0.025994769727054917,
      "grad_norm": 0.9775244840413565,
      "learning_rate": 8.663883089770355e-06,
      "loss": 0.4717,
      "step": 415
    },
    {
      "epoch": 0.026057407726397218,
      "grad_norm": 1.0218755935329171,
      "learning_rate": 8.684759916492695e-06,
      "loss": 0.4704,
      "step": 416
    },
    {
      "epoch": 0.02612004572573952,
      "grad_norm": 1.0144329241935746,
      "learning_rate": 8.705636743215033e-06,
      "loss": 0.5093,
      "step": 417
    },
    {
      "epoch": 0.02618268372508182,
      "grad_norm": 1.0379790069097439,
      "learning_rate": 8.72651356993737e-06,
      "loss": 0.4927,
      "step": 418
    },
    {
      "epoch": 0.026245321724424123,
      "grad_norm": 1.1413911809593913,
      "learning_rate": 8.747390396659708e-06,
      "loss": 0.4624,
      "step": 419
    },
    {
      "epoch": 0.026307959723766424,
      "grad_norm": 1.090739939005204,
      "learning_rate": 8.768267223382046e-06,
      "loss": 0.4669,
      "step": 420
    },
    {
      "epoch": 0.026370597723108725,
      "grad_norm": 1.359887219753948,
      "learning_rate": 8.789144050104385e-06,
      "loss": 0.4675,
      "step": 421
    },
    {
      "epoch": 0.026433235722451025,
      "grad_norm": 1.1462638444121762,
      "learning_rate": 8.810020876826723e-06,
      "loss": 0.4464,
      "step": 422
    },
    {
      "epoch": 0.026495873721793326,
      "grad_norm": 1.0289071005825086,
      "learning_rate": 8.830897703549061e-06,
      "loss": 0.4464,
      "step": 423
    },
    {
      "epoch": 0.026558511721135627,
      "grad_norm": 1.1092552650257865,
      "learning_rate": 8.8517745302714e-06,
      "loss": 0.488,
      "step": 424
    },
    {
      "epoch": 0.026621149720477928,
      "grad_norm": 0.9723700364332956,
      "learning_rate": 8.872651356993738e-06,
      "loss": 0.4403,
      "step": 425
    },
    {
      "epoch": 0.026683787719820228,
      "grad_norm": 1.0884467003196372,
      "learning_rate": 8.893528183716076e-06,
      "loss": 0.4897,
      "step": 426
    },
    {
      "epoch": 0.02674642571916253,
      "grad_norm": 1.0339845314560088,
      "learning_rate": 8.914405010438413e-06,
      "loss": 0.4508,
      "step": 427
    },
    {
      "epoch": 0.02680906371850483,
      "grad_norm": 1.0215559406421708,
      "learning_rate": 8.935281837160751e-06,
      "loss": 0.4455,
      "step": 428
    },
    {
      "epoch": 0.02687170171784713,
      "grad_norm": 1.3859299271398258,
      "learning_rate": 8.95615866388309e-06,
      "loss": 0.4688,
      "step": 429
    },
    {
      "epoch": 0.026934339717189434,
      "grad_norm": 0.9658495508125703,
      "learning_rate": 8.97703549060543e-06,
      "loss": 0.4715,
      "step": 430
    },
    {
      "epoch": 0.026996977716531735,
      "grad_norm": 0.8392638969134845,
      "learning_rate": 8.997912317327767e-06,
      "loss": 0.4373,
      "step": 431
    },
    {
      "epoch": 0.027059615715874036,
      "grad_norm": 1.0236434732686934,
      "learning_rate": 9.018789144050105e-06,
      "loss": 0.4407,
      "step": 432
    },
    {
      "epoch": 0.027122253715216336,
      "grad_norm": 0.9506770278455658,
      "learning_rate": 9.039665970772443e-06,
      "loss": 0.4303,
      "step": 433
    },
    {
      "epoch": 0.027184891714558637,
      "grad_norm": 0.9762568425740348,
      "learning_rate": 9.060542797494782e-06,
      "loss": 0.4551,
      "step": 434
    },
    {
      "epoch": 0.027247529713900938,
      "grad_norm": 0.9493415685861777,
      "learning_rate": 9.08141962421712e-06,
      "loss": 0.4291,
      "step": 435
    },
    {
      "epoch": 0.02731016771324324,
      "grad_norm": 0.98682363472654,
      "learning_rate": 9.102296450939458e-06,
      "loss": 0.4223,
      "step": 436
    },
    {
      "epoch": 0.02737280571258554,
      "grad_norm": 1.0583860429633773,
      "learning_rate": 9.123173277661797e-06,
      "loss": 0.5016,
      "step": 437
    },
    {
      "epoch": 0.02743544371192784,
      "grad_norm": 1.094915578684592,
      "learning_rate": 9.144050104384135e-06,
      "loss": 0.4812,
      "step": 438
    },
    {
      "epoch": 0.02749808171127014,
      "grad_norm": 1.1489910075375958,
      "learning_rate": 9.164926931106473e-06,
      "loss": 0.4448,
      "step": 439
    },
    {
      "epoch": 0.027560719710612445,
      "grad_norm": 1.0651067721530127,
      "learning_rate": 9.18580375782881e-06,
      "loss": 0.451,
      "step": 440
    },
    {
      "epoch": 0.027623357709954745,
      "grad_norm": 0.984838143778904,
      "learning_rate": 9.206680584551148e-06,
      "loss": 0.4447,
      "step": 441
    },
    {
      "epoch": 0.027685995709297046,
      "grad_norm": 0.9919782890265614,
      "learning_rate": 9.227557411273487e-06,
      "loss": 0.4561,
      "step": 442
    },
    {
      "epoch": 0.027748633708639347,
      "grad_norm": 1.0154264201915884,
      "learning_rate": 9.248434237995825e-06,
      "loss": 0.4458,
      "step": 443
    },
    {
      "epoch": 0.027811271707981648,
      "grad_norm": 1.0792354186623045,
      "learning_rate": 9.269311064718163e-06,
      "loss": 0.4239,
      "step": 444
    },
    {
      "epoch": 0.027873909707323948,
      "grad_norm": 1.34376289454518,
      "learning_rate": 9.290187891440502e-06,
      "loss": 0.4765,
      "step": 445
    },
    {
      "epoch": 0.02793654770666625,
      "grad_norm": 1.024778963163423,
      "learning_rate": 9.31106471816284e-06,
      "loss": 0.4298,
      "step": 446
    },
    {
      "epoch": 0.02799918570600855,
      "grad_norm": 1.0031450242070024,
      "learning_rate": 9.331941544885179e-06,
      "loss": 0.4481,
      "step": 447
    },
    {
      "epoch": 0.02806182370535085,
      "grad_norm": 1.0626565643549075,
      "learning_rate": 9.352818371607515e-06,
      "loss": 0.4604,
      "step": 448
    },
    {
      "epoch": 0.02812446170469315,
      "grad_norm": 0.9551350468949873,
      "learning_rate": 9.373695198329854e-06,
      "loss": 0.4503,
      "step": 449
    },
    {
      "epoch": 0.02818709970403545,
      "grad_norm": 1.504403744828678,
      "learning_rate": 9.394572025052194e-06,
      "loss": 0.4255,
      "step": 450
    },
    {
      "epoch": 0.028249737703377756,
      "grad_norm": 0.9780721403717871,
      "learning_rate": 9.415448851774532e-06,
      "loss": 0.4795,
      "step": 451
    },
    {
      "epoch": 0.028312375702720056,
      "grad_norm": 0.9786691712435661,
      "learning_rate": 9.436325678496869e-06,
      "loss": 0.4787,
      "step": 452
    },
    {
      "epoch": 0.028375013702062357,
      "grad_norm": 1.0643419116158872,
      "learning_rate": 9.457202505219207e-06,
      "loss": 0.4649,
      "step": 453
    },
    {
      "epoch": 0.028437651701404658,
      "grad_norm": 1.0986803047589564,
      "learning_rate": 9.478079331941545e-06,
      "loss": 0.4881,
      "step": 454
    },
    {
      "epoch": 0.02850028970074696,
      "grad_norm": 2.2652590419375427,
      "learning_rate": 9.498956158663884e-06,
      "loss": 0.4163,
      "step": 455
    },
    {
      "epoch": 0.02856292770008926,
      "grad_norm": 1.0581246843333987,
      "learning_rate": 9.519832985386222e-06,
      "loss": 0.4753,
      "step": 456
    },
    {
      "epoch": 0.02862556569943156,
      "grad_norm": 0.9339112351436439,
      "learning_rate": 9.54070981210856e-06,
      "loss": 0.4728,
      "step": 457
    },
    {
      "epoch": 0.02868820369877386,
      "grad_norm": 1.0919109524183666,
      "learning_rate": 9.561586638830899e-06,
      "loss": 0.453,
      "step": 458
    },
    {
      "epoch": 0.02875084169811616,
      "grad_norm": 1.1234203968235812,
      "learning_rate": 9.582463465553237e-06,
      "loss": 0.4842,
      "step": 459
    },
    {
      "epoch": 0.028813479697458462,
      "grad_norm": 0.8195552089319333,
      "learning_rate": 9.603340292275575e-06,
      "loss": 0.4697,
      "step": 460
    },
    {
      "epoch": 0.028876117696800763,
      "grad_norm": 0.9771454273604905,
      "learning_rate": 9.624217118997912e-06,
      "loss": 0.468,
      "step": 461
    },
    {
      "epoch": 0.028938755696143067,
      "grad_norm": 0.999655777687778,
      "learning_rate": 9.64509394572025e-06,
      "loss": 0.4898,
      "step": 462
    },
    {
      "epoch": 0.029001393695485368,
      "grad_norm": 1.050439874886279,
      "learning_rate": 9.66597077244259e-06,
      "loss": 0.4998,
      "step": 463
    },
    {
      "epoch": 0.029064031694827668,
      "grad_norm": 1.038096993298549,
      "learning_rate": 9.686847599164927e-06,
      "loss": 0.4304,
      "step": 464
    },
    {
      "epoch": 0.02912666969416997,
      "grad_norm": 0.9590823159093111,
      "learning_rate": 9.707724425887266e-06,
      "loss": 0.4569,
      "step": 465
    },
    {
      "epoch": 0.02918930769351227,
      "grad_norm": 1.0126144191437996,
      "learning_rate": 9.728601252609604e-06,
      "loss": 0.416,
      "step": 466
    },
    {
      "epoch": 0.02925194569285457,
      "grad_norm": 0.990069353375383,
      "learning_rate": 9.749478079331942e-06,
      "loss": 0.4583,
      "step": 467
    },
    {
      "epoch": 0.02931458369219687,
      "grad_norm": 1.0380118484199026,
      "learning_rate": 9.77035490605428e-06,
      "loss": 0.4956,
      "step": 468
    },
    {
      "epoch": 0.02937722169153917,
      "grad_norm": 1.1014460595693187,
      "learning_rate": 9.791231732776619e-06,
      "loss": 0.4977,
      "step": 469
    },
    {
      "epoch": 0.029439859690881472,
      "grad_norm": 1.0085092508405116,
      "learning_rate": 9.812108559498957e-06,
      "loss": 0.4648,
      "step": 470
    },
    {
      "epoch": 0.029502497690223773,
      "grad_norm": 0.9907624588013442,
      "learning_rate": 9.832985386221296e-06,
      "loss": 0.4884,
      "step": 471
    },
    {
      "epoch": 0.029565135689566074,
      "grad_norm": 1.0051374056097386,
      "learning_rate": 9.853862212943634e-06,
      "loss": 0.5023,
      "step": 472
    },
    {
      "epoch": 0.029627773688908378,
      "grad_norm": 0.9108677857108926,
      "learning_rate": 9.87473903966597e-06,
      "loss": 0.4439,
      "step": 473
    },
    {
      "epoch": 0.02969041168825068,
      "grad_norm": 1.0280437212636164,
      "learning_rate": 9.895615866388309e-06,
      "loss": 0.465,
      "step": 474
    },
    {
      "epoch": 0.02975304968759298,
      "grad_norm": 0.9906976080255074,
      "learning_rate": 9.916492693110647e-06,
      "loss": 0.5124,
      "step": 475
    },
    {
      "epoch": 0.02981568768693528,
      "grad_norm": 1.0164079870710059,
      "learning_rate": 9.937369519832987e-06,
      "loss": 0.4487,
      "step": 476
    },
    {
      "epoch": 0.02987832568627758,
      "grad_norm": 1.214954161070769,
      "learning_rate": 9.958246346555324e-06,
      "loss": 0.4621,
      "step": 477
    },
    {
      "epoch": 0.02994096368561988,
      "grad_norm": 0.9466834711550224,
      "learning_rate": 9.979123173277662e-06,
      "loss": 0.4288,
      "step": 478
    },
    {
      "epoch": 0.030003601684962182,
      "grad_norm": 1.0107912584126701,
      "learning_rate": 1e-05,
      "loss": 0.4327,
      "step": 479
    },
    {
      "epoch": 0.030066239684304483,
      "grad_norm": 1.0436277987152458,
      "learning_rate": 9.999999897099536e-06,
      "loss": 0.4946,
      "step": 480
    },
    {
      "epoch": 0.030128877683646783,
      "grad_norm": 1.1407797723300557,
      "learning_rate": 9.999999588398148e-06,
      "loss": 0.5431,
      "step": 481
    },
    {
      "epoch": 0.030191515682989084,
      "grad_norm": 0.9581165982334768,
      "learning_rate": 9.99999907389585e-06,
      "loss": 0.433,
      "step": 482
    },
    {
      "epoch": 0.030254153682331385,
      "grad_norm": 1.0943637577641834,
      "learning_rate": 9.99999835359266e-06,
      "loss": 0.4752,
      "step": 483
    },
    {
      "epoch": 0.03031679168167369,
      "grad_norm": 0.9923849428331267,
      "learning_rate": 9.999997427488612e-06,
      "loss": 0.4821,
      "step": 484
    },
    {
      "epoch": 0.03037942968101599,
      "grad_norm": 0.9175734193478818,
      "learning_rate": 9.999996295583741e-06,
      "loss": 0.4388,
      "step": 485
    },
    {
      "epoch": 0.03044206768035829,
      "grad_norm": 1.0024348356344028,
      "learning_rate": 9.999994957878094e-06,
      "loss": 0.4839,
      "step": 486
    },
    {
      "epoch": 0.03050470567970059,
      "grad_norm": 0.9854766442151004,
      "learning_rate": 9.999993414371726e-06,
      "loss": 0.46,
      "step": 487
    },
    {
      "epoch": 0.03056734367904289,
      "grad_norm": 0.9245639832385998,
      "learning_rate": 9.999991665064703e-06,
      "loss": 0.4281,
      "step": 488
    },
    {
      "epoch": 0.030629981678385192,
      "grad_norm": 0.9248227741637166,
      "learning_rate": 9.999989709957093e-06,
      "loss": 0.4771,
      "step": 489
    },
    {
      "epoch": 0.030692619677727493,
      "grad_norm": 0.92059840901639,
      "learning_rate": 9.99998754904898e-06,
      "loss": 0.4139,
      "step": 490
    },
    {
      "epoch": 0.030755257677069794,
      "grad_norm": 1.0609216515151922,
      "learning_rate": 9.99998518234045e-06,
      "loss": 0.466,
      "step": 491
    },
    {
      "epoch": 0.030817895676412094,
      "grad_norm": 0.9465076246325514,
      "learning_rate": 9.999982609831602e-06,
      "loss": 0.4717,
      "step": 492
    },
    {
      "epoch": 0.030880533675754395,
      "grad_norm": 1.1400113555055207,
      "learning_rate": 9.999979831522542e-06,
      "loss": 0.4905,
      "step": 493
    },
    {
      "epoch": 0.030943171675096696,
      "grad_norm": 0.969603267295193,
      "learning_rate": 9.999976847413384e-06,
      "loss": 0.4653,
      "step": 494
    },
    {
      "epoch": 0.031005809674439,
      "grad_norm": 1.0096136673356448,
      "learning_rate": 9.999973657504252e-06,
      "loss": 0.4862,
      "step": 495
    },
    {
      "epoch": 0.0310684476737813,
      "grad_norm": 1.00760839191466,
      "learning_rate": 9.999970261795275e-06,
      "loss": 0.48,
      "step": 496
    },
    {
      "epoch": 0.0311310856731236,
      "grad_norm": 0.9691567894864449,
      "learning_rate": 9.999966660286595e-06,
      "loss": 0.4521,
      "step": 497
    },
    {
      "epoch": 0.031193723672465902,
      "grad_norm": 1.0578340506578705,
      "learning_rate": 9.999962852978358e-06,
      "loss": 0.4849,
      "step": 498
    },
    {
      "epoch": 0.0312563616718082,
      "grad_norm": 0.9764114767764127,
      "learning_rate": 9.999958839870722e-06,
      "loss": 0.4267,
      "step": 499
    },
    {
      "epoch": 0.0313189996711505,
      "grad_norm": 1.0074380395987035,
      "learning_rate": 9.999954620963855e-06,
      "loss": 0.4799,
      "step": 500
    },
    {
      "epoch": 0.031381637670492804,
      "grad_norm": 1.1811927042275638,
      "learning_rate": 9.999950196257924e-06,
      "loss": 0.4904,
      "step": 501
    },
    {
      "epoch": 0.031444275669835105,
      "grad_norm": 1.0041882768132266,
      "learning_rate": 9.999945565753116e-06,
      "loss": 0.4806,
      "step": 502
    },
    {
      "epoch": 0.031506913669177405,
      "grad_norm": 1.0273531123174042,
      "learning_rate": 9.999940729449622e-06,
      "loss": 0.454,
      "step": 503
    },
    {
      "epoch": 0.031569551668519706,
      "grad_norm": 0.9722350000020039,
      "learning_rate": 9.999935687347638e-06,
      "loss": 0.4295,
      "step": 504
    },
    {
      "epoch": 0.03163218966786201,
      "grad_norm": 0.92030913206606,
      "learning_rate": 9.999930439447372e-06,
      "loss": 0.4687,
      "step": 505
    },
    {
      "epoch": 0.03169482766720431,
      "grad_norm": 1.0924693890570083,
      "learning_rate": 9.999924985749044e-06,
      "loss": 0.4898,
      "step": 506
    },
    {
      "epoch": 0.03175746566654661,
      "grad_norm": 1.0646972647631245,
      "learning_rate": 9.999919326252874e-06,
      "loss": 0.4895,
      "step": 507
    },
    {
      "epoch": 0.03182010366588891,
      "grad_norm": 0.8652912409826867,
      "learning_rate": 9.999913460959097e-06,
      "loss": 0.4549,
      "step": 508
    },
    {
      "epoch": 0.03188274166523121,
      "grad_norm": 0.9624685363914474,
      "learning_rate": 9.999907389867954e-06,
      "loss": 0.4506,
      "step": 509
    },
    {
      "epoch": 0.03194537966457351,
      "grad_norm": 1.0315788411209288,
      "learning_rate": 9.999901112979692e-06,
      "loss": 0.5095,
      "step": 510
    },
    {
      "epoch": 0.03200801766391582,
      "grad_norm": 0.9846071396915839,
      "learning_rate": 9.999894630294576e-06,
      "loss": 0.5179,
      "step": 511
    },
    {
      "epoch": 0.03207065566325812,
      "grad_norm": 1.0627532866412681,
      "learning_rate": 9.999887941812866e-06,
      "loss": 0.488,
      "step": 512
    },
    {
      "epoch": 0.03213329366260042,
      "grad_norm": 0.9742328408810338,
      "learning_rate": 9.999881047534842e-06,
      "loss": 0.4739,
      "step": 513
    },
    {
      "epoch": 0.03219593166194272,
      "grad_norm": 0.9507339577337984,
      "learning_rate": 9.999873947460787e-06,
      "loss": 0.4452,
      "step": 514
    },
    {
      "epoch": 0.03225856966128502,
      "grad_norm": 0.9704327396775589,
      "learning_rate": 9.99986664159099e-06,
      "loss": 0.4349,
      "step": 515
    },
    {
      "epoch": 0.03232120766062732,
      "grad_norm": 0.9318856938466582,
      "learning_rate": 9.999859129925755e-06,
      "loss": 0.465,
      "step": 516
    },
    {
      "epoch": 0.03238384565996962,
      "grad_norm": 1.018203564786288,
      "learning_rate": 9.99985141246539e-06,
      "loss": 0.4696,
      "step": 517
    },
    {
      "epoch": 0.03244648365931192,
      "grad_norm": 1.0303214212289362,
      "learning_rate": 9.999843489210214e-06,
      "loss": 0.4796,
      "step": 518
    },
    {
      "epoch": 0.03250912165865422,
      "grad_norm": 0.8916298851440834,
      "learning_rate": 9.999835360160552e-06,
      "loss": 0.4438,
      "step": 519
    },
    {
      "epoch": 0.032571759657996524,
      "grad_norm": 1.075470573489586,
      "learning_rate": 9.999827025316737e-06,
      "loss": 0.4633,
      "step": 520
    },
    {
      "epoch": 0.032634397657338825,
      "grad_norm": 0.9824834115931615,
      "learning_rate": 9.999818484679115e-06,
      "loss": 0.4318,
      "step": 521
    },
    {
      "epoch": 0.032697035656681125,
      "grad_norm": 0.9773623813408532,
      "learning_rate": 9.999809738248035e-06,
      "loss": 0.4415,
      "step": 522
    },
    {
      "epoch": 0.032759673656023426,
      "grad_norm": 0.9665533542360624,
      "learning_rate": 9.999800786023858e-06,
      "loss": 0.4053,
      "step": 523
    },
    {
      "epoch": 0.03282231165536573,
      "grad_norm": 0.9851886697033954,
      "learning_rate": 9.999791628006953e-06,
      "loss": 0.5264,
      "step": 524
    },
    {
      "epoch": 0.03288494965470803,
      "grad_norm": 1.0079922070618188,
      "learning_rate": 9.999782264197699e-06,
      "loss": 0.4498,
      "step": 525
    },
    {
      "epoch": 0.03294758765405033,
      "grad_norm": 0.9543285247903758,
      "learning_rate": 9.999772694596475e-06,
      "loss": 0.4461,
      "step": 526
    },
    {
      "epoch": 0.03301022565339263,
      "grad_norm": 1.011477439861251,
      "learning_rate": 9.999762919203681e-06,
      "loss": 0.4603,
      "step": 527
    },
    {
      "epoch": 0.03307286365273493,
      "grad_norm": 1.1059121496488515,
      "learning_rate": 9.999752938019718e-06,
      "loss": 0.472,
      "step": 528
    },
    {
      "epoch": 0.03313550165207723,
      "grad_norm": 0.9578748311593964,
      "learning_rate": 9.999742751044993e-06,
      "loss": 0.4874,
      "step": 529
    },
    {
      "epoch": 0.03319813965141953,
      "grad_norm": 1.0079423308019362,
      "learning_rate": 9.999732358279932e-06,
      "loss": 0.4868,
      "step": 530
    },
    {
      "epoch": 0.03326077765076183,
      "grad_norm": 0.9469607821182577,
      "learning_rate": 9.999721759724958e-06,
      "loss": 0.4356,
      "step": 531
    },
    {
      "epoch": 0.03332341565010413,
      "grad_norm": 1.0053609209392569,
      "learning_rate": 9.999710955380509e-06,
      "loss": 0.4679,
      "step": 532
    },
    {
      "epoch": 0.03338605364944644,
      "grad_norm": 0.9762542388891873,
      "learning_rate": 9.999699945247028e-06,
      "loss": 0.4402,
      "step": 533
    },
    {
      "epoch": 0.03344869164878874,
      "grad_norm": 0.9384813781953285,
      "learning_rate": 9.999688729324967e-06,
      "loss": 0.4742,
      "step": 534
    },
    {
      "epoch": 0.03351132964813104,
      "grad_norm": 0.9897891615559052,
      "learning_rate": 9.999677307614793e-06,
      "loss": 0.4712,
      "step": 535
    },
    {
      "epoch": 0.03357396764747334,
      "grad_norm": 0.9885448983496022,
      "learning_rate": 9.999665680116973e-06,
      "loss": 0.4195,
      "step": 536
    },
    {
      "epoch": 0.03363660564681564,
      "grad_norm": 0.991051960999574,
      "learning_rate": 9.999653846831985e-06,
      "loss": 0.4998,
      "step": 537
    },
    {
      "epoch": 0.03369924364615794,
      "grad_norm": 1.0361053951553014,
      "learning_rate": 9.999641807760316e-06,
      "loss": 0.4431,
      "step": 538
    },
    {
      "epoch": 0.033761881645500244,
      "grad_norm": 1.0285461505097981,
      "learning_rate": 9.999629562902463e-06,
      "loss": 0.4672,
      "step": 539
    },
    {
      "epoch": 0.033824519644842545,
      "grad_norm": 0.965644470727518,
      "learning_rate": 9.99961711225893e-06,
      "loss": 0.4202,
      "step": 540
    },
    {
      "epoch": 0.033887157644184845,
      "grad_norm": 0.9560817790554014,
      "learning_rate": 9.999604455830228e-06,
      "loss": 0.4671,
      "step": 541
    },
    {
      "epoch": 0.033949795643527146,
      "grad_norm": 1.1812162057199627,
      "learning_rate": 9.999591593616878e-06,
      "loss": 0.4952,
      "step": 542
    },
    {
      "epoch": 0.03401243364286945,
      "grad_norm": 1.0043295465511102,
      "learning_rate": 9.99957852561941e-06,
      "loss": 0.4761,
      "step": 543
    },
    {
      "epoch": 0.03407507164221175,
      "grad_norm": 1.0737894947783835,
      "learning_rate": 9.999565251838361e-06,
      "loss": 0.5041,
      "step": 544
    },
    {
      "epoch": 0.03413770964155405,
      "grad_norm": 0.8868052825081272,
      "learning_rate": 9.999551772274281e-06,
      "loss": 0.4753,
      "step": 545
    },
    {
      "epoch": 0.03420034764089635,
      "grad_norm": 1.0044086592696277,
      "learning_rate": 9.99953808692772e-06,
      "loss": 0.4926,
      "step": 546
    },
    {
      "epoch": 0.03426298564023865,
      "grad_norm": 0.9703547332596356,
      "learning_rate": 9.999524195799244e-06,
      "loss": 0.4543,
      "step": 547
    },
    {
      "epoch": 0.03432562363958095,
      "grad_norm": 0.9492467948678193,
      "learning_rate": 9.999510098889425e-06,
      "loss": 0.4406,
      "step": 548
    },
    {
      "epoch": 0.03438826163892325,
      "grad_norm": 0.9101995244343161,
      "learning_rate": 9.999495796198842e-06,
      "loss": 0.4883,
      "step": 549
    },
    {
      "epoch": 0.03445089963826555,
      "grad_norm": 0.9384182648318252,
      "learning_rate": 9.999481287728083e-06,
      "loss": 0.4536,
      "step": 550
    },
    {
      "epoch": 0.03451353763760785,
      "grad_norm": 1.0059085044870621,
      "learning_rate": 9.999466573477748e-06,
      "loss": 0.4657,
      "step": 551
    },
    {
      "epoch": 0.03457617563695015,
      "grad_norm": 1.0104732065656994,
      "learning_rate": 9.999451653448438e-06,
      "loss": 0.457,
      "step": 552
    },
    {
      "epoch": 0.034638813636292454,
      "grad_norm": 0.9040445433360217,
      "learning_rate": 9.999436527640772e-06,
      "loss": 0.3829,
      "step": 553
    },
    {
      "epoch": 0.034701451635634754,
      "grad_norm": 0.9930828924993765,
      "learning_rate": 9.99942119605537e-06,
      "loss": 0.4404,
      "step": 554
    },
    {
      "epoch": 0.03476408963497706,
      "grad_norm": 0.9292860631002692,
      "learning_rate": 9.999405658692864e-06,
      "loss": 0.4676,
      "step": 555
    },
    {
      "epoch": 0.03482672763431936,
      "grad_norm": 0.964658134385176,
      "learning_rate": 9.999389915553894e-06,
      "loss": 0.4461,
      "step": 556
    },
    {
      "epoch": 0.03488936563366166,
      "grad_norm": 0.9930273170704527,
      "learning_rate": 9.999373966639106e-06,
      "loss": 0.4422,
      "step": 557
    },
    {
      "epoch": 0.034952003633003964,
      "grad_norm": 0.914622629657591,
      "learning_rate": 9.999357811949158e-06,
      "loss": 0.4806,
      "step": 558
    },
    {
      "epoch": 0.035014641632346265,
      "grad_norm": 1.0415850723617697,
      "learning_rate": 9.999341451484714e-06,
      "loss": 0.4781,
      "step": 559
    },
    {
      "epoch": 0.035077279631688565,
      "grad_norm": 0.9259754796628213,
      "learning_rate": 9.999324885246449e-06,
      "loss": 0.462,
      "step": 560
    },
    {
      "epoch": 0.035139917631030866,
      "grad_norm": 0.905892572369413,
      "learning_rate": 9.999308113235043e-06,
      "loss": 0.4208,
      "step": 561
    },
    {
      "epoch": 0.03520255563037317,
      "grad_norm": 1.0308834747625186,
      "learning_rate": 9.999291135451188e-06,
      "loss": 0.4785,
      "step": 562
    },
    {
      "epoch": 0.03526519362971547,
      "grad_norm": 1.0378162844826708,
      "learning_rate": 9.99927395189558e-06,
      "loss": 0.4321,
      "step": 563
    },
    {
      "epoch": 0.03532783162905777,
      "grad_norm": 1.019531433765153,
      "learning_rate": 9.999256562568932e-06,
      "loss": 0.4782,
      "step": 564
    },
    {
      "epoch": 0.03539046962840007,
      "grad_norm": 1.00416111298788,
      "learning_rate": 9.999238967471952e-06,
      "loss": 0.4604,
      "step": 565
    },
    {
      "epoch": 0.03545310762774237,
      "grad_norm": 0.9412528474484246,
      "learning_rate": 9.99922116660537e-06,
      "loss": 0.4733,
      "step": 566
    },
    {
      "epoch": 0.03551574562708467,
      "grad_norm": 1.015456751882925,
      "learning_rate": 9.999203159969916e-06,
      "loss": 0.476,
      "step": 567
    },
    {
      "epoch": 0.03557838362642697,
      "grad_norm": 0.9795878241595918,
      "learning_rate": 9.999184947566333e-06,
      "loss": 0.4406,
      "step": 568
    },
    {
      "epoch": 0.03564102162576927,
      "grad_norm": 1.0006474992202883,
      "learning_rate": 9.99916652939537e-06,
      "loss": 0.4526,
      "step": 569
    },
    {
      "epoch": 0.03570365962511157,
      "grad_norm": 0.9384636498456076,
      "learning_rate": 9.999147905457782e-06,
      "loss": 0.438,
      "step": 570
    },
    {
      "epoch": 0.03576629762445387,
      "grad_norm": 0.8966149020553249,
      "learning_rate": 9.99912907575434e-06,
      "loss": 0.4327,
      "step": 571
    },
    {
      "epoch": 0.035828935623796174,
      "grad_norm": 1.0951653468182427,
      "learning_rate": 9.99911004028582e-06,
      "loss": 0.49,
      "step": 572
    },
    {
      "epoch": 0.035891573623138474,
      "grad_norm": 0.902266576135075,
      "learning_rate": 9.999090799053e-06,
      "loss": 0.4528,
      "step": 573
    },
    {
      "epoch": 0.035954211622480775,
      "grad_norm": 1.0421962777080522,
      "learning_rate": 9.999071352056676e-06,
      "loss": 0.4651,
      "step": 574
    },
    {
      "epoch": 0.036016849621823076,
      "grad_norm": 0.9597946218997044,
      "learning_rate": 9.999051699297646e-06,
      "loss": 0.4208,
      "step": 575
    },
    {
      "epoch": 0.03607948762116538,
      "grad_norm": 0.8637534029130581,
      "learning_rate": 9.999031840776721e-06,
      "loss": 0.4403,
      "step": 576
    },
    {
      "epoch": 0.036142125620507684,
      "grad_norm": 1.0033480009493443,
      "learning_rate": 9.999011776494719e-06,
      "loss": 0.4952,
      "step": 577
    },
    {
      "epoch": 0.036204763619849985,
      "grad_norm": 1.090690976714218,
      "learning_rate": 9.998991506452462e-06,
      "loss": 0.4763,
      "step": 578
    },
    {
      "epoch": 0.036267401619192285,
      "grad_norm": 1.0459510154470744,
      "learning_rate": 9.998971030650788e-06,
      "loss": 0.4714,
      "step": 579
    },
    {
      "epoch": 0.036330039618534586,
      "grad_norm": 1.1860738438237561,
      "learning_rate": 9.998950349090536e-06,
      "loss": 0.4746,
      "step": 580
    },
    {
      "epoch": 0.03639267761787689,
      "grad_norm": 0.9451826114718472,
      "learning_rate": 9.998929461772564e-06,
      "loss": 0.431,
      "step": 581
    },
    {
      "epoch": 0.03645531561721919,
      "grad_norm": 1.0399613778743524,
      "learning_rate": 9.998908368697725e-06,
      "loss": 0.4568,
      "step": 582
    },
    {
      "epoch": 0.03651795361656149,
      "grad_norm": 0.9938401747308981,
      "learning_rate": 9.998887069866889e-06,
      "loss": 0.4957,
      "step": 583
    },
    {
      "epoch": 0.03658059161590379,
      "grad_norm": 0.9713714096755157,
      "learning_rate": 9.998865565280935e-06,
      "loss": 0.4548,
      "step": 584
    },
    {
      "epoch": 0.03664322961524609,
      "grad_norm": 0.8986037559771676,
      "learning_rate": 9.998843854940746e-06,
      "loss": 0.438,
      "step": 585
    },
    {
      "epoch": 0.03670586761458839,
      "grad_norm": 1.0918995669339344,
      "learning_rate": 9.998821938847216e-06,
      "loss": 0.4701,
      "step": 586
    },
    {
      "epoch": 0.03676850561393069,
      "grad_norm": 1.0070802225957607,
      "learning_rate": 9.998799817001247e-06,
      "loss": 0.4809,
      "step": 587
    },
    {
      "epoch": 0.03683114361327299,
      "grad_norm": 0.9910427723137986,
      "learning_rate": 9.99877748940375e-06,
      "loss": 0.4601,
      "step": 588
    },
    {
      "epoch": 0.03689378161261529,
      "grad_norm": 1.0890501289461592,
      "learning_rate": 9.998754956055645e-06,
      "loss": 0.5149,
      "step": 589
    },
    {
      "epoch": 0.03695641961195759,
      "grad_norm": 0.9331195552361714,
      "learning_rate": 9.998732216957856e-06,
      "loss": 0.4744,
      "step": 590
    },
    {
      "epoch": 0.037019057611299894,
      "grad_norm": 0.973348073584033,
      "learning_rate": 9.998709272111322e-06,
      "loss": 0.476,
      "step": 591
    },
    {
      "epoch": 0.037081695610642194,
      "grad_norm": 1.0317916428820801,
      "learning_rate": 9.998686121516988e-06,
      "loss": 0.5082,
      "step": 592
    },
    {
      "epoch": 0.037144333609984495,
      "grad_norm": 1.0142032508682037,
      "learning_rate": 9.998662765175803e-06,
      "loss": 0.4474,
      "step": 593
    },
    {
      "epoch": 0.037206971609326796,
      "grad_norm": 0.9769792523750915,
      "learning_rate": 9.998639203088732e-06,
      "loss": 0.4842,
      "step": 594
    },
    {
      "epoch": 0.037269609608669096,
      "grad_norm": 0.9487157168800099,
      "learning_rate": 9.998615435256744e-06,
      "loss": 0.4103,
      "step": 595
    },
    {
      "epoch": 0.0373322476080114,
      "grad_norm": 0.8861090868687992,
      "learning_rate": 9.998591461680815e-06,
      "loss": 0.4585,
      "step": 596
    },
    {
      "epoch": 0.0373948856073537,
      "grad_norm": 0.9031499065684753,
      "learning_rate": 9.998567282361935e-06,
      "loss": 0.4237,
      "step": 597
    },
    {
      "epoch": 0.037457523606696005,
      "grad_norm": 0.9088413311422079,
      "learning_rate": 9.998542897301097e-06,
      "loss": 0.4415,
      "step": 598
    },
    {
      "epoch": 0.037520161606038306,
      "grad_norm": 0.9777998365073011,
      "learning_rate": 9.998518306499307e-06,
      "loss": 0.5075,
      "step": 599
    },
    {
      "epoch": 0.03758279960538061,
      "grad_norm": 0.9600785925598556,
      "learning_rate": 9.998493509957577e-06,
      "loss": 0.4243,
      "step": 600
    },
    {
      "epoch": 0.03764543760472291,
      "grad_norm": 1.112337269143292,
      "learning_rate": 9.998468507676924e-06,
      "loss": 0.5082,
      "step": 601
    },
    {
      "epoch": 0.03770807560406521,
      "grad_norm": 0.9255230529345939,
      "learning_rate": 9.99844329965838e-06,
      "loss": 0.432,
      "step": 602
    },
    {
      "epoch": 0.03777071360340751,
      "grad_norm": 0.9657390450813839,
      "learning_rate": 9.998417885902983e-06,
      "loss": 0.503,
      "step": 603
    },
    {
      "epoch": 0.03783335160274981,
      "grad_norm": 0.9914077051594874,
      "learning_rate": 9.998392266411776e-06,
      "loss": 0.5022,
      "step": 604
    },
    {
      "epoch": 0.03789598960209211,
      "grad_norm": 0.9615120100875046,
      "learning_rate": 9.998366441185816e-06,
      "loss": 0.4413,
      "step": 605
    },
    {
      "epoch": 0.03795862760143441,
      "grad_norm": 0.9907185647151263,
      "learning_rate": 9.998340410226166e-06,
      "loss": 0.4439,
      "step": 606
    },
    {
      "epoch": 0.03802126560077671,
      "grad_norm": 1.0645631392736339,
      "learning_rate": 9.998314173533899e-06,
      "loss": 0.4211,
      "step": 607
    },
    {
      "epoch": 0.03808390360011901,
      "grad_norm": 1.1165331680035415,
      "learning_rate": 9.99828773111009e-06,
      "loss": 0.4752,
      "step": 608
    },
    {
      "epoch": 0.03814654159946131,
      "grad_norm": 0.9976949663430053,
      "learning_rate": 9.99826108295583e-06,
      "loss": 0.4809,
      "step": 609
    },
    {
      "epoch": 0.038209179598803614,
      "grad_norm": 0.9060148437491219,
      "learning_rate": 9.998234229072219e-06,
      "loss": 0.4385,
      "step": 610
    },
    {
      "epoch": 0.038271817598145914,
      "grad_norm": 0.9165908417039417,
      "learning_rate": 9.998207169460358e-06,
      "loss": 0.4656,
      "step": 611
    },
    {
      "epoch": 0.038334455597488215,
      "grad_norm": 1.0031112514335099,
      "learning_rate": 9.998179904121361e-06,
      "loss": 0.4856,
      "step": 612
    },
    {
      "epoch": 0.038397093596830516,
      "grad_norm": 1.1455165579127387,
      "learning_rate": 9.998152433056354e-06,
      "loss": 0.4747,
      "step": 613
    },
    {
      "epoch": 0.038459731596172816,
      "grad_norm": 1.0934454697254627,
      "learning_rate": 9.998124756266463e-06,
      "loss": 0.468,
      "step": 614
    },
    {
      "epoch": 0.03852236959551512,
      "grad_norm": 1.0091767455467406,
      "learning_rate": 9.99809687375283e-06,
      "loss": 0.4955,
      "step": 615
    },
    {
      "epoch": 0.03858500759485742,
      "grad_norm": 1.1526015635320352,
      "learning_rate": 9.998068785516601e-06,
      "loss": 0.5141,
      "step": 616
    },
    {
      "epoch": 0.03864764559419972,
      "grad_norm": 0.9864017088146733,
      "learning_rate": 9.998040491558936e-06,
      "loss": 0.4895,
      "step": 617
    },
    {
      "epoch": 0.03871028359354202,
      "grad_norm": 1.0044950574183737,
      "learning_rate": 9.998011991880994e-06,
      "loss": 0.4432,
      "step": 618
    },
    {
      "epoch": 0.03877292159288432,
      "grad_norm": 0.9998424825193412,
      "learning_rate": 9.997983286483953e-06,
      "loss": 0.4293,
      "step": 619
    },
    {
      "epoch": 0.03883555959222663,
      "grad_norm": 1.034036172985092,
      "learning_rate": 9.997954375368989e-06,
      "loss": 0.4835,
      "step": 620
    },
    {
      "epoch": 0.03889819759156893,
      "grad_norm": 1.0515823173171122,
      "learning_rate": 9.997925258537298e-06,
      "loss": 0.5218,
      "step": 621
    },
    {
      "epoch": 0.03896083559091123,
      "grad_norm": 1.0685245540989072,
      "learning_rate": 9.997895935990075e-06,
      "loss": 0.4448,
      "step": 622
    },
    {
      "epoch": 0.03902347359025353,
      "grad_norm": 0.9460773359178397,
      "learning_rate": 9.997866407728527e-06,
      "loss": 0.4359,
      "step": 623
    },
    {
      "epoch": 0.03908611158959583,
      "grad_norm": 0.9143083826136245,
      "learning_rate": 9.997836673753868e-06,
      "loss": 0.456,
      "step": 624
    },
    {
      "epoch": 0.03914874958893813,
      "grad_norm": 0.9963318489173749,
      "learning_rate": 9.997806734067328e-06,
      "loss": 0.5058,
      "step": 625
    },
    {
      "epoch": 0.03921138758828043,
      "grad_norm": 1.0096796677534479,
      "learning_rate": 9.997776588670133e-06,
      "loss": 0.4329,
      "step": 626
    },
    {
      "epoch": 0.03927402558762273,
      "grad_norm": 1.0502358728595,
      "learning_rate": 9.997746237563526e-06,
      "loss": 0.4813,
      "step": 627
    },
    {
      "epoch": 0.03933666358696503,
      "grad_norm": 1.5333964417376655,
      "learning_rate": 9.997715680748757e-06,
      "loss": 0.5102,
      "step": 628
    },
    {
      "epoch": 0.039399301586307334,
      "grad_norm": 1.114156203998625,
      "learning_rate": 9.997684918227083e-06,
      "loss": 0.4615,
      "step": 629
    },
    {
      "epoch": 0.039461939585649634,
      "grad_norm": 1.1449257459646056,
      "learning_rate": 9.997653949999769e-06,
      "loss": 0.4957,
      "step": 630
    },
    {
      "epoch": 0.039524577584991935,
      "grad_norm": 1.0123482057200637,
      "learning_rate": 9.997622776068092e-06,
      "loss": 0.4804,
      "step": 631
    },
    {
      "epoch": 0.039587215584334236,
      "grad_norm": 1.3042639852064444,
      "learning_rate": 9.997591396433333e-06,
      "loss": 0.5095,
      "step": 632
    },
    {
      "epoch": 0.039649853583676536,
      "grad_norm": 1.0845242278561564,
      "learning_rate": 9.997559811096785e-06,
      "loss": 0.4896,
      "step": 633
    },
    {
      "epoch": 0.03971249158301884,
      "grad_norm": 0.9102289622479706,
      "learning_rate": 9.997528020059748e-06,
      "loss": 0.4297,
      "step": 634
    },
    {
      "epoch": 0.03977512958236114,
      "grad_norm": 0.959861066428182,
      "learning_rate": 9.99749602332353e-06,
      "loss": 0.4941,
      "step": 635
    },
    {
      "epoch": 0.03983776758170344,
      "grad_norm": 0.9190787546617656,
      "learning_rate": 9.997463820889447e-06,
      "loss": 0.4645,
      "step": 636
    },
    {
      "epoch": 0.03990040558104574,
      "grad_norm": 0.9388601258662337,
      "learning_rate": 9.997431412758827e-06,
      "loss": 0.4955,
      "step": 637
    },
    {
      "epoch": 0.03996304358038804,
      "grad_norm": 1.0037848348482277,
      "learning_rate": 9.997398798933002e-06,
      "loss": 0.4714,
      "step": 638
    },
    {
      "epoch": 0.04002568157973034,
      "grad_norm": 0.9213487559779558,
      "learning_rate": 9.997365979413314e-06,
      "loss": 0.4566,
      "step": 639
    },
    {
      "epoch": 0.04008831957907264,
      "grad_norm": 1.4342791385857256,
      "learning_rate": 9.997332954201115e-06,
      "loss": 0.5175,
      "step": 640
    },
    {
      "epoch": 0.04015095757841495,
      "grad_norm": 0.9184833642187181,
      "learning_rate": 9.997299723297765e-06,
      "loss": 0.4374,
      "step": 641
    },
    {
      "epoch": 0.04021359557775725,
      "grad_norm": 0.9976762961925686,
      "learning_rate": 9.99726628670463e-06,
      "loss": 0.4289,
      "step": 642
    },
    {
      "epoch": 0.04027623357709955,
      "grad_norm": 1.068931329314942,
      "learning_rate": 9.997232644423089e-06,
      "loss": 0.5068,
      "step": 643
    },
    {
      "epoch": 0.04033887157644185,
      "grad_norm": 0.9900129083905248,
      "learning_rate": 9.997198796454522e-06,
      "loss": 0.442,
      "step": 644
    },
    {
      "epoch": 0.04040150957578415,
      "grad_norm": 0.8981927838528493,
      "learning_rate": 9.997164742800327e-06,
      "loss": 0.4626,
      "step": 645
    },
    {
      "epoch": 0.04046414757512645,
      "grad_norm": 0.8916856873102379,
      "learning_rate": 9.997130483461903e-06,
      "loss": 0.4588,
      "step": 646
    },
    {
      "epoch": 0.04052678557446875,
      "grad_norm": 1.0013684128942837,
      "learning_rate": 9.997096018440661e-06,
      "loss": 0.4628,
      "step": 647
    },
    {
      "epoch": 0.040589423573811054,
      "grad_norm": 0.9528620086703701,
      "learning_rate": 9.997061347738019e-06,
      "loss": 0.4967,
      "step": 648
    },
    {
      "epoch": 0.040652061573153354,
      "grad_norm": 0.9902420493385229,
      "learning_rate": 9.997026471355404e-06,
      "loss": 0.4409,
      "step": 649
    },
    {
      "epoch": 0.040714699572495655,
      "grad_norm": 0.8399190444522047,
      "learning_rate": 9.996991389294253e-06,
      "loss": 0.423,
      "step": 650
    },
    {
      "epoch": 0.040777337571837956,
      "grad_norm": 0.9908145594898994,
      "learning_rate": 9.996956101556009e-06,
      "loss": 0.4565,
      "step": 651
    },
    {
      "epoch": 0.040839975571180256,
      "grad_norm": 0.9793276007170698,
      "learning_rate": 9.996920608142123e-06,
      "loss": 0.4278,
      "step": 652
    },
    {
      "epoch": 0.04090261357052256,
      "grad_norm": 0.9463422669458154,
      "learning_rate": 9.996884909054059e-06,
      "loss": 0.4493,
      "step": 653
    },
    {
      "epoch": 0.04096525156986486,
      "grad_norm": 0.9818598444410425,
      "learning_rate": 9.996849004293283e-06,
      "loss": 0.473,
      "step": 654
    },
    {
      "epoch": 0.04102788956920716,
      "grad_norm": 1.0508552153246706,
      "learning_rate": 9.996812893861276e-06,
      "loss": 0.5317,
      "step": 655
    },
    {
      "epoch": 0.04109052756854946,
      "grad_norm": 0.9274853335495093,
      "learning_rate": 9.996776577759522e-06,
      "loss": 0.429,
      "step": 656
    },
    {
      "epoch": 0.04115316556789176,
      "grad_norm": 0.8521552724711785,
      "learning_rate": 9.996740055989516e-06,
      "loss": 0.434,
      "step": 657
    },
    {
      "epoch": 0.04121580356723406,
      "grad_norm": 0.9248493874067197,
      "learning_rate": 9.996703328552764e-06,
      "loss": 0.4437,
      "step": 658
    },
    {
      "epoch": 0.04127844156657636,
      "grad_norm": 0.912593040838753,
      "learning_rate": 9.996666395450773e-06,
      "loss": 0.4893,
      "step": 659
    },
    {
      "epoch": 0.04134107956591866,
      "grad_norm": 0.9605858141682788,
      "learning_rate": 9.996629256685067e-06,
      "loss": 0.4195,
      "step": 660
    },
    {
      "epoch": 0.04140371756526096,
      "grad_norm": 1.2225560143239458,
      "learning_rate": 9.996591912257175e-06,
      "loss": 0.4545,
      "step": 661
    },
    {
      "epoch": 0.04146635556460326,
      "grad_norm": 0.8644472926459017,
      "learning_rate": 9.99655436216863e-06,
      "loss": 0.4011,
      "step": 662
    },
    {
      "epoch": 0.04152899356394557,
      "grad_norm": 1.1243880529160646,
      "learning_rate": 9.99651660642098e-06,
      "loss": 0.4948,
      "step": 663
    },
    {
      "epoch": 0.04159163156328787,
      "grad_norm": 0.9128976792932832,
      "learning_rate": 9.996478645015779e-06,
      "loss": 0.4293,
      "step": 664
    },
    {
      "epoch": 0.04165426956263017,
      "grad_norm": 1.0085742032382932,
      "learning_rate": 9.996440477954592e-06,
      "loss": 0.49,
      "step": 665
    },
    {
      "epoch": 0.04171690756197247,
      "grad_norm": 0.9138971856367806,
      "learning_rate": 9.996402105238985e-06,
      "loss": 0.4422,
      "step": 666
    },
    {
      "epoch": 0.041779545561314774,
      "grad_norm": 1.0633114091813225,
      "learning_rate": 9.996363526870541e-06,
      "loss": 0.4967,
      "step": 667
    },
    {
      "epoch": 0.041842183560657074,
      "grad_norm": 0.9610155186509937,
      "learning_rate": 9.996324742850848e-06,
      "loss": 0.4577,
      "step": 668
    },
    {
      "epoch": 0.041904821559999375,
      "grad_norm": 0.9824864826920999,
      "learning_rate": 9.996285753181499e-06,
      "loss": 0.4655,
      "step": 669
    },
    {
      "epoch": 0.041967459559341676,
      "grad_norm": 0.934706462278132,
      "learning_rate": 9.996246557864102e-06,
      "loss": 0.4839,
      "step": 670
    },
    {
      "epoch": 0.042030097558683976,
      "grad_norm": 0.9835437236070101,
      "learning_rate": 9.99620715690027e-06,
      "loss": 0.4699,
      "step": 671
    },
    {
      "epoch": 0.04209273555802628,
      "grad_norm": 0.9486628364755014,
      "learning_rate": 9.996167550291624e-06,
      "loss": 0.4678,
      "step": 672
    },
    {
      "epoch": 0.04215537355736858,
      "grad_norm": 1.0128997596890357,
      "learning_rate": 9.996127738039793e-06,
      "loss": 0.4497,
      "step": 673
    },
    {
      "epoch": 0.04221801155671088,
      "grad_norm": 1.049530930318942,
      "learning_rate": 9.99608772014642e-06,
      "loss": 0.4507,
      "step": 674
    },
    {
      "epoch": 0.04228064955605318,
      "grad_norm": 1.06903445550272,
      "learning_rate": 9.996047496613146e-06,
      "loss": 0.4874,
      "step": 675
    },
    {
      "epoch": 0.04234328755539548,
      "grad_norm": 0.9642058510071302,
      "learning_rate": 9.99600706744163e-06,
      "loss": 0.4969,
      "step": 676
    },
    {
      "epoch": 0.04240592555473778,
      "grad_norm": 1.0214254623919068,
      "learning_rate": 9.995966432633537e-06,
      "loss": 0.506,
      "step": 677
    },
    {
      "epoch": 0.04246856355408008,
      "grad_norm": 0.9421611011238127,
      "learning_rate": 9.995925592190537e-06,
      "loss": 0.4824,
      "step": 678
    },
    {
      "epoch": 0.04253120155342238,
      "grad_norm": 0.925849155242043,
      "learning_rate": 9.995884546114315e-06,
      "loss": 0.4865,
      "step": 679
    },
    {
      "epoch": 0.04259383955276468,
      "grad_norm": 1.0146283781562906,
      "learning_rate": 9.995843294406554e-06,
      "loss": 0.4322,
      "step": 680
    },
    {
      "epoch": 0.04265647755210698,
      "grad_norm": 1.0042717374500967,
      "learning_rate": 9.995801837068958e-06,
      "loss": 0.4626,
      "step": 681
    },
    {
      "epoch": 0.042719115551449284,
      "grad_norm": 1.0130780964908586,
      "learning_rate": 9.99576017410323e-06,
      "loss": 0.4568,
      "step": 682
    },
    {
      "epoch": 0.042781753550791585,
      "grad_norm": 0.9802978347810047,
      "learning_rate": 9.995718305511088e-06,
      "loss": 0.4683,
      "step": 683
    },
    {
      "epoch": 0.042844391550133885,
      "grad_norm": 0.9630203858770743,
      "learning_rate": 9.995676231294251e-06,
      "loss": 0.488,
      "step": 684
    },
    {
      "epoch": 0.04290702954947619,
      "grad_norm": 1.035245011056049,
      "learning_rate": 9.995633951454456e-06,
      "loss": 0.4757,
      "step": 685
    },
    {
      "epoch": 0.042969667548818494,
      "grad_norm": 0.9806765192994227,
      "learning_rate": 9.995591465993438e-06,
      "loss": 0.4137,
      "step": 686
    },
    {
      "epoch": 0.043032305548160794,
      "grad_norm": 1.0582608361824482,
      "learning_rate": 9.99554877491295e-06,
      "loss": 0.4664,
      "step": 687
    },
    {
      "epoch": 0.043094943547503095,
      "grad_norm": 0.950494578421443,
      "learning_rate": 9.995505878214744e-06,
      "loss": 0.487,
      "step": 688
    },
    {
      "epoch": 0.043157581546845396,
      "grad_norm": 1.0051273453706044,
      "learning_rate": 9.995462775900591e-06,
      "loss": 0.4444,
      "step": 689
    },
    {
      "epoch": 0.043220219546187696,
      "grad_norm": 0.9189436909288832,
      "learning_rate": 9.995419467972262e-06,
      "loss": 0.4604,
      "step": 690
    },
    {
      "epoch": 0.04328285754553,
      "grad_norm": 0.9722365439422429,
      "learning_rate": 9.99537595443154e-06,
      "loss": 0.4444,
      "step": 691
    },
    {
      "epoch": 0.0433454955448723,
      "grad_norm": 1.0125307967345447,
      "learning_rate": 9.995332235280218e-06,
      "loss": 0.4161,
      "step": 692
    },
    {
      "epoch": 0.0434081335442146,
      "grad_norm": 0.7909757593408102,
      "learning_rate": 9.995288310520094e-06,
      "loss": 0.4613,
      "step": 693
    },
    {
      "epoch": 0.0434707715435569,
      "grad_norm": 1.003720395867555,
      "learning_rate": 9.995244180152975e-06,
      "loss": 0.4664,
      "step": 694
    },
    {
      "epoch": 0.0435334095428992,
      "grad_norm": 0.9804291823667496,
      "learning_rate": 9.99519984418068e-06,
      "loss": 0.4237,
      "step": 695
    },
    {
      "epoch": 0.0435960475422415,
      "grad_norm": 1.0764437751444076,
      "learning_rate": 9.99515530260503e-06,
      "loss": 0.4445,
      "step": 696
    },
    {
      "epoch": 0.0436586855415838,
      "grad_norm": 0.9552801235216631,
      "learning_rate": 9.995110555427863e-06,
      "loss": 0.415,
      "step": 697
    },
    {
      "epoch": 0.0437213235409261,
      "grad_norm": 0.8476546466671363,
      "learning_rate": 9.995065602651017e-06,
      "loss": 0.4556,
      "step": 698
    },
    {
      "epoch": 0.0437839615402684,
      "grad_norm": 0.9968752408477135,
      "learning_rate": 9.995020444276345e-06,
      "loss": 0.5059,
      "step": 699
    },
    {
      "epoch": 0.0438465995396107,
      "grad_norm": 0.9772301507710963,
      "learning_rate": 9.994975080305703e-06,
      "loss": 0.4134,
      "step": 700
    },
    {
      "epoch": 0.043909237538953004,
      "grad_norm": 1.013029721280459,
      "learning_rate": 9.994929510740958e-06,
      "loss": 0.5126,
      "step": 701
    },
    {
      "epoch": 0.043971875538295305,
      "grad_norm": 0.9894687006271955,
      "learning_rate": 9.99488373558399e-06,
      "loss": 0.5247,
      "step": 702
    },
    {
      "epoch": 0.044034513537637605,
      "grad_norm": 1.0058284374838626,
      "learning_rate": 9.99483775483668e-06,
      "loss": 0.4572,
      "step": 703
    },
    {
      "epoch": 0.044097151536979906,
      "grad_norm": 0.9553474741911481,
      "learning_rate": 9.99479156850092e-06,
      "loss": 0.4717,
      "step": 704
    },
    {
      "epoch": 0.04415978953632221,
      "grad_norm": 0.9099344162592342,
      "learning_rate": 9.994745176578611e-06,
      "loss": 0.4335,
      "step": 705
    },
    {
      "epoch": 0.04422242753566451,
      "grad_norm": 0.8999173419831605,
      "learning_rate": 9.994698579071666e-06,
      "loss": 0.4386,
      "step": 706
    },
    {
      "epoch": 0.044285065535006815,
      "grad_norm": 0.8937404180947458,
      "learning_rate": 9.994651775981998e-06,
      "loss": 0.4454,
      "step": 707
    },
    {
      "epoch": 0.044347703534349116,
      "grad_norm": 0.8901454169732728,
      "learning_rate": 9.994604767311538e-06,
      "loss": 0.4259,
      "step": 708
    },
    {
      "epoch": 0.044410341533691416,
      "grad_norm": 1.0008631391069158,
      "learning_rate": 9.994557553062218e-06,
      "loss": 0.4313,
      "step": 709
    },
    {
      "epoch": 0.04447297953303372,
      "grad_norm": 0.8902744588342272,
      "learning_rate": 9.994510133235982e-06,
      "loss": 0.4418,
      "step": 710
    },
    {
      "epoch": 0.04453561753237602,
      "grad_norm": 0.956393726859102,
      "learning_rate": 9.994462507834782e-06,
      "loss": 0.4208,
      "step": 711
    },
    {
      "epoch": 0.04459825553171832,
      "grad_norm": 0.9444992577474345,
      "learning_rate": 9.994414676860578e-06,
      "loss": 0.4711,
      "step": 712
    },
    {
      "epoch": 0.04466089353106062,
      "grad_norm": 0.9681633570757133,
      "learning_rate": 9.994366640315338e-06,
      "loss": 0.478,
      "step": 713
    },
    {
      "epoch": 0.04472353153040292,
      "grad_norm": 0.901541455168364,
      "learning_rate": 9.994318398201042e-06,
      "loss": 0.4011,
      "step": 714
    },
    {
      "epoch": 0.04478616952974522,
      "grad_norm": 0.9857170835026026,
      "learning_rate": 9.994269950519672e-06,
      "loss": 0.4395,
      "step": 715
    },
    {
      "epoch": 0.04484880752908752,
      "grad_norm": 0.9652941667425703,
      "learning_rate": 9.994221297273224e-06,
      "loss": 0.5097,
      "step": 716
    },
    {
      "epoch": 0.04491144552842982,
      "grad_norm": 0.9798649933997702,
      "learning_rate": 9.9941724384637e-06,
      "loss": 0.4546,
      "step": 717
    },
    {
      "epoch": 0.04497408352777212,
      "grad_norm": 0.9877602517758866,
      "learning_rate": 9.994123374093114e-06,
      "loss": 0.4366,
      "step": 718
    },
    {
      "epoch": 0.04503672152711442,
      "grad_norm": 0.9090927155754608,
      "learning_rate": 9.99407410416348e-06,
      "loss": 0.4284,
      "step": 719
    },
    {
      "epoch": 0.045099359526456724,
      "grad_norm": 0.8169542326975429,
      "learning_rate": 9.99402462867683e-06,
      "loss": 0.4561,
      "step": 720
    },
    {
      "epoch": 0.045161997525799025,
      "grad_norm": 0.9611329299292285,
      "learning_rate": 9.993974947635201e-06,
      "loss": 0.4674,
      "step": 721
    },
    {
      "epoch": 0.045224635525141325,
      "grad_norm": 0.934907586188618,
      "learning_rate": 9.993925061040634e-06,
      "loss": 0.5143,
      "step": 722
    },
    {
      "epoch": 0.045287273524483626,
      "grad_norm": 0.9222254506898022,
      "learning_rate": 9.993874968895186e-06,
      "loss": 0.4294,
      "step": 723
    },
    {
      "epoch": 0.04534991152382593,
      "grad_norm": 0.9256901510340121,
      "learning_rate": 9.993824671200916e-06,
      "loss": 0.4591,
      "step": 724
    },
    {
      "epoch": 0.04541254952316823,
      "grad_norm": 0.9583011173465659,
      "learning_rate": 9.993774167959897e-06,
      "loss": 0.4778,
      "step": 725
    },
    {
      "epoch": 0.04547518752251053,
      "grad_norm": 1.044269474115197,
      "learning_rate": 9.993723459174206e-06,
      "loss": 0.4518,
      "step": 726
    },
    {
      "epoch": 0.04553782552185283,
      "grad_norm": 1.0144672018012593,
      "learning_rate": 9.993672544845931e-06,
      "loss": 0.491,
      "step": 727
    },
    {
      "epoch": 0.045600463521195136,
      "grad_norm": 0.9517287607116042,
      "learning_rate": 9.993621424977166e-06,
      "loss": 0.4622,
      "step": 728
    },
    {
      "epoch": 0.04566310152053744,
      "grad_norm": 0.8889620316072456,
      "learning_rate": 9.993570099570017e-06,
      "loss": 0.4592,
      "step": 729
    },
    {
      "epoch": 0.04572573951987974,
      "grad_norm": 0.9687359871375144,
      "learning_rate": 9.993518568626595e-06,
      "loss": 0.4899,
      "step": 730
    },
    {
      "epoch": 0.04578837751922204,
      "grad_norm": 0.9592521453505902,
      "learning_rate": 9.993466832149024e-06,
      "loss": 0.4645,
      "step": 731
    },
    {
      "epoch": 0.04585101551856434,
      "grad_norm": 1.0372334623343942,
      "learning_rate": 9.99341489013943e-06,
      "loss": 0.4612,
      "step": 732
    },
    {
      "epoch": 0.04591365351790664,
      "grad_norm": 0.910917121050653,
      "learning_rate": 9.99336274259995e-06,
      "loss": 0.4619,
      "step": 733
    },
    {
      "epoch": 0.04597629151724894,
      "grad_norm": 0.9595627785906364,
      "learning_rate": 9.993310389532736e-06,
      "loss": 0.4699,
      "step": 734
    },
    {
      "epoch": 0.04603892951659124,
      "grad_norm": 0.9665070308084366,
      "learning_rate": 9.993257830939938e-06,
      "loss": 0.4912,
      "step": 735
    },
    {
      "epoch": 0.04610156751593354,
      "grad_norm": 0.9931497511221541,
      "learning_rate": 9.993205066823722e-06,
      "loss": 0.4597,
      "step": 736
    },
    {
      "epoch": 0.04616420551527584,
      "grad_norm": 0.9811041344546328,
      "learning_rate": 9.993152097186258e-06,
      "loss": 0.4506,
      "step": 737
    },
    {
      "epoch": 0.04622684351461814,
      "grad_norm": 1.0118203248379027,
      "learning_rate": 9.993098922029728e-06,
      "loss": 0.4833,
      "step": 738
    },
    {
      "epoch": 0.046289481513960444,
      "grad_norm": 1.0123269803984023,
      "learning_rate": 9.993045541356318e-06,
      "loss": 0.4309,
      "step": 739
    },
    {
      "epoch": 0.046352119513302745,
      "grad_norm": 0.8153290511181605,
      "learning_rate": 9.992991955168227e-06,
      "loss": 0.4786,
      "step": 740
    },
    {
      "epoch": 0.046414757512645045,
      "grad_norm": 1.001069671521246,
      "learning_rate": 9.99293816346766e-06,
      "loss": 0.4542,
      "step": 741
    },
    {
      "epoch": 0.046477395511987346,
      "grad_norm": 1.0118245256071083,
      "learning_rate": 9.992884166256833e-06,
      "loss": 0.4529,
      "step": 742
    },
    {
      "epoch": 0.04654003351132965,
      "grad_norm": 0.9863031122078769,
      "learning_rate": 9.992829963537965e-06,
      "loss": 0.4479,
      "step": 743
    },
    {
      "epoch": 0.04660267151067195,
      "grad_norm": 0.949604468946847,
      "learning_rate": 9.992775555313291e-06,
      "loss": 0.4768,
      "step": 744
    },
    {
      "epoch": 0.04666530951001425,
      "grad_norm": 0.9515086186659695,
      "learning_rate": 9.992720941585048e-06,
      "loss": 0.4538,
      "step": 745
    },
    {
      "epoch": 0.04672794750935655,
      "grad_norm": 0.8925033977623502,
      "learning_rate": 9.992666122355483e-06,
      "loss": 0.4277,
      "step": 746
    },
    {
      "epoch": 0.04679058550869885,
      "grad_norm": 0.8449580007111595,
      "learning_rate": 9.992611097626856e-06,
      "loss": 0.4191,
      "step": 747
    },
    {
      "epoch": 0.04685322350804115,
      "grad_norm": 0.9315575193510275,
      "learning_rate": 9.992555867401428e-06,
      "loss": 0.4507,
      "step": 748
    },
    {
      "epoch": 0.04691586150738345,
      "grad_norm": 0.9715945762235036,
      "learning_rate": 9.992500431681473e-06,
      "loss": 0.4849,
      "step": 749
    },
    {
      "epoch": 0.04697849950672576,
      "grad_norm": 0.8807236630022497,
      "learning_rate": 9.992444790469276e-06,
      "loss": 0.4454,
      "step": 750
    },
    {
      "epoch": 0.04704113750606806,
      "grad_norm": 0.951856879460865,
      "learning_rate": 9.992388943767122e-06,
      "loss": 0.4617,
      "step": 751
    },
    {
      "epoch": 0.04710377550541036,
      "grad_norm": 0.859940008081188,
      "learning_rate": 9.992332891577314e-06,
      "loss": 0.4456,
      "step": 752
    },
    {
      "epoch": 0.04716641350475266,
      "grad_norm": 1.0223444075246404,
      "learning_rate": 9.992276633902158e-06,
      "loss": 0.4691,
      "step": 753
    },
    {
      "epoch": 0.04722905150409496,
      "grad_norm": 0.9318242058375885,
      "learning_rate": 9.992220170743968e-06,
      "loss": 0.5017,
      "step": 754
    },
    {
      "epoch": 0.04729168950343726,
      "grad_norm": 0.917414841019299,
      "learning_rate": 9.99216350210507e-06,
      "loss": 0.4723,
      "step": 755
    },
    {
      "epoch": 0.04735432750277956,
      "grad_norm": 0.9312686007650006,
      "learning_rate": 9.992106627987795e-06,
      "loss": 0.4175,
      "step": 756
    },
    {
      "epoch": 0.04741696550212186,
      "grad_norm": 0.9841511116003078,
      "learning_rate": 9.992049548394484e-06,
      "loss": 0.4829,
      "step": 757
    },
    {
      "epoch": 0.047479603501464164,
      "grad_norm": 0.9177329234363769,
      "learning_rate": 9.99199226332749e-06,
      "loss": 0.4303,
      "step": 758
    },
    {
      "epoch": 0.047542241500806465,
      "grad_norm": 0.9677126011431414,
      "learning_rate": 9.991934772789164e-06,
      "loss": 0.486,
      "step": 759
    },
    {
      "epoch": 0.047604879500148765,
      "grad_norm": 0.9842216279546642,
      "learning_rate": 9.99187707678188e-06,
      "loss": 0.4701,
      "step": 760
    },
    {
      "epoch": 0.047667517499491066,
      "grad_norm": 0.9663022413044057,
      "learning_rate": 9.991819175308007e-06,
      "loss": 0.4969,
      "step": 761
    },
    {
      "epoch": 0.04773015549883337,
      "grad_norm": 0.9782789835477755,
      "learning_rate": 9.99176106836993e-06,
      "loss": 0.4745,
      "step": 762
    },
    {
      "epoch": 0.04779279349817567,
      "grad_norm": 0.9719607484345565,
      "learning_rate": 9.991702755970041e-06,
      "loss": 0.4475,
      "step": 763
    },
    {
      "epoch": 0.04785543149751797,
      "grad_norm": 0.9974508358469623,
      "learning_rate": 9.991644238110741e-06,
      "loss": 0.4725,
      "step": 764
    },
    {
      "epoch": 0.04791806949686027,
      "grad_norm": 1.0211469862632896,
      "learning_rate": 9.991585514794436e-06,
      "loss": 0.4966,
      "step": 765
    },
    {
      "epoch": 0.04798070749620257,
      "grad_norm": 0.9183209412355604,
      "learning_rate": 9.991526586023546e-06,
      "loss": 0.4582,
      "step": 766
    },
    {
      "epoch": 0.04804334549554487,
      "grad_norm": 0.9453055985536284,
      "learning_rate": 9.991467451800493e-06,
      "loss": 0.4516,
      "step": 767
    },
    {
      "epoch": 0.04810598349488717,
      "grad_norm": 0.8598246741597764,
      "learning_rate": 9.991408112127717e-06,
      "loss": 0.4725,
      "step": 768
    },
    {
      "epoch": 0.04816862149422947,
      "grad_norm": 0.9239525051110906,
      "learning_rate": 9.991348567007655e-06,
      "loss": 0.4651,
      "step": 769
    },
    {
      "epoch": 0.04823125949357177,
      "grad_norm": 0.9315007100535981,
      "learning_rate": 9.99128881644276e-06,
      "loss": 0.4251,
      "step": 770
    },
    {
      "epoch": 0.04829389749291407,
      "grad_norm": 0.9110274864053051,
      "learning_rate": 9.991228860435489e-06,
      "loss": 0.437,
      "step": 771
    },
    {
      "epoch": 0.04835653549225638,
      "grad_norm": 0.9624368296403017,
      "learning_rate": 9.991168698988315e-06,
      "loss": 0.4491,
      "step": 772
    },
    {
      "epoch": 0.04841917349159868,
      "grad_norm": 0.9525438308942169,
      "learning_rate": 9.991108332103707e-06,
      "loss": 0.4364,
      "step": 773
    },
    {
      "epoch": 0.04848181149094098,
      "grad_norm": 0.9521232787190272,
      "learning_rate": 9.991047759784156e-06,
      "loss": 0.4995,
      "step": 774
    },
    {
      "epoch": 0.04854444949028328,
      "grad_norm": 0.9392138333898377,
      "learning_rate": 9.990986982032152e-06,
      "loss": 0.4826,
      "step": 775
    },
    {
      "epoch": 0.04860708748962558,
      "grad_norm": 0.9677686606535415,
      "learning_rate": 9.9909259988502e-06,
      "loss": 0.4423,
      "step": 776
    },
    {
      "epoch": 0.048669725488967884,
      "grad_norm": 0.934787773706382,
      "learning_rate": 9.990864810240803e-06,
      "loss": 0.506,
      "step": 777
    },
    {
      "epoch": 0.048732363488310185,
      "grad_norm": 0.9413665900296336,
      "learning_rate": 9.990803416206487e-06,
      "loss": 0.4535,
      "step": 778
    },
    {
      "epoch": 0.048795001487652485,
      "grad_norm": 0.952415742292826,
      "learning_rate": 9.990741816749776e-06,
      "loss": 0.4603,
      "step": 779
    },
    {
      "epoch": 0.048857639486994786,
      "grad_norm": 0.8699301034896143,
      "learning_rate": 9.990680011873204e-06,
      "loss": 0.4536,
      "step": 780
    },
    {
      "epoch": 0.04892027748633709,
      "grad_norm": 1.0321118027045681,
      "learning_rate": 9.990618001579318e-06,
      "loss": 0.5103,
      "step": 781
    },
    {
      "epoch": 0.04898291548567939,
      "grad_norm": 0.9887789670760242,
      "learning_rate": 9.990555785870667e-06,
      "loss": 0.4659,
      "step": 782
    },
    {
      "epoch": 0.04904555348502169,
      "grad_norm": 1.0482374261301903,
      "learning_rate": 9.990493364749815e-06,
      "loss": 0.5034,
      "step": 783
    },
    {
      "epoch": 0.04910819148436399,
      "grad_norm": 0.914568086350897,
      "learning_rate": 9.99043073821933e-06,
      "loss": 0.4346,
      "step": 784
    },
    {
      "epoch": 0.04917082948370629,
      "grad_norm": 0.9274842904337175,
      "learning_rate": 9.990367906281789e-06,
      "loss": 0.4173,
      "step": 785
    },
    {
      "epoch": 0.04923346748304859,
      "grad_norm": 0.9490526339521382,
      "learning_rate": 9.990304868939778e-06,
      "loss": 0.4902,
      "step": 786
    },
    {
      "epoch": 0.04929610548239089,
      "grad_norm": 0.9628472319649374,
      "learning_rate": 9.990241626195893e-06,
      "loss": 0.4308,
      "step": 787
    },
    {
      "epoch": 0.04935874348173319,
      "grad_norm": 0.9811908804895358,
      "learning_rate": 9.990178178052737e-06,
      "loss": 0.475,
      "step": 788
    },
    {
      "epoch": 0.04942138148107549,
      "grad_norm": 0.9317354366306052,
      "learning_rate": 9.990114524512919e-06,
      "loss": 0.4255,
      "step": 789
    },
    {
      "epoch": 0.04948401948041779,
      "grad_norm": 1.0361662580161788,
      "learning_rate": 9.990050665579063e-06,
      "loss": 0.4812,
      "step": 790
    },
    {
      "epoch": 0.049546657479760094,
      "grad_norm": 0.9852770433881092,
      "learning_rate": 9.989986601253796e-06,
      "loss": 0.4348,
      "step": 791
    },
    {
      "epoch": 0.049609295479102394,
      "grad_norm": 0.934261336623812,
      "learning_rate": 9.989922331539752e-06,
      "loss": 0.4523,
      "step": 792
    },
    {
      "epoch": 0.049671933478444695,
      "grad_norm": 0.911230732617379,
      "learning_rate": 9.98985785643958e-06,
      "loss": 0.4472,
      "step": 793
    },
    {
      "epoch": 0.049734571477787,
      "grad_norm": 0.9331601883361844,
      "learning_rate": 9.989793175955933e-06,
      "loss": 0.4502,
      "step": 794
    },
    {
      "epoch": 0.0497972094771293,
      "grad_norm": 0.953988623327564,
      "learning_rate": 9.989728290091471e-06,
      "loss": 0.4361,
      "step": 795
    },
    {
      "epoch": 0.049859847476471604,
      "grad_norm": 1.0521302596067241,
      "learning_rate": 9.98966319884887e-06,
      "loss": 0.4775,
      "step": 796
    },
    {
      "epoch": 0.049922485475813905,
      "grad_norm": 0.9059320786382553,
      "learning_rate": 9.989597902230801e-06,
      "loss": 0.4153,
      "step": 797
    },
    {
      "epoch": 0.049985123475156205,
      "grad_norm": 0.8649718726457603,
      "learning_rate": 9.989532400239958e-06,
      "loss": 0.4515,
      "step": 798
    },
    {
      "epoch": 0.050047761474498506,
      "grad_norm": 0.9678164424910579,
      "learning_rate": 9.989466692879036e-06,
      "loss": 0.4794,
      "step": 799
    },
    {
      "epoch": 0.05011039947384081,
      "grad_norm": 1.0188879945578437,
      "learning_rate": 9.989400780150739e-06,
      "loss": 0.4725,
      "step": 800
    },
    {
      "epoch": 0.05017303747318311,
      "grad_norm": 0.9543983584099914,
      "learning_rate": 9.98933466205778e-06,
      "loss": 0.497,
      "step": 801
    },
    {
      "epoch": 0.05023567547252541,
      "grad_norm": 0.9448764313321409,
      "learning_rate": 9.989268338602878e-06,
      "loss": 0.5037,
      "step": 802
    },
    {
      "epoch": 0.05029831347186771,
      "grad_norm": 0.9253155645686162,
      "learning_rate": 9.989201809788767e-06,
      "loss": 0.4321,
      "step": 803
    },
    {
      "epoch": 0.05036095147121001,
      "grad_norm": 0.9180777847225409,
      "learning_rate": 9.989135075618183e-06,
      "loss": 0.4593,
      "step": 804
    },
    {
      "epoch": 0.05042358947055231,
      "grad_norm": 0.9269281045104778,
      "learning_rate": 9.989068136093873e-06,
      "loss": 0.4277,
      "step": 805
    },
    {
      "epoch": 0.05048622746989461,
      "grad_norm": 0.9467219408399294,
      "learning_rate": 9.989000991218594e-06,
      "loss": 0.466,
      "step": 806
    },
    {
      "epoch": 0.05054886546923691,
      "grad_norm": 0.9998168099637439,
      "learning_rate": 9.988933640995107e-06,
      "loss": 0.4618,
      "step": 807
    },
    {
      "epoch": 0.05061150346857921,
      "grad_norm": 0.9775540978994027,
      "learning_rate": 9.988866085426185e-06,
      "loss": 0.442,
      "step": 808
    },
    {
      "epoch": 0.05067414146792151,
      "grad_norm": 0.9209643442921713,
      "learning_rate": 9.988798324514608e-06,
      "loss": 0.4618,
      "step": 809
    },
    {
      "epoch": 0.050736779467263814,
      "grad_norm": 1.0642043301623647,
      "learning_rate": 9.98873035826317e-06,
      "loss": 0.4825,
      "step": 810
    },
    {
      "epoch": 0.050799417466606114,
      "grad_norm": 0.9583716071947113,
      "learning_rate": 9.98866218667466e-06,
      "loss": 0.4331,
      "step": 811
    },
    {
      "epoch": 0.050862055465948415,
      "grad_norm": 0.8924599796374214,
      "learning_rate": 9.98859380975189e-06,
      "loss": 0.4594,
      "step": 812
    },
    {
      "epoch": 0.050924693465290716,
      "grad_norm": 0.9701799397023545,
      "learning_rate": 9.988525227497674e-06,
      "loss": 0.4786,
      "step": 813
    },
    {
      "epoch": 0.050987331464633016,
      "grad_norm": 0.9657838502930279,
      "learning_rate": 9.988456439914834e-06,
      "loss": 0.4902,
      "step": 814
    },
    {
      "epoch": 0.051049969463975324,
      "grad_norm": 1.0839392919680693,
      "learning_rate": 9.988387447006198e-06,
      "loss": 0.4925,
      "step": 815
    },
    {
      "epoch": 0.051112607463317625,
      "grad_norm": 0.9235329106765405,
      "learning_rate": 9.988318248774611e-06,
      "loss": 0.4379,
      "step": 816
    },
    {
      "epoch": 0.051175245462659925,
      "grad_norm": 0.9615940648106116,
      "learning_rate": 9.988248845222919e-06,
      "loss": 0.4472,
      "step": 817
    },
    {
      "epoch": 0.051237883462002226,
      "grad_norm": 0.8971279133130212,
      "learning_rate": 9.98817923635398e-06,
      "loss": 0.4361,
      "step": 818
    },
    {
      "epoch": 0.05130052146134453,
      "grad_norm": 0.9104794496195636,
      "learning_rate": 9.988109422170655e-06,
      "loss": 0.4103,
      "step": 819
    },
    {
      "epoch": 0.05136315946068683,
      "grad_norm": 0.9698776167517091,
      "learning_rate": 9.988039402675823e-06,
      "loss": 0.464,
      "step": 820
    },
    {
      "epoch": 0.05142579746002913,
      "grad_norm": 0.9925907447436046,
      "learning_rate": 9.98796917787236e-06,
      "loss": 0.438,
      "step": 821
    },
    {
      "epoch": 0.05148843545937143,
      "grad_norm": 0.9585101506953108,
      "learning_rate": 9.987898747763163e-06,
      "loss": 0.4626,
      "step": 822
    },
    {
      "epoch": 0.05155107345871373,
      "grad_norm": 0.9918104652586125,
      "learning_rate": 9.987828112351125e-06,
      "loss": 0.4248,
      "step": 823
    },
    {
      "epoch": 0.05161371145805603,
      "grad_norm": 1.0231031762527514,
      "learning_rate": 9.987757271639158e-06,
      "loss": 0.4792,
      "step": 824
    },
    {
      "epoch": 0.05167634945739833,
      "grad_norm": 0.837210004291008,
      "learning_rate": 9.987686225630176e-06,
      "loss": 0.449,
      "step": 825
    },
    {
      "epoch": 0.05173898745674063,
      "grad_norm": 0.8395411293023756,
      "learning_rate": 9.987614974327102e-06,
      "loss": 0.4864,
      "step": 826
    },
    {
      "epoch": 0.05180162545608293,
      "grad_norm": 0.9803945422060287,
      "learning_rate": 9.987543517732871e-06,
      "loss": 0.4612,
      "step": 827
    },
    {
      "epoch": 0.05186426345542523,
      "grad_norm": 0.8911917159864156,
      "learning_rate": 9.987471855850424e-06,
      "loss": 0.4387,
      "step": 828
    },
    {
      "epoch": 0.051926901454767534,
      "grad_norm": 0.7484370429750258,
      "learning_rate": 9.987399988682707e-06,
      "loss": 0.5085,
      "step": 829
    },
    {
      "epoch": 0.051989539454109834,
      "grad_norm": 0.895369309064475,
      "learning_rate": 9.987327916232683e-06,
      "loss": 0.405,
      "step": 830
    },
    {
      "epoch": 0.052052177453452135,
      "grad_norm": 1.001566331005194,
      "learning_rate": 9.987255638503314e-06,
      "loss": 0.484,
      "step": 831
    },
    {
      "epoch": 0.052114815452794436,
      "grad_norm": 0.908769636797958,
      "learning_rate": 9.98718315549758e-06,
      "loss": 0.4376,
      "step": 832
    },
    {
      "epoch": 0.052177453452136736,
      "grad_norm": 0.9083590569556048,
      "learning_rate": 9.98711046721846e-06,
      "loss": 0.451,
      "step": 833
    },
    {
      "epoch": 0.05224009145147904,
      "grad_norm": 0.9863352411103733,
      "learning_rate": 9.987037573668948e-06,
      "loss": 0.4654,
      "step": 834
    },
    {
      "epoch": 0.05230272945082134,
      "grad_norm": 0.985803687548831,
      "learning_rate": 9.986964474852043e-06,
      "loss": 0.4475,
      "step": 835
    },
    {
      "epoch": 0.05236536745016364,
      "grad_norm": 0.9554757157724206,
      "learning_rate": 9.986891170770756e-06,
      "loss": 0.461,
      "step": 836
    },
    {
      "epoch": 0.052428005449505946,
      "grad_norm": 1.0269100791115995,
      "learning_rate": 9.986817661428102e-06,
      "loss": 0.4801,
      "step": 837
    },
    {
      "epoch": 0.05249064344884825,
      "grad_norm": 0.9334353419448908,
      "learning_rate": 9.986743946827106e-06,
      "loss": 0.4732,
      "step": 838
    },
    {
      "epoch": 0.05255328144819055,
      "grad_norm": 1.0003128880954313,
      "learning_rate": 9.986670026970806e-06,
      "loss": 0.4629,
      "step": 839
    },
    {
      "epoch": 0.05261591944753285,
      "grad_norm": 0.9072894332531359,
      "learning_rate": 9.986595901862242e-06,
      "loss": 0.4555,
      "step": 840
    },
    {
      "epoch": 0.05267855744687515,
      "grad_norm": 0.991975393674197,
      "learning_rate": 9.986521571504464e-06,
      "loss": 0.4895,
      "step": 841
    },
    {
      "epoch": 0.05274119544621745,
      "grad_norm": 0.9343772633504821,
      "learning_rate": 9.986447035900533e-06,
      "loss": 0.4545,
      "step": 842
    },
    {
      "epoch": 0.05280383344555975,
      "grad_norm": 0.8907924597464965,
      "learning_rate": 9.986372295053517e-06,
      "loss": 0.4212,
      "step": 843
    },
    {
      "epoch": 0.05286647144490205,
      "grad_norm": 0.8620055570097864,
      "learning_rate": 9.98629734896649e-06,
      "loss": 0.4314,
      "step": 844
    },
    {
      "epoch": 0.05292910944424435,
      "grad_norm": 0.9587957974432815,
      "learning_rate": 9.98622219764254e-06,
      "loss": 0.4893,
      "step": 845
    },
    {
      "epoch": 0.05299174744358665,
      "grad_norm": 1.0441473687464549,
      "learning_rate": 9.986146841084758e-06,
      "loss": 0.4968,
      "step": 846
    },
    {
      "epoch": 0.05305438544292895,
      "grad_norm": 0.9638521498249859,
      "learning_rate": 9.986071279296248e-06,
      "loss": 0.4265,
      "step": 847
    },
    {
      "epoch": 0.053117023442271254,
      "grad_norm": 0.9828339058826837,
      "learning_rate": 9.985995512280118e-06,
      "loss": 0.4817,
      "step": 848
    },
    {
      "epoch": 0.053179661441613554,
      "grad_norm": 0.9666892967486925,
      "learning_rate": 9.985919540039487e-06,
      "loss": 0.464,
      "step": 849
    },
    {
      "epoch": 0.053242299440955855,
      "grad_norm": 0.9658057572395716,
      "learning_rate": 9.985843362577484e-06,
      "loss": 0.5194,
      "step": 850
    },
    {
      "epoch": 0.053304937440298156,
      "grad_norm": 0.9196795697942185,
      "learning_rate": 9.98576697989724e-06,
      "loss": 0.4478,
      "step": 851
    },
    {
      "epoch": 0.053367575439640456,
      "grad_norm": 0.8624988862586199,
      "learning_rate": 9.985690392001904e-06,
      "loss": 0.405,
      "step": 852
    },
    {
      "epoch": 0.05343021343898276,
      "grad_norm": 0.9801863545193211,
      "learning_rate": 9.985613598894624e-06,
      "loss": 0.4377,
      "step": 853
    },
    {
      "epoch": 0.05349285143832506,
      "grad_norm": 0.8735660036909861,
      "learning_rate": 9.985536600578563e-06,
      "loss": 0.4472,
      "step": 854
    },
    {
      "epoch": 0.05355548943766736,
      "grad_norm": 0.8404351626312561,
      "learning_rate": 9.985459397056891e-06,
      "loss": 0.4476,
      "step": 855
    },
    {
      "epoch": 0.05361812743700966,
      "grad_norm": 0.8751077488814104,
      "learning_rate": 9.985381988332785e-06,
      "loss": 0.4488,
      "step": 856
    },
    {
      "epoch": 0.05368076543635196,
      "grad_norm": 1.0022121546005114,
      "learning_rate": 9.985304374409432e-06,
      "loss": 0.437,
      "step": 857
    },
    {
      "epoch": 0.05374340343569426,
      "grad_norm": 0.8948434350195908,
      "learning_rate": 9.985226555290025e-06,
      "loss": 0.4449,
      "step": 858
    },
    {
      "epoch": 0.05380604143503657,
      "grad_norm": 0.8913901078583684,
      "learning_rate": 9.985148530977767e-06,
      "loss": 0.4427,
      "step": 859
    },
    {
      "epoch": 0.05386867943437887,
      "grad_norm": 1.0072950677411912,
      "learning_rate": 9.98507030147587e-06,
      "loss": 0.5063,
      "step": 860
    },
    {
      "epoch": 0.05393131743372117,
      "grad_norm": 0.8816308876687692,
      "learning_rate": 9.984991866787555e-06,
      "loss": 0.431,
      "step": 861
    },
    {
      "epoch": 0.05399395543306347,
      "grad_norm": 0.9390691830159609,
      "learning_rate": 9.98491322691605e-06,
      "loss": 0.4671,
      "step": 862
    },
    {
      "epoch": 0.05405659343240577,
      "grad_norm": 0.8913814263454453,
      "learning_rate": 9.98483438186459e-06,
      "loss": 0.4358,
      "step": 863
    },
    {
      "epoch": 0.05411923143174807,
      "grad_norm": 0.9158210914677264,
      "learning_rate": 9.984755331636423e-06,
      "loss": 0.5372,
      "step": 864
    },
    {
      "epoch": 0.05418186943109037,
      "grad_norm": 0.8308840233581802,
      "learning_rate": 9.984676076234799e-06,
      "loss": 0.4347,
      "step": 865
    },
    {
      "epoch": 0.05424450743043267,
      "grad_norm": 0.9063069192921868,
      "learning_rate": 9.984596615662986e-06,
      "loss": 0.4644,
      "step": 866
    },
    {
      "epoch": 0.054307145429774974,
      "grad_norm": 0.9177900298965735,
      "learning_rate": 9.984516949924246e-06,
      "loss": 0.442,
      "step": 867
    },
    {
      "epoch": 0.054369783429117274,
      "grad_norm": 0.9840136431814384,
      "learning_rate": 9.984437079021867e-06,
      "loss": 0.5021,
      "step": 868
    },
    {
      "epoch": 0.054432421428459575,
      "grad_norm": 0.9519764128475355,
      "learning_rate": 9.984357002959131e-06,
      "loss": 0.4747,
      "step": 869
    },
    {
      "epoch": 0.054495059427801876,
      "grad_norm": 1.1098340702992966,
      "learning_rate": 9.984276721739337e-06,
      "loss": 0.449,
      "step": 870
    },
    {
      "epoch": 0.054557697427144176,
      "grad_norm": 0.8875167203147291,
      "learning_rate": 9.984196235365788e-06,
      "loss": 0.4639,
      "step": 871
    },
    {
      "epoch": 0.05462033542648648,
      "grad_norm": 1.0493588142636467,
      "learning_rate": 9.984115543841797e-06,
      "loss": 0.472,
      "step": 872
    },
    {
      "epoch": 0.05468297342582878,
      "grad_norm": 0.9217014604405765,
      "learning_rate": 9.984034647170684e-06,
      "loss": 0.4496,
      "step": 873
    },
    {
      "epoch": 0.05474561142517108,
      "grad_norm": 0.8737062924409414,
      "learning_rate": 9.983953545355781e-06,
      "loss": 0.4287,
      "step": 874
    },
    {
      "epoch": 0.05480824942451338,
      "grad_norm": 0.9701724464473825,
      "learning_rate": 9.983872238400423e-06,
      "loss": 0.4578,
      "step": 875
    },
    {
      "epoch": 0.05487088742385568,
      "grad_norm": 0.9732272823294963,
      "learning_rate": 9.98379072630796e-06,
      "loss": 0.5021,
      "step": 876
    },
    {
      "epoch": 0.05493352542319798,
      "grad_norm": 0.9609475935736345,
      "learning_rate": 9.983709009081745e-06,
      "loss": 0.432,
      "step": 877
    },
    {
      "epoch": 0.05499616342254028,
      "grad_norm": 0.9556755710942036,
      "learning_rate": 9.983627086725145e-06,
      "loss": 0.4591,
      "step": 878
    },
    {
      "epoch": 0.05505880142188258,
      "grad_norm": 1.015899606016432,
      "learning_rate": 9.983544959241525e-06,
      "loss": 0.4976,
      "step": 879
    },
    {
      "epoch": 0.05512143942122489,
      "grad_norm": 0.8956964380893778,
      "learning_rate": 9.983462626634273e-06,
      "loss": 0.4256,
      "step": 880
    },
    {
      "epoch": 0.05518407742056719,
      "grad_norm": 1.1966224134027057,
      "learning_rate": 9.98338008890677e-06,
      "loss": 0.4785,
      "step": 881
    },
    {
      "epoch": 0.05524671541990949,
      "grad_norm": 0.9558357551207638,
      "learning_rate": 9.983297346062422e-06,
      "loss": 0.427,
      "step": 882
    },
    {
      "epoch": 0.05530935341925179,
      "grad_norm": 1.0492479887215465,
      "learning_rate": 9.983214398104628e-06,
      "loss": 0.4975,
      "step": 883
    },
    {
      "epoch": 0.05537199141859409,
      "grad_norm": 1.0580169935887045,
      "learning_rate": 9.983131245036805e-06,
      "loss": 0.49,
      "step": 884
    },
    {
      "epoch": 0.05543462941793639,
      "grad_norm": 0.8940482994434045,
      "learning_rate": 9.983047886862375e-06,
      "loss": 0.4355,
      "step": 885
    },
    {
      "epoch": 0.055497267417278694,
      "grad_norm": 0.9454529619124102,
      "learning_rate": 9.982964323584769e-06,
      "loss": 0.466,
      "step": 886
    },
    {
      "epoch": 0.055559905416620994,
      "grad_norm": 0.9282112449017146,
      "learning_rate": 9.982880555207428e-06,
      "loss": 0.4499,
      "step": 887
    },
    {
      "epoch": 0.055622543415963295,
      "grad_norm": 0.9838921053676487,
      "learning_rate": 9.982796581733796e-06,
      "loss": 0.4896,
      "step": 888
    },
    {
      "epoch": 0.055685181415305596,
      "grad_norm": 0.8960840755270871,
      "learning_rate": 9.982712403167333e-06,
      "loss": 0.4381,
      "step": 889
    },
    {
      "epoch": 0.055747819414647896,
      "grad_norm": 0.9376876245939268,
      "learning_rate": 9.982628019511502e-06,
      "loss": 0.4421,
      "step": 890
    },
    {
      "epoch": 0.0558104574139902,
      "grad_norm": 0.9416246103319993,
      "learning_rate": 9.982543430769778e-06,
      "loss": 0.4838,
      "step": 891
    },
    {
      "epoch": 0.0558730954133325,
      "grad_norm": 0.9405139792428872,
      "learning_rate": 9.982458636945642e-06,
      "loss": 0.4673,
      "step": 892
    },
    {
      "epoch": 0.0559357334126748,
      "grad_norm": 0.9870481542883106,
      "learning_rate": 9.982373638042581e-06,
      "loss": 0.4406,
      "step": 893
    },
    {
      "epoch": 0.0559983714120171,
      "grad_norm": 0.8814328797603032,
      "learning_rate": 9.9822884340641e-06,
      "loss": 0.465,
      "step": 894
    },
    {
      "epoch": 0.0560610094113594,
      "grad_norm": 0.8608494393820718,
      "learning_rate": 9.982203025013698e-06,
      "loss": 0.4297,
      "step": 895
    },
    {
      "epoch": 0.0561236474107017,
      "grad_norm": 0.8372965359533916,
      "learning_rate": 9.982117410894899e-06,
      "loss": 0.442,
      "step": 896
    },
    {
      "epoch": 0.056186285410044,
      "grad_norm": 0.8889810623459387,
      "learning_rate": 9.982031591711222e-06,
      "loss": 0.4296,
      "step": 897
    },
    {
      "epoch": 0.0562489234093863,
      "grad_norm": 0.9423282543438809,
      "learning_rate": 9.981945567466198e-06,
      "loss": 0.4275,
      "step": 898
    },
    {
      "epoch": 0.0563115614087286,
      "grad_norm": 0.9520083825622397,
      "learning_rate": 9.98185933816337e-06,
      "loss": 0.4521,
      "step": 899
    },
    {
      "epoch": 0.0563741994080709,
      "grad_norm": 1.0332149006199065,
      "learning_rate": 9.981772903806288e-06,
      "loss": 0.4889,
      "step": 900
    },
    {
      "epoch": 0.056436837407413204,
      "grad_norm": 0.9363211714614416,
      "learning_rate": 9.981686264398508e-06,
      "loss": 0.4741,
      "step": 901
    },
    {
      "epoch": 0.05649947540675551,
      "grad_norm": 0.9231417419350555,
      "learning_rate": 9.981599419943597e-06,
      "loss": 0.4186,
      "step": 902
    },
    {
      "epoch": 0.05656211340609781,
      "grad_norm": 1.0143888894010589,
      "learning_rate": 9.981512370445128e-06,
      "loss": 0.4909,
      "step": 903
    },
    {
      "epoch": 0.05662475140544011,
      "grad_norm": 0.8755061832538629,
      "learning_rate": 9.981425115906687e-06,
      "loss": 0.4723,
      "step": 904
    },
    {
      "epoch": 0.056687389404782414,
      "grad_norm": 0.9589821237731164,
      "learning_rate": 9.981337656331863e-06,
      "loss": 0.4487,
      "step": 905
    },
    {
      "epoch": 0.056750027404124714,
      "grad_norm": 0.9268258671708899,
      "learning_rate": 9.981249991724256e-06,
      "loss": 0.4775,
      "step": 906
    },
    {
      "epoch": 0.056812665403467015,
      "grad_norm": 0.89910845295711,
      "learning_rate": 9.981162122087475e-06,
      "loss": 0.433,
      "step": 907
    },
    {
      "epoch": 0.056875303402809316,
      "grad_norm": 0.9124447783041417,
      "learning_rate": 9.981074047425138e-06,
      "loss": 0.5107,
      "step": 908
    },
    {
      "epoch": 0.056937941402151616,
      "grad_norm": 0.846643295064803,
      "learning_rate": 9.980985767740867e-06,
      "loss": 0.419,
      "step": 909
    },
    {
      "epoch": 0.05700057940149392,
      "grad_norm": 1.0178178220466456,
      "learning_rate": 9.980897283038297e-06,
      "loss": 0.4145,
      "step": 910
    },
    {
      "epoch": 0.05706321740083622,
      "grad_norm": 0.9162565092013449,
      "learning_rate": 9.980808593321072e-06,
      "loss": 0.4579,
      "step": 911
    },
    {
      "epoch": 0.05712585540017852,
      "grad_norm": 1.1043123941048623,
      "learning_rate": 9.98071969859284e-06,
      "loss": 0.507,
      "step": 912
    },
    {
      "epoch": 0.05718849339952082,
      "grad_norm": 0.9251331955374069,
      "learning_rate": 9.98063059885726e-06,
      "loss": 0.4665,
      "step": 913
    },
    {
      "epoch": 0.05725113139886312,
      "grad_norm": 0.984266513566876,
      "learning_rate": 9.980541294118e-06,
      "loss": 0.459,
      "step": 914
    },
    {
      "epoch": 0.05731376939820542,
      "grad_norm": 1.2232139114826488,
      "learning_rate": 9.980451784378738e-06,
      "loss": 0.4246,
      "step": 915
    },
    {
      "epoch": 0.05737640739754772,
      "grad_norm": 0.9881560985954707,
      "learning_rate": 9.980362069643155e-06,
      "loss": 0.4447,
      "step": 916
    },
    {
      "epoch": 0.05743904539689002,
      "grad_norm": 0.9332074440127756,
      "learning_rate": 9.980272149914943e-06,
      "loss": 0.5025,
      "step": 917
    },
    {
      "epoch": 0.05750168339623232,
      "grad_norm": 0.9456667678809924,
      "learning_rate": 9.980182025197806e-06,
      "loss": 0.4473,
      "step": 918
    },
    {
      "epoch": 0.05756432139557462,
      "grad_norm": 0.9203097111735135,
      "learning_rate": 9.980091695495453e-06,
      "loss": 0.4746,
      "step": 919
    },
    {
      "epoch": 0.057626959394916924,
      "grad_norm": 0.9906186872856871,
      "learning_rate": 9.980001160811603e-06,
      "loss": 0.4589,
      "step": 920
    },
    {
      "epoch": 0.057689597394259225,
      "grad_norm": 1.1380749997020274,
      "learning_rate": 9.979910421149979e-06,
      "loss": 0.4207,
      "step": 921
    },
    {
      "epoch": 0.057752235393601525,
      "grad_norm": 0.9553277263333892,
      "learning_rate": 9.979819476514318e-06,
      "loss": 0.4428,
      "step": 922
    },
    {
      "epoch": 0.057814873392943826,
      "grad_norm": 0.9094571556056714,
      "learning_rate": 9.979728326908361e-06,
      "loss": 0.4612,
      "step": 923
    },
    {
      "epoch": 0.057877511392286134,
      "grad_norm": 0.960454323871676,
      "learning_rate": 9.979636972335865e-06,
      "loss": 0.4668,
      "step": 924
    },
    {
      "epoch": 0.057940149391628434,
      "grad_norm": 0.9756658565003925,
      "learning_rate": 9.979545412800587e-06,
      "loss": 0.4712,
      "step": 925
    },
    {
      "epoch": 0.058002787390970735,
      "grad_norm": 0.8829836319876834,
      "learning_rate": 9.979453648306295e-06,
      "loss": 0.4634,
      "step": 926
    },
    {
      "epoch": 0.058065425390313036,
      "grad_norm": 0.9486913241601279,
      "learning_rate": 9.979361678856764e-06,
      "loss": 0.4469,
      "step": 927
    },
    {
      "epoch": 0.058128063389655336,
      "grad_norm": 0.9857651049633375,
      "learning_rate": 9.979269504455784e-06,
      "loss": 0.454,
      "step": 928
    },
    {
      "epoch": 0.05819070138899764,
      "grad_norm": 1.0210617396080277,
      "learning_rate": 9.979177125107147e-06,
      "loss": 0.4698,
      "step": 929
    },
    {
      "epoch": 0.05825333938833994,
      "grad_norm": 0.9926905640115624,
      "learning_rate": 9.979084540814657e-06,
      "loss": 0.4737,
      "step": 930
    },
    {
      "epoch": 0.05831597738768224,
      "grad_norm": 0.9465138558091019,
      "learning_rate": 9.978991751582122e-06,
      "loss": 0.4674,
      "step": 931
    },
    {
      "epoch": 0.05837861538702454,
      "grad_norm": 0.9575094352989362,
      "learning_rate": 9.97889875741336e-06,
      "loss": 0.4414,
      "step": 932
    },
    {
      "epoch": 0.05844125338636684,
      "grad_norm": 0.9949242789836218,
      "learning_rate": 9.978805558312203e-06,
      "loss": 0.4555,
      "step": 933
    },
    {
      "epoch": 0.05850389138570914,
      "grad_norm": 0.9797992284721461,
      "learning_rate": 9.978712154282487e-06,
      "loss": 0.4223,
      "step": 934
    },
    {
      "epoch": 0.05856652938505144,
      "grad_norm": 0.8474539995753507,
      "learning_rate": 9.978618545328052e-06,
      "loss": 0.4591,
      "step": 935
    },
    {
      "epoch": 0.05862916738439374,
      "grad_norm": 0.9561551229840582,
      "learning_rate": 9.978524731452754e-06,
      "loss": 0.4463,
      "step": 936
    },
    {
      "epoch": 0.05869180538373604,
      "grad_norm": 0.9569443626413291,
      "learning_rate": 9.978430712660453e-06,
      "loss": 0.4098,
      "step": 937
    },
    {
      "epoch": 0.05875444338307834,
      "grad_norm": 0.9030274133337753,
      "learning_rate": 9.978336488955023e-06,
      "loss": 0.4606,
      "step": 938
    },
    {
      "epoch": 0.058817081382420644,
      "grad_norm": 0.9174697376126015,
      "learning_rate": 9.978242060340335e-06,
      "loss": 0.4227,
      "step": 939
    },
    {
      "epoch": 0.058879719381762945,
      "grad_norm": 1.0669177349351004,
      "learning_rate": 9.978147426820284e-06,
      "loss": 0.5048,
      "step": 940
    },
    {
      "epoch": 0.058942357381105245,
      "grad_norm": 0.947858376392348,
      "learning_rate": 9.97805258839876e-06,
      "loss": 0.4332,
      "step": 941
    },
    {
      "epoch": 0.059004995380447546,
      "grad_norm": 0.9612461179689937,
      "learning_rate": 9.977957545079665e-06,
      "loss": 0.455,
      "step": 942
    },
    {
      "epoch": 0.05906763337978985,
      "grad_norm": 0.9218968962769395,
      "learning_rate": 9.977862296866914e-06,
      "loss": 0.4072,
      "step": 943
    },
    {
      "epoch": 0.05913027137913215,
      "grad_norm": 0.9476948402645348,
      "learning_rate": 9.977766843764428e-06,
      "loss": 0.4408,
      "step": 944
    },
    {
      "epoch": 0.05919290937847445,
      "grad_norm": 0.9946133471206838,
      "learning_rate": 9.977671185776135e-06,
      "loss": 0.4723,
      "step": 945
    },
    {
      "epoch": 0.059255547377816756,
      "grad_norm": 0.9595749695167028,
      "learning_rate": 9.977575322905972e-06,
      "loss": 0.4577,
      "step": 946
    },
    {
      "epoch": 0.059318185377159056,
      "grad_norm": 0.965937191803051,
      "learning_rate": 9.977479255157884e-06,
      "loss": 0.4605,
      "step": 947
    },
    {
      "epoch": 0.05938082337650136,
      "grad_norm": 0.8788612669870901,
      "learning_rate": 9.977382982535826e-06,
      "loss": 0.4566,
      "step": 948
    },
    {
      "epoch": 0.05944346137584366,
      "grad_norm": 0.9309368059991531,
      "learning_rate": 9.977286505043761e-06,
      "loss": 0.4714,
      "step": 949
    },
    {
      "epoch": 0.05950609937518596,
      "grad_norm": 0.9512980053668874,
      "learning_rate": 9.97718982268566e-06,
      "loss": 0.4479,
      "step": 950
    },
    {
      "epoch": 0.05956873737452826,
      "grad_norm": 0.9334506816075641,
      "learning_rate": 9.977092935465502e-06,
      "loss": 0.4515,
      "step": 951
    },
    {
      "epoch": 0.05963137537387056,
      "grad_norm": 1.053954600954042,
      "learning_rate": 9.976995843387276e-06,
      "loss": 0.4595,
      "step": 952
    },
    {
      "epoch": 0.05969401337321286,
      "grad_norm": 1.0296916802194738,
      "learning_rate": 9.976898546454977e-06,
      "loss": 0.4754,
      "step": 953
    },
    {
      "epoch": 0.05975665137255516,
      "grad_norm": 1.1303383150369335,
      "learning_rate": 9.976801044672608e-06,
      "loss": 0.4914,
      "step": 954
    },
    {
      "epoch": 0.05981928937189746,
      "grad_norm": 0.906029510011733,
      "learning_rate": 9.976703338044185e-06,
      "loss": 0.4221,
      "step": 955
    },
    {
      "epoch": 0.05988192737123976,
      "grad_norm": 0.8865745701687702,
      "learning_rate": 9.97660542657373e-06,
      "loss": 0.4453,
      "step": 956
    },
    {
      "epoch": 0.05994456537058206,
      "grad_norm": 0.9387242878975702,
      "learning_rate": 9.976507310265271e-06,
      "loss": 0.4934,
      "step": 957
    },
    {
      "epoch": 0.060007203369924364,
      "grad_norm": 0.9049891398671451,
      "learning_rate": 9.976408989122848e-06,
      "loss": 0.4732,
      "step": 958
    },
    {
      "epoch": 0.060069841369266665,
      "grad_norm": 0.9367180008559177,
      "learning_rate": 9.976310463150508e-06,
      "loss": 0.4639,
      "step": 959
    },
    {
      "epoch": 0.060132479368608965,
      "grad_norm": 0.8989167063628697,
      "learning_rate": 9.976211732352304e-06,
      "loss": 0.446,
      "step": 960
    },
    {
      "epoch": 0.060195117367951266,
      "grad_norm": 1.0014260967778823,
      "learning_rate": 9.976112796732301e-06,
      "loss": 0.453,
      "step": 961
    },
    {
      "epoch": 0.06025775536729357,
      "grad_norm": 0.9659638953722826,
      "learning_rate": 9.976013656294575e-06,
      "loss": 0.4238,
      "step": 962
    },
    {
      "epoch": 0.06032039336663587,
      "grad_norm": 0.9039177971765535,
      "learning_rate": 9.9759143110432e-06,
      "loss": 0.4552,
      "step": 963
    },
    {
      "epoch": 0.06038303136597817,
      "grad_norm": 0.880517260929041,
      "learning_rate": 9.975814760982269e-06,
      "loss": 0.3886,
      "step": 964
    },
    {
      "epoch": 0.06044566936532047,
      "grad_norm": 0.9166192865662124,
      "learning_rate": 9.97571500611588e-06,
      "loss": 0.4598,
      "step": 965
    },
    {
      "epoch": 0.06050830736466277,
      "grad_norm": 0.8944555417296142,
      "learning_rate": 9.975615046448137e-06,
      "loss": 0.4945,
      "step": 966
    },
    {
      "epoch": 0.06057094536400508,
      "grad_norm": 0.9709797180527064,
      "learning_rate": 9.975514881983154e-06,
      "loss": 0.4231,
      "step": 967
    },
    {
      "epoch": 0.06063358336334738,
      "grad_norm": 0.954999533709772,
      "learning_rate": 9.975414512725058e-06,
      "loss": 0.4579,
      "step": 968
    },
    {
      "epoch": 0.06069622136268968,
      "grad_norm": 0.8931842778492058,
      "learning_rate": 9.975313938677975e-06,
      "loss": 0.4304,
      "step": 969
    },
    {
      "epoch": 0.06075885936203198,
      "grad_norm": 0.8642292049908066,
      "learning_rate": 9.975213159846048e-06,
      "loss": 0.4202,
      "step": 970
    },
    {
      "epoch": 0.06082149736137428,
      "grad_norm": 0.8977775615294807,
      "learning_rate": 9.975112176233424e-06,
      "loss": 0.4585,
      "step": 971
    },
    {
      "epoch": 0.06088413536071658,
      "grad_norm": 0.9659720478773683,
      "learning_rate": 9.975010987844258e-06,
      "loss": 0.4706,
      "step": 972
    },
    {
      "epoch": 0.06094677336005888,
      "grad_norm": 0.9324741276768244,
      "learning_rate": 9.974909594682716e-06,
      "loss": 0.4271,
      "step": 973
    },
    {
      "epoch": 0.06100941135940118,
      "grad_norm": 0.9268965987266379,
      "learning_rate": 9.974807996752974e-06,
      "loss": 0.4587,
      "step": 974
    },
    {
      "epoch": 0.06107204935874348,
      "grad_norm": 0.8108900599759947,
      "learning_rate": 9.97470619405921e-06,
      "loss": 0.4358,
      "step": 975
    },
    {
      "epoch": 0.06113468735808578,
      "grad_norm": 0.9732965503219843,
      "learning_rate": 9.974604186605616e-06,
      "loss": 0.435,
      "step": 976
    },
    {
      "epoch": 0.061197325357428084,
      "grad_norm": 1.1011568168130204,
      "learning_rate": 9.97450197439639e-06,
      "loss": 0.4677,
      "step": 977
    },
    {
      "epoch": 0.061259963356770385,
      "grad_norm": 0.8867602174200493,
      "learning_rate": 9.97439955743574e-06,
      "loss": 0.4802,
      "step": 978
    },
    {
      "epoch": 0.061322601356112685,
      "grad_norm": 0.8921255658508139,
      "learning_rate": 9.97429693572788e-06,
      "loss": 0.4501,
      "step": 979
    },
    {
      "epoch": 0.061385239355454986,
      "grad_norm": 0.935648865338509,
      "learning_rate": 9.974194109277036e-06,
      "loss": 0.4718,
      "step": 980
    },
    {
      "epoch": 0.06144787735479729,
      "grad_norm": 0.8770261663061795,
      "learning_rate": 9.97409107808744e-06,
      "loss": 0.3942,
      "step": 981
    },
    {
      "epoch": 0.06151051535413959,
      "grad_norm": 0.8714295095284562,
      "learning_rate": 9.97398784216333e-06,
      "loss": 0.4799,
      "step": 982
    },
    {
      "epoch": 0.06157315335348189,
      "grad_norm": 0.9293139694323811,
      "learning_rate": 9.973884401508958e-06,
      "loss": 0.4552,
      "step": 983
    },
    {
      "epoch": 0.06163579135282419,
      "grad_norm": 0.9655245783214117,
      "learning_rate": 9.97378075612858e-06,
      "loss": 0.4652,
      "step": 984
    },
    {
      "epoch": 0.06169842935216649,
      "grad_norm": 0.8889073839939376,
      "learning_rate": 9.973676906026464e-06,
      "loss": 0.4775,
      "step": 985
    },
    {
      "epoch": 0.06176106735150879,
      "grad_norm": 0.7771725776282543,
      "learning_rate": 9.973572851206882e-06,
      "loss": 0.4946,
      "step": 986
    },
    {
      "epoch": 0.06182370535085109,
      "grad_norm": 0.920354583074144,
      "learning_rate": 9.973468591674121e-06,
      "loss": 0.4686,
      "step": 987
    },
    {
      "epoch": 0.06188634335019339,
      "grad_norm": 0.9105984779556204,
      "learning_rate": 9.973364127432468e-06,
      "loss": 0.4635,
      "step": 988
    },
    {
      "epoch": 0.0619489813495357,
      "grad_norm": 0.9050315508186861,
      "learning_rate": 9.973259458486225e-06,
      "loss": 0.4596,
      "step": 989
    },
    {
      "epoch": 0.062011619348878,
      "grad_norm": 0.8844632321456114,
      "learning_rate": 9.973154584839698e-06,
      "loss": 0.4277,
      "step": 990
    },
    {
      "epoch": 0.0620742573482203,
      "grad_norm": 0.9141800110896054,
      "learning_rate": 9.973049506497205e-06,
      "loss": 0.4341,
      "step": 991
    },
    {
      "epoch": 0.0621368953475626,
      "grad_norm": 0.911077269839778,
      "learning_rate": 9.972944223463073e-06,
      "loss": 0.4416,
      "step": 992
    },
    {
      "epoch": 0.0621995333469049,
      "grad_norm": 0.978684313220983,
      "learning_rate": 9.972838735741633e-06,
      "loss": 0.3992,
      "step": 993
    },
    {
      "epoch": 0.0622621713462472,
      "grad_norm": 0.9362980721917727,
      "learning_rate": 9.972733043337228e-06,
      "loss": 0.4737,
      "step": 994
    },
    {
      "epoch": 0.0623248093455895,
      "grad_norm": 0.9751142882755939,
      "learning_rate": 9.972627146254208e-06,
      "loss": 0.4559,
      "step": 995
    },
    {
      "epoch": 0.062387447344931804,
      "grad_norm": 1.009280100128346,
      "learning_rate": 9.972521044496932e-06,
      "loss": 0.4666,
      "step": 996
    },
    {
      "epoch": 0.062450085344274105,
      "grad_norm": 0.9057171229464022,
      "learning_rate": 9.972414738069766e-06,
      "loss": 0.4133,
      "step": 997
    },
    {
      "epoch": 0.0625127233436164,
      "grad_norm": 0.8771620729376317,
      "learning_rate": 9.972308226977086e-06,
      "loss": 0.4736,
      "step": 998
    },
    {
      "epoch": 0.0625753613429587,
      "grad_norm": 0.9838228405145389,
      "learning_rate": 9.972201511223278e-06,
      "loss": 0.4627,
      "step": 999
    },
    {
      "epoch": 0.062637999342301,
      "grad_norm": 0.8821466566844415,
      "learning_rate": 9.972094590812734e-06,
      "loss": 0.4423,
      "step": 1000
    },
    {
      "epoch": 0.06270063734164331,
      "grad_norm": 0.9644488339449918,
      "learning_rate": 9.971987465749851e-06,
      "loss": 0.4878,
      "step": 1001
    },
    {
      "epoch": 0.06276327534098561,
      "grad_norm": 0.9673194926596151,
      "learning_rate": 9.971880136039042e-06,
      "loss": 0.4626,
      "step": 1002
    },
    {
      "epoch": 0.06282591334032792,
      "grad_norm": 0.9178048293334611,
      "learning_rate": 9.971772601684724e-06,
      "loss": 0.4788,
      "step": 1003
    },
    {
      "epoch": 0.06288855133967021,
      "grad_norm": 0.9296480613615702,
      "learning_rate": 9.971664862691322e-06,
      "loss": 0.4383,
      "step": 1004
    },
    {
      "epoch": 0.06295118933901252,
      "grad_norm": 0.903606460617527,
      "learning_rate": 9.971556919063273e-06,
      "loss": 0.4506,
      "step": 1005
    },
    {
      "epoch": 0.06301382733835481,
      "grad_norm": 0.9209403957513969,
      "learning_rate": 9.971448770805017e-06,
      "loss": 0.4542,
      "step": 1006
    },
    {
      "epoch": 0.06307646533769712,
      "grad_norm": 0.9397220390110198,
      "learning_rate": 9.971340417921008e-06,
      "loss": 0.4388,
      "step": 1007
    },
    {
      "epoch": 0.06313910333703941,
      "grad_norm": 0.9904803669120282,
      "learning_rate": 9.971231860415704e-06,
      "loss": 0.442,
      "step": 1008
    },
    {
      "epoch": 0.06320174133638172,
      "grad_norm": 0.9131840867728517,
      "learning_rate": 9.971123098293574e-06,
      "loss": 0.4408,
      "step": 1009
    },
    {
      "epoch": 0.06326437933572401,
      "grad_norm": 0.8986131443348198,
      "learning_rate": 9.971014131559094e-06,
      "loss": 0.4649,
      "step": 1010
    },
    {
      "epoch": 0.06332701733506632,
      "grad_norm": 1.0056334920104384,
      "learning_rate": 9.970904960216751e-06,
      "loss": 0.4705,
      "step": 1011
    },
    {
      "epoch": 0.06338965533440861,
      "grad_norm": 0.9071066584465742,
      "learning_rate": 9.970795584271036e-06,
      "loss": 0.4621,
      "step": 1012
    },
    {
      "epoch": 0.06345229333375092,
      "grad_norm": 0.9036766481701008,
      "learning_rate": 9.970686003726453e-06,
      "loss": 0.4327,
      "step": 1013
    },
    {
      "epoch": 0.06351493133309322,
      "grad_norm": 0.9077034810405973,
      "learning_rate": 9.970576218587511e-06,
      "loss": 0.4538,
      "step": 1014
    },
    {
      "epoch": 0.06357756933243552,
      "grad_norm": 0.957117245868161,
      "learning_rate": 9.97046622885873e-06,
      "loss": 0.4604,
      "step": 1015
    },
    {
      "epoch": 0.06364020733177782,
      "grad_norm": 0.9110448941827765,
      "learning_rate": 9.970356034544636e-06,
      "loss": 0.455,
      "step": 1016
    },
    {
      "epoch": 0.06370284533112013,
      "grad_norm": 0.8248256897242008,
      "learning_rate": 9.970245635649764e-06,
      "loss": 0.4003,
      "step": 1017
    },
    {
      "epoch": 0.06376548333046242,
      "grad_norm": 0.9991489189376284,
      "learning_rate": 9.97013503217866e-06,
      "loss": 0.4794,
      "step": 1018
    },
    {
      "epoch": 0.06382812132980473,
      "grad_norm": 0.9409702245941902,
      "learning_rate": 9.970024224135877e-06,
      "loss": 0.4774,
      "step": 1019
    },
    {
      "epoch": 0.06389075932914702,
      "grad_norm": 0.9028339116343745,
      "learning_rate": 9.969913211525972e-06,
      "loss": 0.4802,
      "step": 1020
    },
    {
      "epoch": 0.06395339732848933,
      "grad_norm": 0.8781710156893328,
      "learning_rate": 9.969801994353519e-06,
      "loss": 0.4277,
      "step": 1021
    },
    {
      "epoch": 0.06401603532783164,
      "grad_norm": 0.8541983134215585,
      "learning_rate": 9.969690572623092e-06,
      "loss": 0.4217,
      "step": 1022
    },
    {
      "epoch": 0.06407867332717393,
      "grad_norm": 0.7660410972984174,
      "learning_rate": 9.969578946339278e-06,
      "loss": 0.4877,
      "step": 1023
    },
    {
      "epoch": 0.06414131132651624,
      "grad_norm": 0.8484072247725958,
      "learning_rate": 9.969467115506674e-06,
      "loss": 0.4517,
      "step": 1024
    },
    {
      "epoch": 0.06420394932585853,
      "grad_norm": 0.8873983109178231,
      "learning_rate": 9.969355080129881e-06,
      "loss": 0.4833,
      "step": 1025
    },
    {
      "epoch": 0.06426658732520084,
      "grad_norm": 0.9061388665907908,
      "learning_rate": 9.96924284021351e-06,
      "loss": 0.4201,
      "step": 1026
    },
    {
      "epoch": 0.06432922532454313,
      "grad_norm": 0.9243026718581335,
      "learning_rate": 9.969130395762182e-06,
      "loss": 0.4285,
      "step": 1027
    },
    {
      "epoch": 0.06439186332388544,
      "grad_norm": 0.8853667719331157,
      "learning_rate": 9.969017746780523e-06,
      "loss": 0.4659,
      "step": 1028
    },
    {
      "epoch": 0.06445450132322773,
      "grad_norm": 0.9006652863226411,
      "learning_rate": 9.968904893273172e-06,
      "loss": 0.468,
      "step": 1029
    },
    {
      "epoch": 0.06451713932257004,
      "grad_norm": 0.8372570179293093,
      "learning_rate": 9.968791835244773e-06,
      "loss": 0.4477,
      "step": 1030
    },
    {
      "epoch": 0.06457977732191233,
      "grad_norm": 0.9128770938664015,
      "learning_rate": 9.968678572699978e-06,
      "loss": 0.4654,
      "step": 1031
    },
    {
      "epoch": 0.06464241532125464,
      "grad_norm": 0.8855725196001685,
      "learning_rate": 9.968565105643456e-06,
      "loss": 0.4609,
      "step": 1032
    },
    {
      "epoch": 0.06470505332059694,
      "grad_norm": 0.9001299675168565,
      "learning_rate": 9.968451434079868e-06,
      "loss": 0.452,
      "step": 1033
    },
    {
      "epoch": 0.06476769131993924,
      "grad_norm": 0.9316845603520797,
      "learning_rate": 9.968337558013898e-06,
      "loss": 0.4792,
      "step": 1034
    },
    {
      "epoch": 0.06483032931928154,
      "grad_norm": 0.87638039153309,
      "learning_rate": 9.968223477450231e-06,
      "loss": 0.4416,
      "step": 1035
    },
    {
      "epoch": 0.06489296731862385,
      "grad_norm": 0.8877389999941736,
      "learning_rate": 9.968109192393564e-06,
      "loss": 0.4432,
      "step": 1036
    },
    {
      "epoch": 0.06495560531796614,
      "grad_norm": 0.9182093475014985,
      "learning_rate": 9.9679947028486e-06,
      "loss": 0.4341,
      "step": 1037
    },
    {
      "epoch": 0.06501824331730845,
      "grad_norm": 0.8484111084510647,
      "learning_rate": 9.967880008820053e-06,
      "loss": 0.4561,
      "step": 1038
    },
    {
      "epoch": 0.06508088131665074,
      "grad_norm": 0.8244522808702833,
      "learning_rate": 9.967765110312642e-06,
      "loss": 0.4406,
      "step": 1039
    },
    {
      "epoch": 0.06514351931599305,
      "grad_norm": 0.9040459731092966,
      "learning_rate": 9.967650007331098e-06,
      "loss": 0.4344,
      "step": 1040
    },
    {
      "epoch": 0.06520615731533534,
      "grad_norm": 0.9244436619548205,
      "learning_rate": 9.967534699880157e-06,
      "loss": 0.4565,
      "step": 1041
    },
    {
      "epoch": 0.06526879531467765,
      "grad_norm": 0.8542188210499615,
      "learning_rate": 9.967419187964567e-06,
      "loss": 0.4203,
      "step": 1042
    },
    {
      "epoch": 0.06533143331401996,
      "grad_norm": 0.9084288471077621,
      "learning_rate": 9.96730347158908e-06,
      "loss": 0.4148,
      "step": 1043
    },
    {
      "epoch": 0.06539407131336225,
      "grad_norm": 1.0000715132494038,
      "learning_rate": 9.967187550758459e-06,
      "loss": 0.4318,
      "step": 1044
    },
    {
      "epoch": 0.06545670931270456,
      "grad_norm": 0.9084262834924828,
      "learning_rate": 9.967071425477477e-06,
      "loss": 0.4506,
      "step": 1045
    },
    {
      "epoch": 0.06551934731204685,
      "grad_norm": 0.9551499610852525,
      "learning_rate": 9.966955095750915e-06,
      "loss": 0.4475,
      "step": 1046
    },
    {
      "epoch": 0.06558198531138916,
      "grad_norm": 0.923131266093865,
      "learning_rate": 9.966838561583559e-06,
      "loss": 0.4447,
      "step": 1047
    },
    {
      "epoch": 0.06564462331073145,
      "grad_norm": 0.873682475985493,
      "learning_rate": 9.966721822980203e-06,
      "loss": 0.4422,
      "step": 1048
    },
    {
      "epoch": 0.06570726131007376,
      "grad_norm": 1.0050883055046294,
      "learning_rate": 9.966604879945659e-06,
      "loss": 0.4771,
      "step": 1049
    },
    {
      "epoch": 0.06576989930941605,
      "grad_norm": 0.9193923258615352,
      "learning_rate": 9.966487732484733e-06,
      "loss": 0.4458,
      "step": 1050
    },
    {
      "epoch": 0.06583253730875836,
      "grad_norm": 0.9431915540285041,
      "learning_rate": 9.966370380602252e-06,
      "loss": 0.4814,
      "step": 1051
    },
    {
      "epoch": 0.06589517530810066,
      "grad_norm": 0.9187129039029993,
      "learning_rate": 9.966252824303044e-06,
      "loss": 0.4401,
      "step": 1052
    },
    {
      "epoch": 0.06595781330744296,
      "grad_norm": 0.8963055310063858,
      "learning_rate": 9.966135063591946e-06,
      "loss": 0.4159,
      "step": 1053
    },
    {
      "epoch": 0.06602045130678526,
      "grad_norm": 0.9744156641165728,
      "learning_rate": 9.96601709847381e-06,
      "loss": 0.5086,
      "step": 1054
    },
    {
      "epoch": 0.06608308930612757,
      "grad_norm": 0.9857139727399274,
      "learning_rate": 9.965898928953486e-06,
      "loss": 0.4799,
      "step": 1055
    },
    {
      "epoch": 0.06614572730546986,
      "grad_norm": 0.8686198172842436,
      "learning_rate": 9.965780555035842e-06,
      "loss": 0.4018,
      "step": 1056
    },
    {
      "epoch": 0.06620836530481217,
      "grad_norm": 0.862618625627495,
      "learning_rate": 9.965661976725748e-06,
      "loss": 0.4551,
      "step": 1057
    },
    {
      "epoch": 0.06627100330415446,
      "grad_norm": 0.9424355279202712,
      "learning_rate": 9.965543194028086e-06,
      "loss": 0.4766,
      "step": 1058
    },
    {
      "epoch": 0.06633364130349677,
      "grad_norm": 0.9353845625708711,
      "learning_rate": 9.965424206947745e-06,
      "loss": 0.4057,
      "step": 1059
    },
    {
      "epoch": 0.06639627930283906,
      "grad_norm": 0.9827181201359431,
      "learning_rate": 9.96530501548962e-06,
      "loss": 0.4888,
      "step": 1060
    },
    {
      "epoch": 0.06645891730218137,
      "grad_norm": 0.9315698778385604,
      "learning_rate": 9.965185619658621e-06,
      "loss": 0.4099,
      "step": 1061
    },
    {
      "epoch": 0.06652155530152366,
      "grad_norm": 0.8863635864541215,
      "learning_rate": 9.96506601945966e-06,
      "loss": 0.4431,
      "step": 1062
    },
    {
      "epoch": 0.06658419330086597,
      "grad_norm": 0.9578244485706328,
      "learning_rate": 9.964946214897659e-06,
      "loss": 0.4743,
      "step": 1063
    },
    {
      "epoch": 0.06664683130020826,
      "grad_norm": 1.0078022609933255,
      "learning_rate": 9.96482620597755e-06,
      "loss": 0.4279,
      "step": 1064
    },
    {
      "epoch": 0.06670946929955057,
      "grad_norm": 0.8946946824065742,
      "learning_rate": 9.964705992704273e-06,
      "loss": 0.4785,
      "step": 1065
    },
    {
      "epoch": 0.06677210729889288,
      "grad_norm": 0.9246503965463188,
      "learning_rate": 9.964585575082777e-06,
      "loss": 0.4267,
      "step": 1066
    },
    {
      "epoch": 0.06683474529823517,
      "grad_norm": 0.8731057352785752,
      "learning_rate": 9.964464953118016e-06,
      "loss": 0.454,
      "step": 1067
    },
    {
      "epoch": 0.06689738329757748,
      "grad_norm": 0.9961925471354837,
      "learning_rate": 9.964344126814958e-06,
      "loss": 0.4336,
      "step": 1068
    },
    {
      "epoch": 0.06696002129691978,
      "grad_norm": 0.9713635591582898,
      "learning_rate": 9.964223096178572e-06,
      "loss": 0.4508,
      "step": 1069
    },
    {
      "epoch": 0.06702265929626208,
      "grad_norm": 0.9571056815436522,
      "learning_rate": 9.964101861213844e-06,
      "loss": 0.4316,
      "step": 1070
    },
    {
      "epoch": 0.06708529729560438,
      "grad_norm": 0.9482825518772832,
      "learning_rate": 9.963980421925761e-06,
      "loss": 0.474,
      "step": 1071
    },
    {
      "epoch": 0.06714793529494668,
      "grad_norm": 0.9170432560258359,
      "learning_rate": 9.963858778319322e-06,
      "loss": 0.4516,
      "step": 1072
    },
    {
      "epoch": 0.06721057329428898,
      "grad_norm": 0.9166343971621109,
      "learning_rate": 9.963736930399535e-06,
      "loss": 0.4589,
      "step": 1073
    },
    {
      "epoch": 0.06727321129363129,
      "grad_norm": 0.8559648795578971,
      "learning_rate": 9.963614878171416e-06,
      "loss": 0.4113,
      "step": 1074
    },
    {
      "epoch": 0.06733584929297358,
      "grad_norm": 0.8612745386116809,
      "learning_rate": 9.963492621639982e-06,
      "loss": 0.463,
      "step": 1075
    },
    {
      "epoch": 0.06739848729231589,
      "grad_norm": 0.9310823501186235,
      "learning_rate": 9.963370160810275e-06,
      "loss": 0.4287,
      "step": 1076
    },
    {
      "epoch": 0.06746112529165818,
      "grad_norm": 0.88813870771171,
      "learning_rate": 9.963247495687331e-06,
      "loss": 0.4282,
      "step": 1077
    },
    {
      "epoch": 0.06752376329100049,
      "grad_norm": 0.9097374814041197,
      "learning_rate": 9.963124626276199e-06,
      "loss": 0.4434,
      "step": 1078
    },
    {
      "epoch": 0.06758640129034278,
      "grad_norm": 0.9612153673565844,
      "learning_rate": 9.963001552581935e-06,
      "loss": 0.4513,
      "step": 1079
    },
    {
      "epoch": 0.06764903928968509,
      "grad_norm": 0.9333535480026945,
      "learning_rate": 9.962878274609605e-06,
      "loss": 0.4479,
      "step": 1080
    },
    {
      "epoch": 0.06771167728902738,
      "grad_norm": 0.8827605912405209,
      "learning_rate": 9.962754792364287e-06,
      "loss": 0.4936,
      "step": 1081
    },
    {
      "epoch": 0.06777431528836969,
      "grad_norm": 0.9584891305113715,
      "learning_rate": 9.962631105851058e-06,
      "loss": 0.4306,
      "step": 1082
    },
    {
      "epoch": 0.06783695328771198,
      "grad_norm": 0.9311372333280447,
      "learning_rate": 9.962507215075012e-06,
      "loss": 0.4072,
      "step": 1083
    },
    {
      "epoch": 0.06789959128705429,
      "grad_norm": 0.9068682767925895,
      "learning_rate": 9.962383120041248e-06,
      "loss": 0.4411,
      "step": 1084
    },
    {
      "epoch": 0.06796222928639659,
      "grad_norm": 0.9137223296077437,
      "learning_rate": 9.962258820754875e-06,
      "loss": 0.474,
      "step": 1085
    },
    {
      "epoch": 0.0680248672857389,
      "grad_norm": 0.8972665858279727,
      "learning_rate": 9.962134317221007e-06,
      "loss": 0.3916,
      "step": 1086
    },
    {
      "epoch": 0.0680875052850812,
      "grad_norm": 0.9548936070325101,
      "learning_rate": 9.962009609444768e-06,
      "loss": 0.4522,
      "step": 1087
    },
    {
      "epoch": 0.0681501432844235,
      "grad_norm": 1.0181335766040742,
      "learning_rate": 9.961884697431294e-06,
      "loss": 0.4348,
      "step": 1088
    },
    {
      "epoch": 0.0682127812837658,
      "grad_norm": 0.8984773758659099,
      "learning_rate": 9.961759581185725e-06,
      "loss": 0.4483,
      "step": 1089
    },
    {
      "epoch": 0.0682754192831081,
      "grad_norm": 0.8649458717935091,
      "learning_rate": 9.961634260713212e-06,
      "loss": 0.4396,
      "step": 1090
    },
    {
      "epoch": 0.0683380572824504,
      "grad_norm": 0.9806224292421863,
      "learning_rate": 9.961508736018908e-06,
      "loss": 0.4714,
      "step": 1091
    },
    {
      "epoch": 0.0684006952817927,
      "grad_norm": 1.014288825771017,
      "learning_rate": 9.961383007107988e-06,
      "loss": 0.476,
      "step": 1092
    },
    {
      "epoch": 0.068463333281135,
      "grad_norm": 0.913084649542963,
      "learning_rate": 9.961257073985619e-06,
      "loss": 0.4614,
      "step": 1093
    },
    {
      "epoch": 0.0685259712804773,
      "grad_norm": 0.8344525450440529,
      "learning_rate": 9.961130936656989e-06,
      "loss": 0.4319,
      "step": 1094
    },
    {
      "epoch": 0.0685886092798196,
      "grad_norm": 0.896882409207514,
      "learning_rate": 9.961004595127291e-06,
      "loss": 0.4215,
      "step": 1095
    },
    {
      "epoch": 0.0686512472791619,
      "grad_norm": 0.9791635581839561,
      "learning_rate": 9.96087804940172e-06,
      "loss": 0.4479,
      "step": 1096
    },
    {
      "epoch": 0.06871388527850421,
      "grad_norm": 0.8448729645162987,
      "learning_rate": 9.960751299485489e-06,
      "loss": 0.4781,
      "step": 1097
    },
    {
      "epoch": 0.0687765232778465,
      "grad_norm": 0.9599394220550328,
      "learning_rate": 9.960624345383813e-06,
      "loss": 0.4645,
      "step": 1098
    },
    {
      "epoch": 0.06883916127718881,
      "grad_norm": 0.9654315503801658,
      "learning_rate": 9.96049718710192e-06,
      "loss": 0.4732,
      "step": 1099
    },
    {
      "epoch": 0.0689017992765311,
      "grad_norm": 0.9964694285345661,
      "learning_rate": 9.960369824645042e-06,
      "loss": 0.4457,
      "step": 1100
    },
    {
      "epoch": 0.06896443727587341,
      "grad_norm": 0.9405536948237645,
      "learning_rate": 9.96024225801842e-06,
      "loss": 0.4447,
      "step": 1101
    },
    {
      "epoch": 0.0690270752752157,
      "grad_norm": 0.9593962207457893,
      "learning_rate": 9.960114487227307e-06,
      "loss": 0.4461,
      "step": 1102
    },
    {
      "epoch": 0.06908971327455801,
      "grad_norm": 0.8650910378535112,
      "learning_rate": 9.95998651227696e-06,
      "loss": 0.4013,
      "step": 1103
    },
    {
      "epoch": 0.0691523512739003,
      "grad_norm": 0.9822624090824733,
      "learning_rate": 9.959858333172647e-06,
      "loss": 0.4565,
      "step": 1104
    },
    {
      "epoch": 0.06921498927324261,
      "grad_norm": 1.0289869184405083,
      "learning_rate": 9.959729949919646e-06,
      "loss": 0.4375,
      "step": 1105
    },
    {
      "epoch": 0.06927762727258491,
      "grad_norm": 0.9297178769887658,
      "learning_rate": 9.959601362523238e-06,
      "loss": 0.4444,
      "step": 1106
    },
    {
      "epoch": 0.06934026527192722,
      "grad_norm": 0.9825713951325267,
      "learning_rate": 9.95947257098872e-06,
      "loss": 0.4598,
      "step": 1107
    },
    {
      "epoch": 0.06940290327126951,
      "grad_norm": 1.0194870028196827,
      "learning_rate": 9.959343575321389e-06,
      "loss": 0.4253,
      "step": 1108
    },
    {
      "epoch": 0.06946554127061182,
      "grad_norm": 0.9502963612492574,
      "learning_rate": 9.959214375526556e-06,
      "loss": 0.428,
      "step": 1109
    },
    {
      "epoch": 0.06952817926995412,
      "grad_norm": 0.8670379745980935,
      "learning_rate": 9.95908497160954e-06,
      "loss": 0.4599,
      "step": 1110
    },
    {
      "epoch": 0.06959081726929642,
      "grad_norm": 0.9238120039163563,
      "learning_rate": 9.958955363575661e-06,
      "loss": 0.4309,
      "step": 1111
    },
    {
      "epoch": 0.06965345526863873,
      "grad_norm": 0.9775400357781101,
      "learning_rate": 9.958825551430265e-06,
      "loss": 0.4628,
      "step": 1112
    },
    {
      "epoch": 0.06971609326798102,
      "grad_norm": 0.925876157090268,
      "learning_rate": 9.958695535178685e-06,
      "loss": 0.4871,
      "step": 1113
    },
    {
      "epoch": 0.06977873126732333,
      "grad_norm": 0.9621446605084049,
      "learning_rate": 9.958565314826279e-06,
      "loss": 0.4714,
      "step": 1114
    },
    {
      "epoch": 0.06984136926666562,
      "grad_norm": 1.0005025514571129,
      "learning_rate": 9.958434890378404e-06,
      "loss": 0.4501,
      "step": 1115
    },
    {
      "epoch": 0.06990400726600793,
      "grad_norm": 0.9935280843101303,
      "learning_rate": 9.958304261840427e-06,
      "loss": 0.4573,
      "step": 1116
    },
    {
      "epoch": 0.06996664526535022,
      "grad_norm": 0.9415024323495434,
      "learning_rate": 9.958173429217727e-06,
      "loss": 0.4385,
      "step": 1117
    },
    {
      "epoch": 0.07002928326469253,
      "grad_norm": 0.9914079369023342,
      "learning_rate": 9.958042392515688e-06,
      "loss": 0.4234,
      "step": 1118
    },
    {
      "epoch": 0.07009192126403482,
      "grad_norm": 0.9035275437428306,
      "learning_rate": 9.957911151739705e-06,
      "loss": 0.4474,
      "step": 1119
    },
    {
      "epoch": 0.07015455926337713,
      "grad_norm": 1.0076690702132527,
      "learning_rate": 9.957779706895177e-06,
      "loss": 0.4693,
      "step": 1120
    },
    {
      "epoch": 0.07021719726271942,
      "grad_norm": 0.9182887107050397,
      "learning_rate": 9.957648057987517e-06,
      "loss": 0.4597,
      "step": 1121
    },
    {
      "epoch": 0.07027983526206173,
      "grad_norm": 0.815784218228752,
      "learning_rate": 9.957516205022144e-06,
      "loss": 0.4545,
      "step": 1122
    },
    {
      "epoch": 0.07034247326140403,
      "grad_norm": 0.973652260628064,
      "learning_rate": 9.957384148004483e-06,
      "loss": 0.4199,
      "step": 1123
    },
    {
      "epoch": 0.07040511126074633,
      "grad_norm": 0.9553005309946088,
      "learning_rate": 9.957251886939969e-06,
      "loss": 0.4401,
      "step": 1124
    },
    {
      "epoch": 0.07046774926008863,
      "grad_norm": 0.9483020112212504,
      "learning_rate": 9.957119421834048e-06,
      "loss": 0.4563,
      "step": 1125
    },
    {
      "epoch": 0.07053038725943094,
      "grad_norm": 0.9882903751244959,
      "learning_rate": 9.956986752692172e-06,
      "loss": 0.442,
      "step": 1126
    },
    {
      "epoch": 0.07059302525877323,
      "grad_norm": 0.907178779271339,
      "learning_rate": 9.9568538795198e-06,
      "loss": 0.3905,
      "step": 1127
    },
    {
      "epoch": 0.07065566325811554,
      "grad_norm": 0.9573002407315578,
      "learning_rate": 9.956720802322403e-06,
      "loss": 0.4589,
      "step": 1128
    },
    {
      "epoch": 0.07071830125745783,
      "grad_norm": 0.8445772019176783,
      "learning_rate": 9.95658752110546e-06,
      "loss": 0.4699,
      "step": 1129
    },
    {
      "epoch": 0.07078093925680014,
      "grad_norm": 0.9549087782177684,
      "learning_rate": 9.956454035874452e-06,
      "loss": 0.4649,
      "step": 1130
    },
    {
      "epoch": 0.07084357725614245,
      "grad_norm": 0.9568305742933522,
      "learning_rate": 9.956320346634877e-06,
      "loss": 0.4826,
      "step": 1131
    },
    {
      "epoch": 0.07090621525548474,
      "grad_norm": 0.9096135562349444,
      "learning_rate": 9.956186453392235e-06,
      "loss": 0.4942,
      "step": 1132
    },
    {
      "epoch": 0.07096885325482705,
      "grad_norm": 0.8795459110597155,
      "learning_rate": 9.956052356152041e-06,
      "loss": 0.4328,
      "step": 1133
    },
    {
      "epoch": 0.07103149125416934,
      "grad_norm": 0.9841528295810771,
      "learning_rate": 9.95591805491981e-06,
      "loss": 0.4739,
      "step": 1134
    },
    {
      "epoch": 0.07109412925351165,
      "grad_norm": 0.9301455559437479,
      "learning_rate": 9.955783549701075e-06,
      "loss": 0.4814,
      "step": 1135
    },
    {
      "epoch": 0.07115676725285394,
      "grad_norm": 0.8548161740828436,
      "learning_rate": 9.955648840501366e-06,
      "loss": 0.4115,
      "step": 1136
    },
    {
      "epoch": 0.07121940525219625,
      "grad_norm": 0.9510972828463805,
      "learning_rate": 9.955513927326233e-06,
      "loss": 0.4569,
      "step": 1137
    },
    {
      "epoch": 0.07128204325153854,
      "grad_norm": 0.86684128366375,
      "learning_rate": 9.955378810181225e-06,
      "loss": 0.4396,
      "step": 1138
    },
    {
      "epoch": 0.07134468125088085,
      "grad_norm": 1.0117445198628543,
      "learning_rate": 9.955243489071906e-06,
      "loss": 0.4851,
      "step": 1139
    },
    {
      "epoch": 0.07140731925022314,
      "grad_norm": 0.9510666838591837,
      "learning_rate": 9.955107964003847e-06,
      "loss": 0.465,
      "step": 1140
    },
    {
      "epoch": 0.07146995724956545,
      "grad_norm": 0.8990530930038272,
      "learning_rate": 9.954972234982622e-06,
      "loss": 0.4296,
      "step": 1141
    },
    {
      "epoch": 0.07153259524890775,
      "grad_norm": 0.8603565891570338,
      "learning_rate": 9.954836302013822e-06,
      "loss": 0.4578,
      "step": 1142
    },
    {
      "epoch": 0.07159523324825005,
      "grad_norm": 0.9302194647674883,
      "learning_rate": 9.95470016510304e-06,
      "loss": 0.4744,
      "step": 1143
    },
    {
      "epoch": 0.07165787124759235,
      "grad_norm": 0.9316150826999832,
      "learning_rate": 9.954563824255879e-06,
      "loss": 0.4912,
      "step": 1144
    },
    {
      "epoch": 0.07172050924693466,
      "grad_norm": 0.9718984299637353,
      "learning_rate": 9.954427279477951e-06,
      "loss": 0.4586,
      "step": 1145
    },
    {
      "epoch": 0.07178314724627695,
      "grad_norm": 0.990188305555676,
      "learning_rate": 9.954290530774877e-06,
      "loss": 0.4922,
      "step": 1146
    },
    {
      "epoch": 0.07184578524561926,
      "grad_norm": 0.9100260376288748,
      "learning_rate": 9.954153578152286e-06,
      "loss": 0.4557,
      "step": 1147
    },
    {
      "epoch": 0.07190842324496155,
      "grad_norm": 0.8663613440671539,
      "learning_rate": 9.954016421615813e-06,
      "loss": 0.462,
      "step": 1148
    },
    {
      "epoch": 0.07197106124430386,
      "grad_norm": 0.9142179382806003,
      "learning_rate": 9.953879061171106e-06,
      "loss": 0.4562,
      "step": 1149
    },
    {
      "epoch": 0.07203369924364615,
      "grad_norm": 0.9397616857576695,
      "learning_rate": 9.953741496823817e-06,
      "loss": 0.4575,
      "step": 1150
    },
    {
      "epoch": 0.07209633724298846,
      "grad_norm": 0.913691245532332,
      "learning_rate": 9.953603728579607e-06,
      "loss": 0.4117,
      "step": 1151
    },
    {
      "epoch": 0.07215897524233077,
      "grad_norm": 0.8825543135902681,
      "learning_rate": 9.953465756444149e-06,
      "loss": 0.4309,
      "step": 1152
    },
    {
      "epoch": 0.07222161324167306,
      "grad_norm": 0.8923141296343525,
      "learning_rate": 9.95332758042312e-06,
      "loss": 0.4331,
      "step": 1153
    },
    {
      "epoch": 0.07228425124101537,
      "grad_norm": 0.9238709178108122,
      "learning_rate": 9.95318920052221e-06,
      "loss": 0.4363,
      "step": 1154
    },
    {
      "epoch": 0.07234688924035766,
      "grad_norm": 0.9459094201711958,
      "learning_rate": 9.953050616747113e-06,
      "loss": 0.4498,
      "step": 1155
    },
    {
      "epoch": 0.07240952723969997,
      "grad_norm": 0.9396152795557088,
      "learning_rate": 9.952911829103532e-06,
      "loss": 0.5072,
      "step": 1156
    },
    {
      "epoch": 0.07247216523904226,
      "grad_norm": 0.8390940859911431,
      "learning_rate": 9.952772837597184e-06,
      "loss": 0.4305,
      "step": 1157
    },
    {
      "epoch": 0.07253480323838457,
      "grad_norm": 0.9714988848845277,
      "learning_rate": 9.952633642233785e-06,
      "loss": 0.5019,
      "step": 1158
    },
    {
      "epoch": 0.07259744123772686,
      "grad_norm": 0.9698975596699668,
      "learning_rate": 9.952494243019064e-06,
      "loss": 0.4481,
      "step": 1159
    },
    {
      "epoch": 0.07266007923706917,
      "grad_norm": 0.998845851989552,
      "learning_rate": 9.952354639958763e-06,
      "loss": 0.4691,
      "step": 1160
    },
    {
      "epoch": 0.07272271723641147,
      "grad_norm": 0.8250987923827157,
      "learning_rate": 9.952214833058624e-06,
      "loss": 0.5005,
      "step": 1161
    },
    {
      "epoch": 0.07278535523575377,
      "grad_norm": 0.9587165557526699,
      "learning_rate": 9.952074822324405e-06,
      "loss": 0.4952,
      "step": 1162
    },
    {
      "epoch": 0.07284799323509607,
      "grad_norm": 0.9679028423552817,
      "learning_rate": 9.951934607761866e-06,
      "loss": 0.447,
      "step": 1163
    },
    {
      "epoch": 0.07291063123443838,
      "grad_norm": 0.8418214590437509,
      "learning_rate": 9.95179418937678e-06,
      "loss": 0.4726,
      "step": 1164
    },
    {
      "epoch": 0.07297326923378067,
      "grad_norm": 0.871812113618552,
      "learning_rate": 9.951653567174926e-06,
      "loss": 0.439,
      "step": 1165
    },
    {
      "epoch": 0.07303590723312298,
      "grad_norm": 1.0064238103868552,
      "learning_rate": 9.951512741162093e-06,
      "loss": 0.4579,
      "step": 1166
    },
    {
      "epoch": 0.07309854523246527,
      "grad_norm": 1.0404181207242498,
      "learning_rate": 9.951371711344076e-06,
      "loss": 0.47,
      "step": 1167
    },
    {
      "epoch": 0.07316118323180758,
      "grad_norm": 0.8798062741709393,
      "learning_rate": 9.951230477726678e-06,
      "loss": 0.4195,
      "step": 1168
    },
    {
      "epoch": 0.07322382123114987,
      "grad_norm": 0.8846510464585108,
      "learning_rate": 9.951089040315718e-06,
      "loss": 0.4369,
      "step": 1169
    },
    {
      "epoch": 0.07328645923049218,
      "grad_norm": 0.7998512183174141,
      "learning_rate": 9.950947399117012e-06,
      "loss": 0.4409,
      "step": 1170
    },
    {
      "epoch": 0.07334909722983447,
      "grad_norm": 0.9490126707161151,
      "learning_rate": 9.950805554136393e-06,
      "loss": 0.4449,
      "step": 1171
    },
    {
      "epoch": 0.07341173522917678,
      "grad_norm": 0.9469023842744038,
      "learning_rate": 9.950663505379699e-06,
      "loss": 0.4668,
      "step": 1172
    },
    {
      "epoch": 0.07347437322851907,
      "grad_norm": 0.9921476776184577,
      "learning_rate": 9.950521252852775e-06,
      "loss": 0.4653,
      "step": 1173
    },
    {
      "epoch": 0.07353701122786138,
      "grad_norm": 0.8606166100754216,
      "learning_rate": 9.950378796561478e-06,
      "loss": 0.4442,
      "step": 1174
    },
    {
      "epoch": 0.07359964922720369,
      "grad_norm": 1.001117763064365,
      "learning_rate": 9.95023613651167e-06,
      "loss": 0.4428,
      "step": 1175
    },
    {
      "epoch": 0.07366228722654598,
      "grad_norm": 0.894424089470169,
      "learning_rate": 9.950093272709224e-06,
      "loss": 0.4829,
      "step": 1176
    },
    {
      "epoch": 0.07372492522588829,
      "grad_norm": 0.9090861431087188,
      "learning_rate": 9.949950205160022e-06,
      "loss": 0.4565,
      "step": 1177
    },
    {
      "epoch": 0.07378756322523058,
      "grad_norm": 0.9723573552227865,
      "learning_rate": 9.949806933869949e-06,
      "loss": 0.5018,
      "step": 1178
    },
    {
      "epoch": 0.07385020122457289,
      "grad_norm": 0.9381742406911614,
      "learning_rate": 9.949663458844904e-06,
      "loss": 0.433,
      "step": 1179
    },
    {
      "epoch": 0.07391283922391519,
      "grad_norm": 1.0242736448952792,
      "learning_rate": 9.949519780090792e-06,
      "loss": 0.4634,
      "step": 1180
    },
    {
      "epoch": 0.0739754772232575,
      "grad_norm": 0.8610151495552374,
      "learning_rate": 9.94937589761353e-06,
      "loss": 0.4226,
      "step": 1181
    },
    {
      "epoch": 0.07403811522259979,
      "grad_norm": 0.8747923216137132,
      "learning_rate": 9.949231811419036e-06,
      "loss": 0.4745,
      "step": 1182
    },
    {
      "epoch": 0.0741007532219421,
      "grad_norm": 0.9318842555194711,
      "learning_rate": 9.94908752151324e-06,
      "loss": 0.4799,
      "step": 1183
    },
    {
      "epoch": 0.07416339122128439,
      "grad_norm": 0.9225389818245735,
      "learning_rate": 9.948943027902085e-06,
      "loss": 0.4577,
      "step": 1184
    },
    {
      "epoch": 0.0742260292206267,
      "grad_norm": 0.8808203161591768,
      "learning_rate": 9.948798330591516e-06,
      "loss": 0.3909,
      "step": 1185
    },
    {
      "epoch": 0.07428866721996899,
      "grad_norm": 0.8694371232544301,
      "learning_rate": 9.948653429587487e-06,
      "loss": 0.4351,
      "step": 1186
    },
    {
      "epoch": 0.0743513052193113,
      "grad_norm": 0.8385342897620334,
      "learning_rate": 9.948508324895967e-06,
      "loss": 0.4181,
      "step": 1187
    },
    {
      "epoch": 0.07441394321865359,
      "grad_norm": 0.9709958983575211,
      "learning_rate": 9.948363016522926e-06,
      "loss": 0.4157,
      "step": 1188
    },
    {
      "epoch": 0.0744765812179959,
      "grad_norm": 0.8604417110685101,
      "learning_rate": 9.948217504474343e-06,
      "loss": 0.4138,
      "step": 1189
    },
    {
      "epoch": 0.07453921921733819,
      "grad_norm": 0.900180203594728,
      "learning_rate": 9.948071788756211e-06,
      "loss": 0.4126,
      "step": 1190
    },
    {
      "epoch": 0.0746018572166805,
      "grad_norm": 0.8048510443890734,
      "learning_rate": 9.947925869374524e-06,
      "loss": 0.4343,
      "step": 1191
    },
    {
      "epoch": 0.0746644952160228,
      "grad_norm": 0.8686742237974772,
      "learning_rate": 9.947779746335291e-06,
      "loss": 0.4321,
      "step": 1192
    },
    {
      "epoch": 0.0747271332153651,
      "grad_norm": 0.936783405606959,
      "learning_rate": 9.947633419644525e-06,
      "loss": 0.4558,
      "step": 1193
    },
    {
      "epoch": 0.0747897712147074,
      "grad_norm": 0.9124320706529733,
      "learning_rate": 9.947486889308249e-06,
      "loss": 0.4546,
      "step": 1194
    },
    {
      "epoch": 0.0748524092140497,
      "grad_norm": 0.9117615591529733,
      "learning_rate": 9.947340155332495e-06,
      "loss": 0.4121,
      "step": 1195
    },
    {
      "epoch": 0.07491504721339201,
      "grad_norm": 0.9744141301753017,
      "learning_rate": 9.9471932177233e-06,
      "loss": 0.4412,
      "step": 1196
    },
    {
      "epoch": 0.0749776852127343,
      "grad_norm": 1.0059087199773467,
      "learning_rate": 9.947046076486715e-06,
      "loss": 0.4682,
      "step": 1197
    },
    {
      "epoch": 0.07504032321207661,
      "grad_norm": 0.8194743559016224,
      "learning_rate": 9.946898731628795e-06,
      "loss": 0.4458,
      "step": 1198
    },
    {
      "epoch": 0.0751029612114189,
      "grad_norm": 0.9752424015351329,
      "learning_rate": 9.946751183155607e-06,
      "loss": 0.4375,
      "step": 1199
    },
    {
      "epoch": 0.07516559921076121,
      "grad_norm": 0.9549432557534264,
      "learning_rate": 9.946603431073221e-06,
      "loss": 0.4359,
      "step": 1200
    },
    {
      "epoch": 0.07522823721010351,
      "grad_norm": 0.9941861250100175,
      "learning_rate": 9.946455475387717e-06,
      "loss": 0.4687,
      "step": 1201
    },
    {
      "epoch": 0.07529087520944582,
      "grad_norm": 0.9022724180212576,
      "learning_rate": 9.94630731610519e-06,
      "loss": 0.4194,
      "step": 1202
    },
    {
      "epoch": 0.07535351320878811,
      "grad_norm": 0.8446332738052422,
      "learning_rate": 9.946158953231737e-06,
      "loss": 0.4484,
      "step": 1203
    },
    {
      "epoch": 0.07541615120813042,
      "grad_norm": 0.7636256876765098,
      "learning_rate": 9.946010386773462e-06,
      "loss": 0.5083,
      "step": 1204
    },
    {
      "epoch": 0.07547878920747271,
      "grad_norm": 0.9112928294176557,
      "learning_rate": 9.94586161673648e-06,
      "loss": 0.4392,
      "step": 1205
    },
    {
      "epoch": 0.07554142720681502,
      "grad_norm": 0.8813121173335156,
      "learning_rate": 9.945712643126919e-06,
      "loss": 0.421,
      "step": 1206
    },
    {
      "epoch": 0.07560406520615731,
      "grad_norm": 0.9123075791013374,
      "learning_rate": 9.945563465950905e-06,
      "loss": 0.4364,
      "step": 1207
    },
    {
      "epoch": 0.07566670320549962,
      "grad_norm": 0.8805298541638007,
      "learning_rate": 9.945414085214582e-06,
      "loss": 0.4412,
      "step": 1208
    },
    {
      "epoch": 0.07572934120484191,
      "grad_norm": 0.8699009932271607,
      "learning_rate": 9.945264500924097e-06,
      "loss": 0.4373,
      "step": 1209
    },
    {
      "epoch": 0.07579197920418422,
      "grad_norm": 0.943685212097875,
      "learning_rate": 9.945114713085607e-06,
      "loss": 0.4486,
      "step": 1210
    },
    {
      "epoch": 0.07585461720352651,
      "grad_norm": 0.8801967481923897,
      "learning_rate": 9.944964721705278e-06,
      "loss": 0.3975,
      "step": 1211
    },
    {
      "epoch": 0.07591725520286882,
      "grad_norm": 0.8223723930928067,
      "learning_rate": 9.944814526789284e-06,
      "loss": 0.477,
      "step": 1212
    },
    {
      "epoch": 0.07597989320221112,
      "grad_norm": 0.8875098819083301,
      "learning_rate": 9.944664128343806e-06,
      "loss": 0.4286,
      "step": 1213
    },
    {
      "epoch": 0.07604253120155342,
      "grad_norm": 0.8960322748018014,
      "learning_rate": 9.944513526375032e-06,
      "loss": 0.455,
      "step": 1214
    },
    {
      "epoch": 0.07610516920089572,
      "grad_norm": 0.9350673298224462,
      "learning_rate": 9.944362720889167e-06,
      "loss": 0.4672,
      "step": 1215
    },
    {
      "epoch": 0.07616780720023802,
      "grad_norm": 0.9466788065437881,
      "learning_rate": 9.944211711892414e-06,
      "loss": 0.4503,
      "step": 1216
    },
    {
      "epoch": 0.07623044519958033,
      "grad_norm": 0.9328923339335707,
      "learning_rate": 9.944060499390989e-06,
      "loss": 0.4613,
      "step": 1217
    },
    {
      "epoch": 0.07629308319892263,
      "grad_norm": 0.9121181139680061,
      "learning_rate": 9.943909083391115e-06,
      "loss": 0.4205,
      "step": 1218
    },
    {
      "epoch": 0.07635572119826493,
      "grad_norm": 0.8687352844174967,
      "learning_rate": 9.943757463899027e-06,
      "loss": 0.4938,
      "step": 1219
    },
    {
      "epoch": 0.07641835919760723,
      "grad_norm": 0.9719311931696941,
      "learning_rate": 9.943605640920962e-06,
      "loss": 0.457,
      "step": 1220
    },
    {
      "epoch": 0.07648099719694954,
      "grad_norm": 0.9217159937198494,
      "learning_rate": 9.943453614463174e-06,
      "loss": 0.4905,
      "step": 1221
    },
    {
      "epoch": 0.07654363519629183,
      "grad_norm": 0.9358187811657068,
      "learning_rate": 9.943301384531916e-06,
      "loss": 0.4387,
      "step": 1222
    },
    {
      "epoch": 0.07660627319563414,
      "grad_norm": 0.917428881965497,
      "learning_rate": 9.943148951133455e-06,
      "loss": 0.4739,
      "step": 1223
    },
    {
      "epoch": 0.07666891119497643,
      "grad_norm": 0.9622594182104627,
      "learning_rate": 9.942996314274067e-06,
      "loss": 0.455,
      "step": 1224
    },
    {
      "epoch": 0.07673154919431874,
      "grad_norm": 0.9184884607029525,
      "learning_rate": 9.942843473960033e-06,
      "loss": 0.4587,
      "step": 1225
    },
    {
      "epoch": 0.07679418719366103,
      "grad_norm": 0.9332928338547054,
      "learning_rate": 9.942690430197644e-06,
      "loss": 0.4077,
      "step": 1226
    },
    {
      "epoch": 0.07685682519300334,
      "grad_norm": 0.9500450505788541,
      "learning_rate": 9.9425371829932e-06,
      "loss": 0.4284,
      "step": 1227
    },
    {
      "epoch": 0.07691946319234563,
      "grad_norm": 0.814791174051221,
      "learning_rate": 9.942383732353007e-06,
      "loss": 0.4808,
      "step": 1228
    },
    {
      "epoch": 0.07698210119168794,
      "grad_norm": 0.987964401462945,
      "learning_rate": 9.942230078283383e-06,
      "loss": 0.465,
      "step": 1229
    },
    {
      "epoch": 0.07704473919103023,
      "grad_norm": 0.9806123709421981,
      "learning_rate": 9.94207622079065e-06,
      "loss": 0.4017,
      "step": 1230
    },
    {
      "epoch": 0.07710737719037254,
      "grad_norm": 0.868216845376054,
      "learning_rate": 9.941922159881146e-06,
      "loss": 0.441,
      "step": 1231
    },
    {
      "epoch": 0.07717001518971484,
      "grad_norm": 0.9290924344040146,
      "learning_rate": 9.941767895561205e-06,
      "loss": 0.4347,
      "step": 1232
    },
    {
      "epoch": 0.07723265318905714,
      "grad_norm": 0.761634539050343,
      "learning_rate": 9.941613427837182e-06,
      "loss": 0.4793,
      "step": 1233
    },
    {
      "epoch": 0.07729529118839944,
      "grad_norm": 0.9008101631017644,
      "learning_rate": 9.941458756715433e-06,
      "loss": 0.4637,
      "step": 1234
    },
    {
      "epoch": 0.07735792918774174,
      "grad_norm": 0.8718394908016194,
      "learning_rate": 9.941303882202325e-06,
      "loss": 0.4229,
      "step": 1235
    },
    {
      "epoch": 0.07742056718708404,
      "grad_norm": 0.9839948034352733,
      "learning_rate": 9.94114880430423e-06,
      "loss": 0.4419,
      "step": 1236
    },
    {
      "epoch": 0.07748320518642635,
      "grad_norm": 0.8509985578547442,
      "learning_rate": 9.940993523027535e-06,
      "loss": 0.4221,
      "step": 1237
    },
    {
      "epoch": 0.07754584318576864,
      "grad_norm": 0.8655700800330552,
      "learning_rate": 9.940838038378628e-06,
      "loss": 0.4801,
      "step": 1238
    },
    {
      "epoch": 0.07760848118511095,
      "grad_norm": 0.9774996127021218,
      "learning_rate": 9.940682350363913e-06,
      "loss": 0.4794,
      "step": 1239
    },
    {
      "epoch": 0.07767111918445326,
      "grad_norm": 0.9627491963876326,
      "learning_rate": 9.940526458989794e-06,
      "loss": 0.4618,
      "step": 1240
    },
    {
      "epoch": 0.07773375718379555,
      "grad_norm": 0.8609968737583685,
      "learning_rate": 9.940370364262689e-06,
      "loss": 0.4305,
      "step": 1241
    },
    {
      "epoch": 0.07779639518313786,
      "grad_norm": 0.8664149335450184,
      "learning_rate": 9.940214066189021e-06,
      "loss": 0.4321,
      "step": 1242
    },
    {
      "epoch": 0.07785903318248015,
      "grad_norm": 0.8928766383367529,
      "learning_rate": 9.940057564775225e-06,
      "loss": 0.4621,
      "step": 1243
    },
    {
      "epoch": 0.07792167118182246,
      "grad_norm": 0.8795047640106914,
      "learning_rate": 9.939900860027745e-06,
      "loss": 0.4803,
      "step": 1244
    },
    {
      "epoch": 0.07798430918116475,
      "grad_norm": 0.9309412281283268,
      "learning_rate": 9.93974395195303e-06,
      "loss": 0.4722,
      "step": 1245
    },
    {
      "epoch": 0.07804694718050706,
      "grad_norm": 0.800948587336539,
      "learning_rate": 9.939586840557533e-06,
      "loss": 0.4114,
      "step": 1246
    },
    {
      "epoch": 0.07810958517984935,
      "grad_norm": 0.906322814345792,
      "learning_rate": 9.939429525847728e-06,
      "loss": 0.4435,
      "step": 1247
    },
    {
      "epoch": 0.07817222317919166,
      "grad_norm": 0.9269134628177128,
      "learning_rate": 9.939272007830086e-06,
      "loss": 0.4992,
      "step": 1248
    },
    {
      "epoch": 0.07823486117853395,
      "grad_norm": 0.8863414158141384,
      "learning_rate": 9.93911428651109e-06,
      "loss": 0.4422,
      "step": 1249
    },
    {
      "epoch": 0.07829749917787626,
      "grad_norm": 0.8877966588323426,
      "learning_rate": 9.938956361897235e-06,
      "loss": 0.4551,
      "step": 1250
    },
    {
      "epoch": 0.07836013717721856,
      "grad_norm": 0.87633714189077,
      "learning_rate": 9.938798233995018e-06,
      "loss": 0.4741,
      "step": 1251
    },
    {
      "epoch": 0.07842277517656086,
      "grad_norm": 0.8763069140427436,
      "learning_rate": 9.938639902810951e-06,
      "loss": 0.4299,
      "step": 1252
    },
    {
      "epoch": 0.07848541317590316,
      "grad_norm": 0.8774112362888605,
      "learning_rate": 9.938481368351547e-06,
      "loss": 0.4309,
      "step": 1253
    },
    {
      "epoch": 0.07854805117524546,
      "grad_norm": 1.1502880938645852,
      "learning_rate": 9.938322630623334e-06,
      "loss": 0.3964,
      "step": 1254
    },
    {
      "epoch": 0.07861068917458776,
      "grad_norm": 0.8727638113279288,
      "learning_rate": 9.938163689632844e-06,
      "loss": 0.4306,
      "step": 1255
    },
    {
      "epoch": 0.07867332717393007,
      "grad_norm": 0.9779276797264517,
      "learning_rate": 9.938004545386622e-06,
      "loss": 0.4282,
      "step": 1256
    },
    {
      "epoch": 0.07873596517327236,
      "grad_norm": 0.8820157332683263,
      "learning_rate": 9.937845197891215e-06,
      "loss": 0.398,
      "step": 1257
    },
    {
      "epoch": 0.07879860317261467,
      "grad_norm": 0.9529335465898677,
      "learning_rate": 9.937685647153184e-06,
      "loss": 0.4527,
      "step": 1258
    },
    {
      "epoch": 0.07886124117195696,
      "grad_norm": 0.9581635974408398,
      "learning_rate": 9.937525893179095e-06,
      "loss": 0.4644,
      "step": 1259
    },
    {
      "epoch": 0.07892387917129927,
      "grad_norm": 0.9282909793018385,
      "learning_rate": 9.937365935975523e-06,
      "loss": 0.4515,
      "step": 1260
    },
    {
      "epoch": 0.07898651717064158,
      "grad_norm": 0.9965946658055406,
      "learning_rate": 9.937205775549052e-06,
      "loss": 0.5241,
      "step": 1261
    },
    {
      "epoch": 0.07904915516998387,
      "grad_norm": 0.8895398121273224,
      "learning_rate": 9.937045411906276e-06,
      "loss": 0.4938,
      "step": 1262
    },
    {
      "epoch": 0.07911179316932618,
      "grad_norm": 0.927802947928921,
      "learning_rate": 9.936884845053796e-06,
      "loss": 0.4549,
      "step": 1263
    },
    {
      "epoch": 0.07917443116866847,
      "grad_norm": 0.9068969745148321,
      "learning_rate": 9.936724074998216e-06,
      "loss": 0.4416,
      "step": 1264
    },
    {
      "epoch": 0.07923706916801078,
      "grad_norm": 0.9291953169177368,
      "learning_rate": 9.93656310174616e-06,
      "loss": 0.4261,
      "step": 1265
    },
    {
      "epoch": 0.07929970716735307,
      "grad_norm": 0.8863548289019633,
      "learning_rate": 9.93640192530425e-06,
      "loss": 0.4519,
      "step": 1266
    },
    {
      "epoch": 0.07936234516669538,
      "grad_norm": 0.9249054663138242,
      "learning_rate": 9.936240545679119e-06,
      "loss": 0.4502,
      "step": 1267
    },
    {
      "epoch": 0.07942498316603767,
      "grad_norm": 0.8953713920354981,
      "learning_rate": 9.936078962877411e-06,
      "loss": 0.4572,
      "step": 1268
    },
    {
      "epoch": 0.07948762116537998,
      "grad_norm": 0.9130688534532823,
      "learning_rate": 9.935917176905778e-06,
      "loss": 0.4346,
      "step": 1269
    },
    {
      "epoch": 0.07955025916472228,
      "grad_norm": 0.8785494083673833,
      "learning_rate": 9.935755187770876e-06,
      "loss": 0.4787,
      "step": 1270
    },
    {
      "epoch": 0.07961289716406458,
      "grad_norm": 0.9881268488793725,
      "learning_rate": 9.935592995479376e-06,
      "loss": 0.4542,
      "step": 1271
    },
    {
      "epoch": 0.07967553516340688,
      "grad_norm": 0.885360927274521,
      "learning_rate": 9.935430600037952e-06,
      "loss": 0.413,
      "step": 1272
    },
    {
      "epoch": 0.07973817316274918,
      "grad_norm": 0.9006873165108479,
      "learning_rate": 9.935268001453289e-06,
      "loss": 0.4212,
      "step": 1273
    },
    {
      "epoch": 0.07980081116209148,
      "grad_norm": 0.900047285868892,
      "learning_rate": 9.935105199732078e-06,
      "loss": 0.3857,
      "step": 1274
    },
    {
      "epoch": 0.07986344916143379,
      "grad_norm": 0.8897202616806049,
      "learning_rate": 9.934942194881021e-06,
      "loss": 0.465,
      "step": 1275
    },
    {
      "epoch": 0.07992608716077608,
      "grad_norm": 0.9246008928186844,
      "learning_rate": 9.934778986906828e-06,
      "loss": 0.449,
      "step": 1276
    },
    {
      "epoch": 0.07998872516011839,
      "grad_norm": 0.9131053405363146,
      "learning_rate": 9.934615575816215e-06,
      "loss": 0.4119,
      "step": 1277
    },
    {
      "epoch": 0.08005136315946068,
      "grad_norm": 1.0304607358113678,
      "learning_rate": 9.934451961615909e-06,
      "loss": 0.4719,
      "step": 1278
    },
    {
      "epoch": 0.08011400115880299,
      "grad_norm": 0.9664580827729918,
      "learning_rate": 9.934288144312645e-06,
      "loss": 0.4717,
      "step": 1279
    },
    {
      "epoch": 0.08017663915814528,
      "grad_norm": 0.9119044109810605,
      "learning_rate": 9.934124123913164e-06,
      "loss": 0.4421,
      "step": 1280
    },
    {
      "epoch": 0.08023927715748759,
      "grad_norm": 0.9049418121460919,
      "learning_rate": 9.93395990042422e-06,
      "loss": 0.4213,
      "step": 1281
    },
    {
      "epoch": 0.0803019151568299,
      "grad_norm": 0.9481312747839066,
      "learning_rate": 9.93379547385257e-06,
      "loss": 0.429,
      "step": 1282
    },
    {
      "epoch": 0.08036455315617219,
      "grad_norm": 0.8311396097185462,
      "learning_rate": 9.933630844204982e-06,
      "loss": 0.4728,
      "step": 1283
    },
    {
      "epoch": 0.0804271911555145,
      "grad_norm": 0.9587217944843172,
      "learning_rate": 9.933466011488233e-06,
      "loss": 0.4413,
      "step": 1284
    },
    {
      "epoch": 0.08048982915485679,
      "grad_norm": 0.9248487172111027,
      "learning_rate": 9.933300975709109e-06,
      "loss": 0.4306,
      "step": 1285
    },
    {
      "epoch": 0.0805524671541991,
      "grad_norm": 0.9379860293473867,
      "learning_rate": 9.933135736874399e-06,
      "loss": 0.4703,
      "step": 1286
    },
    {
      "epoch": 0.0806151051535414,
      "grad_norm": 0.9893959058113893,
      "learning_rate": 9.932970294990907e-06,
      "loss": 0.4706,
      "step": 1287
    },
    {
      "epoch": 0.0806777431528837,
      "grad_norm": 0.9474194553271569,
      "learning_rate": 9.932804650065442e-06,
      "loss": 0.4774,
      "step": 1288
    },
    {
      "epoch": 0.080740381152226,
      "grad_norm": 0.9244822708182773,
      "learning_rate": 9.932638802104822e-06,
      "loss": 0.4958,
      "step": 1289
    },
    {
      "epoch": 0.0808030191515683,
      "grad_norm": 0.7866499122523637,
      "learning_rate": 9.932472751115873e-06,
      "loss": 0.4303,
      "step": 1290
    },
    {
      "epoch": 0.0808656571509106,
      "grad_norm": 0.8189372240258482,
      "learning_rate": 9.93230649710543e-06,
      "loss": 0.404,
      "step": 1291
    },
    {
      "epoch": 0.0809282951502529,
      "grad_norm": 0.9481294493303128,
      "learning_rate": 9.932140040080337e-06,
      "loss": 0.4409,
      "step": 1292
    },
    {
      "epoch": 0.0809909331495952,
      "grad_norm": 0.9847854211311461,
      "learning_rate": 9.931973380047446e-06,
      "loss": 0.4796,
      "step": 1293
    },
    {
      "epoch": 0.0810535711489375,
      "grad_norm": 0.8757564692839778,
      "learning_rate": 9.931806517013612e-06,
      "loss": 0.4291,
      "step": 1294
    },
    {
      "epoch": 0.0811162091482798,
      "grad_norm": 0.786392225815852,
      "learning_rate": 9.931639450985708e-06,
      "loss": 0.386,
      "step": 1295
    },
    {
      "epoch": 0.08117884714762211,
      "grad_norm": 0.882097348131154,
      "learning_rate": 9.93147218197061e-06,
      "loss": 0.4324,
      "step": 1296
    },
    {
      "epoch": 0.0812414851469644,
      "grad_norm": 0.7867297251863316,
      "learning_rate": 9.9313047099752e-06,
      "loss": 0.4454,
      "step": 1297
    },
    {
      "epoch": 0.08130412314630671,
      "grad_norm": 0.8409534747623244,
      "learning_rate": 9.931137035006373e-06,
      "loss": 0.4469,
      "step": 1298
    },
    {
      "epoch": 0.081366761145649,
      "grad_norm": 0.9400948533190642,
      "learning_rate": 9.93096915707103e-06,
      "loss": 0.4318,
      "step": 1299
    },
    {
      "epoch": 0.08142939914499131,
      "grad_norm": 0.9088292303879242,
      "learning_rate": 9.930801076176083e-06,
      "loss": 0.4493,
      "step": 1300
    },
    {
      "epoch": 0.0814920371443336,
      "grad_norm": 0.9081349800162409,
      "learning_rate": 9.930632792328446e-06,
      "loss": 0.4276,
      "step": 1301
    },
    {
      "epoch": 0.08155467514367591,
      "grad_norm": 0.9634553241936091,
      "learning_rate": 9.930464305535051e-06,
      "loss": 0.4517,
      "step": 1302
    },
    {
      "epoch": 0.0816173131430182,
      "grad_norm": 0.8289762816688031,
      "learning_rate": 9.930295615802827e-06,
      "loss": 0.3868,
      "step": 1303
    },
    {
      "epoch": 0.08167995114236051,
      "grad_norm": 0.8967478684707443,
      "learning_rate": 9.930126723138722e-06,
      "loss": 0.4571,
      "step": 1304
    },
    {
      "epoch": 0.08174258914170282,
      "grad_norm": 0.8963523149097937,
      "learning_rate": 9.929957627549686e-06,
      "loss": 0.4417,
      "step": 1305
    },
    {
      "epoch": 0.08180522714104511,
      "grad_norm": 0.9101153980554354,
      "learning_rate": 9.929788329042679e-06,
      "loss": 0.4509,
      "step": 1306
    },
    {
      "epoch": 0.08186786514038742,
      "grad_norm": 0.8317269853181682,
      "learning_rate": 9.92961882762467e-06,
      "loss": 0.4175,
      "step": 1307
    },
    {
      "epoch": 0.08193050313972972,
      "grad_norm": 0.8959554798876624,
      "learning_rate": 9.929449123302633e-06,
      "loss": 0.4938,
      "step": 1308
    },
    {
      "epoch": 0.08199314113907202,
      "grad_norm": 0.8904192439558701,
      "learning_rate": 9.929279216083556e-06,
      "loss": 0.4159,
      "step": 1309
    },
    {
      "epoch": 0.08205577913841432,
      "grad_norm": 0.9513386395678146,
      "learning_rate": 9.929109105974432e-06,
      "loss": 0.4604,
      "step": 1310
    },
    {
      "epoch": 0.08211841713775662,
      "grad_norm": 0.9133337279862005,
      "learning_rate": 9.928938792982263e-06,
      "loss": 0.4739,
      "step": 1311
    },
    {
      "epoch": 0.08218105513709892,
      "grad_norm": 0.9067196757715565,
      "learning_rate": 9.928768277114059e-06,
      "loss": 0.474,
      "step": 1312
    },
    {
      "epoch": 0.08224369313644123,
      "grad_norm": 0.8665390477708504,
      "learning_rate": 9.928597558376836e-06,
      "loss": 0.4165,
      "step": 1313
    },
    {
      "epoch": 0.08230633113578352,
      "grad_norm": 0.8480357180498844,
      "learning_rate": 9.928426636777622e-06,
      "loss": 0.4452,
      "step": 1314
    },
    {
      "epoch": 0.08236896913512583,
      "grad_norm": 0.9529430768707674,
      "learning_rate": 9.928255512323455e-06,
      "loss": 0.4561,
      "step": 1315
    },
    {
      "epoch": 0.08243160713446812,
      "grad_norm": 0.8525850547021465,
      "learning_rate": 9.928084185021374e-06,
      "loss": 0.3786,
      "step": 1316
    },
    {
      "epoch": 0.08249424513381043,
      "grad_norm": 0.8727674759196122,
      "learning_rate": 9.927912654878436e-06,
      "loss": 0.4265,
      "step": 1317
    },
    {
      "epoch": 0.08255688313315272,
      "grad_norm": 0.8859945258540228,
      "learning_rate": 9.927740921901696e-06,
      "loss": 0.435,
      "step": 1318
    },
    {
      "epoch": 0.08261952113249503,
      "grad_norm": 0.8579271657598861,
      "learning_rate": 9.927568986098227e-06,
      "loss": 0.4167,
      "step": 1319
    },
    {
      "epoch": 0.08268215913183732,
      "grad_norm": 0.9209203133023071,
      "learning_rate": 9.927396847475103e-06,
      "loss": 0.4008,
      "step": 1320
    },
    {
      "epoch": 0.08274479713117963,
      "grad_norm": 0.950579500825011,
      "learning_rate": 9.92722450603941e-06,
      "loss": 0.4381,
      "step": 1321
    },
    {
      "epoch": 0.08280743513052193,
      "grad_norm": 0.8841469672713104,
      "learning_rate": 9.927051961798244e-06,
      "loss": 0.44,
      "step": 1322
    },
    {
      "epoch": 0.08287007312986423,
      "grad_norm": 0.9776301900190234,
      "learning_rate": 9.926879214758703e-06,
      "loss": 0.4132,
      "step": 1323
    },
    {
      "epoch": 0.08293271112920653,
      "grad_norm": 0.8789002799842064,
      "learning_rate": 9.926706264927898e-06,
      "loss": 0.445,
      "step": 1324
    },
    {
      "epoch": 0.08299534912854883,
      "grad_norm": 0.8238934117054088,
      "learning_rate": 9.92653311231295e-06,
      "loss": 0.3836,
      "step": 1325
    },
    {
      "epoch": 0.08305798712789114,
      "grad_norm": 0.8682642135144991,
      "learning_rate": 9.926359756920985e-06,
      "loss": 0.4378,
      "step": 1326
    },
    {
      "epoch": 0.08312062512723344,
      "grad_norm": 0.9400329264055638,
      "learning_rate": 9.926186198759136e-06,
      "loss": 0.4377,
      "step": 1327
    },
    {
      "epoch": 0.08318326312657574,
      "grad_norm": 0.8131060153557612,
      "learning_rate": 9.926012437834552e-06,
      "loss": 0.4829,
      "step": 1328
    },
    {
      "epoch": 0.08324590112591804,
      "grad_norm": 0.7975465088742627,
      "learning_rate": 9.92583847415438e-06,
      "loss": 0.4992,
      "step": 1329
    },
    {
      "epoch": 0.08330853912526034,
      "grad_norm": 0.9033009330790217,
      "learning_rate": 9.92566430772578e-06,
      "loss": 0.4689,
      "step": 1330
    },
    {
      "epoch": 0.08337117712460264,
      "grad_norm": 0.9395848150659001,
      "learning_rate": 9.925489938555926e-06,
      "loss": 0.4447,
      "step": 1331
    },
    {
      "epoch": 0.08343381512394495,
      "grad_norm": 0.8972050483590744,
      "learning_rate": 9.92531536665199e-06,
      "loss": 0.4537,
      "step": 1332
    },
    {
      "epoch": 0.08349645312328724,
      "grad_norm": 0.8545733599726286,
      "learning_rate": 9.925140592021161e-06,
      "loss": 0.4396,
      "step": 1333
    },
    {
      "epoch": 0.08355909112262955,
      "grad_norm": 0.8728196893996324,
      "learning_rate": 9.924965614670629e-06,
      "loss": 0.4726,
      "step": 1334
    },
    {
      "epoch": 0.08362172912197184,
      "grad_norm": 0.8394858913774375,
      "learning_rate": 9.924790434607599e-06,
      "loss": 0.4388,
      "step": 1335
    },
    {
      "epoch": 0.08368436712131415,
      "grad_norm": 0.8583505947935397,
      "learning_rate": 9.924615051839281e-06,
      "loss": 0.4035,
      "step": 1336
    },
    {
      "epoch": 0.08374700512065644,
      "grad_norm": 0.8982565530363668,
      "learning_rate": 9.924439466372894e-06,
      "loss": 0.4209,
      "step": 1337
    },
    {
      "epoch": 0.08380964311999875,
      "grad_norm": 0.9092155741048061,
      "learning_rate": 9.924263678215664e-06,
      "loss": 0.4382,
      "step": 1338
    },
    {
      "epoch": 0.08387228111934104,
      "grad_norm": 0.8799977148733958,
      "learning_rate": 9.924087687374826e-06,
      "loss": 0.4319,
      "step": 1339
    },
    {
      "epoch": 0.08393491911868335,
      "grad_norm": 0.8798112161212216,
      "learning_rate": 9.923911493857627e-06,
      "loss": 0.4663,
      "step": 1340
    },
    {
      "epoch": 0.08399755711802565,
      "grad_norm": 0.833742595524073,
      "learning_rate": 9.923735097671315e-06,
      "loss": 0.416,
      "step": 1341
    },
    {
      "epoch": 0.08406019511736795,
      "grad_norm": 0.8446554140721072,
      "learning_rate": 9.923558498823154e-06,
      "loss": 0.4383,
      "step": 1342
    },
    {
      "epoch": 0.08412283311671025,
      "grad_norm": 0.8598310260829837,
      "learning_rate": 9.92338169732041e-06,
      "loss": 0.4416,
      "step": 1343
    },
    {
      "epoch": 0.08418547111605255,
      "grad_norm": 0.9235150111263606,
      "learning_rate": 9.923204693170362e-06,
      "loss": 0.4812,
      "step": 1344
    },
    {
      "epoch": 0.08424810911539485,
      "grad_norm": 0.9270111754100334,
      "learning_rate": 9.923027486380296e-06,
      "loss": 0.5047,
      "step": 1345
    },
    {
      "epoch": 0.08431074711473716,
      "grad_norm": 0.8926653877662165,
      "learning_rate": 9.922850076957505e-06,
      "loss": 0.415,
      "step": 1346
    },
    {
      "epoch": 0.08437338511407945,
      "grad_norm": 0.8142186964379081,
      "learning_rate": 9.922672464909289e-06,
      "loss": 0.4081,
      "step": 1347
    },
    {
      "epoch": 0.08443602311342176,
      "grad_norm": 0.8869852971594008,
      "learning_rate": 9.922494650242962e-06,
      "loss": 0.455,
      "step": 1348
    },
    {
      "epoch": 0.08449866111276406,
      "grad_norm": 0.9369135446675806,
      "learning_rate": 9.922316632965841e-06,
      "loss": 0.4474,
      "step": 1349
    },
    {
      "epoch": 0.08456129911210636,
      "grad_norm": 0.8888503030737285,
      "learning_rate": 9.922138413085254e-06,
      "loss": 0.4679,
      "step": 1350
    },
    {
      "epoch": 0.08462393711144867,
      "grad_norm": 0.8820582315879134,
      "learning_rate": 9.921959990608536e-06,
      "loss": 0.382,
      "step": 1351
    },
    {
      "epoch": 0.08468657511079096,
      "grad_norm": 0.9652105790874645,
      "learning_rate": 9.921781365543033e-06,
      "loss": 0.4272,
      "step": 1352
    },
    {
      "epoch": 0.08474921311013327,
      "grad_norm": 0.8993888060598315,
      "learning_rate": 9.921602537896093e-06,
      "loss": 0.3956,
      "step": 1353
    },
    {
      "epoch": 0.08481185110947556,
      "grad_norm": 0.9249119667557419,
      "learning_rate": 9.921423507675081e-06,
      "loss": 0.4624,
      "step": 1354
    },
    {
      "epoch": 0.08487448910881787,
      "grad_norm": 0.8673504778088543,
      "learning_rate": 9.921244274887363e-06,
      "loss": 0.4481,
      "step": 1355
    },
    {
      "epoch": 0.08493712710816016,
      "grad_norm": 0.9904119354222318,
      "learning_rate": 9.921064839540318e-06,
      "loss": 0.4897,
      "step": 1356
    },
    {
      "epoch": 0.08499976510750247,
      "grad_norm": 0.9422069615993155,
      "learning_rate": 9.920885201641329e-06,
      "loss": 0.4228,
      "step": 1357
    },
    {
      "epoch": 0.08506240310684476,
      "grad_norm": 0.8877408695069506,
      "learning_rate": 9.920705361197793e-06,
      "loss": 0.4772,
      "step": 1358
    },
    {
      "epoch": 0.08512504110618707,
      "grad_norm": 0.885646440879506,
      "learning_rate": 9.920525318217111e-06,
      "loss": 0.4647,
      "step": 1359
    },
    {
      "epoch": 0.08518767910552937,
      "grad_norm": 0.875424777824192,
      "learning_rate": 9.920345072706695e-06,
      "loss": 0.4456,
      "step": 1360
    },
    {
      "epoch": 0.08525031710487167,
      "grad_norm": 0.9034189648162483,
      "learning_rate": 9.92016462467396e-06,
      "loss": 0.4691,
      "step": 1361
    },
    {
      "epoch": 0.08531295510421397,
      "grad_norm": 0.9726799929197013,
      "learning_rate": 9.919983974126336e-06,
      "loss": 0.4846,
      "step": 1362
    },
    {
      "epoch": 0.08537559310355627,
      "grad_norm": 0.9755037733795654,
      "learning_rate": 9.91980312107126e-06,
      "loss": 0.4397,
      "step": 1363
    },
    {
      "epoch": 0.08543823110289857,
      "grad_norm": 0.8138341896033927,
      "learning_rate": 9.919622065516173e-06,
      "loss": 0.405,
      "step": 1364
    },
    {
      "epoch": 0.08550086910224088,
      "grad_norm": 0.917193139745396,
      "learning_rate": 9.91944080746853e-06,
      "loss": 0.4345,
      "step": 1365
    },
    {
      "epoch": 0.08556350710158317,
      "grad_norm": 0.8377839301030301,
      "learning_rate": 9.91925934693579e-06,
      "loss": 0.4406,
      "step": 1366
    },
    {
      "epoch": 0.08562614510092548,
      "grad_norm": 0.8899823493585649,
      "learning_rate": 9.919077683925422e-06,
      "loss": 0.467,
      "step": 1367
    },
    {
      "epoch": 0.08568878310026777,
      "grad_norm": 0.8999683179078123,
      "learning_rate": 9.918895818444903e-06,
      "loss": 0.4971,
      "step": 1368
    },
    {
      "epoch": 0.08575142109961008,
      "grad_norm": 0.8940180727154063,
      "learning_rate": 9.91871375050172e-06,
      "loss": 0.4056,
      "step": 1369
    },
    {
      "epoch": 0.08581405909895239,
      "grad_norm": 0.9125008930517028,
      "learning_rate": 9.918531480103365e-06,
      "loss": 0.4538,
      "step": 1370
    },
    {
      "epoch": 0.08587669709829468,
      "grad_norm": 0.9183663677437494,
      "learning_rate": 9.918349007257341e-06,
      "loss": 0.4976,
      "step": 1371
    },
    {
      "epoch": 0.08593933509763699,
      "grad_norm": 0.8691737245020468,
      "learning_rate": 9.918166331971162e-06,
      "loss": 0.4454,
      "step": 1372
    },
    {
      "epoch": 0.08600197309697928,
      "grad_norm": 0.8946216830516855,
      "learning_rate": 9.91798345425234e-06,
      "loss": 0.4577,
      "step": 1373
    },
    {
      "epoch": 0.08606461109632159,
      "grad_norm": 0.9250315219000506,
      "learning_rate": 9.917800374108408e-06,
      "loss": 0.4001,
      "step": 1374
    },
    {
      "epoch": 0.08612724909566388,
      "grad_norm": 0.8183727896455611,
      "learning_rate": 9.9176170915469e-06,
      "loss": 0.4971,
      "step": 1375
    },
    {
      "epoch": 0.08618988709500619,
      "grad_norm": 0.9525050032681599,
      "learning_rate": 9.91743360657536e-06,
      "loss": 0.4912,
      "step": 1376
    },
    {
      "epoch": 0.08625252509434848,
      "grad_norm": 0.9217854526680355,
      "learning_rate": 9.91724991920134e-06,
      "loss": 0.4685,
      "step": 1377
    },
    {
      "epoch": 0.08631516309369079,
      "grad_norm": 0.8973605701124336,
      "learning_rate": 9.9170660294324e-06,
      "loss": 0.4683,
      "step": 1378
    },
    {
      "epoch": 0.08637780109303309,
      "grad_norm": 0.8564823704358066,
      "learning_rate": 9.91688193727611e-06,
      "loss": 0.5028,
      "step": 1379
    },
    {
      "epoch": 0.08644043909237539,
      "grad_norm": 0.9878513312830047,
      "learning_rate": 9.916697642740047e-06,
      "loss": 0.4667,
      "step": 1380
    },
    {
      "epoch": 0.08650307709171769,
      "grad_norm": 0.9783682440705049,
      "learning_rate": 9.916513145831797e-06,
      "loss": 0.4422,
      "step": 1381
    },
    {
      "epoch": 0.08656571509106,
      "grad_norm": 0.8592352311945686,
      "learning_rate": 9.916328446558951e-06,
      "loss": 0.4457,
      "step": 1382
    },
    {
      "epoch": 0.08662835309040229,
      "grad_norm": 0.8591974827790301,
      "learning_rate": 9.916143544929115e-06,
      "loss": 0.3973,
      "step": 1383
    },
    {
      "epoch": 0.0866909910897446,
      "grad_norm": 0.8699453794572631,
      "learning_rate": 9.915958440949899e-06,
      "loss": 0.455,
      "step": 1384
    },
    {
      "epoch": 0.08675362908908689,
      "grad_norm": 0.806697157483979,
      "learning_rate": 9.915773134628921e-06,
      "loss": 0.3977,
      "step": 1385
    },
    {
      "epoch": 0.0868162670884292,
      "grad_norm": 0.9189914250062717,
      "learning_rate": 9.915587625973807e-06,
      "loss": 0.4466,
      "step": 1386
    },
    {
      "epoch": 0.08687890508777149,
      "grad_norm": 0.8733466181604178,
      "learning_rate": 9.915401914992196e-06,
      "loss": 0.4361,
      "step": 1387
    },
    {
      "epoch": 0.0869415430871138,
      "grad_norm": 0.9079970862157909,
      "learning_rate": 9.915216001691728e-06,
      "loss": 0.456,
      "step": 1388
    },
    {
      "epoch": 0.08700418108645609,
      "grad_norm": 0.8694976093234688,
      "learning_rate": 9.915029886080058e-06,
      "loss": 0.4943,
      "step": 1389
    },
    {
      "epoch": 0.0870668190857984,
      "grad_norm": 0.9637066595052954,
      "learning_rate": 9.914843568164845e-06,
      "loss": 0.4647,
      "step": 1390
    },
    {
      "epoch": 0.08712945708514071,
      "grad_norm": 0.8263936324208578,
      "learning_rate": 9.914657047953758e-06,
      "loss": 0.4377,
      "step": 1391
    },
    {
      "epoch": 0.087192095084483,
      "grad_norm": 0.93944505450495,
      "learning_rate": 9.914470325454476e-06,
      "loss": 0.4696,
      "step": 1392
    },
    {
      "epoch": 0.08725473308382531,
      "grad_norm": 0.810240495384967,
      "learning_rate": 9.914283400674683e-06,
      "loss": 0.3892,
      "step": 1393
    },
    {
      "epoch": 0.0873173710831676,
      "grad_norm": 0.9182884733742914,
      "learning_rate": 9.914096273622072e-06,
      "loss": 0.4375,
      "step": 1394
    },
    {
      "epoch": 0.08738000908250991,
      "grad_norm": 0.9046982775831647,
      "learning_rate": 9.913908944304346e-06,
      "loss": 0.4347,
      "step": 1395
    },
    {
      "epoch": 0.0874426470818522,
      "grad_norm": 0.8632876319351713,
      "learning_rate": 9.913721412729218e-06,
      "loss": 0.4199,
      "step": 1396
    },
    {
      "epoch": 0.08750528508119451,
      "grad_norm": 0.9100003424736227,
      "learning_rate": 9.913533678904403e-06,
      "loss": 0.4757,
      "step": 1397
    },
    {
      "epoch": 0.0875679230805368,
      "grad_norm": 0.9096000087088054,
      "learning_rate": 9.91334574283763e-06,
      "loss": 0.4497,
      "step": 1398
    },
    {
      "epoch": 0.08763056107987911,
      "grad_norm": 0.7716644198701434,
      "learning_rate": 9.913157604536636e-06,
      "loss": 0.4921,
      "step": 1399
    },
    {
      "epoch": 0.0876931990792214,
      "grad_norm": 0.9124959333369587,
      "learning_rate": 9.912969264009162e-06,
      "loss": 0.4347,
      "step": 1400
    },
    {
      "epoch": 0.08775583707856371,
      "grad_norm": 0.9667904475252258,
      "learning_rate": 9.91278072126296e-06,
      "loss": 0.4828,
      "step": 1401
    },
    {
      "epoch": 0.08781847507790601,
      "grad_norm": 0.7857968307570329,
      "learning_rate": 9.912591976305794e-06,
      "loss": 0.5058,
      "step": 1402
    },
    {
      "epoch": 0.08788111307724832,
      "grad_norm": 0.8812932983947236,
      "learning_rate": 9.912403029145428e-06,
      "loss": 0.4393,
      "step": 1403
    },
    {
      "epoch": 0.08794375107659061,
      "grad_norm": 0.7834902756080955,
      "learning_rate": 9.912213879789643e-06,
      "loss": 0.4096,
      "step": 1404
    },
    {
      "epoch": 0.08800638907593292,
      "grad_norm": 0.9606828582098004,
      "learning_rate": 9.912024528246222e-06,
      "loss": 0.488,
      "step": 1405
    },
    {
      "epoch": 0.08806902707527521,
      "grad_norm": 0.9592104482722726,
      "learning_rate": 9.91183497452296e-06,
      "loss": 0.4315,
      "step": 1406
    },
    {
      "epoch": 0.08813166507461752,
      "grad_norm": 0.8799191410829605,
      "learning_rate": 9.911645218627659e-06,
      "loss": 0.4281,
      "step": 1407
    },
    {
      "epoch": 0.08819430307395981,
      "grad_norm": 0.8866016612584962,
      "learning_rate": 9.91145526056813e-06,
      "loss": 0.4863,
      "step": 1408
    },
    {
      "epoch": 0.08825694107330212,
      "grad_norm": 0.8490503006788033,
      "learning_rate": 9.911265100352188e-06,
      "loss": 0.4789,
      "step": 1409
    },
    {
      "epoch": 0.08831957907264441,
      "grad_norm": 0.8544828785789507,
      "learning_rate": 9.911074737987665e-06,
      "loss": 0.4077,
      "step": 1410
    },
    {
      "epoch": 0.08838221707198672,
      "grad_norm": 0.8324897233725786,
      "learning_rate": 9.910884173482395e-06,
      "loss": 0.445,
      "step": 1411
    },
    {
      "epoch": 0.08844485507132901,
      "grad_norm": 0.9380870185865814,
      "learning_rate": 9.91069340684422e-06,
      "loss": 0.4337,
      "step": 1412
    },
    {
      "epoch": 0.08850749307067132,
      "grad_norm": 0.9284636077504759,
      "learning_rate": 9.910502438080993e-06,
      "loss": 0.457,
      "step": 1413
    },
    {
      "epoch": 0.08857013107001363,
      "grad_norm": 0.9537935376045981,
      "learning_rate": 9.910311267200574e-06,
      "loss": 0.4604,
      "step": 1414
    },
    {
      "epoch": 0.08863276906935592,
      "grad_norm": 0.8210276047848626,
      "learning_rate": 9.910119894210833e-06,
      "loss": 0.4312,
      "step": 1415
    },
    {
      "epoch": 0.08869540706869823,
      "grad_norm": 0.8788326891700143,
      "learning_rate": 9.909928319119644e-06,
      "loss": 0.4299,
      "step": 1416
    },
    {
      "epoch": 0.08875804506804053,
      "grad_norm": 0.9051767739822573,
      "learning_rate": 9.909736541934893e-06,
      "loss": 0.457,
      "step": 1417
    },
    {
      "epoch": 0.08882068306738283,
      "grad_norm": 0.9201199948958599,
      "learning_rate": 9.909544562664477e-06,
      "loss": 0.4669,
      "step": 1418
    },
    {
      "epoch": 0.08888332106672513,
      "grad_norm": 0.8803563775633492,
      "learning_rate": 9.909352381316295e-06,
      "loss": 0.4063,
      "step": 1419
    },
    {
      "epoch": 0.08894595906606743,
      "grad_norm": 0.8903313042330325,
      "learning_rate": 9.90915999789826e-06,
      "loss": 0.437,
      "step": 1420
    },
    {
      "epoch": 0.08900859706540973,
      "grad_norm": 0.9590001200987133,
      "learning_rate": 9.908967412418285e-06,
      "loss": 0.46,
      "step": 1421
    },
    {
      "epoch": 0.08907123506475204,
      "grad_norm": 0.8755420771248582,
      "learning_rate": 9.908774624884301e-06,
      "loss": 0.4144,
      "step": 1422
    },
    {
      "epoch": 0.08913387306409433,
      "grad_norm": 0.9814485717329147,
      "learning_rate": 9.908581635304241e-06,
      "loss": 0.4806,
      "step": 1423
    },
    {
      "epoch": 0.08919651106343664,
      "grad_norm": 0.8078865184386038,
      "learning_rate": 9.908388443686053e-06,
      "loss": 0.4408,
      "step": 1424
    },
    {
      "epoch": 0.08925914906277893,
      "grad_norm": 0.86630409433829,
      "learning_rate": 9.908195050037683e-06,
      "loss": 0.4376,
      "step": 1425
    },
    {
      "epoch": 0.08932178706212124,
      "grad_norm": 0.9222234286120634,
      "learning_rate": 9.908001454367096e-06,
      "loss": 0.4613,
      "step": 1426
    },
    {
      "epoch": 0.08938442506146353,
      "grad_norm": 0.9260909584738123,
      "learning_rate": 9.907807656682256e-06,
      "loss": 0.4783,
      "step": 1427
    },
    {
      "epoch": 0.08944706306080584,
      "grad_norm": 0.9026631278853621,
      "learning_rate": 9.907613656991143e-06,
      "loss": 0.4452,
      "step": 1428
    },
    {
      "epoch": 0.08950970106014813,
      "grad_norm": 0.8165509404055101,
      "learning_rate": 9.90741945530174e-06,
      "loss": 0.4284,
      "step": 1429
    },
    {
      "epoch": 0.08957233905949044,
      "grad_norm": 0.8317573036912767,
      "learning_rate": 9.907225051622042e-06,
      "loss": 0.4415,
      "step": 1430
    },
    {
      "epoch": 0.08963497705883273,
      "grad_norm": 0.9601241569944564,
      "learning_rate": 9.90703044596005e-06,
      "loss": 0.4912,
      "step": 1431
    },
    {
      "epoch": 0.08969761505817504,
      "grad_norm": 0.8710323141700268,
      "learning_rate": 9.906835638323775e-06,
      "loss": 0.4729,
      "step": 1432
    },
    {
      "epoch": 0.08976025305751734,
      "grad_norm": 0.8429993595610012,
      "learning_rate": 9.906640628721234e-06,
      "loss": 0.4313,
      "step": 1433
    },
    {
      "epoch": 0.08982289105685964,
      "grad_norm": 0.8917161274637533,
      "learning_rate": 9.906445417160452e-06,
      "loss": 0.4105,
      "step": 1434
    },
    {
      "epoch": 0.08988552905620195,
      "grad_norm": 0.8862198213063751,
      "learning_rate": 9.906250003649469e-06,
      "loss": 0.4122,
      "step": 1435
    },
    {
      "epoch": 0.08994816705554425,
      "grad_norm": 0.9225951552117314,
      "learning_rate": 9.906054388196324e-06,
      "loss": 0.4775,
      "step": 1436
    },
    {
      "epoch": 0.09001080505488655,
      "grad_norm": 0.9221453391753531,
      "learning_rate": 9.90585857080907e-06,
      "loss": 0.4715,
      "step": 1437
    },
    {
      "epoch": 0.09007344305422885,
      "grad_norm": 0.9735940052682739,
      "learning_rate": 9.905662551495766e-06,
      "loss": 0.4651,
      "step": 1438
    },
    {
      "epoch": 0.09013608105357115,
      "grad_norm": 0.9276860588843439,
      "learning_rate": 9.905466330264481e-06,
      "loss": 0.464,
      "step": 1439
    },
    {
      "epoch": 0.09019871905291345,
      "grad_norm": 0.864940857680695,
      "learning_rate": 9.90526990712329e-06,
      "loss": 0.4238,
      "step": 1440
    },
    {
      "epoch": 0.09026135705225576,
      "grad_norm": 0.9008822748292902,
      "learning_rate": 9.905073282080281e-06,
      "loss": 0.4235,
      "step": 1441
    },
    {
      "epoch": 0.09032399505159805,
      "grad_norm": 0.8855782072529085,
      "learning_rate": 9.904876455143545e-06,
      "loss": 0.3939,
      "step": 1442
    },
    {
      "epoch": 0.09038663305094036,
      "grad_norm": 0.788617731769824,
      "learning_rate": 9.904679426321184e-06,
      "loss": 0.485,
      "step": 1443
    },
    {
      "epoch": 0.09044927105028265,
      "grad_norm": 0.8366924188789073,
      "learning_rate": 9.904482195621308e-06,
      "loss": 0.4081,
      "step": 1444
    },
    {
      "epoch": 0.09051190904962496,
      "grad_norm": 0.9083772926928925,
      "learning_rate": 9.904284763052032e-06,
      "loss": 0.4131,
      "step": 1445
    },
    {
      "epoch": 0.09057454704896725,
      "grad_norm": 0.8951185972984395,
      "learning_rate": 9.904087128621487e-06,
      "loss": 0.4466,
      "step": 1446
    },
    {
      "epoch": 0.09063718504830956,
      "grad_norm": 0.9161670139002976,
      "learning_rate": 9.903889292337804e-06,
      "loss": 0.421,
      "step": 1447
    },
    {
      "epoch": 0.09069982304765185,
      "grad_norm": 0.7664988393569152,
      "learning_rate": 9.90369125420913e-06,
      "loss": 0.4715,
      "step": 1448
    },
    {
      "epoch": 0.09076246104699416,
      "grad_norm": 0.8829658941645271,
      "learning_rate": 9.903493014243613e-06,
      "loss": 0.4458,
      "step": 1449
    },
    {
      "epoch": 0.09082509904633645,
      "grad_norm": 0.9648936179194268,
      "learning_rate": 9.903294572449412e-06,
      "loss": 0.4711,
      "step": 1450
    },
    {
      "epoch": 0.09088773704567876,
      "grad_norm": 0.8828475881410391,
      "learning_rate": 9.903095928834696e-06,
      "loss": 0.4517,
      "step": 1451
    },
    {
      "epoch": 0.09095037504502106,
      "grad_norm": 0.9297508385704868,
      "learning_rate": 9.902897083407642e-06,
      "loss": 0.4316,
      "step": 1452
    },
    {
      "epoch": 0.09101301304436336,
      "grad_norm": 0.9711919905025587,
      "learning_rate": 9.902698036176437e-06,
      "loss": 0.4717,
      "step": 1453
    },
    {
      "epoch": 0.09107565104370566,
      "grad_norm": 0.8949900602784688,
      "learning_rate": 9.902498787149268e-06,
      "loss": 0.4497,
      "step": 1454
    },
    {
      "epoch": 0.09113828904304797,
      "grad_norm": 0.9260906587721149,
      "learning_rate": 9.90229933633434e-06,
      "loss": 0.4408,
      "step": 1455
    },
    {
      "epoch": 0.09120092704239027,
      "grad_norm": 0.9368634417647993,
      "learning_rate": 9.902099683739859e-06,
      "loss": 0.4411,
      "step": 1456
    },
    {
      "epoch": 0.09126356504173257,
      "grad_norm": 0.912966210524351,
      "learning_rate": 9.901899829374048e-06,
      "loss": 0.4453,
      "step": 1457
    },
    {
      "epoch": 0.09132620304107487,
      "grad_norm": 0.9242540766879641,
      "learning_rate": 9.901699773245128e-06,
      "loss": 0.3862,
      "step": 1458
    },
    {
      "epoch": 0.09138884104041717,
      "grad_norm": 0.96048669816582,
      "learning_rate": 9.901499515361338e-06,
      "loss": 0.4759,
      "step": 1459
    },
    {
      "epoch": 0.09145147903975948,
      "grad_norm": 0.9207268526676665,
      "learning_rate": 9.901299055730916e-06,
      "loss": 0.4117,
      "step": 1460
    },
    {
      "epoch": 0.09151411703910177,
      "grad_norm": 0.8984347536281135,
      "learning_rate": 9.901098394362117e-06,
      "loss": 0.481,
      "step": 1461
    },
    {
      "epoch": 0.09157675503844408,
      "grad_norm": 0.9894226969579926,
      "learning_rate": 9.900897531263197e-06,
      "loss": 0.4655,
      "step": 1462
    },
    {
      "epoch": 0.09163939303778637,
      "grad_norm": 0.848282126916537,
      "learning_rate": 9.900696466442425e-06,
      "loss": 0.5055,
      "step": 1463
    },
    {
      "epoch": 0.09170203103712868,
      "grad_norm": 0.9072287605583648,
      "learning_rate": 9.900495199908077e-06,
      "loss": 0.4234,
      "step": 1464
    },
    {
      "epoch": 0.09176466903647097,
      "grad_norm": 0.9812647787164273,
      "learning_rate": 9.900293731668437e-06,
      "loss": 0.4999,
      "step": 1465
    },
    {
      "epoch": 0.09182730703581328,
      "grad_norm": 0.9044640721637041,
      "learning_rate": 9.900092061731799e-06,
      "loss": 0.4497,
      "step": 1466
    },
    {
      "epoch": 0.09188994503515557,
      "grad_norm": 0.9120064060300129,
      "learning_rate": 9.899890190106461e-06,
      "loss": 0.4305,
      "step": 1467
    },
    {
      "epoch": 0.09195258303449788,
      "grad_norm": 0.8626294196444891,
      "learning_rate": 9.899688116800732e-06,
      "loss": 0.448,
      "step": 1468
    },
    {
      "epoch": 0.09201522103384017,
      "grad_norm": 0.9443738868187155,
      "learning_rate": 9.899485841822932e-06,
      "loss": 0.419,
      "step": 1469
    },
    {
      "epoch": 0.09207785903318248,
      "grad_norm": 0.9149288654865164,
      "learning_rate": 9.899283365181384e-06,
      "loss": 0.4553,
      "step": 1470
    },
    {
      "epoch": 0.09214049703252478,
      "grad_norm": 0.8862757443050963,
      "learning_rate": 9.899080686884426e-06,
      "loss": 0.4609,
      "step": 1471
    },
    {
      "epoch": 0.09220313503186708,
      "grad_norm": 0.9397660849885395,
      "learning_rate": 9.898877806940396e-06,
      "loss": 0.4614,
      "step": 1472
    },
    {
      "epoch": 0.09226577303120938,
      "grad_norm": 0.931526896673016,
      "learning_rate": 9.898674725357646e-06,
      "loss": 0.4592,
      "step": 1473
    },
    {
      "epoch": 0.09232841103055169,
      "grad_norm": 0.7917876619276295,
      "learning_rate": 9.898471442144535e-06,
      "loss": 0.4122,
      "step": 1474
    },
    {
      "epoch": 0.09239104902989398,
      "grad_norm": 0.9339296768327562,
      "learning_rate": 9.898267957309432e-06,
      "loss": 0.423,
      "step": 1475
    },
    {
      "epoch": 0.09245368702923629,
      "grad_norm": 1.096678304025624,
      "learning_rate": 9.898064270860709e-06,
      "loss": 0.4617,
      "step": 1476
    },
    {
      "epoch": 0.09251632502857858,
      "grad_norm": 1.2474738436209039,
      "learning_rate": 9.897860382806752e-06,
      "loss": 0.47,
      "step": 1477
    },
    {
      "epoch": 0.09257896302792089,
      "grad_norm": 0.9793132893795026,
      "learning_rate": 9.897656293155953e-06,
      "loss": 0.4646,
      "step": 1478
    },
    {
      "epoch": 0.0926416010272632,
      "grad_norm": 0.9849391562004103,
      "learning_rate": 9.89745200191671e-06,
      "loss": 0.4388,
      "step": 1479
    },
    {
      "epoch": 0.09270423902660549,
      "grad_norm": 1.0037002442017227,
      "learning_rate": 9.897247509097436e-06,
      "loss": 0.4717,
      "step": 1480
    },
    {
      "epoch": 0.0927668770259478,
      "grad_norm": 0.8125584336111052,
      "learning_rate": 9.897042814706543e-06,
      "loss": 0.4244,
      "step": 1481
    },
    {
      "epoch": 0.09282951502529009,
      "grad_norm": 0.9190290288335197,
      "learning_rate": 9.896837918752463e-06,
      "loss": 0.435,
      "step": 1482
    },
    {
      "epoch": 0.0928921530246324,
      "grad_norm": 1.0200006074645942,
      "learning_rate": 9.896632821243622e-06,
      "loss": 0.4771,
      "step": 1483
    },
    {
      "epoch": 0.09295479102397469,
      "grad_norm": 0.8917018157143095,
      "learning_rate": 9.896427522188466e-06,
      "loss": 0.457,
      "step": 1484
    },
    {
      "epoch": 0.093017429023317,
      "grad_norm": 0.9446787976929322,
      "learning_rate": 9.896222021595443e-06,
      "loss": 0.4872,
      "step": 1485
    },
    {
      "epoch": 0.0930800670226593,
      "grad_norm": 1.0377954352135303,
      "learning_rate": 9.896016319473015e-06,
      "loss": 0.4613,
      "step": 1486
    },
    {
      "epoch": 0.0931427050220016,
      "grad_norm": 0.8622481990808284,
      "learning_rate": 9.895810415829645e-06,
      "loss": 0.5145,
      "step": 1487
    },
    {
      "epoch": 0.0932053430213439,
      "grad_norm": 0.9456294006331782,
      "learning_rate": 9.895604310673809e-06,
      "loss": 0.4091,
      "step": 1488
    },
    {
      "epoch": 0.0932679810206862,
      "grad_norm": 0.9435843121809769,
      "learning_rate": 9.895398004013994e-06,
      "loss": 0.4865,
      "step": 1489
    },
    {
      "epoch": 0.0933306190200285,
      "grad_norm": 0.9306929711031157,
      "learning_rate": 9.895191495858686e-06,
      "loss": 0.4319,
      "step": 1490
    },
    {
      "epoch": 0.0933932570193708,
      "grad_norm": 0.8789036358482651,
      "learning_rate": 9.894984786216389e-06,
      "loss": 0.4356,
      "step": 1491
    },
    {
      "epoch": 0.0934558950187131,
      "grad_norm": 1.0070410068059266,
      "learning_rate": 9.894777875095609e-06,
      "loss": 0.4554,
      "step": 1492
    },
    {
      "epoch": 0.0935185330180554,
      "grad_norm": 0.9637700323071658,
      "learning_rate": 9.894570762504863e-06,
      "loss": 0.4487,
      "step": 1493
    },
    {
      "epoch": 0.0935811710173977,
      "grad_norm": 0.8355788205250919,
      "learning_rate": 9.894363448452675e-06,
      "loss": 0.4279,
      "step": 1494
    },
    {
      "epoch": 0.09364380901674,
      "grad_norm": 1.0170388582004242,
      "learning_rate": 9.89415593294758e-06,
      "loss": 0.4357,
      "step": 1495
    },
    {
      "epoch": 0.0937064470160823,
      "grad_norm": 0.8979517974333173,
      "learning_rate": 9.893948215998118e-06,
      "loss": 0.4788,
      "step": 1496
    },
    {
      "epoch": 0.09376908501542461,
      "grad_norm": 1.0560972505172115,
      "learning_rate": 9.89374029761284e-06,
      "loss": 0.4457,
      "step": 1497
    },
    {
      "epoch": 0.0938317230147669,
      "grad_norm": 0.9359166153795132,
      "learning_rate": 9.893532177800302e-06,
      "loss": 0.4361,
      "step": 1498
    },
    {
      "epoch": 0.09389436101410921,
      "grad_norm": 0.9549406530855985,
      "learning_rate": 9.893323856569071e-06,
      "loss": 0.4835,
      "step": 1499
    },
    {
      "epoch": 0.09395699901345152,
      "grad_norm": 0.8119845808565949,
      "learning_rate": 9.893115333927723e-06,
      "loss": 0.4816,
      "step": 1500
    },
    {
      "epoch": 0.09401963701279381,
      "grad_norm": 0.8421225662717794,
      "learning_rate": 9.892906609884838e-06,
      "loss": 0.4532,
      "step": 1501
    },
    {
      "epoch": 0.09408227501213612,
      "grad_norm": 0.8562898607430842,
      "learning_rate": 9.892697684449011e-06,
      "loss": 0.4535,
      "step": 1502
    },
    {
      "epoch": 0.09414491301147841,
      "grad_norm": 0.8571592201766343,
      "learning_rate": 9.892488557628838e-06,
      "loss": 0.4526,
      "step": 1503
    },
    {
      "epoch": 0.09420755101082072,
      "grad_norm": 0.8016828152124434,
      "learning_rate": 9.892279229432929e-06,
      "loss": 0.4342,
      "step": 1504
    },
    {
      "epoch": 0.09427018901016301,
      "grad_norm": 0.9507099778753813,
      "learning_rate": 9.892069699869897e-06,
      "loss": 0.4887,
      "step": 1505
    },
    {
      "epoch": 0.09433282700950532,
      "grad_norm": 0.8658002651569511,
      "learning_rate": 9.891859968948368e-06,
      "loss": 0.4856,
      "step": 1506
    },
    {
      "epoch": 0.09439546500884761,
      "grad_norm": 0.9233232235498693,
      "learning_rate": 9.891650036676976e-06,
      "loss": 0.5101,
      "step": 1507
    },
    {
      "epoch": 0.09445810300818992,
      "grad_norm": 0.8993673258117713,
      "learning_rate": 9.891439903064361e-06,
      "loss": 0.4646,
      "step": 1508
    },
    {
      "epoch": 0.09452074100753222,
      "grad_norm": 0.8886501405166498,
      "learning_rate": 9.89122956811917e-06,
      "loss": 0.4157,
      "step": 1509
    },
    {
      "epoch": 0.09458337900687452,
      "grad_norm": 0.9656494805603429,
      "learning_rate": 9.891019031850063e-06,
      "loss": 0.5207,
      "step": 1510
    },
    {
      "epoch": 0.09464601700621682,
      "grad_norm": 0.8889575971982786,
      "learning_rate": 9.890808294265706e-06,
      "loss": 0.4679,
      "step": 1511
    },
    {
      "epoch": 0.09470865500555913,
      "grad_norm": 0.8374231232526697,
      "learning_rate": 9.890597355374769e-06,
      "loss": 0.4794,
      "step": 1512
    },
    {
      "epoch": 0.09477129300490142,
      "grad_norm": 0.9428455338207213,
      "learning_rate": 9.89038621518594e-06,
      "loss": 0.4849,
      "step": 1513
    },
    {
      "epoch": 0.09483393100424373,
      "grad_norm": 0.9317924947339054,
      "learning_rate": 9.890174873707905e-06,
      "loss": 0.4472,
      "step": 1514
    },
    {
      "epoch": 0.09489656900358602,
      "grad_norm": 0.8867083688940639,
      "learning_rate": 9.889963330949366e-06,
      "loss": 0.4402,
      "step": 1515
    },
    {
      "epoch": 0.09495920700292833,
      "grad_norm": 0.9445734555186007,
      "learning_rate": 9.889751586919028e-06,
      "loss": 0.4344,
      "step": 1516
    },
    {
      "epoch": 0.09502184500227062,
      "grad_norm": 0.9875804337068883,
      "learning_rate": 9.889539641625605e-06,
      "loss": 0.45,
      "step": 1517
    },
    {
      "epoch": 0.09508448300161293,
      "grad_norm": 0.8842639210716592,
      "learning_rate": 9.889327495077824e-06,
      "loss": 0.4652,
      "step": 1518
    },
    {
      "epoch": 0.09514712100095522,
      "grad_norm": 0.890292067193579,
      "learning_rate": 9.889115147284417e-06,
      "loss": 0.476,
      "step": 1519
    },
    {
      "epoch": 0.09520975900029753,
      "grad_norm": 0.8141322385731111,
      "learning_rate": 9.888902598254122e-06,
      "loss": 0.3792,
      "step": 1520
    },
    {
      "epoch": 0.09527239699963984,
      "grad_norm": 0.8322514190258518,
      "learning_rate": 9.88868984799569e-06,
      "loss": 0.4303,
      "step": 1521
    },
    {
      "epoch": 0.09533503499898213,
      "grad_norm": 0.8904831853136833,
      "learning_rate": 9.888476896517875e-06,
      "loss": 0.4367,
      "step": 1522
    },
    {
      "epoch": 0.09539767299832444,
      "grad_norm": 0.863877405150155,
      "learning_rate": 9.888263743829445e-06,
      "loss": 0.4587,
      "step": 1523
    },
    {
      "epoch": 0.09546031099766673,
      "grad_norm": 0.8658056910905525,
      "learning_rate": 9.888050389939172e-06,
      "loss": 0.4269,
      "step": 1524
    },
    {
      "epoch": 0.09552294899700904,
      "grad_norm": 0.9134606048027469,
      "learning_rate": 9.887836834855836e-06,
      "loss": 0.4463,
      "step": 1525
    },
    {
      "epoch": 0.09558558699635133,
      "grad_norm": 0.8933774988349712,
      "learning_rate": 9.88762307858823e-06,
      "loss": 0.417,
      "step": 1526
    },
    {
      "epoch": 0.09564822499569364,
      "grad_norm": 0.8838948085061537,
      "learning_rate": 9.88740912114515e-06,
      "loss": 0.4637,
      "step": 1527
    },
    {
      "epoch": 0.09571086299503594,
      "grad_norm": 0.8540811656528821,
      "learning_rate": 9.887194962535405e-06,
      "loss": 0.4077,
      "step": 1528
    },
    {
      "epoch": 0.09577350099437824,
      "grad_norm": 0.9069828909742581,
      "learning_rate": 9.886980602767806e-06,
      "loss": 0.4535,
      "step": 1529
    },
    {
      "epoch": 0.09583613899372054,
      "grad_norm": 0.8791266051327843,
      "learning_rate": 9.886766041851178e-06,
      "loss": 0.4633,
      "step": 1530
    },
    {
      "epoch": 0.09589877699306285,
      "grad_norm": 0.9516000875499592,
      "learning_rate": 9.886551279794356e-06,
      "loss": 0.4509,
      "step": 1531
    },
    {
      "epoch": 0.09596141499240514,
      "grad_norm": 0.8829923576823713,
      "learning_rate": 9.886336316606175e-06,
      "loss": 0.4411,
      "step": 1532
    },
    {
      "epoch": 0.09602405299174745,
      "grad_norm": 0.9865985601591539,
      "learning_rate": 9.886121152295483e-06,
      "loss": 0.4435,
      "step": 1533
    },
    {
      "epoch": 0.09608669099108974,
      "grad_norm": 0.9586963515089149,
      "learning_rate": 9.885905786871138e-06,
      "loss": 0.46,
      "step": 1534
    },
    {
      "epoch": 0.09614932899043205,
      "grad_norm": 0.8626284828213061,
      "learning_rate": 9.885690220342005e-06,
      "loss": 0.5006,
      "step": 1535
    },
    {
      "epoch": 0.09621196698977434,
      "grad_norm": 0.8488840284700826,
      "learning_rate": 9.885474452716955e-06,
      "loss": 0.4228,
      "step": 1536
    },
    {
      "epoch": 0.09627460498911665,
      "grad_norm": 0.9429534463056131,
      "learning_rate": 9.885258484004869e-06,
      "loss": 0.4967,
      "step": 1537
    },
    {
      "epoch": 0.09633724298845894,
      "grad_norm": 0.95221214117684,
      "learning_rate": 9.885042314214637e-06,
      "loss": 0.5043,
      "step": 1538
    },
    {
      "epoch": 0.09639988098780125,
      "grad_norm": 0.9219411365073961,
      "learning_rate": 9.884825943355158e-06,
      "loss": 0.4555,
      "step": 1539
    },
    {
      "epoch": 0.09646251898714354,
      "grad_norm": 0.8950474938169982,
      "learning_rate": 9.884609371435335e-06,
      "loss": 0.4329,
      "step": 1540
    },
    {
      "epoch": 0.09652515698648585,
      "grad_norm": 0.9578574923298676,
      "learning_rate": 9.884392598464086e-06,
      "loss": 0.4499,
      "step": 1541
    },
    {
      "epoch": 0.09658779498582815,
      "grad_norm": 0.8723696327015126,
      "learning_rate": 9.884175624450328e-06,
      "loss": 0.4313,
      "step": 1542
    },
    {
      "epoch": 0.09665043298517045,
      "grad_norm": 0.858397950929006,
      "learning_rate": 9.883958449402997e-06,
      "loss": 0.4645,
      "step": 1543
    },
    {
      "epoch": 0.09671307098451276,
      "grad_norm": 0.9091150484480857,
      "learning_rate": 9.883741073331027e-06,
      "loss": 0.4314,
      "step": 1544
    },
    {
      "epoch": 0.09677570898385505,
      "grad_norm": 0.8805184215137215,
      "learning_rate": 9.883523496243372e-06,
      "loss": 0.4548,
      "step": 1545
    },
    {
      "epoch": 0.09683834698319736,
      "grad_norm": 0.8937919293605776,
      "learning_rate": 9.883305718148981e-06,
      "loss": 0.4455,
      "step": 1546
    },
    {
      "epoch": 0.09690098498253966,
      "grad_norm": 0.9354055691986288,
      "learning_rate": 9.88308773905682e-06,
      "loss": 0.4881,
      "step": 1547
    },
    {
      "epoch": 0.09696362298188196,
      "grad_norm": 0.8898781625685626,
      "learning_rate": 9.882869558975863e-06,
      "loss": 0.4452,
      "step": 1548
    },
    {
      "epoch": 0.09702626098122426,
      "grad_norm": 0.927365062960007,
      "learning_rate": 9.882651177915087e-06,
      "loss": 0.49,
      "step": 1549
    },
    {
      "epoch": 0.09708889898056657,
      "grad_norm": 0.9936598088360935,
      "learning_rate": 9.882432595883482e-06,
      "loss": 0.4522,
      "step": 1550
    },
    {
      "epoch": 0.09715153697990886,
      "grad_norm": 0.911308392594863,
      "learning_rate": 9.882213812890048e-06,
      "loss": 0.4617,
      "step": 1551
    },
    {
      "epoch": 0.09721417497925117,
      "grad_norm": 0.8789885262684483,
      "learning_rate": 9.881994828943784e-06,
      "loss": 0.4469,
      "step": 1552
    },
    {
      "epoch": 0.09727681297859346,
      "grad_norm": 0.8387014288547737,
      "learning_rate": 9.881775644053708e-06,
      "loss": 0.4396,
      "step": 1553
    },
    {
      "epoch": 0.09733945097793577,
      "grad_norm": 0.9624855766607733,
      "learning_rate": 9.881556258228841e-06,
      "loss": 0.4692,
      "step": 1554
    },
    {
      "epoch": 0.09740208897727806,
      "grad_norm": 0.9625210172598724,
      "learning_rate": 9.881336671478211e-06,
      "loss": 0.4763,
      "step": 1555
    },
    {
      "epoch": 0.09746472697662037,
      "grad_norm": 0.9882852689752175,
      "learning_rate": 9.881116883810858e-06,
      "loss": 0.5004,
      "step": 1556
    },
    {
      "epoch": 0.09752736497596266,
      "grad_norm": 0.8163529937277837,
      "learning_rate": 9.880896895235828e-06,
      "loss": 0.4058,
      "step": 1557
    },
    {
      "epoch": 0.09759000297530497,
      "grad_norm": 0.8903999132465609,
      "learning_rate": 9.880676705762178e-06,
      "loss": 0.4232,
      "step": 1558
    },
    {
      "epoch": 0.09765264097464726,
      "grad_norm": 0.8480173531934881,
      "learning_rate": 9.880456315398967e-06,
      "loss": 0.4272,
      "step": 1559
    },
    {
      "epoch": 0.09771527897398957,
      "grad_norm": 0.9598374480060924,
      "learning_rate": 9.880235724155267e-06,
      "loss": 0.4562,
      "step": 1560
    },
    {
      "epoch": 0.09777791697333187,
      "grad_norm": 0.9188573023833896,
      "learning_rate": 9.880014932040161e-06,
      "loss": 0.4419,
      "step": 1561
    },
    {
      "epoch": 0.09784055497267417,
      "grad_norm": 0.8885909492222273,
      "learning_rate": 9.879793939062734e-06,
      "loss": 0.4344,
      "step": 1562
    },
    {
      "epoch": 0.09790319297201647,
      "grad_norm": 0.9671803805328165,
      "learning_rate": 9.879572745232082e-06,
      "loss": 0.4431,
      "step": 1563
    },
    {
      "epoch": 0.09796583097135877,
      "grad_norm": 0.8867169527735584,
      "learning_rate": 9.879351350557311e-06,
      "loss": 0.4514,
      "step": 1564
    },
    {
      "epoch": 0.09802846897070108,
      "grad_norm": 0.8305946197757244,
      "learning_rate": 9.879129755047532e-06,
      "loss": 0.4188,
      "step": 1565
    },
    {
      "epoch": 0.09809110697004338,
      "grad_norm": 0.8997359923356064,
      "learning_rate": 9.878907958711867e-06,
      "loss": 0.4135,
      "step": 1566
    },
    {
      "epoch": 0.09815374496938568,
      "grad_norm": 0.9292691686082959,
      "learning_rate": 9.878685961559445e-06,
      "loss": 0.477,
      "step": 1567
    },
    {
      "epoch": 0.09821638296872798,
      "grad_norm": 0.9524964470534331,
      "learning_rate": 9.878463763599401e-06,
      "loss": 0.4739,
      "step": 1568
    },
    {
      "epoch": 0.09827902096807029,
      "grad_norm": 0.8832153990349538,
      "learning_rate": 9.878241364840887e-06,
      "loss": 0.4855,
      "step": 1569
    },
    {
      "epoch": 0.09834165896741258,
      "grad_norm": 0.996866276785187,
      "learning_rate": 9.878018765293052e-06,
      "loss": 0.4967,
      "step": 1570
    },
    {
      "epoch": 0.09840429696675489,
      "grad_norm": 0.8977586599449298,
      "learning_rate": 9.877795964965058e-06,
      "loss": 0.4396,
      "step": 1571
    },
    {
      "epoch": 0.09846693496609718,
      "grad_norm": 0.8102605556571321,
      "learning_rate": 9.877572963866075e-06,
      "loss": 0.4343,
      "step": 1572
    },
    {
      "epoch": 0.09852957296543949,
      "grad_norm": 0.8653819066194124,
      "learning_rate": 9.877349762005286e-06,
      "loss": 0.4118,
      "step": 1573
    },
    {
      "epoch": 0.09859221096478178,
      "grad_norm": 0.8802786482958497,
      "learning_rate": 9.877126359391875e-06,
      "loss": 0.4677,
      "step": 1574
    },
    {
      "epoch": 0.09865484896412409,
      "grad_norm": 0.8809539880059191,
      "learning_rate": 9.876902756035038e-06,
      "loss": 0.445,
      "step": 1575
    },
    {
      "epoch": 0.09871748696346638,
      "grad_norm": 0.8227556671158442,
      "learning_rate": 9.87667895194398e-06,
      "loss": 0.4327,
      "step": 1576
    },
    {
      "epoch": 0.09878012496280869,
      "grad_norm": 0.8990257252871325,
      "learning_rate": 9.87645494712791e-06,
      "loss": 0.4274,
      "step": 1577
    },
    {
      "epoch": 0.09884276296215098,
      "grad_norm": 0.8561470829658842,
      "learning_rate": 9.876230741596047e-06,
      "loss": 0.4271,
      "step": 1578
    },
    {
      "epoch": 0.09890540096149329,
      "grad_norm": 0.8582507836484214,
      "learning_rate": 9.876006335357625e-06,
      "loss": 0.4381,
      "step": 1579
    },
    {
      "epoch": 0.09896803896083559,
      "grad_norm": 0.9126165872093309,
      "learning_rate": 9.875781728421874e-06,
      "loss": 0.4579,
      "step": 1580
    },
    {
      "epoch": 0.0990306769601779,
      "grad_norm": 0.8722106874453135,
      "learning_rate": 9.875556920798043e-06,
      "loss": 0.4144,
      "step": 1581
    },
    {
      "epoch": 0.09909331495952019,
      "grad_norm": 0.9145094480176127,
      "learning_rate": 9.875331912495385e-06,
      "loss": 0.4456,
      "step": 1582
    },
    {
      "epoch": 0.0991559529588625,
      "grad_norm": 0.8706458609675202,
      "learning_rate": 9.875106703523161e-06,
      "loss": 0.4286,
      "step": 1583
    },
    {
      "epoch": 0.09921859095820479,
      "grad_norm": 0.8802434336539445,
      "learning_rate": 9.87488129389064e-06,
      "loss": 0.4032,
      "step": 1584
    },
    {
      "epoch": 0.0992812289575471,
      "grad_norm": 0.8730111395571392,
      "learning_rate": 9.874655683607099e-06,
      "loss": 0.4378,
      "step": 1585
    },
    {
      "epoch": 0.09934386695688939,
      "grad_norm": 0.9418277769097348,
      "learning_rate": 9.874429872681826e-06,
      "loss": 0.4733,
      "step": 1586
    },
    {
      "epoch": 0.0994065049562317,
      "grad_norm": 0.8731033196585996,
      "learning_rate": 9.874203861124114e-06,
      "loss": 0.4454,
      "step": 1587
    },
    {
      "epoch": 0.099469142955574,
      "grad_norm": 0.8737559782897221,
      "learning_rate": 9.873977648943267e-06,
      "loss": 0.4726,
      "step": 1588
    },
    {
      "epoch": 0.0995317809549163,
      "grad_norm": 0.9210275981827408,
      "learning_rate": 9.873751236148596e-06,
      "loss": 0.4898,
      "step": 1589
    },
    {
      "epoch": 0.0995944189542586,
      "grad_norm": 1.0915458781321115,
      "learning_rate": 9.87352462274942e-06,
      "loss": 0.4408,
      "step": 1590
    },
    {
      "epoch": 0.0996570569536009,
      "grad_norm": 0.8422571111739536,
      "learning_rate": 9.873297808755065e-06,
      "loss": 0.4492,
      "step": 1591
    },
    {
      "epoch": 0.09971969495294321,
      "grad_norm": 0.8808428287782618,
      "learning_rate": 9.873070794174866e-06,
      "loss": 0.4247,
      "step": 1592
    },
    {
      "epoch": 0.0997823329522855,
      "grad_norm": 0.9735610810916026,
      "learning_rate": 9.87284357901817e-06,
      "loss": 0.4776,
      "step": 1593
    },
    {
      "epoch": 0.09984497095162781,
      "grad_norm": 0.9390131059833663,
      "learning_rate": 9.872616163294328e-06,
      "loss": 0.4992,
      "step": 1594
    },
    {
      "epoch": 0.0999076089509701,
      "grad_norm": 0.8827399506654423,
      "learning_rate": 9.872388547012701e-06,
      "loss": 0.4202,
      "step": 1595
    },
    {
      "epoch": 0.09997024695031241,
      "grad_norm": 0.8987906903067735,
      "learning_rate": 9.872160730182656e-06,
      "loss": 0.4048,
      "step": 1596
    },
    {
      "epoch": 0.1000328849496547,
      "grad_norm": 0.8138361145827581,
      "learning_rate": 9.871932712813572e-06,
      "loss": 0.4297,
      "step": 1597
    },
    {
      "epoch": 0.10009552294899701,
      "grad_norm": 0.8130679581312211,
      "learning_rate": 9.871704494914832e-06,
      "loss": 0.4752,
      "step": 1598
    },
    {
      "epoch": 0.1001581609483393,
      "grad_norm": 0.975517713522263,
      "learning_rate": 9.871476076495831e-06,
      "loss": 0.4406,
      "step": 1599
    },
    {
      "epoch": 0.10022079894768161,
      "grad_norm": 0.8951589086817716,
      "learning_rate": 9.871247457565971e-06,
      "loss": 0.439,
      "step": 1600
    },
    {
      "epoch": 0.10028343694702391,
      "grad_norm": 0.9146858997152447,
      "learning_rate": 9.871018638134661e-06,
      "loss": 0.4925,
      "step": 1601
    },
    {
      "epoch": 0.10034607494636621,
      "grad_norm": 0.928208667574649,
      "learning_rate": 9.87078961821132e-06,
      "loss": 0.4433,
      "step": 1602
    },
    {
      "epoch": 0.10040871294570851,
      "grad_norm": 0.997420475943222,
      "learning_rate": 9.870560397805375e-06,
      "loss": 0.4543,
      "step": 1603
    },
    {
      "epoch": 0.10047135094505082,
      "grad_norm": 0.9061383374389848,
      "learning_rate": 9.870330976926261e-06,
      "loss": 0.4476,
      "step": 1604
    },
    {
      "epoch": 0.10053398894439311,
      "grad_norm": 0.8410419745164147,
      "learning_rate": 9.870101355583416e-06,
      "loss": 0.4325,
      "step": 1605
    },
    {
      "epoch": 0.10059662694373542,
      "grad_norm": 0.8533857346363363,
      "learning_rate": 9.8698715337863e-06,
      "loss": 0.439,
      "step": 1606
    },
    {
      "epoch": 0.10065926494307771,
      "grad_norm": 0.8295931134816784,
      "learning_rate": 9.869641511544366e-06,
      "loss": 0.4136,
      "step": 1607
    },
    {
      "epoch": 0.10072190294242002,
      "grad_norm": 0.9321156041665808,
      "learning_rate": 9.86941128886708e-06,
      "loss": 0.4681,
      "step": 1608
    },
    {
      "epoch": 0.10078454094176233,
      "grad_norm": 0.8710737149967902,
      "learning_rate": 9.869180865763926e-06,
      "loss": 0.4439,
      "step": 1609
    },
    {
      "epoch": 0.10084717894110462,
      "grad_norm": 0.8640181492809873,
      "learning_rate": 9.868950242244382e-06,
      "loss": 0.4625,
      "step": 1610
    },
    {
      "epoch": 0.10090981694044693,
      "grad_norm": 0.9643987223518166,
      "learning_rate": 9.868719418317944e-06,
      "loss": 0.4697,
      "step": 1611
    },
    {
      "epoch": 0.10097245493978922,
      "grad_norm": 0.8626840173643526,
      "learning_rate": 9.868488393994109e-06,
      "loss": 0.4334,
      "step": 1612
    },
    {
      "epoch": 0.10103509293913153,
      "grad_norm": 0.950042206748387,
      "learning_rate": 9.86825716928239e-06,
      "loss": 0.4606,
      "step": 1613
    },
    {
      "epoch": 0.10109773093847382,
      "grad_norm": 0.886719828764984,
      "learning_rate": 9.8680257441923e-06,
      "loss": 0.47,
      "step": 1614
    },
    {
      "epoch": 0.10116036893781613,
      "grad_norm": 0.8588788137389599,
      "learning_rate": 9.867794118733368e-06,
      "loss": 0.4278,
      "step": 1615
    },
    {
      "epoch": 0.10122300693715842,
      "grad_norm": 0.8227576623418922,
      "learning_rate": 9.867562292915128e-06,
      "loss": 0.4028,
      "step": 1616
    },
    {
      "epoch": 0.10128564493650073,
      "grad_norm": 0.8981278258606616,
      "learning_rate": 9.867330266747118e-06,
      "loss": 0.4281,
      "step": 1617
    },
    {
      "epoch": 0.10134828293584303,
      "grad_norm": 1.0052308750920933,
      "learning_rate": 9.867098040238892e-06,
      "loss": 0.4357,
      "step": 1618
    },
    {
      "epoch": 0.10141092093518533,
      "grad_norm": 0.907873859413351,
      "learning_rate": 9.866865613400008e-06,
      "loss": 0.4747,
      "step": 1619
    },
    {
      "epoch": 0.10147355893452763,
      "grad_norm": 0.986479635281565,
      "learning_rate": 9.86663298624003e-06,
      "loss": 0.4813,
      "step": 1620
    },
    {
      "epoch": 0.10153619693386993,
      "grad_norm": 0.8610221096948076,
      "learning_rate": 9.866400158768536e-06,
      "loss": 0.4468,
      "step": 1621
    },
    {
      "epoch": 0.10159883493321223,
      "grad_norm": 0.9057490033619454,
      "learning_rate": 9.866167130995108e-06,
      "loss": 0.425,
      "step": 1622
    },
    {
      "epoch": 0.10166147293255454,
      "grad_norm": 0.8898577268240097,
      "learning_rate": 9.86593390292934e-06,
      "loss": 0.4476,
      "step": 1623
    },
    {
      "epoch": 0.10172411093189683,
      "grad_norm": 0.8840136506244884,
      "learning_rate": 9.865700474580827e-06,
      "loss": 0.4379,
      "step": 1624
    },
    {
      "epoch": 0.10178674893123914,
      "grad_norm": 0.9451869427096309,
      "learning_rate": 9.86546684595918e-06,
      "loss": 0.4514,
      "step": 1625
    },
    {
      "epoch": 0.10184938693058143,
      "grad_norm": 0.8657267730157551,
      "learning_rate": 9.865233017074014e-06,
      "loss": 0.4767,
      "step": 1626
    },
    {
      "epoch": 0.10191202492992374,
      "grad_norm": 0.8145133597527424,
      "learning_rate": 9.864998987934954e-06,
      "loss": 0.4214,
      "step": 1627
    },
    {
      "epoch": 0.10197466292926603,
      "grad_norm": 0.8903203735523892,
      "learning_rate": 9.864764758551632e-06,
      "loss": 0.4254,
      "step": 1628
    },
    {
      "epoch": 0.10203730092860834,
      "grad_norm": 0.8914813918201505,
      "learning_rate": 9.864530328933692e-06,
      "loss": 0.4749,
      "step": 1629
    },
    {
      "epoch": 0.10209993892795065,
      "grad_norm": 0.8999176548241218,
      "learning_rate": 9.864295699090779e-06,
      "loss": 0.4487,
      "step": 1630
    },
    {
      "epoch": 0.10216257692729294,
      "grad_norm": 0.870678370015496,
      "learning_rate": 9.864060869032553e-06,
      "loss": 0.4349,
      "step": 1631
    },
    {
      "epoch": 0.10222521492663525,
      "grad_norm": 0.9111666325552786,
      "learning_rate": 9.863825838768678e-06,
      "loss": 0.4421,
      "step": 1632
    },
    {
      "epoch": 0.10228785292597754,
      "grad_norm": 0.8509554426305704,
      "learning_rate": 9.863590608308828e-06,
      "loss": 0.446,
      "step": 1633
    },
    {
      "epoch": 0.10235049092531985,
      "grad_norm": 0.8715878297339201,
      "learning_rate": 9.863355177662689e-06,
      "loss": 0.4531,
      "step": 1634
    },
    {
      "epoch": 0.10241312892466214,
      "grad_norm": 0.8239743870641836,
      "learning_rate": 9.863119546839945e-06,
      "loss": 0.4358,
      "step": 1635
    },
    {
      "epoch": 0.10247576692400445,
      "grad_norm": 0.8742994438071928,
      "learning_rate": 9.862883715850299e-06,
      "loss": 0.4716,
      "step": 1636
    },
    {
      "epoch": 0.10253840492334675,
      "grad_norm": 0.9677685685369322,
      "learning_rate": 9.862647684703457e-06,
      "loss": 0.4637,
      "step": 1637
    },
    {
      "epoch": 0.10260104292268905,
      "grad_norm": 0.8736481579576624,
      "learning_rate": 9.862411453409133e-06,
      "loss": 0.4356,
      "step": 1638
    },
    {
      "epoch": 0.10266368092203135,
      "grad_norm": 0.9229859984915212,
      "learning_rate": 9.862175021977053e-06,
      "loss": 0.4186,
      "step": 1639
    },
    {
      "epoch": 0.10272631892137365,
      "grad_norm": 0.9646057437216833,
      "learning_rate": 9.861938390416944e-06,
      "loss": 0.486,
      "step": 1640
    },
    {
      "epoch": 0.10278895692071595,
      "grad_norm": 0.9016565734915767,
      "learning_rate": 9.861701558738548e-06,
      "loss": 0.4742,
      "step": 1641
    },
    {
      "epoch": 0.10285159492005826,
      "grad_norm": 0.918072803481721,
      "learning_rate": 9.861464526951616e-06,
      "loss": 0.4272,
      "step": 1642
    },
    {
      "epoch": 0.10291423291940055,
      "grad_norm": 0.9346532552311103,
      "learning_rate": 9.861227295065901e-06,
      "loss": 0.437,
      "step": 1643
    },
    {
      "epoch": 0.10297687091874286,
      "grad_norm": 0.9147588719545384,
      "learning_rate": 9.860989863091168e-06,
      "loss": 0.4312,
      "step": 1644
    },
    {
      "epoch": 0.10303950891808515,
      "grad_norm": 0.881686582456261,
      "learning_rate": 9.860752231037189e-06,
      "loss": 0.4022,
      "step": 1645
    },
    {
      "epoch": 0.10310214691742746,
      "grad_norm": 0.9275460045466786,
      "learning_rate": 9.860514398913748e-06,
      "loss": 0.4599,
      "step": 1646
    },
    {
      "epoch": 0.10316478491676975,
      "grad_norm": 0.8935640524063782,
      "learning_rate": 9.86027636673063e-06,
      "loss": 0.4288,
      "step": 1647
    },
    {
      "epoch": 0.10322742291611206,
      "grad_norm": 0.9292660249184228,
      "learning_rate": 9.860038134497635e-06,
      "loss": 0.5014,
      "step": 1648
    },
    {
      "epoch": 0.10329006091545435,
      "grad_norm": 0.8641883373382413,
      "learning_rate": 9.859799702224571e-06,
      "loss": 0.4229,
      "step": 1649
    },
    {
      "epoch": 0.10335269891479666,
      "grad_norm": 0.7863905206583155,
      "learning_rate": 9.859561069921246e-06,
      "loss": 0.5183,
      "step": 1650
    },
    {
      "epoch": 0.10341533691413896,
      "grad_norm": 0.9634587644794977,
      "learning_rate": 9.859322237597488e-06,
      "loss": 0.4605,
      "step": 1651
    },
    {
      "epoch": 0.10347797491348126,
      "grad_norm": 0.9455039075557116,
      "learning_rate": 9.859083205263122e-06,
      "loss": 0.4123,
      "step": 1652
    },
    {
      "epoch": 0.10354061291282357,
      "grad_norm": 0.9692545167239823,
      "learning_rate": 9.858843972927992e-06,
      "loss": 0.4586,
      "step": 1653
    },
    {
      "epoch": 0.10360325091216586,
      "grad_norm": 0.8539553312350842,
      "learning_rate": 9.858604540601941e-06,
      "loss": 0.4044,
      "step": 1654
    },
    {
      "epoch": 0.10366588891150817,
      "grad_norm": 0.8611053881468008,
      "learning_rate": 9.858364908294826e-06,
      "loss": 0.4754,
      "step": 1655
    },
    {
      "epoch": 0.10372852691085047,
      "grad_norm": 0.8098606389588296,
      "learning_rate": 9.858125076016507e-06,
      "loss": 0.413,
      "step": 1656
    },
    {
      "epoch": 0.10379116491019277,
      "grad_norm": 0.8928764148044713,
      "learning_rate": 9.857885043776862e-06,
      "loss": 0.4532,
      "step": 1657
    },
    {
      "epoch": 0.10385380290953507,
      "grad_norm": 0.8560461608484777,
      "learning_rate": 9.857644811585764e-06,
      "loss": 0.4711,
      "step": 1658
    },
    {
      "epoch": 0.10391644090887737,
      "grad_norm": 1.0205851565970248,
      "learning_rate": 9.857404379453106e-06,
      "loss": 0.4214,
      "step": 1659
    },
    {
      "epoch": 0.10397907890821967,
      "grad_norm": 0.9777910485960564,
      "learning_rate": 9.85716374738878e-06,
      "loss": 0.4682,
      "step": 1660
    },
    {
      "epoch": 0.10404171690756198,
      "grad_norm": 1.1197939040366751,
      "learning_rate": 9.856922915402695e-06,
      "loss": 0.4949,
      "step": 1661
    },
    {
      "epoch": 0.10410435490690427,
      "grad_norm": 0.9511377142908352,
      "learning_rate": 9.85668188350476e-06,
      "loss": 0.4832,
      "step": 1662
    },
    {
      "epoch": 0.10416699290624658,
      "grad_norm": 0.8800495258207905,
      "learning_rate": 9.856440651704897e-06,
      "loss": 0.4681,
      "step": 1663
    },
    {
      "epoch": 0.10422963090558887,
      "grad_norm": 0.9522932188927004,
      "learning_rate": 9.856199220013037e-06,
      "loss": 0.478,
      "step": 1664
    },
    {
      "epoch": 0.10429226890493118,
      "grad_norm": 1.054456042521886,
      "learning_rate": 9.855957588439114e-06,
      "loss": 0.5094,
      "step": 1665
    },
    {
      "epoch": 0.10435490690427347,
      "grad_norm": 0.9388326455503359,
      "learning_rate": 9.855715756993074e-06,
      "loss": 0.4092,
      "step": 1666
    },
    {
      "epoch": 0.10441754490361578,
      "grad_norm": 1.1016712697483182,
      "learning_rate": 9.855473725684875e-06,
      "loss": 0.5068,
      "step": 1667
    },
    {
      "epoch": 0.10448018290295807,
      "grad_norm": 0.8790681057726103,
      "learning_rate": 9.855231494524475e-06,
      "loss": 0.4233,
      "step": 1668
    },
    {
      "epoch": 0.10454282090230038,
      "grad_norm": 0.8998848505308056,
      "learning_rate": 9.854989063521846e-06,
      "loss": 0.4077,
      "step": 1669
    },
    {
      "epoch": 0.10460545890164268,
      "grad_norm": 0.9353527747599285,
      "learning_rate": 9.854746432686965e-06,
      "loss": 0.4597,
      "step": 1670
    },
    {
      "epoch": 0.10466809690098498,
      "grad_norm": 0.9155795233593754,
      "learning_rate": 9.854503602029821e-06,
      "loss": 0.3945,
      "step": 1671
    },
    {
      "epoch": 0.10473073490032728,
      "grad_norm": 0.8980389780944664,
      "learning_rate": 9.854260571560408e-06,
      "loss": 0.4396,
      "step": 1672
    },
    {
      "epoch": 0.10479337289966958,
      "grad_norm": 0.8565074431516235,
      "learning_rate": 9.854017341288728e-06,
      "loss": 0.4275,
      "step": 1673
    },
    {
      "epoch": 0.10485601089901189,
      "grad_norm": 0.8903065673027977,
      "learning_rate": 9.853773911224794e-06,
      "loss": 0.4332,
      "step": 1674
    },
    {
      "epoch": 0.10491864889835419,
      "grad_norm": 0.8599586996257699,
      "learning_rate": 9.853530281378623e-06,
      "loss": 0.4441,
      "step": 1675
    },
    {
      "epoch": 0.1049812868976965,
      "grad_norm": 0.8708904812350786,
      "learning_rate": 9.853286451760246e-06,
      "loss": 0.4173,
      "step": 1676
    },
    {
      "epoch": 0.10504392489703879,
      "grad_norm": 0.9565777861560476,
      "learning_rate": 9.853042422379698e-06,
      "loss": 0.4502,
      "step": 1677
    },
    {
      "epoch": 0.1051065628963811,
      "grad_norm": 0.9157878551033322,
      "learning_rate": 9.852798193247024e-06,
      "loss": 0.4912,
      "step": 1678
    },
    {
      "epoch": 0.10516920089572339,
      "grad_norm": 0.9041663690420485,
      "learning_rate": 9.852553764372275e-06,
      "loss": 0.4337,
      "step": 1679
    },
    {
      "epoch": 0.1052318388950657,
      "grad_norm": 0.9531355914253057,
      "learning_rate": 9.85230913576551e-06,
      "loss": 0.4314,
      "step": 1680
    },
    {
      "epoch": 0.10529447689440799,
      "grad_norm": 0.8303868159760173,
      "learning_rate": 9.852064307436803e-06,
      "loss": 0.4241,
      "step": 1681
    },
    {
      "epoch": 0.1053571148937503,
      "grad_norm": 0.8009388689403389,
      "learning_rate": 9.851819279396228e-06,
      "loss": 0.3715,
      "step": 1682
    },
    {
      "epoch": 0.10541975289309259,
      "grad_norm": 0.911416249993186,
      "learning_rate": 9.851574051653871e-06,
      "loss": 0.4084,
      "step": 1683
    },
    {
      "epoch": 0.1054823908924349,
      "grad_norm": 0.9012310852151912,
      "learning_rate": 9.851328624219826e-06,
      "loss": 0.4853,
      "step": 1684
    },
    {
      "epoch": 0.10554502889177719,
      "grad_norm": 0.8130258725707079,
      "learning_rate": 9.851082997104194e-06,
      "loss": 0.4034,
      "step": 1685
    },
    {
      "epoch": 0.1056076668911195,
      "grad_norm": 0.8600995074922225,
      "learning_rate": 9.850837170317084e-06,
      "loss": 0.4274,
      "step": 1686
    },
    {
      "epoch": 0.1056703048904618,
      "grad_norm": 0.9493935449244991,
      "learning_rate": 9.850591143868618e-06,
      "loss": 0.4726,
      "step": 1687
    },
    {
      "epoch": 0.1057329428898041,
      "grad_norm": 0.9932795429894357,
      "learning_rate": 9.85034491776892e-06,
      "loss": 0.4548,
      "step": 1688
    },
    {
      "epoch": 0.1057955808891464,
      "grad_norm": 0.8984649291418174,
      "learning_rate": 9.850098492028122e-06,
      "loss": 0.4618,
      "step": 1689
    },
    {
      "epoch": 0.1058582188884887,
      "grad_norm": 0.855911141728444,
      "learning_rate": 9.849851866656373e-06,
      "loss": 0.4335,
      "step": 1690
    },
    {
      "epoch": 0.105920856887831,
      "grad_norm": 0.9604817416382047,
      "learning_rate": 9.84960504166382e-06,
      "loss": 0.4599,
      "step": 1691
    },
    {
      "epoch": 0.1059834948871733,
      "grad_norm": 0.9105369643646842,
      "learning_rate": 9.849358017060622e-06,
      "loss": 0.417,
      "step": 1692
    },
    {
      "epoch": 0.1060461328865156,
      "grad_norm": 1.2907743981203437,
      "learning_rate": 9.84911079285695e-06,
      "loss": 0.4412,
      "step": 1693
    },
    {
      "epoch": 0.1061087708858579,
      "grad_norm": 0.8565494530855996,
      "learning_rate": 9.848863369062977e-06,
      "loss": 0.4341,
      "step": 1694
    },
    {
      "epoch": 0.10617140888520021,
      "grad_norm": 0.9214419807626887,
      "learning_rate": 9.848615745688888e-06,
      "loss": 0.4332,
      "step": 1695
    },
    {
      "epoch": 0.10623404688454251,
      "grad_norm": 0.8397219296147146,
      "learning_rate": 9.848367922744874e-06,
      "loss": 0.4161,
      "step": 1696
    },
    {
      "epoch": 0.10629668488388481,
      "grad_norm": 0.8741640542570748,
      "learning_rate": 9.848119900241138e-06,
      "loss": 0.3901,
      "step": 1697
    },
    {
      "epoch": 0.10635932288322711,
      "grad_norm": 0.8831692285470235,
      "learning_rate": 9.847871678187885e-06,
      "loss": 0.4401,
      "step": 1698
    },
    {
      "epoch": 0.10642196088256942,
      "grad_norm": 0.8007718646736729,
      "learning_rate": 9.847623256595336e-06,
      "loss": 0.4134,
      "step": 1699
    },
    {
      "epoch": 0.10648459888191171,
      "grad_norm": 0.9303942601553514,
      "learning_rate": 9.847374635473713e-06,
      "loss": 0.4877,
      "step": 1700
    },
    {
      "epoch": 0.10654723688125402,
      "grad_norm": 0.8978394903296093,
      "learning_rate": 9.84712581483325e-06,
      "loss": 0.4511,
      "step": 1701
    },
    {
      "epoch": 0.10660987488059631,
      "grad_norm": 0.9147950077820368,
      "learning_rate": 9.846876794684188e-06,
      "loss": 0.4291,
      "step": 1702
    },
    {
      "epoch": 0.10667251287993862,
      "grad_norm": 0.8631361194654336,
      "learning_rate": 9.846627575036779e-06,
      "loss": 0.4442,
      "step": 1703
    },
    {
      "epoch": 0.10673515087928091,
      "grad_norm": 0.9856092630483868,
      "learning_rate": 9.84637815590128e-06,
      "loss": 0.5205,
      "step": 1704
    },
    {
      "epoch": 0.10679778887862322,
      "grad_norm": 0.8835227542571434,
      "learning_rate": 9.846128537287954e-06,
      "loss": 0.4223,
      "step": 1705
    },
    {
      "epoch": 0.10686042687796551,
      "grad_norm": 0.898073908912535,
      "learning_rate": 9.84587871920708e-06,
      "loss": 0.4831,
      "step": 1706
    },
    {
      "epoch": 0.10692306487730782,
      "grad_norm": 0.9163370923733389,
      "learning_rate": 9.845628701668938e-06,
      "loss": 0.4674,
      "step": 1707
    },
    {
      "epoch": 0.10698570287665012,
      "grad_norm": 0.9204438506746777,
      "learning_rate": 9.845378484683818e-06,
      "loss": 0.4128,
      "step": 1708
    },
    {
      "epoch": 0.10704834087599242,
      "grad_norm": 0.8284473774769301,
      "learning_rate": 9.845128068262023e-06,
      "loss": 0.4274,
      "step": 1709
    },
    {
      "epoch": 0.10711097887533472,
      "grad_norm": 0.843847485275788,
      "learning_rate": 9.844877452413855e-06,
      "loss": 0.4366,
      "step": 1710
    },
    {
      "epoch": 0.10717361687467702,
      "grad_norm": 0.9245624314526578,
      "learning_rate": 9.844626637149633e-06,
      "loss": 0.4577,
      "step": 1711
    },
    {
      "epoch": 0.10723625487401932,
      "grad_norm": 0.9086965655883255,
      "learning_rate": 9.844375622479678e-06,
      "loss": 0.4517,
      "step": 1712
    },
    {
      "epoch": 0.10729889287336163,
      "grad_norm": 0.8766304336561018,
      "learning_rate": 9.844124408414325e-06,
      "loss": 0.4533,
      "step": 1713
    },
    {
      "epoch": 0.10736153087270392,
      "grad_norm": 0.8873359125326057,
      "learning_rate": 9.843872994963912e-06,
      "loss": 0.4555,
      "step": 1714
    },
    {
      "epoch": 0.10742416887204623,
      "grad_norm": 0.8632025998144776,
      "learning_rate": 9.843621382138786e-06,
      "loss": 0.4096,
      "step": 1715
    },
    {
      "epoch": 0.10748680687138852,
      "grad_norm": 0.851485841302409,
      "learning_rate": 9.843369569949305e-06,
      "loss": 0.4682,
      "step": 1716
    },
    {
      "epoch": 0.10754944487073083,
      "grad_norm": 0.8881083174316393,
      "learning_rate": 9.843117558405836e-06,
      "loss": 0.4451,
      "step": 1717
    },
    {
      "epoch": 0.10761208287007314,
      "grad_norm": 0.8988576874504488,
      "learning_rate": 9.842865347518748e-06,
      "loss": 0.4268,
      "step": 1718
    },
    {
      "epoch": 0.10767472086941543,
      "grad_norm": 0.9628945791785181,
      "learning_rate": 9.842612937298422e-06,
      "loss": 0.4303,
      "step": 1719
    },
    {
      "epoch": 0.10773735886875774,
      "grad_norm": 0.9973766002851463,
      "learning_rate": 9.84236032775525e-06,
      "loss": 0.4435,
      "step": 1720
    },
    {
      "epoch": 0.10779999686810003,
      "grad_norm": 0.9011858740406491,
      "learning_rate": 9.842107518899628e-06,
      "loss": 0.4349,
      "step": 1721
    },
    {
      "epoch": 0.10786263486744234,
      "grad_norm": 0.8859749054523879,
      "learning_rate": 9.841854510741962e-06,
      "loss": 0.4333,
      "step": 1722
    },
    {
      "epoch": 0.10792527286678463,
      "grad_norm": 0.8594753411421615,
      "learning_rate": 9.841601303292664e-06,
      "loss": 0.4222,
      "step": 1723
    },
    {
      "epoch": 0.10798791086612694,
      "grad_norm": 0.9130734500730797,
      "learning_rate": 9.84134789656216e-06,
      "loss": 0.4623,
      "step": 1724
    },
    {
      "epoch": 0.10805054886546923,
      "grad_norm": 0.8043543914739791,
      "learning_rate": 9.841094290560878e-06,
      "loss": 0.4331,
      "step": 1725
    },
    {
      "epoch": 0.10811318686481154,
      "grad_norm": 0.7725050648198681,
      "learning_rate": 9.840840485299255e-06,
      "loss": 0.47,
      "step": 1726
    },
    {
      "epoch": 0.10817582486415384,
      "grad_norm": 0.8644720260195033,
      "learning_rate": 9.840586480787741e-06,
      "loss": 0.4829,
      "step": 1727
    },
    {
      "epoch": 0.10823846286349614,
      "grad_norm": 0.9566882242234561,
      "learning_rate": 9.840332277036786e-06,
      "loss": 0.4273,
      "step": 1728
    },
    {
      "epoch": 0.10830110086283844,
      "grad_norm": 0.9761980142861405,
      "learning_rate": 9.840077874056859e-06,
      "loss": 0.452,
      "step": 1729
    },
    {
      "epoch": 0.10836373886218074,
      "grad_norm": 0.8719169850149026,
      "learning_rate": 9.839823271858426e-06,
      "loss": 0.4648,
      "step": 1730
    },
    {
      "epoch": 0.10842637686152304,
      "grad_norm": 0.8969050921120731,
      "learning_rate": 9.83956847045197e-06,
      "loss": 0.4516,
      "step": 1731
    },
    {
      "epoch": 0.10848901486086535,
      "grad_norm": 0.828756635376854,
      "learning_rate": 9.839313469847978e-06,
      "loss": 0.4608,
      "step": 1732
    },
    {
      "epoch": 0.10855165286020764,
      "grad_norm": 0.9081884583948283,
      "learning_rate": 9.839058270056945e-06,
      "loss": 0.4678,
      "step": 1733
    },
    {
      "epoch": 0.10861429085954995,
      "grad_norm": 0.9144341872695031,
      "learning_rate": 9.838802871089374e-06,
      "loss": 0.5011,
      "step": 1734
    },
    {
      "epoch": 0.10867692885889224,
      "grad_norm": 0.9119872641598379,
      "learning_rate": 9.83854727295578e-06,
      "loss": 0.4492,
      "step": 1735
    },
    {
      "epoch": 0.10873956685823455,
      "grad_norm": 0.854625886087694,
      "learning_rate": 9.838291475666684e-06,
      "loss": 0.4779,
      "step": 1736
    },
    {
      "epoch": 0.10880220485757684,
      "grad_norm": 0.9165727896008979,
      "learning_rate": 9.83803547923261e-06,
      "loss": 0.4366,
      "step": 1737
    },
    {
      "epoch": 0.10886484285691915,
      "grad_norm": 1.0070723316251193,
      "learning_rate": 9.837779283664099e-06,
      "loss": 0.4722,
      "step": 1738
    },
    {
      "epoch": 0.10892748085626146,
      "grad_norm": 0.8887584049947701,
      "learning_rate": 9.837522888971694e-06,
      "loss": 0.4933,
      "step": 1739
    },
    {
      "epoch": 0.10899011885560375,
      "grad_norm": 1.0945554215099949,
      "learning_rate": 9.837266295165949e-06,
      "loss": 0.4461,
      "step": 1740
    },
    {
      "epoch": 0.10905275685494606,
      "grad_norm": 0.9280059674521232,
      "learning_rate": 9.837009502257426e-06,
      "loss": 0.47,
      "step": 1741
    },
    {
      "epoch": 0.10911539485428835,
      "grad_norm": 0.9009191779341991,
      "learning_rate": 9.836752510256694e-06,
      "loss": 0.4866,
      "step": 1742
    },
    {
      "epoch": 0.10917803285363066,
      "grad_norm": 0.9573147622727864,
      "learning_rate": 9.836495319174331e-06,
      "loss": 0.4811,
      "step": 1743
    },
    {
      "epoch": 0.10924067085297295,
      "grad_norm": 0.8880653453612596,
      "learning_rate": 9.836237929020922e-06,
      "loss": 0.4774,
      "step": 1744
    },
    {
      "epoch": 0.10930330885231526,
      "grad_norm": 0.9006539444317118,
      "learning_rate": 9.835980339807063e-06,
      "loss": 0.4445,
      "step": 1745
    },
    {
      "epoch": 0.10936594685165756,
      "grad_norm": 0.8500135235833722,
      "learning_rate": 9.835722551543357e-06,
      "loss": 0.4486,
      "step": 1746
    },
    {
      "epoch": 0.10942858485099986,
      "grad_norm": 1.015211072333527,
      "learning_rate": 9.83546456424041e-06,
      "loss": 0.5279,
      "step": 1747
    },
    {
      "epoch": 0.10949122285034216,
      "grad_norm": 1.0234548982738485,
      "learning_rate": 9.835206377908845e-06,
      "loss": 0.4641,
      "step": 1748
    },
    {
      "epoch": 0.10955386084968446,
      "grad_norm": 0.8749914250972367,
      "learning_rate": 9.834947992559291e-06,
      "loss": 0.4098,
      "step": 1749
    },
    {
      "epoch": 0.10961649884902676,
      "grad_norm": 0.952860909544951,
      "learning_rate": 9.834689408202378e-06,
      "loss": 0.4473,
      "step": 1750
    },
    {
      "epoch": 0.10967913684836907,
      "grad_norm": 0.8539255868737581,
      "learning_rate": 9.834430624848752e-06,
      "loss": 0.4826,
      "step": 1751
    },
    {
      "epoch": 0.10974177484771136,
      "grad_norm": 0.8779646998854063,
      "learning_rate": 9.834171642509063e-06,
      "loss": 0.4498,
      "step": 1752
    },
    {
      "epoch": 0.10980441284705367,
      "grad_norm": 0.9406805526413583,
      "learning_rate": 9.833912461193973e-06,
      "loss": 0.4226,
      "step": 1753
    },
    {
      "epoch": 0.10986705084639596,
      "grad_norm": 0.8777633701092064,
      "learning_rate": 9.833653080914149e-06,
      "loss": 0.4091,
      "step": 1754
    },
    {
      "epoch": 0.10992968884573827,
      "grad_norm": 0.8776340875297767,
      "learning_rate": 9.833393501680266e-06,
      "loss": 0.4565,
      "step": 1755
    },
    {
      "epoch": 0.10999232684508056,
      "grad_norm": 0.8082185189860553,
      "learning_rate": 9.83313372350301e-06,
      "loss": 0.4007,
      "step": 1756
    },
    {
      "epoch": 0.11005496484442287,
      "grad_norm": 0.9346803961552255,
      "learning_rate": 9.832873746393073e-06,
      "loss": 0.4306,
      "step": 1757
    },
    {
      "epoch": 0.11011760284376516,
      "grad_norm": 0.8504990623853856,
      "learning_rate": 9.832613570361157e-06,
      "loss": 0.4223,
      "step": 1758
    },
    {
      "epoch": 0.11018024084310747,
      "grad_norm": 0.9068985625592745,
      "learning_rate": 9.832353195417967e-06,
      "loss": 0.4426,
      "step": 1759
    },
    {
      "epoch": 0.11024287884244978,
      "grad_norm": 0.8855831518129471,
      "learning_rate": 9.832092621574224e-06,
      "loss": 0.4122,
      "step": 1760
    },
    {
      "epoch": 0.11030551684179207,
      "grad_norm": 0.9219516343181697,
      "learning_rate": 9.83183184884065e-06,
      "loss": 0.4282,
      "step": 1761
    },
    {
      "epoch": 0.11036815484113438,
      "grad_norm": 0.8799376637555896,
      "learning_rate": 9.831570877227983e-06,
      "loss": 0.4723,
      "step": 1762
    },
    {
      "epoch": 0.11043079284047667,
      "grad_norm": 0.87486849346039,
      "learning_rate": 9.831309706746959e-06,
      "loss": 0.4365,
      "step": 1763
    },
    {
      "epoch": 0.11049343083981898,
      "grad_norm": 0.8654277231578429,
      "learning_rate": 9.831048337408333e-06,
      "loss": 0.4623,
      "step": 1764
    },
    {
      "epoch": 0.11055606883916128,
      "grad_norm": 0.8444809191854944,
      "learning_rate": 9.83078676922286e-06,
      "loss": 0.4785,
      "step": 1765
    },
    {
      "epoch": 0.11061870683850358,
      "grad_norm": 1.1293142437366843,
      "learning_rate": 9.830525002201307e-06,
      "loss": 0.4515,
      "step": 1766
    },
    {
      "epoch": 0.11068134483784588,
      "grad_norm": 0.8900615490833907,
      "learning_rate": 9.830263036354448e-06,
      "loss": 0.4466,
      "step": 1767
    },
    {
      "epoch": 0.11074398283718818,
      "grad_norm": 0.8652821195140946,
      "learning_rate": 9.830000871693065e-06,
      "loss": 0.4259,
      "step": 1768
    },
    {
      "epoch": 0.11080662083653048,
      "grad_norm": 0.8840333613642132,
      "learning_rate": 9.829738508227953e-06,
      "loss": 0.4629,
      "step": 1769
    },
    {
      "epoch": 0.11086925883587279,
      "grad_norm": 0.8125703140841185,
      "learning_rate": 9.829475945969904e-06,
      "loss": 0.4061,
      "step": 1770
    },
    {
      "epoch": 0.11093189683521508,
      "grad_norm": 0.8641317274558977,
      "learning_rate": 9.82921318492973e-06,
      "loss": 0.4103,
      "step": 1771
    },
    {
      "epoch": 0.11099453483455739,
      "grad_norm": 0.8877273206625376,
      "learning_rate": 9.828950225118245e-06,
      "loss": 0.434,
      "step": 1772
    },
    {
      "epoch": 0.11105717283389968,
      "grad_norm": 0.8753463300069221,
      "learning_rate": 9.828687066546273e-06,
      "loss": 0.4823,
      "step": 1773
    },
    {
      "epoch": 0.11111981083324199,
      "grad_norm": 0.8665560786318306,
      "learning_rate": 9.828423709224644e-06,
      "loss": 0.4659,
      "step": 1774
    },
    {
      "epoch": 0.11118244883258428,
      "grad_norm": 0.8536529725184699,
      "learning_rate": 9.8281601531642e-06,
      "loss": 0.451,
      "step": 1775
    },
    {
      "epoch": 0.11124508683192659,
      "grad_norm": 0.94319748321395,
      "learning_rate": 9.827896398375787e-06,
      "loss": 0.428,
      "step": 1776
    },
    {
      "epoch": 0.11130772483126888,
      "grad_norm": 0.9329199143995318,
      "learning_rate": 9.827632444870264e-06,
      "loss": 0.4979,
      "step": 1777
    },
    {
      "epoch": 0.11137036283061119,
      "grad_norm": 1.0859475793029072,
      "learning_rate": 9.82736829265849e-06,
      "loss": 0.4507,
      "step": 1778
    },
    {
      "epoch": 0.11143300082995349,
      "grad_norm": 0.8469204274642447,
      "learning_rate": 9.827103941751344e-06,
      "loss": 0.427,
      "step": 1779
    },
    {
      "epoch": 0.11149563882929579,
      "grad_norm": 0.7489782166092357,
      "learning_rate": 9.826839392159702e-06,
      "loss": 0.4747,
      "step": 1780
    },
    {
      "epoch": 0.11155827682863809,
      "grad_norm": 0.9021802736852541,
      "learning_rate": 9.826574643894457e-06,
      "loss": 0.4451,
      "step": 1781
    },
    {
      "epoch": 0.1116209148279804,
      "grad_norm": 0.9054750028292631,
      "learning_rate": 9.826309696966503e-06,
      "loss": 0.4714,
      "step": 1782
    },
    {
      "epoch": 0.1116835528273227,
      "grad_norm": 0.9119500267574243,
      "learning_rate": 9.826044551386743e-06,
      "loss": 0.4595,
      "step": 1783
    },
    {
      "epoch": 0.111746190826665,
      "grad_norm": 0.8365916935888272,
      "learning_rate": 9.825779207166098e-06,
      "loss": 0.4039,
      "step": 1784
    },
    {
      "epoch": 0.1118088288260073,
      "grad_norm": 0.8685061048231827,
      "learning_rate": 9.825513664315482e-06,
      "loss": 0.496,
      "step": 1785
    },
    {
      "epoch": 0.1118714668253496,
      "grad_norm": 0.8631829656525766,
      "learning_rate": 9.825247922845829e-06,
      "loss": 0.4512,
      "step": 1786
    },
    {
      "epoch": 0.1119341048246919,
      "grad_norm": 0.9195597415227157,
      "learning_rate": 9.824981982768076e-06,
      "loss": 0.4241,
      "step": 1787
    },
    {
      "epoch": 0.1119967428240342,
      "grad_norm": 0.8202935239095179,
      "learning_rate": 9.824715844093169e-06,
      "loss": 0.416,
      "step": 1788
    },
    {
      "epoch": 0.1120593808233765,
      "grad_norm": 0.8842601017209399,
      "learning_rate": 9.824449506832063e-06,
      "loss": 0.4618,
      "step": 1789
    },
    {
      "epoch": 0.1121220188227188,
      "grad_norm": 0.8668653731511038,
      "learning_rate": 9.82418297099572e-06,
      "loss": 0.4359,
      "step": 1790
    },
    {
      "epoch": 0.11218465682206111,
      "grad_norm": 0.8767730634735029,
      "learning_rate": 9.823916236595108e-06,
      "loss": 0.4416,
      "step": 1791
    },
    {
      "epoch": 0.1122472948214034,
      "grad_norm": 0.9289844726162338,
      "learning_rate": 9.82364930364121e-06,
      "loss": 0.4888,
      "step": 1792
    },
    {
      "epoch": 0.11230993282074571,
      "grad_norm": 0.8563483511828351,
      "learning_rate": 9.823382172145012e-06,
      "loss": 0.4164,
      "step": 1793
    },
    {
      "epoch": 0.112372570820088,
      "grad_norm": 0.780248997620945,
      "learning_rate": 9.823114842117507e-06,
      "loss": 0.4463,
      "step": 1794
    },
    {
      "epoch": 0.11243520881943031,
      "grad_norm": 0.8111741736554748,
      "learning_rate": 9.822847313569702e-06,
      "loss": 0.423,
      "step": 1795
    },
    {
      "epoch": 0.1124978468187726,
      "grad_norm": 0.8404707136833853,
      "learning_rate": 9.822579586512603e-06,
      "loss": 0.4513,
      "step": 1796
    },
    {
      "epoch": 0.11256048481811491,
      "grad_norm": 0.8454849622200288,
      "learning_rate": 9.822311660957238e-06,
      "loss": 0.4078,
      "step": 1797
    },
    {
      "epoch": 0.1126231228174572,
      "grad_norm": 0.8107745083119459,
      "learning_rate": 9.822043536914627e-06,
      "loss": 0.4208,
      "step": 1798
    },
    {
      "epoch": 0.11268576081679951,
      "grad_norm": 0.8383428811095872,
      "learning_rate": 9.821775214395807e-06,
      "loss": 0.4327,
      "step": 1799
    },
    {
      "epoch": 0.1127483988161418,
      "grad_norm": 0.9677545367218037,
      "learning_rate": 9.821506693411829e-06,
      "loss": 0.4616,
      "step": 1800
    },
    {
      "epoch": 0.11281103681548411,
      "grad_norm": 0.8924349795888481,
      "learning_rate": 9.821237973973736e-06,
      "loss": 0.4334,
      "step": 1801
    },
    {
      "epoch": 0.11287367481482641,
      "grad_norm": 0.9338246917050314,
      "learning_rate": 9.820969056092597e-06,
      "loss": 0.4324,
      "step": 1802
    },
    {
      "epoch": 0.11293631281416872,
      "grad_norm": 0.9125304707959679,
      "learning_rate": 9.820699939779475e-06,
      "loss": 0.4436,
      "step": 1803
    },
    {
      "epoch": 0.11299895081351102,
      "grad_norm": 0.8977270023078138,
      "learning_rate": 9.820430625045448e-06,
      "loss": 0.4443,
      "step": 1804
    },
    {
      "epoch": 0.11306158881285332,
      "grad_norm": 0.8553912431816587,
      "learning_rate": 9.820161111901604e-06,
      "loss": 0.3957,
      "step": 1805
    },
    {
      "epoch": 0.11312422681219562,
      "grad_norm": 0.9137765647527338,
      "learning_rate": 9.819891400359031e-06,
      "loss": 0.4865,
      "step": 1806
    },
    {
      "epoch": 0.11318686481153792,
      "grad_norm": 0.9094275204817321,
      "learning_rate": 9.819621490428835e-06,
      "loss": 0.4605,
      "step": 1807
    },
    {
      "epoch": 0.11324950281088023,
      "grad_norm": 0.8915574852316523,
      "learning_rate": 9.819351382122125e-06,
      "loss": 0.4228,
      "step": 1808
    },
    {
      "epoch": 0.11331214081022252,
      "grad_norm": 0.8394056472734308,
      "learning_rate": 9.819081075450014e-06,
      "loss": 0.4213,
      "step": 1809
    },
    {
      "epoch": 0.11337477880956483,
      "grad_norm": 0.9438529646140494,
      "learning_rate": 9.818810570423634e-06,
      "loss": 0.4446,
      "step": 1810
    },
    {
      "epoch": 0.11343741680890712,
      "grad_norm": 0.8567747173963876,
      "learning_rate": 9.818539867054117e-06,
      "loss": 0.4417,
      "step": 1811
    },
    {
      "epoch": 0.11350005480824943,
      "grad_norm": 0.8865814135217905,
      "learning_rate": 9.818268965352604e-06,
      "loss": 0.4586,
      "step": 1812
    },
    {
      "epoch": 0.11356269280759172,
      "grad_norm": 0.8615889312188874,
      "learning_rate": 9.817997865330246e-06,
      "loss": 0.441,
      "step": 1813
    },
    {
      "epoch": 0.11362533080693403,
      "grad_norm": 0.8075814745515107,
      "learning_rate": 9.817726566998202e-06,
      "loss": 0.4497,
      "step": 1814
    },
    {
      "epoch": 0.11368796880627632,
      "grad_norm": 0.9338161023367897,
      "learning_rate": 9.817455070367637e-06,
      "loss": 0.4573,
      "step": 1815
    },
    {
      "epoch": 0.11375060680561863,
      "grad_norm": 0.8819476649983128,
      "learning_rate": 9.817183375449729e-06,
      "loss": 0.4471,
      "step": 1816
    },
    {
      "epoch": 0.11381324480496093,
      "grad_norm": 0.839917304445873,
      "learning_rate": 9.816911482255658e-06,
      "loss": 0.4249,
      "step": 1817
    },
    {
      "epoch": 0.11387588280430323,
      "grad_norm": 0.800213665244549,
      "learning_rate": 9.816639390796616e-06,
      "loss": 0.4507,
      "step": 1818
    },
    {
      "epoch": 0.11393852080364553,
      "grad_norm": 0.9242563535970489,
      "learning_rate": 9.816367101083803e-06,
      "loss": 0.431,
      "step": 1819
    },
    {
      "epoch": 0.11400115880298783,
      "grad_norm": 0.8499915501654253,
      "learning_rate": 9.816094613128427e-06,
      "loss": 0.4161,
      "step": 1820
    },
    {
      "epoch": 0.11406379680233013,
      "grad_norm": 0.9061046785795988,
      "learning_rate": 9.815821926941703e-06,
      "loss": 0.4407,
      "step": 1821
    },
    {
      "epoch": 0.11412643480167244,
      "grad_norm": 0.8782420244272203,
      "learning_rate": 9.815549042534854e-06,
      "loss": 0.4231,
      "step": 1822
    },
    {
      "epoch": 0.11418907280101473,
      "grad_norm": 0.877207147384595,
      "learning_rate": 9.815275959919112e-06,
      "loss": 0.4747,
      "step": 1823
    },
    {
      "epoch": 0.11425171080035704,
      "grad_norm": 0.9444735375534098,
      "learning_rate": 9.815002679105718e-06,
      "loss": 0.447,
      "step": 1824
    },
    {
      "epoch": 0.11431434879969933,
      "grad_norm": 0.9840523683625316,
      "learning_rate": 9.81472920010592e-06,
      "loss": 0.4798,
      "step": 1825
    },
    {
      "epoch": 0.11437698679904164,
      "grad_norm": 0.888144664773925,
      "learning_rate": 9.814455522930975e-06,
      "loss": 0.4664,
      "step": 1826
    },
    {
      "epoch": 0.11443962479838395,
      "grad_norm": 0.8732070494029672,
      "learning_rate": 9.814181647592146e-06,
      "loss": 0.4329,
      "step": 1827
    },
    {
      "epoch": 0.11450226279772624,
      "grad_norm": 0.8915971083823488,
      "learning_rate": 9.813907574100709e-06,
      "loss": 0.4463,
      "step": 1828
    },
    {
      "epoch": 0.11456490079706855,
      "grad_norm": 0.7935610782977627,
      "learning_rate": 9.81363330246794e-06,
      "loss": 0.4885,
      "step": 1829
    },
    {
      "epoch": 0.11462753879641084,
      "grad_norm": 0.9061445728996753,
      "learning_rate": 9.813358832705133e-06,
      "loss": 0.4571,
      "step": 1830
    },
    {
      "epoch": 0.11469017679575315,
      "grad_norm": 1.0162545580906128,
      "learning_rate": 9.813084164823583e-06,
      "loss": 0.4604,
      "step": 1831
    },
    {
      "epoch": 0.11475281479509544,
      "grad_norm": 0.8368103257381461,
      "learning_rate": 9.812809298834594e-06,
      "loss": 0.4417,
      "step": 1832
    },
    {
      "epoch": 0.11481545279443775,
      "grad_norm": 0.9126132060443275,
      "learning_rate": 9.812534234749482e-06,
      "loss": 0.4939,
      "step": 1833
    },
    {
      "epoch": 0.11487809079378004,
      "grad_norm": 0.8453153253313673,
      "learning_rate": 9.812258972579567e-06,
      "loss": 0.4521,
      "step": 1834
    },
    {
      "epoch": 0.11494072879312235,
      "grad_norm": 0.7965715860216054,
      "learning_rate": 9.811983512336178e-06,
      "loss": 0.4736,
      "step": 1835
    },
    {
      "epoch": 0.11500336679246465,
      "grad_norm": 0.8912920536376826,
      "learning_rate": 9.811707854030656e-06,
      "loss": 0.4519,
      "step": 1836
    },
    {
      "epoch": 0.11506600479180695,
      "grad_norm": 0.9492911375361305,
      "learning_rate": 9.811431997674347e-06,
      "loss": 0.4384,
      "step": 1837
    },
    {
      "epoch": 0.11512864279114925,
      "grad_norm": 0.8292140892331655,
      "learning_rate": 9.811155943278603e-06,
      "loss": 0.4338,
      "step": 1838
    },
    {
      "epoch": 0.11519128079049155,
      "grad_norm": 0.7740482895225712,
      "learning_rate": 9.810879690854787e-06,
      "loss": 0.4802,
      "step": 1839
    },
    {
      "epoch": 0.11525391878983385,
      "grad_norm": 0.8095614305250138,
      "learning_rate": 9.81060324041427e-06,
      "loss": 0.4345,
      "step": 1840
    },
    {
      "epoch": 0.11531655678917616,
      "grad_norm": 0.8746284578008088,
      "learning_rate": 9.810326591968431e-06,
      "loss": 0.4382,
      "step": 1841
    },
    {
      "epoch": 0.11537919478851845,
      "grad_norm": 0.8577072766322007,
      "learning_rate": 9.810049745528656e-06,
      "loss": 0.4574,
      "step": 1842
    },
    {
      "epoch": 0.11544183278786076,
      "grad_norm": 1.02853805132689,
      "learning_rate": 9.809772701106342e-06,
      "loss": 0.4618,
      "step": 1843
    },
    {
      "epoch": 0.11550447078720305,
      "grad_norm": 0.9983689663120909,
      "learning_rate": 9.80949545871289e-06,
      "loss": 0.407,
      "step": 1844
    },
    {
      "epoch": 0.11556710878654536,
      "grad_norm": 1.0555470442906645,
      "learning_rate": 9.809218018359712e-06,
      "loss": 0.4574,
      "step": 1845
    },
    {
      "epoch": 0.11562974678588765,
      "grad_norm": 0.8873253127472138,
      "learning_rate": 9.808940380058226e-06,
      "loss": 0.4692,
      "step": 1846
    },
    {
      "epoch": 0.11569238478522996,
      "grad_norm": 0.9117737700902694,
      "learning_rate": 9.808662543819864e-06,
      "loss": 0.4317,
      "step": 1847
    },
    {
      "epoch": 0.11575502278457227,
      "grad_norm": 0.8092209775339054,
      "learning_rate": 9.808384509656059e-06,
      "loss": 0.387,
      "step": 1848
    },
    {
      "epoch": 0.11581766078391456,
      "grad_norm": 0.8869226847897876,
      "learning_rate": 9.808106277578255e-06,
      "loss": 0.4172,
      "step": 1849
    },
    {
      "epoch": 0.11588029878325687,
      "grad_norm": 0.8657560217405227,
      "learning_rate": 9.807827847597903e-06,
      "loss": 0.4242,
      "step": 1850
    },
    {
      "epoch": 0.11594293678259916,
      "grad_norm": 0.8826669184181659,
      "learning_rate": 9.807549219726467e-06,
      "loss": 0.4508,
      "step": 1851
    },
    {
      "epoch": 0.11600557478194147,
      "grad_norm": 0.9204109756552029,
      "learning_rate": 9.807270393975412e-06,
      "loss": 0.4877,
      "step": 1852
    },
    {
      "epoch": 0.11606821278128376,
      "grad_norm": 0.8920022310876107,
      "learning_rate": 9.806991370356213e-06,
      "loss": 0.4083,
      "step": 1853
    },
    {
      "epoch": 0.11613085078062607,
      "grad_norm": 0.827755862476535,
      "learning_rate": 9.806712148880359e-06,
      "loss": 0.4348,
      "step": 1854
    },
    {
      "epoch": 0.11619348877996837,
      "grad_norm": 0.8843952033547905,
      "learning_rate": 9.80643272955934e-06,
      "loss": 0.4017,
      "step": 1855
    },
    {
      "epoch": 0.11625612677931067,
      "grad_norm": 0.9583351269670265,
      "learning_rate": 9.806153112404658e-06,
      "loss": 0.4529,
      "step": 1856
    },
    {
      "epoch": 0.11631876477865297,
      "grad_norm": 1.0011670577368086,
      "learning_rate": 9.805873297427822e-06,
      "loss": 0.4808,
      "step": 1857
    },
    {
      "epoch": 0.11638140277799527,
      "grad_norm": 0.9404093424506966,
      "learning_rate": 9.805593284640349e-06,
      "loss": 0.4333,
      "step": 1858
    },
    {
      "epoch": 0.11644404077733757,
      "grad_norm": 0.8522502729715666,
      "learning_rate": 9.805313074053764e-06,
      "loss": 0.4097,
      "step": 1859
    },
    {
      "epoch": 0.11650667877667988,
      "grad_norm": 0.8655955617406346,
      "learning_rate": 9.805032665679601e-06,
      "loss": 0.4388,
      "step": 1860
    },
    {
      "epoch": 0.11656931677602217,
      "grad_norm": 0.928818914049864,
      "learning_rate": 9.804752059529401e-06,
      "loss": 0.4673,
      "step": 1861
    },
    {
      "epoch": 0.11663195477536448,
      "grad_norm": 0.8535197691834867,
      "learning_rate": 9.804471255614716e-06,
      "loss": 0.4163,
      "step": 1862
    },
    {
      "epoch": 0.11669459277470677,
      "grad_norm": 0.9325410239357725,
      "learning_rate": 9.8041902539471e-06,
      "loss": 0.4467,
      "step": 1863
    },
    {
      "epoch": 0.11675723077404908,
      "grad_norm": 0.8673886746712364,
      "learning_rate": 9.803909054538125e-06,
      "loss": 0.3982,
      "step": 1864
    },
    {
      "epoch": 0.11681986877339137,
      "grad_norm": 0.7962051903006204,
      "learning_rate": 9.80362765739936e-06,
      "loss": 0.4719,
      "step": 1865
    },
    {
      "epoch": 0.11688250677273368,
      "grad_norm": 0.8752645488158937,
      "learning_rate": 9.803346062542388e-06,
      "loss": 0.4664,
      "step": 1866
    },
    {
      "epoch": 0.11694514477207597,
      "grad_norm": 0.8464256411165557,
      "learning_rate": 9.803064269978802e-06,
      "loss": 0.426,
      "step": 1867
    },
    {
      "epoch": 0.11700778277141828,
      "grad_norm": 0.9449243371978524,
      "learning_rate": 9.802782279720198e-06,
      "loss": 0.4817,
      "step": 1868
    },
    {
      "epoch": 0.11707042077076059,
      "grad_norm": 0.8327541168629713,
      "learning_rate": 9.802500091778185e-06,
      "loss": 0.4563,
      "step": 1869
    },
    {
      "epoch": 0.11713305877010288,
      "grad_norm": 0.8782458761950669,
      "learning_rate": 9.802217706164376e-06,
      "loss": 0.4533,
      "step": 1870
    },
    {
      "epoch": 0.11719569676944519,
      "grad_norm": 0.8329608458563408,
      "learning_rate": 9.801935122890394e-06,
      "loss": 0.4038,
      "step": 1871
    },
    {
      "epoch": 0.11725833476878748,
      "grad_norm": 0.8669713938456914,
      "learning_rate": 9.801652341967872e-06,
      "loss": 0.4711,
      "step": 1872
    },
    {
      "epoch": 0.11732097276812979,
      "grad_norm": 0.8908329448807645,
      "learning_rate": 9.801369363408448e-06,
      "loss": 0.4449,
      "step": 1873
    },
    {
      "epoch": 0.11738361076747209,
      "grad_norm": 0.9484261268105042,
      "learning_rate": 9.801086187223772e-06,
      "loss": 0.4623,
      "step": 1874
    },
    {
      "epoch": 0.11744624876681439,
      "grad_norm": 0.779672442016922,
      "learning_rate": 9.800802813425495e-06,
      "loss": 0.4073,
      "step": 1875
    },
    {
      "epoch": 0.11750888676615669,
      "grad_norm": 0.8137531800030036,
      "learning_rate": 9.800519242025285e-06,
      "loss": 0.4152,
      "step": 1876
    },
    {
      "epoch": 0.117571524765499,
      "grad_norm": 0.8557024115057666,
      "learning_rate": 9.80023547303481e-06,
      "loss": 0.4374,
      "step": 1877
    },
    {
      "epoch": 0.11763416276484129,
      "grad_norm": 0.9681301507168295,
      "learning_rate": 9.799951506465753e-06,
      "loss": 0.4323,
      "step": 1878
    },
    {
      "epoch": 0.1176968007641836,
      "grad_norm": 0.8669353901520958,
      "learning_rate": 9.7996673423298e-06,
      "loss": 0.4491,
      "step": 1879
    },
    {
      "epoch": 0.11775943876352589,
      "grad_norm": 0.7389160667306787,
      "learning_rate": 9.79938298063865e-06,
      "loss": 0.4874,
      "step": 1880
    },
    {
      "epoch": 0.1178220767628682,
      "grad_norm": 0.7794703842238357,
      "learning_rate": 9.799098421404006e-06,
      "loss": 0.4162,
      "step": 1881
    },
    {
      "epoch": 0.11788471476221049,
      "grad_norm": 0.9019050748078803,
      "learning_rate": 9.79881366463758e-06,
      "loss": 0.4332,
      "step": 1882
    },
    {
      "epoch": 0.1179473527615528,
      "grad_norm": 0.9161745479607387,
      "learning_rate": 9.798528710351089e-06,
      "loss": 0.4401,
      "step": 1883
    },
    {
      "epoch": 0.11800999076089509,
      "grad_norm": 0.8642639711281483,
      "learning_rate": 9.79824355855627e-06,
      "loss": 0.4262,
      "step": 1884
    },
    {
      "epoch": 0.1180726287602374,
      "grad_norm": 0.9304407248712889,
      "learning_rate": 9.797958209264853e-06,
      "loss": 0.4638,
      "step": 1885
    },
    {
      "epoch": 0.1181352667595797,
      "grad_norm": 0.894351778227182,
      "learning_rate": 9.797672662488588e-06,
      "loss": 0.4211,
      "step": 1886
    },
    {
      "epoch": 0.118197904758922,
      "grad_norm": 0.8523632438665623,
      "learning_rate": 9.797386918239225e-06,
      "loss": 0.4567,
      "step": 1887
    },
    {
      "epoch": 0.1182605427582643,
      "grad_norm": 0.8522848105714573,
      "learning_rate": 9.797100976528525e-06,
      "loss": 0.4344,
      "step": 1888
    },
    {
      "epoch": 0.1183231807576066,
      "grad_norm": 0.8555568071677059,
      "learning_rate": 9.796814837368258e-06,
      "loss": 0.4483,
      "step": 1889
    },
    {
      "epoch": 0.1183858187569489,
      "grad_norm": 0.8514662446840497,
      "learning_rate": 9.796528500770203e-06,
      "loss": 0.4128,
      "step": 1890
    },
    {
      "epoch": 0.1184484567562912,
      "grad_norm": 0.8662943577478178,
      "learning_rate": 9.796241966746145e-06,
      "loss": 0.423,
      "step": 1891
    },
    {
      "epoch": 0.11851109475563351,
      "grad_norm": 0.8472479170835439,
      "learning_rate": 9.795955235307876e-06,
      "loss": 0.4383,
      "step": 1892
    },
    {
      "epoch": 0.1185737327549758,
      "grad_norm": 0.880638112145388,
      "learning_rate": 9.7956683064672e-06,
      "loss": 0.4611,
      "step": 1893
    },
    {
      "epoch": 0.11863637075431811,
      "grad_norm": 0.8723754376340148,
      "learning_rate": 9.795381180235929e-06,
      "loss": 0.4536,
      "step": 1894
    },
    {
      "epoch": 0.1186990087536604,
      "grad_norm": 0.8538393903052174,
      "learning_rate": 9.795093856625875e-06,
      "loss": 0.436,
      "step": 1895
    },
    {
      "epoch": 0.11876164675300271,
      "grad_norm": 0.9587448332065376,
      "learning_rate": 9.794806335648868e-06,
      "loss": 0.4761,
      "step": 1896
    },
    {
      "epoch": 0.11882428475234501,
      "grad_norm": 0.8565138521963223,
      "learning_rate": 9.794518617316745e-06,
      "loss": 0.4841,
      "step": 1897
    },
    {
      "epoch": 0.11888692275168732,
      "grad_norm": 0.8705235098323003,
      "learning_rate": 9.794230701641343e-06,
      "loss": 0.4416,
      "step": 1898
    },
    {
      "epoch": 0.11894956075102961,
      "grad_norm": 0.8202464581466415,
      "learning_rate": 9.793942588634518e-06,
      "loss": 0.4538,
      "step": 1899
    },
    {
      "epoch": 0.11901219875037192,
      "grad_norm": 0.8908521424222513,
      "learning_rate": 9.793654278308125e-06,
      "loss": 0.4776,
      "step": 1900
    },
    {
      "epoch": 0.11907483674971421,
      "grad_norm": 0.9511568025639996,
      "learning_rate": 9.793365770674031e-06,
      "loss": 0.4009,
      "step": 1901
    },
    {
      "epoch": 0.11913747474905652,
      "grad_norm": 0.882567568100085,
      "learning_rate": 9.793077065744114e-06,
      "loss": 0.4779,
      "step": 1902
    },
    {
      "epoch": 0.11920011274839881,
      "grad_norm": 0.8571723577326055,
      "learning_rate": 9.792788163530254e-06,
      "loss": 0.4201,
      "step": 1903
    },
    {
      "epoch": 0.11926275074774112,
      "grad_norm": 0.8760476129173063,
      "learning_rate": 9.792499064044343e-06,
      "loss": 0.4523,
      "step": 1904
    },
    {
      "epoch": 0.11932538874708341,
      "grad_norm": 0.7843825700816958,
      "learning_rate": 9.792209767298282e-06,
      "loss": 0.4921,
      "step": 1905
    },
    {
      "epoch": 0.11938802674642572,
      "grad_norm": 0.8116121003920735,
      "learning_rate": 9.79192027330398e-06,
      "loss": 0.4327,
      "step": 1906
    },
    {
      "epoch": 0.11945066474576801,
      "grad_norm": 0.8585154877457689,
      "learning_rate": 9.791630582073345e-06,
      "loss": 0.4264,
      "step": 1907
    },
    {
      "epoch": 0.11951330274511032,
      "grad_norm": 0.7052824910939224,
      "learning_rate": 9.79134069361831e-06,
      "loss": 0.4666,
      "step": 1908
    },
    {
      "epoch": 0.11957594074445262,
      "grad_norm": 0.9319054263297472,
      "learning_rate": 9.7910506079508e-06,
      "loss": 0.4515,
      "step": 1909
    },
    {
      "epoch": 0.11963857874379492,
      "grad_norm": 0.6880721951419185,
      "learning_rate": 9.79076032508276e-06,
      "loss": 0.4598,
      "step": 1910
    },
    {
      "epoch": 0.11970121674313722,
      "grad_norm": 0.8738478308714286,
      "learning_rate": 9.790469845026136e-06,
      "loss": 0.4492,
      "step": 1911
    },
    {
      "epoch": 0.11976385474247953,
      "grad_norm": 0.9009560904629528,
      "learning_rate": 9.790179167792882e-06,
      "loss": 0.4826,
      "step": 1912
    },
    {
      "epoch": 0.11982649274182183,
      "grad_norm": 0.9450338411082336,
      "learning_rate": 9.789888293394965e-06,
      "loss": 0.452,
      "step": 1913
    },
    {
      "epoch": 0.11988913074116413,
      "grad_norm": 0.9554951015526164,
      "learning_rate": 9.789597221844358e-06,
      "loss": 0.4706,
      "step": 1914
    },
    {
      "epoch": 0.11995176874050643,
      "grad_norm": 0.7910428100332522,
      "learning_rate": 9.789305953153038e-06,
      "loss": 0.4817,
      "step": 1915
    },
    {
      "epoch": 0.12001440673984873,
      "grad_norm": 0.8702468802902968,
      "learning_rate": 9.789014487332999e-06,
      "loss": 0.4023,
      "step": 1916
    },
    {
      "epoch": 0.12007704473919104,
      "grad_norm": 0.9150530353493025,
      "learning_rate": 9.788722824396232e-06,
      "loss": 0.4697,
      "step": 1917
    },
    {
      "epoch": 0.12013968273853333,
      "grad_norm": 0.8164398242538387,
      "learning_rate": 9.788430964354747e-06,
      "loss": 0.42,
      "step": 1918
    },
    {
      "epoch": 0.12020232073787564,
      "grad_norm": 0.8910891412318117,
      "learning_rate": 9.788138907220553e-06,
      "loss": 0.4772,
      "step": 1919
    },
    {
      "epoch": 0.12026495873721793,
      "grad_norm": 0.8812851136303951,
      "learning_rate": 9.787846653005675e-06,
      "loss": 0.4645,
      "step": 1920
    },
    {
      "epoch": 0.12032759673656024,
      "grad_norm": 0.8956053236953402,
      "learning_rate": 9.787554201722138e-06,
      "loss": 0.4388,
      "step": 1921
    },
    {
      "epoch": 0.12039023473590253,
      "grad_norm": 0.7617935476842171,
      "learning_rate": 9.787261553381981e-06,
      "loss": 0.5035,
      "step": 1922
    },
    {
      "epoch": 0.12045287273524484,
      "grad_norm": 0.8906855387705271,
      "learning_rate": 9.786968707997253e-06,
      "loss": 0.4784,
      "step": 1923
    },
    {
      "epoch": 0.12051551073458713,
      "grad_norm": 0.877625738642612,
      "learning_rate": 9.786675665580002e-06,
      "loss": 0.4548,
      "step": 1924
    },
    {
      "epoch": 0.12057814873392944,
      "grad_norm": 0.9435432054605591,
      "learning_rate": 9.786382426142292e-06,
      "loss": 0.4458,
      "step": 1925
    },
    {
      "epoch": 0.12064078673327173,
      "grad_norm": 0.8520552182203984,
      "learning_rate": 9.786088989696194e-06,
      "loss": 0.4434,
      "step": 1926
    },
    {
      "epoch": 0.12070342473261404,
      "grad_norm": 0.9017825108790051,
      "learning_rate": 9.785795356253784e-06,
      "loss": 0.454,
      "step": 1927
    },
    {
      "epoch": 0.12076606273195634,
      "grad_norm": 0.8562990074599104,
      "learning_rate": 9.78550152582715e-06,
      "loss": 0.3969,
      "step": 1928
    },
    {
      "epoch": 0.12082870073129864,
      "grad_norm": 0.847255016459296,
      "learning_rate": 9.785207498428384e-06,
      "loss": 0.4704,
      "step": 1929
    },
    {
      "epoch": 0.12089133873064094,
      "grad_norm": 0.7905320037355367,
      "learning_rate": 9.78491327406959e-06,
      "loss": 0.4265,
      "step": 1930
    },
    {
      "epoch": 0.12095397672998325,
      "grad_norm": 0.9115492611743168,
      "learning_rate": 9.784618852762876e-06,
      "loss": 0.448,
      "step": 1931
    },
    {
      "epoch": 0.12101661472932554,
      "grad_norm": 0.8503515998889217,
      "learning_rate": 9.784324234520363e-06,
      "loss": 0.4392,
      "step": 1932
    },
    {
      "epoch": 0.12107925272866785,
      "grad_norm": 0.6695028803100502,
      "learning_rate": 9.784029419354177e-06,
      "loss": 0.4793,
      "step": 1933
    },
    {
      "epoch": 0.12114189072801015,
      "grad_norm": 0.8590726281894936,
      "learning_rate": 9.783734407276452e-06,
      "loss": 0.4175,
      "step": 1934
    },
    {
      "epoch": 0.12120452872735245,
      "grad_norm": 0.8614399218486347,
      "learning_rate": 9.78343919829933e-06,
      "loss": 0.4079,
      "step": 1935
    },
    {
      "epoch": 0.12126716672669476,
      "grad_norm": 0.8852876121090753,
      "learning_rate": 9.783143792434964e-06,
      "loss": 0.413,
      "step": 1936
    },
    {
      "epoch": 0.12132980472603705,
      "grad_norm": 0.8800285009052519,
      "learning_rate": 9.78284818969551e-06,
      "loss": 0.4151,
      "step": 1937
    },
    {
      "epoch": 0.12139244272537936,
      "grad_norm": 0.8580730050358212,
      "learning_rate": 9.782552390093139e-06,
      "loss": 0.4198,
      "step": 1938
    },
    {
      "epoch": 0.12145508072472165,
      "grad_norm": 0.8882665352039608,
      "learning_rate": 9.782256393640022e-06,
      "loss": 0.433,
      "step": 1939
    },
    {
      "epoch": 0.12151771872406396,
      "grad_norm": 0.8309596894341328,
      "learning_rate": 9.781960200348343e-06,
      "loss": 0.4174,
      "step": 1940
    },
    {
      "epoch": 0.12158035672340625,
      "grad_norm": 0.8910879768307287,
      "learning_rate": 9.781663810230298e-06,
      "loss": 0.4145,
      "step": 1941
    },
    {
      "epoch": 0.12164299472274856,
      "grad_norm": 0.9255918586421936,
      "learning_rate": 9.781367223298082e-06,
      "loss": 0.4305,
      "step": 1942
    },
    {
      "epoch": 0.12170563272209085,
      "grad_norm": 0.8610108274097095,
      "learning_rate": 9.781070439563902e-06,
      "loss": 0.4798,
      "step": 1943
    },
    {
      "epoch": 0.12176827072143316,
      "grad_norm": 0.9032538311887406,
      "learning_rate": 9.780773459039977e-06,
      "loss": 0.445,
      "step": 1944
    },
    {
      "epoch": 0.12183090872077545,
      "grad_norm": 0.8850903931272996,
      "learning_rate": 9.780476281738528e-06,
      "loss": 0.4375,
      "step": 1945
    },
    {
      "epoch": 0.12189354672011776,
      "grad_norm": 0.9689633369027005,
      "learning_rate": 9.780178907671788e-06,
      "loss": 0.4555,
      "step": 1946
    },
    {
      "epoch": 0.12195618471946006,
      "grad_norm": 0.7389075555406811,
      "learning_rate": 9.779881336851998e-06,
      "loss": 0.4843,
      "step": 1947
    },
    {
      "epoch": 0.12201882271880236,
      "grad_norm": 0.8998760281243913,
      "learning_rate": 9.779583569291405e-06,
      "loss": 0.4481,
      "step": 1948
    },
    {
      "epoch": 0.12208146071814466,
      "grad_norm": 0.8530380878704589,
      "learning_rate": 9.779285605002265e-06,
      "loss": 0.4965,
      "step": 1949
    },
    {
      "epoch": 0.12214409871748697,
      "grad_norm": 0.8930443159422766,
      "learning_rate": 9.778987443996841e-06,
      "loss": 0.4475,
      "step": 1950
    },
    {
      "epoch": 0.12220673671682926,
      "grad_norm": 0.9361500358245101,
      "learning_rate": 9.778689086287407e-06,
      "loss": 0.477,
      "step": 1951
    },
    {
      "epoch": 0.12226937471617157,
      "grad_norm": 0.8832521336121351,
      "learning_rate": 9.778390531886245e-06,
      "loss": 0.4284,
      "step": 1952
    },
    {
      "epoch": 0.12233201271551386,
      "grad_norm": 0.8431885223713479,
      "learning_rate": 9.77809178080564e-06,
      "loss": 0.4245,
      "step": 1953
    },
    {
      "epoch": 0.12239465071485617,
      "grad_norm": 0.8831172994188349,
      "learning_rate": 9.777792833057893e-06,
      "loss": 0.4174,
      "step": 1954
    },
    {
      "epoch": 0.12245728871419846,
      "grad_norm": 0.7621422290422518,
      "learning_rate": 9.777493688655303e-06,
      "loss": 0.4947,
      "step": 1955
    },
    {
      "epoch": 0.12251992671354077,
      "grad_norm": 0.9511591613154979,
      "learning_rate": 9.777194347610189e-06,
      "loss": 0.4689,
      "step": 1956
    },
    {
      "epoch": 0.12258256471288308,
      "grad_norm": 0.9092383095725791,
      "learning_rate": 9.776894809934866e-06,
      "loss": 0.4595,
      "step": 1957
    },
    {
      "epoch": 0.12264520271222537,
      "grad_norm": 0.8280430115830753,
      "learning_rate": 9.776595075641667e-06,
      "loss": 0.4387,
      "step": 1958
    },
    {
      "epoch": 0.12270784071156768,
      "grad_norm": 0.8346567794777724,
      "learning_rate": 9.77629514474293e-06,
      "loss": 0.4913,
      "step": 1959
    },
    {
      "epoch": 0.12277047871090997,
      "grad_norm": 0.8939037252726106,
      "learning_rate": 9.775995017250995e-06,
      "loss": 0.4191,
      "step": 1960
    },
    {
      "epoch": 0.12283311671025228,
      "grad_norm": 0.954339365250822,
      "learning_rate": 9.775694693178222e-06,
      "loss": 0.4842,
      "step": 1961
    },
    {
      "epoch": 0.12289575470959457,
      "grad_norm": 0.872884066048064,
      "learning_rate": 9.775394172536967e-06,
      "loss": 0.4448,
      "step": 1962
    },
    {
      "epoch": 0.12295839270893688,
      "grad_norm": 0.9347014330890309,
      "learning_rate": 9.775093455339601e-06,
      "loss": 0.4729,
      "step": 1963
    },
    {
      "epoch": 0.12302103070827917,
      "grad_norm": 0.8399675614208689,
      "learning_rate": 9.774792541598504e-06,
      "loss": 0.4048,
      "step": 1964
    },
    {
      "epoch": 0.12308366870762148,
      "grad_norm": 0.8991716812353074,
      "learning_rate": 9.774491431326058e-06,
      "loss": 0.4933,
      "step": 1965
    },
    {
      "epoch": 0.12314630670696378,
      "grad_norm": 0.9256635214755898,
      "learning_rate": 9.77419012453466e-06,
      "loss": 0.4597,
      "step": 1966
    },
    {
      "epoch": 0.12320894470630608,
      "grad_norm": 0.8562747610961979,
      "learning_rate": 9.773888621236708e-06,
      "loss": 0.4536,
      "step": 1967
    },
    {
      "epoch": 0.12327158270564838,
      "grad_norm": 0.8358971795133557,
      "learning_rate": 9.773586921444616e-06,
      "loss": 0.4104,
      "step": 1968
    },
    {
      "epoch": 0.12333422070499069,
      "grad_norm": 0.7596544452598072,
      "learning_rate": 9.773285025170801e-06,
      "loss": 0.4831,
      "step": 1969
    },
    {
      "epoch": 0.12339685870433298,
      "grad_norm": 0.7928230186685794,
      "learning_rate": 9.772982932427689e-06,
      "loss": 0.4108,
      "step": 1970
    },
    {
      "epoch": 0.12345949670367529,
      "grad_norm": 0.8698664952868713,
      "learning_rate": 9.77268064322771e-06,
      "loss": 0.4213,
      "step": 1971
    },
    {
      "epoch": 0.12352213470301758,
      "grad_norm": 0.8598034842677503,
      "learning_rate": 9.772378157583314e-06,
      "loss": 0.4244,
      "step": 1972
    },
    {
      "epoch": 0.12358477270235989,
      "grad_norm": 0.8597939166270584,
      "learning_rate": 9.772075475506943e-06,
      "loss": 0.429,
      "step": 1973
    },
    {
      "epoch": 0.12364741070170218,
      "grad_norm": 0.8559225370518411,
      "learning_rate": 9.771772597011063e-06,
      "loss": 0.4629,
      "step": 1974
    },
    {
      "epoch": 0.12371004870104449,
      "grad_norm": 1.0404646663152395,
      "learning_rate": 9.771469522108137e-06,
      "loss": 0.4765,
      "step": 1975
    },
    {
      "epoch": 0.12377268670038678,
      "grad_norm": 0.8881557984884741,
      "learning_rate": 9.771166250810638e-06,
      "loss": 0.4317,
      "step": 1976
    },
    {
      "epoch": 0.12383532469972909,
      "grad_norm": 0.9135083737814151,
      "learning_rate": 9.770862783131051e-06,
      "loss": 0.4721,
      "step": 1977
    },
    {
      "epoch": 0.1238979626990714,
      "grad_norm": 0.8299129373524771,
      "learning_rate": 9.770559119081869e-06,
      "loss": 0.3922,
      "step": 1978
    },
    {
      "epoch": 0.12396060069841369,
      "grad_norm": 1.0097811319466954,
      "learning_rate": 9.770255258675585e-06,
      "loss": 0.4691,
      "step": 1979
    },
    {
      "epoch": 0.124023238697756,
      "grad_norm": 0.8589812941746178,
      "learning_rate": 9.76995120192471e-06,
      "loss": 0.471,
      "step": 1980
    },
    {
      "epoch": 0.1240858766970983,
      "grad_norm": 0.8892097782697058,
      "learning_rate": 9.769646948841759e-06,
      "loss": 0.411,
      "step": 1981
    },
    {
      "epoch": 0.1241485146964406,
      "grad_norm": 0.836382394889718,
      "learning_rate": 9.769342499439253e-06,
      "loss": 0.4179,
      "step": 1982
    },
    {
      "epoch": 0.1242111526957829,
      "grad_norm": 0.9701303555126329,
      "learning_rate": 9.769037853729726e-06,
      "loss": 0.4337,
      "step": 1983
    },
    {
      "epoch": 0.1242737906951252,
      "grad_norm": 0.8025952015137189,
      "learning_rate": 9.768733011725715e-06,
      "loss": 0.4411,
      "step": 1984
    },
    {
      "epoch": 0.1243364286944675,
      "grad_norm": 0.8717624177547921,
      "learning_rate": 9.768427973439769e-06,
      "loss": 0.45,
      "step": 1985
    },
    {
      "epoch": 0.1243990666938098,
      "grad_norm": 0.9064800083155948,
      "learning_rate": 9.76812273888444e-06,
      "loss": 0.4542,
      "step": 1986
    },
    {
      "epoch": 0.1244617046931521,
      "grad_norm": 0.9797016817669671,
      "learning_rate": 9.767817308072296e-06,
      "loss": 0.4272,
      "step": 1987
    },
    {
      "epoch": 0.1245243426924944,
      "grad_norm": 0.9332144097959241,
      "learning_rate": 9.767511681015907e-06,
      "loss": 0.4648,
      "step": 1988
    },
    {
      "epoch": 0.1245869806918367,
      "grad_norm": 0.8784522514899924,
      "learning_rate": 9.76720585772785e-06,
      "loss": 0.4258,
      "step": 1989
    },
    {
      "epoch": 0.124649618691179,
      "grad_norm": 0.8439068846303541,
      "learning_rate": 9.766899838220718e-06,
      "loss": 0.4425,
      "step": 1990
    },
    {
      "epoch": 0.1247122566905213,
      "grad_norm": 0.7321479731781613,
      "learning_rate": 9.766593622507103e-06,
      "loss": 0.4435,
      "step": 1991
    },
    {
      "epoch": 0.12477489468986361,
      "grad_norm": 0.8417725018487341,
      "learning_rate": 9.76628721059961e-06,
      "loss": 0.4215,
      "step": 1992
    },
    {
      "epoch": 0.1248375326892059,
      "grad_norm": 0.8796977911200452,
      "learning_rate": 9.765980602510849e-06,
      "loss": 0.4233,
      "step": 1993
    },
    {
      "epoch": 0.12490017068854821,
      "grad_norm": 0.7878704244519861,
      "learning_rate": 9.765673798253443e-06,
      "loss": 0.4696,
      "step": 1994
    },
    {
      "epoch": 0.1249628086878905,
      "grad_norm": 0.9176408645587596,
      "learning_rate": 9.76536679784002e-06,
      "loss": 0.4831,
      "step": 1995
    },
    {
      "epoch": 0.1250254466872328,
      "grad_norm": 0.888170379551344,
      "learning_rate": 9.765059601283212e-06,
      "loss": 0.4068,
      "step": 1996
    },
    {
      "epoch": 0.12508808468657512,
      "grad_norm": 0.8629017296950021,
      "learning_rate": 9.764752208595669e-06,
      "loss": 0.4641,
      "step": 1997
    },
    {
      "epoch": 0.1251507226859174,
      "grad_norm": 0.8697727302216106,
      "learning_rate": 9.764444619790041e-06,
      "loss": 0.4255,
      "step": 1998
    },
    {
      "epoch": 0.1252133606852597,
      "grad_norm": 0.8133250765798806,
      "learning_rate": 9.764136834878987e-06,
      "loss": 0.3907,
      "step": 1999
    },
    {
      "epoch": 0.125275998684602,
      "grad_norm": 0.9080153321639883,
      "learning_rate": 9.763828853875177e-06,
      "loss": 0.4315,
      "step": 2000
    },
    {
      "epoch": 0.12533863668394432,
      "grad_norm": 0.6909455843220903,
      "learning_rate": 9.763520676791287e-06,
      "loss": 0.4859,
      "step": 2001
    },
    {
      "epoch": 0.12540127468328663,
      "grad_norm": 0.7896755186114948,
      "learning_rate": 9.763212303640001e-06,
      "loss": 0.4853,
      "step": 2002
    },
    {
      "epoch": 0.1254639126826289,
      "grad_norm": 0.933548381592713,
      "learning_rate": 9.762903734434014e-06,
      "loss": 0.4565,
      "step": 2003
    },
    {
      "epoch": 0.12552655068197122,
      "grad_norm": 0.982648658820914,
      "learning_rate": 9.762594969186024e-06,
      "loss": 0.4907,
      "step": 2004
    },
    {
      "epoch": 0.12558918868131352,
      "grad_norm": 0.9947701639144322,
      "learning_rate": 9.762286007908741e-06,
      "loss": 0.43,
      "step": 2005
    },
    {
      "epoch": 0.12565182668065583,
      "grad_norm": 0.846969810884745,
      "learning_rate": 9.761976850614882e-06,
      "loss": 0.4448,
      "step": 2006
    },
    {
      "epoch": 0.1257144646799981,
      "grad_norm": 0.9013533073968757,
      "learning_rate": 9.761667497317173e-06,
      "loss": 0.4621,
      "step": 2007
    },
    {
      "epoch": 0.12577710267934042,
      "grad_norm": 0.8147084444742309,
      "learning_rate": 9.761357948028347e-06,
      "loss": 0.3919,
      "step": 2008
    },
    {
      "epoch": 0.12583974067868273,
      "grad_norm": 0.8916963068247526,
      "learning_rate": 9.761048202761142e-06,
      "loss": 0.4169,
      "step": 2009
    },
    {
      "epoch": 0.12590237867802503,
      "grad_norm": 0.8260349160941491,
      "learning_rate": 9.76073826152831e-06,
      "loss": 0.4063,
      "step": 2010
    },
    {
      "epoch": 0.12596501667736731,
      "grad_norm": 0.8735279711176607,
      "learning_rate": 9.760428124342608e-06,
      "loss": 0.4459,
      "step": 2011
    },
    {
      "epoch": 0.12602765467670962,
      "grad_norm": 0.9168879496020104,
      "learning_rate": 9.7601177912168e-06,
      "loss": 0.4092,
      "step": 2012
    },
    {
      "epoch": 0.12609029267605193,
      "grad_norm": 0.8548067308993058,
      "learning_rate": 9.759807262163662e-06,
      "loss": 0.4423,
      "step": 2013
    },
    {
      "epoch": 0.12615293067539424,
      "grad_norm": 0.89348098665832,
      "learning_rate": 9.75949653719597e-06,
      "loss": 0.4313,
      "step": 2014
    },
    {
      "epoch": 0.12621556867473652,
      "grad_norm": 0.8598510032170522,
      "learning_rate": 9.75918561632652e-06,
      "loss": 0.4458,
      "step": 2015
    },
    {
      "epoch": 0.12627820667407882,
      "grad_norm": 0.8778185840376277,
      "learning_rate": 9.758874499568106e-06,
      "loss": 0.4361,
      "step": 2016
    },
    {
      "epoch": 0.12634084467342113,
      "grad_norm": 0.8324730732058352,
      "learning_rate": 9.758563186933535e-06,
      "loss": 0.44,
      "step": 2017
    },
    {
      "epoch": 0.12640348267276344,
      "grad_norm": 0.8808563990140768,
      "learning_rate": 9.758251678435617e-06,
      "loss": 0.4678,
      "step": 2018
    },
    {
      "epoch": 0.12646612067210572,
      "grad_norm": 0.9712151362074406,
      "learning_rate": 9.75793997408718e-06,
      "loss": 0.4884,
      "step": 2019
    },
    {
      "epoch": 0.12652875867144803,
      "grad_norm": 0.883091683707926,
      "learning_rate": 9.757628073901049e-06,
      "loss": 0.411,
      "step": 2020
    },
    {
      "epoch": 0.12659139667079033,
      "grad_norm": 0.8657294110379281,
      "learning_rate": 9.757315977890064e-06,
      "loss": 0.4453,
      "step": 2021
    },
    {
      "epoch": 0.12665403467013264,
      "grad_norm": 0.955509872734563,
      "learning_rate": 9.75700368606707e-06,
      "loss": 0.4347,
      "step": 2022
    },
    {
      "epoch": 0.12671667266947495,
      "grad_norm": 0.8545629708570182,
      "learning_rate": 9.756691198444923e-06,
      "loss": 0.4388,
      "step": 2023
    },
    {
      "epoch": 0.12677931066881723,
      "grad_norm": 0.9323397915705518,
      "learning_rate": 9.756378515036481e-06,
      "loss": 0.4408,
      "step": 2024
    },
    {
      "epoch": 0.12684194866815954,
      "grad_norm": 0.9153132757586695,
      "learning_rate": 9.756065635854619e-06,
      "loss": 0.4125,
      "step": 2025
    },
    {
      "epoch": 0.12690458666750185,
      "grad_norm": 0.8580141862048717,
      "learning_rate": 9.75575256091221e-06,
      "loss": 0.4062,
      "step": 2026
    },
    {
      "epoch": 0.12696722466684415,
      "grad_norm": 0.8394931159853855,
      "learning_rate": 9.755439290222144e-06,
      "loss": 0.4564,
      "step": 2027
    },
    {
      "epoch": 0.12702986266618643,
      "grad_norm": 0.8441399555243873,
      "learning_rate": 9.755125823797315e-06,
      "loss": 0.4884,
      "step": 2028
    },
    {
      "epoch": 0.12709250066552874,
      "grad_norm": 0.8803089097707018,
      "learning_rate": 9.754812161650624e-06,
      "loss": 0.46,
      "step": 2029
    },
    {
      "epoch": 0.12715513866487105,
      "grad_norm": 0.9066221584711328,
      "learning_rate": 9.754498303794981e-06,
      "loss": 0.4084,
      "step": 2030
    },
    {
      "epoch": 0.12721777666421336,
      "grad_norm": 0.8310112819886916,
      "learning_rate": 9.754184250243306e-06,
      "loss": 0.4511,
      "step": 2031
    },
    {
      "epoch": 0.12728041466355564,
      "grad_norm": 0.8726795696019146,
      "learning_rate": 9.753870001008526e-06,
      "loss": 0.456,
      "step": 2032
    },
    {
      "epoch": 0.12734305266289794,
      "grad_norm": 0.8023723634057087,
      "learning_rate": 9.75355555610357e-06,
      "loss": 0.3957,
      "step": 2033
    },
    {
      "epoch": 0.12740569066224025,
      "grad_norm": 0.8050130343961314,
      "learning_rate": 9.75324091554139e-06,
      "loss": 0.3921,
      "step": 2034
    },
    {
      "epoch": 0.12746832866158256,
      "grad_norm": 0.9124820458098558,
      "learning_rate": 9.752926079334931e-06,
      "loss": 0.4375,
      "step": 2035
    },
    {
      "epoch": 0.12753096666092484,
      "grad_norm": 0.9110556984116792,
      "learning_rate": 9.75261104749715e-06,
      "loss": 0.419,
      "step": 2036
    },
    {
      "epoch": 0.12759360466026715,
      "grad_norm": 0.8706164931849545,
      "learning_rate": 9.752295820041017e-06,
      "loss": 0.4439,
      "step": 2037
    },
    {
      "epoch": 0.12765624265960945,
      "grad_norm": 0.9570859734750722,
      "learning_rate": 9.751980396979507e-06,
      "loss": 0.4632,
      "step": 2038
    },
    {
      "epoch": 0.12771888065895176,
      "grad_norm": 0.7886553351442371,
      "learning_rate": 9.751664778325602e-06,
      "loss": 0.4388,
      "step": 2039
    },
    {
      "epoch": 0.12778151865829404,
      "grad_norm": 0.8383529158615958,
      "learning_rate": 9.751348964092291e-06,
      "loss": 0.4357,
      "step": 2040
    },
    {
      "epoch": 0.12784415665763635,
      "grad_norm": 0.8799563225724358,
      "learning_rate": 9.751032954292576e-06,
      "loss": 0.4553,
      "step": 2041
    },
    {
      "epoch": 0.12790679465697866,
      "grad_norm": 0.8462974517060525,
      "learning_rate": 9.750716748939463e-06,
      "loss": 0.3834,
      "step": 2042
    },
    {
      "epoch": 0.12796943265632096,
      "grad_norm": 0.9089764984716264,
      "learning_rate": 9.750400348045966e-06,
      "loss": 0.4617,
      "step": 2043
    },
    {
      "epoch": 0.12803207065566327,
      "grad_norm": 0.9237026023491177,
      "learning_rate": 9.75008375162511e-06,
      "loss": 0.4738,
      "step": 2044
    },
    {
      "epoch": 0.12809470865500555,
      "grad_norm": 0.7842969209751549,
      "learning_rate": 9.749766959689923e-06,
      "loss": 0.4474,
      "step": 2045
    },
    {
      "epoch": 0.12815734665434786,
      "grad_norm": 0.8528907040512711,
      "learning_rate": 9.749449972253449e-06,
      "loss": 0.4537,
      "step": 2046
    },
    {
      "epoch": 0.12821998465369017,
      "grad_norm": 0.979993972059209,
      "learning_rate": 9.749132789328732e-06,
      "loss": 0.5089,
      "step": 2047
    },
    {
      "epoch": 0.12828262265303247,
      "grad_norm": 0.8513663397782814,
      "learning_rate": 9.748815410928827e-06,
      "loss": 0.4048,
      "step": 2048
    },
    {
      "epoch": 0.12834526065237475,
      "grad_norm": 0.9671122965066332,
      "learning_rate": 9.748497837066798e-06,
      "loss": 0.4225,
      "step": 2049
    },
    {
      "epoch": 0.12840789865171706,
      "grad_norm": 0.8311741757958695,
      "learning_rate": 9.748180067755717e-06,
      "loss": 0.5078,
      "step": 2050
    },
    {
      "epoch": 0.12847053665105937,
      "grad_norm": 0.8592665184978404,
      "learning_rate": 9.747862103008664e-06,
      "loss": 0.4271,
      "step": 2051
    },
    {
      "epoch": 0.12853317465040168,
      "grad_norm": 0.767003818152189,
      "learning_rate": 9.747543942838723e-06,
      "loss": 0.4459,
      "step": 2052
    },
    {
      "epoch": 0.12859581264974396,
      "grad_norm": 0.8746017007166439,
      "learning_rate": 9.747225587258995e-06,
      "loss": 0.4516,
      "step": 2053
    },
    {
      "epoch": 0.12865845064908626,
      "grad_norm": 0.9243591429428598,
      "learning_rate": 9.74690703628258e-06,
      "loss": 0.4305,
      "step": 2054
    },
    {
      "epoch": 0.12872108864842857,
      "grad_norm": 0.8867291733512515,
      "learning_rate": 9.746588289922591e-06,
      "loss": 0.4531,
      "step": 2055
    },
    {
      "epoch": 0.12878372664777088,
      "grad_norm": 0.9791520813199897,
      "learning_rate": 9.746269348192145e-06,
      "loss": 0.4587,
      "step": 2056
    },
    {
      "epoch": 0.12884636464711316,
      "grad_norm": 0.8915441619125695,
      "learning_rate": 9.745950211104374e-06,
      "loss": 0.4153,
      "step": 2057
    },
    {
      "epoch": 0.12890900264645547,
      "grad_norm": 0.8665394804766472,
      "learning_rate": 9.74563087867241e-06,
      "loss": 0.4283,
      "step": 2058
    },
    {
      "epoch": 0.12897164064579777,
      "grad_norm": 0.8335906962051938,
      "learning_rate": 9.745311350909399e-06,
      "loss": 0.4171,
      "step": 2059
    },
    {
      "epoch": 0.12903427864514008,
      "grad_norm": 0.8843820658663534,
      "learning_rate": 9.74499162782849e-06,
      "loss": 0.4611,
      "step": 2060
    },
    {
      "epoch": 0.12909691664448236,
      "grad_norm": 0.8935214659814371,
      "learning_rate": 9.744671709442847e-06,
      "loss": 0.4535,
      "step": 2061
    },
    {
      "epoch": 0.12915955464382467,
      "grad_norm": 0.8880458998206955,
      "learning_rate": 9.744351595765635e-06,
      "loss": 0.4439,
      "step": 2062
    },
    {
      "epoch": 0.12922219264316698,
      "grad_norm": 0.9103310682799571,
      "learning_rate": 9.74403128681003e-06,
      "loss": 0.4321,
      "step": 2063
    },
    {
      "epoch": 0.12928483064250929,
      "grad_norm": 0.8564431268135086,
      "learning_rate": 9.743710782589217e-06,
      "loss": 0.4615,
      "step": 2064
    },
    {
      "epoch": 0.1293474686418516,
      "grad_norm": 0.8644586404774781,
      "learning_rate": 9.743390083116389e-06,
      "loss": 0.3982,
      "step": 2065
    },
    {
      "epoch": 0.12941010664119387,
      "grad_norm": 0.7673740818181908,
      "learning_rate": 9.743069188404744e-06,
      "loss": 0.4786,
      "step": 2066
    },
    {
      "epoch": 0.12947274464053618,
      "grad_norm": 0.9348737651679134,
      "learning_rate": 9.742748098467492e-06,
      "loss": 0.4401,
      "step": 2067
    },
    {
      "epoch": 0.1295353826398785,
      "grad_norm": 0.8383828559393218,
      "learning_rate": 9.742426813317847e-06,
      "loss": 0.4062,
      "step": 2068
    },
    {
      "epoch": 0.1295980206392208,
      "grad_norm": 0.8917451892204629,
      "learning_rate": 9.742105332969033e-06,
      "loss": 0.4334,
      "step": 2069
    },
    {
      "epoch": 0.12966065863856308,
      "grad_norm": 0.8683136765903321,
      "learning_rate": 9.741783657434286e-06,
      "loss": 0.4332,
      "step": 2070
    },
    {
      "epoch": 0.12972329663790538,
      "grad_norm": 0.9014577489559453,
      "learning_rate": 9.74146178672684e-06,
      "loss": 0.4246,
      "step": 2071
    },
    {
      "epoch": 0.1297859346372477,
      "grad_norm": 0.9589528320282293,
      "learning_rate": 9.741139720859948e-06,
      "loss": 0.4479,
      "step": 2072
    },
    {
      "epoch": 0.12984857263659,
      "grad_norm": 0.7468152516383746,
      "learning_rate": 9.740817459846867e-06,
      "loss": 0.4856,
      "step": 2073
    },
    {
      "epoch": 0.12991121063593228,
      "grad_norm": 0.908066838412693,
      "learning_rate": 9.740495003700859e-06,
      "loss": 0.4287,
      "step": 2074
    },
    {
      "epoch": 0.12997384863527459,
      "grad_norm": 0.8219531984459191,
      "learning_rate": 9.740172352435194e-06,
      "loss": 0.4127,
      "step": 2075
    },
    {
      "epoch": 0.1300364866346169,
      "grad_norm": 0.7768455925740888,
      "learning_rate": 9.739849506063157e-06,
      "loss": 0.4686,
      "step": 2076
    },
    {
      "epoch": 0.1300991246339592,
      "grad_norm": 0.9383218971072638,
      "learning_rate": 9.739526464598034e-06,
      "loss": 0.4391,
      "step": 2077
    },
    {
      "epoch": 0.13016176263330148,
      "grad_norm": 0.8450537384981881,
      "learning_rate": 9.739203228053123e-06,
      "loss": 0.4332,
      "step": 2078
    },
    {
      "epoch": 0.1302244006326438,
      "grad_norm": 0.7596182859917648,
      "learning_rate": 9.738879796441728e-06,
      "loss": 0.502,
      "step": 2079
    },
    {
      "epoch": 0.1302870386319861,
      "grad_norm": 0.9353950815726088,
      "learning_rate": 9.73855616977716e-06,
      "loss": 0.4476,
      "step": 2080
    },
    {
      "epoch": 0.1303496766313284,
      "grad_norm": 0.9957119013716053,
      "learning_rate": 9.73823234807274e-06,
      "loss": 0.4787,
      "step": 2081
    },
    {
      "epoch": 0.13041231463067068,
      "grad_norm": 0.9250461280177235,
      "learning_rate": 9.737908331341798e-06,
      "loss": 0.4398,
      "step": 2082
    },
    {
      "epoch": 0.130474952630013,
      "grad_norm": 0.8722862695734025,
      "learning_rate": 9.73758411959767e-06,
      "loss": 0.4461,
      "step": 2083
    },
    {
      "epoch": 0.1305375906293553,
      "grad_norm": 0.8344402196138767,
      "learning_rate": 9.7372597128537e-06,
      "loss": 0.4028,
      "step": 2084
    },
    {
      "epoch": 0.1306002286286976,
      "grad_norm": 0.9218798034550804,
      "learning_rate": 9.73693511112324e-06,
      "loss": 0.4773,
      "step": 2085
    },
    {
      "epoch": 0.13066286662803991,
      "grad_norm": 0.7687854386751745,
      "learning_rate": 9.736610314419651e-06,
      "loss": 0.4224,
      "step": 2086
    },
    {
      "epoch": 0.1307255046273822,
      "grad_norm": 0.8934784212119946,
      "learning_rate": 9.736285322756305e-06,
      "loss": 0.4482,
      "step": 2087
    },
    {
      "epoch": 0.1307881426267245,
      "grad_norm": 0.9355473488871526,
      "learning_rate": 9.735960136146576e-06,
      "loss": 0.4454,
      "step": 2088
    },
    {
      "epoch": 0.1308507806260668,
      "grad_norm": 0.979981554988764,
      "learning_rate": 9.735634754603846e-06,
      "loss": 0.4655,
      "step": 2089
    },
    {
      "epoch": 0.13091341862540912,
      "grad_norm": 0.9142567062089845,
      "learning_rate": 9.735309178141513e-06,
      "loss": 0.4552,
      "step": 2090
    },
    {
      "epoch": 0.1309760566247514,
      "grad_norm": 0.8741745846148111,
      "learning_rate": 9.734983406772973e-06,
      "loss": 0.4272,
      "step": 2091
    },
    {
      "epoch": 0.1310386946240937,
      "grad_norm": 1.1000956883010664,
      "learning_rate": 9.734657440511638e-06,
      "loss": 0.4095,
      "step": 2092
    },
    {
      "epoch": 0.131101332623436,
      "grad_norm": 0.8415160776035115,
      "learning_rate": 9.734331279370925e-06,
      "loss": 0.4462,
      "step": 2093
    },
    {
      "epoch": 0.13116397062277832,
      "grad_norm": 0.7782170463323025,
      "learning_rate": 9.734004923364258e-06,
      "loss": 0.4942,
      "step": 2094
    },
    {
      "epoch": 0.1312266086221206,
      "grad_norm": 0.7984043822282996,
      "learning_rate": 9.733678372505068e-06,
      "loss": 0.3905,
      "step": 2095
    },
    {
      "epoch": 0.1312892466214629,
      "grad_norm": 0.9109716182097731,
      "learning_rate": 9.733351626806797e-06,
      "loss": 0.4634,
      "step": 2096
    },
    {
      "epoch": 0.13135188462080521,
      "grad_norm": 0.8242302375418116,
      "learning_rate": 9.733024686282898e-06,
      "loss": 0.4369,
      "step": 2097
    },
    {
      "epoch": 0.13141452262014752,
      "grad_norm": 0.9602117753687835,
      "learning_rate": 9.732697550946822e-06,
      "loss": 0.4961,
      "step": 2098
    },
    {
      "epoch": 0.1314771606194898,
      "grad_norm": 0.9457929286076865,
      "learning_rate": 9.732370220812035e-06,
      "loss": 0.4603,
      "step": 2099
    },
    {
      "epoch": 0.1315397986188321,
      "grad_norm": 0.9272297008909214,
      "learning_rate": 9.732042695892014e-06,
      "loss": 0.4219,
      "step": 2100
    },
    {
      "epoch": 0.13160243661817442,
      "grad_norm": 0.9167402340628865,
      "learning_rate": 9.731714976200236e-06,
      "loss": 0.4319,
      "step": 2101
    },
    {
      "epoch": 0.13166507461751673,
      "grad_norm": 0.8560026137950737,
      "learning_rate": 9.731387061750193e-06,
      "loss": 0.401,
      "step": 2102
    },
    {
      "epoch": 0.131727712616859,
      "grad_norm": 0.8833413241614748,
      "learning_rate": 9.731058952555378e-06,
      "loss": 0.4513,
      "step": 2103
    },
    {
      "epoch": 0.1317903506162013,
      "grad_norm": 0.808418940493975,
      "learning_rate": 9.730730648629298e-06,
      "loss": 0.4548,
      "step": 2104
    },
    {
      "epoch": 0.13185298861554362,
      "grad_norm": 0.8659482965154802,
      "learning_rate": 9.730402149985469e-06,
      "loss": 0.4045,
      "step": 2105
    },
    {
      "epoch": 0.13191562661488593,
      "grad_norm": 0.9478601818802093,
      "learning_rate": 9.730073456637408e-06,
      "loss": 0.4629,
      "step": 2106
    },
    {
      "epoch": 0.1319782646142282,
      "grad_norm": 0.8276304840136891,
      "learning_rate": 9.729744568598647e-06,
      "loss": 0.4156,
      "step": 2107
    },
    {
      "epoch": 0.13204090261357052,
      "grad_norm": 0.8405966541006353,
      "learning_rate": 9.72941548588272e-06,
      "loss": 0.4153,
      "step": 2108
    },
    {
      "epoch": 0.13210354061291282,
      "grad_norm": 0.8642937532057018,
      "learning_rate": 9.729086208503174e-06,
      "loss": 0.4368,
      "step": 2109
    },
    {
      "epoch": 0.13216617861225513,
      "grad_norm": 0.8864839571976726,
      "learning_rate": 9.728756736473562e-06,
      "loss": 0.441,
      "step": 2110
    },
    {
      "epoch": 0.13222881661159744,
      "grad_norm": 0.8918146164092077,
      "learning_rate": 9.728427069807445e-06,
      "loss": 0.4166,
      "step": 2111
    },
    {
      "epoch": 0.13229145461093972,
      "grad_norm": 0.8422247028600206,
      "learning_rate": 9.728097208518393e-06,
      "loss": 0.453,
      "step": 2112
    },
    {
      "epoch": 0.13235409261028203,
      "grad_norm": 0.8733152189483934,
      "learning_rate": 9.727767152619982e-06,
      "loss": 0.4331,
      "step": 2113
    },
    {
      "epoch": 0.13241673060962433,
      "grad_norm": 0.9782218870159587,
      "learning_rate": 9.727436902125798e-06,
      "loss": 0.4818,
      "step": 2114
    },
    {
      "epoch": 0.13247936860896664,
      "grad_norm": 0.8379718918800763,
      "learning_rate": 9.727106457049432e-06,
      "loss": 0.4522,
      "step": 2115
    },
    {
      "epoch": 0.13254200660830892,
      "grad_norm": 0.9283562791051059,
      "learning_rate": 9.726775817404488e-06,
      "loss": 0.5075,
      "step": 2116
    },
    {
      "epoch": 0.13260464460765123,
      "grad_norm": 0.8440467725165138,
      "learning_rate": 9.726444983204574e-06,
      "loss": 0.4354,
      "step": 2117
    },
    {
      "epoch": 0.13266728260699354,
      "grad_norm": 0.9012323417565896,
      "learning_rate": 9.726113954463307e-06,
      "loss": 0.4331,
      "step": 2118
    },
    {
      "epoch": 0.13272992060633584,
      "grad_norm": 1.9517194563470441,
      "learning_rate": 9.725782731194312e-06,
      "loss": 0.4242,
      "step": 2119
    },
    {
      "epoch": 0.13279255860567812,
      "grad_norm": 0.9405411734135994,
      "learning_rate": 9.725451313411223e-06,
      "loss": 0.4392,
      "step": 2120
    },
    {
      "epoch": 0.13285519660502043,
      "grad_norm": 0.8593712235727949,
      "learning_rate": 9.725119701127681e-06,
      "loss": 0.4425,
      "step": 2121
    },
    {
      "epoch": 0.13291783460436274,
      "grad_norm": 0.8768825684428532,
      "learning_rate": 9.724787894357335e-06,
      "loss": 0.4017,
      "step": 2122
    },
    {
      "epoch": 0.13298047260370505,
      "grad_norm": 0.8792789583527646,
      "learning_rate": 9.724455893113842e-06,
      "loss": 0.4536,
      "step": 2123
    },
    {
      "epoch": 0.13304311060304733,
      "grad_norm": 0.8799351477849479,
      "learning_rate": 9.724123697410867e-06,
      "loss": 0.4196,
      "step": 2124
    },
    {
      "epoch": 0.13310574860238963,
      "grad_norm": 0.8782838784237665,
      "learning_rate": 9.723791307262086e-06,
      "loss": 0.4627,
      "step": 2125
    },
    {
      "epoch": 0.13316838660173194,
      "grad_norm": 0.846872776750436,
      "learning_rate": 9.723458722681176e-06,
      "loss": 0.4432,
      "step": 2126
    },
    {
      "epoch": 0.13323102460107425,
      "grad_norm": 0.7840915035669741,
      "learning_rate": 9.72312594368183e-06,
      "loss": 0.4002,
      "step": 2127
    },
    {
      "epoch": 0.13329366260041653,
      "grad_norm": 0.8263191234037119,
      "learning_rate": 9.722792970277741e-06,
      "loss": 0.4216,
      "step": 2128
    },
    {
      "epoch": 0.13335630059975884,
      "grad_norm": 0.8476495730573337,
      "learning_rate": 9.72245980248262e-06,
      "loss": 0.4269,
      "step": 2129
    },
    {
      "epoch": 0.13341893859910114,
      "grad_norm": 0.8305705966718464,
      "learning_rate": 9.722126440310173e-06,
      "loss": 0.4047,
      "step": 2130
    },
    {
      "epoch": 0.13348157659844345,
      "grad_norm": 0.8449023565602183,
      "learning_rate": 9.721792883774128e-06,
      "loss": 0.4315,
      "step": 2131
    },
    {
      "epoch": 0.13354421459778576,
      "grad_norm": 0.8798460524317174,
      "learning_rate": 9.721459132888212e-06,
      "loss": 0.4597,
      "step": 2132
    },
    {
      "epoch": 0.13360685259712804,
      "grad_norm": 0.8249873330353092,
      "learning_rate": 9.721125187666161e-06,
      "loss": 0.4339,
      "step": 2133
    },
    {
      "epoch": 0.13366949059647035,
      "grad_norm": 0.9279176723886595,
      "learning_rate": 9.72079104812172e-06,
      "loss": 0.4601,
      "step": 2134
    },
    {
      "epoch": 0.13373212859581265,
      "grad_norm": 0.936381955049695,
      "learning_rate": 9.720456714268643e-06,
      "loss": 0.4548,
      "step": 2135
    },
    {
      "epoch": 0.13379476659515496,
      "grad_norm": 0.8995382763071785,
      "learning_rate": 9.720122186120693e-06,
      "loss": 0.5048,
      "step": 2136
    },
    {
      "epoch": 0.13385740459449724,
      "grad_norm": 0.8385715440903346,
      "learning_rate": 9.719787463691637e-06,
      "loss": 0.3964,
      "step": 2137
    },
    {
      "epoch": 0.13392004259383955,
      "grad_norm": 0.8257095846577366,
      "learning_rate": 9.719452546995252e-06,
      "loss": 0.485,
      "step": 2138
    },
    {
      "epoch": 0.13398268059318186,
      "grad_norm": 1.040061344649781,
      "learning_rate": 9.719117436045327e-06,
      "loss": 0.458,
      "step": 2139
    },
    {
      "epoch": 0.13404531859252417,
      "grad_norm": 0.7992353174131153,
      "learning_rate": 9.71878213085565e-06,
      "loss": 0.4938,
      "step": 2140
    },
    {
      "epoch": 0.13410795659186645,
      "grad_norm": 0.8048263921102105,
      "learning_rate": 9.718446631440025e-06,
      "loss": 0.4078,
      "step": 2141
    },
    {
      "epoch": 0.13417059459120875,
      "grad_norm": 0.8937308109052531,
      "learning_rate": 9.718110937812262e-06,
      "loss": 0.4387,
      "step": 2142
    },
    {
      "epoch": 0.13423323259055106,
      "grad_norm": 0.8307589142258672,
      "learning_rate": 9.717775049986176e-06,
      "loss": 0.3744,
      "step": 2143
    },
    {
      "epoch": 0.13429587058989337,
      "grad_norm": 0.8097826014244687,
      "learning_rate": 9.717438967975593e-06,
      "loss": 0.4269,
      "step": 2144
    },
    {
      "epoch": 0.13435850858923565,
      "grad_norm": 0.8944994559635281,
      "learning_rate": 9.717102691794347e-06,
      "loss": 0.4648,
      "step": 2145
    },
    {
      "epoch": 0.13442114658857796,
      "grad_norm": 0.9658348627088638,
      "learning_rate": 9.716766221456278e-06,
      "loss": 0.4717,
      "step": 2146
    },
    {
      "epoch": 0.13448378458792026,
      "grad_norm": 0.9393262925904621,
      "learning_rate": 9.716429556975237e-06,
      "loss": 0.4466,
      "step": 2147
    },
    {
      "epoch": 0.13454642258726257,
      "grad_norm": 0.8234086764367133,
      "learning_rate": 9.71609269836508e-06,
      "loss": 0.448,
      "step": 2148
    },
    {
      "epoch": 0.13460906058660485,
      "grad_norm": 0.7971101830165479,
      "learning_rate": 9.715755645639671e-06,
      "loss": 0.4575,
      "step": 2149
    },
    {
      "epoch": 0.13467169858594716,
      "grad_norm": 0.9360481326285544,
      "learning_rate": 9.715418398812886e-06,
      "loss": 0.4167,
      "step": 2150
    },
    {
      "epoch": 0.13473433658528947,
      "grad_norm": 0.8971146854140705,
      "learning_rate": 9.715080957898604e-06,
      "loss": 0.4418,
      "step": 2151
    },
    {
      "epoch": 0.13479697458463177,
      "grad_norm": 0.8213756117406368,
      "learning_rate": 9.714743322910714e-06,
      "loss": 0.4235,
      "step": 2152
    },
    {
      "epoch": 0.13485961258397408,
      "grad_norm": 0.8428002272901317,
      "learning_rate": 9.714405493863112e-06,
      "loss": 0.419,
      "step": 2153
    },
    {
      "epoch": 0.13492225058331636,
      "grad_norm": 1.0022217124068777,
      "learning_rate": 9.714067470769708e-06,
      "loss": 0.4504,
      "step": 2154
    },
    {
      "epoch": 0.13498488858265867,
      "grad_norm": 0.8954731967577586,
      "learning_rate": 9.71372925364441e-06,
      "loss": 0.4316,
      "step": 2155
    },
    {
      "epoch": 0.13504752658200098,
      "grad_norm": 0.9022354644447992,
      "learning_rate": 9.713390842501143e-06,
      "loss": 0.4303,
      "step": 2156
    },
    {
      "epoch": 0.13511016458134328,
      "grad_norm": 0.8228634737975942,
      "learning_rate": 9.713052237353833e-06,
      "loss": 0.4387,
      "step": 2157
    },
    {
      "epoch": 0.13517280258068556,
      "grad_norm": 0.9650432420131635,
      "learning_rate": 9.712713438216417e-06,
      "loss": 0.4721,
      "step": 2158
    },
    {
      "epoch": 0.13523544058002787,
      "grad_norm": 0.9045021161912801,
      "learning_rate": 9.712374445102842e-06,
      "loss": 0.4444,
      "step": 2159
    },
    {
      "epoch": 0.13529807857937018,
      "grad_norm": 0.8342035994388977,
      "learning_rate": 9.712035258027061e-06,
      "loss": 0.4275,
      "step": 2160
    },
    {
      "epoch": 0.1353607165787125,
      "grad_norm": 0.8807765871049515,
      "learning_rate": 9.711695877003033e-06,
      "loss": 0.4098,
      "step": 2161
    },
    {
      "epoch": 0.13542335457805477,
      "grad_norm": 0.8575262193259905,
      "learning_rate": 9.711356302044728e-06,
      "loss": 0.444,
      "step": 2162
    },
    {
      "epoch": 0.13548599257739707,
      "grad_norm": 0.8334092422121091,
      "learning_rate": 9.711016533166125e-06,
      "loss": 0.4427,
      "step": 2163
    },
    {
      "epoch": 0.13554863057673938,
      "grad_norm": 0.7888257518994014,
      "learning_rate": 9.710676570381205e-06,
      "loss": 0.4622,
      "step": 2164
    },
    {
      "epoch": 0.1356112685760817,
      "grad_norm": 0.94374525229091,
      "learning_rate": 9.710336413703964e-06,
      "loss": 0.4899,
      "step": 2165
    },
    {
      "epoch": 0.13567390657542397,
      "grad_norm": 0.8434059067039544,
      "learning_rate": 9.709996063148401e-06,
      "loss": 0.3999,
      "step": 2166
    },
    {
      "epoch": 0.13573654457476628,
      "grad_norm": 0.9040013981548884,
      "learning_rate": 9.709655518728526e-06,
      "loss": 0.4743,
      "step": 2167
    },
    {
      "epoch": 0.13579918257410858,
      "grad_norm": 0.8596217902106368,
      "learning_rate": 9.709314780458357e-06,
      "loss": 0.4224,
      "step": 2168
    },
    {
      "epoch": 0.1358618205734509,
      "grad_norm": 0.8824956620121363,
      "learning_rate": 9.708973848351917e-06,
      "loss": 0.4297,
      "step": 2169
    },
    {
      "epoch": 0.13592445857279317,
      "grad_norm": 0.9976441827894507,
      "learning_rate": 9.708632722423238e-06,
      "loss": 0.4558,
      "step": 2170
    },
    {
      "epoch": 0.13598709657213548,
      "grad_norm": 0.8811244329160518,
      "learning_rate": 9.708291402686365e-06,
      "loss": 0.4358,
      "step": 2171
    },
    {
      "epoch": 0.1360497345714778,
      "grad_norm": 0.9030085495743074,
      "learning_rate": 9.707949889155341e-06,
      "loss": 0.4073,
      "step": 2172
    },
    {
      "epoch": 0.1361123725708201,
      "grad_norm": 0.9428667186128505,
      "learning_rate": 9.707608181844228e-06,
      "loss": 0.456,
      "step": 2173
    },
    {
      "epoch": 0.1361750105701624,
      "grad_norm": 0.8664669935624583,
      "learning_rate": 9.707266280767086e-06,
      "loss": 0.4247,
      "step": 2174
    },
    {
      "epoch": 0.13623764856950468,
      "grad_norm": 0.8447772226283399,
      "learning_rate": 9.706924185937993e-06,
      "loss": 0.446,
      "step": 2175
    },
    {
      "epoch": 0.136300286568847,
      "grad_norm": 0.8795845193358592,
      "learning_rate": 9.706581897371025e-06,
      "loss": 0.4119,
      "step": 2176
    },
    {
      "epoch": 0.1363629245681893,
      "grad_norm": 0.8624702789665037,
      "learning_rate": 9.706239415080275e-06,
      "loss": 0.4659,
      "step": 2177
    },
    {
      "epoch": 0.1364255625675316,
      "grad_norm": 0.8349453064877617,
      "learning_rate": 9.705896739079834e-06,
      "loss": 0.437,
      "step": 2178
    },
    {
      "epoch": 0.13648820056687389,
      "grad_norm": 0.827452215672811,
      "learning_rate": 9.705553869383811e-06,
      "loss": 0.4811,
      "step": 2179
    },
    {
      "epoch": 0.1365508385662162,
      "grad_norm": 0.9400347248623704,
      "learning_rate": 9.705210806006317e-06,
      "loss": 0.4919,
      "step": 2180
    },
    {
      "epoch": 0.1366134765655585,
      "grad_norm": 0.8717365463924438,
      "learning_rate": 9.704867548961474e-06,
      "loss": 0.3825,
      "step": 2181
    },
    {
      "epoch": 0.1366761145649008,
      "grad_norm": 0.9083129104353631,
      "learning_rate": 9.704524098263409e-06,
      "loss": 0.48,
      "step": 2182
    },
    {
      "epoch": 0.1367387525642431,
      "grad_norm": 0.8940436638516562,
      "learning_rate": 9.704180453926259e-06,
      "loss": 0.4296,
      "step": 2183
    },
    {
      "epoch": 0.1368013905635854,
      "grad_norm": 0.9084071348706758,
      "learning_rate": 9.703836615964167e-06,
      "loss": 0.4674,
      "step": 2184
    },
    {
      "epoch": 0.1368640285629277,
      "grad_norm": 0.8560383043156078,
      "learning_rate": 9.703492584391287e-06,
      "loss": 0.4124,
      "step": 2185
    },
    {
      "epoch": 0.13692666656227,
      "grad_norm": 0.9280471870093759,
      "learning_rate": 9.70314835922178e-06,
      "loss": 0.4821,
      "step": 2186
    },
    {
      "epoch": 0.1369893045616123,
      "grad_norm": 0.9253114159966243,
      "learning_rate": 9.702803940469815e-06,
      "loss": 0.474,
      "step": 2187
    },
    {
      "epoch": 0.1370519425609546,
      "grad_norm": 0.9186481620054546,
      "learning_rate": 9.702459328149565e-06,
      "loss": 0.4561,
      "step": 2188
    },
    {
      "epoch": 0.1371145805602969,
      "grad_norm": 0.8745714419647423,
      "learning_rate": 9.702114522275216e-06,
      "loss": 0.3969,
      "step": 2189
    },
    {
      "epoch": 0.1371772185596392,
      "grad_norm": 0.8594754579311678,
      "learning_rate": 9.701769522860963e-06,
      "loss": 0.3859,
      "step": 2190
    },
    {
      "epoch": 0.1372398565589815,
      "grad_norm": 0.8158871232871461,
      "learning_rate": 9.701424329921e-06,
      "loss": 0.4359,
      "step": 2191
    },
    {
      "epoch": 0.1373024945583238,
      "grad_norm": 0.9492692006820466,
      "learning_rate": 9.701078943469541e-06,
      "loss": 0.4509,
      "step": 2192
    },
    {
      "epoch": 0.1373651325576661,
      "grad_norm": 0.9081073910917724,
      "learning_rate": 9.700733363520802e-06,
      "loss": 0.4883,
      "step": 2193
    },
    {
      "epoch": 0.13742777055700842,
      "grad_norm": 0.910381231696085,
      "learning_rate": 9.700387590089002e-06,
      "loss": 0.4467,
      "step": 2194
    },
    {
      "epoch": 0.13749040855635072,
      "grad_norm": 0.878024607270846,
      "learning_rate": 9.70004162318838e-06,
      "loss": 0.432,
      "step": 2195
    },
    {
      "epoch": 0.137553046555693,
      "grad_norm": 0.87043172632439,
      "learning_rate": 9.699695462833169e-06,
      "loss": 0.4187,
      "step": 2196
    },
    {
      "epoch": 0.1376156845550353,
      "grad_norm": 1.0506428282059537,
      "learning_rate": 9.69934910903762e-06,
      "loss": 0.4683,
      "step": 2197
    },
    {
      "epoch": 0.13767832255437762,
      "grad_norm": 0.9082872528130357,
      "learning_rate": 9.699002561815992e-06,
      "loss": 0.4021,
      "step": 2198
    },
    {
      "epoch": 0.13774096055371993,
      "grad_norm": 0.8044506720399468,
      "learning_rate": 9.698655821182545e-06,
      "loss": 0.3913,
      "step": 2199
    },
    {
      "epoch": 0.1378035985530622,
      "grad_norm": 0.9201163411558048,
      "learning_rate": 9.698308887151553e-06,
      "loss": 0.4585,
      "step": 2200
    },
    {
      "epoch": 0.13786623655240451,
      "grad_norm": 0.8784158770071561,
      "learning_rate": 9.697961759737296e-06,
      "loss": 0.4432,
      "step": 2201
    },
    {
      "epoch": 0.13792887455174682,
      "grad_norm": 0.9178607453899356,
      "learning_rate": 9.697614438954059e-06,
      "loss": 0.4665,
      "step": 2202
    },
    {
      "epoch": 0.13799151255108913,
      "grad_norm": 0.9245095424322415,
      "learning_rate": 9.69726692481614e-06,
      "loss": 0.4411,
      "step": 2203
    },
    {
      "epoch": 0.1380541505504314,
      "grad_norm": 0.8446563137255306,
      "learning_rate": 9.696919217337843e-06,
      "loss": 0.4222,
      "step": 2204
    },
    {
      "epoch": 0.13811678854977372,
      "grad_norm": 0.935595206797833,
      "learning_rate": 9.69657131653348e-06,
      "loss": 0.5224,
      "step": 2205
    },
    {
      "epoch": 0.13817942654911602,
      "grad_norm": 0.8359433085844524,
      "learning_rate": 9.696223222417369e-06,
      "loss": 0.4253,
      "step": 2206
    },
    {
      "epoch": 0.13824206454845833,
      "grad_norm": 0.8792681929527693,
      "learning_rate": 9.69587493500384e-06,
      "loss": 0.4279,
      "step": 2207
    },
    {
      "epoch": 0.1383047025478006,
      "grad_norm": 0.7472703270081213,
      "learning_rate": 9.695526454307224e-06,
      "loss": 0.477,
      "step": 2208
    },
    {
      "epoch": 0.13836734054714292,
      "grad_norm": 0.8463356530499532,
      "learning_rate": 9.69517778034187e-06,
      "loss": 0.4436,
      "step": 2209
    },
    {
      "epoch": 0.13842997854648523,
      "grad_norm": 0.8470588751770831,
      "learning_rate": 9.694828913122128e-06,
      "loss": 0.4511,
      "step": 2210
    },
    {
      "epoch": 0.13849261654582753,
      "grad_norm": 0.8999146246378132,
      "learning_rate": 9.694479852662355e-06,
      "loss": 0.4707,
      "step": 2211
    },
    {
      "epoch": 0.13855525454516981,
      "grad_norm": 0.7896702992132011,
      "learning_rate": 9.694130598976919e-06,
      "loss": 0.5006,
      "step": 2212
    },
    {
      "epoch": 0.13861789254451212,
      "grad_norm": 0.9847868740422208,
      "learning_rate": 9.693781152080197e-06,
      "loss": 0.4565,
      "step": 2213
    },
    {
      "epoch": 0.13868053054385443,
      "grad_norm": 0.812871343013172,
      "learning_rate": 9.693431511986572e-06,
      "loss": 0.3874,
      "step": 2214
    },
    {
      "epoch": 0.13874316854319674,
      "grad_norm": 0.8985133757051832,
      "learning_rate": 9.693081678710435e-06,
      "loss": 0.4683,
      "step": 2215
    },
    {
      "epoch": 0.13880580654253902,
      "grad_norm": 0.8695409837128261,
      "learning_rate": 9.692731652266184e-06,
      "loss": 0.415,
      "step": 2216
    },
    {
      "epoch": 0.13886844454188133,
      "grad_norm": 0.9286666645304484,
      "learning_rate": 9.692381432668228e-06,
      "loss": 0.4774,
      "step": 2217
    },
    {
      "epoch": 0.13893108254122363,
      "grad_norm": 0.8496265809508814,
      "learning_rate": 9.692031019930981e-06,
      "loss": 0.4513,
      "step": 2218
    },
    {
      "epoch": 0.13899372054056594,
      "grad_norm": 0.8483157555482311,
      "learning_rate": 9.691680414068867e-06,
      "loss": 0.4923,
      "step": 2219
    },
    {
      "epoch": 0.13905635853990825,
      "grad_norm": 0.8693471719479456,
      "learning_rate": 9.691329615096315e-06,
      "loss": 0.4921,
      "step": 2220
    },
    {
      "epoch": 0.13911899653925053,
      "grad_norm": 0.8789127971424072,
      "learning_rate": 9.690978623027767e-06,
      "loss": 0.4712,
      "step": 2221
    },
    {
      "epoch": 0.13918163453859284,
      "grad_norm": 0.8489954412603635,
      "learning_rate": 9.690627437877667e-06,
      "loss": 0.4244,
      "step": 2222
    },
    {
      "epoch": 0.13924427253793514,
      "grad_norm": 0.8608051690058997,
      "learning_rate": 9.690276059660471e-06,
      "loss": 0.4863,
      "step": 2223
    },
    {
      "epoch": 0.13930691053727745,
      "grad_norm": 0.8178938207427763,
      "learning_rate": 9.689924488390643e-06,
      "loss": 0.396,
      "step": 2224
    },
    {
      "epoch": 0.13936954853661973,
      "grad_norm": 0.7575938072179592,
      "learning_rate": 9.689572724082652e-06,
      "loss": 0.4776,
      "step": 2225
    },
    {
      "epoch": 0.13943218653596204,
      "grad_norm": 0.9320278124664211,
      "learning_rate": 9.689220766750977e-06,
      "loss": 0.412,
      "step": 2226
    },
    {
      "epoch": 0.13949482453530435,
      "grad_norm": 0.7923980175478729,
      "learning_rate": 9.688868616410105e-06,
      "loss": 0.4326,
      "step": 2227
    },
    {
      "epoch": 0.13955746253464665,
      "grad_norm": 0.8303756920232234,
      "learning_rate": 9.68851627307453e-06,
      "loss": 0.4713,
      "step": 2228
    },
    {
      "epoch": 0.13962010053398893,
      "grad_norm": 0.8249966179827555,
      "learning_rate": 9.688163736758755e-06,
      "loss": 0.3996,
      "step": 2229
    },
    {
      "epoch": 0.13968273853333124,
      "grad_norm": 0.8058591671027405,
      "learning_rate": 9.68781100747729e-06,
      "loss": 0.4192,
      "step": 2230
    },
    {
      "epoch": 0.13974537653267355,
      "grad_norm": 0.8307514861322093,
      "learning_rate": 9.687458085244654e-06,
      "loss": 0.4429,
      "step": 2231
    },
    {
      "epoch": 0.13980801453201586,
      "grad_norm": 0.8414700697569526,
      "learning_rate": 9.687104970075374e-06,
      "loss": 0.4442,
      "step": 2232
    },
    {
      "epoch": 0.13987065253135814,
      "grad_norm": 0.9248765424615424,
      "learning_rate": 9.686751661983985e-06,
      "loss": 0.4428,
      "step": 2233
    },
    {
      "epoch": 0.13993329053070044,
      "grad_norm": 0.8320833495374479,
      "learning_rate": 9.686398160985024e-06,
      "loss": 0.3943,
      "step": 2234
    },
    {
      "epoch": 0.13999592853004275,
      "grad_norm": 0.943264199577162,
      "learning_rate": 9.686044467093048e-06,
      "loss": 0.4594,
      "step": 2235
    },
    {
      "epoch": 0.14005856652938506,
      "grad_norm": 0.8929484255142927,
      "learning_rate": 9.685690580322612e-06,
      "loss": 0.4667,
      "step": 2236
    },
    {
      "epoch": 0.14012120452872734,
      "grad_norm": 0.8932555578946941,
      "learning_rate": 9.685336500688279e-06,
      "loss": 0.4765,
      "step": 2237
    },
    {
      "epoch": 0.14018384252806965,
      "grad_norm": 0.8786751361475417,
      "learning_rate": 9.684982228204627e-06,
      "loss": 0.4297,
      "step": 2238
    },
    {
      "epoch": 0.14024648052741195,
      "grad_norm": 0.8313577809048117,
      "learning_rate": 9.684627762886239e-06,
      "loss": 0.4041,
      "step": 2239
    },
    {
      "epoch": 0.14030911852675426,
      "grad_norm": 0.9980954618342258,
      "learning_rate": 9.684273104747702e-06,
      "loss": 0.4774,
      "step": 2240
    },
    {
      "epoch": 0.14037175652609657,
      "grad_norm": 0.9047714497836662,
      "learning_rate": 9.683918253803612e-06,
      "loss": 0.4895,
      "step": 2241
    },
    {
      "epoch": 0.14043439452543885,
      "grad_norm": 0.8658790837920405,
      "learning_rate": 9.683563210068581e-06,
      "loss": 0.4129,
      "step": 2242
    },
    {
      "epoch": 0.14049703252478116,
      "grad_norm": 0.9048304258728651,
      "learning_rate": 9.683207973557216e-06,
      "loss": 0.4093,
      "step": 2243
    },
    {
      "epoch": 0.14055967052412346,
      "grad_norm": 0.8785226375530205,
      "learning_rate": 9.682852544284143e-06,
      "loss": 0.4509,
      "step": 2244
    },
    {
      "epoch": 0.14062230852346577,
      "grad_norm": 0.8029841911522505,
      "learning_rate": 9.68249692226399e-06,
      "loss": 0.4208,
      "step": 2245
    },
    {
      "epoch": 0.14068494652280805,
      "grad_norm": 0.9905500653961982,
      "learning_rate": 9.682141107511394e-06,
      "loss": 0.4844,
      "step": 2246
    },
    {
      "epoch": 0.14074758452215036,
      "grad_norm": 0.8420769461406663,
      "learning_rate": 9.681785100041001e-06,
      "loss": 0.4461,
      "step": 2247
    },
    {
      "epoch": 0.14081022252149267,
      "grad_norm": 0.787049299246364,
      "learning_rate": 9.681428899867466e-06,
      "loss": 0.3771,
      "step": 2248
    },
    {
      "epoch": 0.14087286052083497,
      "grad_norm": 0.8946956054677999,
      "learning_rate": 9.681072507005446e-06,
      "loss": 0.4543,
      "step": 2249
    },
    {
      "epoch": 0.14093549852017725,
      "grad_norm": 0.8698706447603998,
      "learning_rate": 9.680715921469615e-06,
      "loss": 0.4233,
      "step": 2250
    },
    {
      "epoch": 0.14099813651951956,
      "grad_norm": 0.8173812336374683,
      "learning_rate": 9.680359143274646e-06,
      "loss": 0.3999,
      "step": 2251
    },
    {
      "epoch": 0.14106077451886187,
      "grad_norm": 0.9048758681462872,
      "learning_rate": 9.680002172435228e-06,
      "loss": 0.4622,
      "step": 2252
    },
    {
      "epoch": 0.14112341251820418,
      "grad_norm": 0.8488149212689138,
      "learning_rate": 9.67964500896605e-06,
      "loss": 0.3992,
      "step": 2253
    },
    {
      "epoch": 0.14118605051754646,
      "grad_norm": 0.7987622880014729,
      "learning_rate": 9.679287652881815e-06,
      "loss": 0.3994,
      "step": 2254
    },
    {
      "epoch": 0.14124868851688877,
      "grad_norm": 0.874658805330724,
      "learning_rate": 9.678930104197233e-06,
      "loss": 0.4336,
      "step": 2255
    },
    {
      "epoch": 0.14131132651623107,
      "grad_norm": 0.9343589650360308,
      "learning_rate": 9.678572362927019e-06,
      "loss": 0.4979,
      "step": 2256
    },
    {
      "epoch": 0.14137396451557338,
      "grad_norm": 0.7799738041422156,
      "learning_rate": 9.678214429085898e-06,
      "loss": 0.4706,
      "step": 2257
    },
    {
      "epoch": 0.14143660251491566,
      "grad_norm": 0.8247002457649069,
      "learning_rate": 9.677856302688603e-06,
      "loss": 0.4175,
      "step": 2258
    },
    {
      "epoch": 0.14149924051425797,
      "grad_norm": 0.8171683842298494,
      "learning_rate": 9.677497983749874e-06,
      "loss": 0.414,
      "step": 2259
    },
    {
      "epoch": 0.14156187851360028,
      "grad_norm": 0.9030862319938098,
      "learning_rate": 9.677139472284461e-06,
      "loss": 0.4389,
      "step": 2260
    },
    {
      "epoch": 0.14162451651294258,
      "grad_norm": 0.9153877789719159,
      "learning_rate": 9.676780768307117e-06,
      "loss": 0.4421,
      "step": 2261
    },
    {
      "epoch": 0.1416871545122849,
      "grad_norm": 0.8751589506147975,
      "learning_rate": 9.67642187183261e-06,
      "loss": 0.5034,
      "step": 2262
    },
    {
      "epoch": 0.14174979251162717,
      "grad_norm": 0.8585508014159221,
      "learning_rate": 9.67606278287571e-06,
      "loss": 0.3833,
      "step": 2263
    },
    {
      "epoch": 0.14181243051096948,
      "grad_norm": 0.7931676884838392,
      "learning_rate": 9.675703501451199e-06,
      "loss": 0.3535,
      "step": 2264
    },
    {
      "epoch": 0.14187506851031179,
      "grad_norm": 0.8367221322271434,
      "learning_rate": 9.675344027573862e-06,
      "loss": 0.4186,
      "step": 2265
    },
    {
      "epoch": 0.1419377065096541,
      "grad_norm": 0.905681798062385,
      "learning_rate": 9.6749843612585e-06,
      "loss": 0.4276,
      "step": 2266
    },
    {
      "epoch": 0.14200034450899637,
      "grad_norm": 0.9101200639522679,
      "learning_rate": 9.674624502519912e-06,
      "loss": 0.4274,
      "step": 2267
    },
    {
      "epoch": 0.14206298250833868,
      "grad_norm": 0.7942583281151744,
      "learning_rate": 9.674264451372912e-06,
      "loss": 0.4034,
      "step": 2268
    },
    {
      "epoch": 0.142125620507681,
      "grad_norm": 0.8092464420763034,
      "learning_rate": 9.67390420783232e-06,
      "loss": 0.4418,
      "step": 2269
    },
    {
      "epoch": 0.1421882585070233,
      "grad_norm": 0.8938713170876549,
      "learning_rate": 9.673543771912964e-06,
      "loss": 0.4477,
      "step": 2270
    },
    {
      "epoch": 0.14225089650636558,
      "grad_norm": 0.8638147614957213,
      "learning_rate": 9.673183143629677e-06,
      "loss": 0.3924,
      "step": 2271
    },
    {
      "epoch": 0.14231353450570788,
      "grad_norm": 0.7781302978492965,
      "learning_rate": 9.672822322997305e-06,
      "loss": 0.4937,
      "step": 2272
    },
    {
      "epoch": 0.1423761725050502,
      "grad_norm": 0.7949022480335897,
      "learning_rate": 9.6724613100307e-06,
      "loss": 0.4228,
      "step": 2273
    },
    {
      "epoch": 0.1424388105043925,
      "grad_norm": 0.8515484193980477,
      "learning_rate": 9.67210010474472e-06,
      "loss": 0.4449,
      "step": 2274
    },
    {
      "epoch": 0.14250144850373478,
      "grad_norm": 0.8726739340231814,
      "learning_rate": 9.67173870715423e-06,
      "loss": 0.4657,
      "step": 2275
    },
    {
      "epoch": 0.1425640865030771,
      "grad_norm": 0.8310758017733653,
      "learning_rate": 9.67137711727411e-06,
      "loss": 0.3929,
      "step": 2276
    },
    {
      "epoch": 0.1426267245024194,
      "grad_norm": 0.8435275578307975,
      "learning_rate": 9.67101533511924e-06,
      "loss": 0.4631,
      "step": 2277
    },
    {
      "epoch": 0.1426893625017617,
      "grad_norm": 0.8883069312600281,
      "learning_rate": 9.670653360704514e-06,
      "loss": 0.4367,
      "step": 2278
    },
    {
      "epoch": 0.14275200050110398,
      "grad_norm": 0.7980096992000741,
      "learning_rate": 9.670291194044828e-06,
      "loss": 0.4154,
      "step": 2279
    },
    {
      "epoch": 0.1428146385004463,
      "grad_norm": 0.9082874766371772,
      "learning_rate": 9.669928835155088e-06,
      "loss": 0.3995,
      "step": 2280
    },
    {
      "epoch": 0.1428772764997886,
      "grad_norm": 0.8583952157488394,
      "learning_rate": 9.66956628405021e-06,
      "loss": 0.4017,
      "step": 2281
    },
    {
      "epoch": 0.1429399144991309,
      "grad_norm": 0.8180762784395122,
      "learning_rate": 9.669203540745117e-06,
      "loss": 0.4772,
      "step": 2282
    },
    {
      "epoch": 0.1430025524984732,
      "grad_norm": 0.9187110631959239,
      "learning_rate": 9.668840605254742e-06,
      "loss": 0.4502,
      "step": 2283
    },
    {
      "epoch": 0.1430651904978155,
      "grad_norm": 0.894525408282474,
      "learning_rate": 9.668477477594021e-06,
      "loss": 0.4464,
      "step": 2284
    },
    {
      "epoch": 0.1431278284971578,
      "grad_norm": 0.8900166911968236,
      "learning_rate": 9.6681141577779e-06,
      "loss": 0.4304,
      "step": 2285
    },
    {
      "epoch": 0.1431904664965001,
      "grad_norm": 0.9252136322129633,
      "learning_rate": 9.667750645821332e-06,
      "loss": 0.4769,
      "step": 2286
    },
    {
      "epoch": 0.14325310449584241,
      "grad_norm": 0.8215437699935754,
      "learning_rate": 9.667386941739283e-06,
      "loss": 0.4374,
      "step": 2287
    },
    {
      "epoch": 0.1433157424951847,
      "grad_norm": 0.9114618504940122,
      "learning_rate": 9.66702304554672e-06,
      "loss": 0.4105,
      "step": 2288
    },
    {
      "epoch": 0.143378380494527,
      "grad_norm": 0.7957094822203379,
      "learning_rate": 9.66665895725862e-06,
      "loss": 0.4086,
      "step": 2289
    },
    {
      "epoch": 0.1434410184938693,
      "grad_norm": 0.8895002181398354,
      "learning_rate": 9.666294676889976e-06,
      "loss": 0.4286,
      "step": 2290
    },
    {
      "epoch": 0.14350365649321162,
      "grad_norm": 0.8611753123260439,
      "learning_rate": 9.665930204455773e-06,
      "loss": 0.4382,
      "step": 2291
    },
    {
      "epoch": 0.1435662944925539,
      "grad_norm": 0.8581690724377423,
      "learning_rate": 9.66556553997102e-06,
      "loss": 0.4256,
      "step": 2292
    },
    {
      "epoch": 0.1436289324918962,
      "grad_norm": 0.8692570522050129,
      "learning_rate": 9.665200683450721e-06,
      "loss": 0.4123,
      "step": 2293
    },
    {
      "epoch": 0.1436915704912385,
      "grad_norm": 0.8439746226052689,
      "learning_rate": 9.664835634909897e-06,
      "loss": 0.4337,
      "step": 2294
    },
    {
      "epoch": 0.14375420849058082,
      "grad_norm": 0.8431104044401114,
      "learning_rate": 9.664470394363571e-06,
      "loss": 0.4435,
      "step": 2295
    },
    {
      "epoch": 0.1438168464899231,
      "grad_norm": 1.013648199738999,
      "learning_rate": 9.664104961826781e-06,
      "loss": 0.4667,
      "step": 2296
    },
    {
      "epoch": 0.1438794844892654,
      "grad_norm": 0.8978294004178626,
      "learning_rate": 9.663739337314563e-06,
      "loss": 0.43,
      "step": 2297
    },
    {
      "epoch": 0.14394212248860772,
      "grad_norm": 0.8349920122563275,
      "learning_rate": 9.663373520841968e-06,
      "loss": 0.3829,
      "step": 2298
    },
    {
      "epoch": 0.14400476048795002,
      "grad_norm": 0.8114681141504396,
      "learning_rate": 9.663007512424053e-06,
      "loss": 0.3863,
      "step": 2299
    },
    {
      "epoch": 0.1440673984872923,
      "grad_norm": 0.8545206705842632,
      "learning_rate": 9.662641312075884e-06,
      "loss": 0.4271,
      "step": 2300
    },
    {
      "epoch": 0.1441300364866346,
      "grad_norm": 0.820906023156429,
      "learning_rate": 9.662274919812535e-06,
      "loss": 0.4151,
      "step": 2301
    },
    {
      "epoch": 0.14419267448597692,
      "grad_norm": 0.8755671797092478,
      "learning_rate": 9.661908335649082e-06,
      "loss": 0.4466,
      "step": 2302
    },
    {
      "epoch": 0.14425531248531923,
      "grad_norm": 0.9095921462898913,
      "learning_rate": 9.661541559600619e-06,
      "loss": 0.4432,
      "step": 2303
    },
    {
      "epoch": 0.14431795048466153,
      "grad_norm": 0.8438632242623119,
      "learning_rate": 9.661174591682238e-06,
      "loss": 0.4295,
      "step": 2304
    },
    {
      "epoch": 0.1443805884840038,
      "grad_norm": 0.8169418421448084,
      "learning_rate": 9.660807431909047e-06,
      "loss": 0.4224,
      "step": 2305
    },
    {
      "epoch": 0.14444322648334612,
      "grad_norm": 0.8801112277840498,
      "learning_rate": 9.660440080296156e-06,
      "loss": 0.4388,
      "step": 2306
    },
    {
      "epoch": 0.14450586448268843,
      "grad_norm": 0.8805325506577353,
      "learning_rate": 9.660072536858687e-06,
      "loss": 0.4697,
      "step": 2307
    },
    {
      "epoch": 0.14456850248203074,
      "grad_norm": 0.9083380947183523,
      "learning_rate": 9.659704801611767e-06,
      "loss": 0.4516,
      "step": 2308
    },
    {
      "epoch": 0.14463114048137302,
      "grad_norm": 0.8757638135949098,
      "learning_rate": 9.659336874570533e-06,
      "loss": 0.4121,
      "step": 2309
    },
    {
      "epoch": 0.14469377848071532,
      "grad_norm": 0.9218877110706876,
      "learning_rate": 9.658968755750129e-06,
      "loss": 0.4009,
      "step": 2310
    },
    {
      "epoch": 0.14475641648005763,
      "grad_norm": 0.8843122645902163,
      "learning_rate": 9.658600445165705e-06,
      "loss": 0.5088,
      "step": 2311
    },
    {
      "epoch": 0.14481905447939994,
      "grad_norm": 0.8845793209583669,
      "learning_rate": 9.658231942832423e-06,
      "loss": 0.4751,
      "step": 2312
    },
    {
      "epoch": 0.14488169247874222,
      "grad_norm": 0.8151727438848101,
      "learning_rate": 9.657863248765448e-06,
      "loss": 0.4258,
      "step": 2313
    },
    {
      "epoch": 0.14494433047808453,
      "grad_norm": 0.8910249107779263,
      "learning_rate": 9.657494362979958e-06,
      "loss": 0.4364,
      "step": 2314
    },
    {
      "epoch": 0.14500696847742683,
      "grad_norm": 0.7873203021090587,
      "learning_rate": 9.657125285491136e-06,
      "loss": 0.4892,
      "step": 2315
    },
    {
      "epoch": 0.14506960647676914,
      "grad_norm": 0.8857021235148637,
      "learning_rate": 9.656756016314173e-06,
      "loss": 0.4235,
      "step": 2316
    },
    {
      "epoch": 0.14513224447611142,
      "grad_norm": 0.7737993176409519,
      "learning_rate": 9.656386555464267e-06,
      "loss": 0.3882,
      "step": 2317
    },
    {
      "epoch": 0.14519488247545373,
      "grad_norm": 0.7799514869719378,
      "learning_rate": 9.656016902956628e-06,
      "loss": 0.4917,
      "step": 2318
    },
    {
      "epoch": 0.14525752047479604,
      "grad_norm": 0.8973805988302151,
      "learning_rate": 9.655647058806468e-06,
      "loss": 0.4447,
      "step": 2319
    },
    {
      "epoch": 0.14532015847413834,
      "grad_norm": 0.8802062428070762,
      "learning_rate": 9.65527702302901e-06,
      "loss": 0.4706,
      "step": 2320
    },
    {
      "epoch": 0.14538279647348062,
      "grad_norm": 0.9330544171777992,
      "learning_rate": 9.654906795639487e-06,
      "loss": 0.4395,
      "step": 2321
    },
    {
      "epoch": 0.14544543447282293,
      "grad_norm": 0.885867749513151,
      "learning_rate": 9.654536376653135e-06,
      "loss": 0.4048,
      "step": 2322
    },
    {
      "epoch": 0.14550807247216524,
      "grad_norm": 0.8209454098008089,
      "learning_rate": 9.654165766085203e-06,
      "loss": 0.4522,
      "step": 2323
    },
    {
      "epoch": 0.14557071047150755,
      "grad_norm": 0.8851687136485553,
      "learning_rate": 9.653794963950944e-06,
      "loss": 0.4232,
      "step": 2324
    },
    {
      "epoch": 0.14563334847084985,
      "grad_norm": 0.8694756484928869,
      "learning_rate": 9.65342397026562e-06,
      "loss": 0.4378,
      "step": 2325
    },
    {
      "epoch": 0.14569598647019213,
      "grad_norm": 0.9481573664617065,
      "learning_rate": 9.653052785044502e-06,
      "loss": 0.4719,
      "step": 2326
    },
    {
      "epoch": 0.14575862446953444,
      "grad_norm": 0.8907635164538238,
      "learning_rate": 9.652681408302867e-06,
      "loss": 0.4383,
      "step": 2327
    },
    {
      "epoch": 0.14582126246887675,
      "grad_norm": 0.8000978403199045,
      "learning_rate": 9.652309840056002e-06,
      "loss": 0.4739,
      "step": 2328
    },
    {
      "epoch": 0.14588390046821906,
      "grad_norm": 0.9085038364201059,
      "learning_rate": 9.651938080319202e-06,
      "loss": 0.4207,
      "step": 2329
    },
    {
      "epoch": 0.14594653846756134,
      "grad_norm": 0.8020984860767632,
      "learning_rate": 9.651566129107765e-06,
      "loss": 0.3948,
      "step": 2330
    },
    {
      "epoch": 0.14600917646690365,
      "grad_norm": 0.8541056741698139,
      "learning_rate": 9.651193986437004e-06,
      "loss": 0.4457,
      "step": 2331
    },
    {
      "epoch": 0.14607181446624595,
      "grad_norm": 0.769231972553952,
      "learning_rate": 9.650821652322236e-06,
      "loss": 0.4207,
      "step": 2332
    },
    {
      "epoch": 0.14613445246558826,
      "grad_norm": 0.8262319782302551,
      "learning_rate": 9.650449126778785e-06,
      "loss": 0.4316,
      "step": 2333
    },
    {
      "epoch": 0.14619709046493054,
      "grad_norm": 0.8754341312897409,
      "learning_rate": 9.650076409821984e-06,
      "loss": 0.4194,
      "step": 2334
    },
    {
      "epoch": 0.14625972846427285,
      "grad_norm": 0.9158670363535011,
      "learning_rate": 9.649703501467174e-06,
      "loss": 0.4103,
      "step": 2335
    },
    {
      "epoch": 0.14632236646361516,
      "grad_norm": 0.8282031772876082,
      "learning_rate": 9.649330401729707e-06,
      "loss": 0.4312,
      "step": 2336
    },
    {
      "epoch": 0.14638500446295746,
      "grad_norm": 0.9525534426862259,
      "learning_rate": 9.648957110624937e-06,
      "loss": 0.461,
      "step": 2337
    },
    {
      "epoch": 0.14644764246229974,
      "grad_norm": 0.8017443857930013,
      "learning_rate": 9.648583628168229e-06,
      "loss": 0.409,
      "step": 2338
    },
    {
      "epoch": 0.14651028046164205,
      "grad_norm": 0.8782016625081,
      "learning_rate": 9.648209954374956e-06,
      "loss": 0.5175,
      "step": 2339
    },
    {
      "epoch": 0.14657291846098436,
      "grad_norm": 0.8865723502101954,
      "learning_rate": 9.647836089260497e-06,
      "loss": 0.4587,
      "step": 2340
    },
    {
      "epoch": 0.14663555646032667,
      "grad_norm": 0.8038028190721217,
      "learning_rate": 9.647462032840245e-06,
      "loss": 0.4277,
      "step": 2341
    },
    {
      "epoch": 0.14669819445966895,
      "grad_norm": 0.8834983866029281,
      "learning_rate": 9.647087785129591e-06,
      "loss": 0.4094,
      "step": 2342
    },
    {
      "epoch": 0.14676083245901125,
      "grad_norm": 0.8870826228719058,
      "learning_rate": 9.646713346143943e-06,
      "loss": 0.484,
      "step": 2343
    },
    {
      "epoch": 0.14682347045835356,
      "grad_norm": 0.8482045660304712,
      "learning_rate": 9.64633871589871e-06,
      "loss": 0.4639,
      "step": 2344
    },
    {
      "epoch": 0.14688610845769587,
      "grad_norm": 0.8628826819093957,
      "learning_rate": 9.645963894409314e-06,
      "loss": 0.4274,
      "step": 2345
    },
    {
      "epoch": 0.14694874645703815,
      "grad_norm": 0.8866958023314192,
      "learning_rate": 9.64558888169118e-06,
      "loss": 0.4596,
      "step": 2346
    },
    {
      "epoch": 0.14701138445638046,
      "grad_norm": 0.9431210913352419,
      "learning_rate": 9.645213677759746e-06,
      "loss": 0.4865,
      "step": 2347
    },
    {
      "epoch": 0.14707402245572276,
      "grad_norm": 0.7302403598809101,
      "learning_rate": 9.644838282630456e-06,
      "loss": 0.4779,
      "step": 2348
    },
    {
      "epoch": 0.14713666045506507,
      "grad_norm": 0.9049631395261968,
      "learning_rate": 9.644462696318759e-06,
      "loss": 0.4214,
      "step": 2349
    },
    {
      "epoch": 0.14719929845440738,
      "grad_norm": 0.7340839389877459,
      "learning_rate": 9.644086918840116e-06,
      "loss": 0.4771,
      "step": 2350
    },
    {
      "epoch": 0.14726193645374966,
      "grad_norm": 0.8747651804543148,
      "learning_rate": 9.643710950209993e-06,
      "loss": 0.4562,
      "step": 2351
    },
    {
      "epoch": 0.14732457445309197,
      "grad_norm": 0.9470050537399178,
      "learning_rate": 9.643334790443866e-06,
      "loss": 0.4577,
      "step": 2352
    },
    {
      "epoch": 0.14738721245243427,
      "grad_norm": 0.8376829792620915,
      "learning_rate": 9.642958439557217e-06,
      "loss": 0.3923,
      "step": 2353
    },
    {
      "epoch": 0.14744985045177658,
      "grad_norm": 0.8470474210482474,
      "learning_rate": 9.642581897565535e-06,
      "loss": 0.4779,
      "step": 2354
    },
    {
      "epoch": 0.14751248845111886,
      "grad_norm": 0.847232525861142,
      "learning_rate": 9.642205164484323e-06,
      "loss": 0.4355,
      "step": 2355
    },
    {
      "epoch": 0.14757512645046117,
      "grad_norm": 0.8684491595549861,
      "learning_rate": 9.641828240329083e-06,
      "loss": 0.4951,
      "step": 2356
    },
    {
      "epoch": 0.14763776444980348,
      "grad_norm": 0.9849481700290545,
      "learning_rate": 9.641451125115333e-06,
      "loss": 0.4755,
      "step": 2357
    },
    {
      "epoch": 0.14770040244914578,
      "grad_norm": 0.8506513466862283,
      "learning_rate": 9.641073818858591e-06,
      "loss": 0.4153,
      "step": 2358
    },
    {
      "epoch": 0.14776304044848806,
      "grad_norm": 0.8373563000863815,
      "learning_rate": 9.64069632157439e-06,
      "loss": 0.4438,
      "step": 2359
    },
    {
      "epoch": 0.14782567844783037,
      "grad_norm": 0.9019029667949023,
      "learning_rate": 9.640318633278266e-06,
      "loss": 0.4265,
      "step": 2360
    },
    {
      "epoch": 0.14788831644717268,
      "grad_norm": 0.8266657327243677,
      "learning_rate": 9.639940753985768e-06,
      "loss": 0.4273,
      "step": 2361
    },
    {
      "epoch": 0.147950954446515,
      "grad_norm": 0.8581105506802421,
      "learning_rate": 9.639562683712446e-06,
      "loss": 0.4517,
      "step": 2362
    },
    {
      "epoch": 0.14801359244585727,
      "grad_norm": 0.8033684802975174,
      "learning_rate": 9.639184422473865e-06,
      "loss": 0.3757,
      "step": 2363
    },
    {
      "epoch": 0.14807623044519957,
      "grad_norm": 0.8270348263420059,
      "learning_rate": 9.63880597028559e-06,
      "loss": 0.433,
      "step": 2364
    },
    {
      "epoch": 0.14813886844454188,
      "grad_norm": 1.1928836555392626,
      "learning_rate": 9.6384273271632e-06,
      "loss": 0.5043,
      "step": 2365
    },
    {
      "epoch": 0.1482015064438842,
      "grad_norm": 0.8478188482045678,
      "learning_rate": 9.638048493122281e-06,
      "loss": 0.4759,
      "step": 2366
    },
    {
      "epoch": 0.14826414444322647,
      "grad_norm": 0.869010494005975,
      "learning_rate": 9.637669468178425e-06,
      "loss": 0.3903,
      "step": 2367
    },
    {
      "epoch": 0.14832678244256878,
      "grad_norm": 0.8759184839374334,
      "learning_rate": 9.637290252347233e-06,
      "loss": 0.3931,
      "step": 2368
    },
    {
      "epoch": 0.14838942044191109,
      "grad_norm": 0.9058054514862914,
      "learning_rate": 9.636910845644315e-06,
      "loss": 0.4364,
      "step": 2369
    },
    {
      "epoch": 0.1484520584412534,
      "grad_norm": 0.9176614748611075,
      "learning_rate": 9.636531248085284e-06,
      "loss": 0.4255,
      "step": 2370
    },
    {
      "epoch": 0.1485146964405957,
      "grad_norm": 0.8907698815948322,
      "learning_rate": 9.636151459685765e-06,
      "loss": 0.4879,
      "step": 2371
    },
    {
      "epoch": 0.14857733443993798,
      "grad_norm": 0.8698524916142856,
      "learning_rate": 9.635771480461393e-06,
      "loss": 0.4476,
      "step": 2372
    },
    {
      "epoch": 0.1486399724392803,
      "grad_norm": 0.8676108149051293,
      "learning_rate": 9.635391310427807e-06,
      "loss": 0.4554,
      "step": 2373
    },
    {
      "epoch": 0.1487026104386226,
      "grad_norm": 0.8811445140313584,
      "learning_rate": 9.635010949600654e-06,
      "loss": 0.4775,
      "step": 2374
    },
    {
      "epoch": 0.1487652484379649,
      "grad_norm": 0.9492777427326136,
      "learning_rate": 9.634630397995589e-06,
      "loss": 0.4233,
      "step": 2375
    },
    {
      "epoch": 0.14882788643730718,
      "grad_norm": 0.9036401016459509,
      "learning_rate": 9.634249655628278e-06,
      "loss": 0.4633,
      "step": 2376
    },
    {
      "epoch": 0.1488905244366495,
      "grad_norm": 0.833608396397723,
      "learning_rate": 9.63386872251439e-06,
      "loss": 0.4371,
      "step": 2377
    },
    {
      "epoch": 0.1489531624359918,
      "grad_norm": 0.9504202152230048,
      "learning_rate": 9.633487598669605e-06,
      "loss": 0.4615,
      "step": 2378
    },
    {
      "epoch": 0.1490158004353341,
      "grad_norm": 0.8516445229515681,
      "learning_rate": 9.633106284109612e-06,
      "loss": 0.4481,
      "step": 2379
    },
    {
      "epoch": 0.14907843843467639,
      "grad_norm": 0.7756864689633367,
      "learning_rate": 9.632724778850103e-06,
      "loss": 0.4192,
      "step": 2380
    },
    {
      "epoch": 0.1491410764340187,
      "grad_norm": 0.8075483737715474,
      "learning_rate": 9.632343082906782e-06,
      "loss": 0.4214,
      "step": 2381
    },
    {
      "epoch": 0.149203714433361,
      "grad_norm": 0.8495226140789064,
      "learning_rate": 9.631961196295361e-06,
      "loss": 0.4407,
      "step": 2382
    },
    {
      "epoch": 0.1492663524327033,
      "grad_norm": 0.8627696535186514,
      "learning_rate": 9.631579119031557e-06,
      "loss": 0.4603,
      "step": 2383
    },
    {
      "epoch": 0.1493289904320456,
      "grad_norm": 0.9833427205260712,
      "learning_rate": 9.631196851131097e-06,
      "loss": 0.48,
      "step": 2384
    },
    {
      "epoch": 0.1493916284313879,
      "grad_norm": 0.8499499621237054,
      "learning_rate": 9.630814392609715e-06,
      "loss": 0.4374,
      "step": 2385
    },
    {
      "epoch": 0.1494542664307302,
      "grad_norm": 0.7928901697601555,
      "learning_rate": 9.630431743483152e-06,
      "loss": 0.425,
      "step": 2386
    },
    {
      "epoch": 0.1495169044300725,
      "grad_norm": 0.789312482596649,
      "learning_rate": 9.63004890376716e-06,
      "loss": 0.3997,
      "step": 2387
    },
    {
      "epoch": 0.1495795424294148,
      "grad_norm": 0.8128200432999614,
      "learning_rate": 9.629665873477497e-06,
      "loss": 0.4733,
      "step": 2388
    },
    {
      "epoch": 0.1496421804287571,
      "grad_norm": 0.7615365811678368,
      "learning_rate": 9.629282652629926e-06,
      "loss": 0.4114,
      "step": 2389
    },
    {
      "epoch": 0.1497048184280994,
      "grad_norm": 0.8589118116776328,
      "learning_rate": 9.62889924124022e-06,
      "loss": 0.4227,
      "step": 2390
    },
    {
      "epoch": 0.14976745642744171,
      "grad_norm": 0.8291738206322965,
      "learning_rate": 9.628515639324165e-06,
      "loss": 0.4203,
      "step": 2391
    },
    {
      "epoch": 0.14983009442678402,
      "grad_norm": 0.8593183521226162,
      "learning_rate": 9.628131846897546e-06,
      "loss": 0.4076,
      "step": 2392
    },
    {
      "epoch": 0.1498927324261263,
      "grad_norm": 0.8611805723601197,
      "learning_rate": 9.62774786397616e-06,
      "loss": 0.4547,
      "step": 2393
    },
    {
      "epoch": 0.1499553704254686,
      "grad_norm": 0.8722064297884812,
      "learning_rate": 9.627363690575815e-06,
      "loss": 0.4357,
      "step": 2394
    },
    {
      "epoch": 0.15001800842481092,
      "grad_norm": 0.6701532249785436,
      "learning_rate": 9.626979326712321e-06,
      "loss": 0.4846,
      "step": 2395
    },
    {
      "epoch": 0.15008064642415322,
      "grad_norm": 0.8677723601696344,
      "learning_rate": 9.626594772401498e-06,
      "loss": 0.4274,
      "step": 2396
    },
    {
      "epoch": 0.1501432844234955,
      "grad_norm": 0.824606491064819,
      "learning_rate": 9.626210027659177e-06,
      "loss": 0.4092,
      "step": 2397
    },
    {
      "epoch": 0.1502059224228378,
      "grad_norm": 1.007816523083989,
      "learning_rate": 9.62582509250119e-06,
      "loss": 0.4573,
      "step": 2398
    },
    {
      "epoch": 0.15026856042218012,
      "grad_norm": 0.9008596761939137,
      "learning_rate": 9.625439966943388e-06,
      "loss": 0.4367,
      "step": 2399
    },
    {
      "epoch": 0.15033119842152243,
      "grad_norm": 0.8481409811805928,
      "learning_rate": 9.625054651001615e-06,
      "loss": 0.4408,
      "step": 2400
    },
    {
      "epoch": 0.1503938364208647,
      "grad_norm": 0.885431701320464,
      "learning_rate": 9.624669144691735e-06,
      "loss": 0.4359,
      "step": 2401
    },
    {
      "epoch": 0.15045647442020701,
      "grad_norm": 0.8792346507499824,
      "learning_rate": 9.624283448029614e-06,
      "loss": 0.4319,
      "step": 2402
    },
    {
      "epoch": 0.15051911241954932,
      "grad_norm": 0.8356309190399823,
      "learning_rate": 9.623897561031129e-06,
      "loss": 0.416,
      "step": 2403
    },
    {
      "epoch": 0.15058175041889163,
      "grad_norm": 0.8227366621498904,
      "learning_rate": 9.623511483712163e-06,
      "loss": 0.47,
      "step": 2404
    },
    {
      "epoch": 0.1506443884182339,
      "grad_norm": 0.8651954493910214,
      "learning_rate": 9.623125216088604e-06,
      "loss": 0.4446,
      "step": 2405
    },
    {
      "epoch": 0.15070702641757622,
      "grad_norm": 0.8956096972733216,
      "learning_rate": 9.622738758176353e-06,
      "loss": 0.4505,
      "step": 2406
    },
    {
      "epoch": 0.15076966441691853,
      "grad_norm": 0.7718114856052984,
      "learning_rate": 9.62235210999132e-06,
      "loss": 0.5232,
      "step": 2407
    },
    {
      "epoch": 0.15083230241626083,
      "grad_norm": 0.9553808029684103,
      "learning_rate": 9.621965271549413e-06,
      "loss": 0.4899,
      "step": 2408
    },
    {
      "epoch": 0.1508949404156031,
      "grad_norm": 0.8595795702334038,
      "learning_rate": 9.621578242866558e-06,
      "loss": 0.4228,
      "step": 2409
    },
    {
      "epoch": 0.15095757841494542,
      "grad_norm": 0.9160868208238513,
      "learning_rate": 9.621191023958685e-06,
      "loss": 0.4552,
      "step": 2410
    },
    {
      "epoch": 0.15102021641428773,
      "grad_norm": 0.861894225923158,
      "learning_rate": 9.620803614841732e-06,
      "loss": 0.4725,
      "step": 2411
    },
    {
      "epoch": 0.15108285441363004,
      "grad_norm": 1.0112295695950564,
      "learning_rate": 9.620416015531644e-06,
      "loss": 0.4231,
      "step": 2412
    },
    {
      "epoch": 0.15114549241297234,
      "grad_norm": 0.8482139232390409,
      "learning_rate": 9.620028226044375e-06,
      "loss": 0.4362,
      "step": 2413
    },
    {
      "epoch": 0.15120813041231462,
      "grad_norm": 0.9044934825325542,
      "learning_rate": 9.619640246395889e-06,
      "loss": 0.4288,
      "step": 2414
    },
    {
      "epoch": 0.15127076841165693,
      "grad_norm": 0.8497697181108379,
      "learning_rate": 9.61925207660215e-06,
      "loss": 0.4106,
      "step": 2415
    },
    {
      "epoch": 0.15133340641099924,
      "grad_norm": 0.8482769671397066,
      "learning_rate": 9.61886371667914e-06,
      "loss": 0.433,
      "step": 2416
    },
    {
      "epoch": 0.15139604441034155,
      "grad_norm": 0.814045684880283,
      "learning_rate": 9.618475166642842e-06,
      "loss": 0.4373,
      "step": 2417
    },
    {
      "epoch": 0.15145868240968383,
      "grad_norm": 0.8727904097314203,
      "learning_rate": 9.618086426509248e-06,
      "loss": 0.4555,
      "step": 2418
    },
    {
      "epoch": 0.15152132040902613,
      "grad_norm": 0.835947309276288,
      "learning_rate": 9.61769749629436e-06,
      "loss": 0.4032,
      "step": 2419
    },
    {
      "epoch": 0.15158395840836844,
      "grad_norm": 0.7115955313179162,
      "learning_rate": 9.617308376014185e-06,
      "loss": 0.466,
      "step": 2420
    },
    {
      "epoch": 0.15164659640771075,
      "grad_norm": 0.8135974158689896,
      "learning_rate": 9.616919065684742e-06,
      "loss": 0.4271,
      "step": 2421
    },
    {
      "epoch": 0.15170923440705303,
      "grad_norm": 0.6969122467335835,
      "learning_rate": 9.616529565322052e-06,
      "loss": 0.482,
      "step": 2422
    },
    {
      "epoch": 0.15177187240639534,
      "grad_norm": 0.9241893841615136,
      "learning_rate": 9.61613987494215e-06,
      "loss": 0.4669,
      "step": 2423
    },
    {
      "epoch": 0.15183451040573764,
      "grad_norm": 0.8772083729573474,
      "learning_rate": 9.615749994561073e-06,
      "loss": 0.4335,
      "step": 2424
    },
    {
      "epoch": 0.15189714840507995,
      "grad_norm": 0.9350423972844112,
      "learning_rate": 9.61535992419487e-06,
      "loss": 0.4994,
      "step": 2425
    },
    {
      "epoch": 0.15195978640442223,
      "grad_norm": 0.8936695294273029,
      "learning_rate": 9.614969663859593e-06,
      "loss": 0.4445,
      "step": 2426
    },
    {
      "epoch": 0.15202242440376454,
      "grad_norm": 0.9665407637378294,
      "learning_rate": 9.614579213571311e-06,
      "loss": 0.4593,
      "step": 2427
    },
    {
      "epoch": 0.15208506240310685,
      "grad_norm": 0.8440668994360617,
      "learning_rate": 9.614188573346091e-06,
      "loss": 0.433,
      "step": 2428
    },
    {
      "epoch": 0.15214770040244915,
      "grad_norm": 0.9084944508370191,
      "learning_rate": 9.613797743200013e-06,
      "loss": 0.4476,
      "step": 2429
    },
    {
      "epoch": 0.15221033840179143,
      "grad_norm": 0.8668678670879548,
      "learning_rate": 9.613406723149164e-06,
      "loss": 0.4325,
      "step": 2430
    },
    {
      "epoch": 0.15227297640113374,
      "grad_norm": 0.8912780229729776,
      "learning_rate": 9.613015513209638e-06,
      "loss": 0.4573,
      "step": 2431
    },
    {
      "epoch": 0.15233561440047605,
      "grad_norm": 0.8119934631695337,
      "learning_rate": 9.612624113397536e-06,
      "loss": 0.4576,
      "step": 2432
    },
    {
      "epoch": 0.15239825239981836,
      "grad_norm": 0.9263038513500895,
      "learning_rate": 9.61223252372897e-06,
      "loss": 0.4561,
      "step": 2433
    },
    {
      "epoch": 0.15246089039916066,
      "grad_norm": 0.8680718733685372,
      "learning_rate": 9.611840744220057e-06,
      "loss": 0.4414,
      "step": 2434
    },
    {
      "epoch": 0.15252352839850294,
      "grad_norm": 0.8579754161401948,
      "learning_rate": 9.611448774886925e-06,
      "loss": 0.4218,
      "step": 2435
    },
    {
      "epoch": 0.15258616639784525,
      "grad_norm": 0.9626726311010398,
      "learning_rate": 9.611056615745705e-06,
      "loss": 0.4636,
      "step": 2436
    },
    {
      "epoch": 0.15264880439718756,
      "grad_norm": 0.8446558390407034,
      "learning_rate": 9.610664266812537e-06,
      "loss": 0.4647,
      "step": 2437
    },
    {
      "epoch": 0.15271144239652987,
      "grad_norm": 0.8934984853225806,
      "learning_rate": 9.610271728103571e-06,
      "loss": 0.4555,
      "step": 2438
    },
    {
      "epoch": 0.15277408039587215,
      "grad_norm": 0.7864004236749215,
      "learning_rate": 9.609878999634968e-06,
      "loss": 0.3946,
      "step": 2439
    },
    {
      "epoch": 0.15283671839521445,
      "grad_norm": 0.8560483680413262,
      "learning_rate": 9.609486081422887e-06,
      "loss": 0.4561,
      "step": 2440
    },
    {
      "epoch": 0.15289935639455676,
      "grad_norm": 0.8856045989333212,
      "learning_rate": 9.609092973483507e-06,
      "loss": 0.4333,
      "step": 2441
    },
    {
      "epoch": 0.15296199439389907,
      "grad_norm": 0.8942781358042332,
      "learning_rate": 9.608699675833002e-06,
      "loss": 0.4621,
      "step": 2442
    },
    {
      "epoch": 0.15302463239324135,
      "grad_norm": 0.7929863034452728,
      "learning_rate": 9.608306188487562e-06,
      "loss": 0.414,
      "step": 2443
    },
    {
      "epoch": 0.15308727039258366,
      "grad_norm": 0.8458806410002347,
      "learning_rate": 9.607912511463387e-06,
      "loss": 0.4532,
      "step": 2444
    },
    {
      "epoch": 0.15314990839192597,
      "grad_norm": 0.8789685769306063,
      "learning_rate": 9.607518644776675e-06,
      "loss": 0.4139,
      "step": 2445
    },
    {
      "epoch": 0.15321254639126827,
      "grad_norm": 0.8081039208801672,
      "learning_rate": 9.607124588443642e-06,
      "loss": 0.391,
      "step": 2446
    },
    {
      "epoch": 0.15327518439061055,
      "grad_norm": 0.8191526921874455,
      "learning_rate": 9.606730342480506e-06,
      "loss": 0.4219,
      "step": 2447
    },
    {
      "epoch": 0.15333782238995286,
      "grad_norm": 0.8461760390351218,
      "learning_rate": 9.606335906903493e-06,
      "loss": 0.4443,
      "step": 2448
    },
    {
      "epoch": 0.15340046038929517,
      "grad_norm": 0.9170891887360324,
      "learning_rate": 9.60594128172884e-06,
      "loss": 0.4832,
      "step": 2449
    },
    {
      "epoch": 0.15346309838863748,
      "grad_norm": 0.8338661583762684,
      "learning_rate": 9.605546466972788e-06,
      "loss": 0.4289,
      "step": 2450
    },
    {
      "epoch": 0.15352573638797976,
      "grad_norm": 0.8765851636256804,
      "learning_rate": 9.60515146265159e-06,
      "loss": 0.4152,
      "step": 2451
    },
    {
      "epoch": 0.15358837438732206,
      "grad_norm": 0.8129336329477193,
      "learning_rate": 9.6047562687815e-06,
      "loss": 0.4403,
      "step": 2452
    },
    {
      "epoch": 0.15365101238666437,
      "grad_norm": 0.8859911281838742,
      "learning_rate": 9.604360885378789e-06,
      "loss": 0.4171,
      "step": 2453
    },
    {
      "epoch": 0.15371365038600668,
      "grad_norm": 0.9306802696233231,
      "learning_rate": 9.603965312459732e-06,
      "loss": 0.465,
      "step": 2454
    },
    {
      "epoch": 0.15377628838534896,
      "grad_norm": 0.8579268760034917,
      "learning_rate": 9.603569550040606e-06,
      "loss": 0.4326,
      "step": 2455
    },
    {
      "epoch": 0.15383892638469127,
      "grad_norm": 0.8885256382651785,
      "learning_rate": 9.603173598137702e-06,
      "loss": 0.4694,
      "step": 2456
    },
    {
      "epoch": 0.15390156438403357,
      "grad_norm": 0.8727621140846769,
      "learning_rate": 9.602777456767319e-06,
      "loss": 0.4411,
      "step": 2457
    },
    {
      "epoch": 0.15396420238337588,
      "grad_norm": 0.8016710456984827,
      "learning_rate": 9.60238112594576e-06,
      "loss": 0.3963,
      "step": 2458
    },
    {
      "epoch": 0.1540268403827182,
      "grad_norm": 0.8367745093692693,
      "learning_rate": 9.601984605689342e-06,
      "loss": 0.4231,
      "step": 2459
    },
    {
      "epoch": 0.15408947838206047,
      "grad_norm": 0.9247589241881171,
      "learning_rate": 9.601587896014383e-06,
      "loss": 0.4816,
      "step": 2460
    },
    {
      "epoch": 0.15415211638140278,
      "grad_norm": 0.8317591122459586,
      "learning_rate": 9.601190996937213e-06,
      "loss": 0.4404,
      "step": 2461
    },
    {
      "epoch": 0.15421475438074508,
      "grad_norm": 0.7865379627024374,
      "learning_rate": 9.600793908474166e-06,
      "loss": 0.4171,
      "step": 2462
    },
    {
      "epoch": 0.1542773923800874,
      "grad_norm": 0.8324815284051958,
      "learning_rate": 9.600396630641589e-06,
      "loss": 0.4265,
      "step": 2463
    },
    {
      "epoch": 0.15434003037942967,
      "grad_norm": 0.9061957444680231,
      "learning_rate": 9.59999916345583e-06,
      "loss": 0.4033,
      "step": 2464
    },
    {
      "epoch": 0.15440266837877198,
      "grad_norm": 0.8212193129654404,
      "learning_rate": 9.599601506933256e-06,
      "loss": 0.4149,
      "step": 2465
    },
    {
      "epoch": 0.1544653063781143,
      "grad_norm": 0.8900114906296487,
      "learning_rate": 9.599203661090226e-06,
      "loss": 0.4571,
      "step": 2466
    },
    {
      "epoch": 0.1545279443774566,
      "grad_norm": 0.9398715076321317,
      "learning_rate": 9.598805625943122e-06,
      "loss": 0.4354,
      "step": 2467
    },
    {
      "epoch": 0.15459058237679887,
      "grad_norm": 0.8760707797943218,
      "learning_rate": 9.598407401508325e-06,
      "loss": 0.4667,
      "step": 2468
    },
    {
      "epoch": 0.15465322037614118,
      "grad_norm": 0.7735964406430383,
      "learning_rate": 9.598008987802226e-06,
      "loss": 0.3842,
      "step": 2469
    },
    {
      "epoch": 0.1547158583754835,
      "grad_norm": 0.9276637589597163,
      "learning_rate": 9.597610384841222e-06,
      "loss": 0.4919,
      "step": 2470
    },
    {
      "epoch": 0.1547784963748258,
      "grad_norm": 0.8137203146992326,
      "learning_rate": 9.597211592641721e-06,
      "loss": 0.4407,
      "step": 2471
    },
    {
      "epoch": 0.15484113437416808,
      "grad_norm": 0.8715943761363466,
      "learning_rate": 9.596812611220139e-06,
      "loss": 0.4599,
      "step": 2472
    },
    {
      "epoch": 0.15490377237351038,
      "grad_norm": 0.9028416295026567,
      "learning_rate": 9.596413440592897e-06,
      "loss": 0.4177,
      "step": 2473
    },
    {
      "epoch": 0.1549664103728527,
      "grad_norm": 0.864732484161656,
      "learning_rate": 9.596014080776424e-06,
      "loss": 0.4412,
      "step": 2474
    },
    {
      "epoch": 0.155029048372195,
      "grad_norm": 0.9034319011181509,
      "learning_rate": 9.595614531787159e-06,
      "loss": 0.4455,
      "step": 2475
    },
    {
      "epoch": 0.15509168637153728,
      "grad_norm": 0.8204208193529091,
      "learning_rate": 9.595214793641546e-06,
      "loss": 0.405,
      "step": 2476
    },
    {
      "epoch": 0.1551543243708796,
      "grad_norm": 0.8500643692932301,
      "learning_rate": 9.594814866356039e-06,
      "loss": 0.4322,
      "step": 2477
    },
    {
      "epoch": 0.1552169623702219,
      "grad_norm": 0.8211132613232758,
      "learning_rate": 9.594414749947099e-06,
      "loss": 0.4461,
      "step": 2478
    },
    {
      "epoch": 0.1552796003695642,
      "grad_norm": 0.8656297821536006,
      "learning_rate": 9.594014444431196e-06,
      "loss": 0.4543,
      "step": 2479
    },
    {
      "epoch": 0.1553422383689065,
      "grad_norm": 0.8931750320715225,
      "learning_rate": 9.593613949824806e-06,
      "loss": 0.4087,
      "step": 2480
    },
    {
      "epoch": 0.1554048763682488,
      "grad_norm": 0.891000085506786,
      "learning_rate": 9.593213266144412e-06,
      "loss": 0.463,
      "step": 2481
    },
    {
      "epoch": 0.1554675143675911,
      "grad_norm": 0.9658209563212905,
      "learning_rate": 9.592812393406509e-06,
      "loss": 0.4494,
      "step": 2482
    },
    {
      "epoch": 0.1555301523669334,
      "grad_norm": 0.852049062205769,
      "learning_rate": 9.592411331627593e-06,
      "loss": 0.4391,
      "step": 2483
    },
    {
      "epoch": 0.1555927903662757,
      "grad_norm": 0.9895709281234618,
      "learning_rate": 9.592010080824177e-06,
      "loss": 0.4898,
      "step": 2484
    },
    {
      "epoch": 0.155655428365618,
      "grad_norm": 1.001154502257721,
      "learning_rate": 9.591608641012773e-06,
      "loss": 0.4654,
      "step": 2485
    },
    {
      "epoch": 0.1557180663649603,
      "grad_norm": 0.8621538640480553,
      "learning_rate": 9.591207012209905e-06,
      "loss": 0.4575,
      "step": 2486
    },
    {
      "epoch": 0.1557807043643026,
      "grad_norm": 0.8592081363336459,
      "learning_rate": 9.590805194432103e-06,
      "loss": 0.4342,
      "step": 2487
    },
    {
      "epoch": 0.15584334236364492,
      "grad_norm": 0.9036978090961402,
      "learning_rate": 9.59040318769591e-06,
      "loss": 0.4761,
      "step": 2488
    },
    {
      "epoch": 0.1559059803629872,
      "grad_norm": 0.8386593291023011,
      "learning_rate": 9.590000992017867e-06,
      "loss": 0.4824,
      "step": 2489
    },
    {
      "epoch": 0.1559686183623295,
      "grad_norm": 0.8750595485449342,
      "learning_rate": 9.589598607414531e-06,
      "loss": 0.3938,
      "step": 2490
    },
    {
      "epoch": 0.1560312563616718,
      "grad_norm": 0.9135826174710197,
      "learning_rate": 9.589196033902465e-06,
      "loss": 0.4757,
      "step": 2491
    },
    {
      "epoch": 0.15609389436101412,
      "grad_norm": 0.8668451770535498,
      "learning_rate": 9.58879327149824e-06,
      "loss": 0.4689,
      "step": 2492
    },
    {
      "epoch": 0.1561565323603564,
      "grad_norm": 0.8848214317959661,
      "learning_rate": 9.58839032021843e-06,
      "loss": 0.466,
      "step": 2493
    },
    {
      "epoch": 0.1562191703596987,
      "grad_norm": 0.8415634066432458,
      "learning_rate": 9.587987180079626e-06,
      "loss": 0.4266,
      "step": 2494
    },
    {
      "epoch": 0.156281808359041,
      "grad_norm": 0.8438172158473972,
      "learning_rate": 9.587583851098414e-06,
      "loss": 0.491,
      "step": 2495
    },
    {
      "epoch": 0.15634444635838332,
      "grad_norm": 0.9542748483301218,
      "learning_rate": 9.587180333291402e-06,
      "loss": 0.3992,
      "step": 2496
    },
    {
      "epoch": 0.1564070843577256,
      "grad_norm": 0.8051876077893441,
      "learning_rate": 9.586776626675195e-06,
      "loss": 0.4045,
      "step": 2497
    },
    {
      "epoch": 0.1564697223570679,
      "grad_norm": 0.8581728797269185,
      "learning_rate": 9.586372731266412e-06,
      "loss": 0.3971,
      "step": 2498
    },
    {
      "epoch": 0.15653236035641022,
      "grad_norm": 0.8369257112601043,
      "learning_rate": 9.585968647081674e-06,
      "loss": 0.424,
      "step": 2499
    },
    {
      "epoch": 0.15659499835575252,
      "grad_norm": 0.8503933370686396,
      "learning_rate": 9.585564374137617e-06,
      "loss": 0.4581,
      "step": 2500
    },
    {
      "epoch": 0.15665763635509483,
      "grad_norm": 0.8279242480143701,
      "learning_rate": 9.58515991245088e-06,
      "loss": 0.4087,
      "step": 2501
    },
    {
      "epoch": 0.1567202743544371,
      "grad_norm": 0.7982035839231489,
      "learning_rate": 9.584755262038106e-06,
      "loss": 0.3957,
      "step": 2502
    },
    {
      "epoch": 0.15678291235377942,
      "grad_norm": 0.8035537044448254,
      "learning_rate": 9.584350422915959e-06,
      "loss": 0.4113,
      "step": 2503
    },
    {
      "epoch": 0.15684555035312173,
      "grad_norm": 0.8769068026035574,
      "learning_rate": 9.583945395101095e-06,
      "loss": 0.4605,
      "step": 2504
    },
    {
      "epoch": 0.15690818835246403,
      "grad_norm": 0.8176030744920416,
      "learning_rate": 9.583540178610189e-06,
      "loss": 0.4266,
      "step": 2505
    },
    {
      "epoch": 0.15697082635180631,
      "grad_norm": 0.912999599012218,
      "learning_rate": 9.583134773459917e-06,
      "loss": 0.4341,
      "step": 2506
    },
    {
      "epoch": 0.15703346435114862,
      "grad_norm": 0.8424329714768108,
      "learning_rate": 9.582729179666967e-06,
      "loss": 0.4471,
      "step": 2507
    },
    {
      "epoch": 0.15709610235049093,
      "grad_norm": 0.8735793136809137,
      "learning_rate": 9.582323397248035e-06,
      "loss": 0.475,
      "step": 2508
    },
    {
      "epoch": 0.15715874034983324,
      "grad_norm": 0.8269125150035503,
      "learning_rate": 9.58191742621982e-06,
      "loss": 0.4137,
      "step": 2509
    },
    {
      "epoch": 0.15722137834917552,
      "grad_norm": 0.8464295914905088,
      "learning_rate": 9.581511266599036e-06,
      "loss": 0.453,
      "step": 2510
    },
    {
      "epoch": 0.15728401634851782,
      "grad_norm": 0.9307725791557825,
      "learning_rate": 9.581104918402395e-06,
      "loss": 0.4758,
      "step": 2511
    },
    {
      "epoch": 0.15734665434786013,
      "grad_norm": 0.8682580263120265,
      "learning_rate": 9.580698381646626e-06,
      "loss": 0.4341,
      "step": 2512
    },
    {
      "epoch": 0.15740929234720244,
      "grad_norm": 0.8535748257768625,
      "learning_rate": 9.580291656348462e-06,
      "loss": 0.4323,
      "step": 2513
    },
    {
      "epoch": 0.15747193034654472,
      "grad_norm": 0.9471617759836005,
      "learning_rate": 9.579884742524644e-06,
      "loss": 0.5298,
      "step": 2514
    },
    {
      "epoch": 0.15753456834588703,
      "grad_norm": 0.8418125312666805,
      "learning_rate": 9.579477640191917e-06,
      "loss": 0.425,
      "step": 2515
    },
    {
      "epoch": 0.15759720634522933,
      "grad_norm": 0.8691617366885266,
      "learning_rate": 9.579070349367044e-06,
      "loss": 0.4347,
      "step": 2516
    },
    {
      "epoch": 0.15765984434457164,
      "grad_norm": 0.8279703310738359,
      "learning_rate": 9.578662870066783e-06,
      "loss": 0.415,
      "step": 2517
    },
    {
      "epoch": 0.15772248234391392,
      "grad_norm": 0.8138718972081685,
      "learning_rate": 9.578255202307909e-06,
      "loss": 0.4152,
      "step": 2518
    },
    {
      "epoch": 0.15778512034325623,
      "grad_norm": 0.8575941917600187,
      "learning_rate": 9.5778473461072e-06,
      "loss": 0.3966,
      "step": 2519
    },
    {
      "epoch": 0.15784775834259854,
      "grad_norm": 0.8118657391396819,
      "learning_rate": 9.577439301481448e-06,
      "loss": 0.422,
      "step": 2520
    },
    {
      "epoch": 0.15791039634194085,
      "grad_norm": 0.8920240296353231,
      "learning_rate": 9.577031068447441e-06,
      "loss": 0.4285,
      "step": 2521
    },
    {
      "epoch": 0.15797303434128315,
      "grad_norm": 0.8470642872928104,
      "learning_rate": 9.576622647021988e-06,
      "loss": 0.4639,
      "step": 2522
    },
    {
      "epoch": 0.15803567234062543,
      "grad_norm": 0.8956732117373375,
      "learning_rate": 9.576214037221895e-06,
      "loss": 0.4508,
      "step": 2523
    },
    {
      "epoch": 0.15809831033996774,
      "grad_norm": 0.8853283669132453,
      "learning_rate": 9.575805239063985e-06,
      "loss": 0.4811,
      "step": 2524
    },
    {
      "epoch": 0.15816094833931005,
      "grad_norm": 0.824320677832317,
      "learning_rate": 9.575396252565083e-06,
      "loss": 0.3958,
      "step": 2525
    },
    {
      "epoch": 0.15822358633865236,
      "grad_norm": 0.8085501844761102,
      "learning_rate": 9.574987077742019e-06,
      "loss": 0.4232,
      "step": 2526
    },
    {
      "epoch": 0.15828622433799464,
      "grad_norm": 0.8160468566400376,
      "learning_rate": 9.57457771461164e-06,
      "loss": 0.4292,
      "step": 2527
    },
    {
      "epoch": 0.15834886233733694,
      "grad_norm": 0.9197057185278228,
      "learning_rate": 9.574168163190792e-06,
      "loss": 0.4714,
      "step": 2528
    },
    {
      "epoch": 0.15841150033667925,
      "grad_norm": 0.911108862126394,
      "learning_rate": 9.573758423496334e-06,
      "loss": 0.471,
      "step": 2529
    },
    {
      "epoch": 0.15847413833602156,
      "grad_norm": 0.9216576968304545,
      "learning_rate": 9.57334849554513e-06,
      "loss": 0.4666,
      "step": 2530
    },
    {
      "epoch": 0.15853677633536384,
      "grad_norm": 0.844955688177425,
      "learning_rate": 9.572938379354053e-06,
      "loss": 0.4315,
      "step": 2531
    },
    {
      "epoch": 0.15859941433470615,
      "grad_norm": 0.9138515211089059,
      "learning_rate": 9.572528074939985e-06,
      "loss": 0.4552,
      "step": 2532
    },
    {
      "epoch": 0.15866205233404845,
      "grad_norm": 0.8223979468975695,
      "learning_rate": 9.57211758231981e-06,
      "loss": 0.426,
      "step": 2533
    },
    {
      "epoch": 0.15872469033339076,
      "grad_norm": 0.8850696018432389,
      "learning_rate": 9.571706901510429e-06,
      "loss": 0.4484,
      "step": 2534
    },
    {
      "epoch": 0.15878732833273304,
      "grad_norm": 0.8804621297362388,
      "learning_rate": 9.571296032528744e-06,
      "loss": 0.4432,
      "step": 2535
    },
    {
      "epoch": 0.15884996633207535,
      "grad_norm": 0.8425866609014727,
      "learning_rate": 9.570884975391662e-06,
      "loss": 0.4449,
      "step": 2536
    },
    {
      "epoch": 0.15891260433141766,
      "grad_norm": 0.8562809460415205,
      "learning_rate": 9.57047373011611e-06,
      "loss": 0.4351,
      "step": 2537
    },
    {
      "epoch": 0.15897524233075996,
      "grad_norm": 0.840665576556803,
      "learning_rate": 9.570062296719009e-06,
      "loss": 0.4128,
      "step": 2538
    },
    {
      "epoch": 0.15903788033010224,
      "grad_norm": 0.8165823805791668,
      "learning_rate": 9.569650675217295e-06,
      "loss": 0.3964,
      "step": 2539
    },
    {
      "epoch": 0.15910051832944455,
      "grad_norm": 0.8931891364782731,
      "learning_rate": 9.569238865627913e-06,
      "loss": 0.423,
      "step": 2540
    },
    {
      "epoch": 0.15916315632878686,
      "grad_norm": 0.8270748795527492,
      "learning_rate": 9.56882686796781e-06,
      "loss": 0.4155,
      "step": 2541
    },
    {
      "epoch": 0.15922579432812917,
      "grad_norm": 0.8369468855852289,
      "learning_rate": 9.568414682253946e-06,
      "loss": 0.4627,
      "step": 2542
    },
    {
      "epoch": 0.15928843232747147,
      "grad_norm": 0.8036040809493677,
      "learning_rate": 9.568002308503285e-06,
      "loss": 0.4903,
      "step": 2543
    },
    {
      "epoch": 0.15935107032681375,
      "grad_norm": 0.8780574109551809,
      "learning_rate": 9.5675897467328e-06,
      "loss": 0.4483,
      "step": 2544
    },
    {
      "epoch": 0.15941370832615606,
      "grad_norm": 0.88611768799201,
      "learning_rate": 9.567176996959475e-06,
      "loss": 0.4309,
      "step": 2545
    },
    {
      "epoch": 0.15947634632549837,
      "grad_norm": 0.7403056035678959,
      "learning_rate": 9.566764059200296e-06,
      "loss": 0.4967,
      "step": 2546
    },
    {
      "epoch": 0.15953898432484068,
      "grad_norm": 0.8779093338235935,
      "learning_rate": 9.566350933472262e-06,
      "loss": 0.3897,
      "step": 2547
    },
    {
      "epoch": 0.15960162232418296,
      "grad_norm": 0.835235471937895,
      "learning_rate": 9.565937619792375e-06,
      "loss": 0.4165,
      "step": 2548
    },
    {
      "epoch": 0.15966426032352526,
      "grad_norm": 0.8403329879364821,
      "learning_rate": 9.56552411817765e-06,
      "loss": 0.4379,
      "step": 2549
    },
    {
      "epoch": 0.15972689832286757,
      "grad_norm": 0.8695072504748369,
      "learning_rate": 9.5651104286451e-06,
      "loss": 0.4467,
      "step": 2550
    },
    {
      "epoch": 0.15978953632220988,
      "grad_norm": 0.8002200594838434,
      "learning_rate": 9.564696551211763e-06,
      "loss": 0.3966,
      "step": 2551
    },
    {
      "epoch": 0.15985217432155216,
      "grad_norm": 0.8944586092047815,
      "learning_rate": 9.564282485894665e-06,
      "loss": 0.409,
      "step": 2552
    },
    {
      "epoch": 0.15991481232089447,
      "grad_norm": 0.8037714890905621,
      "learning_rate": 9.563868232710855e-06,
      "loss": 0.4496,
      "step": 2553
    },
    {
      "epoch": 0.15997745032023677,
      "grad_norm": 0.839842913644358,
      "learning_rate": 9.56345379167738e-06,
      "loss": 0.4222,
      "step": 2554
    },
    {
      "epoch": 0.16004008831957908,
      "grad_norm": 0.8167535139595054,
      "learning_rate": 9.563039162811301e-06,
      "loss": 0.4217,
      "step": 2555
    },
    {
      "epoch": 0.16010272631892136,
      "grad_norm": 0.8519238483656325,
      "learning_rate": 9.562624346129683e-06,
      "loss": 0.4406,
      "step": 2556
    },
    {
      "epoch": 0.16016536431826367,
      "grad_norm": 0.8680749249414098,
      "learning_rate": 9.562209341649598e-06,
      "loss": 0.4526,
      "step": 2557
    },
    {
      "epoch": 0.16022800231760598,
      "grad_norm": 0.8789017541723526,
      "learning_rate": 9.561794149388132e-06,
      "loss": 0.4417,
      "step": 2558
    },
    {
      "epoch": 0.16029064031694829,
      "grad_norm": 0.8723504599285197,
      "learning_rate": 9.561378769362372e-06,
      "loss": 0.4274,
      "step": 2559
    },
    {
      "epoch": 0.16035327831629057,
      "grad_norm": 0.79947706288344,
      "learning_rate": 9.560963201589416e-06,
      "loss": 0.4562,
      "step": 2560
    },
    {
      "epoch": 0.16041591631563287,
      "grad_norm": 0.8652676750965085,
      "learning_rate": 9.560547446086366e-06,
      "loss": 0.4509,
      "step": 2561
    },
    {
      "epoch": 0.16047855431497518,
      "grad_norm": 0.9135956823633377,
      "learning_rate": 9.560131502870335e-06,
      "loss": 0.4638,
      "step": 2562
    },
    {
      "epoch": 0.1605411923143175,
      "grad_norm": 0.9596051762974733,
      "learning_rate": 9.55971537195845e-06,
      "loss": 0.4404,
      "step": 2563
    },
    {
      "epoch": 0.1606038303136598,
      "grad_norm": 0.8404367915446014,
      "learning_rate": 9.55929905336783e-06,
      "loss": 0.4272,
      "step": 2564
    },
    {
      "epoch": 0.16066646831300208,
      "grad_norm": 0.8639319970923328,
      "learning_rate": 9.558882547115616e-06,
      "loss": 0.4671,
      "step": 2565
    },
    {
      "epoch": 0.16072910631234438,
      "grad_norm": 0.8775147043363833,
      "learning_rate": 9.55846585321895e-06,
      "loss": 0.4421,
      "step": 2566
    },
    {
      "epoch": 0.1607917443116867,
      "grad_norm": 0.8676708367020642,
      "learning_rate": 9.558048971694984e-06,
      "loss": 0.3875,
      "step": 2567
    },
    {
      "epoch": 0.160854382311029,
      "grad_norm": 0.8915820871769075,
      "learning_rate": 9.557631902560876e-06,
      "loss": 0.4102,
      "step": 2568
    },
    {
      "epoch": 0.16091702031037128,
      "grad_norm": 0.8349371890279598,
      "learning_rate": 9.557214645833792e-06,
      "loss": 0.4208,
      "step": 2569
    },
    {
      "epoch": 0.16097965830971359,
      "grad_norm": 0.8912421591791062,
      "learning_rate": 9.556797201530908e-06,
      "loss": 0.4547,
      "step": 2570
    },
    {
      "epoch": 0.1610422963090559,
      "grad_norm": 0.8836316722571188,
      "learning_rate": 9.556379569669406e-06,
      "loss": 0.458,
      "step": 2571
    },
    {
      "epoch": 0.1611049343083982,
      "grad_norm": 0.9030332939927537,
      "learning_rate": 9.555961750266475e-06,
      "loss": 0.5007,
      "step": 2572
    },
    {
      "epoch": 0.16116757230774048,
      "grad_norm": 0.8871465403605112,
      "learning_rate": 9.555543743339312e-06,
      "loss": 0.454,
      "step": 2573
    },
    {
      "epoch": 0.1612302103070828,
      "grad_norm": 0.8847752193321187,
      "learning_rate": 9.555125548905123e-06,
      "loss": 0.3685,
      "step": 2574
    },
    {
      "epoch": 0.1612928483064251,
      "grad_norm": 0.8632241759768705,
      "learning_rate": 9.554707166981122e-06,
      "loss": 0.4133,
      "step": 2575
    },
    {
      "epoch": 0.1613554863057674,
      "grad_norm": 1.0240420403273873,
      "learning_rate": 9.55428859758453e-06,
      "loss": 0.4521,
      "step": 2576
    },
    {
      "epoch": 0.16141812430510968,
      "grad_norm": 0.9089817435399401,
      "learning_rate": 9.55386984073257e-06,
      "loss": 0.3985,
      "step": 2577
    },
    {
      "epoch": 0.161480762304452,
      "grad_norm": 0.8665045743026693,
      "learning_rate": 9.553450896442484e-06,
      "loss": 0.4591,
      "step": 2578
    },
    {
      "epoch": 0.1615434003037943,
      "grad_norm": 0.9349600363179519,
      "learning_rate": 9.553031764731516e-06,
      "loss": 0.5081,
      "step": 2579
    },
    {
      "epoch": 0.1616060383031366,
      "grad_norm": 0.8236522905540098,
      "learning_rate": 9.552612445616912e-06,
      "loss": 0.4414,
      "step": 2580
    },
    {
      "epoch": 0.1616686763024789,
      "grad_norm": 0.8376150834385812,
      "learning_rate": 9.552192939115936e-06,
      "loss": 0.3956,
      "step": 2581
    },
    {
      "epoch": 0.1617313143018212,
      "grad_norm": 0.838462065433609,
      "learning_rate": 9.551773245245854e-06,
      "loss": 0.4226,
      "step": 2582
    },
    {
      "epoch": 0.1617939523011635,
      "grad_norm": 0.8599953816256034,
      "learning_rate": 9.55135336402394e-06,
      "loss": 0.4549,
      "step": 2583
    },
    {
      "epoch": 0.1618565903005058,
      "grad_norm": 0.9240649040907671,
      "learning_rate": 9.550933295467478e-06,
      "loss": 0.472,
      "step": 2584
    },
    {
      "epoch": 0.1619192282998481,
      "grad_norm": 0.9273151361803083,
      "learning_rate": 9.550513039593756e-06,
      "loss": 0.4412,
      "step": 2585
    },
    {
      "epoch": 0.1619818662991904,
      "grad_norm": 0.9272536058986685,
      "learning_rate": 9.550092596420073e-06,
      "loss": 0.504,
      "step": 2586
    },
    {
      "epoch": 0.1620445042985327,
      "grad_norm": 0.8247764956278907,
      "learning_rate": 9.549671965963734e-06,
      "loss": 0.4164,
      "step": 2587
    },
    {
      "epoch": 0.162107142297875,
      "grad_norm": 0.9091803752434323,
      "learning_rate": 9.549251148242051e-06,
      "loss": 0.4595,
      "step": 2588
    },
    {
      "epoch": 0.16216978029721732,
      "grad_norm": 0.8278794013319789,
      "learning_rate": 9.548830143272348e-06,
      "loss": 0.4109,
      "step": 2589
    },
    {
      "epoch": 0.1622324182965596,
      "grad_norm": 0.9085954953181858,
      "learning_rate": 9.548408951071953e-06,
      "loss": 0.4271,
      "step": 2590
    },
    {
      "epoch": 0.1622950562959019,
      "grad_norm": 0.8953640267383759,
      "learning_rate": 9.5479875716582e-06,
      "loss": 0.4448,
      "step": 2591
    },
    {
      "epoch": 0.16235769429524421,
      "grad_norm": 0.8471312579845293,
      "learning_rate": 9.547566005048433e-06,
      "loss": 0.4646,
      "step": 2592
    },
    {
      "epoch": 0.16242033229458652,
      "grad_norm": 1.0128670955935837,
      "learning_rate": 9.547144251260007e-06,
      "loss": 0.4663,
      "step": 2593
    },
    {
      "epoch": 0.1624829702939288,
      "grad_norm": 0.9052457765957664,
      "learning_rate": 9.54672231031028e-06,
      "loss": 0.4446,
      "step": 2594
    },
    {
      "epoch": 0.1625456082932711,
      "grad_norm": 0.9113257736885182,
      "learning_rate": 9.546300182216616e-06,
      "loss": 0.4452,
      "step": 2595
    },
    {
      "epoch": 0.16260824629261342,
      "grad_norm": 0.8564936462335121,
      "learning_rate": 9.545877866996396e-06,
      "loss": 0.4437,
      "step": 2596
    },
    {
      "epoch": 0.16267088429195573,
      "grad_norm": 0.9273631353503378,
      "learning_rate": 9.545455364666998e-06,
      "loss": 0.4315,
      "step": 2597
    },
    {
      "epoch": 0.162733522291298,
      "grad_norm": 0.8558732329447561,
      "learning_rate": 9.545032675245814e-06,
      "loss": 0.4476,
      "step": 2598
    },
    {
      "epoch": 0.1627961602906403,
      "grad_norm": 0.9069014959222792,
      "learning_rate": 9.54460979875024e-06,
      "loss": 0.4671,
      "step": 2599
    },
    {
      "epoch": 0.16285879828998262,
      "grad_norm": 0.832777761286989,
      "learning_rate": 9.544186735197686e-06,
      "loss": 0.4284,
      "step": 2600
    },
    {
      "epoch": 0.16292143628932493,
      "grad_norm": 0.9340500413851691,
      "learning_rate": 9.54376348460556e-06,
      "loss": 0.4597,
      "step": 2601
    },
    {
      "epoch": 0.1629840742886672,
      "grad_norm": 0.8483945898265481,
      "learning_rate": 9.543340046991286e-06,
      "loss": 0.449,
      "step": 2602
    },
    {
      "epoch": 0.16304671228800952,
      "grad_norm": 0.8406446581781889,
      "learning_rate": 9.542916422372293e-06,
      "loss": 0.4243,
      "step": 2603
    },
    {
      "epoch": 0.16310935028735182,
      "grad_norm": 0.9238174578296955,
      "learning_rate": 9.542492610766016e-06,
      "loss": 0.4623,
      "step": 2604
    },
    {
      "epoch": 0.16317198828669413,
      "grad_norm": 0.7945145737787388,
      "learning_rate": 9.5420686121899e-06,
      "loss": 0.4174,
      "step": 2605
    },
    {
      "epoch": 0.1632346262860364,
      "grad_norm": 0.8896004678867944,
      "learning_rate": 9.541644426661397e-06,
      "loss": 0.4264,
      "step": 2606
    },
    {
      "epoch": 0.16329726428537872,
      "grad_norm": 0.8898627299334262,
      "learning_rate": 9.541220054197966e-06,
      "loss": 0.423,
      "step": 2607
    },
    {
      "epoch": 0.16335990228472103,
      "grad_norm": 0.8763471913489947,
      "learning_rate": 9.540795494817076e-06,
      "loss": 0.4193,
      "step": 2608
    },
    {
      "epoch": 0.16342254028406333,
      "grad_norm": 0.8436600810945617,
      "learning_rate": 9.540370748536201e-06,
      "loss": 0.4205,
      "step": 2609
    },
    {
      "epoch": 0.16348517828340564,
      "grad_norm": 0.9350106878042955,
      "learning_rate": 9.539945815372822e-06,
      "loss": 0.4369,
      "step": 2610
    },
    {
      "epoch": 0.16354781628274792,
      "grad_norm": 0.8913550842322173,
      "learning_rate": 9.53952069534443e-06,
      "loss": 0.4521,
      "step": 2611
    },
    {
      "epoch": 0.16361045428209023,
      "grad_norm": 0.8629102139337645,
      "learning_rate": 9.539095388468525e-06,
      "loss": 0.4195,
      "step": 2612
    },
    {
      "epoch": 0.16367309228143254,
      "grad_norm": 0.7844790895617535,
      "learning_rate": 9.538669894762611e-06,
      "loss": 0.4785,
      "step": 2613
    },
    {
      "epoch": 0.16373573028077484,
      "grad_norm": 0.9244115302495114,
      "learning_rate": 9.5382442142442e-06,
      "loss": 0.4502,
      "step": 2614
    },
    {
      "epoch": 0.16379836828011712,
      "grad_norm": 0.8810955780893498,
      "learning_rate": 9.537818346930819e-06,
      "loss": 0.4661,
      "step": 2615
    },
    {
      "epoch": 0.16386100627945943,
      "grad_norm": 0.9217815303113598,
      "learning_rate": 9.537392292839989e-06,
      "loss": 0.4545,
      "step": 2616
    },
    {
      "epoch": 0.16392364427880174,
      "grad_norm": 0.9217952262472913,
      "learning_rate": 9.536966051989252e-06,
      "loss": 0.4404,
      "step": 2617
    },
    {
      "epoch": 0.16398628227814405,
      "grad_norm": 0.9715426676094361,
      "learning_rate": 9.536539624396149e-06,
      "loss": 0.4415,
      "step": 2618
    },
    {
      "epoch": 0.16404892027748633,
      "grad_norm": 0.94715564517307,
      "learning_rate": 9.536113010078232e-06,
      "loss": 0.4662,
      "step": 2619
    },
    {
      "epoch": 0.16411155827682863,
      "grad_norm": 0.8472923294929718,
      "learning_rate": 9.535686209053064e-06,
      "loss": 0.4293,
      "step": 2620
    },
    {
      "epoch": 0.16417419627617094,
      "grad_norm": 0.7983988285100745,
      "learning_rate": 9.535259221338206e-06,
      "loss": 0.4267,
      "step": 2621
    },
    {
      "epoch": 0.16423683427551325,
      "grad_norm": 0.8463324824907011,
      "learning_rate": 9.534832046951241e-06,
      "loss": 0.4417,
      "step": 2622
    },
    {
      "epoch": 0.16429947227485553,
      "grad_norm": 0.8748217691673758,
      "learning_rate": 9.534404685909744e-06,
      "loss": 0.4401,
      "step": 2623
    },
    {
      "epoch": 0.16436211027419784,
      "grad_norm": 0.8878847194572562,
      "learning_rate": 9.533977138231312e-06,
      "loss": 0.4396,
      "step": 2624
    },
    {
      "epoch": 0.16442474827354014,
      "grad_norm": 0.8168905382465736,
      "learning_rate": 9.533549403933538e-06,
      "loss": 0.4187,
      "step": 2625
    },
    {
      "epoch": 0.16448738627288245,
      "grad_norm": 0.9100798866096558,
      "learning_rate": 9.533121483034027e-06,
      "loss": 0.4907,
      "step": 2626
    },
    {
      "epoch": 0.16455002427222473,
      "grad_norm": 0.949333722626153,
      "learning_rate": 9.532693375550396e-06,
      "loss": 0.4053,
      "step": 2627
    },
    {
      "epoch": 0.16461266227156704,
      "grad_norm": 0.8532791262632705,
      "learning_rate": 9.532265081500264e-06,
      "loss": 0.4271,
      "step": 2628
    },
    {
      "epoch": 0.16467530027090935,
      "grad_norm": 0.8356120182006105,
      "learning_rate": 9.53183660090126e-06,
      "loss": 0.4581,
      "step": 2629
    },
    {
      "epoch": 0.16473793827025165,
      "grad_norm": 0.8085402242895724,
      "learning_rate": 9.53140793377102e-06,
      "loss": 0.431,
      "step": 2630
    },
    {
      "epoch": 0.16480057626959396,
      "grad_norm": 0.8556743000547639,
      "learning_rate": 9.530979080127189e-06,
      "loss": 0.4607,
      "step": 2631
    },
    {
      "epoch": 0.16486321426893624,
      "grad_norm": 0.8221051175245874,
      "learning_rate": 9.530550039987419e-06,
      "loss": 0.4075,
      "step": 2632
    },
    {
      "epoch": 0.16492585226827855,
      "grad_norm": 0.878864268450772,
      "learning_rate": 9.530120813369368e-06,
      "loss": 0.4407,
      "step": 2633
    },
    {
      "epoch": 0.16498849026762086,
      "grad_norm": 0.880353809754111,
      "learning_rate": 9.529691400290703e-06,
      "loss": 0.448,
      "step": 2634
    },
    {
      "epoch": 0.16505112826696317,
      "grad_norm": 0.8898731108843905,
      "learning_rate": 9.529261800769099e-06,
      "loss": 0.4413,
      "step": 2635
    },
    {
      "epoch": 0.16511376626630545,
      "grad_norm": 0.8350597809636049,
      "learning_rate": 9.528832014822238e-06,
      "loss": 0.4338,
      "step": 2636
    },
    {
      "epoch": 0.16517640426564775,
      "grad_norm": 0.9264286934249504,
      "learning_rate": 9.528402042467812e-06,
      "loss": 0.4787,
      "step": 2637
    },
    {
      "epoch": 0.16523904226499006,
      "grad_norm": 0.9142159447880784,
      "learning_rate": 9.527971883723518e-06,
      "loss": 0.4764,
      "step": 2638
    },
    {
      "epoch": 0.16530168026433237,
      "grad_norm": 0.7918535403814716,
      "learning_rate": 9.527541538607059e-06,
      "loss": 0.4087,
      "step": 2639
    },
    {
      "epoch": 0.16536431826367465,
      "grad_norm": 0.8617239167269862,
      "learning_rate": 9.52711100713615e-06,
      "loss": 0.4401,
      "step": 2640
    },
    {
      "epoch": 0.16542695626301696,
      "grad_norm": 0.9004091311975695,
      "learning_rate": 9.526680289328513e-06,
      "loss": 0.4709,
      "step": 2641
    },
    {
      "epoch": 0.16548959426235926,
      "grad_norm": 0.8504394300225159,
      "learning_rate": 9.526249385201873e-06,
      "loss": 0.4256,
      "step": 2642
    },
    {
      "epoch": 0.16555223226170157,
      "grad_norm": 0.9693514701140441,
      "learning_rate": 9.525818294773971e-06,
      "loss": 0.4475,
      "step": 2643
    },
    {
      "epoch": 0.16561487026104385,
      "grad_norm": 0.8409492227760244,
      "learning_rate": 9.525387018062547e-06,
      "loss": 0.4326,
      "step": 2644
    },
    {
      "epoch": 0.16567750826038616,
      "grad_norm": 0.9348518632782661,
      "learning_rate": 9.524955555085352e-06,
      "loss": 0.4497,
      "step": 2645
    },
    {
      "epoch": 0.16574014625972847,
      "grad_norm": 0.8815572919295553,
      "learning_rate": 9.524523905860147e-06,
      "loss": 0.4317,
      "step": 2646
    },
    {
      "epoch": 0.16580278425907077,
      "grad_norm": 0.926449605694904,
      "learning_rate": 9.5240920704047e-06,
      "loss": 0.4473,
      "step": 2647
    },
    {
      "epoch": 0.16586542225841305,
      "grad_norm": 0.8319380504598587,
      "learning_rate": 9.523660048736782e-06,
      "loss": 0.4298,
      "step": 2648
    },
    {
      "epoch": 0.16592806025775536,
      "grad_norm": 0.8248268655507827,
      "learning_rate": 9.523227840874177e-06,
      "loss": 0.4162,
      "step": 2649
    },
    {
      "epoch": 0.16599069825709767,
      "grad_norm": 0.8375105822311147,
      "learning_rate": 9.522795446834673e-06,
      "loss": 0.4442,
      "step": 2650
    },
    {
      "epoch": 0.16605333625643998,
      "grad_norm": 0.8101901315404193,
      "learning_rate": 9.522362866636073e-06,
      "loss": 0.4353,
      "step": 2651
    },
    {
      "epoch": 0.16611597425578228,
      "grad_norm": 0.7448247633079829,
      "learning_rate": 9.521930100296175e-06,
      "loss": 0.4198,
      "step": 2652
    },
    {
      "epoch": 0.16617861225512456,
      "grad_norm": 0.9079934609694781,
      "learning_rate": 9.521497147832794e-06,
      "loss": 0.4672,
      "step": 2653
    },
    {
      "epoch": 0.16624125025446687,
      "grad_norm": 0.8201563333136723,
      "learning_rate": 9.521064009263755e-06,
      "loss": 0.4243,
      "step": 2654
    },
    {
      "epoch": 0.16630388825380918,
      "grad_norm": 0.9101177370600885,
      "learning_rate": 9.520630684606877e-06,
      "loss": 0.4413,
      "step": 2655
    },
    {
      "epoch": 0.1663665262531515,
      "grad_norm": 0.8752368488643526,
      "learning_rate": 9.520197173880005e-06,
      "loss": 0.4775,
      "step": 2656
    },
    {
      "epoch": 0.16642916425249377,
      "grad_norm": 0.8923787384681517,
      "learning_rate": 9.519763477100976e-06,
      "loss": 0.4395,
      "step": 2657
    },
    {
      "epoch": 0.16649180225183607,
      "grad_norm": 0.915052711413404,
      "learning_rate": 9.519329594287645e-06,
      "loss": 0.463,
      "step": 2658
    },
    {
      "epoch": 0.16655444025117838,
      "grad_norm": 0.8501612124796961,
      "learning_rate": 9.518895525457869e-06,
      "loss": 0.5256,
      "step": 2659
    },
    {
      "epoch": 0.1666170782505207,
      "grad_norm": 0.7703010151970023,
      "learning_rate": 9.518461270629514e-06,
      "loss": 0.4225,
      "step": 2660
    },
    {
      "epoch": 0.16667971624986297,
      "grad_norm": 0.8146315471089793,
      "learning_rate": 9.518026829820454e-06,
      "loss": 0.4052,
      "step": 2661
    },
    {
      "epoch": 0.16674235424920528,
      "grad_norm": 0.8318083223507846,
      "learning_rate": 9.517592203048571e-06,
      "loss": 0.4081,
      "step": 2662
    },
    {
      "epoch": 0.16680499224854758,
      "grad_norm": 0.7587490568793648,
      "learning_rate": 9.517157390331756e-06,
      "loss": 0.4274,
      "step": 2663
    },
    {
      "epoch": 0.1668676302478899,
      "grad_norm": 0.8581095895237623,
      "learning_rate": 9.516722391687903e-06,
      "loss": 0.4273,
      "step": 2664
    },
    {
      "epoch": 0.16693026824723217,
      "grad_norm": 0.8705821254322631,
      "learning_rate": 9.516287207134918e-06,
      "loss": 0.4131,
      "step": 2665
    },
    {
      "epoch": 0.16699290624657448,
      "grad_norm": 0.9580113294793938,
      "learning_rate": 9.515851836690713e-06,
      "loss": 0.4966,
      "step": 2666
    },
    {
      "epoch": 0.1670555442459168,
      "grad_norm": 0.8113628762092572,
      "learning_rate": 9.515416280373209e-06,
      "loss": 0.4188,
      "step": 2667
    },
    {
      "epoch": 0.1671181822452591,
      "grad_norm": 0.7980668413514465,
      "learning_rate": 9.514980538200333e-06,
      "loss": 0.4154,
      "step": 2668
    },
    {
      "epoch": 0.16718082024460137,
      "grad_norm": 0.8152126449134496,
      "learning_rate": 9.51454461019002e-06,
      "loss": 0.4581,
      "step": 2669
    },
    {
      "epoch": 0.16724345824394368,
      "grad_norm": 0.8660753701274988,
      "learning_rate": 9.514108496360212e-06,
      "loss": 0.4442,
      "step": 2670
    },
    {
      "epoch": 0.167306096243286,
      "grad_norm": 0.8772952846645293,
      "learning_rate": 9.51367219672886e-06,
      "loss": 0.4644,
      "step": 2671
    },
    {
      "epoch": 0.1673687342426283,
      "grad_norm": 0.8730339863017681,
      "learning_rate": 9.513235711313923e-06,
      "loss": 0.4669,
      "step": 2672
    },
    {
      "epoch": 0.1674313722419706,
      "grad_norm": 0.8528428467283024,
      "learning_rate": 9.512799040133368e-06,
      "loss": 0.4409,
      "step": 2673
    },
    {
      "epoch": 0.16749401024131289,
      "grad_norm": 0.8536404598700664,
      "learning_rate": 9.512362183205166e-06,
      "loss": 0.4746,
      "step": 2674
    },
    {
      "epoch": 0.1675566482406552,
      "grad_norm": 0.861086522853339,
      "learning_rate": 9.511925140547299e-06,
      "loss": 0.4101,
      "step": 2675
    },
    {
      "epoch": 0.1676192862399975,
      "grad_norm": 0.8307207154292103,
      "learning_rate": 9.511487912177755e-06,
      "loss": 0.4161,
      "step": 2676
    },
    {
      "epoch": 0.1676819242393398,
      "grad_norm": 0.8297401662783211,
      "learning_rate": 9.511050498114532e-06,
      "loss": 0.4321,
      "step": 2677
    },
    {
      "epoch": 0.1677445622386821,
      "grad_norm": 0.869918407861152,
      "learning_rate": 9.510612898375634e-06,
      "loss": 0.4053,
      "step": 2678
    },
    {
      "epoch": 0.1678072002380244,
      "grad_norm": 0.9072667327835706,
      "learning_rate": 9.51017511297907e-06,
      "loss": 0.4396,
      "step": 2679
    },
    {
      "epoch": 0.1678698382373667,
      "grad_norm": 0.9539621929203498,
      "learning_rate": 9.509737141942863e-06,
      "loss": 0.4668,
      "step": 2680
    },
    {
      "epoch": 0.167932476236709,
      "grad_norm": 0.8496498280095651,
      "learning_rate": 9.509298985285037e-06,
      "loss": 0.4209,
      "step": 2681
    },
    {
      "epoch": 0.1679951142360513,
      "grad_norm": 0.8775222447178712,
      "learning_rate": 9.508860643023627e-06,
      "loss": 0.4248,
      "step": 2682
    },
    {
      "epoch": 0.1680577522353936,
      "grad_norm": 0.9159841707679621,
      "learning_rate": 9.50842211517668e-06,
      "loss": 0.3993,
      "step": 2683
    },
    {
      "epoch": 0.1681203902347359,
      "grad_norm": 0.8386196347112842,
      "learning_rate": 9.507983401762238e-06,
      "loss": 0.4002,
      "step": 2684
    },
    {
      "epoch": 0.1681830282340782,
      "grad_norm": 0.8466563462701431,
      "learning_rate": 9.507544502798363e-06,
      "loss": 0.4236,
      "step": 2685
    },
    {
      "epoch": 0.1682456662334205,
      "grad_norm": 0.9458750776593242,
      "learning_rate": 9.50710541830312e-06,
      "loss": 0.4014,
      "step": 2686
    },
    {
      "epoch": 0.1683083042327628,
      "grad_norm": 0.9449676815239555,
      "learning_rate": 9.506666148294582e-06,
      "loss": 0.4403,
      "step": 2687
    },
    {
      "epoch": 0.1683709422321051,
      "grad_norm": 0.8812754302280255,
      "learning_rate": 9.506226692790828e-06,
      "loss": 0.4784,
      "step": 2688
    },
    {
      "epoch": 0.16843358023144742,
      "grad_norm": 0.797327249113039,
      "learning_rate": 9.505787051809947e-06,
      "loss": 0.358,
      "step": 2689
    },
    {
      "epoch": 0.1684962182307897,
      "grad_norm": 0.8452235159112569,
      "learning_rate": 9.505347225370035e-06,
      "loss": 0.4018,
      "step": 2690
    },
    {
      "epoch": 0.168558856230132,
      "grad_norm": 0.8531581737489607,
      "learning_rate": 9.504907213489196e-06,
      "loss": 0.4723,
      "step": 2691
    },
    {
      "epoch": 0.1686214942294743,
      "grad_norm": 0.8329996344794001,
      "learning_rate": 9.50446701618554e-06,
      "loss": 0.4222,
      "step": 2692
    },
    {
      "epoch": 0.16868413222881662,
      "grad_norm": 0.8464989725103249,
      "learning_rate": 9.504026633477183e-06,
      "loss": 0.4399,
      "step": 2693
    },
    {
      "epoch": 0.1687467702281589,
      "grad_norm": 0.9063625773411863,
      "learning_rate": 9.503586065382255e-06,
      "loss": 0.4005,
      "step": 2694
    },
    {
      "epoch": 0.1688094082275012,
      "grad_norm": 0.9013622227876102,
      "learning_rate": 9.503145311918889e-06,
      "loss": 0.4342,
      "step": 2695
    },
    {
      "epoch": 0.16887204622684351,
      "grad_norm": 0.943849925110537,
      "learning_rate": 9.502704373105225e-06,
      "loss": 0.4753,
      "step": 2696
    },
    {
      "epoch": 0.16893468422618582,
      "grad_norm": 0.8987283216433049,
      "learning_rate": 9.502263248959417e-06,
      "loss": 0.4347,
      "step": 2697
    },
    {
      "epoch": 0.16899732222552813,
      "grad_norm": 0.8638821865809057,
      "learning_rate": 9.501821939499614e-06,
      "loss": 0.4277,
      "step": 2698
    },
    {
      "epoch": 0.1690599602248704,
      "grad_norm": 0.823803854902869,
      "learning_rate": 9.501380444743985e-06,
      "loss": 0.4105,
      "step": 2699
    },
    {
      "epoch": 0.16912259822421272,
      "grad_norm": 0.909702353811993,
      "learning_rate": 9.500938764710703e-06,
      "loss": 0.4519,
      "step": 2700
    },
    {
      "epoch": 0.16918523622355502,
      "grad_norm": 0.8804065773834213,
      "learning_rate": 9.500496899417944e-06,
      "loss": 0.4274,
      "step": 2701
    },
    {
      "epoch": 0.16924787422289733,
      "grad_norm": 0.8241543590731245,
      "learning_rate": 9.500054848883899e-06,
      "loss": 0.4164,
      "step": 2702
    },
    {
      "epoch": 0.1693105122222396,
      "grad_norm": 0.8126760789446797,
      "learning_rate": 9.499612613126761e-06,
      "loss": 0.5008,
      "step": 2703
    },
    {
      "epoch": 0.16937315022158192,
      "grad_norm": 0.8794467976569355,
      "learning_rate": 9.499170192164733e-06,
      "loss": 0.4393,
      "step": 2704
    },
    {
      "epoch": 0.16943578822092423,
      "grad_norm": 0.8812069353238116,
      "learning_rate": 9.498727586016023e-06,
      "loss": 0.4231,
      "step": 2705
    },
    {
      "epoch": 0.16949842622026653,
      "grad_norm": 0.9063087877813988,
      "learning_rate": 9.49828479469885e-06,
      "loss": 0.4776,
      "step": 2706
    },
    {
      "epoch": 0.16956106421960881,
      "grad_norm": 0.9483955436697055,
      "learning_rate": 9.497841818231443e-06,
      "loss": 0.482,
      "step": 2707
    },
    {
      "epoch": 0.16962370221895112,
      "grad_norm": 0.9781055189651422,
      "learning_rate": 9.49739865663203e-06,
      "loss": 0.4661,
      "step": 2708
    },
    {
      "epoch": 0.16968634021829343,
      "grad_norm": 0.8911000308079827,
      "learning_rate": 9.496955309918855e-06,
      "loss": 0.4456,
      "step": 2709
    },
    {
      "epoch": 0.16974897821763574,
      "grad_norm": 0.8666668137747247,
      "learning_rate": 9.496511778110164e-06,
      "loss": 0.3971,
      "step": 2710
    },
    {
      "epoch": 0.16981161621697802,
      "grad_norm": 0.849975326328914,
      "learning_rate": 9.496068061224214e-06,
      "loss": 0.3987,
      "step": 2711
    },
    {
      "epoch": 0.16987425421632033,
      "grad_norm": 0.9100904248557948,
      "learning_rate": 9.495624159279267e-06,
      "loss": 0.4896,
      "step": 2712
    },
    {
      "epoch": 0.16993689221566263,
      "grad_norm": 0.7778734698318733,
      "learning_rate": 9.495180072293595e-06,
      "loss": 0.3832,
      "step": 2713
    },
    {
      "epoch": 0.16999953021500494,
      "grad_norm": 0.9285297469187793,
      "learning_rate": 9.494735800285477e-06,
      "loss": 0.4686,
      "step": 2714
    },
    {
      "epoch": 0.17006216821434722,
      "grad_norm": 0.8726131270779746,
      "learning_rate": 9.4942913432732e-06,
      "loss": 0.4309,
      "step": 2715
    },
    {
      "epoch": 0.17012480621368953,
      "grad_norm": 0.8841817844661076,
      "learning_rate": 9.493846701275058e-06,
      "loss": 0.432,
      "step": 2716
    },
    {
      "epoch": 0.17018744421303184,
      "grad_norm": 0.8645684167309261,
      "learning_rate": 9.49340187430935e-06,
      "loss": 0.408,
      "step": 2717
    },
    {
      "epoch": 0.17025008221237414,
      "grad_norm": 0.8105416051029012,
      "learning_rate": 9.492956862394386e-06,
      "loss": 0.4905,
      "step": 2718
    },
    {
      "epoch": 0.17031272021171645,
      "grad_norm": 0.8654478671981315,
      "learning_rate": 9.492511665548487e-06,
      "loss": 0.4359,
      "step": 2719
    },
    {
      "epoch": 0.17037535821105873,
      "grad_norm": 0.8017937602920091,
      "learning_rate": 9.492066283789972e-06,
      "loss": 0.4054,
      "step": 2720
    },
    {
      "epoch": 0.17043799621040104,
      "grad_norm": 0.6993839182729844,
      "learning_rate": 9.491620717137176e-06,
      "loss": 0.4743,
      "step": 2721
    },
    {
      "epoch": 0.17050063420974335,
      "grad_norm": 0.9004845970824682,
      "learning_rate": 9.491174965608434e-06,
      "loss": 0.4779,
      "step": 2722
    },
    {
      "epoch": 0.17056327220908565,
      "grad_norm": 0.7967138422540898,
      "learning_rate": 9.490729029222101e-06,
      "loss": 0.3717,
      "step": 2723
    },
    {
      "epoch": 0.17062591020842793,
      "grad_norm": 0.8278599102603033,
      "learning_rate": 9.490282907996526e-06,
      "loss": 0.4477,
      "step": 2724
    },
    {
      "epoch": 0.17068854820777024,
      "grad_norm": 0.8503512400342429,
      "learning_rate": 9.489836601950073e-06,
      "loss": 0.4184,
      "step": 2725
    },
    {
      "epoch": 0.17075118620711255,
      "grad_norm": 0.8520475271558471,
      "learning_rate": 9.489390111101112e-06,
      "loss": 0.4305,
      "step": 2726
    },
    {
      "epoch": 0.17081382420645486,
      "grad_norm": 0.9185083585985825,
      "learning_rate": 9.488943435468021e-06,
      "loss": 0.4432,
      "step": 2727
    },
    {
      "epoch": 0.17087646220579714,
      "grad_norm": 1.0039507144890025,
      "learning_rate": 9.488496575069185e-06,
      "loss": 0.4315,
      "step": 2728
    },
    {
      "epoch": 0.17093910020513944,
      "grad_norm": 0.8745137918401572,
      "learning_rate": 9.488049529922997e-06,
      "loss": 0.4414,
      "step": 2729
    },
    {
      "epoch": 0.17100173820448175,
      "grad_norm": 0.942480677025271,
      "learning_rate": 9.487602300047858e-06,
      "loss": 0.4418,
      "step": 2730
    },
    {
      "epoch": 0.17106437620382406,
      "grad_norm": 0.8169426452277193,
      "learning_rate": 9.487154885462173e-06,
      "loss": 0.4172,
      "step": 2731
    },
    {
      "epoch": 0.17112701420316634,
      "grad_norm": 0.8454200600634748,
      "learning_rate": 9.486707286184361e-06,
      "loss": 0.4035,
      "step": 2732
    },
    {
      "epoch": 0.17118965220250865,
      "grad_norm": 0.7587290809751136,
      "learning_rate": 9.486259502232847e-06,
      "loss": 0.4335,
      "step": 2733
    },
    {
      "epoch": 0.17125229020185095,
      "grad_norm": 0.854226458644086,
      "learning_rate": 9.485811533626057e-06,
      "loss": 0.4357,
      "step": 2734
    },
    {
      "epoch": 0.17131492820119326,
      "grad_norm": 0.8425139435119016,
      "learning_rate": 9.485363380382432e-06,
      "loss": 0.4003,
      "step": 2735
    },
    {
      "epoch": 0.17137756620053554,
      "grad_norm": 0.8781491663441114,
      "learning_rate": 9.484915042520418e-06,
      "loss": 0.4351,
      "step": 2736
    },
    {
      "epoch": 0.17144020419987785,
      "grad_norm": 0.8331286137971406,
      "learning_rate": 9.484466520058467e-06,
      "loss": 0.3917,
      "step": 2737
    },
    {
      "epoch": 0.17150284219922016,
      "grad_norm": 0.9094144066104425,
      "learning_rate": 9.484017813015043e-06,
      "loss": 0.4217,
      "step": 2738
    },
    {
      "epoch": 0.17156548019856246,
      "grad_norm": 0.8973977739483414,
      "learning_rate": 9.483568921408614e-06,
      "loss": 0.4936,
      "step": 2739
    },
    {
      "epoch": 0.17162811819790477,
      "grad_norm": 0.8476288079547359,
      "learning_rate": 9.483119845257655e-06,
      "loss": 0.3856,
      "step": 2740
    },
    {
      "epoch": 0.17169075619724705,
      "grad_norm": 0.8337823245474411,
      "learning_rate": 9.48267058458065e-06,
      "loss": 0.4538,
      "step": 2741
    },
    {
      "epoch": 0.17175339419658936,
      "grad_norm": 0.9531192064031861,
      "learning_rate": 9.482221139396093e-06,
      "loss": 0.4412,
      "step": 2742
    },
    {
      "epoch": 0.17181603219593167,
      "grad_norm": 0.8336104062088713,
      "learning_rate": 9.481771509722483e-06,
      "loss": 0.4548,
      "step": 2743
    },
    {
      "epoch": 0.17187867019527397,
      "grad_norm": 0.8487844568902482,
      "learning_rate": 9.481321695578324e-06,
      "loss": 0.4277,
      "step": 2744
    },
    {
      "epoch": 0.17194130819461625,
      "grad_norm": 0.911879637973743,
      "learning_rate": 9.480871696982133e-06,
      "loss": 0.4371,
      "step": 2745
    },
    {
      "epoch": 0.17200394619395856,
      "grad_norm": 0.8445294927748577,
      "learning_rate": 9.480421513952432e-06,
      "loss": 0.4306,
      "step": 2746
    },
    {
      "epoch": 0.17206658419330087,
      "grad_norm": 0.8269585946964968,
      "learning_rate": 9.47997114650775e-06,
      "loss": 0.4169,
      "step": 2747
    },
    {
      "epoch": 0.17212922219264318,
      "grad_norm": 0.8925985873628597,
      "learning_rate": 9.479520594666622e-06,
      "loss": 0.4355,
      "step": 2748
    },
    {
      "epoch": 0.17219186019198546,
      "grad_norm": 0.8741637289002904,
      "learning_rate": 9.479069858447596e-06,
      "loss": 0.4138,
      "step": 2749
    },
    {
      "epoch": 0.17225449819132777,
      "grad_norm": 0.9312310454691328,
      "learning_rate": 9.478618937869223e-06,
      "loss": 0.5024,
      "step": 2750
    },
    {
      "epoch": 0.17231713619067007,
      "grad_norm": 0.847729700839907,
      "learning_rate": 9.478167832950066e-06,
      "loss": 0.4512,
      "step": 2751
    },
    {
      "epoch": 0.17237977419001238,
      "grad_norm": 0.9050249952715486,
      "learning_rate": 9.477716543708687e-06,
      "loss": 0.431,
      "step": 2752
    },
    {
      "epoch": 0.17244241218935466,
      "grad_norm": 0.859880386048491,
      "learning_rate": 9.477265070163663e-06,
      "loss": 0.457,
      "step": 2753
    },
    {
      "epoch": 0.17250505018869697,
      "grad_norm": 0.9960221039365595,
      "learning_rate": 9.47681341233358e-06,
      "loss": 0.5147,
      "step": 2754
    },
    {
      "epoch": 0.17256768818803928,
      "grad_norm": 0.8452680497858814,
      "learning_rate": 9.476361570237025e-06,
      "loss": 0.3847,
      "step": 2755
    },
    {
      "epoch": 0.17263032618738158,
      "grad_norm": 0.8968075224592105,
      "learning_rate": 9.475909543892596e-06,
      "loss": 0.4517,
      "step": 2756
    },
    {
      "epoch": 0.17269296418672386,
      "grad_norm": 0.8870725030714036,
      "learning_rate": 9.4754573333189e-06,
      "loss": 0.4688,
      "step": 2757
    },
    {
      "epoch": 0.17275560218606617,
      "grad_norm": 0.8094623476099095,
      "learning_rate": 9.475004938534551e-06,
      "loss": 0.4497,
      "step": 2758
    },
    {
      "epoch": 0.17281824018540848,
      "grad_norm": 0.8628631723628966,
      "learning_rate": 9.474552359558167e-06,
      "loss": 0.4352,
      "step": 2759
    },
    {
      "epoch": 0.17288087818475079,
      "grad_norm": 0.9329117817565882,
      "learning_rate": 9.474099596408377e-06,
      "loss": 0.4253,
      "step": 2760
    },
    {
      "epoch": 0.1729435161840931,
      "grad_norm": 0.9187457255463044,
      "learning_rate": 9.473646649103819e-06,
      "loss": 0.4617,
      "step": 2761
    },
    {
      "epoch": 0.17300615418343537,
      "grad_norm": 0.8009654569402445,
      "learning_rate": 9.473193517663132e-06,
      "loss": 0.368,
      "step": 2762
    },
    {
      "epoch": 0.17306879218277768,
      "grad_norm": 0.8853783868513775,
      "learning_rate": 9.472740202104971e-06,
      "loss": 0.4421,
      "step": 2763
    },
    {
      "epoch": 0.17313143018212,
      "grad_norm": 0.793431769538641,
      "learning_rate": 9.472286702447993e-06,
      "loss": 0.4091,
      "step": 2764
    },
    {
      "epoch": 0.1731940681814623,
      "grad_norm": 0.823293861170222,
      "learning_rate": 9.471833018710865e-06,
      "loss": 0.3717,
      "step": 2765
    },
    {
      "epoch": 0.17325670618080458,
      "grad_norm": 0.8377959642619103,
      "learning_rate": 9.471379150912258e-06,
      "loss": 0.4236,
      "step": 2766
    },
    {
      "epoch": 0.17331934418014688,
      "grad_norm": 0.8982659090698357,
      "learning_rate": 9.470925099070855e-06,
      "loss": 0.4762,
      "step": 2767
    },
    {
      "epoch": 0.1733819821794892,
      "grad_norm": 0.8310244779299916,
      "learning_rate": 9.470470863205347e-06,
      "loss": 0.4017,
      "step": 2768
    },
    {
      "epoch": 0.1734446201788315,
      "grad_norm": 0.8057244123377411,
      "learning_rate": 9.470016443334429e-06,
      "loss": 0.4205,
      "step": 2769
    },
    {
      "epoch": 0.17350725817817378,
      "grad_norm": 0.8551586268753306,
      "learning_rate": 9.469561839476803e-06,
      "loss": 0.4545,
      "step": 2770
    },
    {
      "epoch": 0.1735698961775161,
      "grad_norm": 0.8641409538149325,
      "learning_rate": 9.469107051651181e-06,
      "loss": 0.3876,
      "step": 2771
    },
    {
      "epoch": 0.1736325341768584,
      "grad_norm": 0.789457684916307,
      "learning_rate": 9.468652079876286e-06,
      "loss": 0.4547,
      "step": 2772
    },
    {
      "epoch": 0.1736951721762007,
      "grad_norm": 0.8886053300130071,
      "learning_rate": 9.468196924170841e-06,
      "loss": 0.4329,
      "step": 2773
    },
    {
      "epoch": 0.17375781017554298,
      "grad_norm": 0.8955429908153815,
      "learning_rate": 9.467741584553581e-06,
      "loss": 0.4553,
      "step": 2774
    },
    {
      "epoch": 0.1738204481748853,
      "grad_norm": 0.862489380304807,
      "learning_rate": 9.467286061043247e-06,
      "loss": 0.3831,
      "step": 2775
    },
    {
      "epoch": 0.1738830861742276,
      "grad_norm": 0.8418746714757002,
      "learning_rate": 9.466830353658591e-06,
      "loss": 0.5016,
      "step": 2776
    },
    {
      "epoch": 0.1739457241735699,
      "grad_norm": 0.8954949212290697,
      "learning_rate": 9.466374462418368e-06,
      "loss": 0.4669,
      "step": 2777
    },
    {
      "epoch": 0.17400836217291218,
      "grad_norm": 0.8080812341076752,
      "learning_rate": 9.465918387341344e-06,
      "loss": 0.4205,
      "step": 2778
    },
    {
      "epoch": 0.1740710001722545,
      "grad_norm": 0.884648161494236,
      "learning_rate": 9.465462128446289e-06,
      "loss": 0.487,
      "step": 2779
    },
    {
      "epoch": 0.1741336381715968,
      "grad_norm": 0.7887060982060315,
      "learning_rate": 9.465005685751984e-06,
      "loss": 0.4307,
      "step": 2780
    },
    {
      "epoch": 0.1741962761709391,
      "grad_norm": 1.1289092877607154,
      "learning_rate": 9.464549059277217e-06,
      "loss": 0.4513,
      "step": 2781
    },
    {
      "epoch": 0.17425891417028141,
      "grad_norm": 0.8192169774867829,
      "learning_rate": 9.464092249040781e-06,
      "loss": 0.451,
      "step": 2782
    },
    {
      "epoch": 0.1743215521696237,
      "grad_norm": 0.8729983913783754,
      "learning_rate": 9.463635255061481e-06,
      "loss": 0.3735,
      "step": 2783
    },
    {
      "epoch": 0.174384190168966,
      "grad_norm": 0.8995998442579233,
      "learning_rate": 9.463178077358126e-06,
      "loss": 0.4022,
      "step": 2784
    },
    {
      "epoch": 0.1744468281683083,
      "grad_norm": 0.888759431566588,
      "learning_rate": 9.46272071594953e-06,
      "loss": 0.4482,
      "step": 2785
    },
    {
      "epoch": 0.17450946616765062,
      "grad_norm": 0.7934948779256831,
      "learning_rate": 9.462263170854522e-06,
      "loss": 0.4811,
      "step": 2786
    },
    {
      "epoch": 0.1745721041669929,
      "grad_norm": 0.8790555270031397,
      "learning_rate": 9.461805442091935e-06,
      "loss": 0.4376,
      "step": 2787
    },
    {
      "epoch": 0.1746347421663352,
      "grad_norm": 0.8616073419403925,
      "learning_rate": 9.461347529680607e-06,
      "loss": 0.4664,
      "step": 2788
    },
    {
      "epoch": 0.1746973801656775,
      "grad_norm": 0.8974259888646205,
      "learning_rate": 9.460889433639386e-06,
      "loss": 0.4454,
      "step": 2789
    },
    {
      "epoch": 0.17476001816501982,
      "grad_norm": 0.8223820250563709,
      "learning_rate": 9.46043115398713e-06,
      "loss": 0.458,
      "step": 2790
    },
    {
      "epoch": 0.1748226561643621,
      "grad_norm": 0.8259213603597525,
      "learning_rate": 9.459972690742698e-06,
      "loss": 0.4133,
      "step": 2791
    },
    {
      "epoch": 0.1748852941637044,
      "grad_norm": 0.8420746608188033,
      "learning_rate": 9.459514043924963e-06,
      "loss": 0.4729,
      "step": 2792
    },
    {
      "epoch": 0.17494793216304672,
      "grad_norm": 0.8587065882725121,
      "learning_rate": 9.4590552135528e-06,
      "loss": 0.4445,
      "step": 2793
    },
    {
      "epoch": 0.17501057016238902,
      "grad_norm": 0.8709269432032266,
      "learning_rate": 9.4585961996451e-06,
      "loss": 0.4236,
      "step": 2794
    },
    {
      "epoch": 0.1750732081617313,
      "grad_norm": 0.9131214558216991,
      "learning_rate": 9.458137002220752e-06,
      "loss": 0.4556,
      "step": 2795
    },
    {
      "epoch": 0.1751358461610736,
      "grad_norm": 0.826543672410774,
      "learning_rate": 9.457677621298658e-06,
      "loss": 0.4584,
      "step": 2796
    },
    {
      "epoch": 0.17519848416041592,
      "grad_norm": 0.8375402310535915,
      "learning_rate": 9.457218056897725e-06,
      "loss": 0.4153,
      "step": 2797
    },
    {
      "epoch": 0.17526112215975823,
      "grad_norm": 0.905849911479791,
      "learning_rate": 9.456758309036868e-06,
      "loss": 0.4474,
      "step": 2798
    },
    {
      "epoch": 0.1753237601591005,
      "grad_norm": 0.8860930809865608,
      "learning_rate": 9.456298377735014e-06,
      "loss": 0.4638,
      "step": 2799
    },
    {
      "epoch": 0.1753863981584428,
      "grad_norm": 0.896778413302155,
      "learning_rate": 9.45583826301109e-06,
      "loss": 0.4266,
      "step": 2800
    },
    {
      "epoch": 0.17544903615778512,
      "grad_norm": 0.8123398971167703,
      "learning_rate": 9.455377964884038e-06,
      "loss": 0.4434,
      "step": 2801
    },
    {
      "epoch": 0.17551167415712743,
      "grad_norm": 0.8512884033411453,
      "learning_rate": 9.4549174833728e-06,
      "loss": 0.4677,
      "step": 2802
    },
    {
      "epoch": 0.17557431215646974,
      "grad_norm": 0.9270304309028553,
      "learning_rate": 9.454456818496331e-06,
      "loss": 0.4417,
      "step": 2803
    },
    {
      "epoch": 0.17563695015581202,
      "grad_norm": 0.9387575558591479,
      "learning_rate": 9.453995970273596e-06,
      "loss": 0.407,
      "step": 2804
    },
    {
      "epoch": 0.17569958815515432,
      "grad_norm": 0.8445830601206818,
      "learning_rate": 9.453534938723557e-06,
      "loss": 0.4696,
      "step": 2805
    },
    {
      "epoch": 0.17576222615449663,
      "grad_norm": 0.8711887470237691,
      "learning_rate": 9.453073723865194e-06,
      "loss": 0.399,
      "step": 2806
    },
    {
      "epoch": 0.17582486415383894,
      "grad_norm": 0.825965198227466,
      "learning_rate": 9.45261232571749e-06,
      "loss": 0.4477,
      "step": 2807
    },
    {
      "epoch": 0.17588750215318122,
      "grad_norm": 0.8331262519124429,
      "learning_rate": 9.452150744299437e-06,
      "loss": 0.4349,
      "step": 2808
    },
    {
      "epoch": 0.17595014015252353,
      "grad_norm": 0.7787306675730276,
      "learning_rate": 9.451688979630035e-06,
      "loss": 0.4063,
      "step": 2809
    },
    {
      "epoch": 0.17601277815186583,
      "grad_norm": 0.9135230522854533,
      "learning_rate": 9.451227031728285e-06,
      "loss": 0.4202,
      "step": 2810
    },
    {
      "epoch": 0.17607541615120814,
      "grad_norm": 0.8437755878750033,
      "learning_rate": 9.450764900613205e-06,
      "loss": 0.3796,
      "step": 2811
    },
    {
      "epoch": 0.17613805415055042,
      "grad_norm": 0.8057037271838987,
      "learning_rate": 9.450302586303814e-06,
      "loss": 0.49,
      "step": 2812
    },
    {
      "epoch": 0.17620069214989273,
      "grad_norm": 0.828301859404482,
      "learning_rate": 9.449840088819145e-06,
      "loss": 0.4062,
      "step": 2813
    },
    {
      "epoch": 0.17626333014923504,
      "grad_norm": 0.850083355037954,
      "learning_rate": 9.44937740817823e-06,
      "loss": 0.4217,
      "step": 2814
    },
    {
      "epoch": 0.17632596814857734,
      "grad_norm": 0.843891222765461,
      "learning_rate": 9.448914544400117e-06,
      "loss": 0.4028,
      "step": 2815
    },
    {
      "epoch": 0.17638860614791962,
      "grad_norm": 0.8672263190467742,
      "learning_rate": 9.448451497503855e-06,
      "loss": 0.4394,
      "step": 2816
    },
    {
      "epoch": 0.17645124414726193,
      "grad_norm": 0.8158743153755569,
      "learning_rate": 9.447988267508503e-06,
      "loss": 0.4346,
      "step": 2817
    },
    {
      "epoch": 0.17651388214660424,
      "grad_norm": 0.9168055384326477,
      "learning_rate": 9.44752485443313e-06,
      "loss": 0.4264,
      "step": 2818
    },
    {
      "epoch": 0.17657652014594655,
      "grad_norm": 0.8956509399189175,
      "learning_rate": 9.447061258296807e-06,
      "loss": 0.4107,
      "step": 2819
    },
    {
      "epoch": 0.17663915814528883,
      "grad_norm": 0.792628270539246,
      "learning_rate": 9.446597479118617e-06,
      "loss": 0.4133,
      "step": 2820
    },
    {
      "epoch": 0.17670179614463113,
      "grad_norm": 0.8286522723441212,
      "learning_rate": 9.446133516917651e-06,
      "loss": 0.4371,
      "step": 2821
    },
    {
      "epoch": 0.17676443414397344,
      "grad_norm": 0.8683794813476353,
      "learning_rate": 9.445669371713003e-06,
      "loss": 0.4481,
      "step": 2822
    },
    {
      "epoch": 0.17682707214331575,
      "grad_norm": 0.9925243733252437,
      "learning_rate": 9.44520504352378e-06,
      "loss": 0.4363,
      "step": 2823
    },
    {
      "epoch": 0.17688971014265803,
      "grad_norm": 0.782966325279666,
      "learning_rate": 9.44474053236909e-06,
      "loss": 0.4946,
      "step": 2824
    },
    {
      "epoch": 0.17695234814200034,
      "grad_norm": 0.8242972771198651,
      "learning_rate": 9.444275838268056e-06,
      "loss": 0.3919,
      "step": 2825
    },
    {
      "epoch": 0.17701498614134265,
      "grad_norm": 0.8619044867299531,
      "learning_rate": 9.443810961239804e-06,
      "loss": 0.4303,
      "step": 2826
    },
    {
      "epoch": 0.17707762414068495,
      "grad_norm": 0.7616622044804653,
      "learning_rate": 9.443345901303466e-06,
      "loss": 0.4933,
      "step": 2827
    },
    {
      "epoch": 0.17714026214002726,
      "grad_norm": 0.948440716515132,
      "learning_rate": 9.442880658478188e-06,
      "loss": 0.4801,
      "step": 2828
    },
    {
      "epoch": 0.17720290013936954,
      "grad_norm": 0.8486653291847763,
      "learning_rate": 9.442415232783115e-06,
      "loss": 0.4442,
      "step": 2829
    },
    {
      "epoch": 0.17726553813871185,
      "grad_norm": 0.8411110166353309,
      "learning_rate": 9.44194962423741e-06,
      "loss": 0.4065,
      "step": 2830
    },
    {
      "epoch": 0.17732817613805416,
      "grad_norm": 0.8628570874826459,
      "learning_rate": 9.441483832860232e-06,
      "loss": 0.4446,
      "step": 2831
    },
    {
      "epoch": 0.17739081413739646,
      "grad_norm": 0.857989848281793,
      "learning_rate": 9.441017858670756e-06,
      "loss": 0.4405,
      "step": 2832
    },
    {
      "epoch": 0.17745345213673874,
      "grad_norm": 0.7579267365614566,
      "learning_rate": 9.440551701688158e-06,
      "loss": 0.4634,
      "step": 2833
    },
    {
      "epoch": 0.17751609013608105,
      "grad_norm": 0.822222747566777,
      "learning_rate": 9.44008536193163e-06,
      "loss": 0.4357,
      "step": 2834
    },
    {
      "epoch": 0.17757872813542336,
      "grad_norm": 0.9098380473054793,
      "learning_rate": 9.439618839420365e-06,
      "loss": 0.4403,
      "step": 2835
    },
    {
      "epoch": 0.17764136613476567,
      "grad_norm": 0.852518820025479,
      "learning_rate": 9.439152134173564e-06,
      "loss": 0.4137,
      "step": 2836
    },
    {
      "epoch": 0.17770400413410795,
      "grad_norm": 0.9247550472097025,
      "learning_rate": 9.438685246210437e-06,
      "loss": 0.4324,
      "step": 2837
    },
    {
      "epoch": 0.17776664213345025,
      "grad_norm": 0.8987367142105294,
      "learning_rate": 9.4382181755502e-06,
      "loss": 0.4918,
      "step": 2838
    },
    {
      "epoch": 0.17782928013279256,
      "grad_norm": 0.8070093594699421,
      "learning_rate": 9.437750922212081e-06,
      "loss": 0.4008,
      "step": 2839
    },
    {
      "epoch": 0.17789191813213487,
      "grad_norm": 0.9239780016626321,
      "learning_rate": 9.43728348621531e-06,
      "loss": 0.5182,
      "step": 2840
    },
    {
      "epoch": 0.17795455613147715,
      "grad_norm": 0.7977501741351619,
      "learning_rate": 9.436815867579126e-06,
      "loss": 0.4097,
      "step": 2841
    },
    {
      "epoch": 0.17801719413081946,
      "grad_norm": 0.7947489602946767,
      "learning_rate": 9.436348066322777e-06,
      "loss": 0.4292,
      "step": 2842
    },
    {
      "epoch": 0.17807983213016176,
      "grad_norm": 0.862281816504563,
      "learning_rate": 9.435880082465521e-06,
      "loss": 0.404,
      "step": 2843
    },
    {
      "epoch": 0.17814247012950407,
      "grad_norm": 0.911535096273376,
      "learning_rate": 9.435411916026616e-06,
      "loss": 0.4318,
      "step": 2844
    },
    {
      "epoch": 0.17820510812884635,
      "grad_norm": 0.8708106213340135,
      "learning_rate": 9.434943567025333e-06,
      "loss": 0.4174,
      "step": 2845
    },
    {
      "epoch": 0.17826774612818866,
      "grad_norm": 0.798989367874429,
      "learning_rate": 9.43447503548095e-06,
      "loss": 0.4405,
      "step": 2846
    },
    {
      "epoch": 0.17833038412753097,
      "grad_norm": 0.8357742468813493,
      "learning_rate": 9.43400632141275e-06,
      "loss": 0.4376,
      "step": 2847
    },
    {
      "epoch": 0.17839302212687327,
      "grad_norm": 0.8629926561248715,
      "learning_rate": 9.433537424840029e-06,
      "loss": 0.4464,
      "step": 2848
    },
    {
      "epoch": 0.17845566012621558,
      "grad_norm": 0.8038463870496617,
      "learning_rate": 9.433068345782084e-06,
      "loss": 0.4067,
      "step": 2849
    },
    {
      "epoch": 0.17851829812555786,
      "grad_norm": 0.8049265263579096,
      "learning_rate": 9.432599084258222e-06,
      "loss": 0.4333,
      "step": 2850
    },
    {
      "epoch": 0.17858093612490017,
      "grad_norm": 0.8883625454407474,
      "learning_rate": 9.43212964028776e-06,
      "loss": 0.4201,
      "step": 2851
    },
    {
      "epoch": 0.17864357412424248,
      "grad_norm": 0.8607998775153426,
      "learning_rate": 9.431660013890021e-06,
      "loss": 0.4312,
      "step": 2852
    },
    {
      "epoch": 0.17870621212358478,
      "grad_norm": 0.761797237350617,
      "learning_rate": 9.43119020508433e-06,
      "loss": 0.4701,
      "step": 2853
    },
    {
      "epoch": 0.17876885012292706,
      "grad_norm": 0.8411655619516173,
      "learning_rate": 9.43072021389003e-06,
      "loss": 0.5054,
      "step": 2854
    },
    {
      "epoch": 0.17883148812226937,
      "grad_norm": 0.8413482757433943,
      "learning_rate": 9.430250040326461e-06,
      "loss": 0.3894,
      "step": 2855
    },
    {
      "epoch": 0.17889412612161168,
      "grad_norm": 0.8531479443896964,
      "learning_rate": 9.42977968441298e-06,
      "loss": 0.426,
      "step": 2856
    },
    {
      "epoch": 0.178956764120954,
      "grad_norm": 0.8449469254188176,
      "learning_rate": 9.429309146168946e-06,
      "loss": 0.4044,
      "step": 2857
    },
    {
      "epoch": 0.17901940212029627,
      "grad_norm": 0.836893767284663,
      "learning_rate": 9.428838425613725e-06,
      "loss": 0.4067,
      "step": 2858
    },
    {
      "epoch": 0.17908204011963857,
      "grad_norm": 0.8868661967464119,
      "learning_rate": 9.428367522766691e-06,
      "loss": 0.4258,
      "step": 2859
    },
    {
      "epoch": 0.17914467811898088,
      "grad_norm": 0.895977773883451,
      "learning_rate": 9.427896437647229e-06,
      "loss": 0.4481,
      "step": 2860
    },
    {
      "epoch": 0.1792073161183232,
      "grad_norm": 0.7936359204716122,
      "learning_rate": 9.427425170274726e-06,
      "loss": 0.3662,
      "step": 2861
    },
    {
      "epoch": 0.17926995411766547,
      "grad_norm": 0.8752654475941211,
      "learning_rate": 9.426953720668582e-06,
      "loss": 0.4411,
      "step": 2862
    },
    {
      "epoch": 0.17933259211700778,
      "grad_norm": 0.8584262426256167,
      "learning_rate": 9.4264820888482e-06,
      "loss": 0.4254,
      "step": 2863
    },
    {
      "epoch": 0.17939523011635009,
      "grad_norm": 0.8254954394337519,
      "learning_rate": 9.426010274832997e-06,
      "loss": 0.4025,
      "step": 2864
    },
    {
      "epoch": 0.1794578681156924,
      "grad_norm": 0.8328592322490375,
      "learning_rate": 9.425538278642388e-06,
      "loss": 0.4184,
      "step": 2865
    },
    {
      "epoch": 0.17952050611503467,
      "grad_norm": 0.8678064483885254,
      "learning_rate": 9.425066100295802e-06,
      "loss": 0.444,
      "step": 2866
    },
    {
      "epoch": 0.17958314411437698,
      "grad_norm": 0.977970753065322,
      "learning_rate": 9.424593739812673e-06,
      "loss": 0.418,
      "step": 2867
    },
    {
      "epoch": 0.1796457821137193,
      "grad_norm": 0.8013993485080579,
      "learning_rate": 9.424121197212446e-06,
      "loss": 0.3942,
      "step": 2868
    },
    {
      "epoch": 0.1797084201130616,
      "grad_norm": 0.8778343748958704,
      "learning_rate": 9.423648472514569e-06,
      "loss": 0.4257,
      "step": 2869
    },
    {
      "epoch": 0.1797710581124039,
      "grad_norm": 0.8394865740162272,
      "learning_rate": 9.423175565738498e-06,
      "loss": 0.4718,
      "step": 2870
    },
    {
      "epoch": 0.17983369611174618,
      "grad_norm": 0.8269695835160127,
      "learning_rate": 9.4227024769037e-06,
      "loss": 0.4401,
      "step": 2871
    },
    {
      "epoch": 0.1798963341110885,
      "grad_norm": 0.858561275214598,
      "learning_rate": 9.42222920602965e-06,
      "loss": 0.4224,
      "step": 2872
    },
    {
      "epoch": 0.1799589721104308,
      "grad_norm": 0.8660332551503425,
      "learning_rate": 9.421755753135823e-06,
      "loss": 0.4525,
      "step": 2873
    },
    {
      "epoch": 0.1800216101097731,
      "grad_norm": 0.8648491452766954,
      "learning_rate": 9.421282118241708e-06,
      "loss": 0.4422,
      "step": 2874
    },
    {
      "epoch": 0.18008424810911539,
      "grad_norm": 0.787454484238567,
      "learning_rate": 9.420808301366801e-06,
      "loss": 0.4787,
      "step": 2875
    },
    {
      "epoch": 0.1801468861084577,
      "grad_norm": 0.8967174709994419,
      "learning_rate": 9.420334302530606e-06,
      "loss": 0.4315,
      "step": 2876
    },
    {
      "epoch": 0.1802095241078,
      "grad_norm": 0.8651576552281631,
      "learning_rate": 9.419860121752628e-06,
      "loss": 0.4574,
      "step": 2877
    },
    {
      "epoch": 0.1802721621071423,
      "grad_norm": 0.8037049688445972,
      "learning_rate": 9.419385759052388e-06,
      "loss": 0.3864,
      "step": 2878
    },
    {
      "epoch": 0.1803348001064846,
      "grad_norm": 0.8328448744908445,
      "learning_rate": 9.418911214449409e-06,
      "loss": 0.417,
      "step": 2879
    },
    {
      "epoch": 0.1803974381058269,
      "grad_norm": 0.7397566097714897,
      "learning_rate": 9.418436487963226e-06,
      "loss": 0.4712,
      "step": 2880
    },
    {
      "epoch": 0.1804600761051692,
      "grad_norm": 0.8595115761754731,
      "learning_rate": 9.417961579613374e-06,
      "loss": 0.414,
      "step": 2881
    },
    {
      "epoch": 0.1805227141045115,
      "grad_norm": 0.8452072564129594,
      "learning_rate": 9.417486489419405e-06,
      "loss": 0.384,
      "step": 2882
    },
    {
      "epoch": 0.1805853521038538,
      "grad_norm": 0.8976933965269407,
      "learning_rate": 9.417011217400874e-06,
      "loss": 0.4733,
      "step": 2883
    },
    {
      "epoch": 0.1806479901031961,
      "grad_norm": 0.8039587613600406,
      "learning_rate": 9.416535763577338e-06,
      "loss": 0.469,
      "step": 2884
    },
    {
      "epoch": 0.1807106281025384,
      "grad_norm": 0.8352348530938727,
      "learning_rate": 9.416060127968372e-06,
      "loss": 0.4281,
      "step": 2885
    },
    {
      "epoch": 0.18077326610188071,
      "grad_norm": 0.8250796357720096,
      "learning_rate": 9.415584310593553e-06,
      "loss": 0.4059,
      "step": 2886
    },
    {
      "epoch": 0.180835904101223,
      "grad_norm": 0.8439640688781531,
      "learning_rate": 9.415108311472462e-06,
      "loss": 0.4615,
      "step": 2887
    },
    {
      "epoch": 0.1808985421005653,
      "grad_norm": 0.8554557800597767,
      "learning_rate": 9.414632130624695e-06,
      "loss": 0.4245,
      "step": 2888
    },
    {
      "epoch": 0.1809611800999076,
      "grad_norm": 0.7912134757309447,
      "learning_rate": 9.414155768069848e-06,
      "loss": 0.4134,
      "step": 2889
    },
    {
      "epoch": 0.18102381809924992,
      "grad_norm": 0.87151258581928,
      "learning_rate": 9.413679223827532e-06,
      "loss": 0.4144,
      "step": 2890
    },
    {
      "epoch": 0.18108645609859222,
      "grad_norm": 0.8511841827892873,
      "learning_rate": 9.41320249791736e-06,
      "loss": 0.445,
      "step": 2891
    },
    {
      "epoch": 0.1811490940979345,
      "grad_norm": 0.7991164386137517,
      "learning_rate": 9.412725590358953e-06,
      "loss": 0.4356,
      "step": 2892
    },
    {
      "epoch": 0.1812117320972768,
      "grad_norm": 0.8428514090080735,
      "learning_rate": 9.412248501171942e-06,
      "loss": 0.4138,
      "step": 2893
    },
    {
      "epoch": 0.18127437009661912,
      "grad_norm": 0.7554550063865975,
      "learning_rate": 9.411771230375963e-06,
      "loss": 0.3926,
      "step": 2894
    },
    {
      "epoch": 0.18133700809596143,
      "grad_norm": 0.8073982552754833,
      "learning_rate": 9.411293777990662e-06,
      "loss": 0.4169,
      "step": 2895
    },
    {
      "epoch": 0.1813996460953037,
      "grad_norm": 0.8900754904224207,
      "learning_rate": 9.410816144035691e-06,
      "loss": 0.4382,
      "step": 2896
    },
    {
      "epoch": 0.18146228409464601,
      "grad_norm": 0.7997583427126023,
      "learning_rate": 9.410338328530708e-06,
      "loss": 0.4096,
      "step": 2897
    },
    {
      "epoch": 0.18152492209398832,
      "grad_norm": 0.9072504660088666,
      "learning_rate": 9.409860331495382e-06,
      "loss": 0.4676,
      "step": 2898
    },
    {
      "epoch": 0.18158756009333063,
      "grad_norm": 0.8336212490615138,
      "learning_rate": 9.409382152949384e-06,
      "loss": 0.4259,
      "step": 2899
    },
    {
      "epoch": 0.1816501980926729,
      "grad_norm": 0.8768546934258616,
      "learning_rate": 9.408903792912401e-06,
      "loss": 0.4495,
      "step": 2900
    },
    {
      "epoch": 0.18171283609201522,
      "grad_norm": 0.9177380048201401,
      "learning_rate": 9.408425251404118e-06,
      "loss": 0.4647,
      "step": 2901
    },
    {
      "epoch": 0.18177547409135753,
      "grad_norm": 0.8182244865170593,
      "learning_rate": 9.407946528444234e-06,
      "loss": 0.4078,
      "step": 2902
    },
    {
      "epoch": 0.18183811209069983,
      "grad_norm": 0.7827699658159359,
      "learning_rate": 9.407467624052451e-06,
      "loss": 0.487,
      "step": 2903
    },
    {
      "epoch": 0.1819007500900421,
      "grad_norm": 0.8956690249530793,
      "learning_rate": 9.406988538248484e-06,
      "loss": 0.4162,
      "step": 2904
    },
    {
      "epoch": 0.18196338808938442,
      "grad_norm": 0.9460942378647424,
      "learning_rate": 9.406509271052051e-06,
      "loss": 0.4689,
      "step": 2905
    },
    {
      "epoch": 0.18202602608872673,
      "grad_norm": 0.8976945259041498,
      "learning_rate": 9.406029822482878e-06,
      "loss": 0.477,
      "step": 2906
    },
    {
      "epoch": 0.18208866408806904,
      "grad_norm": 0.8785565276364077,
      "learning_rate": 9.4055501925607e-06,
      "loss": 0.4374,
      "step": 2907
    },
    {
      "epoch": 0.18215130208741132,
      "grad_norm": 0.8893313334011034,
      "learning_rate": 9.405070381305258e-06,
      "loss": 0.445,
      "step": 2908
    },
    {
      "epoch": 0.18221394008675362,
      "grad_norm": 0.8211317253180791,
      "learning_rate": 9.4045903887363e-06,
      "loss": 0.4289,
      "step": 2909
    },
    {
      "epoch": 0.18227657808609593,
      "grad_norm": 0.7702010576733063,
      "learning_rate": 9.404110214873588e-06,
      "loss": 0.5123,
      "step": 2910
    },
    {
      "epoch": 0.18233921608543824,
      "grad_norm": 0.8595713954134403,
      "learning_rate": 9.40362985973688e-06,
      "loss": 0.4664,
      "step": 2911
    },
    {
      "epoch": 0.18240185408478055,
      "grad_norm": 0.8976749879976275,
      "learning_rate": 9.403149323345948e-06,
      "loss": 0.4389,
      "step": 2912
    },
    {
      "epoch": 0.18246449208412283,
      "grad_norm": 0.8538049360669636,
      "learning_rate": 9.402668605720575e-06,
      "loss": 0.4677,
      "step": 2913
    },
    {
      "epoch": 0.18252713008346513,
      "grad_norm": 0.8548285793926348,
      "learning_rate": 9.402187706880544e-06,
      "loss": 0.4368,
      "step": 2914
    },
    {
      "epoch": 0.18258976808280744,
      "grad_norm": 0.8856824206866444,
      "learning_rate": 9.40170662684565e-06,
      "loss": 0.441,
      "step": 2915
    },
    {
      "epoch": 0.18265240608214975,
      "grad_norm": 0.8200693911174156,
      "learning_rate": 9.401225365635693e-06,
      "loss": 0.4228,
      "step": 2916
    },
    {
      "epoch": 0.18271504408149203,
      "grad_norm": 0.8604082865241042,
      "learning_rate": 9.400743923270485e-06,
      "loss": 0.4238,
      "step": 2917
    },
    {
      "epoch": 0.18277768208083434,
      "grad_norm": 0.8954173561928156,
      "learning_rate": 9.40026229976984e-06,
      "loss": 0.4486,
      "step": 2918
    },
    {
      "epoch": 0.18284032008017664,
      "grad_norm": 0.6334809910949074,
      "learning_rate": 9.399780495153582e-06,
      "loss": 0.4717,
      "step": 2919
    },
    {
      "epoch": 0.18290295807951895,
      "grad_norm": 0.864278795779149,
      "learning_rate": 9.399298509441542e-06,
      "loss": 0.4388,
      "step": 2920
    },
    {
      "epoch": 0.18296559607886123,
      "grad_norm": 0.9127449344398062,
      "learning_rate": 9.398816342653558e-06,
      "loss": 0.4264,
      "step": 2921
    },
    {
      "epoch": 0.18302823407820354,
      "grad_norm": 2.1628699486887126,
      "learning_rate": 9.398333994809476e-06,
      "loss": 0.4706,
      "step": 2922
    },
    {
      "epoch": 0.18309087207754585,
      "grad_norm": 0.9025094413982775,
      "learning_rate": 9.397851465929153e-06,
      "loss": 0.4445,
      "step": 2923
    },
    {
      "epoch": 0.18315351007688815,
      "grad_norm": 0.7407921573300011,
      "learning_rate": 9.397368756032445e-06,
      "loss": 0.4674,
      "step": 2924
    },
    {
      "epoch": 0.18321614807623043,
      "grad_norm": 0.8916199476887927,
      "learning_rate": 9.396885865139224e-06,
      "loss": 0.4261,
      "step": 2925
    },
    {
      "epoch": 0.18327878607557274,
      "grad_norm": 1.5859046344609846,
      "learning_rate": 9.396402793269366e-06,
      "loss": 0.425,
      "step": 2926
    },
    {
      "epoch": 0.18334142407491505,
      "grad_norm": 0.8655470953818517,
      "learning_rate": 9.395919540442751e-06,
      "loss": 0.4189,
      "step": 2927
    },
    {
      "epoch": 0.18340406207425736,
      "grad_norm": 0.6917055174060588,
      "learning_rate": 9.395436106679273e-06,
      "loss": 0.4844,
      "step": 2928
    },
    {
      "epoch": 0.18346670007359964,
      "grad_norm": 0.9378696082175291,
      "learning_rate": 9.39495249199883e-06,
      "loss": 0.4413,
      "step": 2929
    },
    {
      "epoch": 0.18352933807294194,
      "grad_norm": 0.9942924155626044,
      "learning_rate": 9.394468696421323e-06,
      "loss": 0.4717,
      "step": 2930
    },
    {
      "epoch": 0.18359197607228425,
      "grad_norm": 0.8958081820314538,
      "learning_rate": 9.393984719966672e-06,
      "loss": 0.4359,
      "step": 2931
    },
    {
      "epoch": 0.18365461407162656,
      "grad_norm": 0.8522338588049286,
      "learning_rate": 9.393500562654794e-06,
      "loss": 0.4197,
      "step": 2932
    },
    {
      "epoch": 0.18371725207096884,
      "grad_norm": 0.9449682451833215,
      "learning_rate": 9.393016224505615e-06,
      "loss": 0.4945,
      "step": 2933
    },
    {
      "epoch": 0.18377989007031115,
      "grad_norm": 0.9126896300961488,
      "learning_rate": 9.392531705539077e-06,
      "loss": 0.4257,
      "step": 2934
    },
    {
      "epoch": 0.18384252806965345,
      "grad_norm": 0.9119960203125869,
      "learning_rate": 9.392047005775116e-06,
      "loss": 0.4505,
      "step": 2935
    },
    {
      "epoch": 0.18390516606899576,
      "grad_norm": 0.8643492757836909,
      "learning_rate": 9.391562125233683e-06,
      "loss": 0.4576,
      "step": 2936
    },
    {
      "epoch": 0.18396780406833807,
      "grad_norm": 0.9330487371855269,
      "learning_rate": 9.39107706393474e-06,
      "loss": 0.4411,
      "step": 2937
    },
    {
      "epoch": 0.18403044206768035,
      "grad_norm": 0.887143126141096,
      "learning_rate": 9.390591821898252e-06,
      "loss": 0.4278,
      "step": 2938
    },
    {
      "epoch": 0.18409308006702266,
      "grad_norm": 0.8796575869447298,
      "learning_rate": 9.390106399144188e-06,
      "loss": 0.4347,
      "step": 2939
    },
    {
      "epoch": 0.18415571806636497,
      "grad_norm": 0.9200348302726111,
      "learning_rate": 9.389620795692529e-06,
      "loss": 0.5477,
      "step": 2940
    },
    {
      "epoch": 0.18421835606570727,
      "grad_norm": 0.7887600016340023,
      "learning_rate": 9.389135011563264e-06,
      "loss": 0.393,
      "step": 2941
    },
    {
      "epoch": 0.18428099406504955,
      "grad_norm": 0.8448794658187407,
      "learning_rate": 9.388649046776387e-06,
      "loss": 0.433,
      "step": 2942
    },
    {
      "epoch": 0.18434363206439186,
      "grad_norm": 0.7997852961128792,
      "learning_rate": 9.3881629013519e-06,
      "loss": 0.3936,
      "step": 2943
    },
    {
      "epoch": 0.18440627006373417,
      "grad_norm": 0.8489304337681914,
      "learning_rate": 9.387676575309815e-06,
      "loss": 0.4468,
      "step": 2944
    },
    {
      "epoch": 0.18446890806307648,
      "grad_norm": 0.8634599276500341,
      "learning_rate": 9.387190068670148e-06,
      "loss": 0.4258,
      "step": 2945
    },
    {
      "epoch": 0.18453154606241876,
      "grad_norm": 0.8196736139182754,
      "learning_rate": 9.386703381452922e-06,
      "loss": 0.4519,
      "step": 2946
    },
    {
      "epoch": 0.18459418406176106,
      "grad_norm": 0.9049834472261471,
      "learning_rate": 9.386216513678171e-06,
      "loss": 0.4325,
      "step": 2947
    },
    {
      "epoch": 0.18465682206110337,
      "grad_norm": 0.835697972354992,
      "learning_rate": 9.385729465365935e-06,
      "loss": 0.4497,
      "step": 2948
    },
    {
      "epoch": 0.18471946006044568,
      "grad_norm": 0.833240097602765,
      "learning_rate": 9.385242236536259e-06,
      "loss": 0.413,
      "step": 2949
    },
    {
      "epoch": 0.18478209805978796,
      "grad_norm": 0.9557401177745967,
      "learning_rate": 9.3847548272092e-06,
      "loss": 0.4854,
      "step": 2950
    },
    {
      "epoch": 0.18484473605913027,
      "grad_norm": 0.982758957231048,
      "learning_rate": 9.384267237404818e-06,
      "loss": 0.4423,
      "step": 2951
    },
    {
      "epoch": 0.18490737405847257,
      "grad_norm": 0.9634258666302347,
      "learning_rate": 9.38377946714318e-06,
      "loss": 0.4844,
      "step": 2952
    },
    {
      "epoch": 0.18497001205781488,
      "grad_norm": 0.8482169402729168,
      "learning_rate": 9.383291516444368e-06,
      "loss": 0.4335,
      "step": 2953
    },
    {
      "epoch": 0.18503265005715716,
      "grad_norm": 0.9175892677814198,
      "learning_rate": 9.382803385328465e-06,
      "loss": 0.472,
      "step": 2954
    },
    {
      "epoch": 0.18509528805649947,
      "grad_norm": 0.8377373251993687,
      "learning_rate": 9.38231507381556e-06,
      "loss": 0.4123,
      "step": 2955
    },
    {
      "epoch": 0.18515792605584178,
      "grad_norm": 0.823708739515769,
      "learning_rate": 9.381826581925753e-06,
      "loss": 0.3999,
      "step": 2956
    },
    {
      "epoch": 0.18522056405518408,
      "grad_norm": 0.8812213953802267,
      "learning_rate": 9.381337909679149e-06,
      "loss": 0.4312,
      "step": 2957
    },
    {
      "epoch": 0.1852832020545264,
      "grad_norm": 0.8536729184972439,
      "learning_rate": 9.380849057095864e-06,
      "loss": 0.4024,
      "step": 2958
    },
    {
      "epoch": 0.18534584005386867,
      "grad_norm": 0.8436501449561388,
      "learning_rate": 9.380360024196021e-06,
      "loss": 0.4787,
      "step": 2959
    },
    {
      "epoch": 0.18540847805321098,
      "grad_norm": 0.8237958752191988,
      "learning_rate": 9.379870810999745e-06,
      "loss": 0.4286,
      "step": 2960
    },
    {
      "epoch": 0.1854711160525533,
      "grad_norm": 0.8960366627067327,
      "learning_rate": 9.379381417527173e-06,
      "loss": 0.4458,
      "step": 2961
    },
    {
      "epoch": 0.1855337540518956,
      "grad_norm": 0.891391677304874,
      "learning_rate": 9.378891843798448e-06,
      "loss": 0.3994,
      "step": 2962
    },
    {
      "epoch": 0.18559639205123787,
      "grad_norm": 0.970085035752376,
      "learning_rate": 9.378402089833724e-06,
      "loss": 0.4587,
      "step": 2963
    },
    {
      "epoch": 0.18565903005058018,
      "grad_norm": 0.8744739985532543,
      "learning_rate": 9.377912155653158e-06,
      "loss": 0.4547,
      "step": 2964
    },
    {
      "epoch": 0.1857216680499225,
      "grad_norm": 0.8742935489839214,
      "learning_rate": 9.377422041276913e-06,
      "loss": 0.4078,
      "step": 2965
    },
    {
      "epoch": 0.1857843060492648,
      "grad_norm": 0.8912496562547669,
      "learning_rate": 9.376931746725164e-06,
      "loss": 0.4327,
      "step": 2966
    },
    {
      "epoch": 0.18584694404860708,
      "grad_norm": 0.839985620737831,
      "learning_rate": 9.376441272018093e-06,
      "loss": 0.4309,
      "step": 2967
    },
    {
      "epoch": 0.18590958204794938,
      "grad_norm": 0.7672454734835588,
      "learning_rate": 9.375950617175888e-06,
      "loss": 0.4918,
      "step": 2968
    },
    {
      "epoch": 0.1859722200472917,
      "grad_norm": 0.9069760899794105,
      "learning_rate": 9.375459782218741e-06,
      "loss": 0.4185,
      "step": 2969
    },
    {
      "epoch": 0.186034858046634,
      "grad_norm": 0.9345623575664578,
      "learning_rate": 9.37496876716686e-06,
      "loss": 0.4242,
      "step": 2970
    },
    {
      "epoch": 0.18609749604597628,
      "grad_norm": 0.8668019833864888,
      "learning_rate": 9.37447757204045e-06,
      "loss": 0.3957,
      "step": 2971
    },
    {
      "epoch": 0.1861601340453186,
      "grad_norm": 0.8752182708023317,
      "learning_rate": 9.373986196859732e-06,
      "loss": 0.4488,
      "step": 2972
    },
    {
      "epoch": 0.1862227720446609,
      "grad_norm": 0.8885281290053582,
      "learning_rate": 9.37349464164493e-06,
      "loss": 0.437,
      "step": 2973
    },
    {
      "epoch": 0.1862854100440032,
      "grad_norm": 0.95950485753342,
      "learning_rate": 9.373002906416278e-06,
      "loss": 0.4716,
      "step": 2974
    },
    {
      "epoch": 0.18634804804334548,
      "grad_norm": 0.9114173054859926,
      "learning_rate": 9.372510991194014e-06,
      "loss": 0.4504,
      "step": 2975
    },
    {
      "epoch": 0.1864106860426878,
      "grad_norm": 0.9440123688671054,
      "learning_rate": 9.372018895998386e-06,
      "loss": 0.415,
      "step": 2976
    },
    {
      "epoch": 0.1864733240420301,
      "grad_norm": 0.8161747567552697,
      "learning_rate": 9.371526620849647e-06,
      "loss": 0.4707,
      "step": 2977
    },
    {
      "epoch": 0.1865359620413724,
      "grad_norm": 0.8532871456519749,
      "learning_rate": 9.371034165768063e-06,
      "loss": 0.4252,
      "step": 2978
    },
    {
      "epoch": 0.1865986000407147,
      "grad_norm": 0.9578869452254659,
      "learning_rate": 9.370541530773903e-06,
      "loss": 0.4909,
      "step": 2979
    },
    {
      "epoch": 0.186661238040057,
      "grad_norm": 0.8176618222697978,
      "learning_rate": 9.370048715887439e-06,
      "loss": 0.3946,
      "step": 2980
    },
    {
      "epoch": 0.1867238760393993,
      "grad_norm": 0.8648413080781157,
      "learning_rate": 9.36955572112896e-06,
      "loss": 0.4198,
      "step": 2981
    },
    {
      "epoch": 0.1867865140387416,
      "grad_norm": 0.6306535793710252,
      "learning_rate": 9.369062546518757e-06,
      "loss": 0.4795,
      "step": 2982
    },
    {
      "epoch": 0.18684915203808392,
      "grad_norm": 0.8435039767011422,
      "learning_rate": 9.368569192077129e-06,
      "loss": 0.4263,
      "step": 2983
    },
    {
      "epoch": 0.1869117900374262,
      "grad_norm": 0.8736246991365362,
      "learning_rate": 9.36807565782438e-06,
      "loss": 0.416,
      "step": 2984
    },
    {
      "epoch": 0.1869744280367685,
      "grad_norm": 0.82716405157532,
      "learning_rate": 9.367581943780828e-06,
      "loss": 0.4467,
      "step": 2985
    },
    {
      "epoch": 0.1870370660361108,
      "grad_norm": 0.8168820413821022,
      "learning_rate": 9.367088049966793e-06,
      "loss": 0.4125,
      "step": 2986
    },
    {
      "epoch": 0.18709970403545312,
      "grad_norm": 0.7441061061638214,
      "learning_rate": 9.366593976402602e-06,
      "loss": 0.4688,
      "step": 2987
    },
    {
      "epoch": 0.1871623420347954,
      "grad_norm": 0.888836178150615,
      "learning_rate": 9.366099723108592e-06,
      "loss": 0.4814,
      "step": 2988
    },
    {
      "epoch": 0.1872249800341377,
      "grad_norm": 0.9384958088507253,
      "learning_rate": 9.36560529010511e-06,
      "loss": 0.4374,
      "step": 2989
    },
    {
      "epoch": 0.18728761803348,
      "grad_norm": 0.9193705942675995,
      "learning_rate": 9.3651106774125e-06,
      "loss": 0.4515,
      "step": 2990
    },
    {
      "epoch": 0.18735025603282232,
      "grad_norm": 0.8772112727189686,
      "learning_rate": 9.364615885051128e-06,
      "loss": 0.4317,
      "step": 2991
    },
    {
      "epoch": 0.1874128940321646,
      "grad_norm": 0.8649184439628993,
      "learning_rate": 9.364120913041353e-06,
      "loss": 0.4139,
      "step": 2992
    },
    {
      "epoch": 0.1874755320315069,
      "grad_norm": 0.8915792829843158,
      "learning_rate": 9.363625761403554e-06,
      "loss": 0.4714,
      "step": 2993
    },
    {
      "epoch": 0.18753817003084922,
      "grad_norm": 0.8256879377176778,
      "learning_rate": 9.363130430158105e-06,
      "loss": 0.4128,
      "step": 2994
    },
    {
      "epoch": 0.18760080803019152,
      "grad_norm": 0.7146391496164469,
      "learning_rate": 9.362634919325402e-06,
      "loss": 0.4619,
      "step": 2995
    },
    {
      "epoch": 0.1876634460295338,
      "grad_norm": 0.9080143543201559,
      "learning_rate": 9.362139228925833e-06,
      "loss": 0.4686,
      "step": 2996
    },
    {
      "epoch": 0.1877260840288761,
      "grad_norm": 0.7937751368453742,
      "learning_rate": 9.361643358979805e-06,
      "loss": 0.3848,
      "step": 2997
    },
    {
      "epoch": 0.18778872202821842,
      "grad_norm": 0.7862591022586987,
      "learning_rate": 9.361147309507726e-06,
      "loss": 0.3868,
      "step": 2998
    },
    {
      "epoch": 0.18785136002756073,
      "grad_norm": 0.8508209448196775,
      "learning_rate": 9.360651080530015e-06,
      "loss": 0.4658,
      "step": 2999
    },
    {
      "epoch": 0.18791399802690303,
      "grad_norm": 0.8186512091890964,
      "learning_rate": 9.360154672067094e-06,
      "loss": 0.4935,
      "step": 3000
    },
    {
      "epoch": 0.18797663602624531,
      "grad_norm": 0.6817637541848756,
      "learning_rate": 9.3596580841394e-06,
      "loss": 0.4679,
      "step": 3001
    },
    {
      "epoch": 0.18803927402558762,
      "grad_norm": 0.7866681133054536,
      "learning_rate": 9.35916131676737e-06,
      "loss": 0.3969,
      "step": 3002
    },
    {
      "epoch": 0.18810191202492993,
      "grad_norm": 0.7920781804804674,
      "learning_rate": 9.358664369971449e-06,
      "loss": 0.4206,
      "step": 3003
    },
    {
      "epoch": 0.18816455002427224,
      "grad_norm": 0.9429228546412305,
      "learning_rate": 9.358167243772095e-06,
      "loss": 0.4603,
      "step": 3004
    },
    {
      "epoch": 0.18822718802361452,
      "grad_norm": 0.8241467257869807,
      "learning_rate": 9.357669938189768e-06,
      "loss": 0.4069,
      "step": 3005
    },
    {
      "epoch": 0.18828982602295682,
      "grad_norm": 0.9468794548996553,
      "learning_rate": 9.357172453244939e-06,
      "loss": 0.4457,
      "step": 3006
    },
    {
      "epoch": 0.18835246402229913,
      "grad_norm": 0.8358511365434284,
      "learning_rate": 9.356674788958082e-06,
      "loss": 0.4502,
      "step": 3007
    },
    {
      "epoch": 0.18841510202164144,
      "grad_norm": 0.7933254255397642,
      "learning_rate": 9.356176945349682e-06,
      "loss": 0.4082,
      "step": 3008
    },
    {
      "epoch": 0.18847774002098372,
      "grad_norm": 0.909375788219589,
      "learning_rate": 9.35567892244023e-06,
      "loss": 0.4803,
      "step": 3009
    },
    {
      "epoch": 0.18854037802032603,
      "grad_norm": 0.8796628997122837,
      "learning_rate": 9.355180720250225e-06,
      "loss": 0.4961,
      "step": 3010
    },
    {
      "epoch": 0.18860301601966833,
      "grad_norm": 0.8845721909827922,
      "learning_rate": 9.354682338800175e-06,
      "loss": 0.4476,
      "step": 3011
    },
    {
      "epoch": 0.18866565401901064,
      "grad_norm": 0.8119657244959361,
      "learning_rate": 9.354183778110589e-06,
      "loss": 0.4131,
      "step": 3012
    },
    {
      "epoch": 0.18872829201835292,
      "grad_norm": 0.7953609806754641,
      "learning_rate": 9.353685038201991e-06,
      "loss": 0.4139,
      "step": 3013
    },
    {
      "epoch": 0.18879093001769523,
      "grad_norm": 0.820104067979418,
      "learning_rate": 9.35318611909491e-06,
      "loss": 0.4348,
      "step": 3014
    },
    {
      "epoch": 0.18885356801703754,
      "grad_norm": 0.8687014629323044,
      "learning_rate": 9.352687020809879e-06,
      "loss": 0.4514,
      "step": 3015
    },
    {
      "epoch": 0.18891620601637985,
      "grad_norm": 0.8681136033298055,
      "learning_rate": 9.352187743367444e-06,
      "loss": 0.4223,
      "step": 3016
    },
    {
      "epoch": 0.18897884401572212,
      "grad_norm": 0.8246536256739053,
      "learning_rate": 9.351688286788152e-06,
      "loss": 0.3869,
      "step": 3017
    },
    {
      "epoch": 0.18904148201506443,
      "grad_norm": 0.6632715269860872,
      "learning_rate": 9.351188651092565e-06,
      "loss": 0.4837,
      "step": 3018
    },
    {
      "epoch": 0.18910412001440674,
      "grad_norm": 0.932282230158109,
      "learning_rate": 9.350688836301242e-06,
      "loss": 0.4384,
      "step": 3019
    },
    {
      "epoch": 0.18916675801374905,
      "grad_norm": 0.9073391397791334,
      "learning_rate": 9.350188842434762e-06,
      "loss": 0.4854,
      "step": 3020
    },
    {
      "epoch": 0.18922939601309136,
      "grad_norm": 0.9263084792757222,
      "learning_rate": 9.3496886695137e-06,
      "loss": 0.4496,
      "step": 3021
    },
    {
      "epoch": 0.18929203401243364,
      "grad_norm": 0.8647271345446552,
      "learning_rate": 9.349188317558646e-06,
      "loss": 0.4278,
      "step": 3022
    },
    {
      "epoch": 0.18935467201177594,
      "grad_norm": 0.9155470716860798,
      "learning_rate": 9.348687786590194e-06,
      "loss": 0.4197,
      "step": 3023
    },
    {
      "epoch": 0.18941731001111825,
      "grad_norm": 0.824650837703512,
      "learning_rate": 9.348187076628942e-06,
      "loss": 0.441,
      "step": 3024
    },
    {
      "epoch": 0.18947994801046056,
      "grad_norm": 0.7397877795798785,
      "learning_rate": 9.347686187695507e-06,
      "loss": 0.5043,
      "step": 3025
    },
    {
      "epoch": 0.18954258600980284,
      "grad_norm": 0.7637500217145495,
      "learning_rate": 9.3471851198105e-06,
      "loss": 0.4533,
      "step": 3026
    },
    {
      "epoch": 0.18960522400914515,
      "grad_norm": 0.830366539560474,
      "learning_rate": 9.346683872994546e-06,
      "loss": 0.4106,
      "step": 3027
    },
    {
      "epoch": 0.18966786200848745,
      "grad_norm": 0.9305339062995966,
      "learning_rate": 9.346182447268278e-06,
      "loss": 0.4439,
      "step": 3028
    },
    {
      "epoch": 0.18973050000782976,
      "grad_norm": 0.8704920758100215,
      "learning_rate": 9.345680842652333e-06,
      "loss": 0.4082,
      "step": 3029
    },
    {
      "epoch": 0.18979313800717204,
      "grad_norm": 0.7461920725349197,
      "learning_rate": 9.34517905916736e-06,
      "loss": 0.4132,
      "step": 3030
    },
    {
      "epoch": 0.18985577600651435,
      "grad_norm": 0.9042773976597814,
      "learning_rate": 9.344677096834009e-06,
      "loss": 0.4587,
      "step": 3031
    },
    {
      "epoch": 0.18991841400585666,
      "grad_norm": 0.8770806743502902,
      "learning_rate": 9.344174955672942e-06,
      "loss": 0.4107,
      "step": 3032
    },
    {
      "epoch": 0.18998105200519896,
      "grad_norm": 0.871111615664811,
      "learning_rate": 9.343672635704827e-06,
      "loss": 0.4342,
      "step": 3033
    },
    {
      "epoch": 0.19004369000454124,
      "grad_norm": 0.8275954110803647,
      "learning_rate": 9.343170136950342e-06,
      "loss": 0.4226,
      "step": 3034
    },
    {
      "epoch": 0.19010632800388355,
      "grad_norm": 0.9443312696641138,
      "learning_rate": 9.342667459430167e-06,
      "loss": 0.4461,
      "step": 3035
    },
    {
      "epoch": 0.19016896600322586,
      "grad_norm": 0.8744675249752464,
      "learning_rate": 9.34216460316499e-06,
      "loss": 0.4204,
      "step": 3036
    },
    {
      "epoch": 0.19023160400256817,
      "grad_norm": 0.8748358302711422,
      "learning_rate": 9.341661568175517e-06,
      "loss": 0.4361,
      "step": 3037
    },
    {
      "epoch": 0.19029424200191045,
      "grad_norm": 0.8472694503526759,
      "learning_rate": 9.341158354482446e-06,
      "loss": 0.4211,
      "step": 3038
    },
    {
      "epoch": 0.19035688000125275,
      "grad_norm": 0.8815773984721492,
      "learning_rate": 9.34065496210649e-06,
      "loss": 0.45,
      "step": 3039
    },
    {
      "epoch": 0.19041951800059506,
      "grad_norm": 0.8727655266516333,
      "learning_rate": 9.340151391068373e-06,
      "loss": 0.4739,
      "step": 3040
    },
    {
      "epoch": 0.19048215599993737,
      "grad_norm": 0.8247862257973978,
      "learning_rate": 9.339647641388818e-06,
      "loss": 0.4379,
      "step": 3041
    },
    {
      "epoch": 0.19054479399927968,
      "grad_norm": 0.8690850717342572,
      "learning_rate": 9.33914371308856e-06,
      "loss": 0.4019,
      "step": 3042
    },
    {
      "epoch": 0.19060743199862196,
      "grad_norm": 0.8722235390323724,
      "learning_rate": 9.338639606188342e-06,
      "loss": 0.4327,
      "step": 3043
    },
    {
      "epoch": 0.19067006999796426,
      "grad_norm": 0.8638857467415523,
      "learning_rate": 9.338135320708912e-06,
      "loss": 0.4287,
      "step": 3044
    },
    {
      "epoch": 0.19073270799730657,
      "grad_norm": 0.8390084625866391,
      "learning_rate": 9.337630856671027e-06,
      "loss": 0.4033,
      "step": 3045
    },
    {
      "epoch": 0.19079534599664888,
      "grad_norm": 0.9346052336881862,
      "learning_rate": 9.337126214095452e-06,
      "loss": 0.4435,
      "step": 3046
    },
    {
      "epoch": 0.19085798399599116,
      "grad_norm": 0.8397197979585252,
      "learning_rate": 9.336621393002954e-06,
      "loss": 0.4705,
      "step": 3047
    },
    {
      "epoch": 0.19092062199533347,
      "grad_norm": 0.9403805279334988,
      "learning_rate": 9.336116393414318e-06,
      "loss": 0.473,
      "step": 3048
    },
    {
      "epoch": 0.19098325999467577,
      "grad_norm": 0.9396772597370433,
      "learning_rate": 9.335611215350325e-06,
      "loss": 0.5023,
      "step": 3049
    },
    {
      "epoch": 0.19104589799401808,
      "grad_norm": 0.8989161833086503,
      "learning_rate": 9.33510585883177e-06,
      "loss": 0.4141,
      "step": 3050
    },
    {
      "epoch": 0.19110853599336036,
      "grad_norm": 0.7931722682600236,
      "learning_rate": 9.334600323879453e-06,
      "loss": 0.4112,
      "step": 3051
    },
    {
      "epoch": 0.19117117399270267,
      "grad_norm": 0.8469773260607972,
      "learning_rate": 9.334094610514182e-06,
      "loss": 0.4596,
      "step": 3052
    },
    {
      "epoch": 0.19123381199204498,
      "grad_norm": 0.8601726239985092,
      "learning_rate": 9.333588718756772e-06,
      "loss": 0.4573,
      "step": 3053
    },
    {
      "epoch": 0.19129644999138729,
      "grad_norm": 0.8076133329432712,
      "learning_rate": 9.333082648628047e-06,
      "loss": 0.4119,
      "step": 3054
    },
    {
      "epoch": 0.19135908799072956,
      "grad_norm": 0.8281470383041887,
      "learning_rate": 9.332576400148836e-06,
      "loss": 0.4105,
      "step": 3055
    },
    {
      "epoch": 0.19142172599007187,
      "grad_norm": 0.9742558317217965,
      "learning_rate": 9.332069973339975e-06,
      "loss": 0.3979,
      "step": 3056
    },
    {
      "epoch": 0.19148436398941418,
      "grad_norm": 0.8974588680086785,
      "learning_rate": 9.33156336822231e-06,
      "loss": 0.4608,
      "step": 3057
    },
    {
      "epoch": 0.1915470019887565,
      "grad_norm": 0.8569994082953107,
      "learning_rate": 9.331056584816693e-06,
      "loss": 0.436,
      "step": 3058
    },
    {
      "epoch": 0.19160963998809877,
      "grad_norm": 0.8824115500720264,
      "learning_rate": 9.330549623143983e-06,
      "loss": 0.4213,
      "step": 3059
    },
    {
      "epoch": 0.19167227798744108,
      "grad_norm": 0.8930722385912591,
      "learning_rate": 9.330042483225046e-06,
      "loss": 0.4592,
      "step": 3060
    },
    {
      "epoch": 0.19173491598678338,
      "grad_norm": 0.857486055719045,
      "learning_rate": 9.329535165080758e-06,
      "loss": 0.4185,
      "step": 3061
    },
    {
      "epoch": 0.1917975539861257,
      "grad_norm": 0.7602386009832971,
      "learning_rate": 9.329027668731999e-06,
      "loss": 0.4153,
      "step": 3062
    },
    {
      "epoch": 0.19186019198546797,
      "grad_norm": 0.8430855886037426,
      "learning_rate": 9.328519994199658e-06,
      "loss": 0.4217,
      "step": 3063
    },
    {
      "epoch": 0.19192282998481028,
      "grad_norm": 0.8327093938161867,
      "learning_rate": 9.32801214150463e-06,
      "loss": 0.4296,
      "step": 3064
    },
    {
      "epoch": 0.19198546798415259,
      "grad_norm": 0.8305535191422616,
      "learning_rate": 9.327504110667818e-06,
      "loss": 0.4158,
      "step": 3065
    },
    {
      "epoch": 0.1920481059834949,
      "grad_norm": 0.8276445484004731,
      "learning_rate": 9.326995901710134e-06,
      "loss": 0.4219,
      "step": 3066
    },
    {
      "epoch": 0.1921107439828372,
      "grad_norm": 0.8609994942086974,
      "learning_rate": 9.326487514652496e-06,
      "loss": 0.4094,
      "step": 3067
    },
    {
      "epoch": 0.19217338198217948,
      "grad_norm": 0.8260368366916747,
      "learning_rate": 9.325978949515831e-06,
      "loss": 0.4444,
      "step": 3068
    },
    {
      "epoch": 0.1922360199815218,
      "grad_norm": 0.8659415993784343,
      "learning_rate": 9.325470206321067e-06,
      "loss": 0.435,
      "step": 3069
    },
    {
      "epoch": 0.1922986579808641,
      "grad_norm": 0.875401372109644,
      "learning_rate": 9.324961285089147e-06,
      "loss": 0.4609,
      "step": 3070
    },
    {
      "epoch": 0.1923612959802064,
      "grad_norm": 0.8621830778910594,
      "learning_rate": 9.324452185841018e-06,
      "loss": 0.4267,
      "step": 3071
    },
    {
      "epoch": 0.19242393397954868,
      "grad_norm": 0.7687702660913278,
      "learning_rate": 9.323942908597633e-06,
      "loss": 0.4045,
      "step": 3072
    },
    {
      "epoch": 0.192486571978891,
      "grad_norm": 0.7842446178429626,
      "learning_rate": 9.323433453379955e-06,
      "loss": 0.4028,
      "step": 3073
    },
    {
      "epoch": 0.1925492099782333,
      "grad_norm": 0.8261225683715523,
      "learning_rate": 9.322923820208955e-06,
      "loss": 0.4128,
      "step": 3074
    },
    {
      "epoch": 0.1926118479775756,
      "grad_norm": 0.830413934992402,
      "learning_rate": 9.32241400910561e-06,
      "loss": 0.3932,
      "step": 3075
    },
    {
      "epoch": 0.1926744859769179,
      "grad_norm": 0.8572303864007756,
      "learning_rate": 9.321904020090899e-06,
      "loss": 0.4381,
      "step": 3076
    },
    {
      "epoch": 0.1927371239762602,
      "grad_norm": 0.7847889895551254,
      "learning_rate": 9.321393853185818e-06,
      "loss": 0.4821,
      "step": 3077
    },
    {
      "epoch": 0.1927997619756025,
      "grad_norm": 0.7726649272218009,
      "learning_rate": 9.320883508411363e-06,
      "loss": 0.4274,
      "step": 3078
    },
    {
      "epoch": 0.1928623999749448,
      "grad_norm": 0.8052274932004644,
      "learning_rate": 9.320372985788541e-06,
      "loss": 0.4064,
      "step": 3079
    },
    {
      "epoch": 0.1929250379742871,
      "grad_norm": 0.8320172656088374,
      "learning_rate": 9.319862285338366e-06,
      "loss": 0.4027,
      "step": 3080
    },
    {
      "epoch": 0.1929876759736294,
      "grad_norm": 0.9300903982577783,
      "learning_rate": 9.319351407081856e-06,
      "loss": 0.4503,
      "step": 3081
    },
    {
      "epoch": 0.1930503139729717,
      "grad_norm": 0.8869817265084241,
      "learning_rate": 9.318840351040041e-06,
      "loss": 0.4508,
      "step": 3082
    },
    {
      "epoch": 0.193112951972314,
      "grad_norm": 0.8871352622775628,
      "learning_rate": 9.318329117233955e-06,
      "loss": 0.4614,
      "step": 3083
    },
    {
      "epoch": 0.1931755899716563,
      "grad_norm": 0.8239932151646637,
      "learning_rate": 9.317817705684643e-06,
      "loss": 0.3987,
      "step": 3084
    },
    {
      "epoch": 0.1932382279709986,
      "grad_norm": 0.8534177520613482,
      "learning_rate": 9.317306116413152e-06,
      "loss": 0.4111,
      "step": 3085
    },
    {
      "epoch": 0.1933008659703409,
      "grad_norm": 0.9193232929662934,
      "learning_rate": 9.31679434944054e-06,
      "loss": 0.4366,
      "step": 3086
    },
    {
      "epoch": 0.19336350396968321,
      "grad_norm": 0.9016515072439987,
      "learning_rate": 9.31628240478787e-06,
      "loss": 0.4948,
      "step": 3087
    },
    {
      "epoch": 0.19342614196902552,
      "grad_norm": 0.8881390049184057,
      "learning_rate": 9.315770282476216e-06,
      "loss": 0.4193,
      "step": 3088
    },
    {
      "epoch": 0.1934887799683678,
      "grad_norm": 0.8721170851459044,
      "learning_rate": 9.315257982526656e-06,
      "loss": 0.4502,
      "step": 3089
    },
    {
      "epoch": 0.1935514179677101,
      "grad_norm": 0.7865021705055056,
      "learning_rate": 9.314745504960276e-06,
      "loss": 0.3728,
      "step": 3090
    },
    {
      "epoch": 0.19361405596705242,
      "grad_norm": 0.8438787786968912,
      "learning_rate": 9.31423284979817e-06,
      "loss": 0.4022,
      "step": 3091
    },
    {
      "epoch": 0.19367669396639473,
      "grad_norm": 0.8842098687395605,
      "learning_rate": 9.31372001706144e-06,
      "loss": 0.4193,
      "step": 3092
    },
    {
      "epoch": 0.193739331965737,
      "grad_norm": 0.8991501517033545,
      "learning_rate": 9.313207006771195e-06,
      "loss": 0.4519,
      "step": 3093
    },
    {
      "epoch": 0.1938019699650793,
      "grad_norm": 0.7673330457476106,
      "learning_rate": 9.312693818948546e-06,
      "loss": 0.4889,
      "step": 3094
    },
    {
      "epoch": 0.19386460796442162,
      "grad_norm": 0.920314176984053,
      "learning_rate": 9.312180453614621e-06,
      "loss": 0.4446,
      "step": 3095
    },
    {
      "epoch": 0.19392724596376393,
      "grad_norm": 0.8154490329474425,
      "learning_rate": 9.311666910790547e-06,
      "loss": 0.4578,
      "step": 3096
    },
    {
      "epoch": 0.1939898839631062,
      "grad_norm": 0.7941508689007526,
      "learning_rate": 9.311153190497464e-06,
      "loss": 0.4339,
      "step": 3097
    },
    {
      "epoch": 0.19405252196244852,
      "grad_norm": 0.8812251868502797,
      "learning_rate": 9.310639292756513e-06,
      "loss": 0.4487,
      "step": 3098
    },
    {
      "epoch": 0.19411515996179082,
      "grad_norm": 0.8686273501612921,
      "learning_rate": 9.31012521758885e-06,
      "loss": 0.4361,
      "step": 3099
    },
    {
      "epoch": 0.19417779796113313,
      "grad_norm": 0.852728731195404,
      "learning_rate": 9.309610965015633e-06,
      "loss": 0.4438,
      "step": 3100
    },
    {
      "epoch": 0.1942404359604754,
      "grad_norm": 0.827399043181498,
      "learning_rate": 9.30909653505803e-06,
      "loss": 0.4285,
      "step": 3101
    },
    {
      "epoch": 0.19430307395981772,
      "grad_norm": 0.8448094151863449,
      "learning_rate": 9.308581927737212e-06,
      "loss": 0.4545,
      "step": 3102
    },
    {
      "epoch": 0.19436571195916003,
      "grad_norm": 0.8780476772392917,
      "learning_rate": 9.308067143074364e-06,
      "loss": 0.4406,
      "step": 3103
    },
    {
      "epoch": 0.19442834995850233,
      "grad_norm": 0.7413438733132895,
      "learning_rate": 9.307552181090671e-06,
      "loss": 0.4962,
      "step": 3104
    },
    {
      "epoch": 0.1944909879578446,
      "grad_norm": 0.8410074268283168,
      "learning_rate": 9.307037041807333e-06,
      "loss": 0.3969,
      "step": 3105
    },
    {
      "epoch": 0.19455362595718692,
      "grad_norm": 0.7786434635251416,
      "learning_rate": 9.306521725245548e-06,
      "loss": 0.3902,
      "step": 3106
    },
    {
      "epoch": 0.19461626395652923,
      "grad_norm": 0.879452395694447,
      "learning_rate": 9.30600623142653e-06,
      "loss": 0.446,
      "step": 3107
    },
    {
      "epoch": 0.19467890195587154,
      "grad_norm": 0.6604266472150047,
      "learning_rate": 9.305490560371497e-06,
      "loss": 0.4738,
      "step": 3108
    },
    {
      "epoch": 0.19474153995521384,
      "grad_norm": 0.9069115229409969,
      "learning_rate": 9.304974712101675e-06,
      "loss": 0.4319,
      "step": 3109
    },
    {
      "epoch": 0.19480417795455612,
      "grad_norm": 0.8996658026964673,
      "learning_rate": 9.304458686638293e-06,
      "loss": 0.4323,
      "step": 3110
    },
    {
      "epoch": 0.19486681595389843,
      "grad_norm": 0.9001088578831788,
      "learning_rate": 9.303942484002592e-06,
      "loss": 0.4632,
      "step": 3111
    },
    {
      "epoch": 0.19492945395324074,
      "grad_norm": 0.8537822160934425,
      "learning_rate": 9.303426104215821e-06,
      "loss": 0.4694,
      "step": 3112
    },
    {
      "epoch": 0.19499209195258305,
      "grad_norm": 0.9374478332940329,
      "learning_rate": 9.30290954729923e-06,
      "loss": 0.4833,
      "step": 3113
    },
    {
      "epoch": 0.19505472995192533,
      "grad_norm": 0.8657315094607193,
      "learning_rate": 9.302392813274087e-06,
      "loss": 0.4321,
      "step": 3114
    },
    {
      "epoch": 0.19511736795126763,
      "grad_norm": 0.8537650889141112,
      "learning_rate": 9.301875902161656e-06,
      "loss": 0.4591,
      "step": 3115
    },
    {
      "epoch": 0.19518000595060994,
      "grad_norm": 0.8338046585366955,
      "learning_rate": 9.301358813983215e-06,
      "loss": 0.4408,
      "step": 3116
    },
    {
      "epoch": 0.19524264394995225,
      "grad_norm": 0.9011751406171186,
      "learning_rate": 9.300841548760044e-06,
      "loss": 0.405,
      "step": 3117
    },
    {
      "epoch": 0.19530528194929453,
      "grad_norm": 0.9890854714662722,
      "learning_rate": 9.300324106513438e-06,
      "loss": 0.4679,
      "step": 3118
    },
    {
      "epoch": 0.19536791994863684,
      "grad_norm": 0.8406900942536542,
      "learning_rate": 9.299806487264695e-06,
      "loss": 0.418,
      "step": 3119
    },
    {
      "epoch": 0.19543055794797914,
      "grad_norm": 0.8130164889679083,
      "learning_rate": 9.29928869103512e-06,
      "loss": 0.4337,
      "step": 3120
    },
    {
      "epoch": 0.19549319594732145,
      "grad_norm": 0.8887683263091417,
      "learning_rate": 9.298770717846023e-06,
      "loss": 0.454,
      "step": 3121
    },
    {
      "epoch": 0.19555583394666373,
      "grad_norm": 0.6773303924239191,
      "learning_rate": 9.298252567718723e-06,
      "loss": 0.4796,
      "step": 3122
    },
    {
      "epoch": 0.19561847194600604,
      "grad_norm": 0.842330051828045,
      "learning_rate": 9.297734240674551e-06,
      "loss": 0.4332,
      "step": 3123
    },
    {
      "epoch": 0.19568110994534835,
      "grad_norm": 0.9042877565429607,
      "learning_rate": 9.297215736734842e-06,
      "loss": 0.4419,
      "step": 3124
    },
    {
      "epoch": 0.19574374794469065,
      "grad_norm": 0.8203520722144101,
      "learning_rate": 9.296697055920934e-06,
      "loss": 0.4265,
      "step": 3125
    },
    {
      "epoch": 0.19580638594403293,
      "grad_norm": 0.9572551405527563,
      "learning_rate": 9.296178198254176e-06,
      "loss": 0.4267,
      "step": 3126
    },
    {
      "epoch": 0.19586902394337524,
      "grad_norm": 0.844319016557032,
      "learning_rate": 9.295659163755929e-06,
      "loss": 0.4305,
      "step": 3127
    },
    {
      "epoch": 0.19593166194271755,
      "grad_norm": 0.8561080305539454,
      "learning_rate": 9.295139952447551e-06,
      "loss": 0.4379,
      "step": 3128
    },
    {
      "epoch": 0.19599429994205986,
      "grad_norm": 0.8476124050192345,
      "learning_rate": 9.294620564350418e-06,
      "loss": 0.4503,
      "step": 3129
    },
    {
      "epoch": 0.19605693794140217,
      "grad_norm": 0.881782620911456,
      "learning_rate": 9.294100999485903e-06,
      "loss": 0.448,
      "step": 3130
    },
    {
      "epoch": 0.19611957594074444,
      "grad_norm": 0.9425507279915378,
      "learning_rate": 9.293581257875395e-06,
      "loss": 0.4501,
      "step": 3131
    },
    {
      "epoch": 0.19618221394008675,
      "grad_norm": 0.856093067212581,
      "learning_rate": 9.293061339540286e-06,
      "loss": 0.4139,
      "step": 3132
    },
    {
      "epoch": 0.19624485193942906,
      "grad_norm": 0.7933404525687834,
      "learning_rate": 9.292541244501974e-06,
      "loss": 0.4048,
      "step": 3133
    },
    {
      "epoch": 0.19630748993877137,
      "grad_norm": 0.9054864883396326,
      "learning_rate": 9.292020972781868e-06,
      "loss": 0.4067,
      "step": 3134
    },
    {
      "epoch": 0.19637012793811365,
      "grad_norm": 0.8832017306985076,
      "learning_rate": 9.291500524401381e-06,
      "loss": 0.4877,
      "step": 3135
    },
    {
      "epoch": 0.19643276593745596,
      "grad_norm": 0.8268120920333365,
      "learning_rate": 9.29097989938194e-06,
      "loss": 0.4076,
      "step": 3136
    },
    {
      "epoch": 0.19649540393679826,
      "grad_norm": 0.8155807045690256,
      "learning_rate": 9.290459097744966e-06,
      "loss": 0.4396,
      "step": 3137
    },
    {
      "epoch": 0.19655804193614057,
      "grad_norm": 0.9247682438397257,
      "learning_rate": 9.289938119511901e-06,
      "loss": 0.4344,
      "step": 3138
    },
    {
      "epoch": 0.19662067993548285,
      "grad_norm": 0.8159477092472619,
      "learning_rate": 9.289416964704186e-06,
      "loss": 0.4337,
      "step": 3139
    },
    {
      "epoch": 0.19668331793482516,
      "grad_norm": 0.9148352085674376,
      "learning_rate": 9.288895633343273e-06,
      "loss": 0.4676,
      "step": 3140
    },
    {
      "epoch": 0.19674595593416747,
      "grad_norm": 0.8103297484839199,
      "learning_rate": 9.288374125450619e-06,
      "loss": 0.4571,
      "step": 3141
    },
    {
      "epoch": 0.19680859393350977,
      "grad_norm": 0.8566518695257539,
      "learning_rate": 9.28785244104769e-06,
      "loss": 0.4526,
      "step": 3142
    },
    {
      "epoch": 0.19687123193285205,
      "grad_norm": 0.8209376032432789,
      "learning_rate": 9.28733058015596e-06,
      "loss": 0.426,
      "step": 3143
    },
    {
      "epoch": 0.19693386993219436,
      "grad_norm": 0.9198633631968351,
      "learning_rate": 9.286808542796906e-06,
      "loss": 0.4618,
      "step": 3144
    },
    {
      "epoch": 0.19699650793153667,
      "grad_norm": 0.846720034104665,
      "learning_rate": 9.286286328992018e-06,
      "loss": 0.4688,
      "step": 3145
    },
    {
      "epoch": 0.19705914593087898,
      "grad_norm": 0.819176995191718,
      "learning_rate": 9.285763938762789e-06,
      "loss": 0.4008,
      "step": 3146
    },
    {
      "epoch": 0.19712178393022126,
      "grad_norm": 0.7627090205884608,
      "learning_rate": 9.28524137213072e-06,
      "loss": 0.4845,
      "step": 3147
    },
    {
      "epoch": 0.19718442192956356,
      "grad_norm": 0.8898614351117483,
      "learning_rate": 9.284718629117322e-06,
      "loss": 0.4558,
      "step": 3148
    },
    {
      "epoch": 0.19724705992890587,
      "grad_norm": 0.8387937521707625,
      "learning_rate": 9.284195709744109e-06,
      "loss": 0.4188,
      "step": 3149
    },
    {
      "epoch": 0.19730969792824818,
      "grad_norm": 0.7874626621048965,
      "learning_rate": 9.283672614032605e-06,
      "loss": 0.3769,
      "step": 3150
    },
    {
      "epoch": 0.1973723359275905,
      "grad_norm": 0.8926433930245751,
      "learning_rate": 9.283149342004342e-06,
      "loss": 0.4355,
      "step": 3151
    },
    {
      "epoch": 0.19743497392693277,
      "grad_norm": 0.8043407498379131,
      "learning_rate": 9.282625893680857e-06,
      "loss": 0.4086,
      "step": 3152
    },
    {
      "epoch": 0.19749761192627507,
      "grad_norm": 0.8291650828070337,
      "learning_rate": 9.282102269083695e-06,
      "loss": 0.4086,
      "step": 3153
    },
    {
      "epoch": 0.19756024992561738,
      "grad_norm": 0.8074363365845668,
      "learning_rate": 9.281578468234408e-06,
      "loss": 0.4109,
      "step": 3154
    },
    {
      "epoch": 0.1976228879249597,
      "grad_norm": 0.8530288343061533,
      "learning_rate": 9.281054491154557e-06,
      "loss": 0.4275,
      "step": 3155
    },
    {
      "epoch": 0.19768552592430197,
      "grad_norm": 0.8959180341427273,
      "learning_rate": 9.280530337865709e-06,
      "loss": 0.4826,
      "step": 3156
    },
    {
      "epoch": 0.19774816392364428,
      "grad_norm": 0.777317562085768,
      "learning_rate": 9.280006008389439e-06,
      "loss": 0.4258,
      "step": 3157
    },
    {
      "epoch": 0.19781080192298658,
      "grad_norm": 0.9230634956881277,
      "learning_rate": 9.279481502747325e-06,
      "loss": 0.4665,
      "step": 3158
    },
    {
      "epoch": 0.1978734399223289,
      "grad_norm": 0.8141987078557152,
      "learning_rate": 9.27895682096096e-06,
      "loss": 0.4197,
      "step": 3159
    },
    {
      "epoch": 0.19793607792167117,
      "grad_norm": 0.8870271054472848,
      "learning_rate": 9.278431963051937e-06,
      "loss": 0.4762,
      "step": 3160
    },
    {
      "epoch": 0.19799871592101348,
      "grad_norm": 0.9262466578813463,
      "learning_rate": 9.27790692904186e-06,
      "loss": 0.4832,
      "step": 3161
    },
    {
      "epoch": 0.1980613539203558,
      "grad_norm": 0.8159137525607528,
      "learning_rate": 9.27738171895234e-06,
      "loss": 0.4447,
      "step": 3162
    },
    {
      "epoch": 0.1981239919196981,
      "grad_norm": 0.8754869567535126,
      "learning_rate": 9.276856332804993e-06,
      "loss": 0.4528,
      "step": 3163
    },
    {
      "epoch": 0.19818662991904037,
      "grad_norm": 0.9495649549166846,
      "learning_rate": 9.27633077062145e-06,
      "loss": 0.4149,
      "step": 3164
    },
    {
      "epoch": 0.19824926791838268,
      "grad_norm": 0.8897951199391345,
      "learning_rate": 9.275805032423336e-06,
      "loss": 0.4682,
      "step": 3165
    },
    {
      "epoch": 0.198311905917725,
      "grad_norm": 0.8798812981728719,
      "learning_rate": 9.275279118232293e-06,
      "loss": 0.4526,
      "step": 3166
    },
    {
      "epoch": 0.1983745439170673,
      "grad_norm": 0.8839182292217876,
      "learning_rate": 9.274753028069967e-06,
      "loss": 0.4829,
      "step": 3167
    },
    {
      "epoch": 0.19843718191640958,
      "grad_norm": 0.9028428960041952,
      "learning_rate": 9.274226761958015e-06,
      "loss": 0.3922,
      "step": 3168
    },
    {
      "epoch": 0.19849981991575189,
      "grad_norm": 0.8344018444820097,
      "learning_rate": 9.273700319918094e-06,
      "loss": 0.4218,
      "step": 3169
    },
    {
      "epoch": 0.1985624579150942,
      "grad_norm": 0.8420940623702444,
      "learning_rate": 9.273173701971876e-06,
      "loss": 0.4112,
      "step": 3170
    },
    {
      "epoch": 0.1986250959144365,
      "grad_norm": 0.8342319100506808,
      "learning_rate": 9.272646908141034e-06,
      "loss": 0.4435,
      "step": 3171
    },
    {
      "epoch": 0.19868773391377878,
      "grad_norm": 0.8424750120724059,
      "learning_rate": 9.272119938447254e-06,
      "loss": 0.405,
      "step": 3172
    },
    {
      "epoch": 0.1987503719131211,
      "grad_norm": 0.8288627218022735,
      "learning_rate": 9.271592792912224e-06,
      "loss": 0.4999,
      "step": 3173
    },
    {
      "epoch": 0.1988130099124634,
      "grad_norm": 0.7953699725498034,
      "learning_rate": 9.27106547155764e-06,
      "loss": 0.4162,
      "step": 3174
    },
    {
      "epoch": 0.1988756479118057,
      "grad_norm": 0.9566658860155697,
      "learning_rate": 9.270537974405212e-06,
      "loss": 0.4237,
      "step": 3175
    },
    {
      "epoch": 0.198938285911148,
      "grad_norm": 0.810101936134063,
      "learning_rate": 9.270010301476646e-06,
      "loss": 0.4604,
      "step": 3176
    },
    {
      "epoch": 0.1990009239104903,
      "grad_norm": 0.8274138954533432,
      "learning_rate": 9.269482452793663e-06,
      "loss": 0.4189,
      "step": 3177
    },
    {
      "epoch": 0.1990635619098326,
      "grad_norm": 0.8164427442389796,
      "learning_rate": 9.26895442837799e-06,
      "loss": 0.3801,
      "step": 3178
    },
    {
      "epoch": 0.1991261999091749,
      "grad_norm": 0.9139564246320285,
      "learning_rate": 9.268426228251361e-06,
      "loss": 0.4539,
      "step": 3179
    },
    {
      "epoch": 0.1991888379085172,
      "grad_norm": 0.8691302713033695,
      "learning_rate": 9.267897852435516e-06,
      "loss": 0.4337,
      "step": 3180
    },
    {
      "epoch": 0.1992514759078595,
      "grad_norm": 0.7958028492366991,
      "learning_rate": 9.267369300952204e-06,
      "loss": 0.4158,
      "step": 3181
    },
    {
      "epoch": 0.1993141139072018,
      "grad_norm": 0.9454790769861078,
      "learning_rate": 9.266840573823178e-06,
      "loss": 0.4807,
      "step": 3182
    },
    {
      "epoch": 0.1993767519065441,
      "grad_norm": 0.8894327380089813,
      "learning_rate": 9.266311671070204e-06,
      "loss": 0.4607,
      "step": 3183
    },
    {
      "epoch": 0.19943938990588642,
      "grad_norm": 0.8876185127534177,
      "learning_rate": 9.26578259271505e-06,
      "loss": 0.3958,
      "step": 3184
    },
    {
      "epoch": 0.1995020279052287,
      "grad_norm": 0.8875544257289466,
      "learning_rate": 9.26525333877949e-06,
      "loss": 0.4507,
      "step": 3185
    },
    {
      "epoch": 0.199564665904571,
      "grad_norm": 0.6949227126025407,
      "learning_rate": 9.264723909285311e-06,
      "loss": 0.4957,
      "step": 3186
    },
    {
      "epoch": 0.1996273039039133,
      "grad_norm": 0.8929264250545792,
      "learning_rate": 9.264194304254306e-06,
      "loss": 0.3702,
      "step": 3187
    },
    {
      "epoch": 0.19968994190325562,
      "grad_norm": 0.7597806520684258,
      "learning_rate": 9.263664523708273e-06,
      "loss": 0.4836,
      "step": 3188
    },
    {
      "epoch": 0.1997525799025979,
      "grad_norm": 0.8566010782520077,
      "learning_rate": 9.263134567669014e-06,
      "loss": 0.4194,
      "step": 3189
    },
    {
      "epoch": 0.1998152179019402,
      "grad_norm": 0.8118780445845543,
      "learning_rate": 9.262604436158347e-06,
      "loss": 0.4363,
      "step": 3190
    },
    {
      "epoch": 0.19987785590128251,
      "grad_norm": 0.8042127339111369,
      "learning_rate": 9.262074129198089e-06,
      "loss": 0.4308,
      "step": 3191
    },
    {
      "epoch": 0.19994049390062482,
      "grad_norm": 0.8253447963679993,
      "learning_rate": 9.26154364681007e-06,
      "loss": 0.4072,
      "step": 3192
    },
    {
      "epoch": 0.2000031318999671,
      "grad_norm": 0.8913013118748445,
      "learning_rate": 9.261012989016123e-06,
      "loss": 0.4284,
      "step": 3193
    },
    {
      "epoch": 0.2000657698993094,
      "grad_norm": 0.8176330209204661,
      "learning_rate": 9.26048215583809e-06,
      "loss": 0.3742,
      "step": 3194
    },
    {
      "epoch": 0.20012840789865172,
      "grad_norm": 0.8677222756749469,
      "learning_rate": 9.25995114729782e-06,
      "loss": 0.4035,
      "step": 3195
    },
    {
      "epoch": 0.20019104589799402,
      "grad_norm": 0.8287920022063134,
      "learning_rate": 9.259419963417172e-06,
      "loss": 0.4721,
      "step": 3196
    },
    {
      "epoch": 0.20025368389733633,
      "grad_norm": 0.9025430016251274,
      "learning_rate": 9.258888604218006e-06,
      "loss": 0.4444,
      "step": 3197
    },
    {
      "epoch": 0.2003163218966786,
      "grad_norm": 0.9165437458903829,
      "learning_rate": 9.258357069722196e-06,
      "loss": 0.4452,
      "step": 3198
    },
    {
      "epoch": 0.20037895989602092,
      "grad_norm": 0.8302601797190812,
      "learning_rate": 9.257825359951618e-06,
      "loss": 0.4166,
      "step": 3199
    },
    {
      "epoch": 0.20044159789536323,
      "grad_norm": 0.96728417349147,
      "learning_rate": 9.257293474928158e-06,
      "loss": 0.4405,
      "step": 3200
    },
    {
      "epoch": 0.20050423589470553,
      "grad_norm": 0.8672100633651084,
      "learning_rate": 9.256761414673708e-06,
      "loss": 0.468,
      "step": 3201
    },
    {
      "epoch": 0.20056687389404781,
      "grad_norm": 0.9579957403327793,
      "learning_rate": 9.256229179210168e-06,
      "loss": 0.4285,
      "step": 3202
    },
    {
      "epoch": 0.20062951189339012,
      "grad_norm": 0.8303615479066816,
      "learning_rate": 9.255696768559445e-06,
      "loss": 0.4172,
      "step": 3203
    },
    {
      "epoch": 0.20069214989273243,
      "grad_norm": 0.8309464596197392,
      "learning_rate": 9.255164182743454e-06,
      "loss": 0.4288,
      "step": 3204
    },
    {
      "epoch": 0.20075478789207474,
      "grad_norm": 0.8561357556257311,
      "learning_rate": 9.254631421784114e-06,
      "loss": 0.461,
      "step": 3205
    },
    {
      "epoch": 0.20081742589141702,
      "grad_norm": 0.8769505334600178,
      "learning_rate": 9.254098485703355e-06,
      "loss": 0.4626,
      "step": 3206
    },
    {
      "epoch": 0.20088006389075933,
      "grad_norm": 0.9284051091790393,
      "learning_rate": 9.253565374523115e-06,
      "loss": 0.4525,
      "step": 3207
    },
    {
      "epoch": 0.20094270189010163,
      "grad_norm": 0.8464030756434526,
      "learning_rate": 9.253032088265333e-06,
      "loss": 0.3803,
      "step": 3208
    },
    {
      "epoch": 0.20100533988944394,
      "grad_norm": 0.8410659166290186,
      "learning_rate": 9.25249862695196e-06,
      "loss": 0.4142,
      "step": 3209
    },
    {
      "epoch": 0.20106797788878622,
      "grad_norm": 0.8636949289835598,
      "learning_rate": 9.251964990604956e-06,
      "loss": 0.5105,
      "step": 3210
    },
    {
      "epoch": 0.20113061588812853,
      "grad_norm": 0.86710828236121,
      "learning_rate": 9.251431179246284e-06,
      "loss": 0.4,
      "step": 3211
    },
    {
      "epoch": 0.20119325388747084,
      "grad_norm": 0.8523340207468089,
      "learning_rate": 9.250897192897915e-06,
      "loss": 0.3634,
      "step": 3212
    },
    {
      "epoch": 0.20125589188681314,
      "grad_norm": 0.807578275802673,
      "learning_rate": 9.250363031581828e-06,
      "loss": 0.4277,
      "step": 3213
    },
    {
      "epoch": 0.20131852988615542,
      "grad_norm": 0.8571194887649893,
      "learning_rate": 9.24982869532001e-06,
      "loss": 0.4046,
      "step": 3214
    },
    {
      "epoch": 0.20138116788549773,
      "grad_norm": 0.7626605227259577,
      "learning_rate": 9.249294184134453e-06,
      "loss": 0.397,
      "step": 3215
    },
    {
      "epoch": 0.20144380588484004,
      "grad_norm": 0.8355847717459856,
      "learning_rate": 9.24875949804716e-06,
      "loss": 0.3957,
      "step": 3216
    },
    {
      "epoch": 0.20150644388418235,
      "grad_norm": 0.879600757167357,
      "learning_rate": 9.248224637080137e-06,
      "loss": 0.4268,
      "step": 3217
    },
    {
      "epoch": 0.20156908188352465,
      "grad_norm": 0.9111939623188466,
      "learning_rate": 9.2476896012554e-06,
      "loss": 0.4807,
      "step": 3218
    },
    {
      "epoch": 0.20163171988286693,
      "grad_norm": 0.8109071899228582,
      "learning_rate": 9.24715439059497e-06,
      "loss": 0.4128,
      "step": 3219
    },
    {
      "epoch": 0.20169435788220924,
      "grad_norm": 0.9227355069528325,
      "learning_rate": 9.246619005120876e-06,
      "loss": 0.4712,
      "step": 3220
    },
    {
      "epoch": 0.20175699588155155,
      "grad_norm": 0.9238201195393845,
      "learning_rate": 9.246083444855158e-06,
      "loss": 0.5004,
      "step": 3221
    },
    {
      "epoch": 0.20181963388089386,
      "grad_norm": 0.8242325885402408,
      "learning_rate": 9.245547709819856e-06,
      "loss": 0.4306,
      "step": 3222
    },
    {
      "epoch": 0.20188227188023614,
      "grad_norm": 0.8711866329206238,
      "learning_rate": 9.245011800037022e-06,
      "loss": 0.431,
      "step": 3223
    },
    {
      "epoch": 0.20194490987957844,
      "grad_norm": 0.8334503514142803,
      "learning_rate": 9.244475715528715e-06,
      "loss": 0.4013,
      "step": 3224
    },
    {
      "epoch": 0.20200754787892075,
      "grad_norm": 0.8202788861741351,
      "learning_rate": 9.243939456317e-06,
      "loss": 0.4334,
      "step": 3225
    },
    {
      "epoch": 0.20207018587826306,
      "grad_norm": 0.8629029070680639,
      "learning_rate": 9.243403022423947e-06,
      "loss": 0.4303,
      "step": 3226
    },
    {
      "epoch": 0.20213282387760534,
      "grad_norm": 0.7871267553124784,
      "learning_rate": 9.242866413871641e-06,
      "loss": 0.5018,
      "step": 3227
    },
    {
      "epoch": 0.20219546187694765,
      "grad_norm": 0.8930075060036754,
      "learning_rate": 9.242329630682163e-06,
      "loss": 0.4582,
      "step": 3228
    },
    {
      "epoch": 0.20225809987628995,
      "grad_norm": 0.8623281922507666,
      "learning_rate": 9.241792672877612e-06,
      "loss": 0.4269,
      "step": 3229
    },
    {
      "epoch": 0.20232073787563226,
      "grad_norm": 0.7790640768760101,
      "learning_rate": 9.241255540480088e-06,
      "loss": 0.4177,
      "step": 3230
    },
    {
      "epoch": 0.20238337587497454,
      "grad_norm": 0.9752679352726026,
      "learning_rate": 9.240718233511697e-06,
      "loss": 0.4704,
      "step": 3231
    },
    {
      "epoch": 0.20244601387431685,
      "grad_norm": 0.8225161919442645,
      "learning_rate": 9.240180751994557e-06,
      "loss": 0.4115,
      "step": 3232
    },
    {
      "epoch": 0.20250865187365916,
      "grad_norm": 0.8606692760182363,
      "learning_rate": 9.239643095950789e-06,
      "loss": 0.423,
      "step": 3233
    },
    {
      "epoch": 0.20257128987300146,
      "grad_norm": 0.8647871189373134,
      "learning_rate": 9.239105265402525e-06,
      "loss": 0.4176,
      "step": 3234
    },
    {
      "epoch": 0.20263392787234374,
      "grad_norm": 0.8899592350668667,
      "learning_rate": 9.238567260371902e-06,
      "loss": 0.4147,
      "step": 3235
    },
    {
      "epoch": 0.20269656587168605,
      "grad_norm": 0.8719673353782664,
      "learning_rate": 9.238029080881063e-06,
      "loss": 0.4457,
      "step": 3236
    },
    {
      "epoch": 0.20275920387102836,
      "grad_norm": 0.8459635230956709,
      "learning_rate": 9.237490726952161e-06,
      "loss": 0.4106,
      "step": 3237
    },
    {
      "epoch": 0.20282184187037067,
      "grad_norm": 0.8488295857027797,
      "learning_rate": 9.236952198607356e-06,
      "loss": 0.4745,
      "step": 3238
    },
    {
      "epoch": 0.20288447986971297,
      "grad_norm": 0.9335063355701069,
      "learning_rate": 9.23641349586881e-06,
      "loss": 0.4474,
      "step": 3239
    },
    {
      "epoch": 0.20294711786905525,
      "grad_norm": 0.8168319726807147,
      "learning_rate": 9.235874618758698e-06,
      "loss": 0.3928,
      "step": 3240
    },
    {
      "epoch": 0.20300975586839756,
      "grad_norm": 0.80422984559777,
      "learning_rate": 9.235335567299202e-06,
      "loss": 0.4371,
      "step": 3241
    },
    {
      "epoch": 0.20307239386773987,
      "grad_norm": 0.7659351550655774,
      "learning_rate": 9.234796341512506e-06,
      "loss": 0.3891,
      "step": 3242
    },
    {
      "epoch": 0.20313503186708218,
      "grad_norm": 0.8406051253424243,
      "learning_rate": 9.234256941420807e-06,
      "loss": 0.4175,
      "step": 3243
    },
    {
      "epoch": 0.20319766986642446,
      "grad_norm": 0.8844273279916403,
      "learning_rate": 9.233717367046308e-06,
      "loss": 0.4783,
      "step": 3244
    },
    {
      "epoch": 0.20326030786576677,
      "grad_norm": 0.8526448568375605,
      "learning_rate": 9.233177618411215e-06,
      "loss": 0.417,
      "step": 3245
    },
    {
      "epoch": 0.20332294586510907,
      "grad_norm": 0.7946169068535859,
      "learning_rate": 9.232637695537745e-06,
      "loss": 0.4299,
      "step": 3246
    },
    {
      "epoch": 0.20338558386445138,
      "grad_norm": 0.9540225842460709,
      "learning_rate": 9.232097598448123e-06,
      "loss": 0.457,
      "step": 3247
    },
    {
      "epoch": 0.20344822186379366,
      "grad_norm": 0.8606512705446555,
      "learning_rate": 9.231557327164577e-06,
      "loss": 0.4391,
      "step": 3248
    },
    {
      "epoch": 0.20351085986313597,
      "grad_norm": 0.8171470401099932,
      "learning_rate": 9.231016881709349e-06,
      "loss": 0.3958,
      "step": 3249
    },
    {
      "epoch": 0.20357349786247828,
      "grad_norm": 0.8559540311928551,
      "learning_rate": 9.230476262104678e-06,
      "loss": 0.4844,
      "step": 3250
    },
    {
      "epoch": 0.20363613586182058,
      "grad_norm": 0.8252453001044721,
      "learning_rate": 9.22993546837282e-06,
      "loss": 0.413,
      "step": 3251
    },
    {
      "epoch": 0.20369877386116286,
      "grad_norm": 0.9498664625832934,
      "learning_rate": 9.229394500536031e-06,
      "loss": 0.4814,
      "step": 3252
    },
    {
      "epoch": 0.20376141186050517,
      "grad_norm": 0.8522777405365718,
      "learning_rate": 9.228853358616583e-06,
      "loss": 0.4238,
      "step": 3253
    },
    {
      "epoch": 0.20382404985984748,
      "grad_norm": 0.838087611597928,
      "learning_rate": 9.228312042636742e-06,
      "loss": 0.3963,
      "step": 3254
    },
    {
      "epoch": 0.20388668785918979,
      "grad_norm": 0.786745448684298,
      "learning_rate": 9.227770552618796e-06,
      "loss": 0.4205,
      "step": 3255
    },
    {
      "epoch": 0.20394932585853207,
      "grad_norm": 0.8465129928236776,
      "learning_rate": 9.227228888585026e-06,
      "loss": 0.5101,
      "step": 3256
    },
    {
      "epoch": 0.20401196385787437,
      "grad_norm": 0.8491993551880797,
      "learning_rate": 9.226687050557733e-06,
      "loss": 0.4435,
      "step": 3257
    },
    {
      "epoch": 0.20407460185721668,
      "grad_norm": 0.8986361216456349,
      "learning_rate": 9.226145038559216e-06,
      "loss": 0.4338,
      "step": 3258
    },
    {
      "epoch": 0.204137239856559,
      "grad_norm": 0.9148932098197372,
      "learning_rate": 9.225602852611783e-06,
      "loss": 0.4312,
      "step": 3259
    },
    {
      "epoch": 0.2041998778559013,
      "grad_norm": 0.7926063305066009,
      "learning_rate": 9.225060492737754e-06,
      "loss": 0.4331,
      "step": 3260
    },
    {
      "epoch": 0.20426251585524358,
      "grad_norm": 0.8215270707517172,
      "learning_rate": 9.224517958959452e-06,
      "loss": 0.448,
      "step": 3261
    },
    {
      "epoch": 0.20432515385458588,
      "grad_norm": 0.8474693878494831,
      "learning_rate": 9.223975251299204e-06,
      "loss": 0.4412,
      "step": 3262
    },
    {
      "epoch": 0.2043877918539282,
      "grad_norm": 0.8880981535723546,
      "learning_rate": 9.223432369779353e-06,
      "loss": 0.4665,
      "step": 3263
    },
    {
      "epoch": 0.2044504298532705,
      "grad_norm": 0.8453393922475091,
      "learning_rate": 9.22288931442224e-06,
      "loss": 0.3856,
      "step": 3264
    },
    {
      "epoch": 0.20451306785261278,
      "grad_norm": 0.8093588130340836,
      "learning_rate": 9.22234608525022e-06,
      "loss": 0.4084,
      "step": 3265
    },
    {
      "epoch": 0.2045757058519551,
      "grad_norm": 0.8433584698680905,
      "learning_rate": 9.22180268228565e-06,
      "loss": 0.4576,
      "step": 3266
    },
    {
      "epoch": 0.2046383438512974,
      "grad_norm": 0.8801963038246357,
      "learning_rate": 9.221259105550899e-06,
      "loss": 0.4288,
      "step": 3267
    },
    {
      "epoch": 0.2047009818506397,
      "grad_norm": 0.8521240231599905,
      "learning_rate": 9.22071535506834e-06,
      "loss": 0.4275,
      "step": 3268
    },
    {
      "epoch": 0.20476361984998198,
      "grad_norm": 0.8885123942630785,
      "learning_rate": 9.220171430860354e-06,
      "loss": 0.4712,
      "step": 3269
    },
    {
      "epoch": 0.2048262578493243,
      "grad_norm": 0.8883169945183278,
      "learning_rate": 9.219627332949329e-06,
      "loss": 0.4407,
      "step": 3270
    },
    {
      "epoch": 0.2048888958486666,
      "grad_norm": 0.841542583321638,
      "learning_rate": 9.219083061357657e-06,
      "loss": 0.4266,
      "step": 3271
    },
    {
      "epoch": 0.2049515338480089,
      "grad_norm": 0.8823332386938695,
      "learning_rate": 9.218538616107745e-06,
      "loss": 0.4497,
      "step": 3272
    },
    {
      "epoch": 0.20501417184735118,
      "grad_norm": 0.8038784409286737,
      "learning_rate": 9.217993997222e-06,
      "loss": 0.435,
      "step": 3273
    },
    {
      "epoch": 0.2050768098466935,
      "grad_norm": 0.9349981274869598,
      "learning_rate": 9.217449204722839e-06,
      "loss": 0.4378,
      "step": 3274
    },
    {
      "epoch": 0.2051394478460358,
      "grad_norm": 0.7841121310918725,
      "learning_rate": 9.216904238632687e-06,
      "loss": 0.3912,
      "step": 3275
    },
    {
      "epoch": 0.2052020858453781,
      "grad_norm": 0.8398200865770461,
      "learning_rate": 9.216359098973974e-06,
      "loss": 0.3856,
      "step": 3276
    },
    {
      "epoch": 0.2052647238447204,
      "grad_norm": 0.9763858015361466,
      "learning_rate": 9.215813785769136e-06,
      "loss": 0.4571,
      "step": 3277
    },
    {
      "epoch": 0.2053273618440627,
      "grad_norm": 0.868716076287844,
      "learning_rate": 9.215268299040622e-06,
      "loss": 0.433,
      "step": 3278
    },
    {
      "epoch": 0.205389999843405,
      "grad_norm": 0.9024598783588731,
      "learning_rate": 9.214722638810882e-06,
      "loss": 0.4288,
      "step": 3279
    },
    {
      "epoch": 0.2054526378427473,
      "grad_norm": 0.8869001389003607,
      "learning_rate": 9.214176805102375e-06,
      "loss": 0.5048,
      "step": 3280
    },
    {
      "epoch": 0.20551527584208962,
      "grad_norm": 0.986829687202644,
      "learning_rate": 9.213630797937569e-06,
      "loss": 0.4409,
      "step": 3281
    },
    {
      "epoch": 0.2055779138414319,
      "grad_norm": 0.7048444942591653,
      "learning_rate": 9.213084617338938e-06,
      "loss": 0.4989,
      "step": 3282
    },
    {
      "epoch": 0.2056405518407742,
      "grad_norm": 0.8741414315116655,
      "learning_rate": 9.212538263328961e-06,
      "loss": 0.439,
      "step": 3283
    },
    {
      "epoch": 0.2057031898401165,
      "grad_norm": 0.8525264120550041,
      "learning_rate": 9.211991735930129e-06,
      "loss": 0.4327,
      "step": 3284
    },
    {
      "epoch": 0.20576582783945882,
      "grad_norm": 0.924715723932997,
      "learning_rate": 9.211445035164935e-06,
      "loss": 0.4395,
      "step": 3285
    },
    {
      "epoch": 0.2058284658388011,
      "grad_norm": 0.8775756381474793,
      "learning_rate": 9.21089816105588e-06,
      "loss": 0.443,
      "step": 3286
    },
    {
      "epoch": 0.2058911038381434,
      "grad_norm": 0.8981642660236029,
      "learning_rate": 9.210351113625475e-06,
      "loss": 0.4215,
      "step": 3287
    },
    {
      "epoch": 0.20595374183748572,
      "grad_norm": 0.9184965540830196,
      "learning_rate": 9.209803892896238e-06,
      "loss": 0.4483,
      "step": 3288
    },
    {
      "epoch": 0.20601637983682802,
      "grad_norm": 0.8195331560864743,
      "learning_rate": 9.209256498890691e-06,
      "loss": 0.4412,
      "step": 3289
    },
    {
      "epoch": 0.2060790178361703,
      "grad_norm": 0.8338889371574072,
      "learning_rate": 9.208708931631365e-06,
      "loss": 0.4134,
      "step": 3290
    },
    {
      "epoch": 0.2061416558355126,
      "grad_norm": 0.9905878520585092,
      "learning_rate": 9.208161191140799e-06,
      "loss": 0.4957,
      "step": 3291
    },
    {
      "epoch": 0.20620429383485492,
      "grad_norm": 0.8642057550673428,
      "learning_rate": 9.207613277441536e-06,
      "loss": 0.4516,
      "step": 3292
    },
    {
      "epoch": 0.20626693183419723,
      "grad_norm": 0.8035659820464052,
      "learning_rate": 9.207065190556131e-06,
      "loss": 0.4857,
      "step": 3293
    },
    {
      "epoch": 0.2063295698335395,
      "grad_norm": 0.9401476363303195,
      "learning_rate": 9.20651693050714e-06,
      "loss": 0.4558,
      "step": 3294
    },
    {
      "epoch": 0.2063922078328818,
      "grad_norm": 0.9528509689532044,
      "learning_rate": 9.205968497317133e-06,
      "loss": 0.4633,
      "step": 3295
    },
    {
      "epoch": 0.20645484583222412,
      "grad_norm": 0.9759149737339148,
      "learning_rate": 9.205419891008682e-06,
      "loss": 0.4653,
      "step": 3296
    },
    {
      "epoch": 0.20651748383156643,
      "grad_norm": 0.934037209336129,
      "learning_rate": 9.204871111604368e-06,
      "loss": 0.4266,
      "step": 3297
    },
    {
      "epoch": 0.2065801218309087,
      "grad_norm": 0.975892419618638,
      "learning_rate": 9.204322159126777e-06,
      "loss": 0.4817,
      "step": 3298
    },
    {
      "epoch": 0.20664275983025102,
      "grad_norm": 0.8682236893656792,
      "learning_rate": 9.203773033598506e-06,
      "loss": 0.4429,
      "step": 3299
    },
    {
      "epoch": 0.20670539782959332,
      "grad_norm": 0.8525460811615844,
      "learning_rate": 9.203223735042157e-06,
      "loss": 0.4394,
      "step": 3300
    },
    {
      "epoch": 0.20676803582893563,
      "grad_norm": 0.8573448814381951,
      "learning_rate": 9.202674263480338e-06,
      "loss": 0.4211,
      "step": 3301
    },
    {
      "epoch": 0.2068306738282779,
      "grad_norm": 0.8449327165079669,
      "learning_rate": 9.202124618935669e-06,
      "loss": 0.3768,
      "step": 3302
    },
    {
      "epoch": 0.20689331182762022,
      "grad_norm": 0.9380842442114333,
      "learning_rate": 9.201574801430768e-06,
      "loss": 0.4225,
      "step": 3303
    },
    {
      "epoch": 0.20695594982696253,
      "grad_norm": 0.8428742377560019,
      "learning_rate": 9.201024810988271e-06,
      "loss": 0.4131,
      "step": 3304
    },
    {
      "epoch": 0.20701858782630483,
      "grad_norm": 0.90573157748622,
      "learning_rate": 9.20047464763081e-06,
      "loss": 0.4676,
      "step": 3305
    },
    {
      "epoch": 0.20708122582564714,
      "grad_norm": 0.9036935557734621,
      "learning_rate": 9.199924311381035e-06,
      "loss": 0.4496,
      "step": 3306
    },
    {
      "epoch": 0.20714386382498942,
      "grad_norm": 1.3376934777703455,
      "learning_rate": 9.199373802261594e-06,
      "loss": 0.5168,
      "step": 3307
    },
    {
      "epoch": 0.20720650182433173,
      "grad_norm": 0.8684602183382305,
      "learning_rate": 9.198823120295149e-06,
      "loss": 0.419,
      "step": 3308
    },
    {
      "epoch": 0.20726913982367404,
      "grad_norm": 0.8163476239561485,
      "learning_rate": 9.198272265504364e-06,
      "loss": 0.4247,
      "step": 3309
    },
    {
      "epoch": 0.20733177782301634,
      "grad_norm": 0.8377682146518003,
      "learning_rate": 9.197721237911915e-06,
      "loss": 0.375,
      "step": 3310
    },
    {
      "epoch": 0.20739441582235862,
      "grad_norm": 0.804596309568741,
      "learning_rate": 9.19717003754048e-06,
      "loss": 0.3952,
      "step": 3311
    },
    {
      "epoch": 0.20745705382170093,
      "grad_norm": 0.8208002583251885,
      "learning_rate": 9.196618664412746e-06,
      "loss": 0.4195,
      "step": 3312
    },
    {
      "epoch": 0.20751969182104324,
      "grad_norm": 0.8287930929769609,
      "learning_rate": 9.19606711855141e-06,
      "loss": 0.4512,
      "step": 3313
    },
    {
      "epoch": 0.20758232982038555,
      "grad_norm": 0.8652480848829999,
      "learning_rate": 9.195515399979174e-06,
      "loss": 0.4339,
      "step": 3314
    },
    {
      "epoch": 0.20764496781972783,
      "grad_norm": 0.878380307756373,
      "learning_rate": 9.194963508718743e-06,
      "loss": 0.4529,
      "step": 3315
    },
    {
      "epoch": 0.20770760581907013,
      "grad_norm": 0.862769453760316,
      "learning_rate": 9.194411444792836e-06,
      "loss": 0.4764,
      "step": 3316
    },
    {
      "epoch": 0.20777024381841244,
      "grad_norm": 0.892252823798781,
      "learning_rate": 9.193859208224176e-06,
      "loss": 0.4357,
      "step": 3317
    },
    {
      "epoch": 0.20783288181775475,
      "grad_norm": 0.8502293822839186,
      "learning_rate": 9.193306799035494e-06,
      "loss": 0.4272,
      "step": 3318
    },
    {
      "epoch": 0.20789551981709703,
      "grad_norm": 0.8407019195103553,
      "learning_rate": 9.192754217249523e-06,
      "loss": 0.4417,
      "step": 3319
    },
    {
      "epoch": 0.20795815781643934,
      "grad_norm": 0.8690700985593811,
      "learning_rate": 9.192201462889012e-06,
      "loss": 0.4968,
      "step": 3320
    },
    {
      "epoch": 0.20802079581578165,
      "grad_norm": 0.8343541475030292,
      "learning_rate": 9.19164853597671e-06,
      "loss": 0.4244,
      "step": 3321
    },
    {
      "epoch": 0.20808343381512395,
      "grad_norm": 0.8785686486060936,
      "learning_rate": 9.191095436535376e-06,
      "loss": 0.4039,
      "step": 3322
    },
    {
      "epoch": 0.20814607181446623,
      "grad_norm": 0.879600344827115,
      "learning_rate": 9.190542164587776e-06,
      "loss": 0.4128,
      "step": 3323
    },
    {
      "epoch": 0.20820870981380854,
      "grad_norm": 0.8402476629803167,
      "learning_rate": 9.189988720156682e-06,
      "loss": 0.4718,
      "step": 3324
    },
    {
      "epoch": 0.20827134781315085,
      "grad_norm": 0.7082814964233262,
      "learning_rate": 9.189435103264876e-06,
      "loss": 0.4816,
      "step": 3325
    },
    {
      "epoch": 0.20833398581249316,
      "grad_norm": 0.8883845935650176,
      "learning_rate": 9.188881313935143e-06,
      "loss": 0.4543,
      "step": 3326
    },
    {
      "epoch": 0.20839662381183546,
      "grad_norm": 0.8621083407413281,
      "learning_rate": 9.188327352190279e-06,
      "loss": 0.4551,
      "step": 3327
    },
    {
      "epoch": 0.20845926181117774,
      "grad_norm": 0.8554286413378765,
      "learning_rate": 9.187773218053082e-06,
      "loss": 0.4267,
      "step": 3328
    },
    {
      "epoch": 0.20852189981052005,
      "grad_norm": 0.8102099193001588,
      "learning_rate": 9.187218911546363e-06,
      "loss": 0.4446,
      "step": 3329
    },
    {
      "epoch": 0.20858453780986236,
      "grad_norm": 0.812868169457042,
      "learning_rate": 9.186664432692935e-06,
      "loss": 0.4057,
      "step": 3330
    },
    {
      "epoch": 0.20864717580920467,
      "grad_norm": 0.8445059810374039,
      "learning_rate": 9.186109781515623e-06,
      "loss": 0.4629,
      "step": 3331
    },
    {
      "epoch": 0.20870981380854695,
      "grad_norm": 0.9196187841604291,
      "learning_rate": 9.185554958037255e-06,
      "loss": 0.4398,
      "step": 3332
    },
    {
      "epoch": 0.20877245180788925,
      "grad_norm": 0.9281409503617892,
      "learning_rate": 9.184999962280667e-06,
      "loss": 0.4453,
      "step": 3333
    },
    {
      "epoch": 0.20883508980723156,
      "grad_norm": 0.7906767513896703,
      "learning_rate": 9.184444794268707e-06,
      "loss": 0.4824,
      "step": 3334
    },
    {
      "epoch": 0.20889772780657387,
      "grad_norm": 0.9014624503120523,
      "learning_rate": 9.18388945402422e-06,
      "loss": 0.4707,
      "step": 3335
    },
    {
      "epoch": 0.20896036580591615,
      "grad_norm": 0.7901486618255228,
      "learning_rate": 9.183333941570067e-06,
      "loss": 0.4539,
      "step": 3336
    },
    {
      "epoch": 0.20902300380525846,
      "grad_norm": 0.772541668779061,
      "learning_rate": 9.182778256929112e-06,
      "loss": 0.4208,
      "step": 3337
    },
    {
      "epoch": 0.20908564180460076,
      "grad_norm": 0.8619055389279087,
      "learning_rate": 9.182222400124229e-06,
      "loss": 0.4491,
      "step": 3338
    },
    {
      "epoch": 0.20914827980394307,
      "grad_norm": 0.8095572377089879,
      "learning_rate": 9.181666371178295e-06,
      "loss": 0.394,
      "step": 3339
    },
    {
      "epoch": 0.20921091780328535,
      "grad_norm": 0.8972102760495322,
      "learning_rate": 9.181110170114197e-06,
      "loss": 0.4437,
      "step": 3340
    },
    {
      "epoch": 0.20927355580262766,
      "grad_norm": 0.9373634758214595,
      "learning_rate": 9.18055379695483e-06,
      "loss": 0.4523,
      "step": 3341
    },
    {
      "epoch": 0.20933619380196997,
      "grad_norm": 0.8112106002338824,
      "learning_rate": 9.17999725172309e-06,
      "loss": 0.4329,
      "step": 3342
    },
    {
      "epoch": 0.20939883180131227,
      "grad_norm": 0.8979828896218737,
      "learning_rate": 9.179440534441887e-06,
      "loss": 0.4406,
      "step": 3343
    },
    {
      "epoch": 0.20946146980065455,
      "grad_norm": 0.8810433535787447,
      "learning_rate": 9.178883645134138e-06,
      "loss": 0.4744,
      "step": 3344
    },
    {
      "epoch": 0.20952410779999686,
      "grad_norm": 0.836520095203867,
      "learning_rate": 9.178326583822762e-06,
      "loss": 0.4357,
      "step": 3345
    },
    {
      "epoch": 0.20958674579933917,
      "grad_norm": 0.8520675726483377,
      "learning_rate": 9.177769350530688e-06,
      "loss": 0.4076,
      "step": 3346
    },
    {
      "epoch": 0.20964938379868148,
      "grad_norm": 0.7338945331881971,
      "learning_rate": 9.177211945280852e-06,
      "loss": 0.3886,
      "step": 3347
    },
    {
      "epoch": 0.20971202179802378,
      "grad_norm": 0.8532459415734124,
      "learning_rate": 9.176654368096198e-06,
      "loss": 0.4692,
      "step": 3348
    },
    {
      "epoch": 0.20977465979736606,
      "grad_norm": 0.8463146254488535,
      "learning_rate": 9.176096618999673e-06,
      "loss": 0.4255,
      "step": 3349
    },
    {
      "epoch": 0.20983729779670837,
      "grad_norm": 0.8256090493144704,
      "learning_rate": 9.175538698014237e-06,
      "loss": 0.4447,
      "step": 3350
    },
    {
      "epoch": 0.20989993579605068,
      "grad_norm": 0.8743625546416893,
      "learning_rate": 9.174980605162855e-06,
      "loss": 0.4125,
      "step": 3351
    },
    {
      "epoch": 0.209962573795393,
      "grad_norm": 0.8357571253111237,
      "learning_rate": 9.174422340468495e-06,
      "loss": 0.4407,
      "step": 3352
    },
    {
      "epoch": 0.21002521179473527,
      "grad_norm": 0.8138941396973759,
      "learning_rate": 9.173863903954135e-06,
      "loss": 0.4585,
      "step": 3353
    },
    {
      "epoch": 0.21008784979407757,
      "grad_norm": 0.854330156929319,
      "learning_rate": 9.173305295642763e-06,
      "loss": 0.4326,
      "step": 3354
    },
    {
      "epoch": 0.21015048779341988,
      "grad_norm": 0.9123262484525455,
      "learning_rate": 9.172746515557371e-06,
      "loss": 0.4583,
      "step": 3355
    },
    {
      "epoch": 0.2102131257927622,
      "grad_norm": 0.8136257241081035,
      "learning_rate": 9.172187563720958e-06,
      "loss": 0.4248,
      "step": 3356
    },
    {
      "epoch": 0.21027576379210447,
      "grad_norm": 0.8141656766190709,
      "learning_rate": 9.171628440156531e-06,
      "loss": 0.4422,
      "step": 3357
    },
    {
      "epoch": 0.21033840179144678,
      "grad_norm": 0.8429831484959491,
      "learning_rate": 9.171069144887101e-06,
      "loss": 0.4396,
      "step": 3358
    },
    {
      "epoch": 0.21040103979078909,
      "grad_norm": 0.9087373046967302,
      "learning_rate": 9.170509677935694e-06,
      "loss": 0.4107,
      "step": 3359
    },
    {
      "epoch": 0.2104636777901314,
      "grad_norm": 0.8009442604051205,
      "learning_rate": 9.169950039325331e-06,
      "loss": 0.4462,
      "step": 3360
    },
    {
      "epoch": 0.21052631578947367,
      "grad_norm": 0.8334550670730261,
      "learning_rate": 9.169390229079054e-06,
      "loss": 0.4115,
      "step": 3361
    },
    {
      "epoch": 0.21058895378881598,
      "grad_norm": 0.802683878193965,
      "learning_rate": 9.168830247219898e-06,
      "loss": 0.4074,
      "step": 3362
    },
    {
      "epoch": 0.2106515917881583,
      "grad_norm": 0.8820812276645772,
      "learning_rate": 9.168270093770918e-06,
      "loss": 0.4428,
      "step": 3363
    },
    {
      "epoch": 0.2107142297875006,
      "grad_norm": 0.7827213323641778,
      "learning_rate": 9.167709768755165e-06,
      "loss": 0.4162,
      "step": 3364
    },
    {
      "epoch": 0.21077686778684288,
      "grad_norm": 0.8065461297720375,
      "learning_rate": 9.167149272195706e-06,
      "loss": 0.4138,
      "step": 3365
    },
    {
      "epoch": 0.21083950578618518,
      "grad_norm": 0.8585240458389404,
      "learning_rate": 9.166588604115607e-06,
      "loss": 0.4578,
      "step": 3366
    },
    {
      "epoch": 0.2109021437855275,
      "grad_norm": 0.8132387600523127,
      "learning_rate": 9.166027764537951e-06,
      "loss": 0.4493,
      "step": 3367
    },
    {
      "epoch": 0.2109647817848698,
      "grad_norm": 0.8312605688871612,
      "learning_rate": 9.165466753485816e-06,
      "loss": 0.439,
      "step": 3368
    },
    {
      "epoch": 0.2110274197842121,
      "grad_norm": 0.8531031008912388,
      "learning_rate": 9.164905570982299e-06,
      "loss": 0.4701,
      "step": 3369
    },
    {
      "epoch": 0.21109005778355439,
      "grad_norm": 0.8580918254940981,
      "learning_rate": 9.164344217050494e-06,
      "loss": 0.4499,
      "step": 3370
    },
    {
      "epoch": 0.2111526957828967,
      "grad_norm": 0.7994863909437608,
      "learning_rate": 9.16378269171351e-06,
      "loss": 0.422,
      "step": 3371
    },
    {
      "epoch": 0.211215333782239,
      "grad_norm": 0.8427078754987986,
      "learning_rate": 9.163220994994455e-06,
      "loss": 0.402,
      "step": 3372
    },
    {
      "epoch": 0.2112779717815813,
      "grad_norm": 0.8410486643059157,
      "learning_rate": 9.162659126916453e-06,
      "loss": 0.4258,
      "step": 3373
    },
    {
      "epoch": 0.2113406097809236,
      "grad_norm": 0.6816545654416111,
      "learning_rate": 9.162097087502627e-06,
      "loss": 0.4902,
      "step": 3374
    },
    {
      "epoch": 0.2114032477802659,
      "grad_norm": 0.8111489812756513,
      "learning_rate": 9.161534876776114e-06,
      "loss": 0.415,
      "step": 3375
    },
    {
      "epoch": 0.2114658857796082,
      "grad_norm": 0.8102885315253535,
      "learning_rate": 9.160972494760052e-06,
      "loss": 0.4018,
      "step": 3376
    },
    {
      "epoch": 0.2115285237789505,
      "grad_norm": 0.9098676367842284,
      "learning_rate": 9.160409941477591e-06,
      "loss": 0.4523,
      "step": 3377
    },
    {
      "epoch": 0.2115911617782928,
      "grad_norm": 0.8730323812458546,
      "learning_rate": 9.159847216951882e-06,
      "loss": 0.4845,
      "step": 3378
    },
    {
      "epoch": 0.2116537997776351,
      "grad_norm": 0.8724320333098525,
      "learning_rate": 9.159284321206091e-06,
      "loss": 0.4708,
      "step": 3379
    },
    {
      "epoch": 0.2117164377769774,
      "grad_norm": 0.9169340995621772,
      "learning_rate": 9.158721254263387e-06,
      "loss": 0.477,
      "step": 3380
    },
    {
      "epoch": 0.21177907577631971,
      "grad_norm": 0.8337585608992336,
      "learning_rate": 9.158158016146942e-06,
      "loss": 0.3943,
      "step": 3381
    },
    {
      "epoch": 0.211841713775662,
      "grad_norm": 0.7313403140342521,
      "learning_rate": 9.157594606879942e-06,
      "loss": 0.4842,
      "step": 3382
    },
    {
      "epoch": 0.2119043517750043,
      "grad_norm": 0.8831134488675165,
      "learning_rate": 9.157031026485577e-06,
      "loss": 0.4753,
      "step": 3383
    },
    {
      "epoch": 0.2119669897743466,
      "grad_norm": 0.8362287423138655,
      "learning_rate": 9.156467274987042e-06,
      "loss": 0.4163,
      "step": 3384
    },
    {
      "epoch": 0.21202962777368892,
      "grad_norm": 0.8407491728843567,
      "learning_rate": 9.155903352407542e-06,
      "loss": 0.4053,
      "step": 3385
    },
    {
      "epoch": 0.2120922657730312,
      "grad_norm": 0.8798734287286547,
      "learning_rate": 9.15533925877029e-06,
      "loss": 0.4131,
      "step": 3386
    },
    {
      "epoch": 0.2121549037723735,
      "grad_norm": 0.9198286098846054,
      "learning_rate": 9.154774994098504e-06,
      "loss": 0.4418,
      "step": 3387
    },
    {
      "epoch": 0.2122175417717158,
      "grad_norm": 0.8261926059944933,
      "learning_rate": 9.154210558415407e-06,
      "loss": 0.4065,
      "step": 3388
    },
    {
      "epoch": 0.21228017977105812,
      "grad_norm": 0.8595585050570093,
      "learning_rate": 9.153645951744232e-06,
      "loss": 0.4509,
      "step": 3389
    },
    {
      "epoch": 0.21234281777040043,
      "grad_norm": 0.8950249323086904,
      "learning_rate": 9.153081174108219e-06,
      "loss": 0.4481,
      "step": 3390
    },
    {
      "epoch": 0.2124054557697427,
      "grad_norm": 0.7710459287174753,
      "learning_rate": 9.152516225530613e-06,
      "loss": 0.3856,
      "step": 3391
    },
    {
      "epoch": 0.21246809376908501,
      "grad_norm": 0.8487066930226548,
      "learning_rate": 9.151951106034668e-06,
      "loss": 0.3947,
      "step": 3392
    },
    {
      "epoch": 0.21253073176842732,
      "grad_norm": 0.8615629119560242,
      "learning_rate": 9.151385815643647e-06,
      "loss": 0.4332,
      "step": 3393
    },
    {
      "epoch": 0.21259336976776963,
      "grad_norm": 0.8957225434087194,
      "learning_rate": 9.150820354380815e-06,
      "loss": 0.4696,
      "step": 3394
    },
    {
      "epoch": 0.2126560077671119,
      "grad_norm": 0.7660150640114933,
      "learning_rate": 9.150254722269446e-06,
      "loss": 0.4167,
      "step": 3395
    },
    {
      "epoch": 0.21271864576645422,
      "grad_norm": 0.8930365066662044,
      "learning_rate": 9.149688919332822e-06,
      "loss": 0.392,
      "step": 3396
    },
    {
      "epoch": 0.21278128376579653,
      "grad_norm": 0.8674537154207222,
      "learning_rate": 9.149122945594233e-06,
      "loss": 0.438,
      "step": 3397
    },
    {
      "epoch": 0.21284392176513883,
      "grad_norm": 0.7604979436586703,
      "learning_rate": 9.148556801076974e-06,
      "loss": 0.3891,
      "step": 3398
    },
    {
      "epoch": 0.2129065597644811,
      "grad_norm": 0.8342086321245238,
      "learning_rate": 9.147990485804345e-06,
      "loss": 0.402,
      "step": 3399
    },
    {
      "epoch": 0.21296919776382342,
      "grad_norm": 0.8426458998916536,
      "learning_rate": 9.147423999799659e-06,
      "loss": 0.4443,
      "step": 3400
    },
    {
      "epoch": 0.21303183576316573,
      "grad_norm": 0.870501162670089,
      "learning_rate": 9.14685734308623e-06,
      "loss": 0.4013,
      "step": 3401
    },
    {
      "epoch": 0.21309447376250804,
      "grad_norm": 0.8717720572077022,
      "learning_rate": 9.146290515687384e-06,
      "loss": 0.4312,
      "step": 3402
    },
    {
      "epoch": 0.21315711176185032,
      "grad_norm": 0.8758666299979452,
      "learning_rate": 9.14572351762645e-06,
      "loss": 0.4481,
      "step": 3403
    },
    {
      "epoch": 0.21321974976119262,
      "grad_norm": 0.8267187233206539,
      "learning_rate": 9.145156348926767e-06,
      "loss": 0.427,
      "step": 3404
    },
    {
      "epoch": 0.21328238776053493,
      "grad_norm": 0.8835038883106409,
      "learning_rate": 9.144589009611681e-06,
      "loss": 0.4115,
      "step": 3405
    },
    {
      "epoch": 0.21334502575987724,
      "grad_norm": 0.8807430083547967,
      "learning_rate": 9.14402149970454e-06,
      "loss": 0.4046,
      "step": 3406
    },
    {
      "epoch": 0.21340766375921952,
      "grad_norm": 0.8671787824107875,
      "learning_rate": 9.143453819228706e-06,
      "loss": 0.4238,
      "step": 3407
    },
    {
      "epoch": 0.21347030175856183,
      "grad_norm": 0.8902092699295743,
      "learning_rate": 9.142885968207542e-06,
      "loss": 0.4296,
      "step": 3408
    },
    {
      "epoch": 0.21353293975790413,
      "grad_norm": 0.8710251170218991,
      "learning_rate": 9.142317946664426e-06,
      "loss": 0.4681,
      "step": 3409
    },
    {
      "epoch": 0.21359557775724644,
      "grad_norm": 0.8458930132793274,
      "learning_rate": 9.14174975462273e-06,
      "loss": 0.427,
      "step": 3410
    },
    {
      "epoch": 0.21365821575658872,
      "grad_norm": 0.861633021515958,
      "learning_rate": 9.141181392105849e-06,
      "loss": 0.4373,
      "step": 3411
    },
    {
      "epoch": 0.21372085375593103,
      "grad_norm": 0.8821957298733285,
      "learning_rate": 9.140612859137173e-06,
      "loss": 0.4364,
      "step": 3412
    },
    {
      "epoch": 0.21378349175527334,
      "grad_norm": 0.8493858490604816,
      "learning_rate": 9.140044155740102e-06,
      "loss": 0.4182,
      "step": 3413
    },
    {
      "epoch": 0.21384612975461564,
      "grad_norm": 0.8059782421793328,
      "learning_rate": 9.139475281938043e-06,
      "loss": 0.4341,
      "step": 3414
    },
    {
      "epoch": 0.21390876775395795,
      "grad_norm": 0.9382274459637886,
      "learning_rate": 9.138906237754416e-06,
      "loss": 0.4907,
      "step": 3415
    },
    {
      "epoch": 0.21397140575330023,
      "grad_norm": 0.8336303229672425,
      "learning_rate": 9.13833702321264e-06,
      "loss": 0.417,
      "step": 3416
    },
    {
      "epoch": 0.21403404375264254,
      "grad_norm": 1.0127760028686827,
      "learning_rate": 9.137767638336144e-06,
      "loss": 0.4474,
      "step": 3417
    },
    {
      "epoch": 0.21409668175198485,
      "grad_norm": 0.8295996979571515,
      "learning_rate": 9.137198083148363e-06,
      "loss": 0.4415,
      "step": 3418
    },
    {
      "epoch": 0.21415931975132715,
      "grad_norm": 0.8187180235820053,
      "learning_rate": 9.136628357672742e-06,
      "loss": 0.4258,
      "step": 3419
    },
    {
      "epoch": 0.21422195775066943,
      "grad_norm": 0.8707674465480058,
      "learning_rate": 9.136058461932729e-06,
      "loss": 0.4446,
      "step": 3420
    },
    {
      "epoch": 0.21428459575001174,
      "grad_norm": 0.8177326928670063,
      "learning_rate": 9.135488395951782e-06,
      "loss": 0.414,
      "step": 3421
    },
    {
      "epoch": 0.21434723374935405,
      "grad_norm": 0.9291643140778464,
      "learning_rate": 9.134918159753365e-06,
      "loss": 0.4578,
      "step": 3422
    },
    {
      "epoch": 0.21440987174869636,
      "grad_norm": 0.8431285196978607,
      "learning_rate": 9.134347753360951e-06,
      "loss": 0.4595,
      "step": 3423
    },
    {
      "epoch": 0.21447250974803864,
      "grad_norm": 0.9038861576905716,
      "learning_rate": 9.133777176798013e-06,
      "loss": 0.4235,
      "step": 3424
    },
    {
      "epoch": 0.21453514774738094,
      "grad_norm": 0.8534990208814436,
      "learning_rate": 9.133206430088042e-06,
      "loss": 0.4517,
      "step": 3425
    },
    {
      "epoch": 0.21459778574672325,
      "grad_norm": 0.8640343961304656,
      "learning_rate": 9.132635513254526e-06,
      "loss": 0.4212,
      "step": 3426
    },
    {
      "epoch": 0.21466042374606556,
      "grad_norm": 0.8837559071623254,
      "learning_rate": 9.132064426320964e-06,
      "loss": 0.4505,
      "step": 3427
    },
    {
      "epoch": 0.21472306174540784,
      "grad_norm": 0.8476600762400688,
      "learning_rate": 9.131493169310866e-06,
      "loss": 0.435,
      "step": 3428
    },
    {
      "epoch": 0.21478569974475015,
      "grad_norm": 0.7953202400025011,
      "learning_rate": 9.13092174224774e-06,
      "loss": 0.3938,
      "step": 3429
    },
    {
      "epoch": 0.21484833774409245,
      "grad_norm": 0.8536921668703055,
      "learning_rate": 9.13035014515511e-06,
      "loss": 0.4023,
      "step": 3430
    },
    {
      "epoch": 0.21491097574343476,
      "grad_norm": 0.8312522508748729,
      "learning_rate": 9.1297783780565e-06,
      "loss": 0.4085,
      "step": 3431
    },
    {
      "epoch": 0.21497361374277704,
      "grad_norm": 0.8346208927733847,
      "learning_rate": 9.129206440975446e-06,
      "loss": 0.4252,
      "step": 3432
    },
    {
      "epoch": 0.21503625174211935,
      "grad_norm": 0.9240573651414102,
      "learning_rate": 9.128634333935488e-06,
      "loss": 0.4665,
      "step": 3433
    },
    {
      "epoch": 0.21509888974146166,
      "grad_norm": 0.8722491223873804,
      "learning_rate": 9.128062056960175e-06,
      "loss": 0.4224,
      "step": 3434
    },
    {
      "epoch": 0.21516152774080397,
      "grad_norm": 0.8557705129082309,
      "learning_rate": 9.127489610073061e-06,
      "loss": 0.4455,
      "step": 3435
    },
    {
      "epoch": 0.21522416574014627,
      "grad_norm": 0.9396296021169515,
      "learning_rate": 9.12691699329771e-06,
      "loss": 0.4608,
      "step": 3436
    },
    {
      "epoch": 0.21528680373948855,
      "grad_norm": 0.8594037577770807,
      "learning_rate": 9.126344206657689e-06,
      "loss": 0.4306,
      "step": 3437
    },
    {
      "epoch": 0.21534944173883086,
      "grad_norm": 0.8457043378197814,
      "learning_rate": 9.125771250176572e-06,
      "loss": 0.4422,
      "step": 3438
    },
    {
      "epoch": 0.21541207973817317,
      "grad_norm": 0.8962233555005542,
      "learning_rate": 9.125198123877947e-06,
      "loss": 0.4551,
      "step": 3439
    },
    {
      "epoch": 0.21547471773751548,
      "grad_norm": 0.7780312415204476,
      "learning_rate": 9.124624827785402e-06,
      "loss": 0.3997,
      "step": 3440
    },
    {
      "epoch": 0.21553735573685776,
      "grad_norm": 0.9376612362411576,
      "learning_rate": 9.124051361922532e-06,
      "loss": 0.4518,
      "step": 3441
    },
    {
      "epoch": 0.21559999373620006,
      "grad_norm": 0.7554140446957337,
      "learning_rate": 9.123477726312943e-06,
      "loss": 0.422,
      "step": 3442
    },
    {
      "epoch": 0.21566263173554237,
      "grad_norm": 0.682904661201874,
      "learning_rate": 9.122903920980247e-06,
      "loss": 0.4793,
      "step": 3443
    },
    {
      "epoch": 0.21572526973488468,
      "grad_norm": 0.8542650851805832,
      "learning_rate": 9.122329945948059e-06,
      "loss": 0.3996,
      "step": 3444
    },
    {
      "epoch": 0.21578790773422696,
      "grad_norm": 0.8456624103392347,
      "learning_rate": 9.121755801240005e-06,
      "loss": 0.4354,
      "step": 3445
    },
    {
      "epoch": 0.21585054573356927,
      "grad_norm": 0.840919459859079,
      "learning_rate": 9.121181486879718e-06,
      "loss": 0.4513,
      "step": 3446
    },
    {
      "epoch": 0.21591318373291157,
      "grad_norm": 0.80663331109863,
      "learning_rate": 9.120607002890835e-06,
      "loss": 0.4202,
      "step": 3447
    },
    {
      "epoch": 0.21597582173225388,
      "grad_norm": 0.79542519780374,
      "learning_rate": 9.120032349297004e-06,
      "loss": 0.4479,
      "step": 3448
    },
    {
      "epoch": 0.21603845973159616,
      "grad_norm": 0.8111201591089079,
      "learning_rate": 9.119457526121876e-06,
      "loss": 0.4051,
      "step": 3449
    },
    {
      "epoch": 0.21610109773093847,
      "grad_norm": 0.8347062592676378,
      "learning_rate": 9.118882533389111e-06,
      "loss": 0.433,
      "step": 3450
    },
    {
      "epoch": 0.21616373573028078,
      "grad_norm": 0.8467809575940478,
      "learning_rate": 9.11830737112238e-06,
      "loss": 0.4349,
      "step": 3451
    },
    {
      "epoch": 0.21622637372962308,
      "grad_norm": 0.8027466730788757,
      "learning_rate": 9.117732039345348e-06,
      "loss": 0.4355,
      "step": 3452
    },
    {
      "epoch": 0.21628901172896536,
      "grad_norm": 0.8517242027847985,
      "learning_rate": 9.117156538081704e-06,
      "loss": 0.4204,
      "step": 3453
    },
    {
      "epoch": 0.21635164972830767,
      "grad_norm": 0.8424873794675967,
      "learning_rate": 9.116580867355133e-06,
      "loss": 0.3881,
      "step": 3454
    },
    {
      "epoch": 0.21641428772764998,
      "grad_norm": 0.8704989857756612,
      "learning_rate": 9.116005027189328e-06,
      "loss": 0.4474,
      "step": 3455
    },
    {
      "epoch": 0.2164769257269923,
      "grad_norm": 0.8030020603232793,
      "learning_rate": 9.115429017607992e-06,
      "loss": 0.4217,
      "step": 3456
    },
    {
      "epoch": 0.2165395637263346,
      "grad_norm": 0.8410652944040922,
      "learning_rate": 9.114852838634836e-06,
      "loss": 0.4435,
      "step": 3457
    },
    {
      "epoch": 0.21660220172567687,
      "grad_norm": 0.8946209317534654,
      "learning_rate": 9.114276490293572e-06,
      "loss": 0.4431,
      "step": 3458
    },
    {
      "epoch": 0.21666483972501918,
      "grad_norm": 0.8453618464522995,
      "learning_rate": 9.113699972607924e-06,
      "loss": 0.3895,
      "step": 3459
    },
    {
      "epoch": 0.2167274777243615,
      "grad_norm": 0.816534635989571,
      "learning_rate": 9.113123285601621e-06,
      "loss": 0.4081,
      "step": 3460
    },
    {
      "epoch": 0.2167901157237038,
      "grad_norm": 0.8076378981435558,
      "learning_rate": 9.1125464292984e-06,
      "loss": 0.4187,
      "step": 3461
    },
    {
      "epoch": 0.21685275372304608,
      "grad_norm": 0.9298844050294148,
      "learning_rate": 9.111969403722007e-06,
      "loss": 0.4238,
      "step": 3462
    },
    {
      "epoch": 0.21691539172238838,
      "grad_norm": 0.8132141821312064,
      "learning_rate": 9.111392208896187e-06,
      "loss": 0.4209,
      "step": 3463
    },
    {
      "epoch": 0.2169780297217307,
      "grad_norm": 0.8333801230299253,
      "learning_rate": 9.110814844844703e-06,
      "loss": 0.4486,
      "step": 3464
    },
    {
      "epoch": 0.217040667721073,
      "grad_norm": 0.9083940731014362,
      "learning_rate": 9.110237311591316e-06,
      "loss": 0.4409,
      "step": 3465
    },
    {
      "epoch": 0.21710330572041528,
      "grad_norm": 0.742053958746206,
      "learning_rate": 9.109659609159798e-06,
      "loss": 0.485,
      "step": 3466
    },
    {
      "epoch": 0.2171659437197576,
      "grad_norm": 0.8054236064154258,
      "learning_rate": 9.109081737573927e-06,
      "loss": 0.4052,
      "step": 3467
    },
    {
      "epoch": 0.2172285817190999,
      "grad_norm": 0.8442103222480151,
      "learning_rate": 9.108503696857491e-06,
      "loss": 0.4381,
      "step": 3468
    },
    {
      "epoch": 0.2172912197184422,
      "grad_norm": 0.8034897770058345,
      "learning_rate": 9.10792548703428e-06,
      "loss": 0.4086,
      "step": 3469
    },
    {
      "epoch": 0.21735385771778448,
      "grad_norm": 0.8625131783662776,
      "learning_rate": 9.107347108128093e-06,
      "loss": 0.4306,
      "step": 3470
    },
    {
      "epoch": 0.2174164957171268,
      "grad_norm": 0.8534591226171914,
      "learning_rate": 9.106768560162737e-06,
      "loss": 0.4369,
      "step": 3471
    },
    {
      "epoch": 0.2174791337164691,
      "grad_norm": 0.7169922369454246,
      "learning_rate": 9.106189843162023e-06,
      "loss": 0.4826,
      "step": 3472
    },
    {
      "epoch": 0.2175417717158114,
      "grad_norm": 0.8444900854642773,
      "learning_rate": 9.105610957149775e-06,
      "loss": 0.4386,
      "step": 3473
    },
    {
      "epoch": 0.21760440971515368,
      "grad_norm": 0.9066087267780036,
      "learning_rate": 9.105031902149816e-06,
      "loss": 0.4266,
      "step": 3474
    },
    {
      "epoch": 0.217667047714496,
      "grad_norm": 0.8088454150074316,
      "learning_rate": 9.104452678185985e-06,
      "loss": 0.4044,
      "step": 3475
    },
    {
      "epoch": 0.2177296857138383,
      "grad_norm": 0.8369484351419599,
      "learning_rate": 9.103873285282117e-06,
      "loss": 0.4466,
      "step": 3476
    },
    {
      "epoch": 0.2177923237131806,
      "grad_norm": 0.6720668789350304,
      "learning_rate": 9.103293723462064e-06,
      "loss": 0.4658,
      "step": 3477
    },
    {
      "epoch": 0.21785496171252292,
      "grad_norm": 0.8278959808920889,
      "learning_rate": 9.10271399274968e-06,
      "loss": 0.4747,
      "step": 3478
    },
    {
      "epoch": 0.2179175997118652,
      "grad_norm": 0.8807131223641256,
      "learning_rate": 9.102134093168827e-06,
      "loss": 0.4489,
      "step": 3479
    },
    {
      "epoch": 0.2179802377112075,
      "grad_norm": 0.8640073296424367,
      "learning_rate": 9.101554024743373e-06,
      "loss": 0.4022,
      "step": 3480
    },
    {
      "epoch": 0.2180428757105498,
      "grad_norm": 0.823051949220351,
      "learning_rate": 9.100973787497193e-06,
      "loss": 0.4376,
      "step": 3481
    },
    {
      "epoch": 0.21810551370989212,
      "grad_norm": 0.7331856074552733,
      "learning_rate": 9.10039338145417e-06,
      "loss": 0.3878,
      "step": 3482
    },
    {
      "epoch": 0.2181681517092344,
      "grad_norm": 0.8782470633456324,
      "learning_rate": 9.099812806638197e-06,
      "loss": 0.4036,
      "step": 3483
    },
    {
      "epoch": 0.2182307897085767,
      "grad_norm": 0.8555190935855986,
      "learning_rate": 9.099232063073167e-06,
      "loss": 0.408,
      "step": 3484
    },
    {
      "epoch": 0.218293427707919,
      "grad_norm": 0.8261633707583838,
      "learning_rate": 9.098651150782985e-06,
      "loss": 0.4606,
      "step": 3485
    },
    {
      "epoch": 0.21835606570726132,
      "grad_norm": 0.7966133572348397,
      "learning_rate": 9.098070069791559e-06,
      "loss": 0.4417,
      "step": 3486
    },
    {
      "epoch": 0.2184187037066036,
      "grad_norm": 0.7574391340404741,
      "learning_rate": 9.097488820122808e-06,
      "loss": 0.3865,
      "step": 3487
    },
    {
      "epoch": 0.2184813417059459,
      "grad_norm": 0.9170723547892927,
      "learning_rate": 9.096907401800658e-06,
      "loss": 0.4403,
      "step": 3488
    },
    {
      "epoch": 0.21854397970528822,
      "grad_norm": 0.8594080587540511,
      "learning_rate": 9.096325814849039e-06,
      "loss": 0.4296,
      "step": 3489
    },
    {
      "epoch": 0.21860661770463052,
      "grad_norm": 0.7661977675186527,
      "learning_rate": 9.09574405929189e-06,
      "loss": 0.4744,
      "step": 3490
    },
    {
      "epoch": 0.2186692557039728,
      "grad_norm": 0.8183689386478324,
      "learning_rate": 9.095162135153153e-06,
      "loss": 0.4354,
      "step": 3491
    },
    {
      "epoch": 0.2187318937033151,
      "grad_norm": 0.8428987961649365,
      "learning_rate": 9.094580042456784e-06,
      "loss": 0.4168,
      "step": 3492
    },
    {
      "epoch": 0.21879453170265742,
      "grad_norm": 0.7865785162284066,
      "learning_rate": 9.09399778122674e-06,
      "loss": 0.3909,
      "step": 3493
    },
    {
      "epoch": 0.21885716970199973,
      "grad_norm": 0.8349117376245395,
      "learning_rate": 9.09341535148699e-06,
      "loss": 0.3959,
      "step": 3494
    },
    {
      "epoch": 0.218919807701342,
      "grad_norm": 0.842257947610258,
      "learning_rate": 9.0928327532615e-06,
      "loss": 0.4185,
      "step": 3495
    },
    {
      "epoch": 0.21898244570068431,
      "grad_norm": 0.8643290731523918,
      "learning_rate": 9.092249986574258e-06,
      "loss": 0.4149,
      "step": 3496
    },
    {
      "epoch": 0.21904508370002662,
      "grad_norm": 0.8540558447650581,
      "learning_rate": 9.091667051449245e-06,
      "loss": 0.4211,
      "step": 3497
    },
    {
      "epoch": 0.21910772169936893,
      "grad_norm": 0.8759626452252091,
      "learning_rate": 9.091083947910455e-06,
      "loss": 0.422,
      "step": 3498
    },
    {
      "epoch": 0.21917035969871124,
      "grad_norm": 0.9130135007877531,
      "learning_rate": 9.090500675981893e-06,
      "loss": 0.4451,
      "step": 3499
    },
    {
      "epoch": 0.21923299769805352,
      "grad_norm": 0.8420195345887747,
      "learning_rate": 9.089917235687564e-06,
      "loss": 0.4122,
      "step": 3500
    },
    {
      "epoch": 0.21929563569739582,
      "grad_norm": 0.8485558215500045,
      "learning_rate": 9.08933362705148e-06,
      "loss": 0.4266,
      "step": 3501
    },
    {
      "epoch": 0.21935827369673813,
      "grad_norm": 0.8368509502320544,
      "learning_rate": 9.088749850097667e-06,
      "loss": 0.432,
      "step": 3502
    },
    {
      "epoch": 0.21942091169608044,
      "grad_norm": 1.1342222713436856,
      "learning_rate": 9.088165904850152e-06,
      "loss": 0.4938,
      "step": 3503
    },
    {
      "epoch": 0.21948354969542272,
      "grad_norm": 0.8719998307040342,
      "learning_rate": 9.087581791332968e-06,
      "loss": 0.4458,
      "step": 3504
    },
    {
      "epoch": 0.21954618769476503,
      "grad_norm": 0.8392854343309148,
      "learning_rate": 9.086997509570159e-06,
      "loss": 0.4251,
      "step": 3505
    },
    {
      "epoch": 0.21960882569410733,
      "grad_norm": 0.7877223605912422,
      "learning_rate": 9.086413059585775e-06,
      "loss": 0.3917,
      "step": 3506
    },
    {
      "epoch": 0.21967146369344964,
      "grad_norm": 0.8994424070247724,
      "learning_rate": 9.085828441403868e-06,
      "loss": 0.4455,
      "step": 3507
    },
    {
      "epoch": 0.21973410169279192,
      "grad_norm": 0.8583773951892993,
      "learning_rate": 9.085243655048507e-06,
      "loss": 0.4365,
      "step": 3508
    },
    {
      "epoch": 0.21979673969213423,
      "grad_norm": 0.838636095076454,
      "learning_rate": 9.084658700543759e-06,
      "loss": 0.404,
      "step": 3509
    },
    {
      "epoch": 0.21985937769147654,
      "grad_norm": 0.7854355973859403,
      "learning_rate": 9.0840735779137e-06,
      "loss": 0.452,
      "step": 3510
    },
    {
      "epoch": 0.21992201569081885,
      "grad_norm": 0.8541503412710453,
      "learning_rate": 9.083488287182414e-06,
      "loss": 0.4725,
      "step": 3511
    },
    {
      "epoch": 0.21998465369016112,
      "grad_norm": 0.8766287396866455,
      "learning_rate": 9.082902828373992e-06,
      "loss": 0.443,
      "step": 3512
    },
    {
      "epoch": 0.22004729168950343,
      "grad_norm": 0.7896997642543903,
      "learning_rate": 9.082317201512532e-06,
      "loss": 0.4657,
      "step": 3513
    },
    {
      "epoch": 0.22010992968884574,
      "grad_norm": 0.7540936268555365,
      "learning_rate": 9.081731406622139e-06,
      "loss": 0.483,
      "step": 3514
    },
    {
      "epoch": 0.22017256768818805,
      "grad_norm": 0.8387532595957249,
      "learning_rate": 9.081145443726922e-06,
      "loss": 0.3781,
      "step": 3515
    },
    {
      "epoch": 0.22023520568753033,
      "grad_norm": 0.8250549216746454,
      "learning_rate": 9.080559312851003e-06,
      "loss": 0.434,
      "step": 3516
    },
    {
      "epoch": 0.22029784368687264,
      "grad_norm": 0.8189301924777083,
      "learning_rate": 9.079973014018504e-06,
      "loss": 0.3675,
      "step": 3517
    },
    {
      "epoch": 0.22036048168621494,
      "grad_norm": 0.8081565401916514,
      "learning_rate": 9.079386547253559e-06,
      "loss": 0.4546,
      "step": 3518
    },
    {
      "epoch": 0.22042311968555725,
      "grad_norm": 0.8933957294437657,
      "learning_rate": 9.078799912580305e-06,
      "loss": 0.4165,
      "step": 3519
    },
    {
      "epoch": 0.22048575768489956,
      "grad_norm": 0.8678317704148982,
      "learning_rate": 9.078213110022893e-06,
      "loss": 0.4355,
      "step": 3520
    },
    {
      "epoch": 0.22054839568424184,
      "grad_norm": 0.8913713955263155,
      "learning_rate": 9.077626139605469e-06,
      "loss": 0.4442,
      "step": 3521
    },
    {
      "epoch": 0.22061103368358415,
      "grad_norm": 0.7965550493765535,
      "learning_rate": 9.077039001352198e-06,
      "loss": 0.364,
      "step": 3522
    },
    {
      "epoch": 0.22067367168292645,
      "grad_norm": 0.7786881766915114,
      "learning_rate": 9.076451695287245e-06,
      "loss": 0.4072,
      "step": 3523
    },
    {
      "epoch": 0.22073630968226876,
      "grad_norm": 0.8122057398501212,
      "learning_rate": 9.075864221434782e-06,
      "loss": 0.4067,
      "step": 3524
    },
    {
      "epoch": 0.22079894768161104,
      "grad_norm": 0.8262303010894142,
      "learning_rate": 9.075276579818993e-06,
      "loss": 0.4323,
      "step": 3525
    },
    {
      "epoch": 0.22086158568095335,
      "grad_norm": 0.857947800928926,
      "learning_rate": 9.074688770464063e-06,
      "loss": 0.4224,
      "step": 3526
    },
    {
      "epoch": 0.22092422368029566,
      "grad_norm": 0.8601311900350356,
      "learning_rate": 9.074100793394188e-06,
      "loss": 0.4576,
      "step": 3527
    },
    {
      "epoch": 0.22098686167963796,
      "grad_norm": 0.869531032885354,
      "learning_rate": 9.073512648633565e-06,
      "loss": 0.4519,
      "step": 3528
    },
    {
      "epoch": 0.22104949967898024,
      "grad_norm": 0.7917673211552705,
      "learning_rate": 9.072924336206408e-06,
      "loss": 0.4222,
      "step": 3529
    },
    {
      "epoch": 0.22111213767832255,
      "grad_norm": 0.8132525170543327,
      "learning_rate": 9.072335856136929e-06,
      "loss": 0.4412,
      "step": 3530
    },
    {
      "epoch": 0.22117477567766486,
      "grad_norm": 0.8183050254380189,
      "learning_rate": 9.07174720844935e-06,
      "loss": 0.4207,
      "step": 3531
    },
    {
      "epoch": 0.22123741367700717,
      "grad_norm": 0.8137215100378563,
      "learning_rate": 9.0711583931679e-06,
      "loss": 0.4394,
      "step": 3532
    },
    {
      "epoch": 0.22130005167634945,
      "grad_norm": 0.8338738408115987,
      "learning_rate": 9.070569410316815e-06,
      "loss": 0.4221,
      "step": 3533
    },
    {
      "epoch": 0.22136268967569175,
      "grad_norm": 0.8675327922517426,
      "learning_rate": 9.069980259920337e-06,
      "loss": 0.445,
      "step": 3534
    },
    {
      "epoch": 0.22142532767503406,
      "grad_norm": 0.7715128703811805,
      "learning_rate": 9.069390942002718e-06,
      "loss": 0.4032,
      "step": 3535
    },
    {
      "epoch": 0.22148796567437637,
      "grad_norm": 0.7762014601525519,
      "learning_rate": 9.06880145658821e-06,
      "loss": 0.4259,
      "step": 3536
    },
    {
      "epoch": 0.22155060367371865,
      "grad_norm": 0.8532224722321907,
      "learning_rate": 9.068211803701079e-06,
      "loss": 0.4657,
      "step": 3537
    },
    {
      "epoch": 0.22161324167306096,
      "grad_norm": 0.8792214686957324,
      "learning_rate": 9.067621983365597e-06,
      "loss": 0.4676,
      "step": 3538
    },
    {
      "epoch": 0.22167587967240326,
      "grad_norm": 0.7652470054453844,
      "learning_rate": 9.067031995606038e-06,
      "loss": 0.4346,
      "step": 3539
    },
    {
      "epoch": 0.22173851767174557,
      "grad_norm": 0.8323376523279931,
      "learning_rate": 9.066441840446687e-06,
      "loss": 0.4431,
      "step": 3540
    },
    {
      "epoch": 0.22180115567108785,
      "grad_norm": 0.9189318788021077,
      "learning_rate": 9.065851517911836e-06,
      "loss": 0.5086,
      "step": 3541
    },
    {
      "epoch": 0.22186379367043016,
      "grad_norm": 0.8732893366796302,
      "learning_rate": 9.065261028025782e-06,
      "loss": 0.4096,
      "step": 3542
    },
    {
      "epoch": 0.22192643166977247,
      "grad_norm": 0.8732373234277946,
      "learning_rate": 9.064670370812828e-06,
      "loss": 0.4619,
      "step": 3543
    },
    {
      "epoch": 0.22198906966911477,
      "grad_norm": 0.8519115218465824,
      "learning_rate": 9.064079546297289e-06,
      "loss": 0.4122,
      "step": 3544
    },
    {
      "epoch": 0.22205170766845708,
      "grad_norm": 0.8896210186449152,
      "learning_rate": 9.06348855450348e-06,
      "loss": 0.4028,
      "step": 3545
    },
    {
      "epoch": 0.22211434566779936,
      "grad_norm": 0.8770545894921812,
      "learning_rate": 9.062897395455728e-06,
      "loss": 0.3954,
      "step": 3546
    },
    {
      "epoch": 0.22217698366714167,
      "grad_norm": 0.917849304025567,
      "learning_rate": 9.062306069178366e-06,
      "loss": 0.4602,
      "step": 3547
    },
    {
      "epoch": 0.22223962166648398,
      "grad_norm": 0.8670614496144201,
      "learning_rate": 9.061714575695731e-06,
      "loss": 0.4034,
      "step": 3548
    },
    {
      "epoch": 0.22230225966582629,
      "grad_norm": 0.8334009102478156,
      "learning_rate": 9.061122915032172e-06,
      "loss": 0.434,
      "step": 3549
    },
    {
      "epoch": 0.22236489766516856,
      "grad_norm": 0.8870824317082204,
      "learning_rate": 9.060531087212039e-06,
      "loss": 0.4356,
      "step": 3550
    },
    {
      "epoch": 0.22242753566451087,
      "grad_norm": 0.8046687835409454,
      "learning_rate": 9.05993909225969e-06,
      "loss": 0.3884,
      "step": 3551
    },
    {
      "epoch": 0.22249017366385318,
      "grad_norm": 0.8805309782215427,
      "learning_rate": 9.0593469301995e-06,
      "loss": 0.4826,
      "step": 3552
    },
    {
      "epoch": 0.2225528116631955,
      "grad_norm": 0.9151182408337164,
      "learning_rate": 9.058754601055833e-06,
      "loss": 0.5034,
      "step": 3553
    },
    {
      "epoch": 0.22261544966253777,
      "grad_norm": 0.8624281356703499,
      "learning_rate": 9.058162104853074e-06,
      "loss": 0.4481,
      "step": 3554
    },
    {
      "epoch": 0.22267808766188008,
      "grad_norm": 0.8214631364137688,
      "learning_rate": 9.05756944161561e-06,
      "loss": 0.4112,
      "step": 3555
    },
    {
      "epoch": 0.22274072566122238,
      "grad_norm": 0.8769581053994999,
      "learning_rate": 9.056976611367833e-06,
      "loss": 0.4549,
      "step": 3556
    },
    {
      "epoch": 0.2228033636605647,
      "grad_norm": 0.8615884019885124,
      "learning_rate": 9.056383614134147e-06,
      "loss": 0.4164,
      "step": 3557
    },
    {
      "epoch": 0.22286600165990697,
      "grad_norm": 0.6767465379553247,
      "learning_rate": 9.05579044993896e-06,
      "loss": 0.4799,
      "step": 3558
    },
    {
      "epoch": 0.22292863965924928,
      "grad_norm": 0.9212694619652625,
      "learning_rate": 9.055197118806684e-06,
      "loss": 0.4557,
      "step": 3559
    },
    {
      "epoch": 0.22299127765859159,
      "grad_norm": 0.8131195213615494,
      "learning_rate": 9.054603620761742e-06,
      "loss": 0.4528,
      "step": 3560
    },
    {
      "epoch": 0.2230539156579339,
      "grad_norm": 0.7280578765614866,
      "learning_rate": 9.054009955828562e-06,
      "loss": 0.4892,
      "step": 3561
    },
    {
      "epoch": 0.22311655365727617,
      "grad_norm": 0.8800246093595495,
      "learning_rate": 9.053416124031579e-06,
      "loss": 0.4505,
      "step": 3562
    },
    {
      "epoch": 0.22317919165661848,
      "grad_norm": 0.8456636246968394,
      "learning_rate": 9.052822125395238e-06,
      "loss": 0.4387,
      "step": 3563
    },
    {
      "epoch": 0.2232418296559608,
      "grad_norm": 0.8141798536308668,
      "learning_rate": 9.052227959943986e-06,
      "loss": 0.3947,
      "step": 3564
    },
    {
      "epoch": 0.2233044676553031,
      "grad_norm": 0.9817574131738336,
      "learning_rate": 9.051633627702278e-06,
      "loss": 0.4268,
      "step": 3565
    },
    {
      "epoch": 0.2233671056546454,
      "grad_norm": 0.8393628019501299,
      "learning_rate": 9.051039128694578e-06,
      "loss": 0.3848,
      "step": 3566
    },
    {
      "epoch": 0.22342974365398768,
      "grad_norm": 0.8590844618084081,
      "learning_rate": 9.050444462945355e-06,
      "loss": 0.4373,
      "step": 3567
    },
    {
      "epoch": 0.22349238165333,
      "grad_norm": 0.9214391222263101,
      "learning_rate": 9.049849630479086e-06,
      "loss": 0.4208,
      "step": 3568
    },
    {
      "epoch": 0.2235550196526723,
      "grad_norm": 0.9100006291748037,
      "learning_rate": 9.049254631320255e-06,
      "loss": 0.4721,
      "step": 3569
    },
    {
      "epoch": 0.2236176576520146,
      "grad_norm": 0.8552548573373048,
      "learning_rate": 9.048659465493353e-06,
      "loss": 0.4122,
      "step": 3570
    },
    {
      "epoch": 0.2236802956513569,
      "grad_norm": 0.9282594252677706,
      "learning_rate": 9.048064133022874e-06,
      "loss": 0.4502,
      "step": 3571
    },
    {
      "epoch": 0.2237429336506992,
      "grad_norm": 0.7715374686384755,
      "learning_rate": 9.047468633933326e-06,
      "loss": 0.4001,
      "step": 3572
    },
    {
      "epoch": 0.2238055716500415,
      "grad_norm": 0.8182428274860272,
      "learning_rate": 9.046872968249217e-06,
      "loss": 0.3953,
      "step": 3573
    },
    {
      "epoch": 0.2238682096493838,
      "grad_norm": 0.8972927719198263,
      "learning_rate": 9.046277135995064e-06,
      "loss": 0.4193,
      "step": 3574
    },
    {
      "epoch": 0.2239308476487261,
      "grad_norm": 0.8232806534029271,
      "learning_rate": 9.045681137195394e-06,
      "loss": 0.4541,
      "step": 3575
    },
    {
      "epoch": 0.2239934856480684,
      "grad_norm": 0.8753724604429279,
      "learning_rate": 9.045084971874738e-06,
      "loss": 0.4725,
      "step": 3576
    },
    {
      "epoch": 0.2240561236474107,
      "grad_norm": 0.8946521396492207,
      "learning_rate": 9.044488640057635e-06,
      "loss": 0.427,
      "step": 3577
    },
    {
      "epoch": 0.224118761646753,
      "grad_norm": 0.8968621426419141,
      "learning_rate": 9.043892141768626e-06,
      "loss": 0.4484,
      "step": 3578
    },
    {
      "epoch": 0.2241813996460953,
      "grad_norm": 0.7818726911530873,
      "learning_rate": 9.043295477032266e-06,
      "loss": 0.3982,
      "step": 3579
    },
    {
      "epoch": 0.2242440376454376,
      "grad_norm": 0.8948651865936197,
      "learning_rate": 9.042698645873116e-06,
      "loss": 0.4388,
      "step": 3580
    },
    {
      "epoch": 0.2243066756447799,
      "grad_norm": 0.8154687905268586,
      "learning_rate": 9.042101648315737e-06,
      "loss": 0.3933,
      "step": 3581
    },
    {
      "epoch": 0.22436931364412221,
      "grad_norm": 0.9178305582296664,
      "learning_rate": 9.041504484384707e-06,
      "loss": 0.4444,
      "step": 3582
    },
    {
      "epoch": 0.2244319516434645,
      "grad_norm": 0.8870594082026341,
      "learning_rate": 9.0409071541046e-06,
      "loss": 0.4479,
      "step": 3583
    },
    {
      "epoch": 0.2244945896428068,
      "grad_norm": 0.8757963562963595,
      "learning_rate": 9.040309657500003e-06,
      "loss": 0.4555,
      "step": 3584
    },
    {
      "epoch": 0.2245572276421491,
      "grad_norm": 0.8546885720926737,
      "learning_rate": 9.039711994595513e-06,
      "loss": 0.467,
      "step": 3585
    },
    {
      "epoch": 0.22461986564149142,
      "grad_norm": 0.8235052791578358,
      "learning_rate": 9.039114165415727e-06,
      "loss": 0.3976,
      "step": 3586
    },
    {
      "epoch": 0.22468250364083373,
      "grad_norm": 0.8406702126396312,
      "learning_rate": 9.038516169985252e-06,
      "loss": 0.4101,
      "step": 3587
    },
    {
      "epoch": 0.224745141640176,
      "grad_norm": 0.859588675685274,
      "learning_rate": 9.037918008328703e-06,
      "loss": 0.4126,
      "step": 3588
    },
    {
      "epoch": 0.2248077796395183,
      "grad_norm": 0.8545704971318758,
      "learning_rate": 9.037319680470699e-06,
      "loss": 0.4132,
      "step": 3589
    },
    {
      "epoch": 0.22487041763886062,
      "grad_norm": 0.8583795738651745,
      "learning_rate": 9.036721186435867e-06,
      "loss": 0.4087,
      "step": 3590
    },
    {
      "epoch": 0.22493305563820293,
      "grad_norm": 0.8464809106548128,
      "learning_rate": 9.036122526248841e-06,
      "loss": 0.4238,
      "step": 3591
    },
    {
      "epoch": 0.2249956936375452,
      "grad_norm": 0.8125138545710179,
      "learning_rate": 9.035523699934263e-06,
      "loss": 0.4394,
      "step": 3592
    },
    {
      "epoch": 0.22505833163688752,
      "grad_norm": 0.8773698418755456,
      "learning_rate": 9.034924707516781e-06,
      "loss": 0.4569,
      "step": 3593
    },
    {
      "epoch": 0.22512096963622982,
      "grad_norm": 0.852973862653054,
      "learning_rate": 9.03432554902105e-06,
      "loss": 0.4678,
      "step": 3594
    },
    {
      "epoch": 0.22518360763557213,
      "grad_norm": 0.857960585047986,
      "learning_rate": 9.033726224471728e-06,
      "loss": 0.4093,
      "step": 3595
    },
    {
      "epoch": 0.2252462456349144,
      "grad_norm": 0.9377309539076115,
      "learning_rate": 9.033126733893489e-06,
      "loss": 0.4439,
      "step": 3596
    },
    {
      "epoch": 0.22530888363425672,
      "grad_norm": 0.8178875687126334,
      "learning_rate": 9.032527077311004e-06,
      "loss": 0.4359,
      "step": 3597
    },
    {
      "epoch": 0.22537152163359903,
      "grad_norm": 0.8309884363364699,
      "learning_rate": 9.031927254748955e-06,
      "loss": 0.4556,
      "step": 3598
    },
    {
      "epoch": 0.22543415963294133,
      "grad_norm": 0.8892743770215512,
      "learning_rate": 9.031327266232034e-06,
      "loss": 0.4171,
      "step": 3599
    },
    {
      "epoch": 0.2254967976322836,
      "grad_norm": 0.8141032607997513,
      "learning_rate": 9.030727111784932e-06,
      "loss": 0.4281,
      "step": 3600
    },
    {
      "epoch": 0.22555943563162592,
      "grad_norm": 0.8993256858410686,
      "learning_rate": 9.030126791432354e-06,
      "loss": 0.4488,
      "step": 3601
    },
    {
      "epoch": 0.22562207363096823,
      "grad_norm": 0.918306813247655,
      "learning_rate": 9.02952630519901e-06,
      "loss": 0.4522,
      "step": 3602
    },
    {
      "epoch": 0.22568471163031054,
      "grad_norm": 0.8367776861586605,
      "learning_rate": 9.028925653109615e-06,
      "loss": 0.4097,
      "step": 3603
    },
    {
      "epoch": 0.22574734962965282,
      "grad_norm": 0.8846374185660012,
      "learning_rate": 9.028324835188893e-06,
      "loss": 0.4178,
      "step": 3604
    },
    {
      "epoch": 0.22580998762899512,
      "grad_norm": 0.727796092650797,
      "learning_rate": 9.027723851461572e-06,
      "loss": 0.4906,
      "step": 3605
    },
    {
      "epoch": 0.22587262562833743,
      "grad_norm": 0.8237345762759789,
      "learning_rate": 9.02712270195239e-06,
      "loss": 0.4152,
      "step": 3606
    },
    {
      "epoch": 0.22593526362767974,
      "grad_norm": 0.7922443345674449,
      "learning_rate": 9.026521386686092e-06,
      "loss": 0.4123,
      "step": 3607
    },
    {
      "epoch": 0.22599790162702205,
      "grad_norm": 0.8440250170093321,
      "learning_rate": 9.025919905687424e-06,
      "loss": 0.4358,
      "step": 3608
    },
    {
      "epoch": 0.22606053962636433,
      "grad_norm": 0.89195818915605,
      "learning_rate": 9.025318258981148e-06,
      "loss": 0.4518,
      "step": 3609
    },
    {
      "epoch": 0.22612317762570663,
      "grad_norm": 0.8998229010206983,
      "learning_rate": 9.024716446592023e-06,
      "loss": 0.4057,
      "step": 3610
    },
    {
      "epoch": 0.22618581562504894,
      "grad_norm": 0.9912987269987427,
      "learning_rate": 9.024114468544822e-06,
      "loss": 0.4364,
      "step": 3611
    },
    {
      "epoch": 0.22624845362439125,
      "grad_norm": 0.8830502862595292,
      "learning_rate": 9.023512324864325e-06,
      "loss": 0.4293,
      "step": 3612
    },
    {
      "epoch": 0.22631109162373353,
      "grad_norm": 0.8593596591472704,
      "learning_rate": 9.022910015575312e-06,
      "loss": 0.4441,
      "step": 3613
    },
    {
      "epoch": 0.22637372962307584,
      "grad_norm": 0.8619290218066976,
      "learning_rate": 9.022307540702576e-06,
      "loss": 0.4657,
      "step": 3614
    },
    {
      "epoch": 0.22643636762241814,
      "grad_norm": 0.8683566535220671,
      "learning_rate": 9.021704900270916e-06,
      "loss": 0.4211,
      "step": 3615
    },
    {
      "epoch": 0.22649900562176045,
      "grad_norm": 0.7806352379146742,
      "learning_rate": 9.021102094305134e-06,
      "loss": 0.3834,
      "step": 3616
    },
    {
      "epoch": 0.22656164362110273,
      "grad_norm": 0.8695393513515355,
      "learning_rate": 9.020499122830045e-06,
      "loss": 0.4369,
      "step": 3617
    },
    {
      "epoch": 0.22662428162044504,
      "grad_norm": 0.8349845769735094,
      "learning_rate": 9.019895985870466e-06,
      "loss": 0.3852,
      "step": 3618
    },
    {
      "epoch": 0.22668691961978735,
      "grad_norm": 0.8545860122953967,
      "learning_rate": 9.019292683451222e-06,
      "loss": 0.47,
      "step": 3619
    },
    {
      "epoch": 0.22674955761912965,
      "grad_norm": 0.7814150720087838,
      "learning_rate": 9.018689215597145e-06,
      "loss": 0.4153,
      "step": 3620
    },
    {
      "epoch": 0.22681219561847193,
      "grad_norm": 0.9190723226545144,
      "learning_rate": 9.018085582333074e-06,
      "loss": 0.4441,
      "step": 3621
    },
    {
      "epoch": 0.22687483361781424,
      "grad_norm": 0.8217340878291491,
      "learning_rate": 9.017481783683854e-06,
      "loss": 0.4484,
      "step": 3622
    },
    {
      "epoch": 0.22693747161715655,
      "grad_norm": 0.8241610497968725,
      "learning_rate": 9.016877819674337e-06,
      "loss": 0.4368,
      "step": 3623
    },
    {
      "epoch": 0.22700010961649886,
      "grad_norm": 0.847841036551124,
      "learning_rate": 9.016273690329385e-06,
      "loss": 0.446,
      "step": 3624
    },
    {
      "epoch": 0.22706274761584114,
      "grad_norm": 0.8229904034180477,
      "learning_rate": 9.015669395673863e-06,
      "loss": 0.4417,
      "step": 3625
    },
    {
      "epoch": 0.22712538561518344,
      "grad_norm": 0.8548250159269043,
      "learning_rate": 9.015064935732643e-06,
      "loss": 0.434,
      "step": 3626
    },
    {
      "epoch": 0.22718802361452575,
      "grad_norm": 0.8159545581228318,
      "learning_rate": 9.014460310530604e-06,
      "loss": 0.3945,
      "step": 3627
    },
    {
      "epoch": 0.22725066161386806,
      "grad_norm": 0.8727113804774039,
      "learning_rate": 9.013855520092635e-06,
      "loss": 0.4566,
      "step": 3628
    },
    {
      "epoch": 0.22731329961321037,
      "grad_norm": 0.8353332483878763,
      "learning_rate": 9.013250564443627e-06,
      "loss": 0.4394,
      "step": 3629
    },
    {
      "epoch": 0.22737593761255265,
      "grad_norm": 0.8624163550950759,
      "learning_rate": 9.01264544360848e-06,
      "loss": 0.461,
      "step": 3630
    },
    {
      "epoch": 0.22743857561189496,
      "grad_norm": 0.8486252892448428,
      "learning_rate": 9.012040157612103e-06,
      "loss": 0.3931,
      "step": 3631
    },
    {
      "epoch": 0.22750121361123726,
      "grad_norm": 0.8179914168690471,
      "learning_rate": 9.011434706479406e-06,
      "loss": 0.377,
      "step": 3632
    },
    {
      "epoch": 0.22756385161057957,
      "grad_norm": 0.8321579826468115,
      "learning_rate": 9.010829090235316e-06,
      "loss": 0.3845,
      "step": 3633
    },
    {
      "epoch": 0.22762648960992185,
      "grad_norm": 0.7961289461798637,
      "learning_rate": 9.010223308904754e-06,
      "loss": 0.3939,
      "step": 3634
    },
    {
      "epoch": 0.22768912760926416,
      "grad_norm": 0.8432789609437337,
      "learning_rate": 9.009617362512657e-06,
      "loss": 0.4285,
      "step": 3635
    },
    {
      "epoch": 0.22775176560860647,
      "grad_norm": 0.8548105518309141,
      "learning_rate": 9.009011251083964e-06,
      "loss": 0.4404,
      "step": 3636
    },
    {
      "epoch": 0.22781440360794877,
      "grad_norm": 0.891173280719501,
      "learning_rate": 9.008404974643625e-06,
      "loss": 0.4701,
      "step": 3637
    },
    {
      "epoch": 0.22787704160729105,
      "grad_norm": 0.8529117588625832,
      "learning_rate": 9.007798533216591e-06,
      "loss": 0.4339,
      "step": 3638
    },
    {
      "epoch": 0.22793967960663336,
      "grad_norm": 0.8525482471791082,
      "learning_rate": 9.007191926827828e-06,
      "loss": 0.4079,
      "step": 3639
    },
    {
      "epoch": 0.22800231760597567,
      "grad_norm": 0.8350564781465142,
      "learning_rate": 9.006585155502301e-06,
      "loss": 0.4063,
      "step": 3640
    },
    {
      "epoch": 0.22806495560531798,
      "grad_norm": 0.8691396553844191,
      "learning_rate": 9.005978219264985e-06,
      "loss": 0.4014,
      "step": 3641
    },
    {
      "epoch": 0.22812759360466026,
      "grad_norm": 0.8671149230432913,
      "learning_rate": 9.005371118140861e-06,
      "loss": 0.443,
      "step": 3642
    },
    {
      "epoch": 0.22819023160400256,
      "grad_norm": 0.895613361996229,
      "learning_rate": 9.00476385215492e-06,
      "loss": 0.4258,
      "step": 3643
    },
    {
      "epoch": 0.22825286960334487,
      "grad_norm": 0.8591192489273214,
      "learning_rate": 9.004156421332156e-06,
      "loss": 0.4507,
      "step": 3644
    },
    {
      "epoch": 0.22831550760268718,
      "grad_norm": 0.8512531113734823,
      "learning_rate": 9.003548825697567e-06,
      "loss": 0.4518,
      "step": 3645
    },
    {
      "epoch": 0.22837814560202946,
      "grad_norm": 0.814154493244863,
      "learning_rate": 9.00294106527617e-06,
      "loss": 0.4565,
      "step": 3646
    },
    {
      "epoch": 0.22844078360137177,
      "grad_norm": 0.8988306919368988,
      "learning_rate": 9.002333140092972e-06,
      "loss": 0.4898,
      "step": 3647
    },
    {
      "epoch": 0.22850342160071407,
      "grad_norm": 0.8543412422958783,
      "learning_rate": 9.001725050173e-06,
      "loss": 0.4379,
      "step": 3648
    },
    {
      "epoch": 0.22856605960005638,
      "grad_norm": 0.9015357731751348,
      "learning_rate": 9.001116795541282e-06,
      "loss": 0.4868,
      "step": 3649
    },
    {
      "epoch": 0.22862869759939866,
      "grad_norm": 0.8030811000272579,
      "learning_rate": 9.000508376222856e-06,
      "loss": 0.4616,
      "step": 3650
    },
    {
      "epoch": 0.22869133559874097,
      "grad_norm": 0.7979115769066775,
      "learning_rate": 8.99989979224276e-06,
      "loss": 0.4071,
      "step": 3651
    },
    {
      "epoch": 0.22875397359808328,
      "grad_norm": 0.886059446643181,
      "learning_rate": 8.999291043626048e-06,
      "loss": 0.3925,
      "step": 3652
    },
    {
      "epoch": 0.22881661159742558,
      "grad_norm": 0.8560719629244559,
      "learning_rate": 8.998682130397773e-06,
      "loss": 0.3936,
      "step": 3653
    },
    {
      "epoch": 0.2288792495967679,
      "grad_norm": 0.8103702380942743,
      "learning_rate": 8.998073052583001e-06,
      "loss": 0.4242,
      "step": 3654
    },
    {
      "epoch": 0.22894188759611017,
      "grad_norm": 0.8459386323519481,
      "learning_rate": 8.9974638102068e-06,
      "loss": 0.4471,
      "step": 3655
    },
    {
      "epoch": 0.22900452559545248,
      "grad_norm": 0.9311299842616617,
      "learning_rate": 8.996854403294246e-06,
      "loss": 0.4491,
      "step": 3656
    },
    {
      "epoch": 0.2290671635947948,
      "grad_norm": 0.7953068491551474,
      "learning_rate": 8.996244831870421e-06,
      "loss": 0.4263,
      "step": 3657
    },
    {
      "epoch": 0.2291298015941371,
      "grad_norm": 0.8209280593774693,
      "learning_rate": 8.99563509596042e-06,
      "loss": 0.4063,
      "step": 3658
    },
    {
      "epoch": 0.22919243959347937,
      "grad_norm": 0.8630643221499814,
      "learning_rate": 8.995025195589334e-06,
      "loss": 0.4336,
      "step": 3659
    },
    {
      "epoch": 0.22925507759282168,
      "grad_norm": 0.8805229550459962,
      "learning_rate": 8.994415130782272e-06,
      "loss": 0.4343,
      "step": 3660
    },
    {
      "epoch": 0.229317715592164,
      "grad_norm": 0.787656389980031,
      "learning_rate": 8.99380490156434e-06,
      "loss": 0.4129,
      "step": 3661
    },
    {
      "epoch": 0.2293803535915063,
      "grad_norm": 0.7847861060729685,
      "learning_rate": 8.993194507960658e-06,
      "loss": 0.4,
      "step": 3662
    },
    {
      "epoch": 0.22944299159084858,
      "grad_norm": 0.8631948816226814,
      "learning_rate": 8.992583949996349e-06,
      "loss": 0.4831,
      "step": 3663
    },
    {
      "epoch": 0.22950562959019088,
      "grad_norm": 0.835542580361206,
      "learning_rate": 8.991973227696543e-06,
      "loss": 0.4329,
      "step": 3664
    },
    {
      "epoch": 0.2295682675895332,
      "grad_norm": 0.9374668403490802,
      "learning_rate": 8.991362341086378e-06,
      "loss": 0.4341,
      "step": 3665
    },
    {
      "epoch": 0.2296309055888755,
      "grad_norm": 0.7640931525883586,
      "learning_rate": 8.990751290190999e-06,
      "loss": 0.4965,
      "step": 3666
    },
    {
      "epoch": 0.22969354358821778,
      "grad_norm": 0.8432421447935764,
      "learning_rate": 8.990140075035555e-06,
      "loss": 0.4429,
      "step": 3667
    },
    {
      "epoch": 0.2297561815875601,
      "grad_norm": 0.8700845524036463,
      "learning_rate": 8.989528695645206e-06,
      "loss": 0.4329,
      "step": 3668
    },
    {
      "epoch": 0.2298188195869024,
      "grad_norm": 0.9078597897264133,
      "learning_rate": 8.988917152045114e-06,
      "loss": 0.4735,
      "step": 3669
    },
    {
      "epoch": 0.2298814575862447,
      "grad_norm": 0.7924441469914381,
      "learning_rate": 8.988305444260453e-06,
      "loss": 0.4153,
      "step": 3670
    },
    {
      "epoch": 0.22994409558558698,
      "grad_norm": 0.8367374938706016,
      "learning_rate": 8.9876935723164e-06,
      "loss": 0.4118,
      "step": 3671
    },
    {
      "epoch": 0.2300067335849293,
      "grad_norm": 0.9022932997261867,
      "learning_rate": 8.987081536238138e-06,
      "loss": 0.4515,
      "step": 3672
    },
    {
      "epoch": 0.2300693715842716,
      "grad_norm": 0.8849605245440336,
      "learning_rate": 8.98646933605086e-06,
      "loss": 0.4414,
      "step": 3673
    },
    {
      "epoch": 0.2301320095836139,
      "grad_norm": 0.8391350947148194,
      "learning_rate": 8.985856971779765e-06,
      "loss": 0.4564,
      "step": 3674
    },
    {
      "epoch": 0.2301946475829562,
      "grad_norm": 0.8756518010482653,
      "learning_rate": 8.985244443450056e-06,
      "loss": 0.4508,
      "step": 3675
    },
    {
      "epoch": 0.2302572855822985,
      "grad_norm": 0.8750934179465625,
      "learning_rate": 8.984631751086949e-06,
      "loss": 0.4473,
      "step": 3676
    },
    {
      "epoch": 0.2303199235816408,
      "grad_norm": 0.7306329782285383,
      "learning_rate": 8.984018894715657e-06,
      "loss": 0.4914,
      "step": 3677
    },
    {
      "epoch": 0.2303825615809831,
      "grad_norm": 0.6600640693909249,
      "learning_rate": 8.983405874361408e-06,
      "loss": 0.4639,
      "step": 3678
    },
    {
      "epoch": 0.23044519958032542,
      "grad_norm": 0.8420698375494975,
      "learning_rate": 8.982792690049434e-06,
      "loss": 0.4087,
      "step": 3679
    },
    {
      "epoch": 0.2305078375796677,
      "grad_norm": 0.8690032093019414,
      "learning_rate": 8.982179341804975e-06,
      "loss": 0.4705,
      "step": 3680
    },
    {
      "epoch": 0.23057047557901,
      "grad_norm": 0.8790340420396889,
      "learning_rate": 8.981565829653274e-06,
      "loss": 0.4335,
      "step": 3681
    },
    {
      "epoch": 0.2306331135783523,
      "grad_norm": 0.9066763166144028,
      "learning_rate": 8.980952153619584e-06,
      "loss": 0.4362,
      "step": 3682
    },
    {
      "epoch": 0.23069575157769462,
      "grad_norm": 0.8774478353005137,
      "learning_rate": 8.980338313729164e-06,
      "loss": 0.4339,
      "step": 3683
    },
    {
      "epoch": 0.2307583895770369,
      "grad_norm": 0.8827137421941279,
      "learning_rate": 8.979724310007281e-06,
      "loss": 0.4673,
      "step": 3684
    },
    {
      "epoch": 0.2308210275763792,
      "grad_norm": 0.8005373431857583,
      "learning_rate": 8.979110142479208e-06,
      "loss": 0.42,
      "step": 3685
    },
    {
      "epoch": 0.23088366557572151,
      "grad_norm": 0.8508724699571976,
      "learning_rate": 8.978495811170221e-06,
      "loss": 0.4053,
      "step": 3686
    },
    {
      "epoch": 0.23094630357506382,
      "grad_norm": 0.8448282506290096,
      "learning_rate": 8.977881316105609e-06,
      "loss": 0.4229,
      "step": 3687
    },
    {
      "epoch": 0.2310089415744061,
      "grad_norm": 0.7681157991176513,
      "learning_rate": 8.977266657310664e-06,
      "loss": 0.4142,
      "step": 3688
    },
    {
      "epoch": 0.2310715795737484,
      "grad_norm": 0.83437527374813,
      "learning_rate": 8.976651834810683e-06,
      "loss": 0.4084,
      "step": 3689
    },
    {
      "epoch": 0.23113421757309072,
      "grad_norm": 0.8922894232275208,
      "learning_rate": 8.976036848630977e-06,
      "loss": 0.441,
      "step": 3690
    },
    {
      "epoch": 0.23119685557243302,
      "grad_norm": 0.80720800699337,
      "learning_rate": 8.975421698796855e-06,
      "loss": 0.4433,
      "step": 3691
    },
    {
      "epoch": 0.2312594935717753,
      "grad_norm": 0.8447285505251921,
      "learning_rate": 8.974806385333638e-06,
      "loss": 0.4303,
      "step": 3692
    },
    {
      "epoch": 0.2313221315711176,
      "grad_norm": 0.8223564555337609,
      "learning_rate": 8.974190908266654e-06,
      "loss": 0.4001,
      "step": 3693
    },
    {
      "epoch": 0.23138476957045992,
      "grad_norm": 0.8967267771608749,
      "learning_rate": 8.973575267621234e-06,
      "loss": 0.4383,
      "step": 3694
    },
    {
      "epoch": 0.23144740756980223,
      "grad_norm": 0.8496258839207315,
      "learning_rate": 8.972959463422716e-06,
      "loss": 0.4142,
      "step": 3695
    },
    {
      "epoch": 0.23151004556914453,
      "grad_norm": 0.9154507392469868,
      "learning_rate": 8.972343495696452e-06,
      "loss": 0.4399,
      "step": 3696
    },
    {
      "epoch": 0.23157268356848681,
      "grad_norm": 0.8650069351910815,
      "learning_rate": 8.97172736446779e-06,
      "loss": 0.4275,
      "step": 3697
    },
    {
      "epoch": 0.23163532156782912,
      "grad_norm": 0.8639368026876769,
      "learning_rate": 8.971111069762096e-06,
      "loss": 0.4555,
      "step": 3698
    },
    {
      "epoch": 0.23169795956717143,
      "grad_norm": 0.9458452056303981,
      "learning_rate": 8.970494611604729e-06,
      "loss": 0.4635,
      "step": 3699
    },
    {
      "epoch": 0.23176059756651374,
      "grad_norm": 0.8597297933808521,
      "learning_rate": 8.96987799002107e-06,
      "loss": 0.4123,
      "step": 3700
    },
    {
      "epoch": 0.23182323556585602,
      "grad_norm": 0.8148703502192166,
      "learning_rate": 8.969261205036494e-06,
      "loss": 0.4405,
      "step": 3701
    },
    {
      "epoch": 0.23188587356519832,
      "grad_norm": 0.8320595041150318,
      "learning_rate": 8.968644256676392e-06,
      "loss": 0.435,
      "step": 3702
    },
    {
      "epoch": 0.23194851156454063,
      "grad_norm": 0.909770632935105,
      "learning_rate": 8.968027144966153e-06,
      "loss": 0.5001,
      "step": 3703
    },
    {
      "epoch": 0.23201114956388294,
      "grad_norm": 0.865786128481643,
      "learning_rate": 8.967409869931183e-06,
      "loss": 0.4559,
      "step": 3704
    },
    {
      "epoch": 0.23207378756322522,
      "grad_norm": 0.8464079994323607,
      "learning_rate": 8.966792431596885e-06,
      "loss": 0.4411,
      "step": 3705
    },
    {
      "epoch": 0.23213642556256753,
      "grad_norm": 0.8609449460268267,
      "learning_rate": 8.966174829988677e-06,
      "loss": 0.4156,
      "step": 3706
    },
    {
      "epoch": 0.23219906356190984,
      "grad_norm": 0.8324236758129632,
      "learning_rate": 8.965557065131973e-06,
      "loss": 0.451,
      "step": 3707
    },
    {
      "epoch": 0.23226170156125214,
      "grad_norm": 0.8642968861077522,
      "learning_rate": 8.964939137052207e-06,
      "loss": 0.4355,
      "step": 3708
    },
    {
      "epoch": 0.23232433956059442,
      "grad_norm": 0.7932557856384131,
      "learning_rate": 8.964321045774808e-06,
      "loss": 0.4069,
      "step": 3709
    },
    {
      "epoch": 0.23238697755993673,
      "grad_norm": 0.7958166263296355,
      "learning_rate": 8.963702791325219e-06,
      "loss": 0.4134,
      "step": 3710
    },
    {
      "epoch": 0.23244961555927904,
      "grad_norm": 0.8526060383000797,
      "learning_rate": 8.963084373728889e-06,
      "loss": 0.4312,
      "step": 3711
    },
    {
      "epoch": 0.23251225355862135,
      "grad_norm": 0.9234953090126835,
      "learning_rate": 8.96246579301127e-06,
      "loss": 0.4767,
      "step": 3712
    },
    {
      "epoch": 0.23257489155796363,
      "grad_norm": 0.792676111705942,
      "learning_rate": 8.961847049197823e-06,
      "loss": 0.3909,
      "step": 3713
    },
    {
      "epoch": 0.23263752955730593,
      "grad_norm": 0.8683274822151229,
      "learning_rate": 8.961228142314015e-06,
      "loss": 0.4179,
      "step": 3714
    },
    {
      "epoch": 0.23270016755664824,
      "grad_norm": 0.8852087875172702,
      "learning_rate": 8.960609072385325e-06,
      "loss": 0.4722,
      "step": 3715
    },
    {
      "epoch": 0.23276280555599055,
      "grad_norm": 0.8221505926088775,
      "learning_rate": 8.959989839437227e-06,
      "loss": 0.4394,
      "step": 3716
    },
    {
      "epoch": 0.23282544355533286,
      "grad_norm": 0.8164151022937488,
      "learning_rate": 8.959370443495215e-06,
      "loss": 0.4793,
      "step": 3717
    },
    {
      "epoch": 0.23288808155467514,
      "grad_norm": 0.802246347088223,
      "learning_rate": 8.958750884584777e-06,
      "loss": 0.3871,
      "step": 3718
    },
    {
      "epoch": 0.23295071955401744,
      "grad_norm": 0.812899693939796,
      "learning_rate": 8.95813116273142e-06,
      "loss": 0.4181,
      "step": 3719
    },
    {
      "epoch": 0.23301335755335975,
      "grad_norm": 0.8832434904783107,
      "learning_rate": 8.957511277960652e-06,
      "loss": 0.4333,
      "step": 3720
    },
    {
      "epoch": 0.23307599555270206,
      "grad_norm": 0.8593174286802718,
      "learning_rate": 8.956891230297981e-06,
      "loss": 0.4396,
      "step": 3721
    },
    {
      "epoch": 0.23313863355204434,
      "grad_norm": 0.8530785962661518,
      "learning_rate": 8.956271019768934e-06,
      "loss": 0.4384,
      "step": 3722
    },
    {
      "epoch": 0.23320127155138665,
      "grad_norm": 0.8855744902922944,
      "learning_rate": 8.95565064639904e-06,
      "loss": 0.3881,
      "step": 3723
    },
    {
      "epoch": 0.23326390955072895,
      "grad_norm": 0.8947122930005271,
      "learning_rate": 8.955030110213827e-06,
      "loss": 0.435,
      "step": 3724
    },
    {
      "epoch": 0.23332654755007126,
      "grad_norm": 0.8109033463622566,
      "learning_rate": 8.954409411238844e-06,
      "loss": 0.4194,
      "step": 3725
    },
    {
      "epoch": 0.23338918554941354,
      "grad_norm": 0.8572174454075353,
      "learning_rate": 8.953788549499634e-06,
      "loss": 0.4779,
      "step": 3726
    },
    {
      "epoch": 0.23345182354875585,
      "grad_norm": 0.8207598573752485,
      "learning_rate": 8.953167525021753e-06,
      "loss": 0.4206,
      "step": 3727
    },
    {
      "epoch": 0.23351446154809816,
      "grad_norm": 0.8544864944702866,
      "learning_rate": 8.952546337830763e-06,
      "loss": 0.422,
      "step": 3728
    },
    {
      "epoch": 0.23357709954744046,
      "grad_norm": 0.9060684144219673,
      "learning_rate": 8.951924987952233e-06,
      "loss": 0.4227,
      "step": 3729
    },
    {
      "epoch": 0.23363973754678274,
      "grad_norm": 0.8155004566814957,
      "learning_rate": 8.951303475411738e-06,
      "loss": 0.4825,
      "step": 3730
    },
    {
      "epoch": 0.23370237554612505,
      "grad_norm": 0.856861357625736,
      "learning_rate": 8.950681800234857e-06,
      "loss": 0.4208,
      "step": 3731
    },
    {
      "epoch": 0.23376501354546736,
      "grad_norm": 0.7795471092510527,
      "learning_rate": 8.95005996244718e-06,
      "loss": 0.377,
      "step": 3732
    },
    {
      "epoch": 0.23382765154480967,
      "grad_norm": 0.8548138407839065,
      "learning_rate": 8.949437962074301e-06,
      "loss": 0.4103,
      "step": 3733
    },
    {
      "epoch": 0.23389028954415195,
      "grad_norm": 0.7934149776548439,
      "learning_rate": 8.948815799141824e-06,
      "loss": 0.4173,
      "step": 3734
    },
    {
      "epoch": 0.23395292754349425,
      "grad_norm": 0.8797667176727282,
      "learning_rate": 8.948193473675355e-06,
      "loss": 0.4423,
      "step": 3735
    },
    {
      "epoch": 0.23401556554283656,
      "grad_norm": 0.8811033603951505,
      "learning_rate": 8.947570985700511e-06,
      "loss": 0.4356,
      "step": 3736
    },
    {
      "epoch": 0.23407820354217887,
      "grad_norm": 0.8268778424750751,
      "learning_rate": 8.946948335242912e-06,
      "loss": 0.3709,
      "step": 3737
    },
    {
      "epoch": 0.23414084154152118,
      "grad_norm": 0.8273624294044409,
      "learning_rate": 8.946325522328186e-06,
      "loss": 0.4196,
      "step": 3738
    },
    {
      "epoch": 0.23420347954086346,
      "grad_norm": 0.8477235710455265,
      "learning_rate": 8.94570254698197e-06,
      "loss": 0.4288,
      "step": 3739
    },
    {
      "epoch": 0.23426611754020576,
      "grad_norm": 0.8668471997529342,
      "learning_rate": 8.945079409229904e-06,
      "loss": 0.471,
      "step": 3740
    },
    {
      "epoch": 0.23432875553954807,
      "grad_norm": 0.8363740730863963,
      "learning_rate": 8.944456109097637e-06,
      "loss": 0.4131,
      "step": 3741
    },
    {
      "epoch": 0.23439139353889038,
      "grad_norm": 0.7833328468630085,
      "learning_rate": 8.943832646610823e-06,
      "loss": 0.3788,
      "step": 3742
    },
    {
      "epoch": 0.23445403153823266,
      "grad_norm": 0.8877817020225685,
      "learning_rate": 8.943209021795129e-06,
      "loss": 0.4666,
      "step": 3743
    },
    {
      "epoch": 0.23451666953757497,
      "grad_norm": 0.8471352249358615,
      "learning_rate": 8.942585234676218e-06,
      "loss": 0.4314,
      "step": 3744
    },
    {
      "epoch": 0.23457930753691728,
      "grad_norm": 0.8736702672410306,
      "learning_rate": 8.941961285279767e-06,
      "loss": 0.4325,
      "step": 3745
    },
    {
      "epoch": 0.23464194553625958,
      "grad_norm": 0.8030600213820606,
      "learning_rate": 8.941337173631457e-06,
      "loss": 0.4334,
      "step": 3746
    },
    {
      "epoch": 0.23470458353560186,
      "grad_norm": 0.8377536715516944,
      "learning_rate": 8.940712899756977e-06,
      "loss": 0.4187,
      "step": 3747
    },
    {
      "epoch": 0.23476722153494417,
      "grad_norm": 0.9225531878474668,
      "learning_rate": 8.940088463682027e-06,
      "loss": 0.4698,
      "step": 3748
    },
    {
      "epoch": 0.23482985953428648,
      "grad_norm": 0.8791047795436393,
      "learning_rate": 8.9394638654323e-06,
      "loss": 0.4077,
      "step": 3749
    },
    {
      "epoch": 0.23489249753362879,
      "grad_norm": 0.8326946890180807,
      "learning_rate": 8.93883910503351e-06,
      "loss": 0.4817,
      "step": 3750
    },
    {
      "epoch": 0.23495513553297107,
      "grad_norm": 0.8370256712328739,
      "learning_rate": 8.938214182511372e-06,
      "loss": 0.4085,
      "step": 3751
    },
    {
      "epoch": 0.23501777353231337,
      "grad_norm": 0.8681022986897976,
      "learning_rate": 8.937589097891607e-06,
      "loss": 0.412,
      "step": 3752
    },
    {
      "epoch": 0.23508041153165568,
      "grad_norm": 0.7871696422310682,
      "learning_rate": 8.936963851199944e-06,
      "loss": 0.4241,
      "step": 3753
    },
    {
      "epoch": 0.235143049530998,
      "grad_norm": 0.8508332562874982,
      "learning_rate": 8.936338442462117e-06,
      "loss": 0.4082,
      "step": 3754
    },
    {
      "epoch": 0.23520568753034027,
      "grad_norm": 0.737942693515263,
      "learning_rate": 8.93571287170387e-06,
      "loss": 0.483,
      "step": 3755
    },
    {
      "epoch": 0.23526832552968258,
      "grad_norm": 0.832644952096058,
      "learning_rate": 8.93508713895095e-06,
      "loss": 0.4628,
      "step": 3756
    },
    {
      "epoch": 0.23533096352902488,
      "grad_norm": 0.8521944499636712,
      "learning_rate": 8.934461244229115e-06,
      "loss": 0.3606,
      "step": 3757
    },
    {
      "epoch": 0.2353936015283672,
      "grad_norm": 0.8710977827642152,
      "learning_rate": 8.933835187564125e-06,
      "loss": 0.4274,
      "step": 3758
    },
    {
      "epoch": 0.2354562395277095,
      "grad_norm": 0.8231412767388027,
      "learning_rate": 8.933208968981746e-06,
      "loss": 0.4297,
      "step": 3759
    },
    {
      "epoch": 0.23551887752705178,
      "grad_norm": 0.8216214486741641,
      "learning_rate": 8.932582588507757e-06,
      "loss": 0.3866,
      "step": 3760
    },
    {
      "epoch": 0.2355815155263941,
      "grad_norm": 0.6981204660776404,
      "learning_rate": 8.93195604616794e-06,
      "loss": 0.4872,
      "step": 3761
    },
    {
      "epoch": 0.2356441535257364,
      "grad_norm": 0.8310533193036078,
      "learning_rate": 8.93132934198808e-06,
      "loss": 0.4489,
      "step": 3762
    },
    {
      "epoch": 0.2357067915250787,
      "grad_norm": 0.9030648349247024,
      "learning_rate": 8.930702475993976e-06,
      "loss": 0.4639,
      "step": 3763
    },
    {
      "epoch": 0.23576942952442098,
      "grad_norm": 0.7436556012972492,
      "learning_rate": 8.930075448211429e-06,
      "loss": 0.4759,
      "step": 3764
    },
    {
      "epoch": 0.2358320675237633,
      "grad_norm": 0.8902968841061338,
      "learning_rate": 8.929448258666246e-06,
      "loss": 0.4121,
      "step": 3765
    },
    {
      "epoch": 0.2358947055231056,
      "grad_norm": 0.8844055947360535,
      "learning_rate": 8.928820907384244e-06,
      "loss": 0.4538,
      "step": 3766
    },
    {
      "epoch": 0.2359573435224479,
      "grad_norm": 0.7669257056824479,
      "learning_rate": 8.928193394391242e-06,
      "loss": 0.3895,
      "step": 3767
    },
    {
      "epoch": 0.23601998152179018,
      "grad_norm": 0.7291211176838776,
      "learning_rate": 8.927565719713074e-06,
      "loss": 0.4787,
      "step": 3768
    },
    {
      "epoch": 0.2360826195211325,
      "grad_norm": 0.8533683796020851,
      "learning_rate": 8.92693788337557e-06,
      "loss": 0.4365,
      "step": 3769
    },
    {
      "epoch": 0.2361452575204748,
      "grad_norm": 0.8945645650563434,
      "learning_rate": 8.926309885404576e-06,
      "loss": 0.4735,
      "step": 3770
    },
    {
      "epoch": 0.2362078955198171,
      "grad_norm": 0.9050142450603613,
      "learning_rate": 8.925681725825937e-06,
      "loss": 0.4473,
      "step": 3771
    },
    {
      "epoch": 0.2362705335191594,
      "grad_norm": 0.7328997684334048,
      "learning_rate": 8.92505340466551e-06,
      "loss": 0.4825,
      "step": 3772
    },
    {
      "epoch": 0.2363331715185017,
      "grad_norm": 0.8168763718248608,
      "learning_rate": 8.924424921949153e-06,
      "loss": 0.4328,
      "step": 3773
    },
    {
      "epoch": 0.236395809517844,
      "grad_norm": 0.8679027765215588,
      "learning_rate": 8.923796277702742e-06,
      "loss": 0.4383,
      "step": 3774
    },
    {
      "epoch": 0.2364584475171863,
      "grad_norm": 0.8159089332478857,
      "learning_rate": 8.923167471952146e-06,
      "loss": 0.3984,
      "step": 3775
    },
    {
      "epoch": 0.2365210855165286,
      "grad_norm": 0.8649249714778,
      "learning_rate": 8.922538504723249e-06,
      "loss": 0.4182,
      "step": 3776
    },
    {
      "epoch": 0.2365837235158709,
      "grad_norm": 0.853995694656672,
      "learning_rate": 8.92190937604194e-06,
      "loss": 0.4189,
      "step": 3777
    },
    {
      "epoch": 0.2366463615152132,
      "grad_norm": 0.8728484594156509,
      "learning_rate": 8.921280085934112e-06,
      "loss": 0.4516,
      "step": 3778
    },
    {
      "epoch": 0.2367089995145555,
      "grad_norm": 0.9624284579625785,
      "learning_rate": 8.920650634425669e-06,
      "loss": 0.4549,
      "step": 3779
    },
    {
      "epoch": 0.2367716375138978,
      "grad_norm": 0.8319758140391752,
      "learning_rate": 8.920021021542516e-06,
      "loss": 0.4645,
      "step": 3780
    },
    {
      "epoch": 0.2368342755132401,
      "grad_norm": 0.7819526285350792,
      "learning_rate": 8.919391247310571e-06,
      "loss": 0.4165,
      "step": 3781
    },
    {
      "epoch": 0.2368969135125824,
      "grad_norm": 0.9375142597300046,
      "learning_rate": 8.918761311755757e-06,
      "loss": 0.4708,
      "step": 3782
    },
    {
      "epoch": 0.23695955151192472,
      "grad_norm": 0.7683988973098524,
      "learning_rate": 8.918131214903999e-06,
      "loss": 0.3845,
      "step": 3783
    },
    {
      "epoch": 0.23702218951126702,
      "grad_norm": 0.8412245374335875,
      "learning_rate": 8.917500956781231e-06,
      "loss": 0.3808,
      "step": 3784
    },
    {
      "epoch": 0.2370848275106093,
      "grad_norm": 0.7821946421706883,
      "learning_rate": 8.916870537413398e-06,
      "loss": 0.4097,
      "step": 3785
    },
    {
      "epoch": 0.2371474655099516,
      "grad_norm": 0.7907454386525287,
      "learning_rate": 8.916239956826447e-06,
      "loss": 0.4065,
      "step": 3786
    },
    {
      "epoch": 0.23721010350929392,
      "grad_norm": 0.7873113190848684,
      "learning_rate": 8.915609215046332e-06,
      "loss": 0.4217,
      "step": 3787
    },
    {
      "epoch": 0.23727274150863623,
      "grad_norm": 0.8339480222097977,
      "learning_rate": 8.914978312099013e-06,
      "loss": 0.3953,
      "step": 3788
    },
    {
      "epoch": 0.2373353795079785,
      "grad_norm": 0.9096204570061515,
      "learning_rate": 8.914347248010465e-06,
      "loss": 0.458,
      "step": 3789
    },
    {
      "epoch": 0.2373980175073208,
      "grad_norm": 0.8595155959549242,
      "learning_rate": 8.913716022806653e-06,
      "loss": 0.4139,
      "step": 3790
    },
    {
      "epoch": 0.23746065550666312,
      "grad_norm": 0.7743715643824469,
      "learning_rate": 8.913084636513565e-06,
      "loss": 0.4233,
      "step": 3791
    },
    {
      "epoch": 0.23752329350600543,
      "grad_norm": 0.823921818747712,
      "learning_rate": 8.912453089157187e-06,
      "loss": 0.4237,
      "step": 3792
    },
    {
      "epoch": 0.2375859315053477,
      "grad_norm": 0.8553433689346553,
      "learning_rate": 8.911821380763513e-06,
      "loss": 0.4077,
      "step": 3793
    },
    {
      "epoch": 0.23764856950469002,
      "grad_norm": 0.8852992877950974,
      "learning_rate": 8.911189511358546e-06,
      "loss": 0.432,
      "step": 3794
    },
    {
      "epoch": 0.23771120750403232,
      "grad_norm": 0.8907109895283777,
      "learning_rate": 8.910557480968292e-06,
      "loss": 0.4385,
      "step": 3795
    },
    {
      "epoch": 0.23777384550337463,
      "grad_norm": 0.6926662343486706,
      "learning_rate": 8.909925289618767e-06,
      "loss": 0.4865,
      "step": 3796
    },
    {
      "epoch": 0.2378364835027169,
      "grad_norm": 0.9065537541015602,
      "learning_rate": 8.90929293733599e-06,
      "loss": 0.4528,
      "step": 3797
    },
    {
      "epoch": 0.23789912150205922,
      "grad_norm": 0.8664291846221461,
      "learning_rate": 8.90866042414599e-06,
      "loss": 0.4335,
      "step": 3798
    },
    {
      "epoch": 0.23796175950140153,
      "grad_norm": 0.8199647244571044,
      "learning_rate": 8.908027750074804e-06,
      "loss": 0.4253,
      "step": 3799
    },
    {
      "epoch": 0.23802439750074383,
      "grad_norm": 0.8206976811217591,
      "learning_rate": 8.907394915148468e-06,
      "loss": 0.4395,
      "step": 3800
    },
    {
      "epoch": 0.2380870355000861,
      "grad_norm": 0.8265379624953524,
      "learning_rate": 8.906761919393031e-06,
      "loss": 0.4159,
      "step": 3801
    },
    {
      "epoch": 0.23814967349942842,
      "grad_norm": 1.0872287858124354,
      "learning_rate": 8.906128762834551e-06,
      "loss": 0.4224,
      "step": 3802
    },
    {
      "epoch": 0.23821231149877073,
      "grad_norm": 0.7885165105810398,
      "learning_rate": 8.905495445499085e-06,
      "loss": 0.4607,
      "step": 3803
    },
    {
      "epoch": 0.23827494949811304,
      "grad_norm": 0.7257883668437796,
      "learning_rate": 8.904861967412702e-06,
      "loss": 0.4605,
      "step": 3804
    },
    {
      "epoch": 0.23833758749745534,
      "grad_norm": 0.8237033861063113,
      "learning_rate": 8.904228328601477e-06,
      "loss": 0.4237,
      "step": 3805
    },
    {
      "epoch": 0.23840022549679762,
      "grad_norm": 0.8121962622981318,
      "learning_rate": 8.903594529091486e-06,
      "loss": 0.4056,
      "step": 3806
    },
    {
      "epoch": 0.23846286349613993,
      "grad_norm": 0.9395083858249613,
      "learning_rate": 8.902960568908823e-06,
      "loss": 0.4705,
      "step": 3807
    },
    {
      "epoch": 0.23852550149548224,
      "grad_norm": 0.8151922642998696,
      "learning_rate": 8.902326448079578e-06,
      "loss": 0.4093,
      "step": 3808
    },
    {
      "epoch": 0.23858813949482455,
      "grad_norm": 0.7809194869351681,
      "learning_rate": 8.901692166629849e-06,
      "loss": 0.4082,
      "step": 3809
    },
    {
      "epoch": 0.23865077749416683,
      "grad_norm": 0.8914021784464983,
      "learning_rate": 8.901057724585748e-06,
      "loss": 0.4605,
      "step": 3810
    },
    {
      "epoch": 0.23871341549350913,
      "grad_norm": 0.7898755068372498,
      "learning_rate": 8.900423121973386e-06,
      "loss": 0.3815,
      "step": 3811
    },
    {
      "epoch": 0.23877605349285144,
      "grad_norm": 0.9428074098595759,
      "learning_rate": 8.899788358818887e-06,
      "loss": 0.4546,
      "step": 3812
    },
    {
      "epoch": 0.23883869149219375,
      "grad_norm": 0.8165775005709776,
      "learning_rate": 8.899153435148375e-06,
      "loss": 0.3969,
      "step": 3813
    },
    {
      "epoch": 0.23890132949153603,
      "grad_norm": 0.8885229584028427,
      "learning_rate": 8.898518350987982e-06,
      "loss": 0.4378,
      "step": 3814
    },
    {
      "epoch": 0.23896396749087834,
      "grad_norm": 0.85390972658363,
      "learning_rate": 8.897883106363852e-06,
      "loss": 0.425,
      "step": 3815
    },
    {
      "epoch": 0.23902660549022064,
      "grad_norm": 0.8061128446868445,
      "learning_rate": 8.897247701302127e-06,
      "loss": 0.4429,
      "step": 3816
    },
    {
      "epoch": 0.23908924348956295,
      "grad_norm": 0.8760738655043303,
      "learning_rate": 8.896612135828965e-06,
      "loss": 0.4088,
      "step": 3817
    },
    {
      "epoch": 0.23915188148890523,
      "grad_norm": 0.7886372460823756,
      "learning_rate": 8.895976409970525e-06,
      "loss": 0.4334,
      "step": 3818
    },
    {
      "epoch": 0.23921451948824754,
      "grad_norm": 0.8381486278530715,
      "learning_rate": 8.895340523752972e-06,
      "loss": 0.4354,
      "step": 3819
    },
    {
      "epoch": 0.23927715748758985,
      "grad_norm": 0.8351802877039732,
      "learning_rate": 8.89470447720248e-06,
      "loss": 0.4413,
      "step": 3820
    },
    {
      "epoch": 0.23933979548693216,
      "grad_norm": 0.869732731065243,
      "learning_rate": 8.894068270345228e-06,
      "loss": 0.4114,
      "step": 3821
    },
    {
      "epoch": 0.23940243348627444,
      "grad_norm": 0.8355683772359691,
      "learning_rate": 8.893431903207405e-06,
      "loss": 0.3825,
      "step": 3822
    },
    {
      "epoch": 0.23946507148561674,
      "grad_norm": 0.879670938967612,
      "learning_rate": 8.892795375815202e-06,
      "loss": 0.4482,
      "step": 3823
    },
    {
      "epoch": 0.23952770948495905,
      "grad_norm": 0.7740745987605056,
      "learning_rate": 8.892158688194818e-06,
      "loss": 0.3892,
      "step": 3824
    },
    {
      "epoch": 0.23959034748430136,
      "grad_norm": 0.7737580584148029,
      "learning_rate": 8.89152184037246e-06,
      "loss": 0.466,
      "step": 3825
    },
    {
      "epoch": 0.23965298548364367,
      "grad_norm": 0.9213460248229102,
      "learning_rate": 8.89088483237434e-06,
      "loss": 0.4933,
      "step": 3826
    },
    {
      "epoch": 0.23971562348298595,
      "grad_norm": 0.8807606404555246,
      "learning_rate": 8.89024766422668e-06,
      "loss": 0.4417,
      "step": 3827
    },
    {
      "epoch": 0.23977826148232825,
      "grad_norm": 0.8528797614810344,
      "learning_rate": 8.889610335955702e-06,
      "loss": 0.4783,
      "step": 3828
    },
    {
      "epoch": 0.23984089948167056,
      "grad_norm": 0.9273056398406895,
      "learning_rate": 8.888972847587643e-06,
      "loss": 0.448,
      "step": 3829
    },
    {
      "epoch": 0.23990353748101287,
      "grad_norm": 0.850272709066075,
      "learning_rate": 8.888335199148739e-06,
      "loss": 0.4271,
      "step": 3830
    },
    {
      "epoch": 0.23996617548035515,
      "grad_norm": 0.8031878318490221,
      "learning_rate": 8.887697390665236e-06,
      "loss": 0.4322,
      "step": 3831
    },
    {
      "epoch": 0.24002881347969746,
      "grad_norm": 0.9413499598439845,
      "learning_rate": 8.887059422163387e-06,
      "loss": 0.4602,
      "step": 3832
    },
    {
      "epoch": 0.24009145147903976,
      "grad_norm": 0.841742805368772,
      "learning_rate": 8.88642129366945e-06,
      "loss": 0.391,
      "step": 3833
    },
    {
      "epoch": 0.24015408947838207,
      "grad_norm": 0.6896283012288239,
      "learning_rate": 8.885783005209692e-06,
      "loss": 0.4876,
      "step": 3834
    },
    {
      "epoch": 0.24021672747772435,
      "grad_norm": 0.8247979890008593,
      "learning_rate": 8.885144556810384e-06,
      "loss": 0.4328,
      "step": 3835
    },
    {
      "epoch": 0.24027936547706666,
      "grad_norm": 0.7665878760302159,
      "learning_rate": 8.884505948497804e-06,
      "loss": 0.4839,
      "step": 3836
    },
    {
      "epoch": 0.24034200347640897,
      "grad_norm": 0.8594458316445888,
      "learning_rate": 8.883867180298238e-06,
      "loss": 0.4014,
      "step": 3837
    },
    {
      "epoch": 0.24040464147575127,
      "grad_norm": 0.8970851412262189,
      "learning_rate": 8.88322825223798e-06,
      "loss": 0.4351,
      "step": 3838
    },
    {
      "epoch": 0.24046727947509355,
      "grad_norm": 0.8368585322432915,
      "learning_rate": 8.882589164343325e-06,
      "loss": 0.4373,
      "step": 3839
    },
    {
      "epoch": 0.24052991747443586,
      "grad_norm": 0.8317616985592973,
      "learning_rate": 8.88194991664058e-06,
      "loss": 0.4279,
      "step": 3840
    },
    {
      "epoch": 0.24059255547377817,
      "grad_norm": 0.8669492161647292,
      "learning_rate": 8.881310509156056e-06,
      "loss": 0.4196,
      "step": 3841
    },
    {
      "epoch": 0.24065519347312048,
      "grad_norm": 0.88448827343652,
      "learning_rate": 8.88067094191607e-06,
      "loss": 0.4603,
      "step": 3842
    },
    {
      "epoch": 0.24071783147246276,
      "grad_norm": 0.8525556784536993,
      "learning_rate": 8.88003121494695e-06,
      "loss": 0.4428,
      "step": 3843
    },
    {
      "epoch": 0.24078046947180506,
      "grad_norm": 0.8698667669843686,
      "learning_rate": 8.879391328275022e-06,
      "loss": 0.4688,
      "step": 3844
    },
    {
      "epoch": 0.24084310747114737,
      "grad_norm": 0.8424058990224111,
      "learning_rate": 8.87875128192663e-06,
      "loss": 0.4361,
      "step": 3845
    },
    {
      "epoch": 0.24090574547048968,
      "grad_norm": 0.8237963731744432,
      "learning_rate": 8.878111075928114e-06,
      "loss": 0.4445,
      "step": 3846
    },
    {
      "epoch": 0.240968383469832,
      "grad_norm": 0.7522373810650228,
      "learning_rate": 8.877470710305825e-06,
      "loss": 0.4293,
      "step": 3847
    },
    {
      "epoch": 0.24103102146917427,
      "grad_norm": 0.8622563120630704,
      "learning_rate": 8.876830185086123e-06,
      "loss": 0.4454,
      "step": 3848
    },
    {
      "epoch": 0.24109365946851657,
      "grad_norm": 0.8317776958427996,
      "learning_rate": 8.87618950029537e-06,
      "loss": 0.4431,
      "step": 3849
    },
    {
      "epoch": 0.24115629746785888,
      "grad_norm": 0.8385817934868515,
      "learning_rate": 8.875548655959937e-06,
      "loss": 0.451,
      "step": 3850
    },
    {
      "epoch": 0.2412189354672012,
      "grad_norm": 0.9147669227784786,
      "learning_rate": 8.874907652106205e-06,
      "loss": 0.4573,
      "step": 3851
    },
    {
      "epoch": 0.24128157346654347,
      "grad_norm": 0.9099132211902602,
      "learning_rate": 8.874266488760553e-06,
      "loss": 0.4221,
      "step": 3852
    },
    {
      "epoch": 0.24134421146588578,
      "grad_norm": 0.8312882693080789,
      "learning_rate": 8.873625165949373e-06,
      "loss": 0.4198,
      "step": 3853
    },
    {
      "epoch": 0.24140684946522808,
      "grad_norm": 0.8428534923811094,
      "learning_rate": 8.872983683699064e-06,
      "loss": 0.3977,
      "step": 3854
    },
    {
      "epoch": 0.2414694874645704,
      "grad_norm": 0.6567939349573766,
      "learning_rate": 8.872342042036026e-06,
      "loss": 0.4666,
      "step": 3855
    },
    {
      "epoch": 0.24153212546391267,
      "grad_norm": 0.8974547084029709,
      "learning_rate": 8.87170024098667e-06,
      "loss": 0.4225,
      "step": 3856
    },
    {
      "epoch": 0.24159476346325498,
      "grad_norm": 0.8408053409769061,
      "learning_rate": 8.871058280577416e-06,
      "loss": 0.3992,
      "step": 3857
    },
    {
      "epoch": 0.2416574014625973,
      "grad_norm": 0.9040891501169723,
      "learning_rate": 8.870416160834683e-06,
      "loss": 0.4589,
      "step": 3858
    },
    {
      "epoch": 0.2417200394619396,
      "grad_norm": 0.8918791005733381,
      "learning_rate": 8.869773881784904e-06,
      "loss": 0.4386,
      "step": 3859
    },
    {
      "epoch": 0.24178267746128188,
      "grad_norm": 0.8198138321764742,
      "learning_rate": 8.869131443454512e-06,
      "loss": 0.3887,
      "step": 3860
    },
    {
      "epoch": 0.24184531546062418,
      "grad_norm": 0.9172244978384675,
      "learning_rate": 8.868488845869953e-06,
      "loss": 0.4619,
      "step": 3861
    },
    {
      "epoch": 0.2419079534599665,
      "grad_norm": 0.8073360904063039,
      "learning_rate": 8.867846089057675e-06,
      "loss": 0.4096,
      "step": 3862
    },
    {
      "epoch": 0.2419705914593088,
      "grad_norm": 0.8294842765906776,
      "learning_rate": 8.867203173044132e-06,
      "loss": 0.4333,
      "step": 3863
    },
    {
      "epoch": 0.24203322945865108,
      "grad_norm": 0.9279090155557428,
      "learning_rate": 8.86656009785579e-06,
      "loss": 0.4464,
      "step": 3864
    },
    {
      "epoch": 0.24209586745799339,
      "grad_norm": 0.8453373608632624,
      "learning_rate": 8.865916863519117e-06,
      "loss": 0.4684,
      "step": 3865
    },
    {
      "epoch": 0.2421585054573357,
      "grad_norm": 0.8891045435212308,
      "learning_rate": 8.865273470060588e-06,
      "loss": 0.4097,
      "step": 3866
    },
    {
      "epoch": 0.242221143456678,
      "grad_norm": 0.7612119776152719,
      "learning_rate": 8.864629917506686e-06,
      "loss": 0.4008,
      "step": 3867
    },
    {
      "epoch": 0.2422837814560203,
      "grad_norm": 0.8425358429077083,
      "learning_rate": 8.8639862058839e-06,
      "loss": 0.4165,
      "step": 3868
    },
    {
      "epoch": 0.2423464194553626,
      "grad_norm": 0.7205146577315554,
      "learning_rate": 8.863342335218722e-06,
      "loss": 0.4869,
      "step": 3869
    },
    {
      "epoch": 0.2424090574547049,
      "grad_norm": 0.8306622293801256,
      "learning_rate": 8.862698305537658e-06,
      "loss": 0.4315,
      "step": 3870
    },
    {
      "epoch": 0.2424716954540472,
      "grad_norm": 0.8114015418503143,
      "learning_rate": 8.862054116867215e-06,
      "loss": 0.4212,
      "step": 3871
    },
    {
      "epoch": 0.2425343334533895,
      "grad_norm": 0.9478172237007644,
      "learning_rate": 8.861409769233907e-06,
      "loss": 0.4434,
      "step": 3872
    },
    {
      "epoch": 0.2425969714527318,
      "grad_norm": 0.8143818867955306,
      "learning_rate": 8.860765262664257e-06,
      "loss": 0.4067,
      "step": 3873
    },
    {
      "epoch": 0.2426596094520741,
      "grad_norm": 0.8337477646800545,
      "learning_rate": 8.860120597184792e-06,
      "loss": 0.3998,
      "step": 3874
    },
    {
      "epoch": 0.2427222474514164,
      "grad_norm": 0.8837882693424783,
      "learning_rate": 8.859475772822047e-06,
      "loss": 0.4227,
      "step": 3875
    },
    {
      "epoch": 0.24278488545075871,
      "grad_norm": 0.8287012545766702,
      "learning_rate": 8.858830789602561e-06,
      "loss": 0.4351,
      "step": 3876
    },
    {
      "epoch": 0.242847523450101,
      "grad_norm": 0.8200347357771085,
      "learning_rate": 8.858185647552887e-06,
      "loss": 0.4495,
      "step": 3877
    },
    {
      "epoch": 0.2429101614494433,
      "grad_norm": 0.828852250148796,
      "learning_rate": 8.857540346699571e-06,
      "loss": 0.407,
      "step": 3878
    },
    {
      "epoch": 0.2429727994487856,
      "grad_norm": 0.8610270834975878,
      "learning_rate": 8.856894887069181e-06,
      "loss": 0.4395,
      "step": 3879
    },
    {
      "epoch": 0.24303543744812792,
      "grad_norm": 0.8388962460597877,
      "learning_rate": 8.856249268688282e-06,
      "loss": 0.4005,
      "step": 3880
    },
    {
      "epoch": 0.2430980754474702,
      "grad_norm": 0.783078199444167,
      "learning_rate": 8.855603491583447e-06,
      "loss": 0.4422,
      "step": 3881
    },
    {
      "epoch": 0.2431607134468125,
      "grad_norm": 0.8738444351118478,
      "learning_rate": 8.854957555781256e-06,
      "loss": 0.4254,
      "step": 3882
    },
    {
      "epoch": 0.2432233514461548,
      "grad_norm": 0.819727910904429,
      "learning_rate": 8.854311461308296e-06,
      "loss": 0.4421,
      "step": 3883
    },
    {
      "epoch": 0.24328598944549712,
      "grad_norm": 0.8080467786789121,
      "learning_rate": 8.853665208191163e-06,
      "loss": 0.4265,
      "step": 3884
    },
    {
      "epoch": 0.2433486274448394,
      "grad_norm": 0.8822159975668089,
      "learning_rate": 8.853018796456455e-06,
      "loss": 0.4599,
      "step": 3885
    },
    {
      "epoch": 0.2434112654441817,
      "grad_norm": 0.8325085342319247,
      "learning_rate": 8.852372226130775e-06,
      "loss": 0.4223,
      "step": 3886
    },
    {
      "epoch": 0.24347390344352401,
      "grad_norm": 0.8384478443931428,
      "learning_rate": 8.851725497240741e-06,
      "loss": 0.4168,
      "step": 3887
    },
    {
      "epoch": 0.24353654144286632,
      "grad_norm": 0.7911631499563913,
      "learning_rate": 8.85107860981297e-06,
      "loss": 0.497,
      "step": 3888
    },
    {
      "epoch": 0.2435991794422086,
      "grad_norm": 0.9388830742867762,
      "learning_rate": 8.850431563874089e-06,
      "loss": 0.4761,
      "step": 3889
    },
    {
      "epoch": 0.2436618174415509,
      "grad_norm": 0.7989900350442909,
      "learning_rate": 8.849784359450731e-06,
      "loss": 0.4249,
      "step": 3890
    },
    {
      "epoch": 0.24372445544089322,
      "grad_norm": 0.7632928168690947,
      "learning_rate": 8.849136996569534e-06,
      "loss": 0.3702,
      "step": 3891
    },
    {
      "epoch": 0.24378709344023552,
      "grad_norm": 0.7916218327649255,
      "learning_rate": 8.848489475257142e-06,
      "loss": 0.4327,
      "step": 3892
    },
    {
      "epoch": 0.24384973143957783,
      "grad_norm": 0.8818518745072749,
      "learning_rate": 8.84784179554021e-06,
      "loss": 0.4202,
      "step": 3893
    },
    {
      "epoch": 0.2439123694389201,
      "grad_norm": 0.8078180508906292,
      "learning_rate": 8.847193957445398e-06,
      "loss": 0.3919,
      "step": 3894
    },
    {
      "epoch": 0.24397500743826242,
      "grad_norm": 1.0134739599609854,
      "learning_rate": 8.846545960999364e-06,
      "loss": 0.384,
      "step": 3895
    },
    {
      "epoch": 0.24403764543760473,
      "grad_norm": 0.8365548186179433,
      "learning_rate": 8.845897806228786e-06,
      "loss": 0.44,
      "step": 3896
    },
    {
      "epoch": 0.24410028343694704,
      "grad_norm": 0.854168979261859,
      "learning_rate": 8.845249493160341e-06,
      "loss": 0.4857,
      "step": 3897
    },
    {
      "epoch": 0.24416292143628932,
      "grad_norm": 0.8773137580608138,
      "learning_rate": 8.844601021820713e-06,
      "loss": 0.3798,
      "step": 3898
    },
    {
      "epoch": 0.24422555943563162,
      "grad_norm": 0.7477792885677021,
      "learning_rate": 8.843952392236595e-06,
      "loss": 0.4886,
      "step": 3899
    },
    {
      "epoch": 0.24428819743497393,
      "grad_norm": 0.8225780492494403,
      "learning_rate": 8.843303604434682e-06,
      "loss": 0.4226,
      "step": 3900
    },
    {
      "epoch": 0.24435083543431624,
      "grad_norm": 0.8697849968139006,
      "learning_rate": 8.842654658441678e-06,
      "loss": 0.4135,
      "step": 3901
    },
    {
      "epoch": 0.24441347343365852,
      "grad_norm": 0.7062216234162791,
      "learning_rate": 8.842005554284296e-06,
      "loss": 0.4873,
      "step": 3902
    },
    {
      "epoch": 0.24447611143300083,
      "grad_norm": 0.8876808929294948,
      "learning_rate": 8.841356291989251e-06,
      "loss": 0.4538,
      "step": 3903
    },
    {
      "epoch": 0.24453874943234313,
      "grad_norm": 0.8615366244216608,
      "learning_rate": 8.84070687158327e-06,
      "loss": 0.4292,
      "step": 3904
    },
    {
      "epoch": 0.24460138743168544,
      "grad_norm": 0.891025666469548,
      "learning_rate": 8.84005729309308e-06,
      "loss": 0.4181,
      "step": 3905
    },
    {
      "epoch": 0.24466402543102772,
      "grad_norm": 0.8541436825031111,
      "learning_rate": 8.83940755654542e-06,
      "loss": 0.4767,
      "step": 3906
    },
    {
      "epoch": 0.24472666343037003,
      "grad_norm": 0.7656786724037616,
      "learning_rate": 8.838757661967033e-06,
      "loss": 0.3934,
      "step": 3907
    },
    {
      "epoch": 0.24478930142971234,
      "grad_norm": 0.893968216946324,
      "learning_rate": 8.838107609384666e-06,
      "loss": 0.4271,
      "step": 3908
    },
    {
      "epoch": 0.24485193942905464,
      "grad_norm": 0.7682434705703433,
      "learning_rate": 8.837457398825079e-06,
      "loss": 0.4907,
      "step": 3909
    },
    {
      "epoch": 0.24491457742839692,
      "grad_norm": 0.8020858450020409,
      "learning_rate": 8.836807030315032e-06,
      "loss": 0.4196,
      "step": 3910
    },
    {
      "epoch": 0.24497721542773923,
      "grad_norm": 0.7952655751421835,
      "learning_rate": 8.836156503881298e-06,
      "loss": 0.4125,
      "step": 3911
    },
    {
      "epoch": 0.24503985342708154,
      "grad_norm": 0.6440790070375955,
      "learning_rate": 8.835505819550647e-06,
      "loss": 0.4695,
      "step": 3912
    },
    {
      "epoch": 0.24510249142642385,
      "grad_norm": 0.6698288711652673,
      "learning_rate": 8.834854977349866e-06,
      "loss": 0.4876,
      "step": 3913
    },
    {
      "epoch": 0.24516512942576615,
      "grad_norm": 0.8294154264524439,
      "learning_rate": 8.834203977305741e-06,
      "loss": 0.3937,
      "step": 3914
    },
    {
      "epoch": 0.24522776742510843,
      "grad_norm": 0.8036901287644359,
      "learning_rate": 8.83355281944507e-06,
      "loss": 0.3912,
      "step": 3915
    },
    {
      "epoch": 0.24529040542445074,
      "grad_norm": 0.8595900797391635,
      "learning_rate": 8.832901503794656e-06,
      "loss": 0.3929,
      "step": 3916
    },
    {
      "epoch": 0.24535304342379305,
      "grad_norm": 0.828693740980089,
      "learning_rate": 8.8322500303813e-06,
      "loss": 0.4286,
      "step": 3917
    },
    {
      "epoch": 0.24541568142313536,
      "grad_norm": 0.7838308412862481,
      "learning_rate": 8.831598399231824e-06,
      "loss": 0.4285,
      "step": 3918
    },
    {
      "epoch": 0.24547831942247764,
      "grad_norm": 0.8505136004189834,
      "learning_rate": 8.830946610373046e-06,
      "loss": 0.4275,
      "step": 3919
    },
    {
      "epoch": 0.24554095742181994,
      "grad_norm": 0.8824651557035378,
      "learning_rate": 8.830294663831795e-06,
      "loss": 0.4496,
      "step": 3920
    },
    {
      "epoch": 0.24560359542116225,
      "grad_norm": 0.8057230806582947,
      "learning_rate": 8.829642559634905e-06,
      "loss": 0.3987,
      "step": 3921
    },
    {
      "epoch": 0.24566623342050456,
      "grad_norm": 0.8337911384859699,
      "learning_rate": 8.828990297809214e-06,
      "loss": 0.4294,
      "step": 3922
    },
    {
      "epoch": 0.24572887141984684,
      "grad_norm": 0.7511799382063948,
      "learning_rate": 8.828337878381575e-06,
      "loss": 0.3503,
      "step": 3923
    },
    {
      "epoch": 0.24579150941918915,
      "grad_norm": 0.8392438260652049,
      "learning_rate": 8.827685301378836e-06,
      "loss": 0.4345,
      "step": 3924
    },
    {
      "epoch": 0.24585414741853145,
      "grad_norm": 0.8301309115251826,
      "learning_rate": 8.82703256682786e-06,
      "loss": 0.4045,
      "step": 3925
    },
    {
      "epoch": 0.24591678541787376,
      "grad_norm": 0.8811628109442372,
      "learning_rate": 8.826379674755514e-06,
      "loss": 0.4426,
      "step": 3926
    },
    {
      "epoch": 0.24597942341721604,
      "grad_norm": 0.7210272937916996,
      "learning_rate": 8.825726625188671e-06,
      "loss": 0.3701,
      "step": 3927
    },
    {
      "epoch": 0.24604206141655835,
      "grad_norm": 0.797446591222875,
      "learning_rate": 8.825073418154209e-06,
      "loss": 0.4889,
      "step": 3928
    },
    {
      "epoch": 0.24610469941590066,
      "grad_norm": 0.8004556543304957,
      "learning_rate": 8.824420053679015e-06,
      "loss": 0.4024,
      "step": 3929
    },
    {
      "epoch": 0.24616733741524296,
      "grad_norm": 0.9398015098761298,
      "learning_rate": 8.823766531789982e-06,
      "loss": 0.4515,
      "step": 3930
    },
    {
      "epoch": 0.24622997541458524,
      "grad_norm": 0.84893555273298,
      "learning_rate": 8.823112852514007e-06,
      "loss": 0.3969,
      "step": 3931
    },
    {
      "epoch": 0.24629261341392755,
      "grad_norm": 0.8082357090096378,
      "learning_rate": 8.822459015878e-06,
      "loss": 0.454,
      "step": 3932
    },
    {
      "epoch": 0.24635525141326986,
      "grad_norm": 0.8149649099216556,
      "learning_rate": 8.82180502190887e-06,
      "loss": 0.3864,
      "step": 3933
    },
    {
      "epoch": 0.24641788941261217,
      "grad_norm": 0.8414220148987954,
      "learning_rate": 8.821150870633534e-06,
      "loss": 0.4383,
      "step": 3934
    },
    {
      "epoch": 0.24648052741195448,
      "grad_norm": 0.8917717845363813,
      "learning_rate": 8.820496562078921e-06,
      "loss": 0.4485,
      "step": 3935
    },
    {
      "epoch": 0.24654316541129676,
      "grad_norm": 0.8557713182163477,
      "learning_rate": 8.81984209627196e-06,
      "loss": 0.4433,
      "step": 3936
    },
    {
      "epoch": 0.24660580341063906,
      "grad_norm": 0.8605832071320939,
      "learning_rate": 8.81918747323959e-06,
      "loss": 0.4047,
      "step": 3937
    },
    {
      "epoch": 0.24666844140998137,
      "grad_norm": 0.8918633731066301,
      "learning_rate": 8.818532693008752e-06,
      "loss": 0.4309,
      "step": 3938
    },
    {
      "epoch": 0.24673107940932368,
      "grad_norm": 0.9052288434213177,
      "learning_rate": 8.817877755606403e-06,
      "loss": 0.4405,
      "step": 3939
    },
    {
      "epoch": 0.24679371740866596,
      "grad_norm": 0.8425531645817728,
      "learning_rate": 8.817222661059494e-06,
      "loss": 0.3848,
      "step": 3940
    },
    {
      "epoch": 0.24685635540800827,
      "grad_norm": 0.88802522596488,
      "learning_rate": 8.816567409394994e-06,
      "loss": 0.4767,
      "step": 3941
    },
    {
      "epoch": 0.24691899340735057,
      "grad_norm": 0.746073172058348,
      "learning_rate": 8.81591200063987e-06,
      "loss": 0.4817,
      "step": 3942
    },
    {
      "epoch": 0.24698163140669288,
      "grad_norm": 0.8839649860605019,
      "learning_rate": 8.8152564348211e-06,
      "loss": 0.4098,
      "step": 3943
    },
    {
      "epoch": 0.24704426940603516,
      "grad_norm": 0.8263046082179349,
      "learning_rate": 8.814600711965667e-06,
      "loss": 0.4048,
      "step": 3944
    },
    {
      "epoch": 0.24710690740537747,
      "grad_norm": 0.8417180032994523,
      "learning_rate": 8.81394483210056e-06,
      "loss": 0.415,
      "step": 3945
    },
    {
      "epoch": 0.24716954540471978,
      "grad_norm": 0.7769645628597653,
      "learning_rate": 8.813288795252778e-06,
      "loss": 0.3928,
      "step": 3946
    },
    {
      "epoch": 0.24723218340406208,
      "grad_norm": 0.8325042292079057,
      "learning_rate": 8.812632601449317e-06,
      "loss": 0.417,
      "step": 3947
    },
    {
      "epoch": 0.24729482140340436,
      "grad_norm": 0.8132390544068415,
      "learning_rate": 8.811976250717194e-06,
      "loss": 0.4517,
      "step": 3948
    },
    {
      "epoch": 0.24735745940274667,
      "grad_norm": 0.7449082219852897,
      "learning_rate": 8.811319743083421e-06,
      "loss": 0.4922,
      "step": 3949
    },
    {
      "epoch": 0.24742009740208898,
      "grad_norm": 0.8699520611751945,
      "learning_rate": 8.810663078575019e-06,
      "loss": 0.4217,
      "step": 3950
    },
    {
      "epoch": 0.2474827354014313,
      "grad_norm": 0.9082335699619455,
      "learning_rate": 8.810006257219017e-06,
      "loss": 0.4465,
      "step": 3951
    },
    {
      "epoch": 0.24754537340077357,
      "grad_norm": 0.9134804603508642,
      "learning_rate": 8.809349279042451e-06,
      "loss": 0.4327,
      "step": 3952
    },
    {
      "epoch": 0.24760801140011587,
      "grad_norm": 0.8338625076085205,
      "learning_rate": 8.808692144072362e-06,
      "loss": 0.4322,
      "step": 3953
    },
    {
      "epoch": 0.24767064939945818,
      "grad_norm": 0.8707962130788458,
      "learning_rate": 8.808034852335798e-06,
      "loss": 0.426,
      "step": 3954
    },
    {
      "epoch": 0.2477332873988005,
      "grad_norm": 0.8923298399780081,
      "learning_rate": 8.807377403859812e-06,
      "loss": 0.4187,
      "step": 3955
    },
    {
      "epoch": 0.2477959253981428,
      "grad_norm": 0.8489196890551522,
      "learning_rate": 8.806719798671464e-06,
      "loss": 0.4145,
      "step": 3956
    },
    {
      "epoch": 0.24785856339748508,
      "grad_norm": 0.860791437030712,
      "learning_rate": 8.806062036797823e-06,
      "loss": 0.3981,
      "step": 3957
    },
    {
      "epoch": 0.24792120139682738,
      "grad_norm": 0.7977683992983085,
      "learning_rate": 8.805404118265964e-06,
      "loss": 0.4438,
      "step": 3958
    },
    {
      "epoch": 0.2479838393961697,
      "grad_norm": 0.8874977226503682,
      "learning_rate": 8.804746043102964e-06,
      "loss": 0.427,
      "step": 3959
    },
    {
      "epoch": 0.248046477395512,
      "grad_norm": 0.8477491964339541,
      "learning_rate": 8.804087811335908e-06,
      "loss": 0.3959,
      "step": 3960
    },
    {
      "epoch": 0.24810911539485428,
      "grad_norm": 0.8516045610718661,
      "learning_rate": 8.803429422991895e-06,
      "loss": 0.4602,
      "step": 3961
    },
    {
      "epoch": 0.2481717533941966,
      "grad_norm": 0.8278297097410464,
      "learning_rate": 8.80277087809802e-06,
      "loss": 0.4298,
      "step": 3962
    },
    {
      "epoch": 0.2482343913935389,
      "grad_norm": 0.8588203406574173,
      "learning_rate": 8.80211217668139e-06,
      "loss": 0.4355,
      "step": 3963
    },
    {
      "epoch": 0.2482970293928812,
      "grad_norm": 0.8643087175432878,
      "learning_rate": 8.801453318769118e-06,
      "loss": 0.4318,
      "step": 3964
    },
    {
      "epoch": 0.24835966739222348,
      "grad_norm": 0.7685317436768062,
      "learning_rate": 8.80079430438832e-06,
      "loss": 0.4113,
      "step": 3965
    },
    {
      "epoch": 0.2484223053915658,
      "grad_norm": 0.8722506696107984,
      "learning_rate": 8.800135133566123e-06,
      "loss": 0.4224,
      "step": 3966
    },
    {
      "epoch": 0.2484849433909081,
      "grad_norm": 0.7940496521156916,
      "learning_rate": 8.799475806329661e-06,
      "loss": 0.4126,
      "step": 3967
    },
    {
      "epoch": 0.2485475813902504,
      "grad_norm": 0.7626570419214805,
      "learning_rate": 8.798816322706068e-06,
      "loss": 0.394,
      "step": 3968
    },
    {
      "epoch": 0.24861021938959268,
      "grad_norm": 0.75770252884847,
      "learning_rate": 8.79815668272249e-06,
      "loss": 0.5147,
      "step": 3969
    },
    {
      "epoch": 0.248672857388935,
      "grad_norm": 0.911203549575555,
      "learning_rate": 8.797496886406077e-06,
      "loss": 0.4127,
      "step": 3970
    },
    {
      "epoch": 0.2487354953882773,
      "grad_norm": 0.8467451446730104,
      "learning_rate": 8.796836933783989e-06,
      "loss": 0.4543,
      "step": 3971
    },
    {
      "epoch": 0.2487981333876196,
      "grad_norm": 0.7886040714026854,
      "learning_rate": 8.796176824883386e-06,
      "loss": 0.4115,
      "step": 3972
    },
    {
      "epoch": 0.2488607713869619,
      "grad_norm": 0.7834720706981753,
      "learning_rate": 8.795516559731443e-06,
      "loss": 0.4885,
      "step": 3973
    },
    {
      "epoch": 0.2489234093863042,
      "grad_norm": 0.8541896988055808,
      "learning_rate": 8.794856138355331e-06,
      "loss": 0.4252,
      "step": 3974
    },
    {
      "epoch": 0.2489860473856465,
      "grad_norm": 0.8234770796537271,
      "learning_rate": 8.79419556078224e-06,
      "loss": 0.4145,
      "step": 3975
    },
    {
      "epoch": 0.2490486853849888,
      "grad_norm": 0.8732682649350209,
      "learning_rate": 8.793534827039352e-06,
      "loss": 0.4057,
      "step": 3976
    },
    {
      "epoch": 0.24911132338433112,
      "grad_norm": 0.7907405192563969,
      "learning_rate": 8.792873937153868e-06,
      "loss": 0.4188,
      "step": 3977
    },
    {
      "epoch": 0.2491739613836734,
      "grad_norm": 0.7661231525999435,
      "learning_rate": 8.792212891152987e-06,
      "loss": 0.3806,
      "step": 3978
    },
    {
      "epoch": 0.2492365993830157,
      "grad_norm": 0.8574193404852932,
      "learning_rate": 8.791551689063922e-06,
      "loss": 0.4554,
      "step": 3979
    },
    {
      "epoch": 0.249299237382358,
      "grad_norm": 0.9210277924247867,
      "learning_rate": 8.790890330913884e-06,
      "loss": 0.4313,
      "step": 3980
    },
    {
      "epoch": 0.24936187538170032,
      "grad_norm": 0.8716864221713031,
      "learning_rate": 8.790228816730097e-06,
      "loss": 0.4004,
      "step": 3981
    },
    {
      "epoch": 0.2494245133810426,
      "grad_norm": 0.8863501967325914,
      "learning_rate": 8.789567146539788e-06,
      "loss": 0.4321,
      "step": 3982
    },
    {
      "epoch": 0.2494871513803849,
      "grad_norm": 0.9192102743791001,
      "learning_rate": 8.788905320370192e-06,
      "loss": 0.4168,
      "step": 3983
    },
    {
      "epoch": 0.24954978937972722,
      "grad_norm": 0.7888293444240405,
      "learning_rate": 8.78824333824855e-06,
      "loss": 0.3546,
      "step": 3984
    },
    {
      "epoch": 0.24961242737906952,
      "grad_norm": 0.8127054007122644,
      "learning_rate": 8.787581200202109e-06,
      "loss": 0.4249,
      "step": 3985
    },
    {
      "epoch": 0.2496750653784118,
      "grad_norm": 0.9123532451116468,
      "learning_rate": 8.786918906258123e-06,
      "loss": 0.4368,
      "step": 3986
    },
    {
      "epoch": 0.2497377033777541,
      "grad_norm": 0.8069692707048782,
      "learning_rate": 8.786256456443852e-06,
      "loss": 0.4341,
      "step": 3987
    },
    {
      "epoch": 0.24980034137709642,
      "grad_norm": 0.8260091602599452,
      "learning_rate": 8.785593850786562e-06,
      "loss": 0.4237,
      "step": 3988
    },
    {
      "epoch": 0.24986297937643873,
      "grad_norm": 0.8275867910886306,
      "learning_rate": 8.784931089313527e-06,
      "loss": 0.4308,
      "step": 3989
    },
    {
      "epoch": 0.249925617375781,
      "grad_norm": 0.8280817081755709,
      "learning_rate": 8.784268172052025e-06,
      "loss": 0.4348,
      "step": 3990
    },
    {
      "epoch": 0.2499882553751233,
      "grad_norm": 0.8525714043223429,
      "learning_rate": 8.783605099029343e-06,
      "loss": 0.3965,
      "step": 3991
    },
    {
      "epoch": 0.2500508933744656,
      "grad_norm": 0.8257929322325309,
      "learning_rate": 8.782941870272772e-06,
      "loss": 0.4339,
      "step": 3992
    },
    {
      "epoch": 0.25011353137380793,
      "grad_norm": 0.835901479929909,
      "learning_rate": 8.782278485809613e-06,
      "loss": 0.4468,
      "step": 3993
    },
    {
      "epoch": 0.25017616937315024,
      "grad_norm": 0.8730464664081355,
      "learning_rate": 8.78161494566717e-06,
      "loss": 0.4609,
      "step": 3994
    },
    {
      "epoch": 0.25023880737249254,
      "grad_norm": 0.8773300857719045,
      "learning_rate": 8.780951249872753e-06,
      "loss": 0.4115,
      "step": 3995
    },
    {
      "epoch": 0.2503014453718348,
      "grad_norm": 0.6778757188281711,
      "learning_rate": 8.78028739845368e-06,
      "loss": 0.4992,
      "step": 3996
    },
    {
      "epoch": 0.2503640833711771,
      "grad_norm": 0.8826481398221894,
      "learning_rate": 8.779623391437277e-06,
      "loss": 0.4332,
      "step": 3997
    },
    {
      "epoch": 0.2504267213705194,
      "grad_norm": 0.8486151805303769,
      "learning_rate": 8.778959228850873e-06,
      "loss": 0.3737,
      "step": 3998
    },
    {
      "epoch": 0.2504893593698617,
      "grad_norm": 0.8068130842353114,
      "learning_rate": 8.778294910721806e-06,
      "loss": 0.4153,
      "step": 3999
    },
    {
      "epoch": 0.250551997369204,
      "grad_norm": 0.8643179050700646,
      "learning_rate": 8.77763043707742e-06,
      "loss": 0.4377,
      "step": 4000
    },
    {
      "epoch": 0.25061463536854633,
      "grad_norm": 0.8337989094718706,
      "learning_rate": 8.776965807945064e-06,
      "loss": 0.3862,
      "step": 4001
    },
    {
      "epoch": 0.25067727336788864,
      "grad_norm": 0.8346370459932116,
      "learning_rate": 8.776301023352092e-06,
      "loss": 0.3882,
      "step": 4002
    },
    {
      "epoch": 0.25073991136723095,
      "grad_norm": 0.8607500658153516,
      "learning_rate": 8.775636083325871e-06,
      "loss": 0.4289,
      "step": 4003
    },
    {
      "epoch": 0.25080254936657326,
      "grad_norm": 0.8527968999257378,
      "learning_rate": 8.774970987893767e-06,
      "loss": 0.4761,
      "step": 4004
    },
    {
      "epoch": 0.2508651873659155,
      "grad_norm": 0.8606083304861996,
      "learning_rate": 8.774305737083158e-06,
      "loss": 0.4274,
      "step": 4005
    },
    {
      "epoch": 0.2509278253652578,
      "grad_norm": 0.8517970371708324,
      "learning_rate": 8.773640330921423e-06,
      "loss": 0.4224,
      "step": 4006
    },
    {
      "epoch": 0.2509904633646001,
      "grad_norm": 0.7871580714097057,
      "learning_rate": 8.772974769435952e-06,
      "loss": 0.4306,
      "step": 4007
    },
    {
      "epoch": 0.25105310136394243,
      "grad_norm": 0.8335593979709541,
      "learning_rate": 8.77230905265414e-06,
      "loss": 0.4382,
      "step": 4008
    },
    {
      "epoch": 0.25111573936328474,
      "grad_norm": 0.8397877536480824,
      "learning_rate": 8.771643180603386e-06,
      "loss": 0.424,
      "step": 4009
    },
    {
      "epoch": 0.25117837736262705,
      "grad_norm": 0.8321607123241023,
      "learning_rate": 8.770977153311098e-06,
      "loss": 0.4277,
      "step": 4010
    },
    {
      "epoch": 0.25124101536196936,
      "grad_norm": 0.8257959228212339,
      "learning_rate": 8.770310970804692e-06,
      "loss": 0.4299,
      "step": 4011
    },
    {
      "epoch": 0.25130365336131166,
      "grad_norm": 0.7027646229933975,
      "learning_rate": 8.769644633111587e-06,
      "loss": 0.5053,
      "step": 4012
    },
    {
      "epoch": 0.2513662913606539,
      "grad_norm": 0.8962008297561905,
      "learning_rate": 8.768978140259208e-06,
      "loss": 0.4248,
      "step": 4013
    },
    {
      "epoch": 0.2514289293599962,
      "grad_norm": 0.8635876235673597,
      "learning_rate": 8.768311492274989e-06,
      "loss": 0.4661,
      "step": 4014
    },
    {
      "epoch": 0.25149156735933853,
      "grad_norm": 0.6984922616904659,
      "learning_rate": 8.76764468918637e-06,
      "loss": 0.4815,
      "step": 4015
    },
    {
      "epoch": 0.25155420535868084,
      "grad_norm": 0.8361043977631712,
      "learning_rate": 8.766977731020797e-06,
      "loss": 0.3846,
      "step": 4016
    },
    {
      "epoch": 0.25161684335802315,
      "grad_norm": 0.9063223319797613,
      "learning_rate": 8.76631061780572e-06,
      "loss": 0.4589,
      "step": 4017
    },
    {
      "epoch": 0.25167948135736545,
      "grad_norm": 0.8109453657166967,
      "learning_rate": 8.7656433495686e-06,
      "loss": 0.4498,
      "step": 4018
    },
    {
      "epoch": 0.25174211935670776,
      "grad_norm": 0.8954964535853436,
      "learning_rate": 8.7649759263369e-06,
      "loss": 0.4001,
      "step": 4019
    },
    {
      "epoch": 0.25180475735605007,
      "grad_norm": 0.8532660204909573,
      "learning_rate": 8.764308348138094e-06,
      "loss": 0.4439,
      "step": 4020
    },
    {
      "epoch": 0.2518673953553924,
      "grad_norm": 0.8250954616622888,
      "learning_rate": 8.763640614999655e-06,
      "loss": 0.4517,
      "step": 4021
    },
    {
      "epoch": 0.25193003335473463,
      "grad_norm": 0.7871148052626863,
      "learning_rate": 8.762972726949072e-06,
      "loss": 0.4039,
      "step": 4022
    },
    {
      "epoch": 0.25199267135407694,
      "grad_norm": 0.8766716692447776,
      "learning_rate": 8.762304684013832e-06,
      "loss": 0.4627,
      "step": 4023
    },
    {
      "epoch": 0.25205530935341924,
      "grad_norm": 0.8329470379563066,
      "learning_rate": 8.761636486221433e-06,
      "loss": 0.4298,
      "step": 4024
    },
    {
      "epoch": 0.25211794735276155,
      "grad_norm": 0.9304834953066388,
      "learning_rate": 8.760968133599377e-06,
      "loss": 0.4579,
      "step": 4025
    },
    {
      "epoch": 0.25218058535210386,
      "grad_norm": 0.8155942807351162,
      "learning_rate": 8.760299626175176e-06,
      "loss": 0.3996,
      "step": 4026
    },
    {
      "epoch": 0.25224322335144617,
      "grad_norm": 0.9068826138902492,
      "learning_rate": 8.759630963976344e-06,
      "loss": 0.4098,
      "step": 4027
    },
    {
      "epoch": 0.2523058613507885,
      "grad_norm": 0.8183351615919899,
      "learning_rate": 8.758962147030403e-06,
      "loss": 0.5059,
      "step": 4028
    },
    {
      "epoch": 0.2523684993501308,
      "grad_norm": 0.7434116897236949,
      "learning_rate": 8.758293175364883e-06,
      "loss": 0.4833,
      "step": 4029
    },
    {
      "epoch": 0.25243113734947303,
      "grad_norm": 0.7202063671348433,
      "learning_rate": 8.757624049007318e-06,
      "loss": 0.4749,
      "step": 4030
    },
    {
      "epoch": 0.25249377534881534,
      "grad_norm": 0.9249141937085156,
      "learning_rate": 8.756954767985251e-06,
      "loss": 0.4474,
      "step": 4031
    },
    {
      "epoch": 0.25255641334815765,
      "grad_norm": 0.9266192695867204,
      "learning_rate": 8.756285332326228e-06,
      "loss": 0.4409,
      "step": 4032
    },
    {
      "epoch": 0.25261905134749996,
      "grad_norm": 0.8239882531541107,
      "learning_rate": 8.755615742057803e-06,
      "loss": 0.4245,
      "step": 4033
    },
    {
      "epoch": 0.25268168934684226,
      "grad_norm": 0.8387740152787051,
      "learning_rate": 8.754945997207536e-06,
      "loss": 0.4273,
      "step": 4034
    },
    {
      "epoch": 0.25274432734618457,
      "grad_norm": 0.8363618110387796,
      "learning_rate": 8.754276097802996e-06,
      "loss": 0.4024,
      "step": 4035
    },
    {
      "epoch": 0.2528069653455269,
      "grad_norm": 0.9088609948881546,
      "learning_rate": 8.753606043871755e-06,
      "loss": 0.423,
      "step": 4036
    },
    {
      "epoch": 0.2528696033448692,
      "grad_norm": 0.8616980330135126,
      "learning_rate": 8.752935835441392e-06,
      "loss": 0.422,
      "step": 4037
    },
    {
      "epoch": 0.25293224134421144,
      "grad_norm": 0.8090802892109882,
      "learning_rate": 8.752265472539493e-06,
      "loss": 0.4214,
      "step": 4038
    },
    {
      "epoch": 0.25299487934355375,
      "grad_norm": 0.8153324212532123,
      "learning_rate": 8.751594955193654e-06,
      "loss": 0.3816,
      "step": 4039
    },
    {
      "epoch": 0.25305751734289605,
      "grad_norm": 0.8653924966583726,
      "learning_rate": 8.750924283431466e-06,
      "loss": 0.4371,
      "step": 4040
    },
    {
      "epoch": 0.25312015534223836,
      "grad_norm": 0.8468946572355669,
      "learning_rate": 8.75025345728054e-06,
      "loss": 0.4253,
      "step": 4041
    },
    {
      "epoch": 0.25318279334158067,
      "grad_norm": 0.80847526608545,
      "learning_rate": 8.749582476768485e-06,
      "loss": 0.4755,
      "step": 4042
    },
    {
      "epoch": 0.253245431340923,
      "grad_norm": 0.840014494452502,
      "learning_rate": 8.748911341922922e-06,
      "loss": 0.3531,
      "step": 4043
    },
    {
      "epoch": 0.2533080693402653,
      "grad_norm": 0.8175760762724443,
      "learning_rate": 8.748240052771469e-06,
      "loss": 0.4451,
      "step": 4044
    },
    {
      "epoch": 0.2533707073396076,
      "grad_norm": 0.8131004741833516,
      "learning_rate": 8.747568609341762e-06,
      "loss": 0.4079,
      "step": 4045
    },
    {
      "epoch": 0.2534333453389499,
      "grad_norm": 0.8363344982493179,
      "learning_rate": 8.746897011661436e-06,
      "loss": 0.4269,
      "step": 4046
    },
    {
      "epoch": 0.25349598333829215,
      "grad_norm": 0.8297624750483213,
      "learning_rate": 8.74622525975813e-06,
      "loss": 0.4243,
      "step": 4047
    },
    {
      "epoch": 0.25355862133763446,
      "grad_norm": 1.1296446521804464,
      "learning_rate": 8.7455533536595e-06,
      "loss": 0.5161,
      "step": 4048
    },
    {
      "epoch": 0.25362125933697677,
      "grad_norm": 0.885714755518661,
      "learning_rate": 8.744881293393197e-06,
      "loss": 0.4163,
      "step": 4049
    },
    {
      "epoch": 0.2536838973363191,
      "grad_norm": 0.9654579974089538,
      "learning_rate": 8.744209078986884e-06,
      "loss": 0.4398,
      "step": 4050
    },
    {
      "epoch": 0.2537465353356614,
      "grad_norm": 0.8834356055245031,
      "learning_rate": 8.743536710468233e-06,
      "loss": 0.4029,
      "step": 4051
    },
    {
      "epoch": 0.2538091733350037,
      "grad_norm": 0.8036281498217388,
      "learning_rate": 8.742864187864915e-06,
      "loss": 0.3972,
      "step": 4052
    },
    {
      "epoch": 0.253871811334346,
      "grad_norm": 0.8433060021172315,
      "learning_rate": 8.742191511204614e-06,
      "loss": 0.3942,
      "step": 4053
    },
    {
      "epoch": 0.2539344493336883,
      "grad_norm": 0.8503594381338249,
      "learning_rate": 8.741518680515013e-06,
      "loss": 0.4612,
      "step": 4054
    },
    {
      "epoch": 0.25399708733303056,
      "grad_norm": 0.87309049906947,
      "learning_rate": 8.74084569582381e-06,
      "loss": 0.4362,
      "step": 4055
    },
    {
      "epoch": 0.25405972533237287,
      "grad_norm": 0.8189840997711705,
      "learning_rate": 8.740172557158704e-06,
      "loss": 0.4173,
      "step": 4056
    },
    {
      "epoch": 0.2541223633317152,
      "grad_norm": 0.8805397433437789,
      "learning_rate": 8.7394992645474e-06,
      "loss": 0.399,
      "step": 4057
    },
    {
      "epoch": 0.2541850013310575,
      "grad_norm": 0.8258244002764825,
      "learning_rate": 8.738825818017615e-06,
      "loss": 0.4161,
      "step": 4058
    },
    {
      "epoch": 0.2542476393303998,
      "grad_norm": 0.8937130087385429,
      "learning_rate": 8.738152217597063e-06,
      "loss": 0.4017,
      "step": 4059
    },
    {
      "epoch": 0.2543102773297421,
      "grad_norm": 0.8783038488295667,
      "learning_rate": 8.737478463313472e-06,
      "loss": 0.4531,
      "step": 4060
    },
    {
      "epoch": 0.2543729153290844,
      "grad_norm": 0.9347821276731777,
      "learning_rate": 8.736804555194574e-06,
      "loss": 0.4315,
      "step": 4061
    },
    {
      "epoch": 0.2544355533284267,
      "grad_norm": 0.8367019681998954,
      "learning_rate": 8.736130493268108e-06,
      "loss": 0.397,
      "step": 4062
    },
    {
      "epoch": 0.25449819132776896,
      "grad_norm": 0.862641869813239,
      "learning_rate": 8.735456277561817e-06,
      "loss": 0.4265,
      "step": 4063
    },
    {
      "epoch": 0.25456082932711127,
      "grad_norm": 0.8167908684271009,
      "learning_rate": 8.734781908103451e-06,
      "loss": 0.4229,
      "step": 4064
    },
    {
      "epoch": 0.2546234673264536,
      "grad_norm": 0.852534272003623,
      "learning_rate": 8.734107384920771e-06,
      "loss": 0.4135,
      "step": 4065
    },
    {
      "epoch": 0.2546861053257959,
      "grad_norm": 0.8614369110999623,
      "learning_rate": 8.733432708041537e-06,
      "loss": 0.4216,
      "step": 4066
    },
    {
      "epoch": 0.2547487433251382,
      "grad_norm": 0.7902745276836605,
      "learning_rate": 8.732757877493519e-06,
      "loss": 0.4132,
      "step": 4067
    },
    {
      "epoch": 0.2548113813244805,
      "grad_norm": 0.7640101407326391,
      "learning_rate": 8.732082893304492e-06,
      "loss": 0.3884,
      "step": 4068
    },
    {
      "epoch": 0.2548740193238228,
      "grad_norm": 0.8751282280712543,
      "learning_rate": 8.731407755502243e-06,
      "loss": 0.4235,
      "step": 4069
    },
    {
      "epoch": 0.2549366573231651,
      "grad_norm": 0.8107143154453778,
      "learning_rate": 8.730732464114558e-06,
      "loss": 0.4183,
      "step": 4070
    },
    {
      "epoch": 0.2549992953225074,
      "grad_norm": 0.8034483967532228,
      "learning_rate": 8.730057019169232e-06,
      "loss": 0.386,
      "step": 4071
    },
    {
      "epoch": 0.2550619333218497,
      "grad_norm": 0.7752438439491688,
      "learning_rate": 8.729381420694066e-06,
      "loss": 0.3959,
      "step": 4072
    },
    {
      "epoch": 0.255124571321192,
      "grad_norm": 0.9525373461801837,
      "learning_rate": 8.728705668716871e-06,
      "loss": 0.4782,
      "step": 4073
    },
    {
      "epoch": 0.2551872093205343,
      "grad_norm": 0.9043844522297888,
      "learning_rate": 8.728029763265456e-06,
      "loss": 0.4127,
      "step": 4074
    },
    {
      "epoch": 0.2552498473198766,
      "grad_norm": 0.8540554133572085,
      "learning_rate": 8.727353704367646e-06,
      "loss": 0.4268,
      "step": 4075
    },
    {
      "epoch": 0.2553124853192189,
      "grad_norm": 0.7951614755917567,
      "learning_rate": 8.726677492051262e-06,
      "loss": 0.392,
      "step": 4076
    },
    {
      "epoch": 0.2553751233185612,
      "grad_norm": 0.8575671118462527,
      "learning_rate": 8.726001126344143e-06,
      "loss": 0.4676,
      "step": 4077
    },
    {
      "epoch": 0.2554377613179035,
      "grad_norm": 0.7534835449443222,
      "learning_rate": 8.725324607274129e-06,
      "loss": 0.4867,
      "step": 4078
    },
    {
      "epoch": 0.25550039931724583,
      "grad_norm": 0.8958933068182268,
      "learning_rate": 8.724647934869059e-06,
      "loss": 0.4287,
      "step": 4079
    },
    {
      "epoch": 0.2555630373165881,
      "grad_norm": 0.8921530702934388,
      "learning_rate": 8.723971109156789e-06,
      "loss": 0.4121,
      "step": 4080
    },
    {
      "epoch": 0.2556256753159304,
      "grad_norm": 0.8842527029377476,
      "learning_rate": 8.723294130165178e-06,
      "loss": 0.4424,
      "step": 4081
    },
    {
      "epoch": 0.2556883133152727,
      "grad_norm": 0.8672251271798975,
      "learning_rate": 8.72261699792209e-06,
      "loss": 0.4439,
      "step": 4082
    },
    {
      "epoch": 0.255750951314615,
      "grad_norm": 0.8174245536099849,
      "learning_rate": 8.721939712455394e-06,
      "loss": 0.3894,
      "step": 4083
    },
    {
      "epoch": 0.2558135893139573,
      "grad_norm": 0.7754546666573133,
      "learning_rate": 8.72126227379297e-06,
      "loss": 0.4306,
      "step": 4084
    },
    {
      "epoch": 0.2558762273132996,
      "grad_norm": 0.8044210960035963,
      "learning_rate": 8.720584681962698e-06,
      "loss": 0.3881,
      "step": 4085
    },
    {
      "epoch": 0.2559388653126419,
      "grad_norm": 0.7987131786574133,
      "learning_rate": 8.719906936992472e-06,
      "loss": 0.3635,
      "step": 4086
    },
    {
      "epoch": 0.25600150331198424,
      "grad_norm": 0.8844059442113197,
      "learning_rate": 8.719229038910183e-06,
      "loss": 0.415,
      "step": 4087
    },
    {
      "epoch": 0.25606414131132654,
      "grad_norm": 0.7987504473841601,
      "learning_rate": 8.71855098774374e-06,
      "loss": 0.436,
      "step": 4088
    },
    {
      "epoch": 0.2561267793106688,
      "grad_norm": 0.9030998077054236,
      "learning_rate": 8.717872783521048e-06,
      "loss": 0.4006,
      "step": 4089
    },
    {
      "epoch": 0.2561894173100111,
      "grad_norm": 0.8946249842989602,
      "learning_rate": 8.71719442627002e-06,
      "loss": 0.47,
      "step": 4090
    },
    {
      "epoch": 0.2562520553093534,
      "grad_norm": 0.8510250742048954,
      "learning_rate": 8.716515916018581e-06,
      "loss": 0.4336,
      "step": 4091
    },
    {
      "epoch": 0.2563146933086957,
      "grad_norm": 0.8498914150960122,
      "learning_rate": 8.715837252794657e-06,
      "loss": 0.4152,
      "step": 4092
    },
    {
      "epoch": 0.256377331308038,
      "grad_norm": 0.8627818081420423,
      "learning_rate": 8.715158436626181e-06,
      "loss": 0.397,
      "step": 4093
    },
    {
      "epoch": 0.25643996930738033,
      "grad_norm": 0.8400234555582621,
      "learning_rate": 8.714479467541094e-06,
      "loss": 0.39,
      "step": 4094
    },
    {
      "epoch": 0.25650260730672264,
      "grad_norm": 0.901751824913567,
      "learning_rate": 8.713800345567345e-06,
      "loss": 0.4521,
      "step": 4095
    },
    {
      "epoch": 0.25656524530606495,
      "grad_norm": 0.8410610868140957,
      "learning_rate": 8.713121070732881e-06,
      "loss": 0.4268,
      "step": 4096
    },
    {
      "epoch": 0.2566278833054072,
      "grad_norm": 0.9228303409803958,
      "learning_rate": 8.712441643065667e-06,
      "loss": 0.4501,
      "step": 4097
    },
    {
      "epoch": 0.2566905213047495,
      "grad_norm": 0.8906105603297703,
      "learning_rate": 8.711762062593664e-06,
      "loss": 0.4607,
      "step": 4098
    },
    {
      "epoch": 0.2567531593040918,
      "grad_norm": 0.8203275100898164,
      "learning_rate": 8.71108232934485e-06,
      "loss": 0.4099,
      "step": 4099
    },
    {
      "epoch": 0.2568157973034341,
      "grad_norm": 0.8296190108190206,
      "learning_rate": 8.710402443347195e-06,
      "loss": 0.4517,
      "step": 4100
    },
    {
      "epoch": 0.25687843530277643,
      "grad_norm": 0.8018344273346267,
      "learning_rate": 8.709722404628687e-06,
      "loss": 0.3913,
      "step": 4101
    },
    {
      "epoch": 0.25694107330211874,
      "grad_norm": 0.9259044836392691,
      "learning_rate": 8.709042213217316e-06,
      "loss": 0.4171,
      "step": 4102
    },
    {
      "epoch": 0.25700371130146105,
      "grad_norm": 0.8666464347723971,
      "learning_rate": 8.70836186914108e-06,
      "loss": 0.4175,
      "step": 4103
    },
    {
      "epoch": 0.25706634930080335,
      "grad_norm": 0.8607833134334429,
      "learning_rate": 8.707681372427979e-06,
      "loss": 0.4402,
      "step": 4104
    },
    {
      "epoch": 0.2571289873001456,
      "grad_norm": 0.7973261926806484,
      "learning_rate": 8.707000723106027e-06,
      "loss": 0.4139,
      "step": 4105
    },
    {
      "epoch": 0.2571916252994879,
      "grad_norm": 0.8448986245044621,
      "learning_rate": 8.706319921203235e-06,
      "loss": 0.4443,
      "step": 4106
    },
    {
      "epoch": 0.2572542632988302,
      "grad_norm": 0.7633207886701704,
      "learning_rate": 8.70563896674763e-06,
      "loss": 0.3917,
      "step": 4107
    },
    {
      "epoch": 0.25731690129817253,
      "grad_norm": 0.9034984782431605,
      "learning_rate": 8.704957859767235e-06,
      "loss": 0.4848,
      "step": 4108
    },
    {
      "epoch": 0.25737953929751484,
      "grad_norm": 0.8526934486911246,
      "learning_rate": 8.704276600290088e-06,
      "loss": 0.4485,
      "step": 4109
    },
    {
      "epoch": 0.25744217729685714,
      "grad_norm": 0.8818459648778565,
      "learning_rate": 8.703595188344229e-06,
      "loss": 0.4285,
      "step": 4110
    },
    {
      "epoch": 0.25750481529619945,
      "grad_norm": 0.8925638947081126,
      "learning_rate": 8.702913623957703e-06,
      "loss": 0.4528,
      "step": 4111
    },
    {
      "epoch": 0.25756745329554176,
      "grad_norm": 0.8201131107101752,
      "learning_rate": 8.702231907158566e-06,
      "loss": 0.4619,
      "step": 4112
    },
    {
      "epoch": 0.25763009129488407,
      "grad_norm": 0.8520130580982318,
      "learning_rate": 8.701550037974877e-06,
      "loss": 0.462,
      "step": 4113
    },
    {
      "epoch": 0.2576927292942263,
      "grad_norm": 0.7681098160989638,
      "learning_rate": 8.7008680164347e-06,
      "loss": 0.4055,
      "step": 4114
    },
    {
      "epoch": 0.2577553672935686,
      "grad_norm": 0.8645986946154711,
      "learning_rate": 8.70018584256611e-06,
      "loss": 0.4149,
      "step": 4115
    },
    {
      "epoch": 0.25781800529291093,
      "grad_norm": 0.8262245243114451,
      "learning_rate": 8.699503516397184e-06,
      "loss": 0.4056,
      "step": 4116
    },
    {
      "epoch": 0.25788064329225324,
      "grad_norm": 0.8466517891488097,
      "learning_rate": 8.698821037956005e-06,
      "loss": 0.4356,
      "step": 4117
    },
    {
      "epoch": 0.25794328129159555,
      "grad_norm": 0.8633767376268112,
      "learning_rate": 8.698138407270666e-06,
      "loss": 0.4784,
      "step": 4118
    },
    {
      "epoch": 0.25800591929093786,
      "grad_norm": 0.8297454861892111,
      "learning_rate": 8.697455624369262e-06,
      "loss": 0.4203,
      "step": 4119
    },
    {
      "epoch": 0.25806855729028016,
      "grad_norm": 0.8107553953232715,
      "learning_rate": 8.696772689279903e-06,
      "loss": 0.4183,
      "step": 4120
    },
    {
      "epoch": 0.2581311952896225,
      "grad_norm": 0.8562540706474615,
      "learning_rate": 8.696089602030691e-06,
      "loss": 0.4328,
      "step": 4121
    },
    {
      "epoch": 0.2581938332889647,
      "grad_norm": 0.8484250509469583,
      "learning_rate": 8.695406362649744e-06,
      "loss": 0.4586,
      "step": 4122
    },
    {
      "epoch": 0.25825647128830703,
      "grad_norm": 0.9030155153604227,
      "learning_rate": 8.694722971165187e-06,
      "loss": 0.45,
      "step": 4123
    },
    {
      "epoch": 0.25831910928764934,
      "grad_norm": 0.9004623687470886,
      "learning_rate": 8.694039427605146e-06,
      "loss": 0.4989,
      "step": 4124
    },
    {
      "epoch": 0.25838174728699165,
      "grad_norm": 0.8115540089432384,
      "learning_rate": 8.693355731997757e-06,
      "loss": 0.4403,
      "step": 4125
    },
    {
      "epoch": 0.25844438528633396,
      "grad_norm": 1.0268228430599038,
      "learning_rate": 8.69267188437116e-06,
      "loss": 0.4844,
      "step": 4126
    },
    {
      "epoch": 0.25850702328567626,
      "grad_norm": 0.7937489705501786,
      "learning_rate": 8.691987884753503e-06,
      "loss": 0.4266,
      "step": 4127
    },
    {
      "epoch": 0.25856966128501857,
      "grad_norm": 0.8950093623928707,
      "learning_rate": 8.69130373317294e-06,
      "loss": 0.444,
      "step": 4128
    },
    {
      "epoch": 0.2586322992843609,
      "grad_norm": 0.8037600332349116,
      "learning_rate": 8.69061942965763e-06,
      "loss": 0.3983,
      "step": 4129
    },
    {
      "epoch": 0.2586949372837032,
      "grad_norm": 0.7765868026213463,
      "learning_rate": 8.689934974235738e-06,
      "loss": 0.4318,
      "step": 4130
    },
    {
      "epoch": 0.25875757528304544,
      "grad_norm": 0.842859437513257,
      "learning_rate": 8.689250366935438e-06,
      "loss": 0.4502,
      "step": 4131
    },
    {
      "epoch": 0.25882021328238775,
      "grad_norm": 0.8427641235286387,
      "learning_rate": 8.68856560778491e-06,
      "loss": 0.4371,
      "step": 4132
    },
    {
      "epoch": 0.25888285128173005,
      "grad_norm": 1.2188537247414912,
      "learning_rate": 8.687880696812334e-06,
      "loss": 0.5379,
      "step": 4133
    },
    {
      "epoch": 0.25894548928107236,
      "grad_norm": 0.9177934193737616,
      "learning_rate": 8.687195634045906e-06,
      "loss": 0.4095,
      "step": 4134
    },
    {
      "epoch": 0.25900812728041467,
      "grad_norm": 0.8325167373275758,
      "learning_rate": 8.686510419513822e-06,
      "loss": 0.4277,
      "step": 4135
    },
    {
      "epoch": 0.259070765279757,
      "grad_norm": 0.8706844256043668,
      "learning_rate": 8.685825053244284e-06,
      "loss": 0.4707,
      "step": 4136
    },
    {
      "epoch": 0.2591334032790993,
      "grad_norm": 0.8675673277198752,
      "learning_rate": 8.685139535265502e-06,
      "loss": 0.438,
      "step": 4137
    },
    {
      "epoch": 0.2591960412784416,
      "grad_norm": 0.7496365438447595,
      "learning_rate": 8.684453865605692e-06,
      "loss": 0.4058,
      "step": 4138
    },
    {
      "epoch": 0.25925867927778384,
      "grad_norm": 0.7999650246923696,
      "learning_rate": 8.68376804429308e-06,
      "loss": 0.4188,
      "step": 4139
    },
    {
      "epoch": 0.25932131727712615,
      "grad_norm": 0.8226388059361087,
      "learning_rate": 8.68308207135589e-06,
      "loss": 0.4201,
      "step": 4140
    },
    {
      "epoch": 0.25938395527646846,
      "grad_norm": 0.8540728374532782,
      "learning_rate": 8.682395946822358e-06,
      "loss": 0.4376,
      "step": 4141
    },
    {
      "epoch": 0.25944659327581077,
      "grad_norm": 0.8687466528594343,
      "learning_rate": 8.681709670720725e-06,
      "loss": 0.4068,
      "step": 4142
    },
    {
      "epoch": 0.2595092312751531,
      "grad_norm": 0.9396313935528866,
      "learning_rate": 8.681023243079238e-06,
      "loss": 0.459,
      "step": 4143
    },
    {
      "epoch": 0.2595718692744954,
      "grad_norm": 0.8090497826409534,
      "learning_rate": 8.680336663926155e-06,
      "loss": 0.4176,
      "step": 4144
    },
    {
      "epoch": 0.2596345072738377,
      "grad_norm": 0.8927850723462247,
      "learning_rate": 8.679649933289727e-06,
      "loss": 0.462,
      "step": 4145
    },
    {
      "epoch": 0.25969714527318,
      "grad_norm": 0.7953404774031424,
      "learning_rate": 8.678963051198227e-06,
      "loss": 0.391,
      "step": 4146
    },
    {
      "epoch": 0.25975978327252225,
      "grad_norm": 0.8626296491290686,
      "learning_rate": 8.678276017679924e-06,
      "loss": 0.4224,
      "step": 4147
    },
    {
      "epoch": 0.25982242127186456,
      "grad_norm": 0.880460625055956,
      "learning_rate": 8.6775888327631e-06,
      "loss": 0.453,
      "step": 4148
    },
    {
      "epoch": 0.25988505927120686,
      "grad_norm": 0.8833298815845291,
      "learning_rate": 8.676901496476034e-06,
      "loss": 0.4149,
      "step": 4149
    },
    {
      "epoch": 0.25994769727054917,
      "grad_norm": 0.8164909467232022,
      "learning_rate": 8.676214008847022e-06,
      "loss": 0.43,
      "step": 4150
    },
    {
      "epoch": 0.2600103352698915,
      "grad_norm": 0.8793905356466585,
      "learning_rate": 8.675526369904356e-06,
      "loss": 0.4763,
      "step": 4151
    },
    {
      "epoch": 0.2600729732692338,
      "grad_norm": 0.8885702701468139,
      "learning_rate": 8.674838579676346e-06,
      "loss": 0.48,
      "step": 4152
    },
    {
      "epoch": 0.2601356112685761,
      "grad_norm": 0.7966827805968251,
      "learning_rate": 8.674150638191296e-06,
      "loss": 0.4047,
      "step": 4153
    },
    {
      "epoch": 0.2601982492679184,
      "grad_norm": 0.7951338182888199,
      "learning_rate": 8.673462545477524e-06,
      "loss": 0.4392,
      "step": 4154
    },
    {
      "epoch": 0.2602608872672607,
      "grad_norm": 0.772022276820005,
      "learning_rate": 8.672774301563353e-06,
      "loss": 0.4027,
      "step": 4155
    },
    {
      "epoch": 0.26032352526660296,
      "grad_norm": 0.8044297475841337,
      "learning_rate": 8.67208590647711e-06,
      "loss": 0.4822,
      "step": 4156
    },
    {
      "epoch": 0.26038616326594527,
      "grad_norm": 0.8545097982104368,
      "learning_rate": 8.671397360247127e-06,
      "loss": 0.4199,
      "step": 4157
    },
    {
      "epoch": 0.2604488012652876,
      "grad_norm": 0.8375375564498295,
      "learning_rate": 8.67070866290175e-06,
      "loss": 0.417,
      "step": 4158
    },
    {
      "epoch": 0.2605114392646299,
      "grad_norm": 0.8347396284269828,
      "learning_rate": 8.670019814469322e-06,
      "loss": 0.4223,
      "step": 4159
    },
    {
      "epoch": 0.2605740772639722,
      "grad_norm": 0.7664141545550234,
      "learning_rate": 8.669330814978197e-06,
      "loss": 0.4044,
      "step": 4160
    },
    {
      "epoch": 0.2606367152633145,
      "grad_norm": 0.8658851472236724,
      "learning_rate": 8.668641664456734e-06,
      "loss": 0.4424,
      "step": 4161
    },
    {
      "epoch": 0.2606993532626568,
      "grad_norm": 0.8073978696710217,
      "learning_rate": 8.6679523629333e-06,
      "loss": 0.3885,
      "step": 4162
    },
    {
      "epoch": 0.2607619912619991,
      "grad_norm": 0.8563867734924991,
      "learning_rate": 8.667262910436267e-06,
      "loss": 0.4602,
      "step": 4163
    },
    {
      "epoch": 0.26082462926134137,
      "grad_norm": 0.9503621934285358,
      "learning_rate": 8.66657330699401e-06,
      "loss": 0.4737,
      "step": 4164
    },
    {
      "epoch": 0.2608872672606837,
      "grad_norm": 0.8601179119840539,
      "learning_rate": 8.665883552634916e-06,
      "loss": 0.4309,
      "step": 4165
    },
    {
      "epoch": 0.260949905260026,
      "grad_norm": 0.89149382750126,
      "learning_rate": 8.665193647387375e-06,
      "loss": 0.3972,
      "step": 4166
    },
    {
      "epoch": 0.2610125432593683,
      "grad_norm": 0.8435038981463181,
      "learning_rate": 8.664503591279783e-06,
      "loss": 0.3994,
      "step": 4167
    },
    {
      "epoch": 0.2610751812587106,
      "grad_norm": 0.8064683501015125,
      "learning_rate": 8.663813384340543e-06,
      "loss": 0.4072,
      "step": 4168
    },
    {
      "epoch": 0.2611378192580529,
      "grad_norm": 0.8349379830736554,
      "learning_rate": 8.663123026598064e-06,
      "loss": 0.4175,
      "step": 4169
    },
    {
      "epoch": 0.2612004572573952,
      "grad_norm": 0.8837154569890563,
      "learning_rate": 8.662432518080761e-06,
      "loss": 0.408,
      "step": 4170
    },
    {
      "epoch": 0.2612630952567375,
      "grad_norm": 0.7687421454382941,
      "learning_rate": 8.661741858817057e-06,
      "loss": 0.4287,
      "step": 4171
    },
    {
      "epoch": 0.26132573325607983,
      "grad_norm": 0.8769753877108003,
      "learning_rate": 8.661051048835376e-06,
      "loss": 0.4313,
      "step": 4172
    },
    {
      "epoch": 0.2613883712554221,
      "grad_norm": 0.8729883544898286,
      "learning_rate": 8.660360088164158e-06,
      "loss": 0.4174,
      "step": 4173
    },
    {
      "epoch": 0.2614510092547644,
      "grad_norm": 0.8275486268212456,
      "learning_rate": 8.659668976831836e-06,
      "loss": 0.4216,
      "step": 4174
    },
    {
      "epoch": 0.2615136472541067,
      "grad_norm": 0.8128953738953694,
      "learning_rate": 8.658977714866861e-06,
      "loss": 0.4117,
      "step": 4175
    },
    {
      "epoch": 0.261576285253449,
      "grad_norm": 0.791090837724322,
      "learning_rate": 8.658286302297683e-06,
      "loss": 0.4,
      "step": 4176
    },
    {
      "epoch": 0.2616389232527913,
      "grad_norm": 0.839313556386332,
      "learning_rate": 8.657594739152763e-06,
      "loss": 0.4058,
      "step": 4177
    },
    {
      "epoch": 0.2617015612521336,
      "grad_norm": 0.9103315511234095,
      "learning_rate": 8.656903025460564e-06,
      "loss": 0.4394,
      "step": 4178
    },
    {
      "epoch": 0.2617641992514759,
      "grad_norm": 0.8085927290552466,
      "learning_rate": 8.656211161249557e-06,
      "loss": 0.4377,
      "step": 4179
    },
    {
      "epoch": 0.26182683725081823,
      "grad_norm": 0.9481105813770321,
      "learning_rate": 8.655519146548222e-06,
      "loss": 0.4532,
      "step": 4180
    },
    {
      "epoch": 0.2618894752501605,
      "grad_norm": 0.8643272623861211,
      "learning_rate": 8.65482698138504e-06,
      "loss": 0.481,
      "step": 4181
    },
    {
      "epoch": 0.2619521132495028,
      "grad_norm": 0.7685862516835523,
      "learning_rate": 8.6541346657885e-06,
      "loss": 0.3665,
      "step": 4182
    },
    {
      "epoch": 0.2620147512488451,
      "grad_norm": 0.8067714918795141,
      "learning_rate": 8.653442199787099e-06,
      "loss": 0.3908,
      "step": 4183
    },
    {
      "epoch": 0.2620773892481874,
      "grad_norm": 0.8596068286285956,
      "learning_rate": 8.65274958340934e-06,
      "loss": 0.4311,
      "step": 4184
    },
    {
      "epoch": 0.2621400272475297,
      "grad_norm": 0.77719757908897,
      "learning_rate": 8.652056816683729e-06,
      "loss": 0.3765,
      "step": 4185
    },
    {
      "epoch": 0.262202665246872,
      "grad_norm": 0.8870181598781006,
      "learning_rate": 8.651363899638783e-06,
      "loss": 0.4159,
      "step": 4186
    },
    {
      "epoch": 0.26226530324621433,
      "grad_norm": 0.7568236260405687,
      "learning_rate": 8.65067083230302e-06,
      "loss": 0.4215,
      "step": 4187
    },
    {
      "epoch": 0.26232794124555664,
      "grad_norm": 0.7905007806850509,
      "learning_rate": 8.649977614704968e-06,
      "loss": 0.419,
      "step": 4188
    },
    {
      "epoch": 0.2623905792448989,
      "grad_norm": 0.8755281481990128,
      "learning_rate": 8.64928424687316e-06,
      "loss": 0.407,
      "step": 4189
    },
    {
      "epoch": 0.2624532172442412,
      "grad_norm": 0.8233142455322221,
      "learning_rate": 8.648590728836135e-06,
      "loss": 0.3924,
      "step": 4190
    },
    {
      "epoch": 0.2625158552435835,
      "grad_norm": 0.8630513574663252,
      "learning_rate": 8.647897060622439e-06,
      "loss": 0.4308,
      "step": 4191
    },
    {
      "epoch": 0.2625784932429258,
      "grad_norm": 0.806159791636785,
      "learning_rate": 8.647203242260622e-06,
      "loss": 0.4044,
      "step": 4192
    },
    {
      "epoch": 0.2626411312422681,
      "grad_norm": 0.8783320090245615,
      "learning_rate": 8.646509273779243e-06,
      "loss": 0.4828,
      "step": 4193
    },
    {
      "epoch": 0.26270376924161043,
      "grad_norm": 0.8713936924556135,
      "learning_rate": 8.645815155206865e-06,
      "loss": 0.4394,
      "step": 4194
    },
    {
      "epoch": 0.26276640724095274,
      "grad_norm": 0.8036090596685788,
      "learning_rate": 8.64512088657206e-06,
      "loss": 0.4184,
      "step": 4195
    },
    {
      "epoch": 0.26282904524029505,
      "grad_norm": 0.8821880999133781,
      "learning_rate": 8.644426467903401e-06,
      "loss": 0.463,
      "step": 4196
    },
    {
      "epoch": 0.26289168323963735,
      "grad_norm": 0.8074322854801484,
      "learning_rate": 8.643731899229474e-06,
      "loss": 0.4544,
      "step": 4197
    },
    {
      "epoch": 0.2629543212389796,
      "grad_norm": 0.9064994349627313,
      "learning_rate": 8.643037180578864e-06,
      "loss": 0.4376,
      "step": 4198
    },
    {
      "epoch": 0.2630169592383219,
      "grad_norm": 0.844446282072924,
      "learning_rate": 8.642342311980168e-06,
      "loss": 0.4309,
      "step": 4199
    },
    {
      "epoch": 0.2630795972376642,
      "grad_norm": 0.8597869107530864,
      "learning_rate": 8.641647293461987e-06,
      "loss": 0.4518,
      "step": 4200
    },
    {
      "epoch": 0.2631422352370065,
      "grad_norm": 0.8379200434904158,
      "learning_rate": 8.640952125052927e-06,
      "loss": 0.4318,
      "step": 4201
    },
    {
      "epoch": 0.26320487323634884,
      "grad_norm": 0.7774764337895741,
      "learning_rate": 8.640256806781602e-06,
      "loss": 0.4987,
      "step": 4202
    },
    {
      "epoch": 0.26326751123569114,
      "grad_norm": 0.8612398170239322,
      "learning_rate": 8.639561338676632e-06,
      "loss": 0.4232,
      "step": 4203
    },
    {
      "epoch": 0.26333014923503345,
      "grad_norm": 0.796637768813782,
      "learning_rate": 8.638865720766639e-06,
      "loss": 0.4329,
      "step": 4204
    },
    {
      "epoch": 0.26339278723437576,
      "grad_norm": 0.8207795238135752,
      "learning_rate": 8.63816995308026e-06,
      "loss": 0.4626,
      "step": 4205
    },
    {
      "epoch": 0.263455425233718,
      "grad_norm": 0.7374919467034509,
      "learning_rate": 8.63747403564613e-06,
      "loss": 0.3942,
      "step": 4206
    },
    {
      "epoch": 0.2635180632330603,
      "grad_norm": 0.9012380822645152,
      "learning_rate": 8.636777968492892e-06,
      "loss": 0.4268,
      "step": 4207
    },
    {
      "epoch": 0.2635807012324026,
      "grad_norm": 0.871016094064237,
      "learning_rate": 8.636081751649199e-06,
      "loss": 0.4266,
      "step": 4208
    },
    {
      "epoch": 0.26364333923174493,
      "grad_norm": 0.8950501548613405,
      "learning_rate": 8.635385385143707e-06,
      "loss": 0.438,
      "step": 4209
    },
    {
      "epoch": 0.26370597723108724,
      "grad_norm": 0.8230179037999306,
      "learning_rate": 8.634688869005076e-06,
      "loss": 0.4495,
      "step": 4210
    },
    {
      "epoch": 0.26376861523042955,
      "grad_norm": 0.7870846204428023,
      "learning_rate": 8.633992203261978e-06,
      "loss": 0.4221,
      "step": 4211
    },
    {
      "epoch": 0.26383125322977186,
      "grad_norm": 0.8257170697418107,
      "learning_rate": 8.633295387943086e-06,
      "loss": 0.4046,
      "step": 4212
    },
    {
      "epoch": 0.26389389122911416,
      "grad_norm": 0.8367182946297216,
      "learning_rate": 8.632598423077081e-06,
      "loss": 0.4395,
      "step": 4213
    },
    {
      "epoch": 0.2639565292284564,
      "grad_norm": 0.7901231113966316,
      "learning_rate": 8.631901308692652e-06,
      "loss": 0.4109,
      "step": 4214
    },
    {
      "epoch": 0.2640191672277987,
      "grad_norm": 0.8198548155304398,
      "learning_rate": 8.631204044818489e-06,
      "loss": 0.4484,
      "step": 4215
    },
    {
      "epoch": 0.26408180522714103,
      "grad_norm": 0.8655978753786937,
      "learning_rate": 8.630506631483296e-06,
      "loss": 0.4166,
      "step": 4216
    },
    {
      "epoch": 0.26414444322648334,
      "grad_norm": 0.6905911960104792,
      "learning_rate": 8.629809068715773e-06,
      "loss": 0.4904,
      "step": 4217
    },
    {
      "epoch": 0.26420708122582565,
      "grad_norm": 0.8781621862808804,
      "learning_rate": 8.629111356544636e-06,
      "loss": 0.4309,
      "step": 4218
    },
    {
      "epoch": 0.26426971922516795,
      "grad_norm": 0.7917872476703761,
      "learning_rate": 8.628413494998605e-06,
      "loss": 0.3876,
      "step": 4219
    },
    {
      "epoch": 0.26433235722451026,
      "grad_norm": 0.8709370089137499,
      "learning_rate": 8.627715484106399e-06,
      "loss": 0.4253,
      "step": 4220
    },
    {
      "epoch": 0.26439499522385257,
      "grad_norm": 0.855308827970046,
      "learning_rate": 8.62701732389675e-06,
      "loss": 0.4043,
      "step": 4221
    },
    {
      "epoch": 0.2644576332231949,
      "grad_norm": 0.7942695789312221,
      "learning_rate": 8.626319014398394e-06,
      "loss": 0.4104,
      "step": 4222
    },
    {
      "epoch": 0.26452027122253713,
      "grad_norm": 0.8496632787991478,
      "learning_rate": 8.625620555640075e-06,
      "loss": 0.4115,
      "step": 4223
    },
    {
      "epoch": 0.26458290922187944,
      "grad_norm": 0.9215991927478038,
      "learning_rate": 8.624921947650541e-06,
      "loss": 0.4459,
      "step": 4224
    },
    {
      "epoch": 0.26464554722122174,
      "grad_norm": 0.7648230629585848,
      "learning_rate": 8.624223190458549e-06,
      "loss": 0.4072,
      "step": 4225
    },
    {
      "epoch": 0.26470818522056405,
      "grad_norm": 0.7834468080981228,
      "learning_rate": 8.623524284092855e-06,
      "loss": 0.3903,
      "step": 4226
    },
    {
      "epoch": 0.26477082321990636,
      "grad_norm": 0.8243846718639596,
      "learning_rate": 8.62282522858223e-06,
      "loss": 0.4199,
      "step": 4227
    },
    {
      "epoch": 0.26483346121924867,
      "grad_norm": 0.8460844779401967,
      "learning_rate": 8.622126023955446e-06,
      "loss": 0.4289,
      "step": 4228
    },
    {
      "epoch": 0.264896099218591,
      "grad_norm": 0.8484052593311894,
      "learning_rate": 8.621426670241282e-06,
      "loss": 0.4436,
      "step": 4229
    },
    {
      "epoch": 0.2649587372179333,
      "grad_norm": 0.8827965573847948,
      "learning_rate": 8.620727167468525e-06,
      "loss": 0.4335,
      "step": 4230
    },
    {
      "epoch": 0.26502137521727553,
      "grad_norm": 0.9043161607186704,
      "learning_rate": 8.620027515665965e-06,
      "loss": 0.4282,
      "step": 4231
    },
    {
      "epoch": 0.26508401321661784,
      "grad_norm": 0.8805284970426956,
      "learning_rate": 8.6193277148624e-06,
      "loss": 0.4611,
      "step": 4232
    },
    {
      "epoch": 0.26514665121596015,
      "grad_norm": 0.8989082094761628,
      "learning_rate": 8.618627765086636e-06,
      "loss": 0.4396,
      "step": 4233
    },
    {
      "epoch": 0.26520928921530246,
      "grad_norm": 0.8856512746845503,
      "learning_rate": 8.61792766636748e-06,
      "loss": 0.4401,
      "step": 4234
    },
    {
      "epoch": 0.26527192721464476,
      "grad_norm": 0.9045312913496782,
      "learning_rate": 8.617227418733751e-06,
      "loss": 0.3991,
      "step": 4235
    },
    {
      "epoch": 0.2653345652139871,
      "grad_norm": 0.8003614352053787,
      "learning_rate": 8.61652702221427e-06,
      "loss": 0.3708,
      "step": 4236
    },
    {
      "epoch": 0.2653972032133294,
      "grad_norm": 0.8587265904835804,
      "learning_rate": 8.615826476837866e-06,
      "loss": 0.4753,
      "step": 4237
    },
    {
      "epoch": 0.2654598412126717,
      "grad_norm": 0.894464540289796,
      "learning_rate": 8.615125782633371e-06,
      "loss": 0.4497,
      "step": 4238
    },
    {
      "epoch": 0.265522479212014,
      "grad_norm": 0.9098209072813513,
      "learning_rate": 8.61442493962963e-06,
      "loss": 0.4018,
      "step": 4239
    },
    {
      "epoch": 0.26558511721135625,
      "grad_norm": 0.8009483087615848,
      "learning_rate": 8.613723947855485e-06,
      "loss": 0.4229,
      "step": 4240
    },
    {
      "epoch": 0.26564775521069856,
      "grad_norm": 0.8200695031279328,
      "learning_rate": 8.613022807339795e-06,
      "loss": 0.456,
      "step": 4241
    },
    {
      "epoch": 0.26571039321004086,
      "grad_norm": 0.804942710569211,
      "learning_rate": 8.612321518111414e-06,
      "loss": 0.417,
      "step": 4242
    },
    {
      "epoch": 0.26577303120938317,
      "grad_norm": 0.7909933325599363,
      "learning_rate": 8.611620080199208e-06,
      "loss": 0.3819,
      "step": 4243
    },
    {
      "epoch": 0.2658356692087255,
      "grad_norm": 0.863703128707603,
      "learning_rate": 8.61091849363205e-06,
      "loss": 0.4467,
      "step": 4244
    },
    {
      "epoch": 0.2658983072080678,
      "grad_norm": 0.8939744527728983,
      "learning_rate": 8.610216758438817e-06,
      "loss": 0.4495,
      "step": 4245
    },
    {
      "epoch": 0.2659609452074101,
      "grad_norm": 0.7662234997899136,
      "learning_rate": 8.609514874648391e-06,
      "loss": 0.3715,
      "step": 4246
    },
    {
      "epoch": 0.2660235832067524,
      "grad_norm": 0.8898196124227771,
      "learning_rate": 8.608812842289662e-06,
      "loss": 0.4751,
      "step": 4247
    },
    {
      "epoch": 0.26608622120609465,
      "grad_norm": 0.7113900109575446,
      "learning_rate": 8.608110661391529e-06,
      "loss": 0.4961,
      "step": 4248
    },
    {
      "epoch": 0.26614885920543696,
      "grad_norm": 0.8822845453818352,
      "learning_rate": 8.60740833198289e-06,
      "loss": 0.4315,
      "step": 4249
    },
    {
      "epoch": 0.26621149720477927,
      "grad_norm": 0.8302210957043594,
      "learning_rate": 8.606705854092656e-06,
      "loss": 0.441,
      "step": 4250
    },
    {
      "epoch": 0.2662741352041216,
      "grad_norm": 0.83671895551264,
      "learning_rate": 8.606003227749738e-06,
      "loss": 0.4283,
      "step": 4251
    },
    {
      "epoch": 0.2663367732034639,
      "grad_norm": 0.8691187218310675,
      "learning_rate": 8.605300452983058e-06,
      "loss": 0.4379,
      "step": 4252
    },
    {
      "epoch": 0.2663994112028062,
      "grad_norm": 0.8745916959042912,
      "learning_rate": 8.604597529821544e-06,
      "loss": 0.4486,
      "step": 4253
    },
    {
      "epoch": 0.2664620492021485,
      "grad_norm": 0.8775110285079845,
      "learning_rate": 8.603894458294124e-06,
      "loss": 0.4804,
      "step": 4254
    },
    {
      "epoch": 0.2665246872014908,
      "grad_norm": 0.9363694253329028,
      "learning_rate": 8.60319123842974e-06,
      "loss": 0.4832,
      "step": 4255
    },
    {
      "epoch": 0.26658732520083306,
      "grad_norm": 0.8396760969808199,
      "learning_rate": 8.602487870257336e-06,
      "loss": 0.46,
      "step": 4256
    },
    {
      "epoch": 0.26664996320017537,
      "grad_norm": 0.8038662924681632,
      "learning_rate": 8.601784353805864e-06,
      "loss": 0.443,
      "step": 4257
    },
    {
      "epoch": 0.2667126011995177,
      "grad_norm": 0.7576364664962547,
      "learning_rate": 8.601080689104278e-06,
      "loss": 0.345,
      "step": 4258
    },
    {
      "epoch": 0.26677523919886,
      "grad_norm": 0.8736343975370163,
      "learning_rate": 8.600376876181542e-06,
      "loss": 0.4426,
      "step": 4259
    },
    {
      "epoch": 0.2668378771982023,
      "grad_norm": 0.83749708367297,
      "learning_rate": 8.599672915066625e-06,
      "loss": 0.4321,
      "step": 4260
    },
    {
      "epoch": 0.2669005151975446,
      "grad_norm": 0.9132267711212684,
      "learning_rate": 8.598968805788505e-06,
      "loss": 0.4442,
      "step": 4261
    },
    {
      "epoch": 0.2669631531968869,
      "grad_norm": 0.8380932871211902,
      "learning_rate": 8.59826454837616e-06,
      "loss": 0.4417,
      "step": 4262
    },
    {
      "epoch": 0.2670257911962292,
      "grad_norm": 0.8680800114554649,
      "learning_rate": 8.597560142858579e-06,
      "loss": 0.4364,
      "step": 4263
    },
    {
      "epoch": 0.2670884291955715,
      "grad_norm": 0.8126035059612606,
      "learning_rate": 8.596855589264752e-06,
      "loss": 0.4189,
      "step": 4264
    },
    {
      "epoch": 0.26715106719491377,
      "grad_norm": 0.9033153448015726,
      "learning_rate": 8.596150887623685e-06,
      "loss": 0.4449,
      "step": 4265
    },
    {
      "epoch": 0.2672137051942561,
      "grad_norm": 0.832674813516797,
      "learning_rate": 8.595446037964377e-06,
      "loss": 0.4269,
      "step": 4266
    },
    {
      "epoch": 0.2672763431935984,
      "grad_norm": 0.7965668088220613,
      "learning_rate": 8.594741040315844e-06,
      "loss": 0.4445,
      "step": 4267
    },
    {
      "epoch": 0.2673389811929407,
      "grad_norm": 0.8415780124540057,
      "learning_rate": 8.594035894707104e-06,
      "loss": 0.4387,
      "step": 4268
    },
    {
      "epoch": 0.267401619192283,
      "grad_norm": 0.8172702337232598,
      "learning_rate": 8.593330601167179e-06,
      "loss": 0.4232,
      "step": 4269
    },
    {
      "epoch": 0.2674642571916253,
      "grad_norm": 0.8529384404297077,
      "learning_rate": 8.592625159725097e-06,
      "loss": 0.4621,
      "step": 4270
    },
    {
      "epoch": 0.2675268951909676,
      "grad_norm": 0.9313700170662795,
      "learning_rate": 8.591919570409898e-06,
      "loss": 0.4305,
      "step": 4271
    },
    {
      "epoch": 0.2675895331903099,
      "grad_norm": 0.8371350272018241,
      "learning_rate": 8.591213833250624e-06,
      "loss": 0.3848,
      "step": 4272
    },
    {
      "epoch": 0.2676521711896522,
      "grad_norm": 0.8151455311265557,
      "learning_rate": 8.590507948276321e-06,
      "loss": 0.4804,
      "step": 4273
    },
    {
      "epoch": 0.2677148091889945,
      "grad_norm": 0.8828129850619955,
      "learning_rate": 8.589801915516044e-06,
      "loss": 0.4077,
      "step": 4274
    },
    {
      "epoch": 0.2677774471883368,
      "grad_norm": 0.817616360555858,
      "learning_rate": 8.589095734998854e-06,
      "loss": 0.4694,
      "step": 4275
    },
    {
      "epoch": 0.2678400851876791,
      "grad_norm": 0.8435096321049345,
      "learning_rate": 8.58838940675382e-06,
      "loss": 0.4186,
      "step": 4276
    },
    {
      "epoch": 0.2679027231870214,
      "grad_norm": 0.8387899240095171,
      "learning_rate": 8.587682930810008e-06,
      "loss": 0.4403,
      "step": 4277
    },
    {
      "epoch": 0.2679653611863637,
      "grad_norm": 0.782298409475879,
      "learning_rate": 8.586976307196502e-06,
      "loss": 0.4058,
      "step": 4278
    },
    {
      "epoch": 0.268027999185706,
      "grad_norm": 0.7934555779287984,
      "learning_rate": 8.586269535942386e-06,
      "loss": 0.3946,
      "step": 4279
    },
    {
      "epoch": 0.26809063718504833,
      "grad_norm": 0.8843046923786375,
      "learning_rate": 8.58556261707675e-06,
      "loss": 0.3939,
      "step": 4280
    },
    {
      "epoch": 0.26815327518439064,
      "grad_norm": 0.8594399124159386,
      "learning_rate": 8.58485555062869e-06,
      "loss": 0.4462,
      "step": 4281
    },
    {
      "epoch": 0.2682159131837329,
      "grad_norm": 0.7481909940123735,
      "learning_rate": 8.584148336627313e-06,
      "loss": 0.3935,
      "step": 4282
    },
    {
      "epoch": 0.2682785511830752,
      "grad_norm": 0.7818775035875271,
      "learning_rate": 8.583440975101723e-06,
      "loss": 0.4318,
      "step": 4283
    },
    {
      "epoch": 0.2683411891824175,
      "grad_norm": 0.8331356523878437,
      "learning_rate": 8.582733466081038e-06,
      "loss": 0.408,
      "step": 4284
    },
    {
      "epoch": 0.2684038271817598,
      "grad_norm": 0.6840882925234901,
      "learning_rate": 8.58202580959438e-06,
      "loss": 0.4671,
      "step": 4285
    },
    {
      "epoch": 0.2684664651811021,
      "grad_norm": 0.8146926945505253,
      "learning_rate": 8.581318005670872e-06,
      "loss": 0.4431,
      "step": 4286
    },
    {
      "epoch": 0.26852910318044443,
      "grad_norm": 0.7863312949891271,
      "learning_rate": 8.580610054339652e-06,
      "loss": 0.4048,
      "step": 4287
    },
    {
      "epoch": 0.26859174117978674,
      "grad_norm": 0.7661518735744125,
      "learning_rate": 8.579901955629856e-06,
      "loss": 0.4769,
      "step": 4288
    },
    {
      "epoch": 0.26865437917912904,
      "grad_norm": 0.8225534850209659,
      "learning_rate": 8.579193709570632e-06,
      "loss": 0.4427,
      "step": 4289
    },
    {
      "epoch": 0.2687170171784713,
      "grad_norm": 0.8510424079985967,
      "learning_rate": 8.57848531619113e-06,
      "loss": 0.4117,
      "step": 4290
    },
    {
      "epoch": 0.2687796551778136,
      "grad_norm": 0.840495128593006,
      "learning_rate": 8.57777677552051e-06,
      "loss": 0.4434,
      "step": 4291
    },
    {
      "epoch": 0.2688422931771559,
      "grad_norm": 0.7760964976634867,
      "learning_rate": 8.577068087587932e-06,
      "loss": 0.4063,
      "step": 4292
    },
    {
      "epoch": 0.2689049311764982,
      "grad_norm": 0.8255561554435197,
      "learning_rate": 8.576359252422568e-06,
      "loss": 0.4187,
      "step": 4293
    },
    {
      "epoch": 0.2689675691758405,
      "grad_norm": 0.7838289068980341,
      "learning_rate": 8.575650270053594e-06,
      "loss": 0.4164,
      "step": 4294
    },
    {
      "epoch": 0.26903020717518283,
      "grad_norm": 0.8206667475366541,
      "learning_rate": 8.574941140510191e-06,
      "loss": 0.4065,
      "step": 4295
    },
    {
      "epoch": 0.26909284517452514,
      "grad_norm": 0.9284700589917293,
      "learning_rate": 8.574231863821547e-06,
      "loss": 0.4552,
      "step": 4296
    },
    {
      "epoch": 0.26915548317386745,
      "grad_norm": 0.8128868940368372,
      "learning_rate": 8.573522440016855e-06,
      "loss": 0.42,
      "step": 4297
    },
    {
      "epoch": 0.2692181211732097,
      "grad_norm": 0.8764933889525925,
      "learning_rate": 8.572812869125318e-06,
      "loss": 0.4264,
      "step": 4298
    },
    {
      "epoch": 0.269280759172552,
      "grad_norm": 0.8034501200882022,
      "learning_rate": 8.572103151176139e-06,
      "loss": 0.4045,
      "step": 4299
    },
    {
      "epoch": 0.2693433971718943,
      "grad_norm": 0.8572400171978521,
      "learning_rate": 8.571393286198531e-06,
      "loss": 0.4002,
      "step": 4300
    },
    {
      "epoch": 0.2694060351712366,
      "grad_norm": 0.8322150493540873,
      "learning_rate": 8.570683274221713e-06,
      "loss": 0.4319,
      "step": 4301
    },
    {
      "epoch": 0.26946867317057893,
      "grad_norm": 0.813650769811573,
      "learning_rate": 8.56997311527491e-06,
      "loss": 0.4309,
      "step": 4302
    },
    {
      "epoch": 0.26953131116992124,
      "grad_norm": 0.9154974454057713,
      "learning_rate": 8.56926280938735e-06,
      "loss": 0.4635,
      "step": 4303
    },
    {
      "epoch": 0.26959394916926355,
      "grad_norm": 0.8585675072638985,
      "learning_rate": 8.568552356588269e-06,
      "loss": 0.4108,
      "step": 4304
    },
    {
      "epoch": 0.26965658716860585,
      "grad_norm": 0.8506526083784324,
      "learning_rate": 8.567841756906912e-06,
      "loss": 0.401,
      "step": 4305
    },
    {
      "epoch": 0.26971922516794816,
      "grad_norm": 0.821973636106998,
      "learning_rate": 8.567131010372524e-06,
      "loss": 0.3991,
      "step": 4306
    },
    {
      "epoch": 0.2697818631672904,
      "grad_norm": 0.777753626596208,
      "learning_rate": 8.566420117014364e-06,
      "loss": 0.4811,
      "step": 4307
    },
    {
      "epoch": 0.2698445011666327,
      "grad_norm": 0.8377560596676171,
      "learning_rate": 8.56570907686169e-06,
      "loss": 0.4387,
      "step": 4308
    },
    {
      "epoch": 0.26990713916597503,
      "grad_norm": 0.9084314075619007,
      "learning_rate": 8.564997889943767e-06,
      "loss": 0.424,
      "step": 4309
    },
    {
      "epoch": 0.26996977716531734,
      "grad_norm": 0.8682674473268072,
      "learning_rate": 8.564286556289871e-06,
      "loss": 0.4583,
      "step": 4310
    },
    {
      "epoch": 0.27003241516465964,
      "grad_norm": 0.9389185251170087,
      "learning_rate": 8.563575075929277e-06,
      "loss": 0.4679,
      "step": 4311
    },
    {
      "epoch": 0.27009505316400195,
      "grad_norm": 0.9077161294919951,
      "learning_rate": 8.562863448891273e-06,
      "loss": 0.4757,
      "step": 4312
    },
    {
      "epoch": 0.27015769116334426,
      "grad_norm": 0.8157951583746036,
      "learning_rate": 8.562151675205146e-06,
      "loss": 0.391,
      "step": 4313
    },
    {
      "epoch": 0.27022032916268657,
      "grad_norm": 0.7605615425093023,
      "learning_rate": 8.561439754900197e-06,
      "loss": 0.4895,
      "step": 4314
    },
    {
      "epoch": 0.2702829671620288,
      "grad_norm": 0.7306773831770704,
      "learning_rate": 8.560727688005725e-06,
      "loss": 0.3574,
      "step": 4315
    },
    {
      "epoch": 0.2703456051613711,
      "grad_norm": 1.0359759112559586,
      "learning_rate": 8.560015474551043e-06,
      "loss": 0.4458,
      "step": 4316
    },
    {
      "epoch": 0.27040824316071344,
      "grad_norm": 0.9350998051228273,
      "learning_rate": 8.559303114565461e-06,
      "loss": 0.4096,
      "step": 4317
    },
    {
      "epoch": 0.27047088116005574,
      "grad_norm": 0.8381379279154557,
      "learning_rate": 8.558590608078303e-06,
      "loss": 0.4263,
      "step": 4318
    },
    {
      "epoch": 0.27053351915939805,
      "grad_norm": 0.8667188977784114,
      "learning_rate": 8.557877955118893e-06,
      "loss": 0.4056,
      "step": 4319
    },
    {
      "epoch": 0.27059615715874036,
      "grad_norm": 0.8054907254637244,
      "learning_rate": 8.55716515571657e-06,
      "loss": 0.4194,
      "step": 4320
    },
    {
      "epoch": 0.27065879515808267,
      "grad_norm": 0.8354465839055676,
      "learning_rate": 8.556452209900665e-06,
      "loss": 0.4545,
      "step": 4321
    },
    {
      "epoch": 0.270721433157425,
      "grad_norm": 0.847299501263117,
      "learning_rate": 8.55573911770053e-06,
      "loss": 0.4044,
      "step": 4322
    },
    {
      "epoch": 0.2707840711567672,
      "grad_norm": 0.8634912908905146,
      "learning_rate": 8.555025879145512e-06,
      "loss": 0.4315,
      "step": 4323
    },
    {
      "epoch": 0.27084670915610953,
      "grad_norm": 0.8481564083296793,
      "learning_rate": 8.554312494264967e-06,
      "loss": 0.4421,
      "step": 4324
    },
    {
      "epoch": 0.27090934715545184,
      "grad_norm": 0.8676621392203895,
      "learning_rate": 8.553598963088264e-06,
      "loss": 0.4408,
      "step": 4325
    },
    {
      "epoch": 0.27097198515479415,
      "grad_norm": 0.6702704056733135,
      "learning_rate": 8.552885285644765e-06,
      "loss": 0.476,
      "step": 4326
    },
    {
      "epoch": 0.27103462315413646,
      "grad_norm": 0.8449099361385716,
      "learning_rate": 8.552171461963848e-06,
      "loss": 0.4208,
      "step": 4327
    },
    {
      "epoch": 0.27109726115347876,
      "grad_norm": 0.8405828921672676,
      "learning_rate": 8.551457492074895e-06,
      "loss": 0.4379,
      "step": 4328
    },
    {
      "epoch": 0.27115989915282107,
      "grad_norm": 0.8835216768387961,
      "learning_rate": 8.550743376007294e-06,
      "loss": 0.4273,
      "step": 4329
    },
    {
      "epoch": 0.2712225371521634,
      "grad_norm": 0.8425296747669313,
      "learning_rate": 8.550029113790434e-06,
      "loss": 0.4455,
      "step": 4330
    },
    {
      "epoch": 0.2712851751515057,
      "grad_norm": 0.7903709113873222,
      "learning_rate": 8.549314705453719e-06,
      "loss": 0.4259,
      "step": 4331
    },
    {
      "epoch": 0.27134781315084794,
      "grad_norm": 0.8408432014966198,
      "learning_rate": 8.54860015102655e-06,
      "loss": 0.4344,
      "step": 4332
    },
    {
      "epoch": 0.27141045115019025,
      "grad_norm": 0.8601942577594287,
      "learning_rate": 8.547885450538341e-06,
      "loss": 0.4253,
      "step": 4333
    },
    {
      "epoch": 0.27147308914953255,
      "grad_norm": 0.6542106540432312,
      "learning_rate": 8.547170604018508e-06,
      "loss": 0.5088,
      "step": 4334
    },
    {
      "epoch": 0.27153572714887486,
      "grad_norm": 0.7996202370130091,
      "learning_rate": 8.546455611496475e-06,
      "loss": 0.4372,
      "step": 4335
    },
    {
      "epoch": 0.27159836514821717,
      "grad_norm": 0.8113937931660986,
      "learning_rate": 8.545740473001671e-06,
      "loss": 0.4221,
      "step": 4336
    },
    {
      "epoch": 0.2716610031475595,
      "grad_norm": 0.8228119278925813,
      "learning_rate": 8.54502518856353e-06,
      "loss": 0.4541,
      "step": 4337
    },
    {
      "epoch": 0.2717236411469018,
      "grad_norm": 0.8749710613965043,
      "learning_rate": 8.544309758211497e-06,
      "loss": 0.4176,
      "step": 4338
    },
    {
      "epoch": 0.2717862791462441,
      "grad_norm": 0.8425931249036832,
      "learning_rate": 8.543594181975014e-06,
      "loss": 0.4193,
      "step": 4339
    },
    {
      "epoch": 0.27184891714558634,
      "grad_norm": 0.8604076669791659,
      "learning_rate": 8.542878459883537e-06,
      "loss": 0.4277,
      "step": 4340
    },
    {
      "epoch": 0.27191155514492865,
      "grad_norm": 0.9924355153120403,
      "learning_rate": 8.542162591966524e-06,
      "loss": 0.4363,
      "step": 4341
    },
    {
      "epoch": 0.27197419314427096,
      "grad_norm": 0.801698149586279,
      "learning_rate": 8.541446578253445e-06,
      "loss": 0.4489,
      "step": 4342
    },
    {
      "epoch": 0.27203683114361327,
      "grad_norm": 0.8470267582297076,
      "learning_rate": 8.540730418773763e-06,
      "loss": 0.4162,
      "step": 4343
    },
    {
      "epoch": 0.2720994691429556,
      "grad_norm": 0.8675551316763114,
      "learning_rate": 8.540014113556963e-06,
      "loss": 0.4165,
      "step": 4344
    },
    {
      "epoch": 0.2721621071422979,
      "grad_norm": 0.8523372026275139,
      "learning_rate": 8.539297662632524e-06,
      "loss": 0.4177,
      "step": 4345
    },
    {
      "epoch": 0.2722247451416402,
      "grad_norm": 0.8986272731705139,
      "learning_rate": 8.538581066029936e-06,
      "loss": 0.476,
      "step": 4346
    },
    {
      "epoch": 0.2722873831409825,
      "grad_norm": 0.845832859040655,
      "learning_rate": 8.537864323778695e-06,
      "loss": 0.4103,
      "step": 4347
    },
    {
      "epoch": 0.2723500211403248,
      "grad_norm": 0.8368511526608505,
      "learning_rate": 8.537147435908302e-06,
      "loss": 0.412,
      "step": 4348
    },
    {
      "epoch": 0.27241265913966706,
      "grad_norm": 0.8677445490529927,
      "learning_rate": 8.536430402448263e-06,
      "loss": 0.4421,
      "step": 4349
    },
    {
      "epoch": 0.27247529713900936,
      "grad_norm": 0.8712912399145514,
      "learning_rate": 8.535713223428092e-06,
      "loss": 0.4143,
      "step": 4350
    },
    {
      "epoch": 0.2725379351383517,
      "grad_norm": 0.8166410335301322,
      "learning_rate": 8.53499589887731e-06,
      "loss": 0.3838,
      "step": 4351
    },
    {
      "epoch": 0.272600573137694,
      "grad_norm": 0.8270923877800821,
      "learning_rate": 8.53427842882544e-06,
      "loss": 0.4377,
      "step": 4352
    },
    {
      "epoch": 0.2726632111370363,
      "grad_norm": 0.8756766567626832,
      "learning_rate": 8.533560813302013e-06,
      "loss": 0.4156,
      "step": 4353
    },
    {
      "epoch": 0.2727258491363786,
      "grad_norm": 0.7702458007390299,
      "learning_rate": 8.532843052336568e-06,
      "loss": 0.4042,
      "step": 4354
    },
    {
      "epoch": 0.2727884871357209,
      "grad_norm": 0.8588557747002017,
      "learning_rate": 8.532125145958647e-06,
      "loss": 0.4581,
      "step": 4355
    },
    {
      "epoch": 0.2728511251350632,
      "grad_norm": 0.8411401291868301,
      "learning_rate": 8.531407094197799e-06,
      "loss": 0.4287,
      "step": 4356
    },
    {
      "epoch": 0.27291376313440546,
      "grad_norm": 0.8887189886078217,
      "learning_rate": 8.53068889708358e-06,
      "loss": 0.4666,
      "step": 4357
    },
    {
      "epoch": 0.27297640113374777,
      "grad_norm": 0.8206871667059457,
      "learning_rate": 8.52997055464555e-06,
      "loss": 0.4376,
      "step": 4358
    },
    {
      "epoch": 0.2730390391330901,
      "grad_norm": 0.8604326012472059,
      "learning_rate": 8.529252066913279e-06,
      "loss": 0.4421,
      "step": 4359
    },
    {
      "epoch": 0.2731016771324324,
      "grad_norm": 0.9176247479407834,
      "learning_rate": 8.528533433916336e-06,
      "loss": 0.4308,
      "step": 4360
    },
    {
      "epoch": 0.2731643151317747,
      "grad_norm": 0.9098869624190429,
      "learning_rate": 8.527814655684303e-06,
      "loss": 0.4484,
      "step": 4361
    },
    {
      "epoch": 0.273226953131117,
      "grad_norm": 0.9032944138012245,
      "learning_rate": 8.527095732246762e-06,
      "loss": 0.3962,
      "step": 4362
    },
    {
      "epoch": 0.2732895911304593,
      "grad_norm": 0.9182842058897489,
      "learning_rate": 8.526376663633307e-06,
      "loss": 0.4063,
      "step": 4363
    },
    {
      "epoch": 0.2733522291298016,
      "grad_norm": 0.8554223863822619,
      "learning_rate": 8.525657449873535e-06,
      "loss": 0.4356,
      "step": 4364
    },
    {
      "epoch": 0.27341486712914387,
      "grad_norm": 0.8639391123872765,
      "learning_rate": 8.524938090997047e-06,
      "loss": 0.426,
      "step": 4365
    },
    {
      "epoch": 0.2734775051284862,
      "grad_norm": 0.814369456907431,
      "learning_rate": 8.524218587033453e-06,
      "loss": 0.4272,
      "step": 4366
    },
    {
      "epoch": 0.2735401431278285,
      "grad_norm": 0.8372925464149109,
      "learning_rate": 8.523498938012367e-06,
      "loss": 0.4798,
      "step": 4367
    },
    {
      "epoch": 0.2736027811271708,
      "grad_norm": 0.846547176526304,
      "learning_rate": 8.522779143963412e-06,
      "loss": 0.4699,
      "step": 4368
    },
    {
      "epoch": 0.2736654191265131,
      "grad_norm": 0.7965750575918236,
      "learning_rate": 8.522059204916213e-06,
      "loss": 0.3849,
      "step": 4369
    },
    {
      "epoch": 0.2737280571258554,
      "grad_norm": 0.8503504189279292,
      "learning_rate": 8.521339120900405e-06,
      "loss": 0.4246,
      "step": 4370
    },
    {
      "epoch": 0.2737906951251977,
      "grad_norm": 0.7685159005936921,
      "learning_rate": 8.520618891945623e-06,
      "loss": 0.3789,
      "step": 4371
    },
    {
      "epoch": 0.27385333312454,
      "grad_norm": 0.8665474206021262,
      "learning_rate": 8.519898518081515e-06,
      "loss": 0.4717,
      "step": 4372
    },
    {
      "epoch": 0.27391597112388233,
      "grad_norm": 0.8492742315468849,
      "learning_rate": 8.51917799933773e-06,
      "loss": 0.4167,
      "step": 4373
    },
    {
      "epoch": 0.2739786091232246,
      "grad_norm": 0.8522639101759041,
      "learning_rate": 8.518457335743927e-06,
      "loss": 0.3968,
      "step": 4374
    },
    {
      "epoch": 0.2740412471225669,
      "grad_norm": 0.8428183661574936,
      "learning_rate": 8.517736527329765e-06,
      "loss": 0.4221,
      "step": 4375
    },
    {
      "epoch": 0.2741038851219092,
      "grad_norm": 0.8575996627981611,
      "learning_rate": 8.517015574124914e-06,
      "loss": 0.4415,
      "step": 4376
    },
    {
      "epoch": 0.2741665231212515,
      "grad_norm": 0.7683905985288342,
      "learning_rate": 8.516294476159052e-06,
      "loss": 0.4326,
      "step": 4377
    },
    {
      "epoch": 0.2742291611205938,
      "grad_norm": 0.8645290344362941,
      "learning_rate": 8.515573233461855e-06,
      "loss": 0.4501,
      "step": 4378
    },
    {
      "epoch": 0.2742917991199361,
      "grad_norm": 0.8297651751621385,
      "learning_rate": 8.514851846063011e-06,
      "loss": 0.4125,
      "step": 4379
    },
    {
      "epoch": 0.2743544371192784,
      "grad_norm": 0.8044575819377482,
      "learning_rate": 8.514130313992214e-06,
      "loss": 0.4359,
      "step": 4380
    },
    {
      "epoch": 0.27441707511862073,
      "grad_norm": 0.8776688291327135,
      "learning_rate": 8.513408637279159e-06,
      "loss": 0.4309,
      "step": 4381
    },
    {
      "epoch": 0.274479713117963,
      "grad_norm": 0.7939127340336007,
      "learning_rate": 8.512686815953554e-06,
      "loss": 0.4255,
      "step": 4382
    },
    {
      "epoch": 0.2745423511173053,
      "grad_norm": 0.939656281118891,
      "learning_rate": 8.511964850045107e-06,
      "loss": 0.4201,
      "step": 4383
    },
    {
      "epoch": 0.2746049891166476,
      "grad_norm": 0.8023818948856254,
      "learning_rate": 8.511242739583535e-06,
      "loss": 0.4085,
      "step": 4384
    },
    {
      "epoch": 0.2746676271159899,
      "grad_norm": 0.8990318968605082,
      "learning_rate": 8.510520484598561e-06,
      "loss": 0.4486,
      "step": 4385
    },
    {
      "epoch": 0.2747302651153322,
      "grad_norm": 0.8533641732611068,
      "learning_rate": 8.509798085119912e-06,
      "loss": 0.4532,
      "step": 4386
    },
    {
      "epoch": 0.2747929031146745,
      "grad_norm": 0.8775336643389041,
      "learning_rate": 8.50907554117732e-06,
      "loss": 0.4341,
      "step": 4387
    },
    {
      "epoch": 0.27485554111401683,
      "grad_norm": 0.8702741772126047,
      "learning_rate": 8.50835285280053e-06,
      "loss": 0.3998,
      "step": 4388
    },
    {
      "epoch": 0.27491817911335914,
      "grad_norm": 0.8360028879842281,
      "learning_rate": 8.507630020019285e-06,
      "loss": 0.416,
      "step": 4389
    },
    {
      "epoch": 0.27498081711270145,
      "grad_norm": 0.815306707388127,
      "learning_rate": 8.506907042863337e-06,
      "loss": 0.4368,
      "step": 4390
    },
    {
      "epoch": 0.2750434551120437,
      "grad_norm": 0.8293627627805774,
      "learning_rate": 8.506183921362443e-06,
      "loss": 0.4391,
      "step": 4391
    },
    {
      "epoch": 0.275106093111386,
      "grad_norm": 0.8234033116082576,
      "learning_rate": 8.50546065554637e-06,
      "loss": 0.4212,
      "step": 4392
    },
    {
      "epoch": 0.2751687311107283,
      "grad_norm": 0.840378318076946,
      "learning_rate": 8.504737245444886e-06,
      "loss": 0.4355,
      "step": 4393
    },
    {
      "epoch": 0.2752313691100706,
      "grad_norm": 0.8737379073319701,
      "learning_rate": 8.504013691087765e-06,
      "loss": 0.441,
      "step": 4394
    },
    {
      "epoch": 0.27529400710941293,
      "grad_norm": 0.8293156595283807,
      "learning_rate": 8.503289992504791e-06,
      "loss": 0.4029,
      "step": 4395
    },
    {
      "epoch": 0.27535664510875524,
      "grad_norm": 0.977572279285155,
      "learning_rate": 8.50256614972575e-06,
      "loss": 0.4598,
      "step": 4396
    },
    {
      "epoch": 0.27541928310809755,
      "grad_norm": 0.8373837333840282,
      "learning_rate": 8.501842162780436e-06,
      "loss": 0.4053,
      "step": 4397
    },
    {
      "epoch": 0.27548192110743985,
      "grad_norm": 0.8679219281525978,
      "learning_rate": 8.50111803169865e-06,
      "loss": 0.457,
      "step": 4398
    },
    {
      "epoch": 0.2755445591067821,
      "grad_norm": 0.8205877433387483,
      "learning_rate": 8.500393756510194e-06,
      "loss": 0.4356,
      "step": 4399
    },
    {
      "epoch": 0.2756071971061244,
      "grad_norm": 0.8245302570332514,
      "learning_rate": 8.499669337244882e-06,
      "loss": 0.4537,
      "step": 4400
    },
    {
      "epoch": 0.2756698351054667,
      "grad_norm": 0.8188661102198597,
      "learning_rate": 8.49894477393253e-06,
      "loss": 0.429,
      "step": 4401
    },
    {
      "epoch": 0.27573247310480903,
      "grad_norm": 0.8684908132316613,
      "learning_rate": 8.498220066602964e-06,
      "loss": 0.4405,
      "step": 4402
    },
    {
      "epoch": 0.27579511110415134,
      "grad_norm": 0.8070790575384066,
      "learning_rate": 8.49749521528601e-06,
      "loss": 0.4092,
      "step": 4403
    },
    {
      "epoch": 0.27585774910349364,
      "grad_norm": 0.8895869294935559,
      "learning_rate": 8.496770220011502e-06,
      "loss": 0.4182,
      "step": 4404
    },
    {
      "epoch": 0.27592038710283595,
      "grad_norm": 0.7461359356799369,
      "learning_rate": 8.496045080809284e-06,
      "loss": 0.4896,
      "step": 4405
    },
    {
      "epoch": 0.27598302510217826,
      "grad_norm": 0.831048856449058,
      "learning_rate": 8.4953197977092e-06,
      "loss": 0.4504,
      "step": 4406
    },
    {
      "epoch": 0.2760456631015205,
      "grad_norm": 0.7022598468574721,
      "learning_rate": 8.494594370741107e-06,
      "loss": 0.4747,
      "step": 4407
    },
    {
      "epoch": 0.2761083011008628,
      "grad_norm": 0.8795832228884478,
      "learning_rate": 8.493868799934859e-06,
      "loss": 0.4408,
      "step": 4408
    },
    {
      "epoch": 0.2761709391002051,
      "grad_norm": 0.8548274695444132,
      "learning_rate": 8.493143085320324e-06,
      "loss": 0.4545,
      "step": 4409
    },
    {
      "epoch": 0.27623357709954743,
      "grad_norm": 0.8786641747589563,
      "learning_rate": 8.49241722692737e-06,
      "loss": 0.4535,
      "step": 4410
    },
    {
      "epoch": 0.27629621509888974,
      "grad_norm": 0.7831456918959225,
      "learning_rate": 8.491691224785874e-06,
      "loss": 0.3915,
      "step": 4411
    },
    {
      "epoch": 0.27635885309823205,
      "grad_norm": 0.9432531018223964,
      "learning_rate": 8.490965078925722e-06,
      "loss": 0.451,
      "step": 4412
    },
    {
      "epoch": 0.27642149109757436,
      "grad_norm": 0.920706662145824,
      "learning_rate": 8.490238789376796e-06,
      "loss": 0.4811,
      "step": 4413
    },
    {
      "epoch": 0.27648412909691666,
      "grad_norm": 0.8885043802990165,
      "learning_rate": 8.489512356168996e-06,
      "loss": 0.4594,
      "step": 4414
    },
    {
      "epoch": 0.27654676709625897,
      "grad_norm": 0.7731624041193329,
      "learning_rate": 8.488785779332221e-06,
      "loss": 0.4038,
      "step": 4415
    },
    {
      "epoch": 0.2766094050956012,
      "grad_norm": 0.8097818054552902,
      "learning_rate": 8.488059058896373e-06,
      "loss": 0.4642,
      "step": 4416
    },
    {
      "epoch": 0.27667204309494353,
      "grad_norm": 0.838561427476391,
      "learning_rate": 8.487332194891366e-06,
      "loss": 0.4519,
      "step": 4417
    },
    {
      "epoch": 0.27673468109428584,
      "grad_norm": 0.812200447370102,
      "learning_rate": 8.486605187347121e-06,
      "loss": 0.395,
      "step": 4418
    },
    {
      "epoch": 0.27679731909362815,
      "grad_norm": 0.8798548414582801,
      "learning_rate": 8.48587803629356e-06,
      "loss": 0.417,
      "step": 4419
    },
    {
      "epoch": 0.27685995709297045,
      "grad_norm": 0.8759399408055073,
      "learning_rate": 8.48515074176061e-06,
      "loss": 0.3728,
      "step": 4420
    },
    {
      "epoch": 0.27692259509231276,
      "grad_norm": 0.8766383812321952,
      "learning_rate": 8.48442330377821e-06,
      "loss": 0.4177,
      "step": 4421
    },
    {
      "epoch": 0.27698523309165507,
      "grad_norm": 0.8286653667948707,
      "learning_rate": 8.483695722376299e-06,
      "loss": 0.4066,
      "step": 4422
    },
    {
      "epoch": 0.2770478710909974,
      "grad_norm": 0.8775012057870281,
      "learning_rate": 8.482967997584825e-06,
      "loss": 0.4178,
      "step": 4423
    },
    {
      "epoch": 0.27711050909033963,
      "grad_norm": 0.9551931461691585,
      "learning_rate": 8.482240129433744e-06,
      "loss": 0.4298,
      "step": 4424
    },
    {
      "epoch": 0.27717314708968194,
      "grad_norm": 0.8899308677773982,
      "learning_rate": 8.48151211795301e-06,
      "loss": 0.4145,
      "step": 4425
    },
    {
      "epoch": 0.27723578508902424,
      "grad_norm": 0.8532260368167534,
      "learning_rate": 8.480783963172593e-06,
      "loss": 0.423,
      "step": 4426
    },
    {
      "epoch": 0.27729842308836655,
      "grad_norm": 0.7866993886935454,
      "learning_rate": 8.480055665122461e-06,
      "loss": 0.4527,
      "step": 4427
    },
    {
      "epoch": 0.27736106108770886,
      "grad_norm": 0.8353013083996946,
      "learning_rate": 8.479327223832592e-06,
      "loss": 0.4508,
      "step": 4428
    },
    {
      "epoch": 0.27742369908705117,
      "grad_norm": 0.7985567995624065,
      "learning_rate": 8.478598639332969e-06,
      "loss": 0.3955,
      "step": 4429
    },
    {
      "epoch": 0.2774863370863935,
      "grad_norm": 0.8137606287561502,
      "learning_rate": 8.477869911653581e-06,
      "loss": 0.4036,
      "step": 4430
    },
    {
      "epoch": 0.2775489750857358,
      "grad_norm": 0.810642709844054,
      "learning_rate": 8.477141040824421e-06,
      "loss": 0.4415,
      "step": 4431
    },
    {
      "epoch": 0.27761161308507804,
      "grad_norm": 0.8845453142279435,
      "learning_rate": 8.476412026875492e-06,
      "loss": 0.4392,
      "step": 4432
    },
    {
      "epoch": 0.27767425108442034,
      "grad_norm": 0.7887707939256526,
      "learning_rate": 8.475682869836797e-06,
      "loss": 0.4017,
      "step": 4433
    },
    {
      "epoch": 0.27773688908376265,
      "grad_norm": 0.7840639384427776,
      "learning_rate": 8.47495356973835e-06,
      "loss": 0.3929,
      "step": 4434
    },
    {
      "epoch": 0.27779952708310496,
      "grad_norm": 0.8629140596461519,
      "learning_rate": 8.474224126610168e-06,
      "loss": 0.4296,
      "step": 4435
    },
    {
      "epoch": 0.27786216508244727,
      "grad_norm": 0.888438469013795,
      "learning_rate": 8.473494540482279e-06,
      "loss": 0.4391,
      "step": 4436
    },
    {
      "epoch": 0.2779248030817896,
      "grad_norm": 0.8029567280590286,
      "learning_rate": 8.472764811384708e-06,
      "loss": 0.3788,
      "step": 4437
    },
    {
      "epoch": 0.2779874410811319,
      "grad_norm": 0.835363335414453,
      "learning_rate": 8.472034939347493e-06,
      "loss": 0.4338,
      "step": 4438
    },
    {
      "epoch": 0.2780500790804742,
      "grad_norm": 0.9116309719211518,
      "learning_rate": 8.471304924400676e-06,
      "loss": 0.4928,
      "step": 4439
    },
    {
      "epoch": 0.2781127170798165,
      "grad_norm": 0.8157635537044745,
      "learning_rate": 8.470574766574304e-06,
      "loss": 0.4119,
      "step": 4440
    },
    {
      "epoch": 0.27817535507915875,
      "grad_norm": 0.8377107975290504,
      "learning_rate": 8.469844465898431e-06,
      "loss": 0.424,
      "step": 4441
    },
    {
      "epoch": 0.27823799307850106,
      "grad_norm": 0.7777005926653328,
      "learning_rate": 8.469114022403115e-06,
      "loss": 0.4105,
      "step": 4442
    },
    {
      "epoch": 0.27830063107784336,
      "grad_norm": 0.9189570360144972,
      "learning_rate": 8.468383436118422e-06,
      "loss": 0.4672,
      "step": 4443
    },
    {
      "epoch": 0.27836326907718567,
      "grad_norm": 0.8280378017251758,
      "learning_rate": 8.467652707074423e-06,
      "loss": 0.3892,
      "step": 4444
    },
    {
      "epoch": 0.278425907076528,
      "grad_norm": 0.8464483908806905,
      "learning_rate": 8.466921835301195e-06,
      "loss": 0.4058,
      "step": 4445
    },
    {
      "epoch": 0.2784885450758703,
      "grad_norm": 0.8802470678553086,
      "learning_rate": 8.46619082082882e-06,
      "loss": 0.3936,
      "step": 4446
    },
    {
      "epoch": 0.2785511830752126,
      "grad_norm": 0.8611961069396535,
      "learning_rate": 8.465459663687387e-06,
      "loss": 0.4565,
      "step": 4447
    },
    {
      "epoch": 0.2786138210745549,
      "grad_norm": 0.8034794121229686,
      "learning_rate": 8.464728363906993e-06,
      "loss": 0.373,
      "step": 4448
    },
    {
      "epoch": 0.27867645907389715,
      "grad_norm": 0.8293760677112963,
      "learning_rate": 8.463996921517736e-06,
      "loss": 0.4453,
      "step": 4449
    },
    {
      "epoch": 0.27873909707323946,
      "grad_norm": 0.8012125694525203,
      "learning_rate": 8.463265336549722e-06,
      "loss": 0.3807,
      "step": 4450
    },
    {
      "epoch": 0.27880173507258177,
      "grad_norm": 0.7833819368043001,
      "learning_rate": 8.462533609033065e-06,
      "loss": 0.3992,
      "step": 4451
    },
    {
      "epoch": 0.2788643730719241,
      "grad_norm": 0.9147642316197814,
      "learning_rate": 8.461801738997881e-06,
      "loss": 0.4197,
      "step": 4452
    },
    {
      "epoch": 0.2789270110712664,
      "grad_norm": 0.8851205384973309,
      "learning_rate": 8.461069726474295e-06,
      "loss": 0.4524,
      "step": 4453
    },
    {
      "epoch": 0.2789896490706087,
      "grad_norm": 0.8281814661081852,
      "learning_rate": 8.460337571492435e-06,
      "loss": 0.4089,
      "step": 4454
    },
    {
      "epoch": 0.279052287069951,
      "grad_norm": 0.9247996872372588,
      "learning_rate": 8.459605274082441e-06,
      "loss": 0.428,
      "step": 4455
    },
    {
      "epoch": 0.2791149250692933,
      "grad_norm": 0.8323314471910815,
      "learning_rate": 8.458872834274452e-06,
      "loss": 0.4143,
      "step": 4456
    },
    {
      "epoch": 0.2791775630686356,
      "grad_norm": 0.8292869878201337,
      "learning_rate": 8.458140252098615e-06,
      "loss": 0.4518,
      "step": 4457
    },
    {
      "epoch": 0.27924020106797787,
      "grad_norm": 0.8882208715215865,
      "learning_rate": 8.457407527585082e-06,
      "loss": 0.4191,
      "step": 4458
    },
    {
      "epoch": 0.2793028390673202,
      "grad_norm": 0.8370285298639575,
      "learning_rate": 8.456674660764013e-06,
      "loss": 0.4582,
      "step": 4459
    },
    {
      "epoch": 0.2793654770666625,
      "grad_norm": 0.8500355667324807,
      "learning_rate": 8.455941651665574e-06,
      "loss": 0.4384,
      "step": 4460
    },
    {
      "epoch": 0.2794281150660048,
      "grad_norm": 0.8469359387901619,
      "learning_rate": 8.455208500319936e-06,
      "loss": 0.4254,
      "step": 4461
    },
    {
      "epoch": 0.2794907530653471,
      "grad_norm": 0.8761183610078126,
      "learning_rate": 8.454475206757275e-06,
      "loss": 0.4426,
      "step": 4462
    },
    {
      "epoch": 0.2795533910646894,
      "grad_norm": 0.8475887025562969,
      "learning_rate": 8.453741771007772e-06,
      "loss": 0.4318,
      "step": 4463
    },
    {
      "epoch": 0.2796160290640317,
      "grad_norm": 0.8232020975370907,
      "learning_rate": 8.453008193101617e-06,
      "loss": 0.4072,
      "step": 4464
    },
    {
      "epoch": 0.279678667063374,
      "grad_norm": 0.8349937227732535,
      "learning_rate": 8.452274473069005e-06,
      "loss": 0.4048,
      "step": 4465
    },
    {
      "epoch": 0.2797413050627163,
      "grad_norm": 0.7430580599265124,
      "learning_rate": 8.451540610940133e-06,
      "loss": 0.4696,
      "step": 4466
    },
    {
      "epoch": 0.2798039430620586,
      "grad_norm": 0.8266480115293178,
      "learning_rate": 8.45080660674521e-06,
      "loss": 0.4142,
      "step": 4467
    },
    {
      "epoch": 0.2798665810614009,
      "grad_norm": 0.7747035461360926,
      "learning_rate": 8.450072460514445e-06,
      "loss": 0.4085,
      "step": 4468
    },
    {
      "epoch": 0.2799292190607432,
      "grad_norm": 0.9183722291204025,
      "learning_rate": 8.44933817227806e-06,
      "loss": 0.4534,
      "step": 4469
    },
    {
      "epoch": 0.2799918570600855,
      "grad_norm": 0.8430629031982132,
      "learning_rate": 8.448603742066272e-06,
      "loss": 0.4276,
      "step": 4470
    },
    {
      "epoch": 0.2800544950594278,
      "grad_norm": 0.8485708248659706,
      "learning_rate": 8.447869169909318e-06,
      "loss": 0.4206,
      "step": 4471
    },
    {
      "epoch": 0.2801171330587701,
      "grad_norm": 0.6542292607688736,
      "learning_rate": 8.447134455837427e-06,
      "loss": 0.4639,
      "step": 4472
    },
    {
      "epoch": 0.2801797710581124,
      "grad_norm": 0.8006173075995354,
      "learning_rate": 8.44639959988084e-06,
      "loss": 0.4152,
      "step": 4473
    },
    {
      "epoch": 0.2802424090574547,
      "grad_norm": 0.8577385149241405,
      "learning_rate": 8.445664602069809e-06,
      "loss": 0.3876,
      "step": 4474
    },
    {
      "epoch": 0.280305047056797,
      "grad_norm": 0.7699279835162418,
      "learning_rate": 8.444929462434582e-06,
      "loss": 0.4081,
      "step": 4475
    },
    {
      "epoch": 0.2803676850561393,
      "grad_norm": 0.8632558988862935,
      "learning_rate": 8.44419418100542e-06,
      "loss": 0.4525,
      "step": 4476
    },
    {
      "epoch": 0.2804303230554816,
      "grad_norm": 0.8174327990788561,
      "learning_rate": 8.443458757812583e-06,
      "loss": 0.439,
      "step": 4477
    },
    {
      "epoch": 0.2804929610548239,
      "grad_norm": 0.8390206924749255,
      "learning_rate": 8.442723192886349e-06,
      "loss": 0.4287,
      "step": 4478
    },
    {
      "epoch": 0.2805555990541662,
      "grad_norm": 0.8895683908965598,
      "learning_rate": 8.441987486256986e-06,
      "loss": 0.4231,
      "step": 4479
    },
    {
      "epoch": 0.2806182370535085,
      "grad_norm": 0.8422907238365351,
      "learning_rate": 8.441251637954781e-06,
      "loss": 0.4119,
      "step": 4480
    },
    {
      "epoch": 0.28068087505285083,
      "grad_norm": 0.8798004012355268,
      "learning_rate": 8.44051564801002e-06,
      "loss": 0.4271,
      "step": 4481
    },
    {
      "epoch": 0.28074351305219314,
      "grad_norm": 0.9162956688051097,
      "learning_rate": 8.439779516452994e-06,
      "loss": 0.4244,
      "step": 4482
    },
    {
      "epoch": 0.2808061510515354,
      "grad_norm": 0.9387709303606743,
      "learning_rate": 8.439043243314006e-06,
      "loss": 0.4411,
      "step": 4483
    },
    {
      "epoch": 0.2808687890508777,
      "grad_norm": 0.7791824355802911,
      "learning_rate": 8.43830682862336e-06,
      "loss": 0.5019,
      "step": 4484
    },
    {
      "epoch": 0.28093142705022,
      "grad_norm": 0.9065703295401388,
      "learning_rate": 8.437570272411368e-06,
      "loss": 0.4135,
      "step": 4485
    },
    {
      "epoch": 0.2809940650495623,
      "grad_norm": 0.7438404973777876,
      "learning_rate": 8.436833574708343e-06,
      "loss": 0.3841,
      "step": 4486
    },
    {
      "epoch": 0.2810567030489046,
      "grad_norm": 0.6638437396309617,
      "learning_rate": 8.436096735544612e-06,
      "loss": 0.4812,
      "step": 4487
    },
    {
      "epoch": 0.28111934104824693,
      "grad_norm": 0.9019747460027647,
      "learning_rate": 8.435359754950501e-06,
      "loss": 0.5011,
      "step": 4488
    },
    {
      "epoch": 0.28118197904758924,
      "grad_norm": 0.780817237652749,
      "learning_rate": 8.434622632956345e-06,
      "loss": 0.3743,
      "step": 4489
    },
    {
      "epoch": 0.28124461704693154,
      "grad_norm": 0.8588540210354703,
      "learning_rate": 8.433885369592482e-06,
      "loss": 0.4133,
      "step": 4490
    },
    {
      "epoch": 0.2813072550462738,
      "grad_norm": 0.8877931370258269,
      "learning_rate": 8.433147964889263e-06,
      "loss": 0.4449,
      "step": 4491
    },
    {
      "epoch": 0.2813698930456161,
      "grad_norm": 0.8869661196646803,
      "learning_rate": 8.432410418877035e-06,
      "loss": 0.4656,
      "step": 4492
    },
    {
      "epoch": 0.2814325310449584,
      "grad_norm": 0.8431416608472331,
      "learning_rate": 8.431672731586157e-06,
      "loss": 0.445,
      "step": 4493
    },
    {
      "epoch": 0.2814951690443007,
      "grad_norm": 0.8928545977082609,
      "learning_rate": 8.430934903046993e-06,
      "loss": 0.4332,
      "step": 4494
    },
    {
      "epoch": 0.281557807043643,
      "grad_norm": 0.861192931468011,
      "learning_rate": 8.43019693328991e-06,
      "loss": 0.4555,
      "step": 4495
    },
    {
      "epoch": 0.28162044504298533,
      "grad_norm": 0.8007904994052102,
      "learning_rate": 8.429458822345287e-06,
      "loss": 0.4468,
      "step": 4496
    },
    {
      "epoch": 0.28168308304232764,
      "grad_norm": 0.8636905088543667,
      "learning_rate": 8.428720570243501e-06,
      "loss": 0.51,
      "step": 4497
    },
    {
      "epoch": 0.28174572104166995,
      "grad_norm": 0.7882641011984228,
      "learning_rate": 8.427982177014941e-06,
      "loss": 0.3521,
      "step": 4498
    },
    {
      "epoch": 0.28180835904101226,
      "grad_norm": 0.88031674780118,
      "learning_rate": 8.427243642689997e-06,
      "loss": 0.4414,
      "step": 4499
    },
    {
      "epoch": 0.2818709970403545,
      "grad_norm": 0.8746543742715582,
      "learning_rate": 8.426504967299067e-06,
      "loss": 0.4238,
      "step": 4500
    },
    {
      "epoch": 0.2819336350396968,
      "grad_norm": 0.8183299525883578,
      "learning_rate": 8.42576615087256e-06,
      "loss": 0.4311,
      "step": 4501
    },
    {
      "epoch": 0.2819962730390391,
      "grad_norm": 0.7774276500562789,
      "learning_rate": 8.42502719344088e-06,
      "loss": 0.4006,
      "step": 4502
    },
    {
      "epoch": 0.28205891103838143,
      "grad_norm": 0.8730530687122681,
      "learning_rate": 8.424288095034446e-06,
      "loss": 0.4367,
      "step": 4503
    },
    {
      "epoch": 0.28212154903772374,
      "grad_norm": 0.7688278470786755,
      "learning_rate": 8.423548855683679e-06,
      "loss": 0.3729,
      "step": 4504
    },
    {
      "epoch": 0.28218418703706605,
      "grad_norm": 0.8177250654066277,
      "learning_rate": 8.422809475419005e-06,
      "loss": 0.4334,
      "step": 4505
    },
    {
      "epoch": 0.28224682503640836,
      "grad_norm": 0.845049820319112,
      "learning_rate": 8.422069954270858e-06,
      "loss": 0.4407,
      "step": 4506
    },
    {
      "epoch": 0.28230946303575066,
      "grad_norm": 0.9153848550415568,
      "learning_rate": 8.421330292269676e-06,
      "loss": 0.4231,
      "step": 4507
    },
    {
      "epoch": 0.2823721010350929,
      "grad_norm": 0.7320751359689285,
      "learning_rate": 8.420590489445903e-06,
      "loss": 0.3572,
      "step": 4508
    },
    {
      "epoch": 0.2824347390344352,
      "grad_norm": 0.8486031290236098,
      "learning_rate": 8.419850545829993e-06,
      "loss": 0.4245,
      "step": 4509
    },
    {
      "epoch": 0.28249737703377753,
      "grad_norm": 0.8186833846273959,
      "learning_rate": 8.419110461452397e-06,
      "loss": 0.4127,
      "step": 4510
    },
    {
      "epoch": 0.28256001503311984,
      "grad_norm": 0.8925574140126645,
      "learning_rate": 8.418370236343583e-06,
      "loss": 0.4614,
      "step": 4511
    },
    {
      "epoch": 0.28262265303246215,
      "grad_norm": 0.8245211518809294,
      "learning_rate": 8.417629870534013e-06,
      "loss": 0.4349,
      "step": 4512
    },
    {
      "epoch": 0.28268529103180445,
      "grad_norm": 0.9031647050548933,
      "learning_rate": 8.416889364054167e-06,
      "loss": 0.4491,
      "step": 4513
    },
    {
      "epoch": 0.28274792903114676,
      "grad_norm": 0.8462950603204374,
      "learning_rate": 8.416148716934516e-06,
      "loss": 0.3845,
      "step": 4514
    },
    {
      "epoch": 0.28281056703048907,
      "grad_norm": 0.8965035297364966,
      "learning_rate": 8.415407929205551e-06,
      "loss": 0.4057,
      "step": 4515
    },
    {
      "epoch": 0.2828732050298313,
      "grad_norm": 0.8325674788136882,
      "learning_rate": 8.414667000897762e-06,
      "loss": 0.4179,
      "step": 4516
    },
    {
      "epoch": 0.28293584302917363,
      "grad_norm": 0.8535993659281004,
      "learning_rate": 8.413925932041647e-06,
      "loss": 0.4373,
      "step": 4517
    },
    {
      "epoch": 0.28299848102851594,
      "grad_norm": 0.8220744629681055,
      "learning_rate": 8.413184722667705e-06,
      "loss": 0.4234,
      "step": 4518
    },
    {
      "epoch": 0.28306111902785824,
      "grad_norm": 0.8491761830580135,
      "learning_rate": 8.412443372806448e-06,
      "loss": 0.4098,
      "step": 4519
    },
    {
      "epoch": 0.28312375702720055,
      "grad_norm": 0.8116566590517976,
      "learning_rate": 8.411701882488385e-06,
      "loss": 0.4253,
      "step": 4520
    },
    {
      "epoch": 0.28318639502654286,
      "grad_norm": 0.9024882625815372,
      "learning_rate": 8.410960251744042e-06,
      "loss": 0.3905,
      "step": 4521
    },
    {
      "epoch": 0.28324903302588517,
      "grad_norm": 0.7699582910260983,
      "learning_rate": 8.410218480603942e-06,
      "loss": 0.3922,
      "step": 4522
    },
    {
      "epoch": 0.2833116710252275,
      "grad_norm": 0.8499652700254379,
      "learning_rate": 8.409476569098616e-06,
      "loss": 0.4968,
      "step": 4523
    },
    {
      "epoch": 0.2833743090245698,
      "grad_norm": 0.8327557432156951,
      "learning_rate": 8.4087345172586e-06,
      "loss": 0.407,
      "step": 4524
    },
    {
      "epoch": 0.28343694702391203,
      "grad_norm": 0.7259577120830983,
      "learning_rate": 8.407992325114442e-06,
      "loss": 0.4808,
      "step": 4525
    },
    {
      "epoch": 0.28349958502325434,
      "grad_norm": 0.8609384529022338,
      "learning_rate": 8.407249992696685e-06,
      "loss": 0.4313,
      "step": 4526
    },
    {
      "epoch": 0.28356222302259665,
      "grad_norm": 0.8055980159047935,
      "learning_rate": 8.406507520035887e-06,
      "loss": 0.4247,
      "step": 4527
    },
    {
      "epoch": 0.28362486102193896,
      "grad_norm": 0.9501564570764383,
      "learning_rate": 8.405764907162607e-06,
      "loss": 0.4494,
      "step": 4528
    },
    {
      "epoch": 0.28368749902128126,
      "grad_norm": 0.8406642545027252,
      "learning_rate": 8.40502215410741e-06,
      "loss": 0.4623,
      "step": 4529
    },
    {
      "epoch": 0.28375013702062357,
      "grad_norm": 0.7823595518914992,
      "learning_rate": 8.40427926090087e-06,
      "loss": 0.3841,
      "step": 4530
    },
    {
      "epoch": 0.2838127750199659,
      "grad_norm": 0.8055630711773765,
      "learning_rate": 8.403536227573564e-06,
      "loss": 0.4318,
      "step": 4531
    },
    {
      "epoch": 0.2838754130193082,
      "grad_norm": 0.7538077845481955,
      "learning_rate": 8.402793054156076e-06,
      "loss": 0.4678,
      "step": 4532
    },
    {
      "epoch": 0.28393805101865044,
      "grad_norm": 0.8627177219962536,
      "learning_rate": 8.402049740678992e-06,
      "loss": 0.4606,
      "step": 4533
    },
    {
      "epoch": 0.28400068901799275,
      "grad_norm": 0.8812955485177362,
      "learning_rate": 8.40130628717291e-06,
      "loss": 0.4377,
      "step": 4534
    },
    {
      "epoch": 0.28406332701733505,
      "grad_norm": 0.8985200832206445,
      "learning_rate": 8.40056269366843e-06,
      "loss": 0.4412,
      "step": 4535
    },
    {
      "epoch": 0.28412596501667736,
      "grad_norm": 0.8046156077725234,
      "learning_rate": 8.399818960196159e-06,
      "loss": 0.4286,
      "step": 4536
    },
    {
      "epoch": 0.28418860301601967,
      "grad_norm": 0.8968933332815846,
      "learning_rate": 8.399075086786708e-06,
      "loss": 0.4302,
      "step": 4537
    },
    {
      "epoch": 0.284251241015362,
      "grad_norm": 0.8975745534516543,
      "learning_rate": 8.398331073470696e-06,
      "loss": 0.4355,
      "step": 4538
    },
    {
      "epoch": 0.2843138790147043,
      "grad_norm": 0.8825580612614147,
      "learning_rate": 8.397586920278747e-06,
      "loss": 0.4311,
      "step": 4539
    },
    {
      "epoch": 0.2843765170140466,
      "grad_norm": 0.7985635382766482,
      "learning_rate": 8.396842627241488e-06,
      "loss": 0.4364,
      "step": 4540
    },
    {
      "epoch": 0.28443915501338884,
      "grad_norm": 0.8404366891398226,
      "learning_rate": 8.396098194389556e-06,
      "loss": 0.4093,
      "step": 4541
    },
    {
      "epoch": 0.28450179301273115,
      "grad_norm": 0.8410694777437387,
      "learning_rate": 8.395353621753592e-06,
      "loss": 0.3949,
      "step": 4542
    },
    {
      "epoch": 0.28456443101207346,
      "grad_norm": 0.8213176287800736,
      "learning_rate": 8.394608909364243e-06,
      "loss": 0.4171,
      "step": 4543
    },
    {
      "epoch": 0.28462706901141577,
      "grad_norm": 0.8918335480504039,
      "learning_rate": 8.393864057252162e-06,
      "loss": 0.3983,
      "step": 4544
    },
    {
      "epoch": 0.2846897070107581,
      "grad_norm": 0.6608846592114112,
      "learning_rate": 8.393119065448008e-06,
      "loss": 0.4887,
      "step": 4545
    },
    {
      "epoch": 0.2847523450101004,
      "grad_norm": 0.7979826588457201,
      "learning_rate": 8.39237393398244e-06,
      "loss": 0.4014,
      "step": 4546
    },
    {
      "epoch": 0.2848149830094427,
      "grad_norm": 0.8366003094445484,
      "learning_rate": 8.391628662886134e-06,
      "loss": 0.4426,
      "step": 4547
    },
    {
      "epoch": 0.284877621008785,
      "grad_norm": 0.860255340205257,
      "learning_rate": 8.39088325218976e-06,
      "loss": 0.4476,
      "step": 4548
    },
    {
      "epoch": 0.2849402590081273,
      "grad_norm": 0.781542890893512,
      "learning_rate": 8.390137701924003e-06,
      "loss": 0.4114,
      "step": 4549
    },
    {
      "epoch": 0.28500289700746956,
      "grad_norm": 0.8309438586306174,
      "learning_rate": 8.38939201211955e-06,
      "loss": 0.4511,
      "step": 4550
    },
    {
      "epoch": 0.28506553500681187,
      "grad_norm": 0.9276841823885593,
      "learning_rate": 8.388646182807092e-06,
      "loss": 0.4488,
      "step": 4551
    },
    {
      "epoch": 0.2851281730061542,
      "grad_norm": 0.8469701553182943,
      "learning_rate": 8.387900214017328e-06,
      "loss": 0.4422,
      "step": 4552
    },
    {
      "epoch": 0.2851908110054965,
      "grad_norm": 0.8474472166244699,
      "learning_rate": 8.387154105780963e-06,
      "loss": 0.4242,
      "step": 4553
    },
    {
      "epoch": 0.2852534490048388,
      "grad_norm": 0.7861425166930543,
      "learning_rate": 8.386407858128707e-06,
      "loss": 0.4261,
      "step": 4554
    },
    {
      "epoch": 0.2853160870041811,
      "grad_norm": 0.8468881876423646,
      "learning_rate": 8.385661471091273e-06,
      "loss": 0.4328,
      "step": 4555
    },
    {
      "epoch": 0.2853787250035234,
      "grad_norm": 0.8775769448572558,
      "learning_rate": 8.384914944699385e-06,
      "loss": 0.4693,
      "step": 4556
    },
    {
      "epoch": 0.2854413630028657,
      "grad_norm": 0.7478369485962452,
      "learning_rate": 8.384168278983772e-06,
      "loss": 0.3636,
      "step": 4557
    },
    {
      "epoch": 0.28550400100220796,
      "grad_norm": 0.6479622556458028,
      "learning_rate": 8.383421473975162e-06,
      "loss": 0.4846,
      "step": 4558
    },
    {
      "epoch": 0.28556663900155027,
      "grad_norm": 0.6921687063237609,
      "learning_rate": 8.382674529704296e-06,
      "loss": 0.4908,
      "step": 4559
    },
    {
      "epoch": 0.2856292770008926,
      "grad_norm": 0.8045103021598321,
      "learning_rate": 8.381927446201919e-06,
      "loss": 0.45,
      "step": 4560
    },
    {
      "epoch": 0.2856919150002349,
      "grad_norm": 0.792991574508108,
      "learning_rate": 8.38118022349878e-06,
      "loss": 0.3953,
      "step": 4561
    },
    {
      "epoch": 0.2857545529995772,
      "grad_norm": 0.860182308584712,
      "learning_rate": 8.380432861625637e-06,
      "loss": 0.425,
      "step": 4562
    },
    {
      "epoch": 0.2858171909989195,
      "grad_norm": 0.8894312609468864,
      "learning_rate": 8.379685360613248e-06,
      "loss": 0.463,
      "step": 4563
    },
    {
      "epoch": 0.2858798289982618,
      "grad_norm": 0.7935107035960371,
      "learning_rate": 8.378937720492384e-06,
      "loss": 0.4184,
      "step": 4564
    },
    {
      "epoch": 0.2859424669976041,
      "grad_norm": 0.9280838626678545,
      "learning_rate": 8.378189941293815e-06,
      "loss": 0.4164,
      "step": 4565
    },
    {
      "epoch": 0.2860051049969464,
      "grad_norm": 0.686159480813328,
      "learning_rate": 8.377442023048323e-06,
      "loss": 0.4928,
      "step": 4566
    },
    {
      "epoch": 0.2860677429962887,
      "grad_norm": 0.8358448000674422,
      "learning_rate": 8.376693965786688e-06,
      "loss": 0.4346,
      "step": 4567
    },
    {
      "epoch": 0.286130380995631,
      "grad_norm": 0.8292946810615583,
      "learning_rate": 8.375945769539704e-06,
      "loss": 0.4203,
      "step": 4568
    },
    {
      "epoch": 0.2861930189949733,
      "grad_norm": 0.8528568944769392,
      "learning_rate": 8.375197434338164e-06,
      "loss": 0.4726,
      "step": 4569
    },
    {
      "epoch": 0.2862556569943156,
      "grad_norm": 0.8729948069040444,
      "learning_rate": 8.374448960212873e-06,
      "loss": 0.4689,
      "step": 4570
    },
    {
      "epoch": 0.2863182949936579,
      "grad_norm": 0.6992213001896993,
      "learning_rate": 8.373700347194637e-06,
      "loss": 0.4695,
      "step": 4571
    },
    {
      "epoch": 0.2863809329930002,
      "grad_norm": 0.8805033902749299,
      "learning_rate": 8.372951595314268e-06,
      "loss": 0.4218,
      "step": 4572
    },
    {
      "epoch": 0.2864435709923425,
      "grad_norm": 0.8194417206296722,
      "learning_rate": 8.372202704602585e-06,
      "loss": 0.3927,
      "step": 4573
    },
    {
      "epoch": 0.28650620899168483,
      "grad_norm": 0.8556483700079822,
      "learning_rate": 8.371453675090411e-06,
      "loss": 0.4244,
      "step": 4574
    },
    {
      "epoch": 0.2865688469910271,
      "grad_norm": 0.8108283602021576,
      "learning_rate": 8.37070450680858e-06,
      "loss": 0.4307,
      "step": 4575
    },
    {
      "epoch": 0.2866314849903694,
      "grad_norm": 0.8208834526299716,
      "learning_rate": 8.369955199787925e-06,
      "loss": 0.4067,
      "step": 4576
    },
    {
      "epoch": 0.2866941229897117,
      "grad_norm": 0.8082671786589473,
      "learning_rate": 8.369205754059289e-06,
      "loss": 0.4583,
      "step": 4577
    },
    {
      "epoch": 0.286756760989054,
      "grad_norm": 0.8377806156967093,
      "learning_rate": 8.368456169653519e-06,
      "loss": 0.4429,
      "step": 4578
    },
    {
      "epoch": 0.2868193989883963,
      "grad_norm": 0.7605655215949123,
      "learning_rate": 8.367706446601466e-06,
      "loss": 0.3964,
      "step": 4579
    },
    {
      "epoch": 0.2868820369877386,
      "grad_norm": 0.8666580154759712,
      "learning_rate": 8.366956584933992e-06,
      "loss": 0.4352,
      "step": 4580
    },
    {
      "epoch": 0.2869446749870809,
      "grad_norm": 0.8260359629848951,
      "learning_rate": 8.36620658468196e-06,
      "loss": 0.3861,
      "step": 4581
    },
    {
      "epoch": 0.28700731298642324,
      "grad_norm": 0.8292055298147408,
      "learning_rate": 8.36545644587624e-06,
      "loss": 0.3986,
      "step": 4582
    },
    {
      "epoch": 0.2870699509857655,
      "grad_norm": 0.7950863570016041,
      "learning_rate": 8.36470616854771e-06,
      "loss": 0.4094,
      "step": 4583
    },
    {
      "epoch": 0.2871325889851078,
      "grad_norm": 0.8432318095562448,
      "learning_rate": 8.363955752727247e-06,
      "loss": 0.4512,
      "step": 4584
    },
    {
      "epoch": 0.2871952269844501,
      "grad_norm": 0.8510153136061384,
      "learning_rate": 8.36320519844574e-06,
      "loss": 0.3953,
      "step": 4585
    },
    {
      "epoch": 0.2872578649837924,
      "grad_norm": 0.8923641748955161,
      "learning_rate": 8.362454505734085e-06,
      "loss": 0.4168,
      "step": 4586
    },
    {
      "epoch": 0.2873205029831347,
      "grad_norm": 0.9021367066478626,
      "learning_rate": 8.361703674623179e-06,
      "loss": 0.4198,
      "step": 4587
    },
    {
      "epoch": 0.287383140982477,
      "grad_norm": 0.8183372061811466,
      "learning_rate": 8.360952705143924e-06,
      "loss": 0.4085,
      "step": 4588
    },
    {
      "epoch": 0.28744577898181933,
      "grad_norm": 0.6517471790667465,
      "learning_rate": 8.360201597327234e-06,
      "loss": 0.4901,
      "step": 4589
    },
    {
      "epoch": 0.28750841698116164,
      "grad_norm": 0.8552492816403398,
      "learning_rate": 8.35945035120402e-06,
      "loss": 0.4677,
      "step": 4590
    },
    {
      "epoch": 0.28757105498050395,
      "grad_norm": 0.7936449470875007,
      "learning_rate": 8.358698966805208e-06,
      "loss": 0.4349,
      "step": 4591
    },
    {
      "epoch": 0.2876336929798462,
      "grad_norm": 0.721881400775441,
      "learning_rate": 8.35794744416172e-06,
      "loss": 0.4736,
      "step": 4592
    },
    {
      "epoch": 0.2876963309791885,
      "grad_norm": 0.9148032123242769,
      "learning_rate": 8.357195783304495e-06,
      "loss": 0.4317,
      "step": 4593
    },
    {
      "epoch": 0.2877589689785308,
      "grad_norm": 0.7878970544853151,
      "learning_rate": 8.356443984264468e-06,
      "loss": 0.4082,
      "step": 4594
    },
    {
      "epoch": 0.2878216069778731,
      "grad_norm": 0.7997682581998039,
      "learning_rate": 8.355692047072584e-06,
      "loss": 0.4329,
      "step": 4595
    },
    {
      "epoch": 0.28788424497721543,
      "grad_norm": 0.8090189333634814,
      "learning_rate": 8.35493997175979e-06,
      "loss": 0.4536,
      "step": 4596
    },
    {
      "epoch": 0.28794688297655774,
      "grad_norm": 0.8312142684419676,
      "learning_rate": 8.354187758357046e-06,
      "loss": 0.4694,
      "step": 4597
    },
    {
      "epoch": 0.28800952097590005,
      "grad_norm": 0.8739342088039931,
      "learning_rate": 8.353435406895312e-06,
      "loss": 0.4078,
      "step": 4598
    },
    {
      "epoch": 0.28807215897524235,
      "grad_norm": 0.8363434223357454,
      "learning_rate": 8.352682917405552e-06,
      "loss": 0.4718,
      "step": 4599
    },
    {
      "epoch": 0.2881347969745846,
      "grad_norm": 0.8176272164661651,
      "learning_rate": 8.351930289918743e-06,
      "loss": 0.396,
      "step": 4600
    },
    {
      "epoch": 0.2881974349739269,
      "grad_norm": 0.7513051720032988,
      "learning_rate": 8.35117752446586e-06,
      "loss": 0.3926,
      "step": 4601
    },
    {
      "epoch": 0.2882600729732692,
      "grad_norm": 0.8482313169861686,
      "learning_rate": 8.350424621077889e-06,
      "loss": 0.4136,
      "step": 4602
    },
    {
      "epoch": 0.28832271097261153,
      "grad_norm": 0.7760307598353499,
      "learning_rate": 8.349671579785817e-06,
      "loss": 0.3968,
      "step": 4603
    },
    {
      "epoch": 0.28838534897195384,
      "grad_norm": 0.8535403679621641,
      "learning_rate": 8.348918400620643e-06,
      "loss": 0.3908,
      "step": 4604
    },
    {
      "epoch": 0.28844798697129614,
      "grad_norm": 0.7603145667110754,
      "learning_rate": 8.348165083613365e-06,
      "loss": 0.3862,
      "step": 4605
    },
    {
      "epoch": 0.28851062497063845,
      "grad_norm": 0.7195484834137589,
      "learning_rate": 8.34741162879499e-06,
      "loss": 0.4711,
      "step": 4606
    },
    {
      "epoch": 0.28857326296998076,
      "grad_norm": 0.8102594974796218,
      "learning_rate": 8.346658036196533e-06,
      "loss": 0.3762,
      "step": 4607
    },
    {
      "epoch": 0.28863590096932307,
      "grad_norm": 0.7465374118885104,
      "learning_rate": 8.345904305849008e-06,
      "loss": 0.4322,
      "step": 4608
    },
    {
      "epoch": 0.2886985389686653,
      "grad_norm": 0.807764072812133,
      "learning_rate": 8.345150437783442e-06,
      "loss": 0.4202,
      "step": 4609
    },
    {
      "epoch": 0.2887611769680076,
      "grad_norm": 0.818110504961382,
      "learning_rate": 8.344396432030864e-06,
      "loss": 0.4221,
      "step": 4610
    },
    {
      "epoch": 0.28882381496734993,
      "grad_norm": 0.8487899842265424,
      "learning_rate": 8.343642288622305e-06,
      "loss": 0.4459,
      "step": 4611
    },
    {
      "epoch": 0.28888645296669224,
      "grad_norm": 0.8498942389786289,
      "learning_rate": 8.34288800758881e-06,
      "loss": 0.4093,
      "step": 4612
    },
    {
      "epoch": 0.28894909096603455,
      "grad_norm": 0.6768746653230164,
      "learning_rate": 8.342133588961425e-06,
      "loss": 0.4939,
      "step": 4613
    },
    {
      "epoch": 0.28901172896537686,
      "grad_norm": 0.7655078484042608,
      "learning_rate": 8.341379032771202e-06,
      "loss": 0.3852,
      "step": 4614
    },
    {
      "epoch": 0.28907436696471916,
      "grad_norm": 0.7983857594247828,
      "learning_rate": 8.340624339049196e-06,
      "loss": 0.3658,
      "step": 4615
    },
    {
      "epoch": 0.2891370049640615,
      "grad_norm": 0.8465241297815904,
      "learning_rate": 8.339869507826474e-06,
      "loss": 0.4305,
      "step": 4616
    },
    {
      "epoch": 0.2891996429634037,
      "grad_norm": 0.9291578381833316,
      "learning_rate": 8.3391145391341e-06,
      "loss": 0.4183,
      "step": 4617
    },
    {
      "epoch": 0.28926228096274603,
      "grad_norm": 0.8674285395494015,
      "learning_rate": 8.338359433003155e-06,
      "loss": 0.4394,
      "step": 4618
    },
    {
      "epoch": 0.28932491896208834,
      "grad_norm": 0.7872338508372421,
      "learning_rate": 8.337604189464716e-06,
      "loss": 0.3858,
      "step": 4619
    },
    {
      "epoch": 0.28938755696143065,
      "grad_norm": 0.8445375094790136,
      "learning_rate": 8.336848808549866e-06,
      "loss": 0.4395,
      "step": 4620
    },
    {
      "epoch": 0.28945019496077296,
      "grad_norm": 0.7471784282009074,
      "learning_rate": 8.336093290289703e-06,
      "loss": 0.3975,
      "step": 4621
    },
    {
      "epoch": 0.28951283296011526,
      "grad_norm": 0.8807929869245109,
      "learning_rate": 8.33533763471532e-06,
      "loss": 0.4162,
      "step": 4622
    },
    {
      "epoch": 0.28957547095945757,
      "grad_norm": 0.7415396553270062,
      "learning_rate": 8.334581841857821e-06,
      "loss": 0.4722,
      "step": 4623
    },
    {
      "epoch": 0.2896381089587999,
      "grad_norm": 0.6978168918434557,
      "learning_rate": 8.333825911748314e-06,
      "loss": 0.4656,
      "step": 4624
    },
    {
      "epoch": 0.28970074695814213,
      "grad_norm": 0.8031343390110353,
      "learning_rate": 8.333069844417914e-06,
      "loss": 0.3742,
      "step": 4625
    },
    {
      "epoch": 0.28976338495748444,
      "grad_norm": 0.8204884685893606,
      "learning_rate": 8.33231363989774e-06,
      "loss": 0.3975,
      "step": 4626
    },
    {
      "epoch": 0.28982602295682675,
      "grad_norm": 0.9092791041395636,
      "learning_rate": 8.331557298218917e-06,
      "loss": 0.4395,
      "step": 4627
    },
    {
      "epoch": 0.28988866095616905,
      "grad_norm": 0.8917595959460629,
      "learning_rate": 8.33080081941258e-06,
      "loss": 0.3937,
      "step": 4628
    },
    {
      "epoch": 0.28995129895551136,
      "grad_norm": 0.8469314222348321,
      "learning_rate": 8.330044203509862e-06,
      "loss": 0.4239,
      "step": 4629
    },
    {
      "epoch": 0.29001393695485367,
      "grad_norm": 0.866630860024013,
      "learning_rate": 8.329287450541905e-06,
      "loss": 0.4748,
      "step": 4630
    },
    {
      "epoch": 0.290076574954196,
      "grad_norm": 0.8391019617923429,
      "learning_rate": 8.32853056053986e-06,
      "loss": 0.3986,
      "step": 4631
    },
    {
      "epoch": 0.2901392129535383,
      "grad_norm": 0.9204742856913263,
      "learning_rate": 8.327773533534879e-06,
      "loss": 0.4072,
      "step": 4632
    },
    {
      "epoch": 0.2902018509528806,
      "grad_norm": 0.807241427691445,
      "learning_rate": 8.327016369558122e-06,
      "loss": 0.3906,
      "step": 4633
    },
    {
      "epoch": 0.29026448895222284,
      "grad_norm": 0.8805439644033244,
      "learning_rate": 8.326259068640755e-06,
      "loss": 0.4293,
      "step": 4634
    },
    {
      "epoch": 0.29032712695156515,
      "grad_norm": 0.9439031884767,
      "learning_rate": 8.325501630813945e-06,
      "loss": 0.4415,
      "step": 4635
    },
    {
      "epoch": 0.29038976495090746,
      "grad_norm": 0.8125735544480082,
      "learning_rate": 8.32474405610887e-06,
      "loss": 0.3933,
      "step": 4636
    },
    {
      "epoch": 0.29045240295024977,
      "grad_norm": 0.9493186220113874,
      "learning_rate": 8.323986344556716e-06,
      "loss": 0.4722,
      "step": 4637
    },
    {
      "epoch": 0.2905150409495921,
      "grad_norm": 0.8432451575107256,
      "learning_rate": 8.323228496188665e-06,
      "loss": 0.4272,
      "step": 4638
    },
    {
      "epoch": 0.2905776789489344,
      "grad_norm": 0.8508921751247125,
      "learning_rate": 8.322470511035912e-06,
      "loss": 0.4272,
      "step": 4639
    },
    {
      "epoch": 0.2906403169482767,
      "grad_norm": 0.8654976020577734,
      "learning_rate": 8.321712389129656e-06,
      "loss": 0.4622,
      "step": 4640
    },
    {
      "epoch": 0.290702954947619,
      "grad_norm": 0.8449308164063765,
      "learning_rate": 8.320954130501102e-06,
      "loss": 0.4431,
      "step": 4641
    },
    {
      "epoch": 0.29076559294696125,
      "grad_norm": 0.7752847781656992,
      "learning_rate": 8.32019573518146e-06,
      "loss": 0.4149,
      "step": 4642
    },
    {
      "epoch": 0.29082823094630356,
      "grad_norm": 0.807125065309348,
      "learning_rate": 8.319437203201944e-06,
      "loss": 0.4405,
      "step": 4643
    },
    {
      "epoch": 0.29089086894564586,
      "grad_norm": 0.849890624402011,
      "learning_rate": 8.318678534593778e-06,
      "loss": 0.4341,
      "step": 4644
    },
    {
      "epoch": 0.29095350694498817,
      "grad_norm": 0.7779799131768563,
      "learning_rate": 8.317919729388186e-06,
      "loss": 0.3962,
      "step": 4645
    },
    {
      "epoch": 0.2910161449443305,
      "grad_norm": 0.7505664193244607,
      "learning_rate": 8.317160787616404e-06,
      "loss": 0.3901,
      "step": 4646
    },
    {
      "epoch": 0.2910787829436728,
      "grad_norm": 0.8882407394526716,
      "learning_rate": 8.316401709309667e-06,
      "loss": 0.4194,
      "step": 4647
    },
    {
      "epoch": 0.2911414209430151,
      "grad_norm": 0.8305383810416257,
      "learning_rate": 8.31564249449922e-06,
      "loss": 0.4436,
      "step": 4648
    },
    {
      "epoch": 0.2912040589423574,
      "grad_norm": 0.7736816748939043,
      "learning_rate": 8.314883143216315e-06,
      "loss": 0.3901,
      "step": 4649
    },
    {
      "epoch": 0.2912666969416997,
      "grad_norm": 0.8459933326178389,
      "learning_rate": 8.314123655492201e-06,
      "loss": 0.4564,
      "step": 4650
    },
    {
      "epoch": 0.29132933494104196,
      "grad_norm": 0.8146871945820008,
      "learning_rate": 8.313364031358145e-06,
      "loss": 0.4009,
      "step": 4651
    },
    {
      "epoch": 0.29139197294038427,
      "grad_norm": 0.807431689432583,
      "learning_rate": 8.312604270845409e-06,
      "loss": 0.4097,
      "step": 4652
    },
    {
      "epoch": 0.2914546109397266,
      "grad_norm": 0.7491819353441986,
      "learning_rate": 8.311844373985266e-06,
      "loss": 0.4853,
      "step": 4653
    },
    {
      "epoch": 0.2915172489390689,
      "grad_norm": 0.7732419158434022,
      "learning_rate": 8.311084340808996e-06,
      "loss": 0.4007,
      "step": 4654
    },
    {
      "epoch": 0.2915798869384112,
      "grad_norm": 0.8060414413580401,
      "learning_rate": 8.310324171347877e-06,
      "loss": 0.4361,
      "step": 4655
    },
    {
      "epoch": 0.2916425249377535,
      "grad_norm": 0.8482817240676441,
      "learning_rate": 8.309563865633203e-06,
      "loss": 0.4002,
      "step": 4656
    },
    {
      "epoch": 0.2917051629370958,
      "grad_norm": 0.8658207788068841,
      "learning_rate": 8.308803423696264e-06,
      "loss": 0.4131,
      "step": 4657
    },
    {
      "epoch": 0.2917678009364381,
      "grad_norm": 0.8559303334419806,
      "learning_rate": 8.308042845568362e-06,
      "loss": 0.4423,
      "step": 4658
    },
    {
      "epoch": 0.29183043893578037,
      "grad_norm": 0.8064858303171292,
      "learning_rate": 8.307282131280805e-06,
      "loss": 0.4019,
      "step": 4659
    },
    {
      "epoch": 0.2918930769351227,
      "grad_norm": 0.8423262112215374,
      "learning_rate": 8.3065212808649e-06,
      "loss": 0.4121,
      "step": 4660
    },
    {
      "epoch": 0.291955714934465,
      "grad_norm": 0.8263361933712459,
      "learning_rate": 8.305760294351965e-06,
      "loss": 0.4341,
      "step": 4661
    },
    {
      "epoch": 0.2920183529338073,
      "grad_norm": 0.786307338474322,
      "learning_rate": 8.304999171773323e-06,
      "loss": 0.3891,
      "step": 4662
    },
    {
      "epoch": 0.2920809909331496,
      "grad_norm": 0.8363507872715031,
      "learning_rate": 8.304237913160302e-06,
      "loss": 0.4638,
      "step": 4663
    },
    {
      "epoch": 0.2921436289324919,
      "grad_norm": 0.7653592210398874,
      "learning_rate": 8.303476518544236e-06,
      "loss": 0.4136,
      "step": 4664
    },
    {
      "epoch": 0.2922062669318342,
      "grad_norm": 0.7227407131607244,
      "learning_rate": 8.302714987956462e-06,
      "loss": 0.4743,
      "step": 4665
    },
    {
      "epoch": 0.2922689049311765,
      "grad_norm": 0.8068256291290727,
      "learning_rate": 8.301953321428327e-06,
      "loss": 0.441,
      "step": 4666
    },
    {
      "epoch": 0.2923315429305188,
      "grad_norm": 0.7978319766235816,
      "learning_rate": 8.30119151899118e-06,
      "loss": 0.4235,
      "step": 4667
    },
    {
      "epoch": 0.2923941809298611,
      "grad_norm": 0.8558393692471259,
      "learning_rate": 8.300429580676376e-06,
      "loss": 0.4245,
      "step": 4668
    },
    {
      "epoch": 0.2924568189292034,
      "grad_norm": 0.840825423358506,
      "learning_rate": 8.299667506515281e-06,
      "loss": 0.4431,
      "step": 4669
    },
    {
      "epoch": 0.2925194569285457,
      "grad_norm": 0.8216079466630785,
      "learning_rate": 8.298905296539257e-06,
      "loss": 0.4201,
      "step": 4670
    },
    {
      "epoch": 0.292582094927888,
      "grad_norm": 0.8278957404850623,
      "learning_rate": 8.298142950779678e-06,
      "loss": 0.4061,
      "step": 4671
    },
    {
      "epoch": 0.2926447329272303,
      "grad_norm": 0.8750995530321125,
      "learning_rate": 8.297380469267923e-06,
      "loss": 0.4625,
      "step": 4672
    },
    {
      "epoch": 0.2927073709265726,
      "grad_norm": 0.8240331402054177,
      "learning_rate": 8.296617852035377e-06,
      "loss": 0.4369,
      "step": 4673
    },
    {
      "epoch": 0.2927700089259149,
      "grad_norm": 0.7134190594652738,
      "learning_rate": 8.295855099113427e-06,
      "loss": 0.4962,
      "step": 4674
    },
    {
      "epoch": 0.29283264692525723,
      "grad_norm": 0.8215323597152484,
      "learning_rate": 8.29509221053347e-06,
      "loss": 0.4402,
      "step": 4675
    },
    {
      "epoch": 0.2928952849245995,
      "grad_norm": 0.8279998448240088,
      "learning_rate": 8.294329186326908e-06,
      "loss": 0.4236,
      "step": 4676
    },
    {
      "epoch": 0.2929579229239418,
      "grad_norm": 0.8108895627034814,
      "learning_rate": 8.293566026525143e-06,
      "loss": 0.4094,
      "step": 4677
    },
    {
      "epoch": 0.2930205609232841,
      "grad_norm": 0.9618262861236937,
      "learning_rate": 8.29280273115959e-06,
      "loss": 0.4538,
      "step": 4678
    },
    {
      "epoch": 0.2930831989226264,
      "grad_norm": 0.726429719904748,
      "learning_rate": 8.292039300261667e-06,
      "loss": 0.4698,
      "step": 4679
    },
    {
      "epoch": 0.2931458369219687,
      "grad_norm": 0.9899008468987311,
      "learning_rate": 8.29127573386279e-06,
      "loss": 0.4352,
      "step": 4680
    },
    {
      "epoch": 0.293208474921311,
      "grad_norm": 0.8283517665505474,
      "learning_rate": 8.290512031994397e-06,
      "loss": 0.3612,
      "step": 4681
    },
    {
      "epoch": 0.29327111292065333,
      "grad_norm": 0.8898500891696635,
      "learning_rate": 8.289748194687918e-06,
      "loss": 0.444,
      "step": 4682
    },
    {
      "epoch": 0.29333375091999564,
      "grad_norm": 0.7188795302486999,
      "learning_rate": 8.288984221974793e-06,
      "loss": 0.4861,
      "step": 4683
    },
    {
      "epoch": 0.2933963889193379,
      "grad_norm": 0.8542828278885054,
      "learning_rate": 8.288220113886464e-06,
      "loss": 0.4173,
      "step": 4684
    },
    {
      "epoch": 0.2934590269186802,
      "grad_norm": 0.8840187199370168,
      "learning_rate": 8.28745587045439e-06,
      "loss": 0.4248,
      "step": 4685
    },
    {
      "epoch": 0.2935216649180225,
      "grad_norm": 0.8203772733907794,
      "learning_rate": 8.286691491710017e-06,
      "loss": 0.4017,
      "step": 4686
    },
    {
      "epoch": 0.2935843029173648,
      "grad_norm": 0.9187487962232189,
      "learning_rate": 8.285926977684816e-06,
      "loss": 0.4378,
      "step": 4687
    },
    {
      "epoch": 0.2936469409167071,
      "grad_norm": 0.839259553189397,
      "learning_rate": 8.285162328410248e-06,
      "loss": 0.3888,
      "step": 4688
    },
    {
      "epoch": 0.29370957891604943,
      "grad_norm": 0.759083196443782,
      "learning_rate": 8.284397543917789e-06,
      "loss": 0.3764,
      "step": 4689
    },
    {
      "epoch": 0.29377221691539174,
      "grad_norm": 0.8006321331309094,
      "learning_rate": 8.283632624238917e-06,
      "loss": 0.4146,
      "step": 4690
    },
    {
      "epoch": 0.29383485491473404,
      "grad_norm": 0.8198037996735498,
      "learning_rate": 8.28286756940512e-06,
      "loss": 0.46,
      "step": 4691
    },
    {
      "epoch": 0.2938974929140763,
      "grad_norm": 0.8080281907894534,
      "learning_rate": 8.282102379447882e-06,
      "loss": 0.3773,
      "step": 4692
    },
    {
      "epoch": 0.2939601309134186,
      "grad_norm": 0.8413652170957195,
      "learning_rate": 8.281337054398702e-06,
      "loss": 0.4121,
      "step": 4693
    },
    {
      "epoch": 0.2940227689127609,
      "grad_norm": 0.8164615798272002,
      "learning_rate": 8.280571594289077e-06,
      "loss": 0.4598,
      "step": 4694
    },
    {
      "epoch": 0.2940854069121032,
      "grad_norm": 0.8612456261550063,
      "learning_rate": 8.279805999150519e-06,
      "loss": 0.4074,
      "step": 4695
    },
    {
      "epoch": 0.2941480449114455,
      "grad_norm": 0.7967845340924328,
      "learning_rate": 8.279040269014538e-06,
      "loss": 0.4116,
      "step": 4696
    },
    {
      "epoch": 0.29421068291078784,
      "grad_norm": 0.8227985067808806,
      "learning_rate": 8.278274403912649e-06,
      "loss": 0.3976,
      "step": 4697
    },
    {
      "epoch": 0.29427332091013014,
      "grad_norm": 0.821493857872847,
      "learning_rate": 8.277508403876377e-06,
      "loss": 0.4118,
      "step": 4698
    },
    {
      "epoch": 0.29433595890947245,
      "grad_norm": 0.8898107008529212,
      "learning_rate": 8.276742268937252e-06,
      "loss": 0.4342,
      "step": 4699
    },
    {
      "epoch": 0.29439859690881476,
      "grad_norm": 0.7941023499816265,
      "learning_rate": 8.275975999126807e-06,
      "loss": 0.3992,
      "step": 4700
    },
    {
      "epoch": 0.294461234908157,
      "grad_norm": 0.8298192077333765,
      "learning_rate": 8.27520959447658e-06,
      "loss": 0.4223,
      "step": 4701
    },
    {
      "epoch": 0.2945238729074993,
      "grad_norm": 0.8885386678950024,
      "learning_rate": 8.27444305501812e-06,
      "loss": 0.4351,
      "step": 4702
    },
    {
      "epoch": 0.2945865109068416,
      "grad_norm": 0.8704685783743942,
      "learning_rate": 8.273676380782975e-06,
      "loss": 0.4113,
      "step": 4703
    },
    {
      "epoch": 0.29464914890618393,
      "grad_norm": 0.668126025116446,
      "learning_rate": 8.272909571802703e-06,
      "loss": 0.4876,
      "step": 4704
    },
    {
      "epoch": 0.29471178690552624,
      "grad_norm": 0.7900211101541844,
      "learning_rate": 8.272142628108866e-06,
      "loss": 0.4329,
      "step": 4705
    },
    {
      "epoch": 0.29477442490486855,
      "grad_norm": 0.8831766896045606,
      "learning_rate": 8.271375549733031e-06,
      "loss": 0.3973,
      "step": 4706
    },
    {
      "epoch": 0.29483706290421086,
      "grad_norm": 0.8546121525653743,
      "learning_rate": 8.27060833670677e-06,
      "loss": 0.4013,
      "step": 4707
    },
    {
      "epoch": 0.29489970090355316,
      "grad_norm": 0.7931836788042963,
      "learning_rate": 8.269840989061664e-06,
      "loss": 0.3612,
      "step": 4708
    },
    {
      "epoch": 0.2949623389028954,
      "grad_norm": 0.8195227160827858,
      "learning_rate": 8.269073506829296e-06,
      "loss": 0.4047,
      "step": 4709
    },
    {
      "epoch": 0.2950249769022377,
      "grad_norm": 0.8959139707075572,
      "learning_rate": 8.268305890041256e-06,
      "loss": 0.397,
      "step": 4710
    },
    {
      "epoch": 0.29508761490158003,
      "grad_norm": 0.8863124011573619,
      "learning_rate": 8.267538138729137e-06,
      "loss": 0.4244,
      "step": 4711
    },
    {
      "epoch": 0.29515025290092234,
      "grad_norm": 0.8350412832429028,
      "learning_rate": 8.266770252924543e-06,
      "loss": 0.4138,
      "step": 4712
    },
    {
      "epoch": 0.29521289090026465,
      "grad_norm": 0.8851264126354216,
      "learning_rate": 8.266002232659079e-06,
      "loss": 0.4634,
      "step": 4713
    },
    {
      "epoch": 0.29527552889960695,
      "grad_norm": 0.8060326177736885,
      "learning_rate": 8.265234077964356e-06,
      "loss": 0.4252,
      "step": 4714
    },
    {
      "epoch": 0.29533816689894926,
      "grad_norm": 0.7744147991697258,
      "learning_rate": 8.264465788871994e-06,
      "loss": 0.4155,
      "step": 4715
    },
    {
      "epoch": 0.29540080489829157,
      "grad_norm": 0.9061806952338898,
      "learning_rate": 8.263697365413614e-06,
      "loss": 0.4129,
      "step": 4716
    },
    {
      "epoch": 0.2954634428976339,
      "grad_norm": 0.8279778082601156,
      "learning_rate": 8.262928807620843e-06,
      "loss": 0.4345,
      "step": 4717
    },
    {
      "epoch": 0.29552608089697613,
      "grad_norm": 0.8431018264006284,
      "learning_rate": 8.262160115525319e-06,
      "loss": 0.4666,
      "step": 4718
    },
    {
      "epoch": 0.29558871889631844,
      "grad_norm": 0.8190873177356165,
      "learning_rate": 8.261391289158678e-06,
      "loss": 0.3965,
      "step": 4719
    },
    {
      "epoch": 0.29565135689566074,
      "grad_norm": 0.877895606878929,
      "learning_rate": 8.260622328552566e-06,
      "loss": 0.4329,
      "step": 4720
    },
    {
      "epoch": 0.29571399489500305,
      "grad_norm": 0.9211563845406315,
      "learning_rate": 8.259853233738633e-06,
      "loss": 0.4114,
      "step": 4721
    },
    {
      "epoch": 0.29577663289434536,
      "grad_norm": 0.8423267673283132,
      "learning_rate": 8.259084004748536e-06,
      "loss": 0.3923,
      "step": 4722
    },
    {
      "epoch": 0.29583927089368767,
      "grad_norm": 0.884865234915486,
      "learning_rate": 8.258314641613937e-06,
      "loss": 0.4281,
      "step": 4723
    },
    {
      "epoch": 0.29590190889303,
      "grad_norm": 1.317497980610835,
      "learning_rate": 8.257545144366505e-06,
      "loss": 0.4667,
      "step": 4724
    },
    {
      "epoch": 0.2959645468923723,
      "grad_norm": 0.8565417030561198,
      "learning_rate": 8.256775513037907e-06,
      "loss": 0.4398,
      "step": 4725
    },
    {
      "epoch": 0.29602718489171453,
      "grad_norm": 0.8277713792251994,
      "learning_rate": 8.256005747659825e-06,
      "loss": 0.4547,
      "step": 4726
    },
    {
      "epoch": 0.29608982289105684,
      "grad_norm": 0.7805897015784916,
      "learning_rate": 8.255235848263943e-06,
      "loss": 0.4276,
      "step": 4727
    },
    {
      "epoch": 0.29615246089039915,
      "grad_norm": 0.9434115309574751,
      "learning_rate": 8.25446581488195e-06,
      "loss": 0.4521,
      "step": 4728
    },
    {
      "epoch": 0.29621509888974146,
      "grad_norm": 0.9201587369374226,
      "learning_rate": 8.253695647545538e-06,
      "loss": 0.478,
      "step": 4729
    },
    {
      "epoch": 0.29627773688908376,
      "grad_norm": 0.9334825678237567,
      "learning_rate": 8.252925346286409e-06,
      "loss": 0.4319,
      "step": 4730
    },
    {
      "epoch": 0.2963403748884261,
      "grad_norm": 0.9295124154781305,
      "learning_rate": 8.25215491113627e-06,
      "loss": 0.4574,
      "step": 4731
    },
    {
      "epoch": 0.2964030128877684,
      "grad_norm": 0.8402146740866543,
      "learning_rate": 8.251384342126831e-06,
      "loss": 0.4723,
      "step": 4732
    },
    {
      "epoch": 0.2964656508871107,
      "grad_norm": 0.8348208506812645,
      "learning_rate": 8.25061363928981e-06,
      "loss": 0.3951,
      "step": 4733
    },
    {
      "epoch": 0.29652828888645294,
      "grad_norm": 0.960101951463965,
      "learning_rate": 8.249842802656927e-06,
      "loss": 0.4385,
      "step": 4734
    },
    {
      "epoch": 0.29659092688579525,
      "grad_norm": 0.8260345788532651,
      "learning_rate": 8.249071832259911e-06,
      "loss": 0.4377,
      "step": 4735
    },
    {
      "epoch": 0.29665356488513756,
      "grad_norm": 0.8922425512648903,
      "learning_rate": 8.248300728130496e-06,
      "loss": 0.4364,
      "step": 4736
    },
    {
      "epoch": 0.29671620288447986,
      "grad_norm": 0.8596019573375334,
      "learning_rate": 8.247529490300422e-06,
      "loss": 0.4085,
      "step": 4737
    },
    {
      "epoch": 0.29677884088382217,
      "grad_norm": 0.7952553555021092,
      "learning_rate": 8.246758118801428e-06,
      "loss": 0.3888,
      "step": 4738
    },
    {
      "epoch": 0.2968414788831645,
      "grad_norm": 0.8223329050084007,
      "learning_rate": 8.245986613665269e-06,
      "loss": 0.4044,
      "step": 4739
    },
    {
      "epoch": 0.2969041168825068,
      "grad_norm": 0.7966263161013798,
      "learning_rate": 8.245214974923696e-06,
      "loss": 0.4039,
      "step": 4740
    },
    {
      "epoch": 0.2969667548818491,
      "grad_norm": 0.7962596086901801,
      "learning_rate": 8.244443202608475e-06,
      "loss": 0.4089,
      "step": 4741
    },
    {
      "epoch": 0.2970293928811914,
      "grad_norm": 0.8564701481121144,
      "learning_rate": 8.24367129675137e-06,
      "loss": 0.4538,
      "step": 4742
    },
    {
      "epoch": 0.29709203088053365,
      "grad_norm": 0.8424297684063751,
      "learning_rate": 8.242899257384152e-06,
      "loss": 0.4305,
      "step": 4743
    },
    {
      "epoch": 0.29715466887987596,
      "grad_norm": 0.9158770874651943,
      "learning_rate": 8.242127084538597e-06,
      "loss": 0.4038,
      "step": 4744
    },
    {
      "epoch": 0.29721730687921827,
      "grad_norm": 0.9837219172759751,
      "learning_rate": 8.24135477824649e-06,
      "loss": 0.4259,
      "step": 4745
    },
    {
      "epoch": 0.2972799448785606,
      "grad_norm": 0.9038729849652434,
      "learning_rate": 8.24058233853962e-06,
      "loss": 0.4454,
      "step": 4746
    },
    {
      "epoch": 0.2973425828779029,
      "grad_norm": 0.861771243264598,
      "learning_rate": 8.239809765449778e-06,
      "loss": 0.4369,
      "step": 4747
    },
    {
      "epoch": 0.2974052208772452,
      "grad_norm": 0.836841888223182,
      "learning_rate": 8.239037059008765e-06,
      "loss": 0.3748,
      "step": 4748
    },
    {
      "epoch": 0.2974678588765875,
      "grad_norm": 0.8443352383434148,
      "learning_rate": 8.238264219248385e-06,
      "loss": 0.4183,
      "step": 4749
    },
    {
      "epoch": 0.2975304968759298,
      "grad_norm": 0.8075544259324587,
      "learning_rate": 8.23749124620045e-06,
      "loss": 0.4316,
      "step": 4750
    },
    {
      "epoch": 0.29759313487527206,
      "grad_norm": 0.8022350214931576,
      "learning_rate": 8.236718139896773e-06,
      "loss": 0.3974,
      "step": 4751
    },
    {
      "epoch": 0.29765577287461437,
      "grad_norm": 0.8857233494965696,
      "learning_rate": 8.23594490036918e-06,
      "loss": 0.4083,
      "step": 4752
    },
    {
      "epoch": 0.2977184108739567,
      "grad_norm": 0.8598938966633237,
      "learning_rate": 8.23517152764949e-06,
      "loss": 0.4288,
      "step": 4753
    },
    {
      "epoch": 0.297781048873299,
      "grad_norm": 0.7746531806251533,
      "learning_rate": 8.234398021769541e-06,
      "loss": 0.4111,
      "step": 4754
    },
    {
      "epoch": 0.2978436868726413,
      "grad_norm": 0.7866063498055118,
      "learning_rate": 8.23362438276117e-06,
      "loss": 0.4037,
      "step": 4755
    },
    {
      "epoch": 0.2979063248719836,
      "grad_norm": 0.8279048114811394,
      "learning_rate": 8.232850610656218e-06,
      "loss": 0.3838,
      "step": 4756
    },
    {
      "epoch": 0.2979689628713259,
      "grad_norm": 0.7011076429893535,
      "learning_rate": 8.232076705486537e-06,
      "loss": 0.4776,
      "step": 4757
    },
    {
      "epoch": 0.2980316008706682,
      "grad_norm": 0.8241154150989908,
      "learning_rate": 8.231302667283976e-06,
      "loss": 0.4036,
      "step": 4758
    },
    {
      "epoch": 0.2980942388700105,
      "grad_norm": 0.8011876045915408,
      "learning_rate": 8.230528496080397e-06,
      "loss": 0.4273,
      "step": 4759
    },
    {
      "epoch": 0.29815687686935277,
      "grad_norm": 0.8068255058063365,
      "learning_rate": 8.229754191907669e-06,
      "loss": 0.4532,
      "step": 4760
    },
    {
      "epoch": 0.2982195148686951,
      "grad_norm": 0.897957926655788,
      "learning_rate": 8.228979754797657e-06,
      "loss": 0.4553,
      "step": 4761
    },
    {
      "epoch": 0.2982821528680374,
      "grad_norm": 0.8878247705510075,
      "learning_rate": 8.228205184782239e-06,
      "loss": 0.4251,
      "step": 4762
    },
    {
      "epoch": 0.2983447908673797,
      "grad_norm": 0.8643003654771906,
      "learning_rate": 8.227430481893295e-06,
      "loss": 0.4378,
      "step": 4763
    },
    {
      "epoch": 0.298407428866722,
      "grad_norm": 0.7961872255463938,
      "learning_rate": 8.226655646162715e-06,
      "loss": 0.3989,
      "step": 4764
    },
    {
      "epoch": 0.2984700668660643,
      "grad_norm": 0.8926825493282843,
      "learning_rate": 8.225880677622391e-06,
      "loss": 0.4516,
      "step": 4765
    },
    {
      "epoch": 0.2985327048654066,
      "grad_norm": 0.8348825471494002,
      "learning_rate": 8.225105576304218e-06,
      "loss": 0.4176,
      "step": 4766
    },
    {
      "epoch": 0.2985953428647489,
      "grad_norm": 0.6783020631375238,
      "learning_rate": 8.224330342240099e-06,
      "loss": 0.4703,
      "step": 4767
    },
    {
      "epoch": 0.2986579808640912,
      "grad_norm": 0.85722243813747,
      "learning_rate": 8.223554975461947e-06,
      "loss": 0.4451,
      "step": 4768
    },
    {
      "epoch": 0.2987206188634335,
      "grad_norm": 0.8016467662536704,
      "learning_rate": 8.222779476001674e-06,
      "loss": 0.3919,
      "step": 4769
    },
    {
      "epoch": 0.2987832568627758,
      "grad_norm": 0.7157037419351648,
      "learning_rate": 8.2220038438912e-06,
      "loss": 0.4786,
      "step": 4770
    },
    {
      "epoch": 0.2988458948621181,
      "grad_norm": 0.8568678178649315,
      "learning_rate": 8.221228079162449e-06,
      "loss": 0.4236,
      "step": 4771
    },
    {
      "epoch": 0.2989085328614604,
      "grad_norm": 0.8316278926566025,
      "learning_rate": 8.220452181847351e-06,
      "loss": 0.3893,
      "step": 4772
    },
    {
      "epoch": 0.2989711708608027,
      "grad_norm": 0.9010324763356219,
      "learning_rate": 8.219676151977844e-06,
      "loss": 0.4624,
      "step": 4773
    },
    {
      "epoch": 0.299033808860145,
      "grad_norm": 0.7263168173523297,
      "learning_rate": 8.21889998958587e-06,
      "loss": 0.3271,
      "step": 4774
    },
    {
      "epoch": 0.29909644685948733,
      "grad_norm": 0.8348840949641461,
      "learning_rate": 8.218123694703375e-06,
      "loss": 0.3989,
      "step": 4775
    },
    {
      "epoch": 0.2991590848588296,
      "grad_norm": 0.7913390365302424,
      "learning_rate": 8.217347267362311e-06,
      "loss": 0.4196,
      "step": 4776
    },
    {
      "epoch": 0.2992217228581719,
      "grad_norm": 0.8941159651022715,
      "learning_rate": 8.216570707594636e-06,
      "loss": 0.391,
      "step": 4777
    },
    {
      "epoch": 0.2992843608575142,
      "grad_norm": 0.7520331704664364,
      "learning_rate": 8.215794015432314e-06,
      "loss": 0.379,
      "step": 4778
    },
    {
      "epoch": 0.2993469988568565,
      "grad_norm": 0.7977596792448313,
      "learning_rate": 8.215017190907315e-06,
      "loss": 0.4163,
      "step": 4779
    },
    {
      "epoch": 0.2994096368561988,
      "grad_norm": 0.8055332701795884,
      "learning_rate": 8.21424023405161e-06,
      "loss": 0.441,
      "step": 4780
    },
    {
      "epoch": 0.2994722748555411,
      "grad_norm": 0.84292347134879,
      "learning_rate": 8.213463144897179e-06,
      "loss": 0.4261,
      "step": 4781
    },
    {
      "epoch": 0.29953491285488343,
      "grad_norm": 0.8118064690916408,
      "learning_rate": 8.21268592347601e-06,
      "loss": 0.4395,
      "step": 4782
    },
    {
      "epoch": 0.29959755085422574,
      "grad_norm": 0.8349935270719717,
      "learning_rate": 8.211908569820092e-06,
      "loss": 0.4426,
      "step": 4783
    },
    {
      "epoch": 0.29966018885356804,
      "grad_norm": 0.7953420036288309,
      "learning_rate": 8.21113108396142e-06,
      "loss": 0.4039,
      "step": 4784
    },
    {
      "epoch": 0.2997228268529103,
      "grad_norm": 0.856848167880259,
      "learning_rate": 8.210353465931999e-06,
      "loss": 0.3954,
      "step": 4785
    },
    {
      "epoch": 0.2997854648522526,
      "grad_norm": 0.8479212700252556,
      "learning_rate": 8.209575715763832e-06,
      "loss": 0.4051,
      "step": 4786
    },
    {
      "epoch": 0.2998481028515949,
      "grad_norm": 0.7726235869591852,
      "learning_rate": 8.208797833488933e-06,
      "loss": 0.4346,
      "step": 4787
    },
    {
      "epoch": 0.2999107408509372,
      "grad_norm": 0.7721455658618833,
      "learning_rate": 8.20801981913932e-06,
      "loss": 0.4236,
      "step": 4788
    },
    {
      "epoch": 0.2999733788502795,
      "grad_norm": 0.87365066102569,
      "learning_rate": 8.207241672747015e-06,
      "loss": 0.4557,
      "step": 4789
    },
    {
      "epoch": 0.30003601684962183,
      "grad_norm": 0.8643389791368363,
      "learning_rate": 8.206463394344048e-06,
      "loss": 0.4295,
      "step": 4790
    },
    {
      "epoch": 0.30009865484896414,
      "grad_norm": 0.8401956980417803,
      "learning_rate": 8.205684983962453e-06,
      "loss": 0.4248,
      "step": 4791
    },
    {
      "epoch": 0.30016129284830645,
      "grad_norm": 0.8936212497634842,
      "learning_rate": 8.204906441634268e-06,
      "loss": 0.443,
      "step": 4792
    },
    {
      "epoch": 0.3002239308476487,
      "grad_norm": 0.7876317412949028,
      "learning_rate": 8.20412776739154e-06,
      "loss": 0.357,
      "step": 4793
    },
    {
      "epoch": 0.300286568846991,
      "grad_norm": 1.0216568571415405,
      "learning_rate": 8.203348961266317e-06,
      "loss": 0.4806,
      "step": 4794
    },
    {
      "epoch": 0.3003492068463333,
      "grad_norm": 0.8390659766020679,
      "learning_rate": 8.202570023290658e-06,
      "loss": 0.4167,
      "step": 4795
    },
    {
      "epoch": 0.3004118448456756,
      "grad_norm": 0.8518660639480424,
      "learning_rate": 8.201790953496621e-06,
      "loss": 0.4023,
      "step": 4796
    },
    {
      "epoch": 0.30047448284501793,
      "grad_norm": 0.8859856333132752,
      "learning_rate": 8.201011751916275e-06,
      "loss": 0.4453,
      "step": 4797
    },
    {
      "epoch": 0.30053712084436024,
      "grad_norm": 0.8927487391667446,
      "learning_rate": 8.200232418581691e-06,
      "loss": 0.4482,
      "step": 4798
    },
    {
      "epoch": 0.30059975884370255,
      "grad_norm": 0.7799457866327495,
      "learning_rate": 8.199452953524947e-06,
      "loss": 0.392,
      "step": 4799
    },
    {
      "epoch": 0.30066239684304485,
      "grad_norm": 0.8510166274100202,
      "learning_rate": 8.198673356778124e-06,
      "loss": 0.45,
      "step": 4800
    },
    {
      "epoch": 0.3007250348423871,
      "grad_norm": 0.7975167351032469,
      "learning_rate": 8.197893628373312e-06,
      "loss": 0.4562,
      "step": 4801
    },
    {
      "epoch": 0.3007876728417294,
      "grad_norm": 0.8314077210308375,
      "learning_rate": 8.197113768342606e-06,
      "loss": 0.3685,
      "step": 4802
    },
    {
      "epoch": 0.3008503108410717,
      "grad_norm": 0.9132547792182526,
      "learning_rate": 8.196333776718105e-06,
      "loss": 0.4343,
      "step": 4803
    },
    {
      "epoch": 0.30091294884041403,
      "grad_norm": 0.7858181038325729,
      "learning_rate": 8.195553653531909e-06,
      "loss": 0.383,
      "step": 4804
    },
    {
      "epoch": 0.30097558683975634,
      "grad_norm": 0.8534051271985902,
      "learning_rate": 8.194773398816133e-06,
      "loss": 0.4076,
      "step": 4805
    },
    {
      "epoch": 0.30103822483909864,
      "grad_norm": 0.825207466406068,
      "learning_rate": 8.193993012602894e-06,
      "loss": 0.4756,
      "step": 4806
    },
    {
      "epoch": 0.30110086283844095,
      "grad_norm": 0.7792227649305923,
      "learning_rate": 8.193212494924306e-06,
      "loss": 0.4089,
      "step": 4807
    },
    {
      "epoch": 0.30116350083778326,
      "grad_norm": 0.8937861787419573,
      "learning_rate": 8.192431845812501e-06,
      "loss": 0.4254,
      "step": 4808
    },
    {
      "epoch": 0.30122613883712557,
      "grad_norm": 0.8487317770610026,
      "learning_rate": 8.191651065299605e-06,
      "loss": 0.4562,
      "step": 4809
    },
    {
      "epoch": 0.3012887768364678,
      "grad_norm": 0.7856147574281219,
      "learning_rate": 8.190870153417765e-06,
      "loss": 0.3994,
      "step": 4810
    },
    {
      "epoch": 0.3013514148358101,
      "grad_norm": 0.7950153699465518,
      "learning_rate": 8.190089110199113e-06,
      "loss": 0.4002,
      "step": 4811
    },
    {
      "epoch": 0.30141405283515244,
      "grad_norm": 0.8441467012037421,
      "learning_rate": 8.189307935675801e-06,
      "loss": 0.4185,
      "step": 4812
    },
    {
      "epoch": 0.30147669083449474,
      "grad_norm": 0.8451710383126633,
      "learning_rate": 8.188526629879982e-06,
      "loss": 0.3996,
      "step": 4813
    },
    {
      "epoch": 0.30153932883383705,
      "grad_norm": 0.745399398799224,
      "learning_rate": 8.187745192843818e-06,
      "loss": 0.4961,
      "step": 4814
    },
    {
      "epoch": 0.30160196683317936,
      "grad_norm": 0.7753220686914618,
      "learning_rate": 8.186963624599468e-06,
      "loss": 0.3789,
      "step": 4815
    },
    {
      "epoch": 0.30166460483252167,
      "grad_norm": 0.8681661878056526,
      "learning_rate": 8.186181925179105e-06,
      "loss": 0.3959,
      "step": 4816
    },
    {
      "epoch": 0.301727242831864,
      "grad_norm": 0.7718721939163875,
      "learning_rate": 8.185400094614901e-06,
      "loss": 0.3996,
      "step": 4817
    },
    {
      "epoch": 0.3017898808312062,
      "grad_norm": 0.8194811569017296,
      "learning_rate": 8.18461813293904e-06,
      "loss": 0.4035,
      "step": 4818
    },
    {
      "epoch": 0.30185251883054853,
      "grad_norm": 0.8156483990774613,
      "learning_rate": 8.183836040183703e-06,
      "loss": 0.4286,
      "step": 4819
    },
    {
      "epoch": 0.30191515682989084,
      "grad_norm": 0.8469462115984194,
      "learning_rate": 8.183053816381086e-06,
      "loss": 0.4471,
      "step": 4820
    },
    {
      "epoch": 0.30197779482923315,
      "grad_norm": 0.8104533135128699,
      "learning_rate": 8.182271461563383e-06,
      "loss": 0.394,
      "step": 4821
    },
    {
      "epoch": 0.30204043282857546,
      "grad_norm": 0.8372552872857792,
      "learning_rate": 8.181488975762795e-06,
      "loss": 0.4344,
      "step": 4822
    },
    {
      "epoch": 0.30210307082791776,
      "grad_norm": 0.9129281988546251,
      "learning_rate": 8.180706359011529e-06,
      "loss": 0.4263,
      "step": 4823
    },
    {
      "epoch": 0.30216570882726007,
      "grad_norm": 0.828267372060393,
      "learning_rate": 8.1799236113418e-06,
      "loss": 0.398,
      "step": 4824
    },
    {
      "epoch": 0.3022283468266024,
      "grad_norm": 0.8948491511765189,
      "learning_rate": 8.179140732785826e-06,
      "loss": 0.4267,
      "step": 4825
    },
    {
      "epoch": 0.3022909848259447,
      "grad_norm": 0.8219026891906757,
      "learning_rate": 8.178357723375829e-06,
      "loss": 0.4536,
      "step": 4826
    },
    {
      "epoch": 0.30235362282528694,
      "grad_norm": 0.8223538629737535,
      "learning_rate": 8.177574583144037e-06,
      "loss": 0.4006,
      "step": 4827
    },
    {
      "epoch": 0.30241626082462925,
      "grad_norm": 0.8510620946183169,
      "learning_rate": 8.176791312122686e-06,
      "loss": 0.4455,
      "step": 4828
    },
    {
      "epoch": 0.30247889882397155,
      "grad_norm": 0.7100935691511454,
      "learning_rate": 8.176007910344015e-06,
      "loss": 0.4765,
      "step": 4829
    },
    {
      "epoch": 0.30254153682331386,
      "grad_norm": 0.8807525612488563,
      "learning_rate": 8.17522437784027e-06,
      "loss": 0.4079,
      "step": 4830
    },
    {
      "epoch": 0.30260417482265617,
      "grad_norm": 0.8245686686873493,
      "learning_rate": 8.174440714643699e-06,
      "loss": 0.4248,
      "step": 4831
    },
    {
      "epoch": 0.3026668128219985,
      "grad_norm": 0.8195109045239336,
      "learning_rate": 8.17365692078656e-06,
      "loss": 0.4321,
      "step": 4832
    },
    {
      "epoch": 0.3027294508213408,
      "grad_norm": 0.808952013876439,
      "learning_rate": 8.172872996301111e-06,
      "loss": 0.4241,
      "step": 4833
    },
    {
      "epoch": 0.3027920888206831,
      "grad_norm": 0.7878016177245367,
      "learning_rate": 8.172088941219621e-06,
      "loss": 0.4043,
      "step": 4834
    },
    {
      "epoch": 0.30285472682002534,
      "grad_norm": 0.8413070000413724,
      "learning_rate": 8.171304755574363e-06,
      "loss": 0.4397,
      "step": 4835
    },
    {
      "epoch": 0.30291736481936765,
      "grad_norm": 0.7574825886367877,
      "learning_rate": 8.170520439397612e-06,
      "loss": 0.3745,
      "step": 4836
    },
    {
      "epoch": 0.30298000281870996,
      "grad_norm": 0.8405075311521469,
      "learning_rate": 8.16973599272165e-06,
      "loss": 0.426,
      "step": 4837
    },
    {
      "epoch": 0.30304264081805227,
      "grad_norm": 0.803868364849622,
      "learning_rate": 8.168951415578767e-06,
      "loss": 0.4511,
      "step": 4838
    },
    {
      "epoch": 0.3031052788173946,
      "grad_norm": 0.8257762530645486,
      "learning_rate": 8.168166708001257e-06,
      "loss": 0.4017,
      "step": 4839
    },
    {
      "epoch": 0.3031679168167369,
      "grad_norm": 0.8657759361181306,
      "learning_rate": 8.167381870021415e-06,
      "loss": 0.4333,
      "step": 4840
    },
    {
      "epoch": 0.3032305548160792,
      "grad_norm": 0.8166982430632441,
      "learning_rate": 8.16659690167155e-06,
      "loss": 0.4048,
      "step": 4841
    },
    {
      "epoch": 0.3032931928154215,
      "grad_norm": 0.8316262975743812,
      "learning_rate": 8.165811802983967e-06,
      "loss": 0.4175,
      "step": 4842
    },
    {
      "epoch": 0.30335583081476375,
      "grad_norm": 0.8154739324475015,
      "learning_rate": 8.165026573990982e-06,
      "loss": 0.4193,
      "step": 4843
    },
    {
      "epoch": 0.30341846881410606,
      "grad_norm": 0.8355682516152629,
      "learning_rate": 8.164241214724917e-06,
      "loss": 0.4062,
      "step": 4844
    },
    {
      "epoch": 0.30348110681344836,
      "grad_norm": 0.8314498950611395,
      "learning_rate": 8.163455725218098e-06,
      "loss": 0.4199,
      "step": 4845
    },
    {
      "epoch": 0.3035437448127907,
      "grad_norm": 0.8835295189671241,
      "learning_rate": 8.16267010550285e-06,
      "loss": 0.4255,
      "step": 4846
    },
    {
      "epoch": 0.303606382812133,
      "grad_norm": 0.8282337397008477,
      "learning_rate": 8.161884355611516e-06,
      "loss": 0.4221,
      "step": 4847
    },
    {
      "epoch": 0.3036690208114753,
      "grad_norm": 0.8070132383959294,
      "learning_rate": 8.161098475576435e-06,
      "loss": 0.4179,
      "step": 4848
    },
    {
      "epoch": 0.3037316588108176,
      "grad_norm": 0.8647898691389194,
      "learning_rate": 8.160312465429952e-06,
      "loss": 0.4578,
      "step": 4849
    },
    {
      "epoch": 0.3037942968101599,
      "grad_norm": 0.8415725175066151,
      "learning_rate": 8.159526325204424e-06,
      "loss": 0.4458,
      "step": 4850
    },
    {
      "epoch": 0.3038569348095022,
      "grad_norm": 0.8020430368726048,
      "learning_rate": 8.158740054932204e-06,
      "loss": 0.4447,
      "step": 4851
    },
    {
      "epoch": 0.30391957280884446,
      "grad_norm": 0.6846838471925597,
      "learning_rate": 8.157953654645658e-06,
      "loss": 0.4892,
      "step": 4852
    },
    {
      "epoch": 0.30398221080818677,
      "grad_norm": 0.8128116596439973,
      "learning_rate": 8.157167124377153e-06,
      "loss": 0.4132,
      "step": 4853
    },
    {
      "epoch": 0.3040448488075291,
      "grad_norm": 0.8792337177361875,
      "learning_rate": 8.156380464159063e-06,
      "loss": 0.4349,
      "step": 4854
    },
    {
      "epoch": 0.3041074868068714,
      "grad_norm": 0.8579506875745047,
      "learning_rate": 8.155593674023769e-06,
      "loss": 0.419,
      "step": 4855
    },
    {
      "epoch": 0.3041701248062137,
      "grad_norm": 0.8438261302993842,
      "learning_rate": 8.154806754003652e-06,
      "loss": 0.4632,
      "step": 4856
    },
    {
      "epoch": 0.304232762805556,
      "grad_norm": 0.9034882829501002,
      "learning_rate": 8.154019704131103e-06,
      "loss": 0.3934,
      "step": 4857
    },
    {
      "epoch": 0.3042954008048983,
      "grad_norm": 0.8704857461514026,
      "learning_rate": 8.153232524438519e-06,
      "loss": 0.4822,
      "step": 4858
    },
    {
      "epoch": 0.3043580388042406,
      "grad_norm": 0.7644034748880972,
      "learning_rate": 8.152445214958298e-06,
      "loss": 0.4,
      "step": 4859
    },
    {
      "epoch": 0.30442067680358287,
      "grad_norm": 0.8025965186801952,
      "learning_rate": 8.151657775722847e-06,
      "loss": 0.4226,
      "step": 4860
    },
    {
      "epoch": 0.3044833148029252,
      "grad_norm": 0.8522059749343252,
      "learning_rate": 8.15087020676458e-06,
      "loss": 0.4036,
      "step": 4861
    },
    {
      "epoch": 0.3045459528022675,
      "grad_norm": 0.8282562800850097,
      "learning_rate": 8.150082508115906e-06,
      "loss": 0.4091,
      "step": 4862
    },
    {
      "epoch": 0.3046085908016098,
      "grad_norm": 0.8247350685404883,
      "learning_rate": 8.149294679809254e-06,
      "loss": 0.4167,
      "step": 4863
    },
    {
      "epoch": 0.3046712288009521,
      "grad_norm": 0.8465407921276551,
      "learning_rate": 8.148506721877048e-06,
      "loss": 0.4184,
      "step": 4864
    },
    {
      "epoch": 0.3047338668002944,
      "grad_norm": 0.8501785813551659,
      "learning_rate": 8.14771863435172e-06,
      "loss": 0.3791,
      "step": 4865
    },
    {
      "epoch": 0.3047965047996367,
      "grad_norm": 0.737686495391819,
      "learning_rate": 8.146930417265708e-06,
      "loss": 0.4845,
      "step": 4866
    },
    {
      "epoch": 0.304859142798979,
      "grad_norm": 0.811910324233938,
      "learning_rate": 8.146142070651458e-06,
      "loss": 0.3944,
      "step": 4867
    },
    {
      "epoch": 0.30492178079832133,
      "grad_norm": 0.8205297037800524,
      "learning_rate": 8.145353594541416e-06,
      "loss": 0.4239,
      "step": 4868
    },
    {
      "epoch": 0.3049844187976636,
      "grad_norm": 0.827749799655784,
      "learning_rate": 8.144564988968035e-06,
      "loss": 0.4057,
      "step": 4869
    },
    {
      "epoch": 0.3050470567970059,
      "grad_norm": 0.8208243295446684,
      "learning_rate": 8.143776253963777e-06,
      "loss": 0.4332,
      "step": 4870
    },
    {
      "epoch": 0.3051096947963482,
      "grad_norm": 0.8047862181221954,
      "learning_rate": 8.142987389561104e-06,
      "loss": 0.425,
      "step": 4871
    },
    {
      "epoch": 0.3051723327956905,
      "grad_norm": 0.8421713949520846,
      "learning_rate": 8.142198395792487e-06,
      "loss": 0.4429,
      "step": 4872
    },
    {
      "epoch": 0.3052349707950328,
      "grad_norm": 0.8641319904787239,
      "learning_rate": 8.1414092726904e-06,
      "loss": 0.4496,
      "step": 4873
    },
    {
      "epoch": 0.3052976087943751,
      "grad_norm": 0.8615982798893081,
      "learning_rate": 8.140620020287324e-06,
      "loss": 0.4401,
      "step": 4874
    },
    {
      "epoch": 0.3053602467937174,
      "grad_norm": 0.8720021815550819,
      "learning_rate": 8.139830638615747e-06,
      "loss": 0.4183,
      "step": 4875
    },
    {
      "epoch": 0.30542288479305973,
      "grad_norm": 0.8949589539902064,
      "learning_rate": 8.139041127708155e-06,
      "loss": 0.4192,
      "step": 4876
    },
    {
      "epoch": 0.305485522792402,
      "grad_norm": 0.8153068451611218,
      "learning_rate": 8.138251487597051e-06,
      "loss": 0.3912,
      "step": 4877
    },
    {
      "epoch": 0.3055481607917443,
      "grad_norm": 0.7699242223670766,
      "learning_rate": 8.137461718314932e-06,
      "loss": 0.3932,
      "step": 4878
    },
    {
      "epoch": 0.3056107987910866,
      "grad_norm": 0.8676352900325563,
      "learning_rate": 8.136671819894307e-06,
      "loss": 0.4376,
      "step": 4879
    },
    {
      "epoch": 0.3056734367904289,
      "grad_norm": 0.8499862727696043,
      "learning_rate": 8.135881792367686e-06,
      "loss": 0.4082,
      "step": 4880
    },
    {
      "epoch": 0.3057360747897712,
      "grad_norm": 0.8511750615593016,
      "learning_rate": 8.135091635767589e-06,
      "loss": 0.4266,
      "step": 4881
    },
    {
      "epoch": 0.3057987127891135,
      "grad_norm": 0.834779852432792,
      "learning_rate": 8.13430135012654e-06,
      "loss": 0.4102,
      "step": 4882
    },
    {
      "epoch": 0.30586135078845583,
      "grad_norm": 0.8437414074147557,
      "learning_rate": 8.133510935477063e-06,
      "loss": 0.4157,
      "step": 4883
    },
    {
      "epoch": 0.30592398878779814,
      "grad_norm": 0.7739026625781004,
      "learning_rate": 8.132720391851696e-06,
      "loss": 0.3809,
      "step": 4884
    },
    {
      "epoch": 0.3059866267871404,
      "grad_norm": 0.982987870814428,
      "learning_rate": 8.131929719282974e-06,
      "loss": 0.4,
      "step": 4885
    },
    {
      "epoch": 0.3060492647864827,
      "grad_norm": 0.8169195918062997,
      "learning_rate": 8.131138917803448e-06,
      "loss": 0.4141,
      "step": 4886
    },
    {
      "epoch": 0.306111902785825,
      "grad_norm": 0.8224316188866233,
      "learning_rate": 8.13034798744566e-06,
      "loss": 0.4304,
      "step": 4887
    },
    {
      "epoch": 0.3061745407851673,
      "grad_norm": 0.8563683679183645,
      "learning_rate": 8.129556928242166e-06,
      "loss": 0.4193,
      "step": 4888
    },
    {
      "epoch": 0.3062371787845096,
      "grad_norm": 0.8223657513399143,
      "learning_rate": 8.12876574022553e-06,
      "loss": 0.4299,
      "step": 4889
    },
    {
      "epoch": 0.30629981678385193,
      "grad_norm": 0.8559318147003453,
      "learning_rate": 8.127974423428316e-06,
      "loss": 0.4237,
      "step": 4890
    },
    {
      "epoch": 0.30636245478319424,
      "grad_norm": 0.9492392729698913,
      "learning_rate": 8.127182977883095e-06,
      "loss": 0.441,
      "step": 4891
    },
    {
      "epoch": 0.30642509278253655,
      "grad_norm": 0.7891187433988368,
      "learning_rate": 8.126391403622439e-06,
      "loss": 0.3831,
      "step": 4892
    },
    {
      "epoch": 0.30648773078187885,
      "grad_norm": 0.8613166224224634,
      "learning_rate": 8.125599700678935e-06,
      "loss": 0.4213,
      "step": 4893
    },
    {
      "epoch": 0.3065503687812211,
      "grad_norm": 0.8964463814273704,
      "learning_rate": 8.124807869085167e-06,
      "loss": 0.4507,
      "step": 4894
    },
    {
      "epoch": 0.3066130067805634,
      "grad_norm": 0.8586646557267611,
      "learning_rate": 8.124015908873727e-06,
      "loss": 0.4176,
      "step": 4895
    },
    {
      "epoch": 0.3066756447799057,
      "grad_norm": 0.9011308420791188,
      "learning_rate": 8.123223820077211e-06,
      "loss": 0.4625,
      "step": 4896
    },
    {
      "epoch": 0.30673828277924803,
      "grad_norm": 0.8624691662705563,
      "learning_rate": 8.122431602728222e-06,
      "loss": 0.4475,
      "step": 4897
    },
    {
      "epoch": 0.30680092077859034,
      "grad_norm": 0.8770930044909918,
      "learning_rate": 8.121639256859372e-06,
      "loss": 0.4576,
      "step": 4898
    },
    {
      "epoch": 0.30686355877793264,
      "grad_norm": 0.7349527806781032,
      "learning_rate": 8.12084678250327e-06,
      "loss": 0.4945,
      "step": 4899
    },
    {
      "epoch": 0.30692619677727495,
      "grad_norm": 0.8109728363454414,
      "learning_rate": 8.120054179692534e-06,
      "loss": 0.4286,
      "step": 4900
    },
    {
      "epoch": 0.30698883477661726,
      "grad_norm": 0.8341864809412843,
      "learning_rate": 8.119261448459788e-06,
      "loss": 0.4364,
      "step": 4901
    },
    {
      "epoch": 0.3070514727759595,
      "grad_norm": 0.812603069417161,
      "learning_rate": 8.118468588837663e-06,
      "loss": 0.4364,
      "step": 4902
    },
    {
      "epoch": 0.3071141107753018,
      "grad_norm": 0.8722771493661652,
      "learning_rate": 8.117675600858793e-06,
      "loss": 0.4474,
      "step": 4903
    },
    {
      "epoch": 0.3071767487746441,
      "grad_norm": 0.8086919310020935,
      "learning_rate": 8.116882484555815e-06,
      "loss": 0.445,
      "step": 4904
    },
    {
      "epoch": 0.30723938677398643,
      "grad_norm": 0.9045613745945024,
      "learning_rate": 8.116089239961376e-06,
      "loss": 0.441,
      "step": 4905
    },
    {
      "epoch": 0.30730202477332874,
      "grad_norm": 0.8209165952825479,
      "learning_rate": 8.115295867108126e-06,
      "loss": 0.4255,
      "step": 4906
    },
    {
      "epoch": 0.30736466277267105,
      "grad_norm": 0.8109696389609491,
      "learning_rate": 8.11450236602872e-06,
      "loss": 0.4515,
      "step": 4907
    },
    {
      "epoch": 0.30742730077201336,
      "grad_norm": 0.7825493640798826,
      "learning_rate": 8.113708736755819e-06,
      "loss": 0.4153,
      "step": 4908
    },
    {
      "epoch": 0.30748993877135566,
      "grad_norm": 0.7837181072822184,
      "learning_rate": 8.112914979322087e-06,
      "loss": 0.4137,
      "step": 4909
    },
    {
      "epoch": 0.3075525767706979,
      "grad_norm": 0.8446351270227476,
      "learning_rate": 8.112121093760197e-06,
      "loss": 0.3823,
      "step": 4910
    },
    {
      "epoch": 0.3076152147700402,
      "grad_norm": 0.8244494445234959,
      "learning_rate": 8.111327080102824e-06,
      "loss": 0.4163,
      "step": 4911
    },
    {
      "epoch": 0.30767785276938253,
      "grad_norm": 0.8248440900260504,
      "learning_rate": 8.110532938382654e-06,
      "loss": 0.3989,
      "step": 4912
    },
    {
      "epoch": 0.30774049076872484,
      "grad_norm": 0.8822043230474996,
      "learning_rate": 8.109738668632369e-06,
      "loss": 0.4179,
      "step": 4913
    },
    {
      "epoch": 0.30780312876806715,
      "grad_norm": 0.7754561758218487,
      "learning_rate": 8.108944270884665e-06,
      "loss": 0.4074,
      "step": 4914
    },
    {
      "epoch": 0.30786576676740945,
      "grad_norm": 0.8478004413051494,
      "learning_rate": 8.108149745172233e-06,
      "loss": 0.438,
      "step": 4915
    },
    {
      "epoch": 0.30792840476675176,
      "grad_norm": 0.882776108166062,
      "learning_rate": 8.107355091527785e-06,
      "loss": 0.4652,
      "step": 4916
    },
    {
      "epoch": 0.30799104276609407,
      "grad_norm": 0.8134648813030242,
      "learning_rate": 8.106560309984025e-06,
      "loss": 0.4189,
      "step": 4917
    },
    {
      "epoch": 0.3080536807654364,
      "grad_norm": 0.8286704318742771,
      "learning_rate": 8.105765400573663e-06,
      "loss": 0.4292,
      "step": 4918
    },
    {
      "epoch": 0.30811631876477863,
      "grad_norm": 0.8631943854325177,
      "learning_rate": 8.104970363329422e-06,
      "loss": 0.4409,
      "step": 4919
    },
    {
      "epoch": 0.30817895676412094,
      "grad_norm": 0.8065374803593921,
      "learning_rate": 8.104175198284024e-06,
      "loss": 0.4027,
      "step": 4920
    },
    {
      "epoch": 0.30824159476346324,
      "grad_norm": 0.9359900005169771,
      "learning_rate": 8.1033799054702e-06,
      "loss": 0.4657,
      "step": 4921
    },
    {
      "epoch": 0.30830423276280555,
      "grad_norm": 0.6495660463836165,
      "learning_rate": 8.102584484920683e-06,
      "loss": 0.468,
      "step": 4922
    },
    {
      "epoch": 0.30836687076214786,
      "grad_norm": 0.8296938629969983,
      "learning_rate": 8.101788936668213e-06,
      "loss": 0.4572,
      "step": 4923
    },
    {
      "epoch": 0.30842950876149017,
      "grad_norm": 0.87624273888366,
      "learning_rate": 8.100993260745532e-06,
      "loss": 0.3662,
      "step": 4924
    },
    {
      "epoch": 0.3084921467608325,
      "grad_norm": 0.9470618530481331,
      "learning_rate": 8.100197457185396e-06,
      "loss": 0.4159,
      "step": 4925
    },
    {
      "epoch": 0.3085547847601748,
      "grad_norm": 0.7171048436576927,
      "learning_rate": 8.099401526020556e-06,
      "loss": 0.4846,
      "step": 4926
    },
    {
      "epoch": 0.30861742275951703,
      "grad_norm": 0.7859598921002718,
      "learning_rate": 8.098605467283773e-06,
      "loss": 0.4232,
      "step": 4927
    },
    {
      "epoch": 0.30868006075885934,
      "grad_norm": 0.9226309554378108,
      "learning_rate": 8.097809281007814e-06,
      "loss": 0.4413,
      "step": 4928
    },
    {
      "epoch": 0.30874269875820165,
      "grad_norm": 0.8388959154788819,
      "learning_rate": 8.09701296722545e-06,
      "loss": 0.3902,
      "step": 4929
    },
    {
      "epoch": 0.30880533675754396,
      "grad_norm": 0.78888245830845,
      "learning_rate": 8.09621652596946e-06,
      "loss": 0.4237,
      "step": 4930
    },
    {
      "epoch": 0.30886797475688627,
      "grad_norm": 0.8536990399530642,
      "learning_rate": 8.095419957272619e-06,
      "loss": 0.4184,
      "step": 4931
    },
    {
      "epoch": 0.3089306127562286,
      "grad_norm": 0.8445106759290737,
      "learning_rate": 8.094623261167722e-06,
      "loss": 0.3845,
      "step": 4932
    },
    {
      "epoch": 0.3089932507555709,
      "grad_norm": 0.8771987619248268,
      "learning_rate": 8.093826437687555e-06,
      "loss": 0.4263,
      "step": 4933
    },
    {
      "epoch": 0.3090558887549132,
      "grad_norm": 0.835366935443226,
      "learning_rate": 8.093029486864917e-06,
      "loss": 0.4285,
      "step": 4934
    },
    {
      "epoch": 0.3091185267542555,
      "grad_norm": 0.8138177576936849,
      "learning_rate": 8.092232408732612e-06,
      "loss": 0.4434,
      "step": 4935
    },
    {
      "epoch": 0.30918116475359775,
      "grad_norm": 0.828039562562437,
      "learning_rate": 8.091435203323448e-06,
      "loss": 0.4138,
      "step": 4936
    },
    {
      "epoch": 0.30924380275294006,
      "grad_norm": 0.7974810682386841,
      "learning_rate": 8.090637870670236e-06,
      "loss": 0.4076,
      "step": 4937
    },
    {
      "epoch": 0.30930644075228236,
      "grad_norm": 0.8119097884084192,
      "learning_rate": 8.089840410805797e-06,
      "loss": 0.418,
      "step": 4938
    },
    {
      "epoch": 0.30936907875162467,
      "grad_norm": 0.7750906123057526,
      "learning_rate": 8.089042823762951e-06,
      "loss": 0.431,
      "step": 4939
    },
    {
      "epoch": 0.309431716750967,
      "grad_norm": 0.8509699959805482,
      "learning_rate": 8.088245109574531e-06,
      "loss": 0.3907,
      "step": 4940
    },
    {
      "epoch": 0.3094943547503093,
      "grad_norm": 0.8628915253878398,
      "learning_rate": 8.087447268273368e-06,
      "loss": 0.449,
      "step": 4941
    },
    {
      "epoch": 0.3095569927496516,
      "grad_norm": 0.813293333033416,
      "learning_rate": 8.086649299892303e-06,
      "loss": 0.4343,
      "step": 4942
    },
    {
      "epoch": 0.3096196307489939,
      "grad_norm": 0.8655645067881812,
      "learning_rate": 8.08585120446418e-06,
      "loss": 0.4539,
      "step": 4943
    },
    {
      "epoch": 0.30968226874833615,
      "grad_norm": 0.7128880550623127,
      "learning_rate": 8.085052982021849e-06,
      "loss": 0.4958,
      "step": 4944
    },
    {
      "epoch": 0.30974490674767846,
      "grad_norm": 0.8450090951052504,
      "learning_rate": 8.084254632598163e-06,
      "loss": 0.3973,
      "step": 4945
    },
    {
      "epoch": 0.30980754474702077,
      "grad_norm": 0.8859861618411089,
      "learning_rate": 8.083456156225985e-06,
      "loss": 0.4457,
      "step": 4946
    },
    {
      "epoch": 0.3098701827463631,
      "grad_norm": 0.7792504052742428,
      "learning_rate": 8.08265755293818e-06,
      "loss": 0.3705,
      "step": 4947
    },
    {
      "epoch": 0.3099328207457054,
      "grad_norm": 0.9210790691779724,
      "learning_rate": 8.081858822767616e-06,
      "loss": 0.4791,
      "step": 4948
    },
    {
      "epoch": 0.3099954587450477,
      "grad_norm": 0.8251513425269472,
      "learning_rate": 8.081059965747174e-06,
      "loss": 0.4256,
      "step": 4949
    },
    {
      "epoch": 0.31005809674439,
      "grad_norm": 0.866054015075676,
      "learning_rate": 8.080260981909729e-06,
      "loss": 0.429,
      "step": 4950
    },
    {
      "epoch": 0.3101207347437323,
      "grad_norm": 0.8185218214986945,
      "learning_rate": 8.07946187128817e-06,
      "loss": 0.4004,
      "step": 4951
    },
    {
      "epoch": 0.31018337274307456,
      "grad_norm": 0.7835128840009582,
      "learning_rate": 8.07866263391539e-06,
      "loss": 0.408,
      "step": 4952
    },
    {
      "epoch": 0.31024601074241687,
      "grad_norm": 0.8338749777146633,
      "learning_rate": 8.077863269824285e-06,
      "loss": 0.4437,
      "step": 4953
    },
    {
      "epoch": 0.3103086487417592,
      "grad_norm": 0.8574901148791424,
      "learning_rate": 8.077063779047755e-06,
      "loss": 0.4266,
      "step": 4954
    },
    {
      "epoch": 0.3103712867411015,
      "grad_norm": 1.1355841884685645,
      "learning_rate": 8.07626416161871e-06,
      "loss": 0.3942,
      "step": 4955
    },
    {
      "epoch": 0.3104339247404438,
      "grad_norm": 0.815336252835358,
      "learning_rate": 8.07546441757006e-06,
      "loss": 0.406,
      "step": 4956
    },
    {
      "epoch": 0.3104965627397861,
      "grad_norm": 0.8442167252326819,
      "learning_rate": 8.074664546934726e-06,
      "loss": 0.4014,
      "step": 4957
    },
    {
      "epoch": 0.3105592007391284,
      "grad_norm": 0.8388072079468631,
      "learning_rate": 8.073864549745627e-06,
      "loss": 0.3847,
      "step": 4958
    },
    {
      "epoch": 0.3106218387384707,
      "grad_norm": 0.8492360737290412,
      "learning_rate": 8.073064426035692e-06,
      "loss": 0.4605,
      "step": 4959
    },
    {
      "epoch": 0.310684476737813,
      "grad_norm": 0.9035413859872073,
      "learning_rate": 8.072264175837857e-06,
      "loss": 0.4385,
      "step": 4960
    },
    {
      "epoch": 0.31074711473715527,
      "grad_norm": 0.9270639923554235,
      "learning_rate": 8.071463799185057e-06,
      "loss": 0.5253,
      "step": 4961
    },
    {
      "epoch": 0.3108097527364976,
      "grad_norm": 0.8748521826830172,
      "learning_rate": 8.070663296110237e-06,
      "loss": 0.3878,
      "step": 4962
    },
    {
      "epoch": 0.3108723907358399,
      "grad_norm": 0.8541719602950977,
      "learning_rate": 8.069862666646346e-06,
      "loss": 0.4334,
      "step": 4963
    },
    {
      "epoch": 0.3109350287351822,
      "grad_norm": 0.7634609223676169,
      "learning_rate": 8.069061910826336e-06,
      "loss": 0.402,
      "step": 4964
    },
    {
      "epoch": 0.3109976667345245,
      "grad_norm": 0.763142426438047,
      "learning_rate": 8.06826102868317e-06,
      "loss": 0.4287,
      "step": 4965
    },
    {
      "epoch": 0.3110603047338668,
      "grad_norm": 0.8751859156125904,
      "learning_rate": 8.06746002024981e-06,
      "loss": 0.4403,
      "step": 4966
    },
    {
      "epoch": 0.3111229427332091,
      "grad_norm": 0.8965053476778267,
      "learning_rate": 8.066658885559226e-06,
      "loss": 0.4378,
      "step": 4967
    },
    {
      "epoch": 0.3111855807325514,
      "grad_norm": 0.8162322267952861,
      "learning_rate": 8.065857624644394e-06,
      "loss": 0.3905,
      "step": 4968
    },
    {
      "epoch": 0.3112482187318937,
      "grad_norm": 0.838199710988239,
      "learning_rate": 8.065056237538294e-06,
      "loss": 0.4063,
      "step": 4969
    },
    {
      "epoch": 0.311310856731236,
      "grad_norm": 0.9022551477948377,
      "learning_rate": 8.064254724273908e-06,
      "loss": 0.4566,
      "step": 4970
    },
    {
      "epoch": 0.3113734947305783,
      "grad_norm": 0.8344377487047416,
      "learning_rate": 8.06345308488423e-06,
      "loss": 0.3854,
      "step": 4971
    },
    {
      "epoch": 0.3114361327299206,
      "grad_norm": 0.8391274184291014,
      "learning_rate": 8.062651319402254e-06,
      "loss": 0.4257,
      "step": 4972
    },
    {
      "epoch": 0.3114987707292629,
      "grad_norm": 0.8354804147292167,
      "learning_rate": 8.061849427860982e-06,
      "loss": 0.4106,
      "step": 4973
    },
    {
      "epoch": 0.3115614087286052,
      "grad_norm": 0.8729725925010979,
      "learning_rate": 8.061047410293417e-06,
      "loss": 0.4466,
      "step": 4974
    },
    {
      "epoch": 0.3116240467279475,
      "grad_norm": 0.8533980613985834,
      "learning_rate": 8.060245266732573e-06,
      "loss": 0.415,
      "step": 4975
    },
    {
      "epoch": 0.31168668472728983,
      "grad_norm": 0.8723275814264664,
      "learning_rate": 8.059442997211468e-06,
      "loss": 0.4301,
      "step": 4976
    },
    {
      "epoch": 0.31174932272663214,
      "grad_norm": 0.8480071115350224,
      "learning_rate": 8.05864060176312e-06,
      "loss": 0.3996,
      "step": 4977
    },
    {
      "epoch": 0.3118119607259744,
      "grad_norm": 0.7488636693689433,
      "learning_rate": 8.057838080420559e-06,
      "loss": 0.3991,
      "step": 4978
    },
    {
      "epoch": 0.3118745987253167,
      "grad_norm": 0.8979794161322747,
      "learning_rate": 8.057035433216812e-06,
      "loss": 0.4712,
      "step": 4979
    },
    {
      "epoch": 0.311937236724659,
      "grad_norm": 0.7914175697472948,
      "learning_rate": 8.056232660184921e-06,
      "loss": 0.3923,
      "step": 4980
    },
    {
      "epoch": 0.3119998747240013,
      "grad_norm": 0.8801331297320059,
      "learning_rate": 8.055429761357926e-06,
      "loss": 0.4658,
      "step": 4981
    },
    {
      "epoch": 0.3120625127233436,
      "grad_norm": 0.7228825133718395,
      "learning_rate": 8.054626736768875e-06,
      "loss": 0.4714,
      "step": 4982
    },
    {
      "epoch": 0.31212515072268593,
      "grad_norm": 0.8957796904533588,
      "learning_rate": 8.05382358645082e-06,
      "loss": 0.4191,
      "step": 4983
    },
    {
      "epoch": 0.31218778872202824,
      "grad_norm": 0.9476719254737364,
      "learning_rate": 8.05302031043682e-06,
      "loss": 0.4366,
      "step": 4984
    },
    {
      "epoch": 0.31225042672137054,
      "grad_norm": 0.816339749790841,
      "learning_rate": 8.052216908759937e-06,
      "loss": 0.4103,
      "step": 4985
    },
    {
      "epoch": 0.3123130647207128,
      "grad_norm": 0.8127545411986151,
      "learning_rate": 8.05141338145324e-06,
      "loss": 0.4374,
      "step": 4986
    },
    {
      "epoch": 0.3123757027200551,
      "grad_norm": 0.8467434151950716,
      "learning_rate": 8.050609728549802e-06,
      "loss": 0.4326,
      "step": 4987
    },
    {
      "epoch": 0.3124383407193974,
      "grad_norm": 0.8782682546326628,
      "learning_rate": 8.0498059500827e-06,
      "loss": 0.3701,
      "step": 4988
    },
    {
      "epoch": 0.3125009787187397,
      "grad_norm": 0.7818101168476022,
      "learning_rate": 8.04900204608502e-06,
      "loss": 0.4166,
      "step": 4989
    },
    {
      "epoch": 0.312563616718082,
      "grad_norm": 0.8000255094968827,
      "learning_rate": 8.04819801658985e-06,
      "loss": 0.3931,
      "step": 4990
    },
    {
      "epoch": 0.31262625471742433,
      "grad_norm": 0.7929852991914422,
      "learning_rate": 8.047393861630284e-06,
      "loss": 0.3914,
      "step": 4991
    },
    {
      "epoch": 0.31268889271676664,
      "grad_norm": 0.8911560014367439,
      "learning_rate": 8.046589581239421e-06,
      "loss": 0.4407,
      "step": 4992
    },
    {
      "epoch": 0.31275153071610895,
      "grad_norm": 0.8703301936657539,
      "learning_rate": 8.045785175450365e-06,
      "loss": 0.4572,
      "step": 4993
    },
    {
      "epoch": 0.3128141687154512,
      "grad_norm": 0.7660051621870526,
      "learning_rate": 8.044980644296225e-06,
      "loss": 0.3642,
      "step": 4994
    },
    {
      "epoch": 0.3128768067147935,
      "grad_norm": 0.768686875156736,
      "learning_rate": 8.044175987810118e-06,
      "loss": 0.4281,
      "step": 4995
    },
    {
      "epoch": 0.3129394447141358,
      "grad_norm": 0.8045800405696867,
      "learning_rate": 8.043371206025161e-06,
      "loss": 0.4427,
      "step": 4996
    },
    {
      "epoch": 0.3130020827134781,
      "grad_norm": 0.784087317835657,
      "learning_rate": 8.042566298974482e-06,
      "loss": 0.3912,
      "step": 4997
    },
    {
      "epoch": 0.31306472071282043,
      "grad_norm": 0.7887204878921159,
      "learning_rate": 8.041761266691208e-06,
      "loss": 0.4062,
      "step": 4998
    },
    {
      "epoch": 0.31312735871216274,
      "grad_norm": 0.8566052689562,
      "learning_rate": 8.040956109208476e-06,
      "loss": 0.4128,
      "step": 4999
    },
    {
      "epoch": 0.31318999671150505,
      "grad_norm": 0.8476388561263404,
      "learning_rate": 8.040150826559427e-06,
      "loss": 0.3902,
      "step": 5000
    },
    {
      "epoch": 0.31325263471084736,
      "grad_norm": 0.8511144446920266,
      "learning_rate": 8.039345418777205e-06,
      "loss": 0.4283,
      "step": 5001
    },
    {
      "epoch": 0.31331527271018966,
      "grad_norm": 0.7976906055021273,
      "learning_rate": 8.03853988589496e-06,
      "loss": 0.4281,
      "step": 5002
    },
    {
      "epoch": 0.3133779107095319,
      "grad_norm": 0.7740770995445977,
      "learning_rate": 8.03773422794585e-06,
      "loss": 0.4219,
      "step": 5003
    },
    {
      "epoch": 0.3134405487088742,
      "grad_norm": 0.8435551376003861,
      "learning_rate": 8.036928444963037e-06,
      "loss": 0.4506,
      "step": 5004
    },
    {
      "epoch": 0.31350318670821653,
      "grad_norm": 0.7807971578976349,
      "learning_rate": 8.036122536979686e-06,
      "loss": 0.3669,
      "step": 5005
    },
    {
      "epoch": 0.31356582470755884,
      "grad_norm": 0.8243852545914679,
      "learning_rate": 8.035316504028966e-06,
      "loss": 0.4348,
      "step": 5006
    },
    {
      "epoch": 0.31362846270690115,
      "grad_norm": 0.7853002109598962,
      "learning_rate": 8.034510346144056e-06,
      "loss": 0.3925,
      "step": 5007
    },
    {
      "epoch": 0.31369110070624345,
      "grad_norm": 0.9128611347641988,
      "learning_rate": 8.033704063358136e-06,
      "loss": 0.4522,
      "step": 5008
    },
    {
      "epoch": 0.31375373870558576,
      "grad_norm": 0.846373018516144,
      "learning_rate": 8.032897655704394e-06,
      "loss": 0.4015,
      "step": 5009
    },
    {
      "epoch": 0.31381637670492807,
      "grad_norm": 0.8241957984544122,
      "learning_rate": 8.032091123216022e-06,
      "loss": 0.4437,
      "step": 5010
    },
    {
      "epoch": 0.3138790147042703,
      "grad_norm": 0.866024374581297,
      "learning_rate": 8.031284465926217e-06,
      "loss": 0.4325,
      "step": 5011
    },
    {
      "epoch": 0.31394165270361263,
      "grad_norm": 0.8228674787924728,
      "learning_rate": 8.030477683868178e-06,
      "loss": 0.4185,
      "step": 5012
    },
    {
      "epoch": 0.31400429070295494,
      "grad_norm": 0.8096838335560873,
      "learning_rate": 8.029670777075119e-06,
      "loss": 0.4358,
      "step": 5013
    },
    {
      "epoch": 0.31406692870229724,
      "grad_norm": 0.8341614990585298,
      "learning_rate": 8.028863745580246e-06,
      "loss": 0.4275,
      "step": 5014
    },
    {
      "epoch": 0.31412956670163955,
      "grad_norm": 0.8178754093532089,
      "learning_rate": 8.02805658941678e-06,
      "loss": 0.378,
      "step": 5015
    },
    {
      "epoch": 0.31419220470098186,
      "grad_norm": 0.8652222054014166,
      "learning_rate": 8.02724930861794e-06,
      "loss": 0.43,
      "step": 5016
    },
    {
      "epoch": 0.31425484270032417,
      "grad_norm": 0.8340277150523977,
      "learning_rate": 8.026441903216962e-06,
      "loss": 0.4271,
      "step": 5017
    },
    {
      "epoch": 0.3143174806996665,
      "grad_norm": 0.7893920333173438,
      "learning_rate": 8.02563437324707e-06,
      "loss": 0.4124,
      "step": 5018
    },
    {
      "epoch": 0.3143801186990087,
      "grad_norm": 0.915062720661596,
      "learning_rate": 8.024826718741507e-06,
      "loss": 0.4825,
      "step": 5019
    },
    {
      "epoch": 0.31444275669835103,
      "grad_norm": 0.8390900119629268,
      "learning_rate": 8.024018939733513e-06,
      "loss": 0.4412,
      "step": 5020
    },
    {
      "epoch": 0.31450539469769334,
      "grad_norm": 0.8671743173281535,
      "learning_rate": 8.02321103625634e-06,
      "loss": 0.4738,
      "step": 5021
    },
    {
      "epoch": 0.31456803269703565,
      "grad_norm": 0.8276453929865151,
      "learning_rate": 8.022403008343238e-06,
      "loss": 0.445,
      "step": 5022
    },
    {
      "epoch": 0.31463067069637796,
      "grad_norm": 0.8893083116293428,
      "learning_rate": 8.021594856027468e-06,
      "loss": 0.4457,
      "step": 5023
    },
    {
      "epoch": 0.31469330869572026,
      "grad_norm": 0.8091303699900442,
      "learning_rate": 8.020786579342295e-06,
      "loss": 0.3866,
      "step": 5024
    },
    {
      "epoch": 0.31475594669506257,
      "grad_norm": 0.7859800257243339,
      "learning_rate": 8.019978178320982e-06,
      "loss": 0.4135,
      "step": 5025
    },
    {
      "epoch": 0.3148185846944049,
      "grad_norm": 0.8794379582756335,
      "learning_rate": 8.01916965299681e-06,
      "loss": 0.4463,
      "step": 5026
    },
    {
      "epoch": 0.3148812226937472,
      "grad_norm": 0.8864412235489758,
      "learning_rate": 8.01836100340305e-06,
      "loss": 0.425,
      "step": 5027
    },
    {
      "epoch": 0.31494386069308944,
      "grad_norm": 0.8225971659117948,
      "learning_rate": 8.017552229572996e-06,
      "loss": 0.3955,
      "step": 5028
    },
    {
      "epoch": 0.31500649869243175,
      "grad_norm": 0.7928575098918874,
      "learning_rate": 8.016743331539929e-06,
      "loss": 0.3641,
      "step": 5029
    },
    {
      "epoch": 0.31506913669177405,
      "grad_norm": 0.872350969151839,
      "learning_rate": 8.015934309337148e-06,
      "loss": 0.4071,
      "step": 5030
    },
    {
      "epoch": 0.31513177469111636,
      "grad_norm": 0.7867699515649699,
      "learning_rate": 8.01512516299795e-06,
      "loss": 0.4158,
      "step": 5031
    },
    {
      "epoch": 0.31519441269045867,
      "grad_norm": 0.8529030243877322,
      "learning_rate": 8.014315892555642e-06,
      "loss": 0.4277,
      "step": 5032
    },
    {
      "epoch": 0.315257050689801,
      "grad_norm": 0.8255450829934592,
      "learning_rate": 8.01350649804353e-06,
      "loss": 0.4216,
      "step": 5033
    },
    {
      "epoch": 0.3153196886891433,
      "grad_norm": 0.799898929060876,
      "learning_rate": 8.012696979494934e-06,
      "loss": 0.4203,
      "step": 5034
    },
    {
      "epoch": 0.3153823266884856,
      "grad_norm": 0.7817476789962179,
      "learning_rate": 8.011887336943169e-06,
      "loss": 0.412,
      "step": 5035
    },
    {
      "epoch": 0.31544496468782784,
      "grad_norm": 0.7814810926697635,
      "learning_rate": 8.011077570421564e-06,
      "loss": 0.4144,
      "step": 5036
    },
    {
      "epoch": 0.31550760268717015,
      "grad_norm": 0.8589610132282892,
      "learning_rate": 8.010267679963445e-06,
      "loss": 0.4502,
      "step": 5037
    },
    {
      "epoch": 0.31557024068651246,
      "grad_norm": 0.8215749151957307,
      "learning_rate": 8.009457665602153e-06,
      "loss": 0.4183,
      "step": 5038
    },
    {
      "epoch": 0.31563287868585477,
      "grad_norm": 0.825091821072601,
      "learning_rate": 8.008647527371022e-06,
      "loss": 0.3643,
      "step": 5039
    },
    {
      "epoch": 0.3156955166851971,
      "grad_norm": 0.7860292259965894,
      "learning_rate": 8.007837265303402e-06,
      "loss": 0.4071,
      "step": 5040
    },
    {
      "epoch": 0.3157581546845394,
      "grad_norm": 0.7346543523363515,
      "learning_rate": 8.00702687943264e-06,
      "loss": 0.3598,
      "step": 5041
    },
    {
      "epoch": 0.3158207926838817,
      "grad_norm": 0.7951234030947529,
      "learning_rate": 8.006216369792096e-06,
      "loss": 0.4241,
      "step": 5042
    },
    {
      "epoch": 0.315883430683224,
      "grad_norm": 0.8986690494037133,
      "learning_rate": 8.005405736415127e-06,
      "loss": 0.4475,
      "step": 5043
    },
    {
      "epoch": 0.3159460686825663,
      "grad_norm": 0.8170577202694469,
      "learning_rate": 8.004594979335099e-06,
      "loss": 0.4424,
      "step": 5044
    },
    {
      "epoch": 0.31600870668190856,
      "grad_norm": 0.804111531451695,
      "learning_rate": 8.003784098585386e-06,
      "loss": 0.3828,
      "step": 5045
    },
    {
      "epoch": 0.31607134468125087,
      "grad_norm": 0.8617713450491025,
      "learning_rate": 8.00297309419936e-06,
      "loss": 0.4115,
      "step": 5046
    },
    {
      "epoch": 0.3161339826805932,
      "grad_norm": 0.8777830918735803,
      "learning_rate": 8.002161966210406e-06,
      "loss": 0.4659,
      "step": 5047
    },
    {
      "epoch": 0.3161966206799355,
      "grad_norm": 0.8452242205253307,
      "learning_rate": 8.001350714651905e-06,
      "loss": 0.4272,
      "step": 5048
    },
    {
      "epoch": 0.3162592586792778,
      "grad_norm": 0.7850319669913206,
      "learning_rate": 8.000539339557254e-06,
      "loss": 0.409,
      "step": 5049
    },
    {
      "epoch": 0.3163218966786201,
      "grad_norm": 0.834149838463763,
      "learning_rate": 7.999727840959848e-06,
      "loss": 0.435,
      "step": 5050
    },
    {
      "epoch": 0.3163845346779624,
      "grad_norm": 0.9453386659790933,
      "learning_rate": 7.998916218893084e-06,
      "loss": 0.4021,
      "step": 5051
    },
    {
      "epoch": 0.3164471726773047,
      "grad_norm": 0.841534784778502,
      "learning_rate": 7.998104473390376e-06,
      "loss": 0.4165,
      "step": 5052
    },
    {
      "epoch": 0.31650981067664696,
      "grad_norm": 0.8195936584125011,
      "learning_rate": 7.997292604485127e-06,
      "loss": 0.4112,
      "step": 5053
    },
    {
      "epoch": 0.31657244867598927,
      "grad_norm": 0.8555665142356441,
      "learning_rate": 7.996480612210761e-06,
      "loss": 0.4079,
      "step": 5054
    },
    {
      "epoch": 0.3166350866753316,
      "grad_norm": 0.8272620369619935,
      "learning_rate": 7.995668496600697e-06,
      "loss": 0.4236,
      "step": 5055
    },
    {
      "epoch": 0.3166977246746739,
      "grad_norm": 0.8401699108011883,
      "learning_rate": 7.99485625768836e-06,
      "loss": 0.4284,
      "step": 5056
    },
    {
      "epoch": 0.3167603626740162,
      "grad_norm": 0.7583817945665231,
      "learning_rate": 7.994043895507184e-06,
      "loss": 0.4261,
      "step": 5057
    },
    {
      "epoch": 0.3168230006733585,
      "grad_norm": 0.840327118381637,
      "learning_rate": 7.993231410090605e-06,
      "loss": 0.4375,
      "step": 5058
    },
    {
      "epoch": 0.3168856386727008,
      "grad_norm": 0.7857419956629302,
      "learning_rate": 7.992418801472066e-06,
      "loss": 0.4035,
      "step": 5059
    },
    {
      "epoch": 0.3169482766720431,
      "grad_norm": 0.7721464536042302,
      "learning_rate": 7.991606069685016e-06,
      "loss": 0.4295,
      "step": 5060
    },
    {
      "epoch": 0.31701091467138537,
      "grad_norm": 0.7813907753958246,
      "learning_rate": 7.990793214762902e-06,
      "loss": 0.3931,
      "step": 5061
    },
    {
      "epoch": 0.3170735526707277,
      "grad_norm": 0.8761601871025742,
      "learning_rate": 7.989980236739185e-06,
      "loss": 0.4666,
      "step": 5062
    },
    {
      "epoch": 0.31713619067007,
      "grad_norm": 0.8499312699852934,
      "learning_rate": 7.989167135647327e-06,
      "loss": 0.3971,
      "step": 5063
    },
    {
      "epoch": 0.3171988286694123,
      "grad_norm": 0.8853208479016473,
      "learning_rate": 7.988353911520794e-06,
      "loss": 0.428,
      "step": 5064
    },
    {
      "epoch": 0.3172614666687546,
      "grad_norm": 0.8109396967026704,
      "learning_rate": 7.98754056439306e-06,
      "loss": 0.4253,
      "step": 5065
    },
    {
      "epoch": 0.3173241046680969,
      "grad_norm": 0.7571396257356571,
      "learning_rate": 7.986727094297601e-06,
      "loss": 0.3928,
      "step": 5066
    },
    {
      "epoch": 0.3173867426674392,
      "grad_norm": 0.7258914342160417,
      "learning_rate": 7.985913501267901e-06,
      "loss": 0.4888,
      "step": 5067
    },
    {
      "epoch": 0.3174493806667815,
      "grad_norm": 0.7227400191419782,
      "learning_rate": 7.985099785337447e-06,
      "loss": 0.4823,
      "step": 5068
    },
    {
      "epoch": 0.31751201866612383,
      "grad_norm": 0.870311710580583,
      "learning_rate": 7.984285946539733e-06,
      "loss": 0.4319,
      "step": 5069
    },
    {
      "epoch": 0.3175746566654661,
      "grad_norm": 0.7870891849908813,
      "learning_rate": 7.983471984908254e-06,
      "loss": 0.4241,
      "step": 5070
    },
    {
      "epoch": 0.3176372946648084,
      "grad_norm": 0.7652804801750462,
      "learning_rate": 7.982657900476514e-06,
      "loss": 0.3623,
      "step": 5071
    },
    {
      "epoch": 0.3176999326641507,
      "grad_norm": 0.7454536378953547,
      "learning_rate": 7.981843693278022e-06,
      "loss": 0.3678,
      "step": 5072
    },
    {
      "epoch": 0.317762570663493,
      "grad_norm": 0.8222413615877726,
      "learning_rate": 7.98102936334629e-06,
      "loss": 0.4307,
      "step": 5073
    },
    {
      "epoch": 0.3178252086628353,
      "grad_norm": 0.8167878270057956,
      "learning_rate": 7.980214910714838e-06,
      "loss": 0.4125,
      "step": 5074
    },
    {
      "epoch": 0.3178878466621776,
      "grad_norm": 0.7921746290636332,
      "learning_rate": 7.979400335417185e-06,
      "loss": 0.4308,
      "step": 5075
    },
    {
      "epoch": 0.3179504846615199,
      "grad_norm": 0.9115981966697431,
      "learning_rate": 7.978585637486861e-06,
      "loss": 0.4456,
      "step": 5076
    },
    {
      "epoch": 0.31801312266086224,
      "grad_norm": 0.817251161767991,
      "learning_rate": 7.977770816957399e-06,
      "loss": 0.3928,
      "step": 5077
    },
    {
      "epoch": 0.3180757606602045,
      "grad_norm": 0.8752118185182527,
      "learning_rate": 7.97695587386234e-06,
      "loss": 0.4763,
      "step": 5078
    },
    {
      "epoch": 0.3181383986595468,
      "grad_norm": 0.8228199238801946,
      "learning_rate": 7.976140808235222e-06,
      "loss": 0.4236,
      "step": 5079
    },
    {
      "epoch": 0.3182010366588891,
      "grad_norm": 0.8354934210154288,
      "learning_rate": 7.975325620109597e-06,
      "loss": 0.3972,
      "step": 5080
    },
    {
      "epoch": 0.3182636746582314,
      "grad_norm": 0.764551719396803,
      "learning_rate": 7.974510309519017e-06,
      "loss": 0.3866,
      "step": 5081
    },
    {
      "epoch": 0.3183263126575737,
      "grad_norm": 0.8183191563884643,
      "learning_rate": 7.97369487649704e-06,
      "loss": 0.4049,
      "step": 5082
    },
    {
      "epoch": 0.318388950656916,
      "grad_norm": 0.8192069499086723,
      "learning_rate": 7.972879321077231e-06,
      "loss": 0.3928,
      "step": 5083
    },
    {
      "epoch": 0.31845158865625833,
      "grad_norm": 0.8749200938098107,
      "learning_rate": 7.972063643293158e-06,
      "loss": 0.4721,
      "step": 5084
    },
    {
      "epoch": 0.31851422665560064,
      "grad_norm": 0.8331134401759855,
      "learning_rate": 7.971247843178393e-06,
      "loss": 0.4593,
      "step": 5085
    },
    {
      "epoch": 0.31857686465494295,
      "grad_norm": 0.693272219480031,
      "learning_rate": 7.970431920766515e-06,
      "loss": 0.4641,
      "step": 5086
    },
    {
      "epoch": 0.3186395026542852,
      "grad_norm": 0.766556389996861,
      "learning_rate": 7.969615876091107e-06,
      "loss": 0.4108,
      "step": 5087
    },
    {
      "epoch": 0.3187021406536275,
      "grad_norm": 0.7688057877985655,
      "learning_rate": 7.96879970918576e-06,
      "loss": 0.4192,
      "step": 5088
    },
    {
      "epoch": 0.3187647786529698,
      "grad_norm": 0.8625380555648381,
      "learning_rate": 7.967983420084065e-06,
      "loss": 0.3959,
      "step": 5089
    },
    {
      "epoch": 0.3188274166523121,
      "grad_norm": 0.7795298325638965,
      "learning_rate": 7.967167008819622e-06,
      "loss": 0.3985,
      "step": 5090
    },
    {
      "epoch": 0.31889005465165443,
      "grad_norm": 0.8000182880707414,
      "learning_rate": 7.966350475426034e-06,
      "loss": 0.4073,
      "step": 5091
    },
    {
      "epoch": 0.31895269265099674,
      "grad_norm": 0.7639855399568938,
      "learning_rate": 7.965533819936911e-06,
      "loss": 0.4267,
      "step": 5092
    },
    {
      "epoch": 0.31901533065033905,
      "grad_norm": 0.8634515792513638,
      "learning_rate": 7.964717042385864e-06,
      "loss": 0.3667,
      "step": 5093
    },
    {
      "epoch": 0.31907796864968135,
      "grad_norm": 0.9153722687850432,
      "learning_rate": 7.963900142806513e-06,
      "loss": 0.4343,
      "step": 5094
    },
    {
      "epoch": 0.3191406066490236,
      "grad_norm": 0.806969290057027,
      "learning_rate": 7.963083121232482e-06,
      "loss": 0.4374,
      "step": 5095
    },
    {
      "epoch": 0.3192032446483659,
      "grad_norm": 0.8011560804805132,
      "learning_rate": 7.962265977697401e-06,
      "loss": 0.3964,
      "step": 5096
    },
    {
      "epoch": 0.3192658826477082,
      "grad_norm": 0.8194022514167234,
      "learning_rate": 7.961448712234902e-06,
      "loss": 0.4141,
      "step": 5097
    },
    {
      "epoch": 0.31932852064705053,
      "grad_norm": 0.7889112894621251,
      "learning_rate": 7.960631324878624e-06,
      "loss": 0.405,
      "step": 5098
    },
    {
      "epoch": 0.31939115864639284,
      "grad_norm": 0.8884315233226531,
      "learning_rate": 7.959813815662211e-06,
      "loss": 0.4279,
      "step": 5099
    },
    {
      "epoch": 0.31945379664573514,
      "grad_norm": 0.878282475662147,
      "learning_rate": 7.958996184619312e-06,
      "loss": 0.4255,
      "step": 5100
    },
    {
      "epoch": 0.31951643464507745,
      "grad_norm": 0.7843157168071532,
      "learning_rate": 7.95817843178358e-06,
      "loss": 0.3832,
      "step": 5101
    },
    {
      "epoch": 0.31957907264441976,
      "grad_norm": 0.8243283523146547,
      "learning_rate": 7.957360557188678e-06,
      "loss": 0.3883,
      "step": 5102
    },
    {
      "epoch": 0.319641710643762,
      "grad_norm": 0.8811077643788153,
      "learning_rate": 7.956542560868264e-06,
      "loss": 0.434,
      "step": 5103
    },
    {
      "epoch": 0.3197043486431043,
      "grad_norm": 0.8353962638517818,
      "learning_rate": 7.95572444285601e-06,
      "loss": 0.4046,
      "step": 5104
    },
    {
      "epoch": 0.3197669866424466,
      "grad_norm": 0.8912945681717526,
      "learning_rate": 7.95490620318559e-06,
      "loss": 0.4379,
      "step": 5105
    },
    {
      "epoch": 0.31982962464178893,
      "grad_norm": 0.8430068866310548,
      "learning_rate": 7.954087841890681e-06,
      "loss": 0.414,
      "step": 5106
    },
    {
      "epoch": 0.31989226264113124,
      "grad_norm": 0.8035373109116811,
      "learning_rate": 7.953269359004968e-06,
      "loss": 0.4228,
      "step": 5107
    },
    {
      "epoch": 0.31995490064047355,
      "grad_norm": 0.8471922514389083,
      "learning_rate": 7.95245075456214e-06,
      "loss": 0.4239,
      "step": 5108
    },
    {
      "epoch": 0.32001753863981586,
      "grad_norm": 0.9214329088778995,
      "learning_rate": 7.951632028595893e-06,
      "loss": 0.4291,
      "step": 5109
    },
    {
      "epoch": 0.32008017663915816,
      "grad_norm": 0.816719037346325,
      "learning_rate": 7.950813181139921e-06,
      "loss": 0.4168,
      "step": 5110
    },
    {
      "epoch": 0.3201428146385005,
      "grad_norm": 0.8806120189220769,
      "learning_rate": 7.949994212227932e-06,
      "loss": 0.4418,
      "step": 5111
    },
    {
      "epoch": 0.3202054526378427,
      "grad_norm": 0.8486901878545502,
      "learning_rate": 7.949175121893635e-06,
      "loss": 0.4064,
      "step": 5112
    },
    {
      "epoch": 0.32026809063718503,
      "grad_norm": 0.8593090948505816,
      "learning_rate": 7.94835591017074e-06,
      "loss": 0.4099,
      "step": 5113
    },
    {
      "epoch": 0.32033072863652734,
      "grad_norm": 0.8798046875153371,
      "learning_rate": 7.947536577092971e-06,
      "loss": 0.4434,
      "step": 5114
    },
    {
      "epoch": 0.32039336663586965,
      "grad_norm": 0.822305067817988,
      "learning_rate": 7.946717122694048e-06,
      "loss": 0.451,
      "step": 5115
    },
    {
      "epoch": 0.32045600463521196,
      "grad_norm": 0.766169081319551,
      "learning_rate": 7.945897547007702e-06,
      "loss": 0.3676,
      "step": 5116
    },
    {
      "epoch": 0.32051864263455426,
      "grad_norm": 0.7887326480840396,
      "learning_rate": 7.945077850067665e-06,
      "loss": 0.3708,
      "step": 5117
    },
    {
      "epoch": 0.32058128063389657,
      "grad_norm": 0.7612562744926782,
      "learning_rate": 7.944258031907678e-06,
      "loss": 0.3763,
      "step": 5118
    },
    {
      "epoch": 0.3206439186332389,
      "grad_norm": 0.8767722188366937,
      "learning_rate": 7.943438092561483e-06,
      "loss": 0.4076,
      "step": 5119
    },
    {
      "epoch": 0.32070655663258113,
      "grad_norm": 0.8558357678496453,
      "learning_rate": 7.942618032062831e-06,
      "loss": 0.411,
      "step": 5120
    },
    {
      "epoch": 0.32076919463192344,
      "grad_norm": 0.8976153762458131,
      "learning_rate": 7.941797850445474e-06,
      "loss": 0.4434,
      "step": 5121
    },
    {
      "epoch": 0.32083183263126575,
      "grad_norm": 0.8304389417641109,
      "learning_rate": 7.940977547743172e-06,
      "loss": 0.3989,
      "step": 5122
    },
    {
      "epoch": 0.32089447063060805,
      "grad_norm": 0.7633313427952682,
      "learning_rate": 7.940157123989687e-06,
      "loss": 0.3787,
      "step": 5123
    },
    {
      "epoch": 0.32095710862995036,
      "grad_norm": 0.8240260637543002,
      "learning_rate": 7.93933657921879e-06,
      "loss": 0.3975,
      "step": 5124
    },
    {
      "epoch": 0.32101974662929267,
      "grad_norm": 0.9156991106961266,
      "learning_rate": 7.938515913464255e-06,
      "loss": 0.4525,
      "step": 5125
    },
    {
      "epoch": 0.321082384628635,
      "grad_norm": 0.8423178679928036,
      "learning_rate": 7.937695126759856e-06,
      "loss": 0.4215,
      "step": 5126
    },
    {
      "epoch": 0.3211450226279773,
      "grad_norm": 0.7585947538839439,
      "learning_rate": 7.936874219139384e-06,
      "loss": 0.4144,
      "step": 5127
    },
    {
      "epoch": 0.3212076606273196,
      "grad_norm": 0.7542764520504074,
      "learning_rate": 7.936053190636623e-06,
      "loss": 0.4218,
      "step": 5128
    },
    {
      "epoch": 0.32127029862666184,
      "grad_norm": 0.8255419977216665,
      "learning_rate": 7.935232041285367e-06,
      "loss": 0.4554,
      "step": 5129
    },
    {
      "epoch": 0.32133293662600415,
      "grad_norm": 0.9513257202795389,
      "learning_rate": 7.934410771119416e-06,
      "loss": 0.4777,
      "step": 5130
    },
    {
      "epoch": 0.32139557462534646,
      "grad_norm": 0.8714961997552328,
      "learning_rate": 7.933589380172573e-06,
      "loss": 0.3996,
      "step": 5131
    },
    {
      "epoch": 0.32145821262468877,
      "grad_norm": 0.8387507151359753,
      "learning_rate": 7.932767868478645e-06,
      "loss": 0.4168,
      "step": 5132
    },
    {
      "epoch": 0.3215208506240311,
      "grad_norm": 0.9113427581211256,
      "learning_rate": 7.93194623607145e-06,
      "loss": 0.4449,
      "step": 5133
    },
    {
      "epoch": 0.3215834886233734,
      "grad_norm": 0.9384277128389048,
      "learning_rate": 7.931124482984802e-06,
      "loss": 0.4113,
      "step": 5134
    },
    {
      "epoch": 0.3216461266227157,
      "grad_norm": 0.8155859865800021,
      "learning_rate": 7.930302609252527e-06,
      "loss": 0.4019,
      "step": 5135
    },
    {
      "epoch": 0.321708764622058,
      "grad_norm": 0.8678883633706171,
      "learning_rate": 7.929480614908452e-06,
      "loss": 0.394,
      "step": 5136
    },
    {
      "epoch": 0.32177140262140025,
      "grad_norm": 0.838704309371698,
      "learning_rate": 7.928658499986413e-06,
      "loss": 0.388,
      "step": 5137
    },
    {
      "epoch": 0.32183404062074256,
      "grad_norm": 0.8041954657203737,
      "learning_rate": 7.927836264520243e-06,
      "loss": 0.4109,
      "step": 5138
    },
    {
      "epoch": 0.32189667862008486,
      "grad_norm": 0.8162250247171499,
      "learning_rate": 7.927013908543791e-06,
      "loss": 0.3978,
      "step": 5139
    },
    {
      "epoch": 0.32195931661942717,
      "grad_norm": 0.705940218522006,
      "learning_rate": 7.926191432090904e-06,
      "loss": 0.4794,
      "step": 5140
    },
    {
      "epoch": 0.3220219546187695,
      "grad_norm": 0.8106034938412581,
      "learning_rate": 7.925368835195433e-06,
      "loss": 0.437,
      "step": 5141
    },
    {
      "epoch": 0.3220845926181118,
      "grad_norm": 0.8220792280345631,
      "learning_rate": 7.92454611789124e-06,
      "loss": 0.3992,
      "step": 5142
    },
    {
      "epoch": 0.3221472306174541,
      "grad_norm": 0.9381709283294348,
      "learning_rate": 7.923723280212183e-06,
      "loss": 0.4565,
      "step": 5143
    },
    {
      "epoch": 0.3222098686167964,
      "grad_norm": 0.8004151985912675,
      "learning_rate": 7.922900322192136e-06,
      "loss": 0.4075,
      "step": 5144
    },
    {
      "epoch": 0.32227250661613865,
      "grad_norm": 0.8175780075222279,
      "learning_rate": 7.922077243864967e-06,
      "loss": 0.3918,
      "step": 5145
    },
    {
      "epoch": 0.32233514461548096,
      "grad_norm": 0.9086848845550608,
      "learning_rate": 7.921254045264557e-06,
      "loss": 0.4068,
      "step": 5146
    },
    {
      "epoch": 0.32239778261482327,
      "grad_norm": 0.8571611337233178,
      "learning_rate": 7.920430726424788e-06,
      "loss": 0.4283,
      "step": 5147
    },
    {
      "epoch": 0.3224604206141656,
      "grad_norm": 0.8099183880975428,
      "learning_rate": 7.91960728737955e-06,
      "loss": 0.4185,
      "step": 5148
    },
    {
      "epoch": 0.3225230586135079,
      "grad_norm": 0.8294622248588632,
      "learning_rate": 7.918783728162732e-06,
      "loss": 0.4279,
      "step": 5149
    },
    {
      "epoch": 0.3225856966128502,
      "grad_norm": 0.8426016627548362,
      "learning_rate": 7.917960048808234e-06,
      "loss": 0.4142,
      "step": 5150
    },
    {
      "epoch": 0.3226483346121925,
      "grad_norm": 0.8556031218288799,
      "learning_rate": 7.91713624934996e-06,
      "loss": 0.3956,
      "step": 5151
    },
    {
      "epoch": 0.3227109726115348,
      "grad_norm": 0.8273624424819198,
      "learning_rate": 7.916312329821816e-06,
      "loss": 0.3747,
      "step": 5152
    },
    {
      "epoch": 0.3227736106108771,
      "grad_norm": 0.8293687354486595,
      "learning_rate": 7.915488290257716e-06,
      "loss": 0.4028,
      "step": 5153
    },
    {
      "epoch": 0.32283624861021937,
      "grad_norm": 0.8477889067520357,
      "learning_rate": 7.914664130691574e-06,
      "loss": 0.4319,
      "step": 5154
    },
    {
      "epoch": 0.3228988866095617,
      "grad_norm": 0.9067019239208159,
      "learning_rate": 7.913839851157317e-06,
      "loss": 0.4516,
      "step": 5155
    },
    {
      "epoch": 0.322961524608904,
      "grad_norm": 0.840318193242577,
      "learning_rate": 7.913015451688872e-06,
      "loss": 0.4026,
      "step": 5156
    },
    {
      "epoch": 0.3230241626082463,
      "grad_norm": 0.8432644516422114,
      "learning_rate": 7.912190932320169e-06,
      "loss": 0.4483,
      "step": 5157
    },
    {
      "epoch": 0.3230868006075886,
      "grad_norm": 0.8088876450774706,
      "learning_rate": 7.91136629308515e-06,
      "loss": 0.4212,
      "step": 5158
    },
    {
      "epoch": 0.3231494386069309,
      "grad_norm": 0.8268646435789943,
      "learning_rate": 7.91054153401775e-06,
      "loss": 0.4075,
      "step": 5159
    },
    {
      "epoch": 0.3232120766062732,
      "grad_norm": 0.8145328858992478,
      "learning_rate": 7.909716655151923e-06,
      "loss": 0.4057,
      "step": 5160
    },
    {
      "epoch": 0.3232747146056155,
      "grad_norm": 0.8571044209264677,
      "learning_rate": 7.908891656521616e-06,
      "loss": 0.4159,
      "step": 5161
    },
    {
      "epoch": 0.3233373526049578,
      "grad_norm": 0.8438819477544031,
      "learning_rate": 7.90806653816079e-06,
      "loss": 0.4235,
      "step": 5162
    },
    {
      "epoch": 0.3233999906043001,
      "grad_norm": 0.823826467282001,
      "learning_rate": 7.907241300103404e-06,
      "loss": 0.4161,
      "step": 5163
    },
    {
      "epoch": 0.3234626286036424,
      "grad_norm": 0.826184893237902,
      "learning_rate": 7.906415942383428e-06,
      "loss": 0.4019,
      "step": 5164
    },
    {
      "epoch": 0.3235252666029847,
      "grad_norm": 0.79229232498212,
      "learning_rate": 7.90559046503483e-06,
      "loss": 0.4009,
      "step": 5165
    },
    {
      "epoch": 0.323587904602327,
      "grad_norm": 0.822214733564842,
      "learning_rate": 7.904764868091592e-06,
      "loss": 0.4148,
      "step": 5166
    },
    {
      "epoch": 0.3236505426016693,
      "grad_norm": 0.8352103384290857,
      "learning_rate": 7.903939151587694e-06,
      "loss": 0.437,
      "step": 5167
    },
    {
      "epoch": 0.3237131806010116,
      "grad_norm": 0.8570159767797313,
      "learning_rate": 7.903113315557118e-06,
      "loss": 0.4458,
      "step": 5168
    },
    {
      "epoch": 0.3237758186003539,
      "grad_norm": 0.8379369054574387,
      "learning_rate": 7.90228736003386e-06,
      "loss": 0.4388,
      "step": 5169
    },
    {
      "epoch": 0.3238384565996962,
      "grad_norm": 0.8474233460325179,
      "learning_rate": 7.901461285051917e-06,
      "loss": 0.4382,
      "step": 5170
    },
    {
      "epoch": 0.3239010945990385,
      "grad_norm": 0.8086763967693164,
      "learning_rate": 7.900635090645288e-06,
      "loss": 0.4122,
      "step": 5171
    },
    {
      "epoch": 0.3239637325983808,
      "grad_norm": 0.8574321962283564,
      "learning_rate": 7.899808776847979e-06,
      "loss": 0.4328,
      "step": 5172
    },
    {
      "epoch": 0.3240263705977231,
      "grad_norm": 0.8537750876897777,
      "learning_rate": 7.898982343694006e-06,
      "loss": 0.4013,
      "step": 5173
    },
    {
      "epoch": 0.3240890085970654,
      "grad_norm": 0.7537857920242367,
      "learning_rate": 7.89815579121738e-06,
      "loss": 0.4022,
      "step": 5174
    },
    {
      "epoch": 0.3241516465964077,
      "grad_norm": 0.8020932699553239,
      "learning_rate": 7.897329119452123e-06,
      "loss": 0.4133,
      "step": 5175
    },
    {
      "epoch": 0.32421428459575,
      "grad_norm": 0.8909573668487926,
      "learning_rate": 7.89650232843226e-06,
      "loss": 0.4251,
      "step": 5176
    },
    {
      "epoch": 0.32427692259509233,
      "grad_norm": 0.8252855755328361,
      "learning_rate": 7.895675418191826e-06,
      "loss": 0.4502,
      "step": 5177
    },
    {
      "epoch": 0.32433956059443464,
      "grad_norm": 0.7606499580307765,
      "learning_rate": 7.894848388764854e-06,
      "loss": 0.4147,
      "step": 5178
    },
    {
      "epoch": 0.3244021985937769,
      "grad_norm": 0.7608191523306237,
      "learning_rate": 7.894021240185384e-06,
      "loss": 0.4247,
      "step": 5179
    },
    {
      "epoch": 0.3244648365931192,
      "grad_norm": 0.8516529021772351,
      "learning_rate": 7.893193972487465e-06,
      "loss": 0.4602,
      "step": 5180
    },
    {
      "epoch": 0.3245274745924615,
      "grad_norm": 0.8437478543755268,
      "learning_rate": 7.892366585705142e-06,
      "loss": 0.4378,
      "step": 5181
    },
    {
      "epoch": 0.3245901125918038,
      "grad_norm": 0.7588707808856535,
      "learning_rate": 7.891539079872475e-06,
      "loss": 0.4104,
      "step": 5182
    },
    {
      "epoch": 0.3246527505911461,
      "grad_norm": 0.8433952312948759,
      "learning_rate": 7.890711455023523e-06,
      "loss": 0.4228,
      "step": 5183
    },
    {
      "epoch": 0.32471538859048843,
      "grad_norm": 0.6971345138150729,
      "learning_rate": 7.889883711192351e-06,
      "loss": 0.4725,
      "step": 5184
    },
    {
      "epoch": 0.32477802658983074,
      "grad_norm": 0.8971534143218326,
      "learning_rate": 7.88905584841303e-06,
      "loss": 0.4203,
      "step": 5185
    },
    {
      "epoch": 0.32484066458917304,
      "grad_norm": 0.8464316643830126,
      "learning_rate": 7.888227866719632e-06,
      "loss": 0.4246,
      "step": 5186
    },
    {
      "epoch": 0.3249033025885153,
      "grad_norm": 0.8053237142071493,
      "learning_rate": 7.887399766146238e-06,
      "loss": 0.4429,
      "step": 5187
    },
    {
      "epoch": 0.3249659405878576,
      "grad_norm": 0.8312932704903643,
      "learning_rate": 7.886571546726937e-06,
      "loss": 0.4555,
      "step": 5188
    },
    {
      "epoch": 0.3250285785871999,
      "grad_norm": 0.847991799994537,
      "learning_rate": 7.885743208495812e-06,
      "loss": 0.4215,
      "step": 5189
    },
    {
      "epoch": 0.3250912165865422,
      "grad_norm": 0.8395216531459325,
      "learning_rate": 7.884914751486965e-06,
      "loss": 0.383,
      "step": 5190
    },
    {
      "epoch": 0.3251538545858845,
      "grad_norm": 0.7954324970136867,
      "learning_rate": 7.884086175734488e-06,
      "loss": 0.3952,
      "step": 5191
    },
    {
      "epoch": 0.32521649258522684,
      "grad_norm": 0.7494853546601952,
      "learning_rate": 7.883257481272491e-06,
      "loss": 0.3926,
      "step": 5192
    },
    {
      "epoch": 0.32527913058456914,
      "grad_norm": 0.8766569986337366,
      "learning_rate": 7.882428668135078e-06,
      "loss": 0.4079,
      "step": 5193
    },
    {
      "epoch": 0.32534176858391145,
      "grad_norm": 0.8844268047526014,
      "learning_rate": 7.88159973635637e-06,
      "loss": 0.4368,
      "step": 5194
    },
    {
      "epoch": 0.32540440658325376,
      "grad_norm": 0.8427083573910293,
      "learning_rate": 7.880770685970478e-06,
      "loss": 0.4473,
      "step": 5195
    },
    {
      "epoch": 0.325467044582596,
      "grad_norm": 0.7743768442109207,
      "learning_rate": 7.879941517011531e-06,
      "loss": 0.4246,
      "step": 5196
    },
    {
      "epoch": 0.3255296825819383,
      "grad_norm": 0.8113787078393224,
      "learning_rate": 7.879112229513658e-06,
      "loss": 0.4347,
      "step": 5197
    },
    {
      "epoch": 0.3255923205812806,
      "grad_norm": 1.1686974398568224,
      "learning_rate": 7.87828282351099e-06,
      "loss": 0.4002,
      "step": 5198
    },
    {
      "epoch": 0.32565495858062293,
      "grad_norm": 0.7523577953243661,
      "learning_rate": 7.877453299037667e-06,
      "loss": 0.3869,
      "step": 5199
    },
    {
      "epoch": 0.32571759657996524,
      "grad_norm": 0.7965113550704139,
      "learning_rate": 7.876623656127833e-06,
      "loss": 0.4173,
      "step": 5200
    },
    {
      "epoch": 0.32578023457930755,
      "grad_norm": 0.8047413241699718,
      "learning_rate": 7.875793894815635e-06,
      "loss": 0.3914,
      "step": 5201
    },
    {
      "epoch": 0.32584287257864986,
      "grad_norm": 0.8711412063980055,
      "learning_rate": 7.874964015135225e-06,
      "loss": 0.4461,
      "step": 5202
    },
    {
      "epoch": 0.32590551057799216,
      "grad_norm": 0.8471799604052368,
      "learning_rate": 7.874134017120765e-06,
      "loss": 0.3971,
      "step": 5203
    },
    {
      "epoch": 0.3259681485773344,
      "grad_norm": 0.9086624330844046,
      "learning_rate": 7.873303900806415e-06,
      "loss": 0.409,
      "step": 5204
    },
    {
      "epoch": 0.3260307865766767,
      "grad_norm": 0.8610280774038157,
      "learning_rate": 7.872473666226342e-06,
      "loss": 0.4323,
      "step": 5205
    },
    {
      "epoch": 0.32609342457601903,
      "grad_norm": 0.7975886151189318,
      "learning_rate": 7.871643313414718e-06,
      "loss": 0.4098,
      "step": 5206
    },
    {
      "epoch": 0.32615606257536134,
      "grad_norm": 0.8221172437951643,
      "learning_rate": 7.870812842405727e-06,
      "loss": 0.4046,
      "step": 5207
    },
    {
      "epoch": 0.32621870057470365,
      "grad_norm": 0.8957067861082115,
      "learning_rate": 7.869982253233542e-06,
      "loss": 0.4347,
      "step": 5208
    },
    {
      "epoch": 0.32628133857404595,
      "grad_norm": 0.8137599892687507,
      "learning_rate": 7.869151545932357e-06,
      "loss": 0.4035,
      "step": 5209
    },
    {
      "epoch": 0.32634397657338826,
      "grad_norm": 0.8776500273764155,
      "learning_rate": 7.868320720536361e-06,
      "loss": 0.4224,
      "step": 5210
    },
    {
      "epoch": 0.32640661457273057,
      "grad_norm": 0.7441117667789882,
      "learning_rate": 7.867489777079753e-06,
      "loss": 0.4798,
      "step": 5211
    },
    {
      "epoch": 0.3264692525720728,
      "grad_norm": 0.8800383922663001,
      "learning_rate": 7.866658715596731e-06,
      "loss": 0.4608,
      "step": 5212
    },
    {
      "epoch": 0.32653189057141513,
      "grad_norm": 0.7949715905612069,
      "learning_rate": 7.865827536121507e-06,
      "loss": 0.4272,
      "step": 5213
    },
    {
      "epoch": 0.32659452857075744,
      "grad_norm": 0.8637128771942791,
      "learning_rate": 7.864996238688288e-06,
      "loss": 0.4559,
      "step": 5214
    },
    {
      "epoch": 0.32665716657009974,
      "grad_norm": 0.8031605191192978,
      "learning_rate": 7.864164823331293e-06,
      "loss": 0.4692,
      "step": 5215
    },
    {
      "epoch": 0.32671980456944205,
      "grad_norm": 0.882526402663248,
      "learning_rate": 7.863333290084742e-06,
      "loss": 0.3796,
      "step": 5216
    },
    {
      "epoch": 0.32678244256878436,
      "grad_norm": 0.8041391292570598,
      "learning_rate": 7.86250163898286e-06,
      "loss": 0.4459,
      "step": 5217
    },
    {
      "epoch": 0.32684508056812667,
      "grad_norm": 0.8174635772643305,
      "learning_rate": 7.86166987005988e-06,
      "loss": 0.4201,
      "step": 5218
    },
    {
      "epoch": 0.326907718567469,
      "grad_norm": 0.830859391813682,
      "learning_rate": 7.860837983350038e-06,
      "loss": 0.4025,
      "step": 5219
    },
    {
      "epoch": 0.3269703565668113,
      "grad_norm": 0.8377250084170831,
      "learning_rate": 7.860005978887574e-06,
      "loss": 0.4449,
      "step": 5220
    },
    {
      "epoch": 0.32703299456615353,
      "grad_norm": 0.836468019483843,
      "learning_rate": 7.859173856706731e-06,
      "loss": 0.4162,
      "step": 5221
    },
    {
      "epoch": 0.32709563256549584,
      "grad_norm": 0.8321822606085888,
      "learning_rate": 7.858341616841764e-06,
      "loss": 0.4282,
      "step": 5222
    },
    {
      "epoch": 0.32715827056483815,
      "grad_norm": 0.8462278141045863,
      "learning_rate": 7.857509259326923e-06,
      "loss": 0.4168,
      "step": 5223
    },
    {
      "epoch": 0.32722090856418046,
      "grad_norm": 0.7879583520291029,
      "learning_rate": 7.856676784196472e-06,
      "loss": 0.4216,
      "step": 5224
    },
    {
      "epoch": 0.32728354656352276,
      "grad_norm": 0.8171829645752693,
      "learning_rate": 7.855844191484673e-06,
      "loss": 0.3906,
      "step": 5225
    },
    {
      "epoch": 0.3273461845628651,
      "grad_norm": 0.8110149520865066,
      "learning_rate": 7.855011481225798e-06,
      "loss": 0.4349,
      "step": 5226
    },
    {
      "epoch": 0.3274088225622074,
      "grad_norm": 0.7403512867452526,
      "learning_rate": 7.854178653454121e-06,
      "loss": 0.39,
      "step": 5227
    },
    {
      "epoch": 0.3274714605615497,
      "grad_norm": 1.518850867008762,
      "learning_rate": 7.85334570820392e-06,
      "loss": 0.4213,
      "step": 5228
    },
    {
      "epoch": 0.32753409856089194,
      "grad_norm": 0.7465356385351459,
      "learning_rate": 7.85251264550948e-06,
      "loss": 0.4701,
      "step": 5229
    },
    {
      "epoch": 0.32759673656023425,
      "grad_norm": 0.7517685486742868,
      "learning_rate": 7.851679465405091e-06,
      "loss": 0.4616,
      "step": 5230
    },
    {
      "epoch": 0.32765937455957655,
      "grad_norm": 1.017334999609808,
      "learning_rate": 7.850846167925046e-06,
      "loss": 0.4295,
      "step": 5231
    },
    {
      "epoch": 0.32772201255891886,
      "grad_norm": 0.8136609292698928,
      "learning_rate": 7.850012753103641e-06,
      "loss": 0.4254,
      "step": 5232
    },
    {
      "epoch": 0.32778465055826117,
      "grad_norm": 0.8873884659514182,
      "learning_rate": 7.849179220975185e-06,
      "loss": 0.4352,
      "step": 5233
    },
    {
      "epoch": 0.3278472885576035,
      "grad_norm": 0.8842457946901302,
      "learning_rate": 7.848345571573983e-06,
      "loss": 0.4425,
      "step": 5234
    },
    {
      "epoch": 0.3279099265569458,
      "grad_norm": 0.9572974206899777,
      "learning_rate": 7.847511804934347e-06,
      "loss": 0.375,
      "step": 5235
    },
    {
      "epoch": 0.3279725645562881,
      "grad_norm": 0.796723952865629,
      "learning_rate": 7.846677921090597e-06,
      "loss": 0.3728,
      "step": 5236
    },
    {
      "epoch": 0.3280352025556304,
      "grad_norm": 1.0073057410781978,
      "learning_rate": 7.845843920077057e-06,
      "loss": 0.4122,
      "step": 5237
    },
    {
      "epoch": 0.32809784055497265,
      "grad_norm": 0.8222227857596126,
      "learning_rate": 7.845009801928051e-06,
      "loss": 0.417,
      "step": 5238
    },
    {
      "epoch": 0.32816047855431496,
      "grad_norm": 0.8809618416070407,
      "learning_rate": 7.844175566677915e-06,
      "loss": 0.4292,
      "step": 5239
    },
    {
      "epoch": 0.32822311655365727,
      "grad_norm": 0.8322989468295334,
      "learning_rate": 7.843341214360985e-06,
      "loss": 0.3764,
      "step": 5240
    },
    {
      "epoch": 0.3282857545529996,
      "grad_norm": 0.8262459146177296,
      "learning_rate": 7.842506745011601e-06,
      "loss": 0.3759,
      "step": 5241
    },
    {
      "epoch": 0.3283483925523419,
      "grad_norm": 0.8370993609297789,
      "learning_rate": 7.841672158664113e-06,
      "loss": 0.4547,
      "step": 5242
    },
    {
      "epoch": 0.3284110305516842,
      "grad_norm": 0.8259332394883399,
      "learning_rate": 7.840837455352872e-06,
      "loss": 0.4331,
      "step": 5243
    },
    {
      "epoch": 0.3284736685510265,
      "grad_norm": 0.9162984362838499,
      "learning_rate": 7.840002635112233e-06,
      "loss": 0.422,
      "step": 5244
    },
    {
      "epoch": 0.3285363065503688,
      "grad_norm": 0.8191400188471161,
      "learning_rate": 7.839167697976559e-06,
      "loss": 0.3916,
      "step": 5245
    },
    {
      "epoch": 0.32859894454971106,
      "grad_norm": 0.8870667257104597,
      "learning_rate": 7.838332643980215e-06,
      "loss": 0.3885,
      "step": 5246
    },
    {
      "epoch": 0.32866158254905337,
      "grad_norm": 0.8167830827298292,
      "learning_rate": 7.837497473157574e-06,
      "loss": 0.3816,
      "step": 5247
    },
    {
      "epoch": 0.3287242205483957,
      "grad_norm": 0.9421881461466497,
      "learning_rate": 7.83666218554301e-06,
      "loss": 0.424,
      "step": 5248
    },
    {
      "epoch": 0.328786858547738,
      "grad_norm": 0.8422240636836577,
      "learning_rate": 7.835826781170903e-06,
      "loss": 0.4399,
      "step": 5249
    },
    {
      "epoch": 0.3288494965470803,
      "grad_norm": 0.8394548976321369,
      "learning_rate": 7.834991260075638e-06,
      "loss": 0.3862,
      "step": 5250
    },
    {
      "epoch": 0.3289121345464226,
      "grad_norm": 0.8895964809934912,
      "learning_rate": 7.834155622291609e-06,
      "loss": 0.4098,
      "step": 5251
    },
    {
      "epoch": 0.3289747725457649,
      "grad_norm": 0.92602263488399,
      "learning_rate": 7.833319867853207e-06,
      "loss": 0.4195,
      "step": 5252
    },
    {
      "epoch": 0.3290374105451072,
      "grad_norm": 0.8736709189905953,
      "learning_rate": 7.832483996794834e-06,
      "loss": 0.4272,
      "step": 5253
    },
    {
      "epoch": 0.32910004854444946,
      "grad_norm": 0.8361452617077532,
      "learning_rate": 7.831648009150894e-06,
      "loss": 0.4092,
      "step": 5254
    },
    {
      "epoch": 0.32916268654379177,
      "grad_norm": 0.8525898991933369,
      "learning_rate": 7.830811904955796e-06,
      "loss": 0.4637,
      "step": 5255
    },
    {
      "epoch": 0.3292253245431341,
      "grad_norm": 0.8860817477998135,
      "learning_rate": 7.829975684243956e-06,
      "loss": 0.4122,
      "step": 5256
    },
    {
      "epoch": 0.3292879625424764,
      "grad_norm": 0.8616325227522224,
      "learning_rate": 7.82913934704979e-06,
      "loss": 0.4323,
      "step": 5257
    },
    {
      "epoch": 0.3293506005418187,
      "grad_norm": 0.8444483587443595,
      "learning_rate": 7.828302893407725e-06,
      "loss": 0.4067,
      "step": 5258
    },
    {
      "epoch": 0.329413238541161,
      "grad_norm": 0.8130613271146194,
      "learning_rate": 7.827466323352186e-06,
      "loss": 0.4111,
      "step": 5259
    },
    {
      "epoch": 0.3294758765405033,
      "grad_norm": 0.8074452274458871,
      "learning_rate": 7.826629636917609e-06,
      "loss": 0.4504,
      "step": 5260
    },
    {
      "epoch": 0.3295385145398456,
      "grad_norm": 0.8113267849793148,
      "learning_rate": 7.825792834138431e-06,
      "loss": 0.4276,
      "step": 5261
    },
    {
      "epoch": 0.3296011525391879,
      "grad_norm": 0.8555183193713496,
      "learning_rate": 7.824955915049096e-06,
      "loss": 0.4394,
      "step": 5262
    },
    {
      "epoch": 0.3296637905385302,
      "grad_norm": 0.927641215953857,
      "learning_rate": 7.824118879684053e-06,
      "loss": 0.4541,
      "step": 5263
    },
    {
      "epoch": 0.3297264285378725,
      "grad_norm": 0.8806460333548842,
      "learning_rate": 7.823281728077749e-06,
      "loss": 0.4341,
      "step": 5264
    },
    {
      "epoch": 0.3297890665372148,
      "grad_norm": 0.7794235266674235,
      "learning_rate": 7.822444460264648e-06,
      "loss": 0.3934,
      "step": 5265
    },
    {
      "epoch": 0.3298517045365571,
      "grad_norm": 0.8242736527052328,
      "learning_rate": 7.821607076279207e-06,
      "loss": 0.4231,
      "step": 5266
    },
    {
      "epoch": 0.3299143425358994,
      "grad_norm": 0.86387680885079,
      "learning_rate": 7.820769576155898e-06,
      "loss": 0.4233,
      "step": 5267
    },
    {
      "epoch": 0.3299769805352417,
      "grad_norm": 0.8016958874656288,
      "learning_rate": 7.819931959929187e-06,
      "loss": 0.3869,
      "step": 5268
    },
    {
      "epoch": 0.330039618534584,
      "grad_norm": 0.7776368769168822,
      "learning_rate": 7.819094227633553e-06,
      "loss": 0.4356,
      "step": 5269
    },
    {
      "epoch": 0.33010225653392633,
      "grad_norm": 0.8628280989468848,
      "learning_rate": 7.818256379303479e-06,
      "loss": 0.407,
      "step": 5270
    },
    {
      "epoch": 0.3301648945332686,
      "grad_norm": 0.8644638646345032,
      "learning_rate": 7.81741841497345e-06,
      "loss": 0.4283,
      "step": 5271
    },
    {
      "epoch": 0.3302275325326109,
      "grad_norm": 0.9055212450755559,
      "learning_rate": 7.816580334677955e-06,
      "loss": 0.4747,
      "step": 5272
    },
    {
      "epoch": 0.3302901705319532,
      "grad_norm": 0.8391024644217355,
      "learning_rate": 7.81574213845149e-06,
      "loss": 0.3875,
      "step": 5273
    },
    {
      "epoch": 0.3303528085312955,
      "grad_norm": 0.8461718638131563,
      "learning_rate": 7.814903826328555e-06,
      "loss": 0.4207,
      "step": 5274
    },
    {
      "epoch": 0.3304154465306378,
      "grad_norm": 0.763033826979776,
      "learning_rate": 7.814065398343657e-06,
      "loss": 0.4142,
      "step": 5275
    },
    {
      "epoch": 0.3304780845299801,
      "grad_norm": 0.8865711837599028,
      "learning_rate": 7.813226854531306e-06,
      "loss": 0.4338,
      "step": 5276
    },
    {
      "epoch": 0.33054072252932243,
      "grad_norm": 0.7621239729471987,
      "learning_rate": 7.812388194926015e-06,
      "loss": 0.3697,
      "step": 5277
    },
    {
      "epoch": 0.33060336052866474,
      "grad_norm": 0.8324375980494948,
      "learning_rate": 7.811549419562304e-06,
      "loss": 0.4201,
      "step": 5278
    },
    {
      "epoch": 0.330665998528007,
      "grad_norm": 0.8962010197349815,
      "learning_rate": 7.810710528474696e-06,
      "loss": 0.4563,
      "step": 5279
    },
    {
      "epoch": 0.3307286365273493,
      "grad_norm": 0.8871361168008738,
      "learning_rate": 7.809871521697723e-06,
      "loss": 0.4587,
      "step": 5280
    },
    {
      "epoch": 0.3307912745266916,
      "grad_norm": 0.8585204397023767,
      "learning_rate": 7.809032399265914e-06,
      "loss": 0.4238,
      "step": 5281
    },
    {
      "epoch": 0.3308539125260339,
      "grad_norm": 0.7616702948915002,
      "learning_rate": 7.808193161213812e-06,
      "loss": 0.3786,
      "step": 5282
    },
    {
      "epoch": 0.3309165505253762,
      "grad_norm": 0.8495316515687915,
      "learning_rate": 7.807353807575958e-06,
      "loss": 0.402,
      "step": 5283
    },
    {
      "epoch": 0.3309791885247185,
      "grad_norm": 0.8527649062176175,
      "learning_rate": 7.8065143383869e-06,
      "loss": 0.4639,
      "step": 5284
    },
    {
      "epoch": 0.33104182652406083,
      "grad_norm": 0.769312448297072,
      "learning_rate": 7.805674753681191e-06,
      "loss": 0.4084,
      "step": 5285
    },
    {
      "epoch": 0.33110446452340314,
      "grad_norm": 0.7538009481780182,
      "learning_rate": 7.804835053493388e-06,
      "loss": 0.4117,
      "step": 5286
    },
    {
      "epoch": 0.33116710252274545,
      "grad_norm": 0.8988491710738418,
      "learning_rate": 7.803995237858053e-06,
      "loss": 0.4394,
      "step": 5287
    },
    {
      "epoch": 0.3312297405220877,
      "grad_norm": 0.8289098134535405,
      "learning_rate": 7.803155306809755e-06,
      "loss": 0.4003,
      "step": 5288
    },
    {
      "epoch": 0.33129237852143,
      "grad_norm": 0.9571522899742656,
      "learning_rate": 7.802315260383064e-06,
      "loss": 0.4295,
      "step": 5289
    },
    {
      "epoch": 0.3313550165207723,
      "grad_norm": 0.8463563591999669,
      "learning_rate": 7.801475098612556e-06,
      "loss": 0.4797,
      "step": 5290
    },
    {
      "epoch": 0.3314176545201146,
      "grad_norm": 0.8792863509379935,
      "learning_rate": 7.800634821532813e-06,
      "loss": 0.4396,
      "step": 5291
    },
    {
      "epoch": 0.33148029251945693,
      "grad_norm": 0.8055661368951407,
      "learning_rate": 7.799794429178423e-06,
      "loss": 0.3837,
      "step": 5292
    },
    {
      "epoch": 0.33154293051879924,
      "grad_norm": 0.8739990114371848,
      "learning_rate": 7.798953921583972e-06,
      "loss": 0.4312,
      "step": 5293
    },
    {
      "epoch": 0.33160556851814155,
      "grad_norm": 0.8191310282840503,
      "learning_rate": 7.798113298784058e-06,
      "loss": 0.4054,
      "step": 5294
    },
    {
      "epoch": 0.33166820651748385,
      "grad_norm": 0.861654123807044,
      "learning_rate": 7.797272560813283e-06,
      "loss": 0.4137,
      "step": 5295
    },
    {
      "epoch": 0.3317308445168261,
      "grad_norm": 0.8580202362167622,
      "learning_rate": 7.79643170770625e-06,
      "loss": 0.4059,
      "step": 5296
    },
    {
      "epoch": 0.3317934825161684,
      "grad_norm": 0.8281764288626364,
      "learning_rate": 7.795590739497567e-06,
      "loss": 0.4031,
      "step": 5297
    },
    {
      "epoch": 0.3318561205155107,
      "grad_norm": 0.7595449868474237,
      "learning_rate": 7.794749656221851e-06,
      "loss": 0.4959,
      "step": 5298
    },
    {
      "epoch": 0.33191875851485303,
      "grad_norm": 0.7826860116540257,
      "learning_rate": 7.793908457913723e-06,
      "loss": 0.3928,
      "step": 5299
    },
    {
      "epoch": 0.33198139651419534,
      "grad_norm": 0.8187036041967871,
      "learning_rate": 7.793067144607802e-06,
      "loss": 0.4572,
      "step": 5300
    },
    {
      "epoch": 0.33204403451353764,
      "grad_norm": 0.8150982500977352,
      "learning_rate": 7.792225716338718e-06,
      "loss": 0.4371,
      "step": 5301
    },
    {
      "epoch": 0.33210667251287995,
      "grad_norm": 0.893897737829077,
      "learning_rate": 7.791384173141108e-06,
      "loss": 0.4272,
      "step": 5302
    },
    {
      "epoch": 0.33216931051222226,
      "grad_norm": 0.8327933101313686,
      "learning_rate": 7.790542515049604e-06,
      "loss": 0.4,
      "step": 5303
    },
    {
      "epoch": 0.33223194851156457,
      "grad_norm": 0.8605401178206988,
      "learning_rate": 7.789700742098856e-06,
      "loss": 0.4006,
      "step": 5304
    },
    {
      "epoch": 0.3322945865109068,
      "grad_norm": 0.8665384179168405,
      "learning_rate": 7.788858854323505e-06,
      "loss": 0.4005,
      "step": 5305
    },
    {
      "epoch": 0.3323572245102491,
      "grad_norm": 0.8936708633527918,
      "learning_rate": 7.788016851758206e-06,
      "loss": 0.4212,
      "step": 5306
    },
    {
      "epoch": 0.33241986250959144,
      "grad_norm": 0.7739345529507033,
      "learning_rate": 7.787174734437618e-06,
      "loss": 0.4078,
      "step": 5307
    },
    {
      "epoch": 0.33248250050893374,
      "grad_norm": 0.8070583889376994,
      "learning_rate": 7.786332502396398e-06,
      "loss": 0.4216,
      "step": 5308
    },
    {
      "epoch": 0.33254513850827605,
      "grad_norm": 0.8543673134055818,
      "learning_rate": 7.785490155669216e-06,
      "loss": 0.4422,
      "step": 5309
    },
    {
      "epoch": 0.33260777650761836,
      "grad_norm": 0.838114741121054,
      "learning_rate": 7.784647694290742e-06,
      "loss": 0.3874,
      "step": 5310
    },
    {
      "epoch": 0.33267041450696067,
      "grad_norm": 0.8628592033848987,
      "learning_rate": 7.783805118295653e-06,
      "loss": 0.3802,
      "step": 5311
    },
    {
      "epoch": 0.332733052506303,
      "grad_norm": 0.7910875789324188,
      "learning_rate": 7.782962427718627e-06,
      "loss": 0.4235,
      "step": 5312
    },
    {
      "epoch": 0.3327956905056452,
      "grad_norm": 0.8430032327645138,
      "learning_rate": 7.782119622594353e-06,
      "loss": 0.4078,
      "step": 5313
    },
    {
      "epoch": 0.33285832850498753,
      "grad_norm": 0.8884438862316457,
      "learning_rate": 7.781276702957517e-06,
      "loss": 0.4314,
      "step": 5314
    },
    {
      "epoch": 0.33292096650432984,
      "grad_norm": 0.7863190680077679,
      "learning_rate": 7.780433668842815e-06,
      "loss": 0.4239,
      "step": 5315
    },
    {
      "epoch": 0.33298360450367215,
      "grad_norm": 0.8426529313027998,
      "learning_rate": 7.77959052028495e-06,
      "loss": 0.4264,
      "step": 5316
    },
    {
      "epoch": 0.33304624250301446,
      "grad_norm": 0.8353103629380714,
      "learning_rate": 7.778747257318622e-06,
      "loss": 0.4389,
      "step": 5317
    },
    {
      "epoch": 0.33310888050235676,
      "grad_norm": 0.7508009760810682,
      "learning_rate": 7.777903879978541e-06,
      "loss": 0.416,
      "step": 5318
    },
    {
      "epoch": 0.33317151850169907,
      "grad_norm": 0.812327811966331,
      "learning_rate": 7.77706038829942e-06,
      "loss": 0.4009,
      "step": 5319
    },
    {
      "epoch": 0.3332341565010414,
      "grad_norm": 0.8216174104043901,
      "learning_rate": 7.77621678231598e-06,
      "loss": 0.433,
      "step": 5320
    },
    {
      "epoch": 0.33329679450038363,
      "grad_norm": 0.8725715884364085,
      "learning_rate": 7.775373062062939e-06,
      "loss": 0.4421,
      "step": 5321
    },
    {
      "epoch": 0.33335943249972594,
      "grad_norm": 0.7209670584364476,
      "learning_rate": 7.77452922757503e-06,
      "loss": 0.4663,
      "step": 5322
    },
    {
      "epoch": 0.33342207049906825,
      "grad_norm": 0.863631851725819,
      "learning_rate": 7.773685278886982e-06,
      "loss": 0.476,
      "step": 5323
    },
    {
      "epoch": 0.33348470849841055,
      "grad_norm": 0.8111508226288096,
      "learning_rate": 7.772841216033534e-06,
      "loss": 0.4366,
      "step": 5324
    },
    {
      "epoch": 0.33354734649775286,
      "grad_norm": 0.7873849714248582,
      "learning_rate": 7.771997039049426e-06,
      "loss": 0.432,
      "step": 5325
    },
    {
      "epoch": 0.33360998449709517,
      "grad_norm": 0.8528204610578393,
      "learning_rate": 7.771152747969406e-06,
      "loss": 0.4264,
      "step": 5326
    },
    {
      "epoch": 0.3336726224964375,
      "grad_norm": 0.8318885558307701,
      "learning_rate": 7.770308342828223e-06,
      "loss": 0.3989,
      "step": 5327
    },
    {
      "epoch": 0.3337352604957798,
      "grad_norm": 0.7386677513580496,
      "learning_rate": 7.769463823660636e-06,
      "loss": 0.4821,
      "step": 5328
    },
    {
      "epoch": 0.3337978984951221,
      "grad_norm": 0.8039080529307877,
      "learning_rate": 7.768619190501404e-06,
      "loss": 0.4294,
      "step": 5329
    },
    {
      "epoch": 0.33386053649446434,
      "grad_norm": 0.8546931182843677,
      "learning_rate": 7.767774443385291e-06,
      "loss": 0.4457,
      "step": 5330
    },
    {
      "epoch": 0.33392317449380665,
      "grad_norm": 0.8295138059319225,
      "learning_rate": 7.766929582347068e-06,
      "loss": 0.4368,
      "step": 5331
    },
    {
      "epoch": 0.33398581249314896,
      "grad_norm": 0.996781871412655,
      "learning_rate": 7.766084607421512e-06,
      "loss": 0.4609,
      "step": 5332
    },
    {
      "epoch": 0.33404845049249127,
      "grad_norm": 0.8058142325110019,
      "learning_rate": 7.7652395186434e-06,
      "loss": 0.3921,
      "step": 5333
    },
    {
      "epoch": 0.3341110884918336,
      "grad_norm": 0.8366858900547761,
      "learning_rate": 7.764394316047515e-06,
      "loss": 0.4339,
      "step": 5334
    },
    {
      "epoch": 0.3341737264911759,
      "grad_norm": 0.8666057801764222,
      "learning_rate": 7.763548999668646e-06,
      "loss": 0.429,
      "step": 5335
    },
    {
      "epoch": 0.3342363644905182,
      "grad_norm": 0.7599205422262141,
      "learning_rate": 7.762703569541587e-06,
      "loss": 0.4052,
      "step": 5336
    },
    {
      "epoch": 0.3342990024898605,
      "grad_norm": 0.8589746486096054,
      "learning_rate": 7.761858025701139e-06,
      "loss": 0.4132,
      "step": 5337
    },
    {
      "epoch": 0.33436164048920275,
      "grad_norm": 0.9204255307534548,
      "learning_rate": 7.7610123681821e-06,
      "loss": 0.4253,
      "step": 5338
    },
    {
      "epoch": 0.33442427848854506,
      "grad_norm": 0.8357141245211003,
      "learning_rate": 7.760166597019282e-06,
      "loss": 0.4169,
      "step": 5339
    },
    {
      "epoch": 0.33448691648788736,
      "grad_norm": 0.8701085531681148,
      "learning_rate": 7.759320712247491e-06,
      "loss": 0.4594,
      "step": 5340
    },
    {
      "epoch": 0.3345495544872297,
      "grad_norm": 0.8013979997969015,
      "learning_rate": 7.75847471390155e-06,
      "loss": 0.3932,
      "step": 5341
    },
    {
      "epoch": 0.334612192486572,
      "grad_norm": 0.9185204954221268,
      "learning_rate": 7.757628602016277e-06,
      "loss": 0.4233,
      "step": 5342
    },
    {
      "epoch": 0.3346748304859143,
      "grad_norm": 0.8422496253555444,
      "learning_rate": 7.756782376626497e-06,
      "loss": 0.4598,
      "step": 5343
    },
    {
      "epoch": 0.3347374684852566,
      "grad_norm": 0.9269775437795906,
      "learning_rate": 7.755936037767044e-06,
      "loss": 0.4497,
      "step": 5344
    },
    {
      "epoch": 0.3348001064845989,
      "grad_norm": 0.8311804589628298,
      "learning_rate": 7.755089585472752e-06,
      "loss": 0.383,
      "step": 5345
    },
    {
      "epoch": 0.3348627444839412,
      "grad_norm": 0.8215625405449356,
      "learning_rate": 7.754243019778462e-06,
      "loss": 0.4247,
      "step": 5346
    },
    {
      "epoch": 0.33492538248328346,
      "grad_norm": 0.873192303209613,
      "learning_rate": 7.753396340719017e-06,
      "loss": 0.4476,
      "step": 5347
    },
    {
      "epoch": 0.33498802048262577,
      "grad_norm": 0.7820572346237162,
      "learning_rate": 7.752549548329268e-06,
      "loss": 0.4112,
      "step": 5348
    },
    {
      "epoch": 0.3350506584819681,
      "grad_norm": 0.8298898234825853,
      "learning_rate": 7.75170264264407e-06,
      "loss": 0.4263,
      "step": 5349
    },
    {
      "epoch": 0.3351132964813104,
      "grad_norm": 0.8272622648040023,
      "learning_rate": 7.75085562369828e-06,
      "loss": 0.4274,
      "step": 5350
    },
    {
      "epoch": 0.3351759344806527,
      "grad_norm": 0.7912289446521692,
      "learning_rate": 7.750008491526762e-06,
      "loss": 0.4194,
      "step": 5351
    },
    {
      "epoch": 0.335238572479995,
      "grad_norm": 0.8439861596822192,
      "learning_rate": 7.749161246164383e-06,
      "loss": 0.4234,
      "step": 5352
    },
    {
      "epoch": 0.3353012104793373,
      "grad_norm": 0.8272855743884115,
      "learning_rate": 7.748313887646018e-06,
      "loss": 0.3932,
      "step": 5353
    },
    {
      "epoch": 0.3353638484786796,
      "grad_norm": 0.750605646629579,
      "learning_rate": 7.747466416006541e-06,
      "loss": 0.3673,
      "step": 5354
    },
    {
      "epoch": 0.33542648647802187,
      "grad_norm": 0.840677996618983,
      "learning_rate": 7.746618831280839e-06,
      "loss": 0.4162,
      "step": 5355
    },
    {
      "epoch": 0.3354891244773642,
      "grad_norm": 0.7693757115495204,
      "learning_rate": 7.745771133503796e-06,
      "loss": 0.3932,
      "step": 5356
    },
    {
      "epoch": 0.3355517624767065,
      "grad_norm": 0.8413944710892881,
      "learning_rate": 7.744923322710302e-06,
      "loss": 0.4168,
      "step": 5357
    },
    {
      "epoch": 0.3356144004760488,
      "grad_norm": 0.8703654428222514,
      "learning_rate": 7.744075398935257e-06,
      "loss": 0.3986,
      "step": 5358
    },
    {
      "epoch": 0.3356770384753911,
      "grad_norm": 0.7763988385180106,
      "learning_rate": 7.743227362213556e-06,
      "loss": 0.3932,
      "step": 5359
    },
    {
      "epoch": 0.3357396764747334,
      "grad_norm": 0.8713063939739846,
      "learning_rate": 7.74237921258011e-06,
      "loss": 0.4052,
      "step": 5360
    },
    {
      "epoch": 0.3358023144740757,
      "grad_norm": 0.8838498897382357,
      "learning_rate": 7.741530950069824e-06,
      "loss": 0.4132,
      "step": 5361
    },
    {
      "epoch": 0.335864952473418,
      "grad_norm": 0.7566030573779345,
      "learning_rate": 7.740682574717617e-06,
      "loss": 0.4933,
      "step": 5362
    },
    {
      "epoch": 0.3359275904727603,
      "grad_norm": 0.8252001977795974,
      "learning_rate": 7.739834086558407e-06,
      "loss": 0.4376,
      "step": 5363
    },
    {
      "epoch": 0.3359902284721026,
      "grad_norm": 0.8012953990561389,
      "learning_rate": 7.738985485627117e-06,
      "loss": 0.4341,
      "step": 5364
    },
    {
      "epoch": 0.3360528664714449,
      "grad_norm": 0.7942360756966773,
      "learning_rate": 7.738136771958675e-06,
      "loss": 0.3668,
      "step": 5365
    },
    {
      "epoch": 0.3361155044707872,
      "grad_norm": 0.7839479083702037,
      "learning_rate": 7.737287945588016e-06,
      "loss": 0.3984,
      "step": 5366
    },
    {
      "epoch": 0.3361781424701295,
      "grad_norm": 0.7310883005015003,
      "learning_rate": 7.736439006550078e-06,
      "loss": 0.392,
      "step": 5367
    },
    {
      "epoch": 0.3362407804694718,
      "grad_norm": 0.7557786072236349,
      "learning_rate": 7.7355899548798e-06,
      "loss": 0.3795,
      "step": 5368
    },
    {
      "epoch": 0.3363034184688141,
      "grad_norm": 0.8895310876558375,
      "learning_rate": 7.734740790612137e-06,
      "loss": 0.4288,
      "step": 5369
    },
    {
      "epoch": 0.3363660564681564,
      "grad_norm": 0.7722326562272243,
      "learning_rate": 7.73389151378203e-06,
      "loss": 0.4155,
      "step": 5370
    },
    {
      "epoch": 0.33642869446749873,
      "grad_norm": 0.7178149559722218,
      "learning_rate": 7.733042124424444e-06,
      "loss": 0.4822,
      "step": 5371
    },
    {
      "epoch": 0.336491332466841,
      "grad_norm": 0.7726871430457412,
      "learning_rate": 7.732192622574334e-06,
      "loss": 0.4196,
      "step": 5372
    },
    {
      "epoch": 0.3365539704661833,
      "grad_norm": 0.8835476673888274,
      "learning_rate": 7.731343008266672e-06,
      "loss": 0.4339,
      "step": 5373
    },
    {
      "epoch": 0.3366166084655256,
      "grad_norm": 0.7721786957780604,
      "learning_rate": 7.730493281536422e-06,
      "loss": 0.4383,
      "step": 5374
    },
    {
      "epoch": 0.3366792464648679,
      "grad_norm": 0.8351599877347983,
      "learning_rate": 7.729643442418563e-06,
      "loss": 0.4104,
      "step": 5375
    },
    {
      "epoch": 0.3367418844642102,
      "grad_norm": 0.7936414691642906,
      "learning_rate": 7.728793490948074e-06,
      "loss": 0.4062,
      "step": 5376
    },
    {
      "epoch": 0.3368045224635525,
      "grad_norm": 0.6347564420599564,
      "learning_rate": 7.727943427159937e-06,
      "loss": 0.4488,
      "step": 5377
    },
    {
      "epoch": 0.33686716046289483,
      "grad_norm": 0.8101896293222082,
      "learning_rate": 7.727093251089143e-06,
      "loss": 0.4191,
      "step": 5378
    },
    {
      "epoch": 0.33692979846223714,
      "grad_norm": 0.8137598780747437,
      "learning_rate": 7.726242962770684e-06,
      "loss": 0.401,
      "step": 5379
    },
    {
      "epoch": 0.3369924364615794,
      "grad_norm": 0.8627567676550569,
      "learning_rate": 7.72539256223956e-06,
      "loss": 0.4271,
      "step": 5380
    },
    {
      "epoch": 0.3370550744609217,
      "grad_norm": 0.7277141961963569,
      "learning_rate": 7.72454204953077e-06,
      "loss": 0.3989,
      "step": 5381
    },
    {
      "epoch": 0.337117712460264,
      "grad_norm": 0.8962053192182641,
      "learning_rate": 7.723691424679325e-06,
      "loss": 0.4263,
      "step": 5382
    },
    {
      "epoch": 0.3371803504596063,
      "grad_norm": 0.7953172417322293,
      "learning_rate": 7.722840687720234e-06,
      "loss": 0.44,
      "step": 5383
    },
    {
      "epoch": 0.3372429884589486,
      "grad_norm": 0.8559091772292255,
      "learning_rate": 7.721989838688517e-06,
      "loss": 0.398,
      "step": 5384
    },
    {
      "epoch": 0.33730562645829093,
      "grad_norm": 0.80560444089184,
      "learning_rate": 7.721138877619192e-06,
      "loss": 0.4378,
      "step": 5385
    },
    {
      "epoch": 0.33736826445763324,
      "grad_norm": 0.8509758370680516,
      "learning_rate": 7.720287804547286e-06,
      "loss": 0.4252,
      "step": 5386
    },
    {
      "epoch": 0.33743090245697555,
      "grad_norm": 0.8031309697222552,
      "learning_rate": 7.71943661950783e-06,
      "loss": 0.4325,
      "step": 5387
    },
    {
      "epoch": 0.3374935404563178,
      "grad_norm": 0.8071405914211807,
      "learning_rate": 7.718585322535857e-06,
      "loss": 0.3971,
      "step": 5388
    },
    {
      "epoch": 0.3375561784556601,
      "grad_norm": 0.8274765846928236,
      "learning_rate": 7.717733913666406e-06,
      "loss": 0.4197,
      "step": 5389
    },
    {
      "epoch": 0.3376188164550024,
      "grad_norm": 0.8473581624347716,
      "learning_rate": 7.716882392934523e-06,
      "loss": 0.4348,
      "step": 5390
    },
    {
      "epoch": 0.3376814544543447,
      "grad_norm": 0.866765378536873,
      "learning_rate": 7.716030760375259e-06,
      "loss": 0.4377,
      "step": 5391
    },
    {
      "epoch": 0.33774409245368703,
      "grad_norm": 0.7396856945315529,
      "learning_rate": 7.715179016023664e-06,
      "loss": 0.4831,
      "step": 5392
    },
    {
      "epoch": 0.33780673045302934,
      "grad_norm": 0.7530163294256795,
      "learning_rate": 7.714327159914796e-06,
      "loss": 0.3678,
      "step": 5393
    },
    {
      "epoch": 0.33786936845237164,
      "grad_norm": 0.8093598563265546,
      "learning_rate": 7.713475192083716e-06,
      "loss": 0.3971,
      "step": 5394
    },
    {
      "epoch": 0.33793200645171395,
      "grad_norm": 0.8664218204340495,
      "learning_rate": 7.712623112565497e-06,
      "loss": 0.4229,
      "step": 5395
    },
    {
      "epoch": 0.33799464445105626,
      "grad_norm": 0.8849959730857683,
      "learning_rate": 7.711770921395207e-06,
      "loss": 0.4299,
      "step": 5396
    },
    {
      "epoch": 0.3380572824503985,
      "grad_norm": 0.7701481615794841,
      "learning_rate": 7.710918618607923e-06,
      "loss": 0.3774,
      "step": 5397
    },
    {
      "epoch": 0.3381199204497408,
      "grad_norm": 0.8253163100283495,
      "learning_rate": 7.710066204238722e-06,
      "loss": 0.4171,
      "step": 5398
    },
    {
      "epoch": 0.3381825584490831,
      "grad_norm": 0.6296538860665624,
      "learning_rate": 7.709213678322696e-06,
      "loss": 0.4837,
      "step": 5399
    },
    {
      "epoch": 0.33824519644842543,
      "grad_norm": 0.8881630262771957,
      "learning_rate": 7.70836104089493e-06,
      "loss": 0.4115,
      "step": 5400
    },
    {
      "epoch": 0.33830783444776774,
      "grad_norm": 0.8368362996902028,
      "learning_rate": 7.707508291990523e-06,
      "loss": 0.4092,
      "step": 5401
    },
    {
      "epoch": 0.33837047244711005,
      "grad_norm": 0.8110209648497023,
      "learning_rate": 7.706655431644572e-06,
      "loss": 0.4492,
      "step": 5402
    },
    {
      "epoch": 0.33843311044645236,
      "grad_norm": 0.7931749371267746,
      "learning_rate": 7.70580245989218e-06,
      "loss": 0.408,
      "step": 5403
    },
    {
      "epoch": 0.33849574844579466,
      "grad_norm": 0.7510733242373532,
      "learning_rate": 7.704949376768456e-06,
      "loss": 0.4143,
      "step": 5404
    },
    {
      "epoch": 0.3385583864451369,
      "grad_norm": 0.8361386694068532,
      "learning_rate": 7.704096182308516e-06,
      "loss": 0.4078,
      "step": 5405
    },
    {
      "epoch": 0.3386210244444792,
      "grad_norm": 0.704240115610989,
      "learning_rate": 7.703242876547472e-06,
      "loss": 0.4738,
      "step": 5406
    },
    {
      "epoch": 0.33868366244382153,
      "grad_norm": 0.8864449156342458,
      "learning_rate": 7.702389459520452e-06,
      "loss": 0.4431,
      "step": 5407
    },
    {
      "epoch": 0.33874630044316384,
      "grad_norm": 0.8734114142868725,
      "learning_rate": 7.70153593126258e-06,
      "loss": 0.4199,
      "step": 5408
    },
    {
      "epoch": 0.33880893844250615,
      "grad_norm": 0.6741652651562087,
      "learning_rate": 7.700682291808986e-06,
      "loss": 0.466,
      "step": 5409
    },
    {
      "epoch": 0.33887157644184845,
      "grad_norm": 0.8064459421097978,
      "learning_rate": 7.699828541194809e-06,
      "loss": 0.3904,
      "step": 5410
    },
    {
      "epoch": 0.33893421444119076,
      "grad_norm": 0.8033575621292249,
      "learning_rate": 7.698974679455188e-06,
      "loss": 0.3993,
      "step": 5411
    },
    {
      "epoch": 0.33899685244053307,
      "grad_norm": 0.856593553559791,
      "learning_rate": 7.698120706625267e-06,
      "loss": 0.4253,
      "step": 5412
    },
    {
      "epoch": 0.3390594904398754,
      "grad_norm": 0.8513432646374332,
      "learning_rate": 7.697266622740199e-06,
      "loss": 0.415,
      "step": 5413
    },
    {
      "epoch": 0.33912212843921763,
      "grad_norm": 0.7541088790086903,
      "learning_rate": 7.696412427835133e-06,
      "loss": 0.3906,
      "step": 5414
    },
    {
      "epoch": 0.33918476643855994,
      "grad_norm": 0.8699719283201349,
      "learning_rate": 7.695558121945234e-06,
      "loss": 0.4218,
      "step": 5415
    },
    {
      "epoch": 0.33924740443790224,
      "grad_norm": 0.8054283491627323,
      "learning_rate": 7.694703705105661e-06,
      "loss": 0.4202,
      "step": 5416
    },
    {
      "epoch": 0.33931004243724455,
      "grad_norm": 0.8427659711672542,
      "learning_rate": 7.693849177351584e-06,
      "loss": 0.4572,
      "step": 5417
    },
    {
      "epoch": 0.33937268043658686,
      "grad_norm": 0.9074793068072399,
      "learning_rate": 7.692994538718175e-06,
      "loss": 0.4323,
      "step": 5418
    },
    {
      "epoch": 0.33943531843592917,
      "grad_norm": 0.7980270586412737,
      "learning_rate": 7.692139789240611e-06,
      "loss": 0.4056,
      "step": 5419
    },
    {
      "epoch": 0.3394979564352715,
      "grad_norm": 0.7642496170362,
      "learning_rate": 7.691284928954074e-06,
      "loss": 0.4188,
      "step": 5420
    },
    {
      "epoch": 0.3395605944346138,
      "grad_norm": 0.7274164848148186,
      "learning_rate": 7.690429957893749e-06,
      "loss": 0.3303,
      "step": 5421
    },
    {
      "epoch": 0.33962323243395603,
      "grad_norm": 0.8278760259188338,
      "learning_rate": 7.689574876094828e-06,
      "loss": 0.4175,
      "step": 5422
    },
    {
      "epoch": 0.33968587043329834,
      "grad_norm": 0.817262839122317,
      "learning_rate": 7.688719683592506e-06,
      "loss": 0.4198,
      "step": 5423
    },
    {
      "epoch": 0.33974850843264065,
      "grad_norm": 0.8606501412547473,
      "learning_rate": 7.687864380421981e-06,
      "loss": 0.3911,
      "step": 5424
    },
    {
      "epoch": 0.33981114643198296,
      "grad_norm": 0.8032556799355705,
      "learning_rate": 7.687008966618461e-06,
      "loss": 0.3852,
      "step": 5425
    },
    {
      "epoch": 0.33987378443132527,
      "grad_norm": 0.8501740259165823,
      "learning_rate": 7.686153442217153e-06,
      "loss": 0.4372,
      "step": 5426
    },
    {
      "epoch": 0.3399364224306676,
      "grad_norm": 0.7837413619232704,
      "learning_rate": 7.685297807253269e-06,
      "loss": 0.4112,
      "step": 5427
    },
    {
      "epoch": 0.3399990604300099,
      "grad_norm": 0.8368342940126086,
      "learning_rate": 7.68444206176203e-06,
      "loss": 0.3876,
      "step": 5428
    },
    {
      "epoch": 0.3400616984293522,
      "grad_norm": 0.888831560856548,
      "learning_rate": 7.683586205778657e-06,
      "loss": 0.446,
      "step": 5429
    },
    {
      "epoch": 0.34012433642869444,
      "grad_norm": 0.8278191884546064,
      "learning_rate": 7.682730239338379e-06,
      "loss": 0.4515,
      "step": 5430
    },
    {
      "epoch": 0.34018697442803675,
      "grad_norm": 0.8134627629120762,
      "learning_rate": 7.681874162476426e-06,
      "loss": 0.3939,
      "step": 5431
    },
    {
      "epoch": 0.34024961242737906,
      "grad_norm": 0.8858747371000765,
      "learning_rate": 7.681017975228034e-06,
      "loss": 0.3922,
      "step": 5432
    },
    {
      "epoch": 0.34031225042672136,
      "grad_norm": 0.8122411039539352,
      "learning_rate": 7.680161677628446e-06,
      "loss": 0.4149,
      "step": 5433
    },
    {
      "epoch": 0.34037488842606367,
      "grad_norm": 0.7932695164963309,
      "learning_rate": 7.679305269712905e-06,
      "loss": 0.4349,
      "step": 5434
    },
    {
      "epoch": 0.340437526425406,
      "grad_norm": 0.8717485829481567,
      "learning_rate": 7.67844875151666e-06,
      "loss": 0.3959,
      "step": 5435
    },
    {
      "epoch": 0.3405001644247483,
      "grad_norm": 0.9437319489876956,
      "learning_rate": 7.677592123074969e-06,
      "loss": 0.4356,
      "step": 5436
    },
    {
      "epoch": 0.3405628024240906,
      "grad_norm": 0.8455462713178707,
      "learning_rate": 7.67673538442309e-06,
      "loss": 0.4246,
      "step": 5437
    },
    {
      "epoch": 0.3406254404234329,
      "grad_norm": 0.7489378434680304,
      "learning_rate": 7.675878535596287e-06,
      "loss": 0.4359,
      "step": 5438
    },
    {
      "epoch": 0.34068807842277515,
      "grad_norm": 0.8183158579017404,
      "learning_rate": 7.675021576629823e-06,
      "loss": 0.426,
      "step": 5439
    },
    {
      "epoch": 0.34075071642211746,
      "grad_norm": 0.6704290690162501,
      "learning_rate": 7.674164507558976e-06,
      "loss": 0.4696,
      "step": 5440
    },
    {
      "epoch": 0.34081335442145977,
      "grad_norm": 0.8754645600382787,
      "learning_rate": 7.673307328419021e-06,
      "loss": 0.4373,
      "step": 5441
    },
    {
      "epoch": 0.3408759924208021,
      "grad_norm": 0.8766035617980605,
      "learning_rate": 7.672450039245242e-06,
      "loss": 0.4073,
      "step": 5442
    },
    {
      "epoch": 0.3409386304201444,
      "grad_norm": 0.8289728860485438,
      "learning_rate": 7.671592640072923e-06,
      "loss": 0.4359,
      "step": 5443
    },
    {
      "epoch": 0.3410012684194867,
      "grad_norm": 0.8507956644419364,
      "learning_rate": 7.670735130937353e-06,
      "loss": 0.3969,
      "step": 5444
    },
    {
      "epoch": 0.341063906418829,
      "grad_norm": 0.8307449293419378,
      "learning_rate": 7.66987751187383e-06,
      "loss": 0.4493,
      "step": 5445
    },
    {
      "epoch": 0.3411265444181713,
      "grad_norm": 0.7920721300451884,
      "learning_rate": 7.669019782917653e-06,
      "loss": 0.4179,
      "step": 5446
    },
    {
      "epoch": 0.34118918241751356,
      "grad_norm": 0.8817467291396222,
      "learning_rate": 7.668161944104128e-06,
      "loss": 0.4613,
      "step": 5447
    },
    {
      "epoch": 0.34125182041685587,
      "grad_norm": 0.7999269089128082,
      "learning_rate": 7.66730399546856e-06,
      "loss": 0.4002,
      "step": 5448
    },
    {
      "epoch": 0.3413144584161982,
      "grad_norm": 0.877492997403083,
      "learning_rate": 7.666445937046266e-06,
      "loss": 0.4302,
      "step": 5449
    },
    {
      "epoch": 0.3413770964155405,
      "grad_norm": 0.7830490076781468,
      "learning_rate": 7.665587768872563e-06,
      "loss": 0.3752,
      "step": 5450
    },
    {
      "epoch": 0.3414397344148828,
      "grad_norm": 0.8360849064660894,
      "learning_rate": 7.664729490982772e-06,
      "loss": 0.3696,
      "step": 5451
    },
    {
      "epoch": 0.3415023724142251,
      "grad_norm": 0.8296759111212884,
      "learning_rate": 7.66387110341222e-06,
      "loss": 0.3901,
      "step": 5452
    },
    {
      "epoch": 0.3415650104135674,
      "grad_norm": 0.8250101567956314,
      "learning_rate": 7.663012606196238e-06,
      "loss": 0.3882,
      "step": 5453
    },
    {
      "epoch": 0.3416276484129097,
      "grad_norm": 0.8391093624110277,
      "learning_rate": 7.662153999370163e-06,
      "loss": 0.3828,
      "step": 5454
    },
    {
      "epoch": 0.341690286412252,
      "grad_norm": 0.7554355232828602,
      "learning_rate": 7.661295282969337e-06,
      "loss": 0.3696,
      "step": 5455
    },
    {
      "epoch": 0.34175292441159427,
      "grad_norm": 0.7504994101078319,
      "learning_rate": 7.660436457029104e-06,
      "loss": 0.4056,
      "step": 5456
    },
    {
      "epoch": 0.3418155624109366,
      "grad_norm": 0.8432218153482814,
      "learning_rate": 7.659577521584809e-06,
      "loss": 0.4044,
      "step": 5457
    },
    {
      "epoch": 0.3418782004102789,
      "grad_norm": 0.741342001393949,
      "learning_rate": 7.658718476671812e-06,
      "loss": 0.37,
      "step": 5458
    },
    {
      "epoch": 0.3419408384096212,
      "grad_norm": 0.8164167662648025,
      "learning_rate": 7.657859322325469e-06,
      "loss": 0.4229,
      "step": 5459
    },
    {
      "epoch": 0.3420034764089635,
      "grad_norm": 0.810661074778895,
      "learning_rate": 7.657000058581142e-06,
      "loss": 0.357,
      "step": 5460
    },
    {
      "epoch": 0.3420661144083058,
      "grad_norm": 0.7905288263073328,
      "learning_rate": 7.6561406854742e-06,
      "loss": 0.4325,
      "step": 5461
    },
    {
      "epoch": 0.3421287524076481,
      "grad_norm": 0.8879638993479201,
      "learning_rate": 7.655281203040014e-06,
      "loss": 0.441,
      "step": 5462
    },
    {
      "epoch": 0.3421913904069904,
      "grad_norm": 0.8779210262969016,
      "learning_rate": 7.654421611313961e-06,
      "loss": 0.4251,
      "step": 5463
    },
    {
      "epoch": 0.3422540284063327,
      "grad_norm": 0.7372878745437778,
      "learning_rate": 7.653561910331422e-06,
      "loss": 0.3955,
      "step": 5464
    },
    {
      "epoch": 0.342316666405675,
      "grad_norm": 0.8486530425873063,
      "learning_rate": 7.652702100127783e-06,
      "loss": 0.4734,
      "step": 5465
    },
    {
      "epoch": 0.3423793044050173,
      "grad_norm": 0.7433722755031043,
      "learning_rate": 7.651842180738432e-06,
      "loss": 0.4787,
      "step": 5466
    },
    {
      "epoch": 0.3424419424043596,
      "grad_norm": 0.7934739196559898,
      "learning_rate": 7.650982152198765e-06,
      "loss": 0.3961,
      "step": 5467
    },
    {
      "epoch": 0.3425045804037019,
      "grad_norm": 0.6529942974274798,
      "learning_rate": 7.65012201454418e-06,
      "loss": 0.4826,
      "step": 5468
    },
    {
      "epoch": 0.3425672184030442,
      "grad_norm": 0.7889865969837013,
      "learning_rate": 7.649261767810082e-06,
      "loss": 0.394,
      "step": 5469
    },
    {
      "epoch": 0.3426298564023865,
      "grad_norm": 0.8521095070186899,
      "learning_rate": 7.648401412031878e-06,
      "loss": 0.4318,
      "step": 5470
    },
    {
      "epoch": 0.34269249440172883,
      "grad_norm": 0.7814160347159455,
      "learning_rate": 7.64754094724498e-06,
      "loss": 0.3888,
      "step": 5471
    },
    {
      "epoch": 0.3427551324010711,
      "grad_norm": 0.7709433633248635,
      "learning_rate": 7.646680373484805e-06,
      "loss": 0.3777,
      "step": 5472
    },
    {
      "epoch": 0.3428177704004134,
      "grad_norm": 0.8181645822122546,
      "learning_rate": 7.645819690786776e-06,
      "loss": 0.4002,
      "step": 5473
    },
    {
      "epoch": 0.3428804083997557,
      "grad_norm": 0.8829273762355285,
      "learning_rate": 7.644958899186315e-06,
      "loss": 0.4109,
      "step": 5474
    },
    {
      "epoch": 0.342943046399098,
      "grad_norm": 0.8087624966146467,
      "learning_rate": 7.644097998718856e-06,
      "loss": 0.3688,
      "step": 5475
    },
    {
      "epoch": 0.3430056843984403,
      "grad_norm": 0.8595896729376733,
      "learning_rate": 7.643236989419831e-06,
      "loss": 0.3825,
      "step": 5476
    },
    {
      "epoch": 0.3430683223977826,
      "grad_norm": 0.8238663874645211,
      "learning_rate": 7.642375871324683e-06,
      "loss": 0.434,
      "step": 5477
    },
    {
      "epoch": 0.34313096039712493,
      "grad_norm": 0.8534709575425896,
      "learning_rate": 7.641514644468853e-06,
      "loss": 0.4189,
      "step": 5478
    },
    {
      "epoch": 0.34319359839646724,
      "grad_norm": 0.8338664564570073,
      "learning_rate": 7.64065330888779e-06,
      "loss": 0.3679,
      "step": 5479
    },
    {
      "epoch": 0.34325623639580954,
      "grad_norm": 0.8636972684060491,
      "learning_rate": 7.639791864616948e-06,
      "loss": 0.4296,
      "step": 5480
    },
    {
      "epoch": 0.3433188743951518,
      "grad_norm": 0.8642460887203769,
      "learning_rate": 7.63893031169178e-06,
      "loss": 0.442,
      "step": 5481
    },
    {
      "epoch": 0.3433815123944941,
      "grad_norm": 0.9313440476611863,
      "learning_rate": 7.638068650147753e-06,
      "loss": 0.4355,
      "step": 5482
    },
    {
      "epoch": 0.3434441503938364,
      "grad_norm": 0.7500768060306824,
      "learning_rate": 7.637206880020329e-06,
      "loss": 0.4802,
      "step": 5483
    },
    {
      "epoch": 0.3435067883931787,
      "grad_norm": 0.7346250244470804,
      "learning_rate": 7.636345001344983e-06,
      "loss": 0.4839,
      "step": 5484
    },
    {
      "epoch": 0.343569426392521,
      "grad_norm": 0.832091825352677,
      "learning_rate": 7.635483014157183e-06,
      "loss": 0.3802,
      "step": 5485
    },
    {
      "epoch": 0.34363206439186333,
      "grad_norm": 0.8237562289061159,
      "learning_rate": 7.634620918492417e-06,
      "loss": 0.395,
      "step": 5486
    },
    {
      "epoch": 0.34369470239120564,
      "grad_norm": 0.9132640293540571,
      "learning_rate": 7.633758714386163e-06,
      "loss": 0.4379,
      "step": 5487
    },
    {
      "epoch": 0.34375734039054795,
      "grad_norm": 0.8427968935415014,
      "learning_rate": 7.632896401873913e-06,
      "loss": 0.399,
      "step": 5488
    },
    {
      "epoch": 0.3438199783898902,
      "grad_norm": 0.8849637883127536,
      "learning_rate": 7.632033980991156e-06,
      "loss": 0.3929,
      "step": 5489
    },
    {
      "epoch": 0.3438826163892325,
      "grad_norm": 0.836918971346629,
      "learning_rate": 7.631171451773394e-06,
      "loss": 0.43,
      "step": 5490
    },
    {
      "epoch": 0.3439452543885748,
      "grad_norm": 0.8059752468926502,
      "learning_rate": 7.630308814256124e-06,
      "loss": 0.4176,
      "step": 5491
    },
    {
      "epoch": 0.3440078923879171,
      "grad_norm": 0.8060587833367376,
      "learning_rate": 7.629446068474858e-06,
      "loss": 0.4018,
      "step": 5492
    },
    {
      "epoch": 0.34407053038725943,
      "grad_norm": 0.7875433574251917,
      "learning_rate": 7.628583214465104e-06,
      "loss": 0.426,
      "step": 5493
    },
    {
      "epoch": 0.34413316838660174,
      "grad_norm": 0.8424979401897935,
      "learning_rate": 7.627720252262376e-06,
      "loss": 0.4303,
      "step": 5494
    },
    {
      "epoch": 0.34419580638594405,
      "grad_norm": 0.8572692889835675,
      "learning_rate": 7.6268571819021945e-06,
      "loss": 0.4377,
      "step": 5495
    },
    {
      "epoch": 0.34425844438528636,
      "grad_norm": 0.7774409467404066,
      "learning_rate": 7.625994003420082e-06,
      "loss": 0.472,
      "step": 5496
    },
    {
      "epoch": 0.3443210823846286,
      "grad_norm": 0.8159318500200293,
      "learning_rate": 7.625130716851571e-06,
      "loss": 0.4089,
      "step": 5497
    },
    {
      "epoch": 0.3443837203839709,
      "grad_norm": 0.8218675176111843,
      "learning_rate": 7.624267322232193e-06,
      "loss": 0.4261,
      "step": 5498
    },
    {
      "epoch": 0.3444463583833132,
      "grad_norm": 0.7964246981519245,
      "learning_rate": 7.623403819597484e-06,
      "loss": 0.4234,
      "step": 5499
    },
    {
      "epoch": 0.34450899638265553,
      "grad_norm": 0.8335868230029775,
      "learning_rate": 7.622540208982985e-06,
      "loss": 0.4315,
      "step": 5500
    },
    {
      "epoch": 0.34457163438199784,
      "grad_norm": 0.7487537720787436,
      "learning_rate": 7.621676490424247e-06,
      "loss": 0.4299,
      "step": 5501
    },
    {
      "epoch": 0.34463427238134015,
      "grad_norm": 0.8487150716253518,
      "learning_rate": 7.620812663956818e-06,
      "loss": 0.4021,
      "step": 5502
    },
    {
      "epoch": 0.34469691038068245,
      "grad_norm": 0.8126540608163572,
      "learning_rate": 7.619948729616251e-06,
      "loss": 0.4117,
      "step": 5503
    },
    {
      "epoch": 0.34475954838002476,
      "grad_norm": 0.8074328919256535,
      "learning_rate": 7.61908468743811e-06,
      "loss": 0.4185,
      "step": 5504
    },
    {
      "epoch": 0.34482218637936707,
      "grad_norm": 0.7883477836178467,
      "learning_rate": 7.6182205374579564e-06,
      "loss": 0.4328,
      "step": 5505
    },
    {
      "epoch": 0.3448848243787093,
      "grad_norm": 0.7637949209650231,
      "learning_rate": 7.61735627971136e-06,
      "loss": 0.4759,
      "step": 5506
    },
    {
      "epoch": 0.34494746237805163,
      "grad_norm": 0.7967077855185555,
      "learning_rate": 7.616491914233894e-06,
      "loss": 0.4106,
      "step": 5507
    },
    {
      "epoch": 0.34501010037739394,
      "grad_norm": 0.8509370565585398,
      "learning_rate": 7.615627441061133e-06,
      "loss": 0.3912,
      "step": 5508
    },
    {
      "epoch": 0.34507273837673624,
      "grad_norm": 0.8615819670423556,
      "learning_rate": 7.6147628602286615e-06,
      "loss": 0.439,
      "step": 5509
    },
    {
      "epoch": 0.34513537637607855,
      "grad_norm": 0.9090902146333607,
      "learning_rate": 7.6138981717720675e-06,
      "loss": 0.4214,
      "step": 5510
    },
    {
      "epoch": 0.34519801437542086,
      "grad_norm": 0.8671818291851687,
      "learning_rate": 7.613033375726937e-06,
      "loss": 0.438,
      "step": 5511
    },
    {
      "epoch": 0.34526065237476317,
      "grad_norm": 0.8717363724689163,
      "learning_rate": 7.612168472128869e-06,
      "loss": 0.4582,
      "step": 5512
    },
    {
      "epoch": 0.3453232903741055,
      "grad_norm": 0.7773691544372587,
      "learning_rate": 7.611303461013462e-06,
      "loss": 0.378,
      "step": 5513
    },
    {
      "epoch": 0.3453859283734477,
      "grad_norm": 0.8129926999202555,
      "learning_rate": 7.61043834241632e-06,
      "loss": 0.4438,
      "step": 5514
    },
    {
      "epoch": 0.34544856637279003,
      "grad_norm": 0.8546393835651444,
      "learning_rate": 7.6095731163730505e-06,
      "loss": 0.4098,
      "step": 5515
    },
    {
      "epoch": 0.34551120437213234,
      "grad_norm": 0.8753131067019376,
      "learning_rate": 7.608707782919267e-06,
      "loss": 0.4161,
      "step": 5516
    },
    {
      "epoch": 0.34557384237147465,
      "grad_norm": 0.7659399975431069,
      "learning_rate": 7.6078423420905875e-06,
      "loss": 0.394,
      "step": 5517
    },
    {
      "epoch": 0.34563648037081696,
      "grad_norm": 0.8661957423644814,
      "learning_rate": 7.606976793922633e-06,
      "loss": 0.4372,
      "step": 5518
    },
    {
      "epoch": 0.34569911837015926,
      "grad_norm": 0.7526656768656645,
      "learning_rate": 7.606111138451031e-06,
      "loss": 0.3603,
      "step": 5519
    },
    {
      "epoch": 0.34576175636950157,
      "grad_norm": 0.7952322095822639,
      "learning_rate": 7.60524537571141e-06,
      "loss": 0.3883,
      "step": 5520
    },
    {
      "epoch": 0.3458243943688439,
      "grad_norm": 1.1050798752956899,
      "learning_rate": 7.604379505739406e-06,
      "loss": 0.4626,
      "step": 5521
    },
    {
      "epoch": 0.3458870323681862,
      "grad_norm": 0.8322182895846107,
      "learning_rate": 7.603513528570656e-06,
      "loss": 0.4376,
      "step": 5522
    },
    {
      "epoch": 0.34594967036752844,
      "grad_norm": 0.8031252205047525,
      "learning_rate": 7.6026474442408085e-06,
      "loss": 0.3639,
      "step": 5523
    },
    {
      "epoch": 0.34601230836687075,
      "grad_norm": 0.8177811524143744,
      "learning_rate": 7.6017812527855095e-06,
      "loss": 0.4026,
      "step": 5524
    },
    {
      "epoch": 0.34607494636621305,
      "grad_norm": 0.7481666920352648,
      "learning_rate": 7.6009149542404085e-06,
      "loss": 0.3613,
      "step": 5525
    },
    {
      "epoch": 0.34613758436555536,
      "grad_norm": 0.7952339091045558,
      "learning_rate": 7.600048548641168e-06,
      "loss": 0.3979,
      "step": 5526
    },
    {
      "epoch": 0.34620022236489767,
      "grad_norm": 0.7850935903715607,
      "learning_rate": 7.599182036023445e-06,
      "loss": 0.5111,
      "step": 5527
    },
    {
      "epoch": 0.34626286036424,
      "grad_norm": 0.8435654078486758,
      "learning_rate": 7.598315416422908e-06,
      "loss": 0.4122,
      "step": 5528
    },
    {
      "epoch": 0.3463254983635823,
      "grad_norm": 0.862163617978835,
      "learning_rate": 7.597448689875224e-06,
      "loss": 0.4376,
      "step": 5529
    },
    {
      "epoch": 0.3463881363629246,
      "grad_norm": 0.6971891570090111,
      "learning_rate": 7.596581856416072e-06,
      "loss": 0.4793,
      "step": 5530
    },
    {
      "epoch": 0.34645077436226684,
      "grad_norm": 0.8395506449904372,
      "learning_rate": 7.595714916081127e-06,
      "loss": 0.4261,
      "step": 5531
    },
    {
      "epoch": 0.34651341236160915,
      "grad_norm": 0.9071732292041191,
      "learning_rate": 7.594847868906076e-06,
      "loss": 0.4643,
      "step": 5532
    },
    {
      "epoch": 0.34657605036095146,
      "grad_norm": 0.8492994793191098,
      "learning_rate": 7.5939807149266055e-06,
      "loss": 0.4739,
      "step": 5533
    },
    {
      "epoch": 0.34663868836029377,
      "grad_norm": 0.811555054853143,
      "learning_rate": 7.5931134541784056e-06,
      "loss": 0.4019,
      "step": 5534
    },
    {
      "epoch": 0.3467013263596361,
      "grad_norm": 0.9050161587848232,
      "learning_rate": 7.5922460866971755e-06,
      "loss": 0.401,
      "step": 5535
    },
    {
      "epoch": 0.3467639643589784,
      "grad_norm": 0.8062735018562666,
      "learning_rate": 7.591378612518615e-06,
      "loss": 0.3738,
      "step": 5536
    },
    {
      "epoch": 0.3468266023583207,
      "grad_norm": 0.7984969672813276,
      "learning_rate": 7.590511031678431e-06,
      "loss": 0.4118,
      "step": 5537
    },
    {
      "epoch": 0.346889240357663,
      "grad_norm": 0.7739638528685552,
      "learning_rate": 7.58964334421233e-06,
      "loss": 0.4317,
      "step": 5538
    },
    {
      "epoch": 0.34695187835700525,
      "grad_norm": 0.903617359463641,
      "learning_rate": 7.58877555015603e-06,
      "loss": 0.4538,
      "step": 5539
    },
    {
      "epoch": 0.34701451635634756,
      "grad_norm": 0.8502971114961537,
      "learning_rate": 7.587907649545248e-06,
      "loss": 0.4589,
      "step": 5540
    },
    {
      "epoch": 0.34707715435568987,
      "grad_norm": 0.8437101547561329,
      "learning_rate": 7.587039642415706e-06,
      "loss": 0.4156,
      "step": 5541
    },
    {
      "epoch": 0.3471397923550322,
      "grad_norm": 0.8559737713855234,
      "learning_rate": 7.586171528803132e-06,
      "loss": 0.3998,
      "step": 5542
    },
    {
      "epoch": 0.3472024303543745,
      "grad_norm": 0.7502583505525121,
      "learning_rate": 7.585303308743257e-06,
      "loss": 0.3811,
      "step": 5543
    },
    {
      "epoch": 0.3472650683537168,
      "grad_norm": 0.8418066727286625,
      "learning_rate": 7.58443498227182e-06,
      "loss": 0.4848,
      "step": 5544
    },
    {
      "epoch": 0.3473277063530591,
      "grad_norm": 0.8288971487446554,
      "learning_rate": 7.583566549424558e-06,
      "loss": 0.412,
      "step": 5545
    },
    {
      "epoch": 0.3473903443524014,
      "grad_norm": 0.8264610786064398,
      "learning_rate": 7.582698010237218e-06,
      "loss": 0.3719,
      "step": 5546
    },
    {
      "epoch": 0.3474529823517437,
      "grad_norm": 0.8040868304917782,
      "learning_rate": 7.581829364745548e-06,
      "loss": 0.4261,
      "step": 5547
    },
    {
      "epoch": 0.34751562035108596,
      "grad_norm": 0.802610396326057,
      "learning_rate": 7.580960612985303e-06,
      "loss": 0.3734,
      "step": 5548
    },
    {
      "epoch": 0.34757825835042827,
      "grad_norm": 0.8226891108978885,
      "learning_rate": 7.580091754992241e-06,
      "loss": 0.4154,
      "step": 5549
    },
    {
      "epoch": 0.3476408963497706,
      "grad_norm": 0.857761133500959,
      "learning_rate": 7.579222790802121e-06,
      "loss": 0.4103,
      "step": 5550
    },
    {
      "epoch": 0.3477035343491129,
      "grad_norm": 0.8586351840568583,
      "learning_rate": 7.578353720450713e-06,
      "loss": 0.3948,
      "step": 5551
    },
    {
      "epoch": 0.3477661723484552,
      "grad_norm": 0.812796863684969,
      "learning_rate": 7.5774845439737874e-06,
      "loss": 0.41,
      "step": 5552
    },
    {
      "epoch": 0.3478288103477975,
      "grad_norm": 0.8282526589701269,
      "learning_rate": 7.576615261407121e-06,
      "loss": 0.4355,
      "step": 5553
    },
    {
      "epoch": 0.3478914483471398,
      "grad_norm": 0.8279330877254666,
      "learning_rate": 7.575745872786489e-06,
      "loss": 0.3905,
      "step": 5554
    },
    {
      "epoch": 0.3479540863464821,
      "grad_norm": 0.8118292702858652,
      "learning_rate": 7.574876378147681e-06,
      "loss": 0.4299,
      "step": 5555
    },
    {
      "epoch": 0.34801672434582437,
      "grad_norm": 0.777081593538002,
      "learning_rate": 7.574006777526484e-06,
      "loss": 0.356,
      "step": 5556
    },
    {
      "epoch": 0.3480793623451667,
      "grad_norm": 0.8536432205532797,
      "learning_rate": 7.573137070958689e-06,
      "loss": 0.4373,
      "step": 5557
    },
    {
      "epoch": 0.348142000344509,
      "grad_norm": 0.6613734257030194,
      "learning_rate": 7.572267258480095e-06,
      "loss": 0.4559,
      "step": 5558
    },
    {
      "epoch": 0.3482046383438513,
      "grad_norm": 0.7931299232524233,
      "learning_rate": 7.571397340126503e-06,
      "loss": 0.4244,
      "step": 5559
    },
    {
      "epoch": 0.3482672763431936,
      "grad_norm": 0.842028987366599,
      "learning_rate": 7.570527315933719e-06,
      "loss": 0.4362,
      "step": 5560
    },
    {
      "epoch": 0.3483299143425359,
      "grad_norm": 0.8409704659980595,
      "learning_rate": 7.569657185937553e-06,
      "loss": 0.4025,
      "step": 5561
    },
    {
      "epoch": 0.3483925523418782,
      "grad_norm": 0.7112862795221685,
      "learning_rate": 7.568786950173823e-06,
      "loss": 0.5065,
      "step": 5562
    },
    {
      "epoch": 0.3484551903412205,
      "grad_norm": 0.8445363810804172,
      "learning_rate": 7.567916608678343e-06,
      "loss": 0.435,
      "step": 5563
    },
    {
      "epoch": 0.34851782834056283,
      "grad_norm": 0.7999924819754477,
      "learning_rate": 7.567046161486939e-06,
      "loss": 0.4154,
      "step": 5564
    },
    {
      "epoch": 0.3485804663399051,
      "grad_norm": 0.8940172183681098,
      "learning_rate": 7.566175608635439e-06,
      "loss": 0.392,
      "step": 5565
    },
    {
      "epoch": 0.3486431043392474,
      "grad_norm": 0.7921573453373009,
      "learning_rate": 7.565304950159676e-06,
      "loss": 0.4001,
      "step": 5566
    },
    {
      "epoch": 0.3487057423385897,
      "grad_norm": 0.7786946995227614,
      "learning_rate": 7.564434186095483e-06,
      "loss": 0.437,
      "step": 5567
    },
    {
      "epoch": 0.348768380337932,
      "grad_norm": 0.7693615224349294,
      "learning_rate": 7.563563316478702e-06,
      "loss": 0.4071,
      "step": 5568
    },
    {
      "epoch": 0.3488310183372743,
      "grad_norm": 0.78837305582,
      "learning_rate": 7.562692341345182e-06,
      "loss": 0.3798,
      "step": 5569
    },
    {
      "epoch": 0.3488936563366166,
      "grad_norm": 0.7679854906393965,
      "learning_rate": 7.561821260730767e-06,
      "loss": 0.3902,
      "step": 5570
    },
    {
      "epoch": 0.3489562943359589,
      "grad_norm": 0.8621496717285919,
      "learning_rate": 7.560950074671315e-06,
      "loss": 0.3949,
      "step": 5571
    },
    {
      "epoch": 0.34901893233530124,
      "grad_norm": 0.8948271664712341,
      "learning_rate": 7.560078783202682e-06,
      "loss": 0.4361,
      "step": 5572
    },
    {
      "epoch": 0.3490815703346435,
      "grad_norm": 0.8453287770033234,
      "learning_rate": 7.559207386360731e-06,
      "loss": 0.4173,
      "step": 5573
    },
    {
      "epoch": 0.3491442083339858,
      "grad_norm": 0.8029652339472014,
      "learning_rate": 7.558335884181329e-06,
      "loss": 0.4101,
      "step": 5574
    },
    {
      "epoch": 0.3492068463333281,
      "grad_norm": 0.8140901847156597,
      "learning_rate": 7.5574642767003485e-06,
      "loss": 0.4196,
      "step": 5575
    },
    {
      "epoch": 0.3492694843326704,
      "grad_norm": 0.8240548321595843,
      "learning_rate": 7.556592563953662e-06,
      "loss": 0.3771,
      "step": 5576
    },
    {
      "epoch": 0.3493321223320127,
      "grad_norm": 0.811731753829463,
      "learning_rate": 7.555720745977151e-06,
      "loss": 0.4363,
      "step": 5577
    },
    {
      "epoch": 0.349394760331355,
      "grad_norm": 0.8324323434713622,
      "learning_rate": 7.554848822806699e-06,
      "loss": 0.4215,
      "step": 5578
    },
    {
      "epoch": 0.34945739833069733,
      "grad_norm": 0.7844045750978696,
      "learning_rate": 7.553976794478199e-06,
      "loss": 0.3931,
      "step": 5579
    },
    {
      "epoch": 0.34952003633003964,
      "grad_norm": 0.7876374460826636,
      "learning_rate": 7.5531046610275375e-06,
      "loss": 0.3676,
      "step": 5580
    },
    {
      "epoch": 0.3495826743293819,
      "grad_norm": 0.8462052002736564,
      "learning_rate": 7.552232422490613e-06,
      "loss": 0.3872,
      "step": 5581
    },
    {
      "epoch": 0.3496453123287242,
      "grad_norm": 0.7911929997198875,
      "learning_rate": 7.551360078903329e-06,
      "loss": 0.4269,
      "step": 5582
    },
    {
      "epoch": 0.3497079503280665,
      "grad_norm": 0.8394500995462816,
      "learning_rate": 7.550487630301593e-06,
      "loss": 0.44,
      "step": 5583
    },
    {
      "epoch": 0.3497705883274088,
      "grad_norm": 0.8030892286091592,
      "learning_rate": 7.549615076721311e-06,
      "loss": 0.3973,
      "step": 5584
    },
    {
      "epoch": 0.3498332263267511,
      "grad_norm": 0.8303077812694794,
      "learning_rate": 7.548742418198398e-06,
      "loss": 0.4482,
      "step": 5585
    },
    {
      "epoch": 0.34989586432609343,
      "grad_norm": 0.8710020386102354,
      "learning_rate": 7.547869654768776e-06,
      "loss": 0.4452,
      "step": 5586
    },
    {
      "epoch": 0.34995850232543574,
      "grad_norm": 0.8353248065898808,
      "learning_rate": 7.546996786468366e-06,
      "loss": 0.4041,
      "step": 5587
    },
    {
      "epoch": 0.35002114032477805,
      "grad_norm": 0.8811168581531634,
      "learning_rate": 7.546123813333095e-06,
      "loss": 0.4477,
      "step": 5588
    },
    {
      "epoch": 0.35008377832412035,
      "grad_norm": 0.6645374909981564,
      "learning_rate": 7.545250735398895e-06,
      "loss": 0.4866,
      "step": 5589
    },
    {
      "epoch": 0.3501464163234626,
      "grad_norm": 0.8385627043196684,
      "learning_rate": 7.5443775527017024e-06,
      "loss": 0.4207,
      "step": 5590
    },
    {
      "epoch": 0.3502090543228049,
      "grad_norm": 0.8773141947492662,
      "learning_rate": 7.543504265277457e-06,
      "loss": 0.4247,
      "step": 5591
    },
    {
      "epoch": 0.3502716923221472,
      "grad_norm": 0.8407072946225689,
      "learning_rate": 7.542630873162104e-06,
      "loss": 0.3585,
      "step": 5592
    },
    {
      "epoch": 0.35033433032148953,
      "grad_norm": 0.8263017230096831,
      "learning_rate": 7.541757376391593e-06,
      "loss": 0.3903,
      "step": 5593
    },
    {
      "epoch": 0.35039696832083184,
      "grad_norm": 0.7718370536570259,
      "learning_rate": 7.540883775001877e-06,
      "loss": 0.3896,
      "step": 5594
    },
    {
      "epoch": 0.35045960632017414,
      "grad_norm": 0.8077262503521131,
      "learning_rate": 7.540010069028913e-06,
      "loss": 0.4002,
      "step": 5595
    },
    {
      "epoch": 0.35052224431951645,
      "grad_norm": 0.8460759356861391,
      "learning_rate": 7.5391362585086615e-06,
      "loss": 0.4168,
      "step": 5596
    },
    {
      "epoch": 0.35058488231885876,
      "grad_norm": 0.849168072376639,
      "learning_rate": 7.538262343477092e-06,
      "loss": 0.4326,
      "step": 5597
    },
    {
      "epoch": 0.350647520318201,
      "grad_norm": 0.86876935518197,
      "learning_rate": 7.537388323970173e-06,
      "loss": 0.4163,
      "step": 5598
    },
    {
      "epoch": 0.3507101583175433,
      "grad_norm": 0.8461947274251624,
      "learning_rate": 7.53651420002388e-06,
      "loss": 0.433,
      "step": 5599
    },
    {
      "epoch": 0.3507727963168856,
      "grad_norm": 0.858015964817803,
      "learning_rate": 7.535639971674191e-06,
      "loss": 0.4232,
      "step": 5600
    },
    {
      "epoch": 0.35083543431622793,
      "grad_norm": 0.8032187892183488,
      "learning_rate": 7.534765638957091e-06,
      "loss": 0.3728,
      "step": 5601
    },
    {
      "epoch": 0.35089807231557024,
      "grad_norm": 0.8295230927820374,
      "learning_rate": 7.533891201908565e-06,
      "loss": 0.436,
      "step": 5602
    },
    {
      "epoch": 0.35096071031491255,
      "grad_norm": 0.8479115206381318,
      "learning_rate": 7.533016660564608e-06,
      "loss": 0.4191,
      "step": 5603
    },
    {
      "epoch": 0.35102334831425486,
      "grad_norm": 0.8661556091864963,
      "learning_rate": 7.532142014961216e-06,
      "loss": 0.4245,
      "step": 5604
    },
    {
      "epoch": 0.35108598631359716,
      "grad_norm": 0.8283708295938161,
      "learning_rate": 7.531267265134388e-06,
      "loss": 0.4424,
      "step": 5605
    },
    {
      "epoch": 0.3511486243129395,
      "grad_norm": 0.8568452094749072,
      "learning_rate": 7.530392411120128e-06,
      "loss": 0.3774,
      "step": 5606
    },
    {
      "epoch": 0.3512112623122817,
      "grad_norm": 0.8942284379404295,
      "learning_rate": 7.529517452954449e-06,
      "loss": 0.4545,
      "step": 5607
    },
    {
      "epoch": 0.35127390031162403,
      "grad_norm": 0.7845076006537623,
      "learning_rate": 7.528642390673361e-06,
      "loss": 0.4016,
      "step": 5608
    },
    {
      "epoch": 0.35133653831096634,
      "grad_norm": 0.8749380829608026,
      "learning_rate": 7.527767224312883e-06,
      "loss": 0.4327,
      "step": 5609
    },
    {
      "epoch": 0.35139917631030865,
      "grad_norm": 0.747695654837569,
      "learning_rate": 7.526891953909036e-06,
      "loss": 0.3831,
      "step": 5610
    },
    {
      "epoch": 0.35146181430965096,
      "grad_norm": 0.8817307271344651,
      "learning_rate": 7.5260165794978476e-06,
      "loss": 0.4271,
      "step": 5611
    },
    {
      "epoch": 0.35152445230899326,
      "grad_norm": 0.8136011941471594,
      "learning_rate": 7.52514110111535e-06,
      "loss": 0.3894,
      "step": 5612
    },
    {
      "epoch": 0.35158709030833557,
      "grad_norm": 0.8305791730217351,
      "learning_rate": 7.5242655187975744e-06,
      "loss": 0.4021,
      "step": 5613
    },
    {
      "epoch": 0.3516497283076779,
      "grad_norm": 0.8998248885618982,
      "learning_rate": 7.523389832580561e-06,
      "loss": 0.4499,
      "step": 5614
    },
    {
      "epoch": 0.35171236630702013,
      "grad_norm": 0.8320974372541124,
      "learning_rate": 7.522514042500353e-06,
      "loss": 0.4177,
      "step": 5615
    },
    {
      "epoch": 0.35177500430636244,
      "grad_norm": 0.8405822323805016,
      "learning_rate": 7.521638148593001e-06,
      "loss": 0.4271,
      "step": 5616
    },
    {
      "epoch": 0.35183764230570475,
      "grad_norm": 0.8159909787176922,
      "learning_rate": 7.520762150894554e-06,
      "loss": 0.3982,
      "step": 5617
    },
    {
      "epoch": 0.35190028030504705,
      "grad_norm": 0.8971951339679737,
      "learning_rate": 7.51988604944107e-06,
      "loss": 0.433,
      "step": 5618
    },
    {
      "epoch": 0.35196291830438936,
      "grad_norm": 0.7547752632123363,
      "learning_rate": 7.519009844268605e-06,
      "loss": 0.3716,
      "step": 5619
    },
    {
      "epoch": 0.35202555630373167,
      "grad_norm": 0.7784585799297568,
      "learning_rate": 7.51813353541323e-06,
      "loss": 0.4288,
      "step": 5620
    },
    {
      "epoch": 0.352088194303074,
      "grad_norm": 0.7995662766151819,
      "learning_rate": 7.517257122911011e-06,
      "loss": 0.4287,
      "step": 5621
    },
    {
      "epoch": 0.3521508323024163,
      "grad_norm": 0.8433200671178743,
      "learning_rate": 7.516380606798022e-06,
      "loss": 0.4154,
      "step": 5622
    },
    {
      "epoch": 0.35221347030175854,
      "grad_norm": 0.8559890033354324,
      "learning_rate": 7.51550398711034e-06,
      "loss": 0.4474,
      "step": 5623
    },
    {
      "epoch": 0.35227610830110084,
      "grad_norm": 0.8225210237691848,
      "learning_rate": 7.514627263884045e-06,
      "loss": 0.4098,
      "step": 5624
    },
    {
      "epoch": 0.35233874630044315,
      "grad_norm": 0.873414002293103,
      "learning_rate": 7.513750437155227e-06,
      "loss": 0.4913,
      "step": 5625
    },
    {
      "epoch": 0.35240138429978546,
      "grad_norm": 0.8151223164219492,
      "learning_rate": 7.5128735069599745e-06,
      "loss": 0.417,
      "step": 5626
    },
    {
      "epoch": 0.35246402229912777,
      "grad_norm": 0.8438878799398686,
      "learning_rate": 7.511996473334382e-06,
      "loss": 0.4004,
      "step": 5627
    },
    {
      "epoch": 0.3525266602984701,
      "grad_norm": 0.841482960438345,
      "learning_rate": 7.511119336314547e-06,
      "loss": 0.4232,
      "step": 5628
    },
    {
      "epoch": 0.3525892982978124,
      "grad_norm": 0.8175113425223685,
      "learning_rate": 7.510242095936575e-06,
      "loss": 0.425,
      "step": 5629
    },
    {
      "epoch": 0.3526519362971547,
      "grad_norm": 0.816079634966281,
      "learning_rate": 7.509364752236574e-06,
      "loss": 0.3859,
      "step": 5630
    },
    {
      "epoch": 0.352714574296497,
      "grad_norm": 0.7798798870629529,
      "learning_rate": 7.5084873052506525e-06,
      "loss": 0.4059,
      "step": 5631
    },
    {
      "epoch": 0.35277721229583925,
      "grad_norm": 0.8116179168423437,
      "learning_rate": 7.507609755014927e-06,
      "loss": 0.4154,
      "step": 5632
    },
    {
      "epoch": 0.35283985029518156,
      "grad_norm": 0.7765970698205323,
      "learning_rate": 7.506732101565519e-06,
      "loss": 0.3732,
      "step": 5633
    },
    {
      "epoch": 0.35290248829452386,
      "grad_norm": 0.7448512634996391,
      "learning_rate": 7.505854344938555e-06,
      "loss": 0.3734,
      "step": 5634
    },
    {
      "epoch": 0.35296512629386617,
      "grad_norm": 0.8372485608918793,
      "learning_rate": 7.504976485170159e-06,
      "loss": 0.414,
      "step": 5635
    },
    {
      "epoch": 0.3530277642932085,
      "grad_norm": 0.8787520500902823,
      "learning_rate": 7.504098522296464e-06,
      "loss": 0.4323,
      "step": 5636
    },
    {
      "epoch": 0.3530904022925508,
      "grad_norm": 0.8525024702997288,
      "learning_rate": 7.50322045635361e-06,
      "loss": 0.3899,
      "step": 5637
    },
    {
      "epoch": 0.3531530402918931,
      "grad_norm": 0.8154736150681071,
      "learning_rate": 7.502342287377739e-06,
      "loss": 0.405,
      "step": 5638
    },
    {
      "epoch": 0.3532156782912354,
      "grad_norm": 0.8943359449059987,
      "learning_rate": 7.501464015404996e-06,
      "loss": 0.4334,
      "step": 5639
    },
    {
      "epoch": 0.35327831629057765,
      "grad_norm": 0.8264293292897433,
      "learning_rate": 7.500585640471527e-06,
      "loss": 0.4271,
      "step": 5640
    },
    {
      "epoch": 0.35334095428991996,
      "grad_norm": 0.7939491936617681,
      "learning_rate": 7.499707162613491e-06,
      "loss": 0.4118,
      "step": 5641
    },
    {
      "epoch": 0.35340359228926227,
      "grad_norm": 0.8334957626123541,
      "learning_rate": 7.498828581867043e-06,
      "loss": 0.3864,
      "step": 5642
    },
    {
      "epoch": 0.3534662302886046,
      "grad_norm": 0.7698527499914908,
      "learning_rate": 7.497949898268346e-06,
      "loss": 0.3867,
      "step": 5643
    },
    {
      "epoch": 0.3535288682879469,
      "grad_norm": 0.8234914676080041,
      "learning_rate": 7.49707111185357e-06,
      "loss": 0.367,
      "step": 5644
    },
    {
      "epoch": 0.3535915062872892,
      "grad_norm": 0.8781733733795352,
      "learning_rate": 7.496192222658881e-06,
      "loss": 0.4181,
      "step": 5645
    },
    {
      "epoch": 0.3536541442866315,
      "grad_norm": 0.8971065383898187,
      "learning_rate": 7.495313230720459e-06,
      "loss": 0.4313,
      "step": 5646
    },
    {
      "epoch": 0.3537167822859738,
      "grad_norm": 0.8712249646655686,
      "learning_rate": 7.4944341360744795e-06,
      "loss": 0.4314,
      "step": 5647
    },
    {
      "epoch": 0.35377942028531606,
      "grad_norm": 0.6618227819140626,
      "learning_rate": 7.493554938757129e-06,
      "loss": 0.4614,
      "step": 5648
    },
    {
      "epoch": 0.35384205828465837,
      "grad_norm": 0.8131328700778558,
      "learning_rate": 7.492675638804594e-06,
      "loss": 0.3961,
      "step": 5649
    },
    {
      "epoch": 0.3539046962840007,
      "grad_norm": 0.8183217696455036,
      "learning_rate": 7.4917962362530675e-06,
      "loss": 0.4005,
      "step": 5650
    },
    {
      "epoch": 0.353967334283343,
      "grad_norm": 0.8816309342445946,
      "learning_rate": 7.490916731138744e-06,
      "loss": 0.4043,
      "step": 5651
    },
    {
      "epoch": 0.3540299722826853,
      "grad_norm": 0.8824507920353514,
      "learning_rate": 7.490037123497828e-06,
      "loss": 0.415,
      "step": 5652
    },
    {
      "epoch": 0.3540926102820276,
      "grad_norm": 0.9390184248528646,
      "learning_rate": 7.489157413366519e-06,
      "loss": 0.4466,
      "step": 5653
    },
    {
      "epoch": 0.3541552482813699,
      "grad_norm": 0.6738931191183793,
      "learning_rate": 7.488277600781031e-06,
      "loss": 0.4638,
      "step": 5654
    },
    {
      "epoch": 0.3542178862807122,
      "grad_norm": 0.9302153164071008,
      "learning_rate": 7.487397685777575e-06,
      "loss": 0.4364,
      "step": 5655
    },
    {
      "epoch": 0.3542805242800545,
      "grad_norm": 0.8282397676004909,
      "learning_rate": 7.486517668392367e-06,
      "loss": 0.423,
      "step": 5656
    },
    {
      "epoch": 0.3543431622793968,
      "grad_norm": 0.8757809555930274,
      "learning_rate": 7.485637548661632e-06,
      "loss": 0.4765,
      "step": 5657
    },
    {
      "epoch": 0.3544058002787391,
      "grad_norm": 0.8233331660676295,
      "learning_rate": 7.484757326621592e-06,
      "loss": 0.4446,
      "step": 5658
    },
    {
      "epoch": 0.3544684382780814,
      "grad_norm": 0.8347564605381815,
      "learning_rate": 7.483877002308482e-06,
      "loss": 0.4409,
      "step": 5659
    },
    {
      "epoch": 0.3545310762774237,
      "grad_norm": 0.8700257196005199,
      "learning_rate": 7.482996575758531e-06,
      "loss": 0.4507,
      "step": 5660
    },
    {
      "epoch": 0.354593714276766,
      "grad_norm": 0.9339781562913889,
      "learning_rate": 7.482116047007981e-06,
      "loss": 0.453,
      "step": 5661
    },
    {
      "epoch": 0.3546563522761083,
      "grad_norm": 0.8535346428277744,
      "learning_rate": 7.481235416093075e-06,
      "loss": 0.4298,
      "step": 5662
    },
    {
      "epoch": 0.3547189902754506,
      "grad_norm": 0.7872395600135368,
      "learning_rate": 7.4803546830500576e-06,
      "loss": 0.3974,
      "step": 5663
    },
    {
      "epoch": 0.3547816282747929,
      "grad_norm": 0.8491596400887164,
      "learning_rate": 7.479473847915182e-06,
      "loss": 0.4187,
      "step": 5664
    },
    {
      "epoch": 0.3548442662741352,
      "grad_norm": 0.8687019991627913,
      "learning_rate": 7.4785929107247e-06,
      "loss": 0.4305,
      "step": 5665
    },
    {
      "epoch": 0.3549069042734775,
      "grad_norm": 0.9590327054854777,
      "learning_rate": 7.477711871514876e-06,
      "loss": 0.4087,
      "step": 5666
    },
    {
      "epoch": 0.3549695422728198,
      "grad_norm": 0.8737071612185303,
      "learning_rate": 7.476830730321971e-06,
      "loss": 0.4134,
      "step": 5667
    },
    {
      "epoch": 0.3550321802721621,
      "grad_norm": 0.81178147717957,
      "learning_rate": 7.475949487182255e-06,
      "loss": 0.4287,
      "step": 5668
    },
    {
      "epoch": 0.3550948182715044,
      "grad_norm": 0.7848668177486939,
      "learning_rate": 7.475068142131996e-06,
      "loss": 0.416,
      "step": 5669
    },
    {
      "epoch": 0.3551574562708467,
      "grad_norm": 0.8931840081050557,
      "learning_rate": 7.474186695207474e-06,
      "loss": 0.4236,
      "step": 5670
    },
    {
      "epoch": 0.355220094270189,
      "grad_norm": 0.8167871413651073,
      "learning_rate": 7.473305146444968e-06,
      "loss": 0.4209,
      "step": 5671
    },
    {
      "epoch": 0.35528273226953133,
      "grad_norm": 0.8081293034695419,
      "learning_rate": 7.472423495880763e-06,
      "loss": 0.3912,
      "step": 5672
    },
    {
      "epoch": 0.35534537026887364,
      "grad_norm": 0.8035962345311985,
      "learning_rate": 7.47154174355115e-06,
      "loss": 0.3759,
      "step": 5673
    },
    {
      "epoch": 0.3554080082682159,
      "grad_norm": 0.8454381722528947,
      "learning_rate": 7.4706598894924175e-06,
      "loss": 0.4367,
      "step": 5674
    },
    {
      "epoch": 0.3554706462675582,
      "grad_norm": 0.8183284628454623,
      "learning_rate": 7.4697779337408645e-06,
      "loss": 0.3867,
      "step": 5675
    },
    {
      "epoch": 0.3555332842669005,
      "grad_norm": 0.8599125359393487,
      "learning_rate": 7.468895876332797e-06,
      "loss": 0.4036,
      "step": 5676
    },
    {
      "epoch": 0.3555959222662428,
      "grad_norm": 0.8579432156876312,
      "learning_rate": 7.468013717304515e-06,
      "loss": 0.44,
      "step": 5677
    },
    {
      "epoch": 0.3556585602655851,
      "grad_norm": 0.8864954610873553,
      "learning_rate": 7.467131456692331e-06,
      "loss": 0.4237,
      "step": 5678
    },
    {
      "epoch": 0.35572119826492743,
      "grad_norm": 0.78408753325875,
      "learning_rate": 7.466249094532556e-06,
      "loss": 0.4588,
      "step": 5679
    },
    {
      "epoch": 0.35578383626426974,
      "grad_norm": 0.9346666032856303,
      "learning_rate": 7.465366630861514e-06,
      "loss": 0.4562,
      "step": 5680
    },
    {
      "epoch": 0.35584647426361204,
      "grad_norm": 0.8518179851460292,
      "learning_rate": 7.464484065715521e-06,
      "loss": 0.3991,
      "step": 5681
    },
    {
      "epoch": 0.3559091122629543,
      "grad_norm": 0.7935288603941473,
      "learning_rate": 7.463601399130909e-06,
      "loss": 0.3996,
      "step": 5682
    },
    {
      "epoch": 0.3559717502622966,
      "grad_norm": 0.8351900851829116,
      "learning_rate": 7.462718631144003e-06,
      "loss": 0.4366,
      "step": 5683
    },
    {
      "epoch": 0.3560343882616389,
      "grad_norm": 0.8772243938733965,
      "learning_rate": 7.461835761791143e-06,
      "loss": 0.3965,
      "step": 5684
    },
    {
      "epoch": 0.3560970262609812,
      "grad_norm": 0.8228178071466743,
      "learning_rate": 7.4609527911086665e-06,
      "loss": 0.4378,
      "step": 5685
    },
    {
      "epoch": 0.3561596642603235,
      "grad_norm": 0.7724116761191941,
      "learning_rate": 7.460069719132916e-06,
      "loss": 0.3961,
      "step": 5686
    },
    {
      "epoch": 0.35622230225966584,
      "grad_norm": 0.7930201869313552,
      "learning_rate": 7.459186545900239e-06,
      "loss": 0.4117,
      "step": 5687
    },
    {
      "epoch": 0.35628494025900814,
      "grad_norm": 1.0305273081208268,
      "learning_rate": 7.458303271446987e-06,
      "loss": 0.4179,
      "step": 5688
    },
    {
      "epoch": 0.35634757825835045,
      "grad_norm": 0.8440484601827306,
      "learning_rate": 7.457419895809517e-06,
      "loss": 0.3939,
      "step": 5689
    },
    {
      "epoch": 0.3564102162576927,
      "grad_norm": 0.7794336123178631,
      "learning_rate": 7.456536419024188e-06,
      "loss": 0.4388,
      "step": 5690
    },
    {
      "epoch": 0.356472854257035,
      "grad_norm": 0.8636884559473316,
      "learning_rate": 7.455652841127362e-06,
      "loss": 0.4594,
      "step": 5691
    },
    {
      "epoch": 0.3565354922563773,
      "grad_norm": 0.843679139954803,
      "learning_rate": 7.454769162155411e-06,
      "loss": 0.3952,
      "step": 5692
    },
    {
      "epoch": 0.3565981302557196,
      "grad_norm": 0.8368292402802597,
      "learning_rate": 7.453885382144706e-06,
      "loss": 0.413,
      "step": 5693
    },
    {
      "epoch": 0.35666076825506193,
      "grad_norm": 0.7827048928531934,
      "learning_rate": 7.453001501131623e-06,
      "loss": 0.4011,
      "step": 5694
    },
    {
      "epoch": 0.35672340625440424,
      "grad_norm": 0.7639747942948917,
      "learning_rate": 7.452117519152542e-06,
      "loss": 0.3701,
      "step": 5695
    },
    {
      "epoch": 0.35678604425374655,
      "grad_norm": 0.9163166981380584,
      "learning_rate": 7.451233436243849e-06,
      "loss": 0.4619,
      "step": 5696
    },
    {
      "epoch": 0.35684868225308886,
      "grad_norm": 0.8657455955173209,
      "learning_rate": 7.450349252441932e-06,
      "loss": 0.4436,
      "step": 5697
    },
    {
      "epoch": 0.35691132025243116,
      "grad_norm": 0.7863725616471414,
      "learning_rate": 7.449464967783186e-06,
      "loss": 0.4117,
      "step": 5698
    },
    {
      "epoch": 0.3569739582517734,
      "grad_norm": 0.8049466803812365,
      "learning_rate": 7.448580582304008e-06,
      "loss": 0.3855,
      "step": 5699
    },
    {
      "epoch": 0.3570365962511157,
      "grad_norm": 0.7777560198582516,
      "learning_rate": 7.4476960960407974e-06,
      "loss": 0.409,
      "step": 5700
    },
    {
      "epoch": 0.35709923425045803,
      "grad_norm": 0.8125195935279501,
      "learning_rate": 7.446811509029963e-06,
      "loss": 0.4304,
      "step": 5701
    },
    {
      "epoch": 0.35716187224980034,
      "grad_norm": 0.8108497185016142,
      "learning_rate": 7.445926821307911e-06,
      "loss": 0.3943,
      "step": 5702
    },
    {
      "epoch": 0.35722451024914265,
      "grad_norm": 0.7999966598207707,
      "learning_rate": 7.4450420329110575e-06,
      "loss": 0.4171,
      "step": 5703
    },
    {
      "epoch": 0.35728714824848495,
      "grad_norm": 0.8395916445272443,
      "learning_rate": 7.44415714387582e-06,
      "loss": 0.3696,
      "step": 5704
    },
    {
      "epoch": 0.35734978624782726,
      "grad_norm": 0.8627990500658659,
      "learning_rate": 7.443272154238621e-06,
      "loss": 0.4416,
      "step": 5705
    },
    {
      "epoch": 0.35741242424716957,
      "grad_norm": 0.8487857354390858,
      "learning_rate": 7.442387064035886e-06,
      "loss": 0.4462,
      "step": 5706
    },
    {
      "epoch": 0.3574750622465118,
      "grad_norm": 0.8080522183707828,
      "learning_rate": 7.441501873304048e-06,
      "loss": 0.4375,
      "step": 5707
    },
    {
      "epoch": 0.35753770024585413,
      "grad_norm": 0.8677391298633034,
      "learning_rate": 7.440616582079539e-06,
      "loss": 0.4402,
      "step": 5708
    },
    {
      "epoch": 0.35760033824519644,
      "grad_norm": 0.8248975271312655,
      "learning_rate": 7.439731190398797e-06,
      "loss": 0.4088,
      "step": 5709
    },
    {
      "epoch": 0.35766297624453874,
      "grad_norm": 0.8524577032277924,
      "learning_rate": 7.438845698298267e-06,
      "loss": 0.4304,
      "step": 5710
    },
    {
      "epoch": 0.35772561424388105,
      "grad_norm": 0.8435182009389761,
      "learning_rate": 7.437960105814396e-06,
      "loss": 0.4366,
      "step": 5711
    },
    {
      "epoch": 0.35778825224322336,
      "grad_norm": 0.6799912318851833,
      "learning_rate": 7.437074412983635e-06,
      "loss": 0.4972,
      "step": 5712
    },
    {
      "epoch": 0.35785089024256567,
      "grad_norm": 0.8539132900650681,
      "learning_rate": 7.436188619842437e-06,
      "loss": 0.3955,
      "step": 5713
    },
    {
      "epoch": 0.357913528241908,
      "grad_norm": 0.8762537849851417,
      "learning_rate": 7.435302726427265e-06,
      "loss": 0.43,
      "step": 5714
    },
    {
      "epoch": 0.3579761662412503,
      "grad_norm": 0.8862405907040904,
      "learning_rate": 7.434416732774581e-06,
      "loss": 0.4612,
      "step": 5715
    },
    {
      "epoch": 0.35803880424059253,
      "grad_norm": 0.789525085673593,
      "learning_rate": 7.433530638920851e-06,
      "loss": 0.3925,
      "step": 5716
    },
    {
      "epoch": 0.35810144223993484,
      "grad_norm": 0.812194702053464,
      "learning_rate": 7.432644444902548e-06,
      "loss": 0.4524,
      "step": 5717
    },
    {
      "epoch": 0.35816408023927715,
      "grad_norm": 0.7836515573708729,
      "learning_rate": 7.431758150756149e-06,
      "loss": 0.4069,
      "step": 5718
    },
    {
      "epoch": 0.35822671823861946,
      "grad_norm": 0.8524090117850027,
      "learning_rate": 7.430871756518133e-06,
      "loss": 0.4497,
      "step": 5719
    },
    {
      "epoch": 0.35828935623796176,
      "grad_norm": 0.8103653881257393,
      "learning_rate": 7.429985262224983e-06,
      "loss": 0.4427,
      "step": 5720
    },
    {
      "epoch": 0.3583519942373041,
      "grad_norm": 0.8646339161747995,
      "learning_rate": 7.429098667913191e-06,
      "loss": 0.4324,
      "step": 5721
    },
    {
      "epoch": 0.3584146322366464,
      "grad_norm": 0.8078994230005493,
      "learning_rate": 7.428211973619246e-06,
      "loss": 0.4095,
      "step": 5722
    },
    {
      "epoch": 0.3584772702359887,
      "grad_norm": 0.8232795640546924,
      "learning_rate": 7.427325179379645e-06,
      "loss": 0.4322,
      "step": 5723
    },
    {
      "epoch": 0.35853990823533094,
      "grad_norm": 0.8090124094175934,
      "learning_rate": 7.426438285230891e-06,
      "loss": 0.4235,
      "step": 5724
    },
    {
      "epoch": 0.35860254623467325,
      "grad_norm": 0.8567743103280401,
      "learning_rate": 7.425551291209484e-06,
      "loss": 0.4253,
      "step": 5725
    },
    {
      "epoch": 0.35866518423401555,
      "grad_norm": 0.7609921519897238,
      "learning_rate": 7.424664197351936e-06,
      "loss": 0.3594,
      "step": 5726
    },
    {
      "epoch": 0.35872782223335786,
      "grad_norm": 0.8327894303290452,
      "learning_rate": 7.423777003694761e-06,
      "loss": 0.3901,
      "step": 5727
    },
    {
      "epoch": 0.35879046023270017,
      "grad_norm": 0.7697749276805892,
      "learning_rate": 7.422889710274475e-06,
      "loss": 0.4004,
      "step": 5728
    },
    {
      "epoch": 0.3588530982320425,
      "grad_norm": 0.8266805062329827,
      "learning_rate": 7.422002317127597e-06,
      "loss": 0.4253,
      "step": 5729
    },
    {
      "epoch": 0.3589157362313848,
      "grad_norm": 0.7945306242140134,
      "learning_rate": 7.421114824290654e-06,
      "loss": 0.3752,
      "step": 5730
    },
    {
      "epoch": 0.3589783742307271,
      "grad_norm": 0.8635246588338701,
      "learning_rate": 7.4202272318001775e-06,
      "loss": 0.4476,
      "step": 5731
    },
    {
      "epoch": 0.35904101223006935,
      "grad_norm": 0.8400818996093828,
      "learning_rate": 7.419339539692698e-06,
      "loss": 0.4357,
      "step": 5732
    },
    {
      "epoch": 0.35910365022941165,
      "grad_norm": 0.823570624101091,
      "learning_rate": 7.4184517480047545e-06,
      "loss": 0.4384,
      "step": 5733
    },
    {
      "epoch": 0.35916628822875396,
      "grad_norm": 0.8464233916310302,
      "learning_rate": 7.417563856772887e-06,
      "loss": 0.4259,
      "step": 5734
    },
    {
      "epoch": 0.35922892622809627,
      "grad_norm": 0.8253391130942974,
      "learning_rate": 7.4166758660336424e-06,
      "loss": 0.4119,
      "step": 5735
    },
    {
      "epoch": 0.3592915642274386,
      "grad_norm": 0.8024268054426582,
      "learning_rate": 7.415787775823572e-06,
      "loss": 0.4036,
      "step": 5736
    },
    {
      "epoch": 0.3593542022267809,
      "grad_norm": 0.8190574895790289,
      "learning_rate": 7.414899586179227e-06,
      "loss": 0.3859,
      "step": 5737
    },
    {
      "epoch": 0.3594168402261232,
      "grad_norm": 0.7522354962365168,
      "learning_rate": 7.414011297137168e-06,
      "loss": 0.3999,
      "step": 5738
    },
    {
      "epoch": 0.3594794782254655,
      "grad_norm": 0.8595328706299682,
      "learning_rate": 7.413122908733954e-06,
      "loss": 0.4303,
      "step": 5739
    },
    {
      "epoch": 0.3595421162248078,
      "grad_norm": 0.8177265344739019,
      "learning_rate": 7.412234421006155e-06,
      "loss": 0.42,
      "step": 5740
    },
    {
      "epoch": 0.35960475422415006,
      "grad_norm": 0.7922046285495647,
      "learning_rate": 7.41134583399034e-06,
      "loss": 0.4487,
      "step": 5741
    },
    {
      "epoch": 0.35966739222349237,
      "grad_norm": 0.8374800154602203,
      "learning_rate": 7.4104571477230825e-06,
      "loss": 0.4579,
      "step": 5742
    },
    {
      "epoch": 0.3597300302228347,
      "grad_norm": 0.8592953838868962,
      "learning_rate": 7.409568362240958e-06,
      "loss": 0.4321,
      "step": 5743
    },
    {
      "epoch": 0.359792668222177,
      "grad_norm": 0.8585822730825504,
      "learning_rate": 7.408679477580556e-06,
      "loss": 0.4349,
      "step": 5744
    },
    {
      "epoch": 0.3598553062215193,
      "grad_norm": 0.7856225873720631,
      "learning_rate": 7.4077904937784595e-06,
      "loss": 0.3912,
      "step": 5745
    },
    {
      "epoch": 0.3599179442208616,
      "grad_norm": 0.7931840113389638,
      "learning_rate": 7.40690141087126e-06,
      "loss": 0.3829,
      "step": 5746
    },
    {
      "epoch": 0.3599805822202039,
      "grad_norm": 0.8895156337722715,
      "learning_rate": 7.406012228895549e-06,
      "loss": 0.4116,
      "step": 5747
    },
    {
      "epoch": 0.3600432202195462,
      "grad_norm": 0.8006874178243293,
      "learning_rate": 7.405122947887929e-06,
      "loss": 0.4911,
      "step": 5748
    },
    {
      "epoch": 0.36010585821888846,
      "grad_norm": 0.8200963105638199,
      "learning_rate": 7.404233567885004e-06,
      "loss": 0.3982,
      "step": 5749
    },
    {
      "epoch": 0.36016849621823077,
      "grad_norm": 0.8108287638415568,
      "learning_rate": 7.4033440889233775e-06,
      "loss": 0.3946,
      "step": 5750
    },
    {
      "epoch": 0.3602311342175731,
      "grad_norm": 0.8635074998438425,
      "learning_rate": 7.402454511039662e-06,
      "loss": 0.4076,
      "step": 5751
    },
    {
      "epoch": 0.3602937722169154,
      "grad_norm": 0.8199622735463522,
      "learning_rate": 7.4015648342704726e-06,
      "loss": 0.4217,
      "step": 5752
    },
    {
      "epoch": 0.3603564102162577,
      "grad_norm": 0.8452473617028118,
      "learning_rate": 7.40067505865243e-06,
      "loss": 0.4493,
      "step": 5753
    },
    {
      "epoch": 0.3604190482156,
      "grad_norm": 0.8278559957704651,
      "learning_rate": 7.399785184222157e-06,
      "loss": 0.42,
      "step": 5754
    },
    {
      "epoch": 0.3604816862149423,
      "grad_norm": 0.8523512383397397,
      "learning_rate": 7.398895211016279e-06,
      "loss": 0.4695,
      "step": 5755
    },
    {
      "epoch": 0.3605443242142846,
      "grad_norm": 0.8302486373842175,
      "learning_rate": 7.398005139071429e-06,
      "loss": 0.4062,
      "step": 5756
    },
    {
      "epoch": 0.36060696221362687,
      "grad_norm": 0.7831676388384917,
      "learning_rate": 7.3971149684242415e-06,
      "loss": 0.4016,
      "step": 5757
    },
    {
      "epoch": 0.3606696002129692,
      "grad_norm": 0.88673683421489,
      "learning_rate": 7.39622469911136e-06,
      "loss": 0.4089,
      "step": 5758
    },
    {
      "epoch": 0.3607322382123115,
      "grad_norm": 0.8267418976912764,
      "learning_rate": 7.395334331169423e-06,
      "loss": 0.4126,
      "step": 5759
    },
    {
      "epoch": 0.3607948762116538,
      "grad_norm": 0.8069706735346918,
      "learning_rate": 7.3944438646350805e-06,
      "loss": 0.3946,
      "step": 5760
    },
    {
      "epoch": 0.3608575142109961,
      "grad_norm": 0.8971818136509114,
      "learning_rate": 7.393553299544984e-06,
      "loss": 0.4293,
      "step": 5761
    },
    {
      "epoch": 0.3609201522103384,
      "grad_norm": 0.8096515303167039,
      "learning_rate": 7.392662635935789e-06,
      "loss": 0.4,
      "step": 5762
    },
    {
      "epoch": 0.3609827902096807,
      "grad_norm": 0.9163410836689029,
      "learning_rate": 7.3917718738441565e-06,
      "loss": 0.4118,
      "step": 5763
    },
    {
      "epoch": 0.361045428209023,
      "grad_norm": 0.7978167514907651,
      "learning_rate": 7.39088101330675e-06,
      "loss": 0.3784,
      "step": 5764
    },
    {
      "epoch": 0.36110806620836533,
      "grad_norm": 0.82454137011966,
      "learning_rate": 7.3899900543602364e-06,
      "loss": 0.4083,
      "step": 5765
    },
    {
      "epoch": 0.3611707042077076,
      "grad_norm": 0.876557109486338,
      "learning_rate": 7.389098997041288e-06,
      "loss": 0.4435,
      "step": 5766
    },
    {
      "epoch": 0.3612333422070499,
      "grad_norm": 0.9111743810309809,
      "learning_rate": 7.388207841386583e-06,
      "loss": 0.4047,
      "step": 5767
    },
    {
      "epoch": 0.3612959802063922,
      "grad_norm": 0.7849318530182237,
      "learning_rate": 7.3873165874328005e-06,
      "loss": 0.4138,
      "step": 5768
    },
    {
      "epoch": 0.3613586182057345,
      "grad_norm": 0.8380888978147147,
      "learning_rate": 7.386425235216623e-06,
      "loss": 0.4336,
      "step": 5769
    },
    {
      "epoch": 0.3614212562050768,
      "grad_norm": 0.8143981906393711,
      "learning_rate": 7.38553378477474e-06,
      "loss": 0.4096,
      "step": 5770
    },
    {
      "epoch": 0.3614838942044191,
      "grad_norm": 0.85350584569826,
      "learning_rate": 7.384642236143843e-06,
      "loss": 0.3721,
      "step": 5771
    },
    {
      "epoch": 0.36154653220376143,
      "grad_norm": 0.8640320708221502,
      "learning_rate": 7.38375058936063e-06,
      "loss": 0.4119,
      "step": 5772
    },
    {
      "epoch": 0.36160917020310374,
      "grad_norm": 0.8292969052620064,
      "learning_rate": 7.3828588444618e-06,
      "loss": 0.4307,
      "step": 5773
    },
    {
      "epoch": 0.361671808202446,
      "grad_norm": 0.7202020549364183,
      "learning_rate": 7.381967001484057e-06,
      "loss": 0.3531,
      "step": 5774
    },
    {
      "epoch": 0.3617344462017883,
      "grad_norm": 0.7819569622450845,
      "learning_rate": 7.38107506046411e-06,
      "loss": 0.3779,
      "step": 5775
    },
    {
      "epoch": 0.3617970842011306,
      "grad_norm": 0.8017148472522285,
      "learning_rate": 7.380183021438673e-06,
      "loss": 0.4432,
      "step": 5776
    },
    {
      "epoch": 0.3618597222004729,
      "grad_norm": 0.7642819400070583,
      "learning_rate": 7.3792908844444596e-06,
      "loss": 0.3954,
      "step": 5777
    },
    {
      "epoch": 0.3619223601998152,
      "grad_norm": 0.7034636008226939,
      "learning_rate": 7.378398649518193e-06,
      "loss": 0.4711,
      "step": 5778
    },
    {
      "epoch": 0.3619849981991575,
      "grad_norm": 0.7988594195660303,
      "learning_rate": 7.377506316696596e-06,
      "loss": 0.4008,
      "step": 5779
    },
    {
      "epoch": 0.36204763619849983,
      "grad_norm": 0.77879272726771,
      "learning_rate": 7.376613886016397e-06,
      "loss": 0.381,
      "step": 5780
    },
    {
      "epoch": 0.36211027419784214,
      "grad_norm": 0.8783896651420762,
      "learning_rate": 7.375721357514332e-06,
      "loss": 0.4084,
      "step": 5781
    },
    {
      "epoch": 0.36217291219718445,
      "grad_norm": 0.9010149762858556,
      "learning_rate": 7.374828731227131e-06,
      "loss": 0.4026,
      "step": 5782
    },
    {
      "epoch": 0.3622355501965267,
      "grad_norm": 0.8820445097272199,
      "learning_rate": 7.373936007191543e-06,
      "loss": 0.4107,
      "step": 5783
    },
    {
      "epoch": 0.362298188195869,
      "grad_norm": 0.8215495338436027,
      "learning_rate": 7.373043185444305e-06,
      "loss": 0.4483,
      "step": 5784
    },
    {
      "epoch": 0.3623608261952113,
      "grad_norm": 0.8083781516946685,
      "learning_rate": 7.372150266022169e-06,
      "loss": 0.3943,
      "step": 5785
    },
    {
      "epoch": 0.3624234641945536,
      "grad_norm": 0.8315231412923928,
      "learning_rate": 7.371257248961888e-06,
      "loss": 0.4507,
      "step": 5786
    },
    {
      "epoch": 0.36248610219389593,
      "grad_norm": 0.8216093323594205,
      "learning_rate": 7.370364134300221e-06,
      "loss": 0.3836,
      "step": 5787
    },
    {
      "epoch": 0.36254874019323824,
      "grad_norm": 0.79723920736812,
      "learning_rate": 7.3694709220739245e-06,
      "loss": 0.4223,
      "step": 5788
    },
    {
      "epoch": 0.36261137819258055,
      "grad_norm": 0.7385850929622145,
      "learning_rate": 7.368577612319764e-06,
      "loss": 0.4645,
      "step": 5789
    },
    {
      "epoch": 0.36267401619192285,
      "grad_norm": 0.8470352507252109,
      "learning_rate": 7.36768420507451e-06,
      "loss": 0.4516,
      "step": 5790
    },
    {
      "epoch": 0.3627366541912651,
      "grad_norm": 0.8076105137359078,
      "learning_rate": 7.366790700374935e-06,
      "loss": 0.4187,
      "step": 5791
    },
    {
      "epoch": 0.3627992921906074,
      "grad_norm": 0.763478732323477,
      "learning_rate": 7.365897098257817e-06,
      "loss": 0.3935,
      "step": 5792
    },
    {
      "epoch": 0.3628619301899497,
      "grad_norm": 0.831956689166088,
      "learning_rate": 7.365003398759933e-06,
      "loss": 0.4334,
      "step": 5793
    },
    {
      "epoch": 0.36292456818929203,
      "grad_norm": 0.8556300177484369,
      "learning_rate": 7.364109601918071e-06,
      "loss": 0.4291,
      "step": 5794
    },
    {
      "epoch": 0.36298720618863434,
      "grad_norm": 0.9864569970439634,
      "learning_rate": 7.363215707769018e-06,
      "loss": 0.4349,
      "step": 5795
    },
    {
      "epoch": 0.36304984418797664,
      "grad_norm": 1.021058655069945,
      "learning_rate": 7.36232171634957e-06,
      "loss": 0.4409,
      "step": 5796
    },
    {
      "epoch": 0.36311248218731895,
      "grad_norm": 0.7708827836524246,
      "learning_rate": 7.36142762769652e-06,
      "loss": 0.4094,
      "step": 5797
    },
    {
      "epoch": 0.36317512018666126,
      "grad_norm": 0.7267692896778561,
      "learning_rate": 7.360533441846671e-06,
      "loss": 0.4666,
      "step": 5798
    },
    {
      "epoch": 0.3632377581860035,
      "grad_norm": 0.8135731451056707,
      "learning_rate": 7.359639158836828e-06,
      "loss": 0.4531,
      "step": 5799
    },
    {
      "epoch": 0.3633003961853458,
      "grad_norm": 0.8943867404493698,
      "learning_rate": 7.358744778703798e-06,
      "loss": 0.4025,
      "step": 5800
    },
    {
      "epoch": 0.3633630341846881,
      "grad_norm": 0.8251800709100707,
      "learning_rate": 7.357850301484397e-06,
      "loss": 0.4348,
      "step": 5801
    },
    {
      "epoch": 0.36342567218403043,
      "grad_norm": 0.8573312744480504,
      "learning_rate": 7.356955727215438e-06,
      "loss": 0.4072,
      "step": 5802
    },
    {
      "epoch": 0.36348831018337274,
      "grad_norm": 0.7668219700790058,
      "learning_rate": 7.356061055933744e-06,
      "loss": 0.3813,
      "step": 5803
    },
    {
      "epoch": 0.36355094818271505,
      "grad_norm": 0.8623986841295415,
      "learning_rate": 7.355166287676141e-06,
      "loss": 0.396,
      "step": 5804
    },
    {
      "epoch": 0.36361358618205736,
      "grad_norm": 0.8301702038869436,
      "learning_rate": 7.354271422479455e-06,
      "loss": 0.4193,
      "step": 5805
    },
    {
      "epoch": 0.36367622418139967,
      "grad_norm": 0.6609038000162436,
      "learning_rate": 7.3533764603805214e-06,
      "loss": 0.5061,
      "step": 5806
    },
    {
      "epoch": 0.363738862180742,
      "grad_norm": 0.8570145812874964,
      "learning_rate": 7.352481401416175e-06,
      "loss": 0.4609,
      "step": 5807
    },
    {
      "epoch": 0.3638015001800842,
      "grad_norm": 0.8586510401414492,
      "learning_rate": 7.351586245623258e-06,
      "loss": 0.4187,
      "step": 5808
    },
    {
      "epoch": 0.36386413817942653,
      "grad_norm": 0.8767083788001556,
      "learning_rate": 7.350690993038615e-06,
      "loss": 0.4186,
      "step": 5809
    },
    {
      "epoch": 0.36392677617876884,
      "grad_norm": 0.8020881982498563,
      "learning_rate": 7.3497956436990934e-06,
      "loss": 0.3956,
      "step": 5810
    },
    {
      "epoch": 0.36398941417811115,
      "grad_norm": 0.8586912471994703,
      "learning_rate": 7.348900197641548e-06,
      "loss": 0.4584,
      "step": 5811
    },
    {
      "epoch": 0.36405205217745346,
      "grad_norm": 0.8312683529608528,
      "learning_rate": 7.348004654902833e-06,
      "loss": 0.4097,
      "step": 5812
    },
    {
      "epoch": 0.36411469017679576,
      "grad_norm": 0.8087418026138633,
      "learning_rate": 7.347109015519812e-06,
      "loss": 0.4343,
      "step": 5813
    },
    {
      "epoch": 0.36417732817613807,
      "grad_norm": 0.815462457550286,
      "learning_rate": 7.346213279529349e-06,
      "loss": 0.4157,
      "step": 5814
    },
    {
      "epoch": 0.3642399661754804,
      "grad_norm": 0.7722242127854525,
      "learning_rate": 7.34531744696831e-06,
      "loss": 0.3775,
      "step": 5815
    },
    {
      "epoch": 0.36430260417482263,
      "grad_norm": 0.8759691293868902,
      "learning_rate": 7.34442151787357e-06,
      "loss": 0.4209,
      "step": 5816
    },
    {
      "epoch": 0.36436524217416494,
      "grad_norm": 0.6853713436639753,
      "learning_rate": 7.343525492282005e-06,
      "loss": 0.4864,
      "step": 5817
    },
    {
      "epoch": 0.36442788017350725,
      "grad_norm": 0.6757809452374269,
      "learning_rate": 7.342629370230497e-06,
      "loss": 0.4838,
      "step": 5818
    },
    {
      "epoch": 0.36449051817284955,
      "grad_norm": 0.8125929688384032,
      "learning_rate": 7.3417331517559295e-06,
      "loss": 0.4067,
      "step": 5819
    },
    {
      "epoch": 0.36455315617219186,
      "grad_norm": 0.8666562889749656,
      "learning_rate": 7.340836836895189e-06,
      "loss": 0.4492,
      "step": 5820
    },
    {
      "epoch": 0.36461579417153417,
      "grad_norm": 0.8527590338337774,
      "learning_rate": 7.33994042568517e-06,
      "loss": 0.4157,
      "step": 5821
    },
    {
      "epoch": 0.3646784321708765,
      "grad_norm": 0.7937788529369343,
      "learning_rate": 7.339043918162771e-06,
      "loss": 0.4265,
      "step": 5822
    },
    {
      "epoch": 0.3647410701702188,
      "grad_norm": 0.8164547089102786,
      "learning_rate": 7.338147314364889e-06,
      "loss": 0.4455,
      "step": 5823
    },
    {
      "epoch": 0.3648037081695611,
      "grad_norm": 0.8094705184169673,
      "learning_rate": 7.337250614328428e-06,
      "loss": 0.4293,
      "step": 5824
    },
    {
      "epoch": 0.36486634616890334,
      "grad_norm": 0.7716458269561888,
      "learning_rate": 7.3363538180903e-06,
      "loss": 0.3712,
      "step": 5825
    },
    {
      "epoch": 0.36492898416824565,
      "grad_norm": 0.7473933574553272,
      "learning_rate": 7.3354569256874145e-06,
      "loss": 0.4631,
      "step": 5826
    },
    {
      "epoch": 0.36499162216758796,
      "grad_norm": 0.8393860766530554,
      "learning_rate": 7.334559937156687e-06,
      "loss": 0.4059,
      "step": 5827
    },
    {
      "epoch": 0.36505426016693027,
      "grad_norm": 0.9043140282885366,
      "learning_rate": 7.33366285253504e-06,
      "loss": 0.4373,
      "step": 5828
    },
    {
      "epoch": 0.3651168981662726,
      "grad_norm": 0.8934888126198045,
      "learning_rate": 7.332765671859399e-06,
      "loss": 0.3904,
      "step": 5829
    },
    {
      "epoch": 0.3651795361656149,
      "grad_norm": 0.8015520732754772,
      "learning_rate": 7.331868395166688e-06,
      "loss": 0.3928,
      "step": 5830
    },
    {
      "epoch": 0.3652421741649572,
      "grad_norm": 0.9596202268888693,
      "learning_rate": 7.330971022493841e-06,
      "loss": 0.4156,
      "step": 5831
    },
    {
      "epoch": 0.3653048121642995,
      "grad_norm": 0.8584335353056964,
      "learning_rate": 7.330073553877794e-06,
      "loss": 0.4199,
      "step": 5832
    },
    {
      "epoch": 0.36536745016364175,
      "grad_norm": 0.8966873572523307,
      "learning_rate": 7.329175989355488e-06,
      "loss": 0.4283,
      "step": 5833
    },
    {
      "epoch": 0.36543008816298406,
      "grad_norm": 0.9279153280143996,
      "learning_rate": 7.328278328963866e-06,
      "loss": 0.4225,
      "step": 5834
    },
    {
      "epoch": 0.36549272616232636,
      "grad_norm": 0.7696098666303545,
      "learning_rate": 7.327380572739875e-06,
      "loss": 0.4127,
      "step": 5835
    },
    {
      "epoch": 0.36555536416166867,
      "grad_norm": 0.8567700847572169,
      "learning_rate": 7.326482720720468e-06,
      "loss": 0.4045,
      "step": 5836
    },
    {
      "epoch": 0.365618002161011,
      "grad_norm": 0.9717105661405655,
      "learning_rate": 7.3255847729426e-06,
      "loss": 0.4284,
      "step": 5837
    },
    {
      "epoch": 0.3656806401603533,
      "grad_norm": 0.8694473437236772,
      "learning_rate": 7.324686729443231e-06,
      "loss": 0.4238,
      "step": 5838
    },
    {
      "epoch": 0.3657432781596956,
      "grad_norm": 0.665677355192219,
      "learning_rate": 7.323788590259326e-06,
      "loss": 0.5047,
      "step": 5839
    },
    {
      "epoch": 0.3658059161590379,
      "grad_norm": 0.8686653147360286,
      "learning_rate": 7.3228903554278495e-06,
      "loss": 0.4308,
      "step": 5840
    },
    {
      "epoch": 0.36586855415838015,
      "grad_norm": 0.6488696459559853,
      "learning_rate": 7.3219920249857755e-06,
      "loss": 0.4815,
      "step": 5841
    },
    {
      "epoch": 0.36593119215772246,
      "grad_norm": 0.8093000889942316,
      "learning_rate": 7.32109359897008e-06,
      "loss": 0.4046,
      "step": 5842
    },
    {
      "epoch": 0.36599383015706477,
      "grad_norm": 0.75684909936627,
      "learning_rate": 7.32019507741774e-06,
      "loss": 0.3895,
      "step": 5843
    },
    {
      "epoch": 0.3660564681564071,
      "grad_norm": 0.6603426259183253,
      "learning_rate": 7.31929646036574e-06,
      "loss": 0.4647,
      "step": 5844
    },
    {
      "epoch": 0.3661191061557494,
      "grad_norm": 0.8647438027883455,
      "learning_rate": 7.318397747851067e-06,
      "loss": 0.4715,
      "step": 5845
    },
    {
      "epoch": 0.3661817441550917,
      "grad_norm": 0.8464675668910929,
      "learning_rate": 7.317498939910712e-06,
      "loss": 0.4269,
      "step": 5846
    },
    {
      "epoch": 0.366244382154434,
      "grad_norm": 0.8466271226624354,
      "learning_rate": 7.316600036581673e-06,
      "loss": 0.4122,
      "step": 5847
    },
    {
      "epoch": 0.3663070201537763,
      "grad_norm": 0.8141653428358245,
      "learning_rate": 7.3157010379009444e-06,
      "loss": 0.4294,
      "step": 5848
    },
    {
      "epoch": 0.3663696581531186,
      "grad_norm": 0.8576372888742911,
      "learning_rate": 7.314801943905531e-06,
      "loss": 0.4527,
      "step": 5849
    },
    {
      "epoch": 0.36643229615246087,
      "grad_norm": 0.7730206480443678,
      "learning_rate": 7.3139027546324405e-06,
      "loss": 0.3646,
      "step": 5850
    },
    {
      "epoch": 0.3664949341518032,
      "grad_norm": 0.8474971628587161,
      "learning_rate": 7.313003470118684e-06,
      "loss": 0.4418,
      "step": 5851
    },
    {
      "epoch": 0.3665575721511455,
      "grad_norm": 0.8183426060446829,
      "learning_rate": 7.312104090401274e-06,
      "loss": 0.4112,
      "step": 5852
    },
    {
      "epoch": 0.3666202101504878,
      "grad_norm": 0.8808331895823226,
      "learning_rate": 7.311204615517231e-06,
      "loss": 0.4113,
      "step": 5853
    },
    {
      "epoch": 0.3666828481498301,
      "grad_norm": 0.7669357855449815,
      "learning_rate": 7.310305045503576e-06,
      "loss": 0.4134,
      "step": 5854
    },
    {
      "epoch": 0.3667454861491724,
      "grad_norm": 0.8250859737907706,
      "learning_rate": 7.309405380397338e-06,
      "loss": 0.4152,
      "step": 5855
    },
    {
      "epoch": 0.3668081241485147,
      "grad_norm": 0.8346846183548109,
      "learning_rate": 7.308505620235546e-06,
      "loss": 0.4188,
      "step": 5856
    },
    {
      "epoch": 0.366870762147857,
      "grad_norm": 0.8212067233711811,
      "learning_rate": 7.307605765055235e-06,
      "loss": 0.4562,
      "step": 5857
    },
    {
      "epoch": 0.3669334001471993,
      "grad_norm": 0.7128974662224422,
      "learning_rate": 7.30670581489344e-06,
      "loss": 0.3592,
      "step": 5858
    },
    {
      "epoch": 0.3669960381465416,
      "grad_norm": 0.691902260068284,
      "learning_rate": 7.305805769787207e-06,
      "loss": 0.4879,
      "step": 5859
    },
    {
      "epoch": 0.3670586761458839,
      "grad_norm": 0.8323101708959559,
      "learning_rate": 7.30490562977358e-06,
      "loss": 0.4147,
      "step": 5860
    },
    {
      "epoch": 0.3671213141452262,
      "grad_norm": 0.8076837671059329,
      "learning_rate": 7.30400539488961e-06,
      "loss": 0.4192,
      "step": 5861
    },
    {
      "epoch": 0.3671839521445685,
      "grad_norm": 0.6537604821161068,
      "learning_rate": 7.303105065172349e-06,
      "loss": 0.4846,
      "step": 5862
    },
    {
      "epoch": 0.3672465901439108,
      "grad_norm": 0.908680488952784,
      "learning_rate": 7.302204640658856e-06,
      "loss": 0.4429,
      "step": 5863
    },
    {
      "epoch": 0.3673092281432531,
      "grad_norm": 0.759758158721983,
      "learning_rate": 7.301304121386194e-06,
      "loss": 0.4068,
      "step": 5864
    },
    {
      "epoch": 0.3673718661425954,
      "grad_norm": 0.7964326359627816,
      "learning_rate": 7.300403507391427e-06,
      "loss": 0.3914,
      "step": 5865
    },
    {
      "epoch": 0.3674345041419377,
      "grad_norm": 0.9341805349739256,
      "learning_rate": 7.299502798711625e-06,
      "loss": 0.4387,
      "step": 5866
    },
    {
      "epoch": 0.36749714214128,
      "grad_norm": 0.7930358232193784,
      "learning_rate": 7.29860199538386e-06,
      "loss": 0.3616,
      "step": 5867
    },
    {
      "epoch": 0.3675597801406223,
      "grad_norm": 0.8141205904485523,
      "learning_rate": 7.297701097445211e-06,
      "loss": 0.4209,
      "step": 5868
    },
    {
      "epoch": 0.3676224181399646,
      "grad_norm": 0.7310754969714806,
      "learning_rate": 7.2968001049327575e-06,
      "loss": 0.4701,
      "step": 5869
    },
    {
      "epoch": 0.3676850561393069,
      "grad_norm": 0.8375548393329895,
      "learning_rate": 7.2958990178835875e-06,
      "loss": 0.4315,
      "step": 5870
    },
    {
      "epoch": 0.3677476941386492,
      "grad_norm": 0.8826472485521139,
      "learning_rate": 7.294997836334785e-06,
      "loss": 0.4139,
      "step": 5871
    },
    {
      "epoch": 0.3678103321379915,
      "grad_norm": 0.8821608638098751,
      "learning_rate": 7.294096560323447e-06,
      "loss": 0.4424,
      "step": 5872
    },
    {
      "epoch": 0.36787297013733383,
      "grad_norm": 0.9092539790642856,
      "learning_rate": 7.293195189886668e-06,
      "loss": 0.4543,
      "step": 5873
    },
    {
      "epoch": 0.36793560813667614,
      "grad_norm": 0.8422887332268084,
      "learning_rate": 7.292293725061551e-06,
      "loss": 0.4809,
      "step": 5874
    },
    {
      "epoch": 0.3679982461360184,
      "grad_norm": 0.8002230735990625,
      "learning_rate": 7.291392165885197e-06,
      "loss": 0.4089,
      "step": 5875
    },
    {
      "epoch": 0.3680608841353607,
      "grad_norm": 0.8117942020032257,
      "learning_rate": 7.2904905123947176e-06,
      "loss": 0.4005,
      "step": 5876
    },
    {
      "epoch": 0.368123522134703,
      "grad_norm": 0.8098001324248731,
      "learning_rate": 7.289588764627222e-06,
      "loss": 0.3741,
      "step": 5877
    },
    {
      "epoch": 0.3681861601340453,
      "grad_norm": 0.8016710952990651,
      "learning_rate": 7.28868692261983e-06,
      "loss": 0.4286,
      "step": 5878
    },
    {
      "epoch": 0.3682487981333876,
      "grad_norm": 0.8269977001371119,
      "learning_rate": 7.2877849864096585e-06,
      "loss": 0.3982,
      "step": 5879
    },
    {
      "epoch": 0.36831143613272993,
      "grad_norm": 0.7979685358693286,
      "learning_rate": 7.286882956033831e-06,
      "loss": 0.4013,
      "step": 5880
    },
    {
      "epoch": 0.36837407413207224,
      "grad_norm": 0.8062268842875226,
      "learning_rate": 7.2859808315294776e-06,
      "loss": 0.4014,
      "step": 5881
    },
    {
      "epoch": 0.36843671213141455,
      "grad_norm": 0.7519757497505836,
      "learning_rate": 7.285078612933729e-06,
      "loss": 0.3816,
      "step": 5882
    },
    {
      "epoch": 0.3684993501307568,
      "grad_norm": 0.8184588524919215,
      "learning_rate": 7.284176300283721e-06,
      "loss": 0.4338,
      "step": 5883
    },
    {
      "epoch": 0.3685619881300991,
      "grad_norm": 0.8488849599847593,
      "learning_rate": 7.283273893616591e-06,
      "loss": 0.4166,
      "step": 5884
    },
    {
      "epoch": 0.3686246261294414,
      "grad_norm": 0.8449615099653043,
      "learning_rate": 7.282371392969485e-06,
      "loss": 0.4289,
      "step": 5885
    },
    {
      "epoch": 0.3686872641287837,
      "grad_norm": 0.8107458560945113,
      "learning_rate": 7.281468798379549e-06,
      "loss": 0.4326,
      "step": 5886
    },
    {
      "epoch": 0.36874990212812603,
      "grad_norm": 0.6948261731518378,
      "learning_rate": 7.280566109883933e-06,
      "loss": 0.4679,
      "step": 5887
    },
    {
      "epoch": 0.36881254012746834,
      "grad_norm": 0.775914721634007,
      "learning_rate": 7.279663327519792e-06,
      "loss": 0.4071,
      "step": 5888
    },
    {
      "epoch": 0.36887517812681064,
      "grad_norm": 0.8737911301108886,
      "learning_rate": 7.278760451324287e-06,
      "loss": 0.4199,
      "step": 5889
    },
    {
      "epoch": 0.36893781612615295,
      "grad_norm": 0.8115215156419473,
      "learning_rate": 7.2778574813345785e-06,
      "loss": 0.3871,
      "step": 5890
    },
    {
      "epoch": 0.36900045412549526,
      "grad_norm": 0.8152902092139018,
      "learning_rate": 7.2769544175878325e-06,
      "loss": 0.3965,
      "step": 5891
    },
    {
      "epoch": 0.3690630921248375,
      "grad_norm": 0.798762199245858,
      "learning_rate": 7.27605126012122e-06,
      "loss": 0.3654,
      "step": 5892
    },
    {
      "epoch": 0.3691257301241798,
      "grad_norm": 0.6598247756317287,
      "learning_rate": 7.275148008971916e-06,
      "loss": 0.4628,
      "step": 5893
    },
    {
      "epoch": 0.3691883681235221,
      "grad_norm": 0.8391945367941989,
      "learning_rate": 7.2742446641770985e-06,
      "loss": 0.4211,
      "step": 5894
    },
    {
      "epoch": 0.36925100612286443,
      "grad_norm": 0.7959421676677423,
      "learning_rate": 7.273341225773947e-06,
      "loss": 0.4259,
      "step": 5895
    },
    {
      "epoch": 0.36931364412220674,
      "grad_norm": 0.9527469021722614,
      "learning_rate": 7.2724376937996485e-06,
      "loss": 0.4002,
      "step": 5896
    },
    {
      "epoch": 0.36937628212154905,
      "grad_norm": 0.8926921970620123,
      "learning_rate": 7.271534068291393e-06,
      "loss": 0.4206,
      "step": 5897
    },
    {
      "epoch": 0.36943892012089136,
      "grad_norm": 0.8218255582659042,
      "learning_rate": 7.2706303492863736e-06,
      "loss": 0.4506,
      "step": 5898
    },
    {
      "epoch": 0.36950155812023366,
      "grad_norm": 0.811171425744123,
      "learning_rate": 7.2697265368217886e-06,
      "loss": 0.3765,
      "step": 5899
    },
    {
      "epoch": 0.3695641961195759,
      "grad_norm": 0.8379330333341313,
      "learning_rate": 7.268822630934837e-06,
      "loss": 0.423,
      "step": 5900
    },
    {
      "epoch": 0.3696268341189182,
      "grad_norm": 0.801316722967744,
      "learning_rate": 7.267918631662725e-06,
      "loss": 0.4043,
      "step": 5901
    },
    {
      "epoch": 0.36968947211826053,
      "grad_norm": 0.8289039724573347,
      "learning_rate": 7.267014539042662e-06,
      "loss": 0.4292,
      "step": 5902
    },
    {
      "epoch": 0.36975211011760284,
      "grad_norm": 0.783131399507439,
      "learning_rate": 7.2661103531118595e-06,
      "loss": 0.4341,
      "step": 5903
    },
    {
      "epoch": 0.36981474811694515,
      "grad_norm": 0.7862667731355101,
      "learning_rate": 7.265206073907533e-06,
      "loss": 0.4064,
      "step": 5904
    },
    {
      "epoch": 0.36987738611628745,
      "grad_norm": 0.8595762475799927,
      "learning_rate": 7.264301701466903e-06,
      "loss": 0.4327,
      "step": 5905
    },
    {
      "epoch": 0.36994002411562976,
      "grad_norm": 0.8638465984249799,
      "learning_rate": 7.263397235827199e-06,
      "loss": 0.4193,
      "step": 5906
    },
    {
      "epoch": 0.37000266211497207,
      "grad_norm": 0.6134742201482536,
      "learning_rate": 7.262492677025642e-06,
      "loss": 0.4861,
      "step": 5907
    },
    {
      "epoch": 0.3700653001143143,
      "grad_norm": 0.857971705967142,
      "learning_rate": 7.261588025099466e-06,
      "loss": 0.4223,
      "step": 5908
    },
    {
      "epoch": 0.37012793811365663,
      "grad_norm": 0.8779013468859642,
      "learning_rate": 7.260683280085909e-06,
      "loss": 0.4056,
      "step": 5909
    },
    {
      "epoch": 0.37019057611299894,
      "grad_norm": 0.8281734347793294,
      "learning_rate": 7.2597784420222065e-06,
      "loss": 0.4237,
      "step": 5910
    },
    {
      "epoch": 0.37025321411234124,
      "grad_norm": 0.7663966940390077,
      "learning_rate": 7.2588735109456055e-06,
      "loss": 0.4015,
      "step": 5911
    },
    {
      "epoch": 0.37031585211168355,
      "grad_norm": 0.8399477330063523,
      "learning_rate": 7.257968486893351e-06,
      "loss": 0.3749,
      "step": 5912
    },
    {
      "epoch": 0.37037849011102586,
      "grad_norm": 0.8327990658869963,
      "learning_rate": 7.257063369902694e-06,
      "loss": 0.403,
      "step": 5913
    },
    {
      "epoch": 0.37044112811036817,
      "grad_norm": 0.737305480290575,
      "learning_rate": 7.2561581600108885e-06,
      "loss": 0.3846,
      "step": 5914
    },
    {
      "epoch": 0.3705037661097105,
      "grad_norm": 0.8631055803683375,
      "learning_rate": 7.255252857255197e-06,
      "loss": 0.461,
      "step": 5915
    },
    {
      "epoch": 0.3705664041090528,
      "grad_norm": 0.8078871194487709,
      "learning_rate": 7.254347461672879e-06,
      "loss": 0.4235,
      "step": 5916
    },
    {
      "epoch": 0.37062904210839503,
      "grad_norm": 0.8453306979348787,
      "learning_rate": 7.253441973301201e-06,
      "loss": 0.4152,
      "step": 5917
    },
    {
      "epoch": 0.37069168010773734,
      "grad_norm": 0.7988285169889959,
      "learning_rate": 7.252536392177431e-06,
      "loss": 0.3926,
      "step": 5918
    },
    {
      "epoch": 0.37075431810707965,
      "grad_norm": 0.8695886070928325,
      "learning_rate": 7.251630718338847e-06,
      "loss": 0.4405,
      "step": 5919
    },
    {
      "epoch": 0.37081695610642196,
      "grad_norm": 0.8587568407842109,
      "learning_rate": 7.250724951822724e-06,
      "loss": 0.407,
      "step": 5920
    },
    {
      "epoch": 0.37087959410576427,
      "grad_norm": 0.8804560514303142,
      "learning_rate": 7.2498190926663446e-06,
      "loss": 0.4198,
      "step": 5921
    },
    {
      "epoch": 0.3709422321051066,
      "grad_norm": 0.911632468108049,
      "learning_rate": 7.2489131409069935e-06,
      "loss": 0.4298,
      "step": 5922
    },
    {
      "epoch": 0.3710048701044489,
      "grad_norm": 0.8092471983223791,
      "learning_rate": 7.248007096581959e-06,
      "loss": 0.3866,
      "step": 5923
    },
    {
      "epoch": 0.3710675081037912,
      "grad_norm": 0.8006646524489005,
      "learning_rate": 7.247100959728537e-06,
      "loss": 0.4586,
      "step": 5924
    },
    {
      "epoch": 0.37113014610313344,
      "grad_norm": 0.793886067128531,
      "learning_rate": 7.246194730384022e-06,
      "loss": 0.3867,
      "step": 5925
    },
    {
      "epoch": 0.37119278410247575,
      "grad_norm": 0.849931268035709,
      "learning_rate": 7.245288408585714e-06,
      "loss": 0.4385,
      "step": 5926
    },
    {
      "epoch": 0.37125542210181806,
      "grad_norm": 0.7998762718192635,
      "learning_rate": 7.244381994370919e-06,
      "loss": 0.4262,
      "step": 5927
    },
    {
      "epoch": 0.37131806010116036,
      "grad_norm": 0.7938811117898124,
      "learning_rate": 7.243475487776945e-06,
      "loss": 0.4126,
      "step": 5928
    },
    {
      "epoch": 0.37138069810050267,
      "grad_norm": 0.7727012484048016,
      "learning_rate": 7.242568888841104e-06,
      "loss": 0.3942,
      "step": 5929
    },
    {
      "epoch": 0.371443336099845,
      "grad_norm": 0.7435566419437892,
      "learning_rate": 7.24166219760071e-06,
      "loss": 0.394,
      "step": 5930
    },
    {
      "epoch": 0.3715059740991873,
      "grad_norm": 0.8146411481506421,
      "learning_rate": 7.2407554140930834e-06,
      "loss": 0.4013,
      "step": 5931
    },
    {
      "epoch": 0.3715686120985296,
      "grad_norm": 0.7518512796368698,
      "learning_rate": 7.239848538355549e-06,
      "loss": 0.4827,
      "step": 5932
    },
    {
      "epoch": 0.3716312500978719,
      "grad_norm": 0.66645832751465,
      "learning_rate": 7.2389415704254325e-06,
      "loss": 0.4838,
      "step": 5933
    },
    {
      "epoch": 0.37169388809721415,
      "grad_norm": 0.8231231242399363,
      "learning_rate": 7.238034510340068e-06,
      "loss": 0.4136,
      "step": 5934
    },
    {
      "epoch": 0.37175652609655646,
      "grad_norm": 0.7837267627297182,
      "learning_rate": 7.237127358136784e-06,
      "loss": 0.393,
      "step": 5935
    },
    {
      "epoch": 0.37181916409589877,
      "grad_norm": 0.8284244008144561,
      "learning_rate": 7.236220113852926e-06,
      "loss": 0.4107,
      "step": 5936
    },
    {
      "epoch": 0.3718818020952411,
      "grad_norm": 0.8960754602901276,
      "learning_rate": 7.235312777525829e-06,
      "loss": 0.4461,
      "step": 5937
    },
    {
      "epoch": 0.3719444400945834,
      "grad_norm": 0.7196738019444244,
      "learning_rate": 7.234405349192847e-06,
      "loss": 0.4832,
      "step": 5938
    },
    {
      "epoch": 0.3720070780939257,
      "grad_norm": 0.947720121256737,
      "learning_rate": 7.233497828891324e-06,
      "loss": 0.4372,
      "step": 5939
    },
    {
      "epoch": 0.372069716093268,
      "grad_norm": 0.8272155161589095,
      "learning_rate": 7.232590216658617e-06,
      "loss": 0.3737,
      "step": 5940
    },
    {
      "epoch": 0.3721323540926103,
      "grad_norm": 0.7948729943203034,
      "learning_rate": 7.231682512532081e-06,
      "loss": 0.389,
      "step": 5941
    },
    {
      "epoch": 0.37219499209195256,
      "grad_norm": 0.7219801482895765,
      "learning_rate": 7.230774716549079e-06,
      "loss": 0.3745,
      "step": 5942
    },
    {
      "epoch": 0.37225763009129487,
      "grad_norm": 0.8597338524172178,
      "learning_rate": 7.229866828746977e-06,
      "loss": 0.4547,
      "step": 5943
    },
    {
      "epoch": 0.3723202680906372,
      "grad_norm": 0.8713738015157348,
      "learning_rate": 7.22895884916314e-06,
      "loss": 0.4272,
      "step": 5944
    },
    {
      "epoch": 0.3723829060899795,
      "grad_norm": 0.7753103696600225,
      "learning_rate": 7.228050777834945e-06,
      "loss": 0.3684,
      "step": 5945
    },
    {
      "epoch": 0.3724455440893218,
      "grad_norm": 0.8176179768446045,
      "learning_rate": 7.227142614799765e-06,
      "loss": 0.4281,
      "step": 5946
    },
    {
      "epoch": 0.3725081820886641,
      "grad_norm": 0.7762795415319533,
      "learning_rate": 7.226234360094983e-06,
      "loss": 0.4073,
      "step": 5947
    },
    {
      "epoch": 0.3725708200880064,
      "grad_norm": 0.853264080194063,
      "learning_rate": 7.2253260137579805e-06,
      "loss": 0.4269,
      "step": 5948
    },
    {
      "epoch": 0.3726334580873487,
      "grad_norm": 0.8328935710135116,
      "learning_rate": 7.224417575826147e-06,
      "loss": 0.4074,
      "step": 5949
    },
    {
      "epoch": 0.37269609608669096,
      "grad_norm": 0.8589495442026321,
      "learning_rate": 7.223509046336872e-06,
      "loss": 0.4854,
      "step": 5950
    },
    {
      "epoch": 0.37275873408603327,
      "grad_norm": 0.7819843938545763,
      "learning_rate": 7.222600425327553e-06,
      "loss": 0.4095,
      "step": 5951
    },
    {
      "epoch": 0.3728213720853756,
      "grad_norm": 0.8609183123759093,
      "learning_rate": 7.221691712835588e-06,
      "loss": 0.409,
      "step": 5952
    },
    {
      "epoch": 0.3728840100847179,
      "grad_norm": 0.8704140615612768,
      "learning_rate": 7.220782908898379e-06,
      "loss": 0.4503,
      "step": 5953
    },
    {
      "epoch": 0.3729466480840602,
      "grad_norm": 0.8156248217896908,
      "learning_rate": 7.219874013553335e-06,
      "loss": 0.3938,
      "step": 5954
    },
    {
      "epoch": 0.3730092860834025,
      "grad_norm": 0.8840268382419364,
      "learning_rate": 7.218965026837863e-06,
      "loss": 0.4638,
      "step": 5955
    },
    {
      "epoch": 0.3730719240827448,
      "grad_norm": 0.7638120258682345,
      "learning_rate": 7.218055948789379e-06,
      "loss": 0.4215,
      "step": 5956
    },
    {
      "epoch": 0.3731345620820871,
      "grad_norm": 0.810301367076806,
      "learning_rate": 7.2171467794453e-06,
      "loss": 0.3892,
      "step": 5957
    },
    {
      "epoch": 0.3731972000814294,
      "grad_norm": 0.8050189611253956,
      "learning_rate": 7.216237518843049e-06,
      "loss": 0.4424,
      "step": 5958
    },
    {
      "epoch": 0.3732598380807717,
      "grad_norm": 0.7672124704992737,
      "learning_rate": 7.2153281670200505e-06,
      "loss": 0.4002,
      "step": 5959
    },
    {
      "epoch": 0.373322476080114,
      "grad_norm": 0.9374977004603784,
      "learning_rate": 7.214418724013732e-06,
      "loss": 0.4358,
      "step": 5960
    },
    {
      "epoch": 0.3733851140794563,
      "grad_norm": 0.7422521219893502,
      "learning_rate": 7.213509189861527e-06,
      "loss": 0.389,
      "step": 5961
    },
    {
      "epoch": 0.3734477520787986,
      "grad_norm": 0.8584309409135739,
      "learning_rate": 7.212599564600877e-06,
      "loss": 0.4312,
      "step": 5962
    },
    {
      "epoch": 0.3735103900781409,
      "grad_norm": 0.8567057792489023,
      "learning_rate": 7.2116898482692145e-06,
      "loss": 0.4265,
      "step": 5963
    },
    {
      "epoch": 0.3735730280774832,
      "grad_norm": 0.7766606978563286,
      "learning_rate": 7.210780040903988e-06,
      "loss": 0.3859,
      "step": 5964
    },
    {
      "epoch": 0.3736356660768255,
      "grad_norm": 0.8246403208046869,
      "learning_rate": 7.209870142542645e-06,
      "loss": 0.4215,
      "step": 5965
    },
    {
      "epoch": 0.37369830407616783,
      "grad_norm": 0.8653312184058856,
      "learning_rate": 7.208960153222637e-06,
      "loss": 0.4078,
      "step": 5966
    },
    {
      "epoch": 0.3737609420755101,
      "grad_norm": 0.8072252660618803,
      "learning_rate": 7.2080500729814205e-06,
      "loss": 0.4296,
      "step": 5967
    },
    {
      "epoch": 0.3738235800748524,
      "grad_norm": 0.790967828347879,
      "learning_rate": 7.207139901856453e-06,
      "loss": 0.416,
      "step": 5968
    },
    {
      "epoch": 0.3738862180741947,
      "grad_norm": 0.7770474160195014,
      "learning_rate": 7.206229639885195e-06,
      "loss": 0.3468,
      "step": 5969
    },
    {
      "epoch": 0.373948856073537,
      "grad_norm": 0.8783666762491366,
      "learning_rate": 7.2053192871051156e-06,
      "loss": 0.421,
      "step": 5970
    },
    {
      "epoch": 0.3740114940728793,
      "grad_norm": 0.7500110549377933,
      "learning_rate": 7.204408843553688e-06,
      "loss": 0.3776,
      "step": 5971
    },
    {
      "epoch": 0.3740741320722216,
      "grad_norm": 0.8362908942539149,
      "learning_rate": 7.203498309268381e-06,
      "loss": 0.401,
      "step": 5972
    },
    {
      "epoch": 0.37413677007156393,
      "grad_norm": 0.7720378421086996,
      "learning_rate": 7.202587684286675e-06,
      "loss": 0.378,
      "step": 5973
    },
    {
      "epoch": 0.37419940807090624,
      "grad_norm": 0.7795369900634718,
      "learning_rate": 7.201676968646049e-06,
      "loss": 0.3943,
      "step": 5974
    },
    {
      "epoch": 0.3742620460702485,
      "grad_norm": 0.840661895920009,
      "learning_rate": 7.200766162383994e-06,
      "loss": 0.4833,
      "step": 5975
    },
    {
      "epoch": 0.3743246840695908,
      "grad_norm": 0.8623420522385441,
      "learning_rate": 7.199855265537994e-06,
      "loss": 0.4229,
      "step": 5976
    },
    {
      "epoch": 0.3743873220689331,
      "grad_norm": 0.7619395228810419,
      "learning_rate": 7.198944278145542e-06,
      "loss": 0.3791,
      "step": 5977
    },
    {
      "epoch": 0.3744499600682754,
      "grad_norm": 0.8291581368871479,
      "learning_rate": 7.198033200244136e-06,
      "loss": 0.425,
      "step": 5978
    },
    {
      "epoch": 0.3745125980676177,
      "grad_norm": 0.8407653629188556,
      "learning_rate": 7.197122031871275e-06,
      "loss": 0.4389,
      "step": 5979
    },
    {
      "epoch": 0.37457523606696,
      "grad_norm": 0.8003554860253593,
      "learning_rate": 7.196210773064465e-06,
      "loss": 0.3946,
      "step": 5980
    },
    {
      "epoch": 0.37463787406630233,
      "grad_norm": 0.7716708086916421,
      "learning_rate": 7.195299423861212e-06,
      "loss": 0.3947,
      "step": 5981
    },
    {
      "epoch": 0.37470051206564464,
      "grad_norm": 0.8014232892123095,
      "learning_rate": 7.194387984299025e-06,
      "loss": 0.4068,
      "step": 5982
    },
    {
      "epoch": 0.37476315006498695,
      "grad_norm": 0.7953109508165411,
      "learning_rate": 7.193476454415422e-06,
      "loss": 0.4037,
      "step": 5983
    },
    {
      "epoch": 0.3748257880643292,
      "grad_norm": 0.8450707997875546,
      "learning_rate": 7.192564834247921e-06,
      "loss": 0.4103,
      "step": 5984
    },
    {
      "epoch": 0.3748884260636715,
      "grad_norm": 0.7667314863367545,
      "learning_rate": 7.191653123834046e-06,
      "loss": 0.4196,
      "step": 5985
    },
    {
      "epoch": 0.3749510640630138,
      "grad_norm": 0.827551687652118,
      "learning_rate": 7.19074132321132e-06,
      "loss": 0.4371,
      "step": 5986
    },
    {
      "epoch": 0.3750137020623561,
      "grad_norm": 0.8600924769392306,
      "learning_rate": 7.189829432417275e-06,
      "loss": 0.4342,
      "step": 5987
    },
    {
      "epoch": 0.37507634006169843,
      "grad_norm": 0.8405123451978119,
      "learning_rate": 7.188917451489443e-06,
      "loss": 0.411,
      "step": 5988
    },
    {
      "epoch": 0.37513897806104074,
      "grad_norm": 0.8024844057152233,
      "learning_rate": 7.188005380465365e-06,
      "loss": 0.3909,
      "step": 5989
    },
    {
      "epoch": 0.37520161606038305,
      "grad_norm": 0.8132854521487614,
      "learning_rate": 7.187093219382577e-06,
      "loss": 0.3606,
      "step": 5990
    },
    {
      "epoch": 0.37526425405972536,
      "grad_norm": 0.8163970649569208,
      "learning_rate": 7.186180968278628e-06,
      "loss": 0.4015,
      "step": 5991
    },
    {
      "epoch": 0.3753268920590676,
      "grad_norm": 0.8238552855729523,
      "learning_rate": 7.185268627191062e-06,
      "loss": 0.3767,
      "step": 5992
    },
    {
      "epoch": 0.3753895300584099,
      "grad_norm": 0.9019066648495111,
      "learning_rate": 7.184356196157436e-06,
      "loss": 0.4374,
      "step": 5993
    },
    {
      "epoch": 0.3754521680577522,
      "grad_norm": 0.8273080552809318,
      "learning_rate": 7.183443675215304e-06,
      "loss": 0.4264,
      "step": 5994
    },
    {
      "epoch": 0.37551480605709453,
      "grad_norm": 0.8205476478094733,
      "learning_rate": 7.182531064402223e-06,
      "loss": 0.4394,
      "step": 5995
    },
    {
      "epoch": 0.37557744405643684,
      "grad_norm": 0.8162142829424549,
      "learning_rate": 7.181618363755759e-06,
      "loss": 0.4364,
      "step": 5996
    },
    {
      "epoch": 0.37564008205577915,
      "grad_norm": 0.8572402509933548,
      "learning_rate": 7.180705573313479e-06,
      "loss": 0.421,
      "step": 5997
    },
    {
      "epoch": 0.37570272005512145,
      "grad_norm": 0.8480598814961163,
      "learning_rate": 7.179792693112953e-06,
      "loss": 0.4183,
      "step": 5998
    },
    {
      "epoch": 0.37576535805446376,
      "grad_norm": 0.6516216564562132,
      "learning_rate": 7.178879723191754e-06,
      "loss": 0.4646,
      "step": 5999
    },
    {
      "epoch": 0.37582799605380607,
      "grad_norm": 0.8533128125278401,
      "learning_rate": 7.177966663587461e-06,
      "loss": 0.4271,
      "step": 6000
    },
    {
      "epoch": 0.3758906340531483,
      "grad_norm": 0.7500389432066854,
      "learning_rate": 7.177053514337657e-06,
      "loss": 0.3945,
      "step": 6001
    },
    {
      "epoch": 0.37595327205249063,
      "grad_norm": 0.6790989836247983,
      "learning_rate": 7.176140275479926e-06,
      "loss": 0.4564,
      "step": 6002
    },
    {
      "epoch": 0.37601591005183294,
      "grad_norm": 0.7913516350851612,
      "learning_rate": 7.175226947051856e-06,
      "loss": 0.395,
      "step": 6003
    },
    {
      "epoch": 0.37607854805117524,
      "grad_norm": 0.6901631625392178,
      "learning_rate": 7.174313529091044e-06,
      "loss": 0.4873,
      "step": 6004
    },
    {
      "epoch": 0.37614118605051755,
      "grad_norm": 0.8415366647191069,
      "learning_rate": 7.173400021635082e-06,
      "loss": 0.4249,
      "step": 6005
    },
    {
      "epoch": 0.37620382404985986,
      "grad_norm": 0.7691009992300232,
      "learning_rate": 7.172486424721571e-06,
      "loss": 0.434,
      "step": 6006
    },
    {
      "epoch": 0.37626646204920217,
      "grad_norm": 0.7665853876412939,
      "learning_rate": 7.1715727383881165e-06,
      "loss": 0.3958,
      "step": 6007
    },
    {
      "epoch": 0.3763291000485445,
      "grad_norm": 0.79566184586161,
      "learning_rate": 7.170658962672323e-06,
      "loss": 0.4003,
      "step": 6008
    },
    {
      "epoch": 0.3763917380478867,
      "grad_norm": 0.621799697887271,
      "learning_rate": 7.1697450976118065e-06,
      "loss": 0.5002,
      "step": 6009
    },
    {
      "epoch": 0.37645437604722903,
      "grad_norm": 0.8005438642345978,
      "learning_rate": 7.1688311432441774e-06,
      "loss": 0.3872,
      "step": 6010
    },
    {
      "epoch": 0.37651701404657134,
      "grad_norm": 0.8353808793679727,
      "learning_rate": 7.1679170996070565e-06,
      "loss": 0.387,
      "step": 6011
    },
    {
      "epoch": 0.37657965204591365,
      "grad_norm": 0.8165017290074329,
      "learning_rate": 7.167002966738065e-06,
      "loss": 0.4217,
      "step": 6012
    },
    {
      "epoch": 0.37664229004525596,
      "grad_norm": 0.824631185079661,
      "learning_rate": 7.166088744674829e-06,
      "loss": 0.4276,
      "step": 6013
    },
    {
      "epoch": 0.37670492804459826,
      "grad_norm": 0.7953273581169849,
      "learning_rate": 7.16517443345498e-06,
      "loss": 0.3902,
      "step": 6014
    },
    {
      "epoch": 0.37676756604394057,
      "grad_norm": 0.7513046869493039,
      "learning_rate": 7.164260033116147e-06,
      "loss": 0.3664,
      "step": 6015
    },
    {
      "epoch": 0.3768302040432829,
      "grad_norm": 0.7248581425946624,
      "learning_rate": 7.16334554369597e-06,
      "loss": 0.3975,
      "step": 6016
    },
    {
      "epoch": 0.37689284204262513,
      "grad_norm": 0.662733540418089,
      "learning_rate": 7.16243096523209e-06,
      "loss": 0.4917,
      "step": 6017
    },
    {
      "epoch": 0.37695548004196744,
      "grad_norm": 0.7407243690856311,
      "learning_rate": 7.16151629776215e-06,
      "loss": 0.4148,
      "step": 6018
    },
    {
      "epoch": 0.37701811804130975,
      "grad_norm": 0.8894366151147665,
      "learning_rate": 7.160601541323797e-06,
      "loss": 0.471,
      "step": 6019
    },
    {
      "epoch": 0.37708075604065205,
      "grad_norm": 0.783531186829769,
      "learning_rate": 7.159686695954684e-06,
      "loss": 0.3892,
      "step": 6020
    },
    {
      "epoch": 0.37714339403999436,
      "grad_norm": 0.7920167042021498,
      "learning_rate": 7.158771761692464e-06,
      "loss": 0.3526,
      "step": 6021
    },
    {
      "epoch": 0.37720603203933667,
      "grad_norm": 0.804753911667462,
      "learning_rate": 7.157856738574801e-06,
      "loss": 0.427,
      "step": 6022
    },
    {
      "epoch": 0.377268670038679,
      "grad_norm": 0.8439319425716394,
      "learning_rate": 7.156941626639353e-06,
      "loss": 0.4189,
      "step": 6023
    },
    {
      "epoch": 0.3773313080380213,
      "grad_norm": 0.8362083394373203,
      "learning_rate": 7.156026425923785e-06,
      "loss": 0.4245,
      "step": 6024
    },
    {
      "epoch": 0.3773939460373636,
      "grad_norm": 0.8507503982336692,
      "learning_rate": 7.155111136465771e-06,
      "loss": 0.4199,
      "step": 6025
    },
    {
      "epoch": 0.37745658403670584,
      "grad_norm": 0.8983512662788217,
      "learning_rate": 7.154195758302983e-06,
      "loss": 0.4212,
      "step": 6026
    },
    {
      "epoch": 0.37751922203604815,
      "grad_norm": 0.8549694407339172,
      "learning_rate": 7.153280291473097e-06,
      "loss": 0.4204,
      "step": 6027
    },
    {
      "epoch": 0.37758186003539046,
      "grad_norm": 0.7679684503858496,
      "learning_rate": 7.152364736013796e-06,
      "loss": 0.392,
      "step": 6028
    },
    {
      "epoch": 0.37764449803473277,
      "grad_norm": 0.8124547118744072,
      "learning_rate": 7.15144909196276e-06,
      "loss": 0.3691,
      "step": 6029
    },
    {
      "epoch": 0.3777071360340751,
      "grad_norm": 0.8069992411561813,
      "learning_rate": 7.150533359357682e-06,
      "loss": 0.3921,
      "step": 6030
    },
    {
      "epoch": 0.3777697740334174,
      "grad_norm": 0.8439876763190461,
      "learning_rate": 7.149617538236252e-06,
      "loss": 0.3977,
      "step": 6031
    },
    {
      "epoch": 0.3778324120327597,
      "grad_norm": 0.7975277540331157,
      "learning_rate": 7.148701628636165e-06,
      "loss": 0.4157,
      "step": 6032
    },
    {
      "epoch": 0.377895050032102,
      "grad_norm": 0.8495659059748366,
      "learning_rate": 7.14778563059512e-06,
      "loss": 0.4446,
      "step": 6033
    },
    {
      "epoch": 0.37795768803144425,
      "grad_norm": 0.7858396658114315,
      "learning_rate": 7.146869544150818e-06,
      "loss": 0.4277,
      "step": 6034
    },
    {
      "epoch": 0.37802032603078656,
      "grad_norm": 0.863810610363887,
      "learning_rate": 7.14595336934097e-06,
      "loss": 0.4377,
      "step": 6035
    },
    {
      "epoch": 0.37808296403012887,
      "grad_norm": 0.7975924003095732,
      "learning_rate": 7.145037106203282e-06,
      "loss": 0.4343,
      "step": 6036
    },
    {
      "epoch": 0.3781456020294712,
      "grad_norm": 0.8913888278656966,
      "learning_rate": 7.144120754775468e-06,
      "loss": 0.4134,
      "step": 6037
    },
    {
      "epoch": 0.3782082400288135,
      "grad_norm": 0.8011774048866088,
      "learning_rate": 7.143204315095245e-06,
      "loss": 0.3431,
      "step": 6038
    },
    {
      "epoch": 0.3782708780281558,
      "grad_norm": 0.8149423804321617,
      "learning_rate": 7.142287787200336e-06,
      "loss": 0.4339,
      "step": 6039
    },
    {
      "epoch": 0.3783335160274981,
      "grad_norm": 0.8450040590287069,
      "learning_rate": 7.141371171128464e-06,
      "loss": 0.4449,
      "step": 6040
    },
    {
      "epoch": 0.3783961540268404,
      "grad_norm": 0.8524737030850843,
      "learning_rate": 7.140454466917357e-06,
      "loss": 0.4409,
      "step": 6041
    },
    {
      "epoch": 0.3784587920261827,
      "grad_norm": 0.8800384138308949,
      "learning_rate": 7.139537674604746e-06,
      "loss": 0.4431,
      "step": 6042
    },
    {
      "epoch": 0.37852143002552496,
      "grad_norm": 0.8240992346393788,
      "learning_rate": 7.138620794228368e-06,
      "loss": 0.386,
      "step": 6043
    },
    {
      "epoch": 0.37858406802486727,
      "grad_norm": 0.7650402635696368,
      "learning_rate": 7.137703825825961e-06,
      "loss": 0.3689,
      "step": 6044
    },
    {
      "epoch": 0.3786467060242096,
      "grad_norm": 0.8390782982950227,
      "learning_rate": 7.136786769435269e-06,
      "loss": 0.3998,
      "step": 6045
    },
    {
      "epoch": 0.3787093440235519,
      "grad_norm": 0.826284393384444,
      "learning_rate": 7.135869625094035e-06,
      "loss": 0.4163,
      "step": 6046
    },
    {
      "epoch": 0.3787719820228942,
      "grad_norm": 0.8056540492792158,
      "learning_rate": 7.1349523928400114e-06,
      "loss": 0.4263,
      "step": 6047
    },
    {
      "epoch": 0.3788346200222365,
      "grad_norm": 0.9206344473842335,
      "learning_rate": 7.134035072710951e-06,
      "loss": 0.4056,
      "step": 6048
    },
    {
      "epoch": 0.3788972580215788,
      "grad_norm": 0.7090988845078441,
      "learning_rate": 7.133117664744611e-06,
      "loss": 0.4744,
      "step": 6049
    },
    {
      "epoch": 0.3789598960209211,
      "grad_norm": 0.8466200679897938,
      "learning_rate": 7.132200168978752e-06,
      "loss": 0.4266,
      "step": 6050
    },
    {
      "epoch": 0.37902253402026337,
      "grad_norm": 0.9011767089468742,
      "learning_rate": 7.1312825854511395e-06,
      "loss": 0.4004,
      "step": 6051
    },
    {
      "epoch": 0.3790851720196057,
      "grad_norm": 0.8678158051692165,
      "learning_rate": 7.130364914199538e-06,
      "loss": 0.4356,
      "step": 6052
    },
    {
      "epoch": 0.379147810018948,
      "grad_norm": 0.8956379109553363,
      "learning_rate": 7.129447155261723e-06,
      "loss": 0.4001,
      "step": 6053
    },
    {
      "epoch": 0.3792104480182903,
      "grad_norm": 0.8202795961050825,
      "learning_rate": 7.128529308675467e-06,
      "loss": 0.4318,
      "step": 6054
    },
    {
      "epoch": 0.3792730860176326,
      "grad_norm": 0.8880094846561293,
      "learning_rate": 7.127611374478549e-06,
      "loss": 0.4193,
      "step": 6055
    },
    {
      "epoch": 0.3793357240169749,
      "grad_norm": 0.810752501470387,
      "learning_rate": 7.126693352708753e-06,
      "loss": 0.3901,
      "step": 6056
    },
    {
      "epoch": 0.3793983620163172,
      "grad_norm": 0.6655308593998541,
      "learning_rate": 7.125775243403863e-06,
      "loss": 0.4981,
      "step": 6057
    },
    {
      "epoch": 0.3794610000156595,
      "grad_norm": 0.7972077241264358,
      "learning_rate": 7.124857046601671e-06,
      "loss": 0.3944,
      "step": 6058
    },
    {
      "epoch": 0.3795236380150018,
      "grad_norm": 0.8059590489260231,
      "learning_rate": 7.123938762339967e-06,
      "loss": 0.3682,
      "step": 6059
    },
    {
      "epoch": 0.3795862760143441,
      "grad_norm": 0.816379415383212,
      "learning_rate": 7.1230203906565495e-06,
      "loss": 0.4445,
      "step": 6060
    },
    {
      "epoch": 0.3796489140136864,
      "grad_norm": 0.9455543711074325,
      "learning_rate": 7.122101931589219e-06,
      "loss": 0.377,
      "step": 6061
    },
    {
      "epoch": 0.3797115520130287,
      "grad_norm": 0.7964918327990895,
      "learning_rate": 7.1211833851757795e-06,
      "loss": 0.4088,
      "step": 6062
    },
    {
      "epoch": 0.379774190012371,
      "grad_norm": 0.8948309114763255,
      "learning_rate": 7.120264751454037e-06,
      "loss": 0.4487,
      "step": 6063
    },
    {
      "epoch": 0.3798368280117133,
      "grad_norm": 0.8628859167454173,
      "learning_rate": 7.119346030461806e-06,
      "loss": 0.4241,
      "step": 6064
    },
    {
      "epoch": 0.3798994660110556,
      "grad_norm": 0.8692288096954564,
      "learning_rate": 7.118427222236897e-06,
      "loss": 0.3919,
      "step": 6065
    },
    {
      "epoch": 0.3799621040103979,
      "grad_norm": 0.8324405352079225,
      "learning_rate": 7.117508326817131e-06,
      "loss": 0.409,
      "step": 6066
    },
    {
      "epoch": 0.38002474200974024,
      "grad_norm": 0.867942230473971,
      "learning_rate": 7.116589344240329e-06,
      "loss": 0.3967,
      "step": 6067
    },
    {
      "epoch": 0.3800873800090825,
      "grad_norm": 0.8234615891186724,
      "learning_rate": 7.115670274544316e-06,
      "loss": 0.3778,
      "step": 6068
    },
    {
      "epoch": 0.3801500180084248,
      "grad_norm": 0.7341679970163366,
      "learning_rate": 7.114751117766924e-06,
      "loss": 0.4168,
      "step": 6069
    },
    {
      "epoch": 0.3802126560077671,
      "grad_norm": 0.7862873491237222,
      "learning_rate": 7.113831873945981e-06,
      "loss": 0.4018,
      "step": 6070
    },
    {
      "epoch": 0.3802752940071094,
      "grad_norm": 0.903022434882879,
      "learning_rate": 7.112912543119328e-06,
      "loss": 0.4129,
      "step": 6071
    },
    {
      "epoch": 0.3803379320064517,
      "grad_norm": 0.8840631762578425,
      "learning_rate": 7.111993125324801e-06,
      "loss": 0.3941,
      "step": 6072
    },
    {
      "epoch": 0.380400570005794,
      "grad_norm": 0.8234099102447939,
      "learning_rate": 7.111073620600245e-06,
      "loss": 0.3821,
      "step": 6073
    },
    {
      "epoch": 0.38046320800513633,
      "grad_norm": 0.8519210462753598,
      "learning_rate": 7.110154028983508e-06,
      "loss": 0.4047,
      "step": 6074
    },
    {
      "epoch": 0.38052584600447864,
      "grad_norm": 0.7818051967010954,
      "learning_rate": 7.109234350512439e-06,
      "loss": 0.3973,
      "step": 6075
    },
    {
      "epoch": 0.3805884840038209,
      "grad_norm": 0.7817464632214571,
      "learning_rate": 7.1083145852248915e-06,
      "loss": 0.4937,
      "step": 6076
    },
    {
      "epoch": 0.3806511220031632,
      "grad_norm": 0.8907566173857909,
      "learning_rate": 7.107394733158726e-06,
      "loss": 0.4422,
      "step": 6077
    },
    {
      "epoch": 0.3807137600025055,
      "grad_norm": 0.6889556536234162,
      "learning_rate": 7.106474794351801e-06,
      "loss": 0.464,
      "step": 6078
    },
    {
      "epoch": 0.3807763980018478,
      "grad_norm": 0.8877523582237647,
      "learning_rate": 7.1055547688419825e-06,
      "loss": 0.4647,
      "step": 6079
    },
    {
      "epoch": 0.3808390360011901,
      "grad_norm": 0.8343597020353921,
      "learning_rate": 7.10463465666714e-06,
      "loss": 0.4223,
      "step": 6080
    },
    {
      "epoch": 0.38090167400053243,
      "grad_norm": 0.6776265749064624,
      "learning_rate": 7.103714457865143e-06,
      "loss": 0.4922,
      "step": 6081
    },
    {
      "epoch": 0.38096431199987474,
      "grad_norm": 0.8087224227186447,
      "learning_rate": 7.1027941724738705e-06,
      "loss": 0.3582,
      "step": 6082
    },
    {
      "epoch": 0.38102694999921705,
      "grad_norm": 0.8662593380042622,
      "learning_rate": 7.1018738005311985e-06,
      "loss": 0.4606,
      "step": 6083
    },
    {
      "epoch": 0.38108958799855935,
      "grad_norm": 0.843764945010773,
      "learning_rate": 7.10095334207501e-06,
      "loss": 0.4279,
      "step": 6084
    },
    {
      "epoch": 0.3811522259979016,
      "grad_norm": 0.7516773743128801,
      "learning_rate": 7.100032797143192e-06,
      "loss": 0.3889,
      "step": 6085
    },
    {
      "epoch": 0.3812148639972439,
      "grad_norm": 0.7823528932997977,
      "learning_rate": 7.099112165773635e-06,
      "loss": 0.3795,
      "step": 6086
    },
    {
      "epoch": 0.3812775019965862,
      "grad_norm": 0.8065662231496112,
      "learning_rate": 7.098191448004231e-06,
      "loss": 0.3735,
      "step": 6087
    },
    {
      "epoch": 0.38134013999592853,
      "grad_norm": 0.8633143535134019,
      "learning_rate": 7.097270643872878e-06,
      "loss": 0.4215,
      "step": 6088
    },
    {
      "epoch": 0.38140277799527084,
      "grad_norm": 0.7991308122247683,
      "learning_rate": 7.096349753417475e-06,
      "loss": 0.4102,
      "step": 6089
    },
    {
      "epoch": 0.38146541599461314,
      "grad_norm": 0.8322083999885583,
      "learning_rate": 7.095428776675929e-06,
      "loss": 0.4435,
      "step": 6090
    },
    {
      "epoch": 0.38152805399395545,
      "grad_norm": 0.7962328746914612,
      "learning_rate": 7.094507713686145e-06,
      "loss": 0.3982,
      "step": 6091
    },
    {
      "epoch": 0.38159069199329776,
      "grad_norm": 0.8267405626242065,
      "learning_rate": 7.093586564486034e-06,
      "loss": 0.4371,
      "step": 6092
    },
    {
      "epoch": 0.38165332999264,
      "grad_norm": 0.8415291011420623,
      "learning_rate": 7.0926653291135115e-06,
      "loss": 0.4557,
      "step": 6093
    },
    {
      "epoch": 0.3817159679919823,
      "grad_norm": 0.8074275585494274,
      "learning_rate": 7.091744007606496e-06,
      "loss": 0.4136,
      "step": 6094
    },
    {
      "epoch": 0.3817786059913246,
      "grad_norm": 0.8892885651399276,
      "learning_rate": 7.090822600002909e-06,
      "loss": 0.4131,
      "step": 6095
    },
    {
      "epoch": 0.38184124399066693,
      "grad_norm": 0.7711199105216994,
      "learning_rate": 7.089901106340677e-06,
      "loss": 0.3948,
      "step": 6096
    },
    {
      "epoch": 0.38190388199000924,
      "grad_norm": 0.8471987244294158,
      "learning_rate": 7.088979526657725e-06,
      "loss": 0.4503,
      "step": 6097
    },
    {
      "epoch": 0.38196651998935155,
      "grad_norm": 0.7956049177348885,
      "learning_rate": 7.088057860991988e-06,
      "loss": 0.4792,
      "step": 6098
    },
    {
      "epoch": 0.38202915798869386,
      "grad_norm": 0.8563233654901828,
      "learning_rate": 7.0871361093814036e-06,
      "loss": 0.4261,
      "step": 6099
    },
    {
      "epoch": 0.38209179598803616,
      "grad_norm": 0.8225772620302215,
      "learning_rate": 7.086214271863909e-06,
      "loss": 0.3863,
      "step": 6100
    },
    {
      "epoch": 0.3821544339873784,
      "grad_norm": 0.9061651649292842,
      "learning_rate": 7.085292348477448e-06,
      "loss": 0.4553,
      "step": 6101
    },
    {
      "epoch": 0.3822170719867207,
      "grad_norm": 0.8549339113020955,
      "learning_rate": 7.084370339259966e-06,
      "loss": 0.4255,
      "step": 6102
    },
    {
      "epoch": 0.38227970998606303,
      "grad_norm": 0.8329869533002395,
      "learning_rate": 7.0834482442494155e-06,
      "loss": 0.4051,
      "step": 6103
    },
    {
      "epoch": 0.38234234798540534,
      "grad_norm": 0.8139988035058909,
      "learning_rate": 7.082526063483748e-06,
      "loss": 0.4111,
      "step": 6104
    },
    {
      "epoch": 0.38240498598474765,
      "grad_norm": 0.8958073937193602,
      "learning_rate": 7.081603797000921e-06,
      "loss": 0.4393,
      "step": 6105
    },
    {
      "epoch": 0.38246762398408995,
      "grad_norm": 0.8617039177681631,
      "learning_rate": 7.080681444838895e-06,
      "loss": 0.4247,
      "step": 6106
    },
    {
      "epoch": 0.38253026198343226,
      "grad_norm": 0.8565327629378452,
      "learning_rate": 7.0797590070356334e-06,
      "loss": 0.415,
      "step": 6107
    },
    {
      "epoch": 0.38259289998277457,
      "grad_norm": 0.8206057971242239,
      "learning_rate": 7.078836483629107e-06,
      "loss": 0.3918,
      "step": 6108
    },
    {
      "epoch": 0.3826555379821169,
      "grad_norm": 0.8036698706371129,
      "learning_rate": 7.0779138746572855e-06,
      "loss": 0.4209,
      "step": 6109
    },
    {
      "epoch": 0.38271817598145913,
      "grad_norm": 0.8307803580587131,
      "learning_rate": 7.07699118015814e-06,
      "loss": 0.4382,
      "step": 6110
    },
    {
      "epoch": 0.38278081398080144,
      "grad_norm": 0.786656307109097,
      "learning_rate": 7.076068400169654e-06,
      "loss": 0.4331,
      "step": 6111
    },
    {
      "epoch": 0.38284345198014375,
      "grad_norm": 0.7957678689175984,
      "learning_rate": 7.075145534729808e-06,
      "loss": 0.4284,
      "step": 6112
    },
    {
      "epoch": 0.38290608997948605,
      "grad_norm": 0.823625052383805,
      "learning_rate": 7.074222583876586e-06,
      "loss": 0.3997,
      "step": 6113
    },
    {
      "epoch": 0.38296872797882836,
      "grad_norm": 0.8279795507655386,
      "learning_rate": 7.073299547647978e-06,
      "loss": 0.4213,
      "step": 6114
    },
    {
      "epoch": 0.38303136597817067,
      "grad_norm": 0.8696556029233287,
      "learning_rate": 7.072376426081975e-06,
      "loss": 0.4284,
      "step": 6115
    },
    {
      "epoch": 0.383094003977513,
      "grad_norm": 0.756378195930642,
      "learning_rate": 7.071453219216574e-06,
      "loss": 0.408,
      "step": 6116
    },
    {
      "epoch": 0.3831566419768553,
      "grad_norm": 0.8820947095354198,
      "learning_rate": 7.070529927089775e-06,
      "loss": 0.4269,
      "step": 6117
    },
    {
      "epoch": 0.38321927997619754,
      "grad_norm": 0.8795352633507372,
      "learning_rate": 7.069606549739578e-06,
      "loss": 0.4027,
      "step": 6118
    },
    {
      "epoch": 0.38328191797553984,
      "grad_norm": 0.8040210790714822,
      "learning_rate": 7.0686830872039926e-06,
      "loss": 0.4287,
      "step": 6119
    },
    {
      "epoch": 0.38334455597488215,
      "grad_norm": 0.8083023530909176,
      "learning_rate": 7.067759539521027e-06,
      "loss": 0.3863,
      "step": 6120
    },
    {
      "epoch": 0.38340719397422446,
      "grad_norm": 0.8146731471027779,
      "learning_rate": 7.066835906728694e-06,
      "loss": 0.4008,
      "step": 6121
    },
    {
      "epoch": 0.38346983197356677,
      "grad_norm": 0.7680538179510763,
      "learning_rate": 7.065912188865013e-06,
      "loss": 0.4203,
      "step": 6122
    },
    {
      "epoch": 0.3835324699729091,
      "grad_norm": 0.8828789169504319,
      "learning_rate": 7.064988385968001e-06,
      "loss": 0.4185,
      "step": 6123
    },
    {
      "epoch": 0.3835951079722514,
      "grad_norm": 0.6712466349729105,
      "learning_rate": 7.064064498075687e-06,
      "loss": 0.4772,
      "step": 6124
    },
    {
      "epoch": 0.3836577459715937,
      "grad_norm": 0.68733228881146,
      "learning_rate": 7.063140525226092e-06,
      "loss": 0.4507,
      "step": 6125
    },
    {
      "epoch": 0.38372038397093594,
      "grad_norm": 0.8094313406382148,
      "learning_rate": 7.062216467457252e-06,
      "loss": 0.4382,
      "step": 6126
    },
    {
      "epoch": 0.38378302197027825,
      "grad_norm": 0.8275936428364348,
      "learning_rate": 7.061292324807197e-06,
      "loss": 0.4166,
      "step": 6127
    },
    {
      "epoch": 0.38384565996962056,
      "grad_norm": 0.8204922246018035,
      "learning_rate": 7.06036809731397e-06,
      "loss": 0.4102,
      "step": 6128
    },
    {
      "epoch": 0.38390829796896286,
      "grad_norm": 0.914871574862147,
      "learning_rate": 7.0594437850156096e-06,
      "loss": 0.4324,
      "step": 6129
    },
    {
      "epoch": 0.38397093596830517,
      "grad_norm": 0.7886774891560075,
      "learning_rate": 7.058519387950158e-06,
      "loss": 0.3982,
      "step": 6130
    },
    {
      "epoch": 0.3840335739676475,
      "grad_norm": 0.6666009870063716,
      "learning_rate": 7.057594906155669e-06,
      "loss": 0.4912,
      "step": 6131
    },
    {
      "epoch": 0.3840962119669898,
      "grad_norm": 0.8047449025901291,
      "learning_rate": 7.056670339670191e-06,
      "loss": 0.4174,
      "step": 6132
    },
    {
      "epoch": 0.3841588499663321,
      "grad_norm": 0.8370873642364914,
      "learning_rate": 7.05574568853178e-06,
      "loss": 0.4138,
      "step": 6133
    },
    {
      "epoch": 0.3842214879656744,
      "grad_norm": 0.7738710054180674,
      "learning_rate": 7.054820952778494e-06,
      "loss": 0.4259,
      "step": 6134
    },
    {
      "epoch": 0.38428412596501665,
      "grad_norm": 0.9058332050643645,
      "learning_rate": 7.053896132448397e-06,
      "loss": 0.4107,
      "step": 6135
    },
    {
      "epoch": 0.38434676396435896,
      "grad_norm": 0.8203265181207906,
      "learning_rate": 7.0529712275795535e-06,
      "loss": 0.4287,
      "step": 6136
    },
    {
      "epoch": 0.38440940196370127,
      "grad_norm": 0.8220741627375615,
      "learning_rate": 7.052046238210034e-06,
      "loss": 0.3761,
      "step": 6137
    },
    {
      "epoch": 0.3844720399630436,
      "grad_norm": 0.818124139525692,
      "learning_rate": 7.051121164377911e-06,
      "loss": 0.4291,
      "step": 6138
    },
    {
      "epoch": 0.3845346779623859,
      "grad_norm": 0.8385456722420321,
      "learning_rate": 7.0501960061212595e-06,
      "loss": 0.3834,
      "step": 6139
    },
    {
      "epoch": 0.3845973159617282,
      "grad_norm": 0.8398561034804021,
      "learning_rate": 7.049270763478157e-06,
      "loss": 0.3944,
      "step": 6140
    },
    {
      "epoch": 0.3846599539610705,
      "grad_norm": 0.7402479612206025,
      "learning_rate": 7.048345436486694e-06,
      "loss": 0.5004,
      "step": 6141
    },
    {
      "epoch": 0.3847225919604128,
      "grad_norm": 0.81743265418052,
      "learning_rate": 7.047420025184951e-06,
      "loss": 0.445,
      "step": 6142
    },
    {
      "epoch": 0.38478522995975506,
      "grad_norm": 0.744576977889003,
      "learning_rate": 7.04649452961102e-06,
      "loss": 0.3947,
      "step": 6143
    },
    {
      "epoch": 0.38484786795909737,
      "grad_norm": 0.697630556748134,
      "learning_rate": 7.045568949802993e-06,
      "loss": 0.4513,
      "step": 6144
    },
    {
      "epoch": 0.3849105059584397,
      "grad_norm": 0.7706973240211059,
      "learning_rate": 7.0446432857989686e-06,
      "loss": 0.3779,
      "step": 6145
    },
    {
      "epoch": 0.384973143957782,
      "grad_norm": 0.7822950018232786,
      "learning_rate": 7.04371753763705e-06,
      "loss": 0.4409,
      "step": 6146
    },
    {
      "epoch": 0.3850357819571243,
      "grad_norm": 0.8052279125208087,
      "learning_rate": 7.042791705355335e-06,
      "loss": 0.4029,
      "step": 6147
    },
    {
      "epoch": 0.3850984199564666,
      "grad_norm": 0.8453720757953296,
      "learning_rate": 7.041865788991935e-06,
      "loss": 0.4252,
      "step": 6148
    },
    {
      "epoch": 0.3851610579558089,
      "grad_norm": 0.8980409516993794,
      "learning_rate": 7.04093978858496e-06,
      "loss": 0.4426,
      "step": 6149
    },
    {
      "epoch": 0.3852236959551512,
      "grad_norm": 0.8548432268392293,
      "learning_rate": 7.040013704172525e-06,
      "loss": 0.4703,
      "step": 6150
    },
    {
      "epoch": 0.3852863339544935,
      "grad_norm": 0.8346243577711209,
      "learning_rate": 7.039087535792747e-06,
      "loss": 0.3969,
      "step": 6151
    },
    {
      "epoch": 0.3853489719538358,
      "grad_norm": 0.8860690897156427,
      "learning_rate": 7.038161283483749e-06,
      "loss": 0.4143,
      "step": 6152
    },
    {
      "epoch": 0.3854116099531781,
      "grad_norm": 0.8022423626587407,
      "learning_rate": 7.037234947283651e-06,
      "loss": 0.4129,
      "step": 6153
    },
    {
      "epoch": 0.3854742479525204,
      "grad_norm": 0.8228900317146465,
      "learning_rate": 7.036308527230586e-06,
      "loss": 0.3714,
      "step": 6154
    },
    {
      "epoch": 0.3855368859518627,
      "grad_norm": 0.7681466136332357,
      "learning_rate": 7.035382023362684e-06,
      "loss": 0.3902,
      "step": 6155
    },
    {
      "epoch": 0.385599523951205,
      "grad_norm": 0.8460435553822424,
      "learning_rate": 7.034455435718082e-06,
      "loss": 0.393,
      "step": 6156
    },
    {
      "epoch": 0.3856621619505473,
      "grad_norm": 0.6687627688391062,
      "learning_rate": 7.0335287643349145e-06,
      "loss": 0.4969,
      "step": 6157
    },
    {
      "epoch": 0.3857247999498896,
      "grad_norm": 0.9608722976147317,
      "learning_rate": 7.032602009251326e-06,
      "loss": 0.4204,
      "step": 6158
    },
    {
      "epoch": 0.3857874379492319,
      "grad_norm": 0.7767507687843853,
      "learning_rate": 7.031675170505462e-06,
      "loss": 0.403,
      "step": 6159
    },
    {
      "epoch": 0.3858500759485742,
      "grad_norm": 0.8993542552838526,
      "learning_rate": 7.030748248135472e-06,
      "loss": 0.3935,
      "step": 6160
    },
    {
      "epoch": 0.3859127139479165,
      "grad_norm": 0.8353725635021992,
      "learning_rate": 7.029821242179504e-06,
      "loss": 0.3914,
      "step": 6161
    },
    {
      "epoch": 0.3859753519472588,
      "grad_norm": 0.8280702780361766,
      "learning_rate": 7.028894152675719e-06,
      "loss": 0.4084,
      "step": 6162
    },
    {
      "epoch": 0.3860379899466011,
      "grad_norm": 0.80787186896438,
      "learning_rate": 7.027966979662274e-06,
      "loss": 0.4401,
      "step": 6163
    },
    {
      "epoch": 0.3861006279459434,
      "grad_norm": 0.8651213322129924,
      "learning_rate": 7.027039723177332e-06,
      "loss": 0.4276,
      "step": 6164
    },
    {
      "epoch": 0.3861632659452857,
      "grad_norm": 0.7981814999362484,
      "learning_rate": 7.026112383259059e-06,
      "loss": 0.4328,
      "step": 6165
    },
    {
      "epoch": 0.386225903944628,
      "grad_norm": 0.8177546363711925,
      "learning_rate": 7.025184959945624e-06,
      "loss": 0.4158,
      "step": 6166
    },
    {
      "epoch": 0.38628854194397033,
      "grad_norm": 0.8050742772884729,
      "learning_rate": 7.024257453275198e-06,
      "loss": 0.4173,
      "step": 6167
    },
    {
      "epoch": 0.3863511799433126,
      "grad_norm": 0.7958635651178614,
      "learning_rate": 7.023329863285962e-06,
      "loss": 0.367,
      "step": 6168
    },
    {
      "epoch": 0.3864138179426549,
      "grad_norm": 0.9280203434439677,
      "learning_rate": 7.022402190016093e-06,
      "loss": 0.4824,
      "step": 6169
    },
    {
      "epoch": 0.3864764559419972,
      "grad_norm": 0.9563501837714657,
      "learning_rate": 7.021474433503774e-06,
      "loss": 0.477,
      "step": 6170
    },
    {
      "epoch": 0.3865390939413395,
      "grad_norm": 0.8391430497915747,
      "learning_rate": 7.020546593787192e-06,
      "loss": 0.4223,
      "step": 6171
    },
    {
      "epoch": 0.3866017319406818,
      "grad_norm": 0.800374053585278,
      "learning_rate": 7.019618670904538e-06,
      "loss": 0.4049,
      "step": 6172
    },
    {
      "epoch": 0.3866643699400241,
      "grad_norm": 0.7518359052188687,
      "learning_rate": 7.018690664894004e-06,
      "loss": 0.3728,
      "step": 6173
    },
    {
      "epoch": 0.38672700793936643,
      "grad_norm": 0.8295153872007655,
      "learning_rate": 7.017762575793786e-06,
      "loss": 0.4266,
      "step": 6174
    },
    {
      "epoch": 0.38678964593870874,
      "grad_norm": 0.8359729522811751,
      "learning_rate": 7.016834403642088e-06,
      "loss": 0.4233,
      "step": 6175
    },
    {
      "epoch": 0.38685228393805104,
      "grad_norm": 0.829842998837805,
      "learning_rate": 7.015906148477111e-06,
      "loss": 0.4415,
      "step": 6176
    },
    {
      "epoch": 0.3869149219373933,
      "grad_norm": 0.728342774037764,
      "learning_rate": 7.014977810337063e-06,
      "loss": 0.4031,
      "step": 6177
    },
    {
      "epoch": 0.3869775599367356,
      "grad_norm": 0.7660761987400988,
      "learning_rate": 7.014049389260154e-06,
      "loss": 0.3888,
      "step": 6178
    },
    {
      "epoch": 0.3870401979360779,
      "grad_norm": 0.8505024623864532,
      "learning_rate": 7.013120885284599e-06,
      "loss": 0.4607,
      "step": 6179
    },
    {
      "epoch": 0.3871028359354202,
      "grad_norm": 0.9060968345122337,
      "learning_rate": 7.012192298448613e-06,
      "loss": 0.444,
      "step": 6180
    },
    {
      "epoch": 0.3871654739347625,
      "grad_norm": 0.6741519834689613,
      "learning_rate": 7.01126362879042e-06,
      "loss": 0.4889,
      "step": 6181
    },
    {
      "epoch": 0.38722811193410483,
      "grad_norm": 0.8847204493527717,
      "learning_rate": 7.010334876348243e-06,
      "loss": 0.4154,
      "step": 6182
    },
    {
      "epoch": 0.38729074993344714,
      "grad_norm": 0.8201690555468174,
      "learning_rate": 7.009406041160308e-06,
      "loss": 0.3938,
      "step": 6183
    },
    {
      "epoch": 0.38735338793278945,
      "grad_norm": 0.8446472067726634,
      "learning_rate": 7.008477123264849e-06,
      "loss": 0.433,
      "step": 6184
    },
    {
      "epoch": 0.3874160259321317,
      "grad_norm": 0.8280300181115833,
      "learning_rate": 7.0075481227000966e-06,
      "loss": 0.4472,
      "step": 6185
    },
    {
      "epoch": 0.387478663931474,
      "grad_norm": 0.7929782981711644,
      "learning_rate": 7.006619039504292e-06,
      "loss": 0.4077,
      "step": 6186
    },
    {
      "epoch": 0.3875413019308163,
      "grad_norm": 0.7805953791143646,
      "learning_rate": 7.005689873715676e-06,
      "loss": 0.4157,
      "step": 6187
    },
    {
      "epoch": 0.3876039399301586,
      "grad_norm": 0.8598275831396661,
      "learning_rate": 7.004760625372492e-06,
      "loss": 0.4377,
      "step": 6188
    },
    {
      "epoch": 0.38766657792950093,
      "grad_norm": 0.8062958300132573,
      "learning_rate": 7.0038312945129885e-06,
      "loss": 0.4018,
      "step": 6189
    },
    {
      "epoch": 0.38772921592884324,
      "grad_norm": 0.8258440223864086,
      "learning_rate": 7.002901881175417e-06,
      "loss": 0.4357,
      "step": 6190
    },
    {
      "epoch": 0.38779185392818555,
      "grad_norm": 0.8109436589767919,
      "learning_rate": 7.001972385398031e-06,
      "loss": 0.395,
      "step": 6191
    },
    {
      "epoch": 0.38785449192752786,
      "grad_norm": 0.7751756437660384,
      "learning_rate": 7.001042807219093e-06,
      "loss": 0.3738,
      "step": 6192
    },
    {
      "epoch": 0.38791712992687016,
      "grad_norm": 0.7911603467940191,
      "learning_rate": 7.00011314667686e-06,
      "loss": 0.39,
      "step": 6193
    },
    {
      "epoch": 0.3879797679262124,
      "grad_norm": 0.8242666196084655,
      "learning_rate": 6.999183403809597e-06,
      "loss": 0.3786,
      "step": 6194
    },
    {
      "epoch": 0.3880424059255547,
      "grad_norm": 0.8202294883203314,
      "learning_rate": 6.998253578655578e-06,
      "loss": 0.4363,
      "step": 6195
    },
    {
      "epoch": 0.38810504392489703,
      "grad_norm": 0.68041982945756,
      "learning_rate": 6.997323671253067e-06,
      "loss": 0.4731,
      "step": 6196
    },
    {
      "epoch": 0.38816768192423934,
      "grad_norm": 0.9025962885443595,
      "learning_rate": 6.996393681640346e-06,
      "loss": 0.4345,
      "step": 6197
    },
    {
      "epoch": 0.38823031992358165,
      "grad_norm": 0.7026575801946802,
      "learning_rate": 6.9954636098556886e-06,
      "loss": 0.3614,
      "step": 6198
    },
    {
      "epoch": 0.38829295792292395,
      "grad_norm": 0.7795281538896568,
      "learning_rate": 6.994533455937378e-06,
      "loss": 0.4028,
      "step": 6199
    },
    {
      "epoch": 0.38835559592226626,
      "grad_norm": 0.8195641870178829,
      "learning_rate": 6.9936032199237015e-06,
      "loss": 0.4145,
      "step": 6200
    },
    {
      "epoch": 0.38841823392160857,
      "grad_norm": 0.8209605295960006,
      "learning_rate": 6.992672901852946e-06,
      "loss": 0.4588,
      "step": 6201
    },
    {
      "epoch": 0.3884808719209508,
      "grad_norm": 0.7683160910929471,
      "learning_rate": 6.991742501763405e-06,
      "loss": 0.3766,
      "step": 6202
    },
    {
      "epoch": 0.38854350992029313,
      "grad_norm": 0.8454600185663477,
      "learning_rate": 6.9908120196933716e-06,
      "loss": 0.4008,
      "step": 6203
    },
    {
      "epoch": 0.38860614791963544,
      "grad_norm": 0.8319243568497579,
      "learning_rate": 6.989881455681145e-06,
      "loss": 0.4146,
      "step": 6204
    },
    {
      "epoch": 0.38866878591897774,
      "grad_norm": 0.7966691451218918,
      "learning_rate": 6.98895080976503e-06,
      "loss": 0.3518,
      "step": 6205
    },
    {
      "epoch": 0.38873142391832005,
      "grad_norm": 0.8163031467449161,
      "learning_rate": 6.988020081983331e-06,
      "loss": 0.3931,
      "step": 6206
    },
    {
      "epoch": 0.38879406191766236,
      "grad_norm": 0.793206962290555,
      "learning_rate": 6.987089272374357e-06,
      "loss": 0.4141,
      "step": 6207
    },
    {
      "epoch": 0.38885669991700467,
      "grad_norm": 0.7967672359928603,
      "learning_rate": 6.986158380976417e-06,
      "loss": 0.4023,
      "step": 6208
    },
    {
      "epoch": 0.388919337916347,
      "grad_norm": 0.7691215855398174,
      "learning_rate": 6.98522740782783e-06,
      "loss": 0.3843,
      "step": 6209
    },
    {
      "epoch": 0.3889819759156892,
      "grad_norm": 0.6748149288994163,
      "learning_rate": 6.984296352966916e-06,
      "loss": 0.4726,
      "step": 6210
    },
    {
      "epoch": 0.38904461391503153,
      "grad_norm": 0.917656407363133,
      "learning_rate": 6.983365216431995e-06,
      "loss": 0.4209,
      "step": 6211
    },
    {
      "epoch": 0.38910725191437384,
      "grad_norm": 0.8761516832216083,
      "learning_rate": 6.982433998261393e-06,
      "loss": 0.4492,
      "step": 6212
    },
    {
      "epoch": 0.38916988991371615,
      "grad_norm": 0.7841174853666925,
      "learning_rate": 6.98150269849344e-06,
      "loss": 0.4276,
      "step": 6213
    },
    {
      "epoch": 0.38923252791305846,
      "grad_norm": 0.792445421600834,
      "learning_rate": 6.980571317166469e-06,
      "loss": 0.395,
      "step": 6214
    },
    {
      "epoch": 0.38929516591240076,
      "grad_norm": 0.8129982319459191,
      "learning_rate": 6.979639854318815e-06,
      "loss": 0.4079,
      "step": 6215
    },
    {
      "epoch": 0.3893578039117431,
      "grad_norm": 0.7760939403744748,
      "learning_rate": 6.978708309988816e-06,
      "loss": 0.4412,
      "step": 6216
    },
    {
      "epoch": 0.3894204419110854,
      "grad_norm": 0.685485237757321,
      "learning_rate": 6.977776684214815e-06,
      "loss": 0.4921,
      "step": 6217
    },
    {
      "epoch": 0.3894830799104277,
      "grad_norm": 0.8040299540441401,
      "learning_rate": 6.976844977035161e-06,
      "loss": 0.4142,
      "step": 6218
    },
    {
      "epoch": 0.38954571790976994,
      "grad_norm": 0.8028324955885181,
      "learning_rate": 6.975913188488199e-06,
      "loss": 0.4007,
      "step": 6219
    },
    {
      "epoch": 0.38960835590911225,
      "grad_norm": 0.6733709496030699,
      "learning_rate": 6.974981318612285e-06,
      "loss": 0.465,
      "step": 6220
    },
    {
      "epoch": 0.38967099390845455,
      "grad_norm": 0.6725166604511087,
      "learning_rate": 6.974049367445772e-06,
      "loss": 0.4846,
      "step": 6221
    },
    {
      "epoch": 0.38973363190779686,
      "grad_norm": 0.8409732686886441,
      "learning_rate": 6.973117335027021e-06,
      "loss": 0.3663,
      "step": 6222
    },
    {
      "epoch": 0.38979626990713917,
      "grad_norm": 0.8978737774289759,
      "learning_rate": 6.972185221394394e-06,
      "loss": 0.4158,
      "step": 6223
    },
    {
      "epoch": 0.3898589079064815,
      "grad_norm": 0.8544033433022894,
      "learning_rate": 6.971253026586259e-06,
      "loss": 0.369,
      "step": 6224
    },
    {
      "epoch": 0.3899215459058238,
      "grad_norm": 0.7624093092995738,
      "learning_rate": 6.970320750640981e-06,
      "loss": 0.3749,
      "step": 6225
    },
    {
      "epoch": 0.3899841839051661,
      "grad_norm": 0.8216447555378344,
      "learning_rate": 6.969388393596937e-06,
      "loss": 0.4212,
      "step": 6226
    },
    {
      "epoch": 0.39004682190450835,
      "grad_norm": 0.8840790601311261,
      "learning_rate": 6.968455955492499e-06,
      "loss": 0.3849,
      "step": 6227
    },
    {
      "epoch": 0.39010945990385065,
      "grad_norm": 0.8239479727059336,
      "learning_rate": 6.96752343636605e-06,
      "loss": 0.4169,
      "step": 6228
    },
    {
      "epoch": 0.39017209790319296,
      "grad_norm": 0.8185490885071799,
      "learning_rate": 6.966590836255973e-06,
      "loss": 0.379,
      "step": 6229
    },
    {
      "epoch": 0.39023473590253527,
      "grad_norm": 0.8351929139637599,
      "learning_rate": 6.965658155200649e-06,
      "loss": 0.4419,
      "step": 6230
    },
    {
      "epoch": 0.3902973739018776,
      "grad_norm": 0.8302529713926861,
      "learning_rate": 6.964725393238472e-06,
      "loss": 0.4377,
      "step": 6231
    },
    {
      "epoch": 0.3903600119012199,
      "grad_norm": 0.8255675963778419,
      "learning_rate": 6.963792550407833e-06,
      "loss": 0.441,
      "step": 6232
    },
    {
      "epoch": 0.3904226499005622,
      "grad_norm": 0.8516348009485282,
      "learning_rate": 6.9628596267471285e-06,
      "loss": 0.4087,
      "step": 6233
    },
    {
      "epoch": 0.3904852878999045,
      "grad_norm": 0.8385894303553048,
      "learning_rate": 6.961926622294757e-06,
      "loss": 0.4507,
      "step": 6234
    },
    {
      "epoch": 0.39054792589924675,
      "grad_norm": 0.8190229689448724,
      "learning_rate": 6.960993537089121e-06,
      "loss": 0.3801,
      "step": 6235
    },
    {
      "epoch": 0.39061056389858906,
      "grad_norm": 0.8330272632321322,
      "learning_rate": 6.960060371168627e-06,
      "loss": 0.3941,
      "step": 6236
    },
    {
      "epoch": 0.39067320189793137,
      "grad_norm": 0.7062158096420558,
      "learning_rate": 6.959127124571686e-06,
      "loss": 0.3807,
      "step": 6237
    },
    {
      "epoch": 0.3907358398972737,
      "grad_norm": 0.7894032928933101,
      "learning_rate": 6.9581937973367074e-06,
      "loss": 0.4101,
      "step": 6238
    },
    {
      "epoch": 0.390798477896616,
      "grad_norm": 0.8459236580768467,
      "learning_rate": 6.95726038950211e-06,
      "loss": 0.427,
      "step": 6239
    },
    {
      "epoch": 0.3908611158959583,
      "grad_norm": 0.8634530879690346,
      "learning_rate": 6.956326901106309e-06,
      "loss": 0.3853,
      "step": 6240
    },
    {
      "epoch": 0.3909237538953006,
      "grad_norm": 0.8107272114882706,
      "learning_rate": 6.9553933321877325e-06,
      "loss": 0.4073,
      "step": 6241
    },
    {
      "epoch": 0.3909863918946429,
      "grad_norm": 0.7701886107564521,
      "learning_rate": 6.954459682784803e-06,
      "loss": 0.4079,
      "step": 6242
    },
    {
      "epoch": 0.3910490298939852,
      "grad_norm": 0.9043468302895145,
      "learning_rate": 6.953525952935949e-06,
      "loss": 0.4474,
      "step": 6243
    },
    {
      "epoch": 0.39111166789332746,
      "grad_norm": 0.8834605689517924,
      "learning_rate": 6.952592142679605e-06,
      "loss": 0.454,
      "step": 6244
    },
    {
      "epoch": 0.39117430589266977,
      "grad_norm": 0.8802085865646595,
      "learning_rate": 6.9516582520542055e-06,
      "loss": 0.468,
      "step": 6245
    },
    {
      "epoch": 0.3912369438920121,
      "grad_norm": 0.828391494054391,
      "learning_rate": 6.950724281098191e-06,
      "loss": 0.4212,
      "step": 6246
    },
    {
      "epoch": 0.3912995818913544,
      "grad_norm": 0.7784739758908243,
      "learning_rate": 6.949790229850002e-06,
      "loss": 0.4002,
      "step": 6247
    },
    {
      "epoch": 0.3913622198906967,
      "grad_norm": 0.8360607604258662,
      "learning_rate": 6.948856098348086e-06,
      "loss": 0.4441,
      "step": 6248
    },
    {
      "epoch": 0.391424857890039,
      "grad_norm": 0.7063344554148981,
      "learning_rate": 6.9479218866308905e-06,
      "loss": 0.4705,
      "step": 6249
    },
    {
      "epoch": 0.3914874958893813,
      "grad_norm": 0.8351601665884644,
      "learning_rate": 6.946987594736869e-06,
      "loss": 0.408,
      "step": 6250
    },
    {
      "epoch": 0.3915501338887236,
      "grad_norm": 0.8225845214498992,
      "learning_rate": 6.946053222704477e-06,
      "loss": 0.4434,
      "step": 6251
    },
    {
      "epoch": 0.39161277188806587,
      "grad_norm": 0.8375422099805985,
      "learning_rate": 6.945118770572172e-06,
      "loss": 0.4132,
      "step": 6252
    },
    {
      "epoch": 0.3916754098874082,
      "grad_norm": 0.8165865022341944,
      "learning_rate": 6.94418423837842e-06,
      "loss": 0.3886,
      "step": 6253
    },
    {
      "epoch": 0.3917380478867505,
      "grad_norm": 0.8229880719036343,
      "learning_rate": 6.943249626161681e-06,
      "loss": 0.4298,
      "step": 6254
    },
    {
      "epoch": 0.3918006858860928,
      "grad_norm": 0.8193245164742149,
      "learning_rate": 6.942314933960428e-06,
      "loss": 0.4382,
      "step": 6255
    },
    {
      "epoch": 0.3918633238854351,
      "grad_norm": 0.899105039205608,
      "learning_rate": 6.941380161813132e-06,
      "loss": 0.4344,
      "step": 6256
    },
    {
      "epoch": 0.3919259618847774,
      "grad_norm": 0.8561719388550229,
      "learning_rate": 6.940445309758268e-06,
      "loss": 0.4188,
      "step": 6257
    },
    {
      "epoch": 0.3919885998841197,
      "grad_norm": 0.8085016924731511,
      "learning_rate": 6.939510377834316e-06,
      "loss": 0.3997,
      "step": 6258
    },
    {
      "epoch": 0.392051237883462,
      "grad_norm": 0.8239753640132296,
      "learning_rate": 6.938575366079754e-06,
      "loss": 0.406,
      "step": 6259
    },
    {
      "epoch": 0.39211387588280433,
      "grad_norm": 0.835055642731441,
      "learning_rate": 6.9376402745330705e-06,
      "loss": 0.43,
      "step": 6260
    },
    {
      "epoch": 0.3921765138821466,
      "grad_norm": 0.8137353876346224,
      "learning_rate": 6.936705103232756e-06,
      "loss": 0.4296,
      "step": 6261
    },
    {
      "epoch": 0.3922391518814889,
      "grad_norm": 0.8364752332479508,
      "learning_rate": 6.9357698522173004e-06,
      "loss": 0.4289,
      "step": 6262
    },
    {
      "epoch": 0.3923017898808312,
      "grad_norm": 0.7719259676208826,
      "learning_rate": 6.934834521525195e-06,
      "loss": 0.3645,
      "step": 6263
    },
    {
      "epoch": 0.3923644278801735,
      "grad_norm": 0.831436714382028,
      "learning_rate": 6.933899111194943e-06,
      "loss": 0.4495,
      "step": 6264
    },
    {
      "epoch": 0.3924270658795158,
      "grad_norm": 0.8103971554551445,
      "learning_rate": 6.9329636212650455e-06,
      "loss": 0.3996,
      "step": 6265
    },
    {
      "epoch": 0.3924897038788581,
      "grad_norm": 0.8007845627653513,
      "learning_rate": 6.932028051774007e-06,
      "loss": 0.4151,
      "step": 6266
    },
    {
      "epoch": 0.39255234187820043,
      "grad_norm": 0.8013647174239807,
      "learning_rate": 6.931092402760335e-06,
      "loss": 0.4106,
      "step": 6267
    },
    {
      "epoch": 0.39261497987754274,
      "grad_norm": 0.8425587594328153,
      "learning_rate": 6.930156674262541e-06,
      "loss": 0.4205,
      "step": 6268
    },
    {
      "epoch": 0.392677617876885,
      "grad_norm": 0.7642327964599537,
      "learning_rate": 6.929220866319137e-06,
      "loss": 0.3906,
      "step": 6269
    },
    {
      "epoch": 0.3927402558762273,
      "grad_norm": 0.766845201065648,
      "learning_rate": 6.928284978968648e-06,
      "loss": 0.4014,
      "step": 6270
    },
    {
      "epoch": 0.3928028938755696,
      "grad_norm": 0.7728481022408388,
      "learning_rate": 6.927349012249591e-06,
      "loss": 0.3864,
      "step": 6271
    },
    {
      "epoch": 0.3928655318749119,
      "grad_norm": 0.7197859669037694,
      "learning_rate": 6.9264129662004886e-06,
      "loss": 0.4625,
      "step": 6272
    },
    {
      "epoch": 0.3929281698742542,
      "grad_norm": 0.8326686001844251,
      "learning_rate": 6.925476840859873e-06,
      "loss": 0.3997,
      "step": 6273
    },
    {
      "epoch": 0.3929908078735965,
      "grad_norm": 0.858597000997595,
      "learning_rate": 6.924540636266272e-06,
      "loss": 0.4112,
      "step": 6274
    },
    {
      "epoch": 0.39305344587293883,
      "grad_norm": 0.8692843268267555,
      "learning_rate": 6.9236043524582234e-06,
      "loss": 0.4278,
      "step": 6275
    },
    {
      "epoch": 0.39311608387228114,
      "grad_norm": 0.8359767232663443,
      "learning_rate": 6.92266798947426e-06,
      "loss": 0.4237,
      "step": 6276
    },
    {
      "epoch": 0.3931787218716234,
      "grad_norm": 0.8749015613486474,
      "learning_rate": 6.921731547352926e-06,
      "loss": 0.438,
      "step": 6277
    },
    {
      "epoch": 0.3932413598709657,
      "grad_norm": 0.8190468036565202,
      "learning_rate": 6.920795026132765e-06,
      "loss": 0.3987,
      "step": 6278
    },
    {
      "epoch": 0.393303997870308,
      "grad_norm": 0.849794921119129,
      "learning_rate": 6.919858425852326e-06,
      "loss": 0.4249,
      "step": 6279
    },
    {
      "epoch": 0.3933666358696503,
      "grad_norm": 0.8518208309307036,
      "learning_rate": 6.918921746550155e-06,
      "loss": 0.3985,
      "step": 6280
    },
    {
      "epoch": 0.3934292738689926,
      "grad_norm": 0.7468173729992125,
      "learning_rate": 6.917984988264811e-06,
      "loss": 0.3846,
      "step": 6281
    },
    {
      "epoch": 0.39349191186833493,
      "grad_norm": 0.85795921420482,
      "learning_rate": 6.917048151034847e-06,
      "loss": 0.4471,
      "step": 6282
    },
    {
      "epoch": 0.39355454986767724,
      "grad_norm": 0.8258276094844708,
      "learning_rate": 6.916111234898827e-06,
      "loss": 0.392,
      "step": 6283
    },
    {
      "epoch": 0.39361718786701955,
      "grad_norm": 0.7910217169238906,
      "learning_rate": 6.915174239895313e-06,
      "loss": 0.4017,
      "step": 6284
    },
    {
      "epoch": 0.39367982586636185,
      "grad_norm": 0.8378723488869755,
      "learning_rate": 6.9142371660628695e-06,
      "loss": 0.3805,
      "step": 6285
    },
    {
      "epoch": 0.3937424638657041,
      "grad_norm": 0.855887676250797,
      "learning_rate": 6.9133000134400695e-06,
      "loss": 0.4011,
      "step": 6286
    },
    {
      "epoch": 0.3938051018650464,
      "grad_norm": 0.8118153918444826,
      "learning_rate": 6.9123627820654875e-06,
      "loss": 0.4043,
      "step": 6287
    },
    {
      "epoch": 0.3938677398643887,
      "grad_norm": 0.8148850499156327,
      "learning_rate": 6.911425471977697e-06,
      "loss": 0.3866,
      "step": 6288
    },
    {
      "epoch": 0.39393037786373103,
      "grad_norm": 0.8442215005396841,
      "learning_rate": 6.910488083215281e-06,
      "loss": 0.4324,
      "step": 6289
    },
    {
      "epoch": 0.39399301586307334,
      "grad_norm": 0.7971181812518628,
      "learning_rate": 6.90955061581682e-06,
      "loss": 0.4229,
      "step": 6290
    },
    {
      "epoch": 0.39405565386241564,
      "grad_norm": 0.7811350998662991,
      "learning_rate": 6.9086130698208995e-06,
      "loss": 0.3772,
      "step": 6291
    },
    {
      "epoch": 0.39411829186175795,
      "grad_norm": 0.7512863780681799,
      "learning_rate": 6.907675445266113e-06,
      "loss": 0.421,
      "step": 6292
    },
    {
      "epoch": 0.39418092986110026,
      "grad_norm": 0.7254876580819134,
      "learning_rate": 6.906737742191051e-06,
      "loss": 0.4671,
      "step": 6293
    },
    {
      "epoch": 0.3942435678604425,
      "grad_norm": 0.780558196411235,
      "learning_rate": 6.905799960634308e-06,
      "loss": 0.4385,
      "step": 6294
    },
    {
      "epoch": 0.3943062058597848,
      "grad_norm": 0.823899157434386,
      "learning_rate": 6.904862100634488e-06,
      "loss": 0.4388,
      "step": 6295
    },
    {
      "epoch": 0.3943688438591271,
      "grad_norm": 0.876835322333112,
      "learning_rate": 6.9039241622301866e-06,
      "loss": 0.4177,
      "step": 6296
    },
    {
      "epoch": 0.39443148185846943,
      "grad_norm": 0.8766837319300567,
      "learning_rate": 6.902986145460016e-06,
      "loss": 0.427,
      "step": 6297
    },
    {
      "epoch": 0.39449411985781174,
      "grad_norm": 0.8664674932316826,
      "learning_rate": 6.902048050362581e-06,
      "loss": 0.4061,
      "step": 6298
    },
    {
      "epoch": 0.39455675785715405,
      "grad_norm": 0.7094547444779266,
      "learning_rate": 6.901109876976495e-06,
      "loss": 0.4669,
      "step": 6299
    },
    {
      "epoch": 0.39461939585649636,
      "grad_norm": 0.8189903832470341,
      "learning_rate": 6.900171625340374e-06,
      "loss": 0.4012,
      "step": 6300
    },
    {
      "epoch": 0.39468203385583867,
      "grad_norm": 0.796573092479714,
      "learning_rate": 6.899233295492838e-06,
      "loss": 0.3648,
      "step": 6301
    },
    {
      "epoch": 0.394744671855181,
      "grad_norm": 0.8224362481539954,
      "learning_rate": 6.898294887472504e-06,
      "loss": 0.4208,
      "step": 6302
    },
    {
      "epoch": 0.3948073098545232,
      "grad_norm": 0.8203043657480685,
      "learning_rate": 6.897356401318002e-06,
      "loss": 0.4169,
      "step": 6303
    },
    {
      "epoch": 0.39486994785386553,
      "grad_norm": 0.8313681627637585,
      "learning_rate": 6.896417837067959e-06,
      "loss": 0.4218,
      "step": 6304
    },
    {
      "epoch": 0.39493258585320784,
      "grad_norm": 0.836634535414638,
      "learning_rate": 6.895479194761003e-06,
      "loss": 0.4584,
      "step": 6305
    },
    {
      "epoch": 0.39499522385255015,
      "grad_norm": 0.6872763201262445,
      "learning_rate": 6.894540474435775e-06,
      "loss": 0.4537,
      "step": 6306
    },
    {
      "epoch": 0.39505786185189246,
      "grad_norm": 0.8303498340580105,
      "learning_rate": 6.8936016761309065e-06,
      "loss": 0.452,
      "step": 6307
    },
    {
      "epoch": 0.39512049985123476,
      "grad_norm": 0.8143878840544966,
      "learning_rate": 6.892662799885044e-06,
      "loss": 0.4183,
      "step": 6308
    },
    {
      "epoch": 0.39518313785057707,
      "grad_norm": 0.8717294475794997,
      "learning_rate": 6.891723845736828e-06,
      "loss": 0.391,
      "step": 6309
    },
    {
      "epoch": 0.3952457758499194,
      "grad_norm": 0.7124160135916298,
      "learning_rate": 6.890784813724907e-06,
      "loss": 0.4898,
      "step": 6310
    },
    {
      "epoch": 0.39530841384926163,
      "grad_norm": 0.7876945181929432,
      "learning_rate": 6.889845703887932e-06,
      "loss": 0.4113,
      "step": 6311
    },
    {
      "epoch": 0.39537105184860394,
      "grad_norm": 0.8536426052503389,
      "learning_rate": 6.888906516264559e-06,
      "loss": 0.4176,
      "step": 6312
    },
    {
      "epoch": 0.39543368984794625,
      "grad_norm": 0.6708749857189626,
      "learning_rate": 6.887967250893441e-06,
      "loss": 0.481,
      "step": 6313
    },
    {
      "epoch": 0.39549632784728855,
      "grad_norm": 0.7869677222775309,
      "learning_rate": 6.887027907813241e-06,
      "loss": 0.4269,
      "step": 6314
    },
    {
      "epoch": 0.39555896584663086,
      "grad_norm": 0.8168080957712988,
      "learning_rate": 6.886088487062622e-06,
      "loss": 0.4285,
      "step": 6315
    },
    {
      "epoch": 0.39562160384597317,
      "grad_norm": 2.0450401033313454,
      "learning_rate": 6.885148988680253e-06,
      "loss": 0.3944,
      "step": 6316
    },
    {
      "epoch": 0.3956842418453155,
      "grad_norm": 0.7015108381608076,
      "learning_rate": 6.8842094127048e-06,
      "loss": 0.4533,
      "step": 6317
    },
    {
      "epoch": 0.3957468798446578,
      "grad_norm": 0.7718621416442655,
      "learning_rate": 6.883269759174938e-06,
      "loss": 0.377,
      "step": 6318
    },
    {
      "epoch": 0.39580951784400004,
      "grad_norm": 0.8312395770818326,
      "learning_rate": 6.882330028129342e-06,
      "loss": 0.4129,
      "step": 6319
    },
    {
      "epoch": 0.39587215584334234,
      "grad_norm": 0.9328840886304656,
      "learning_rate": 6.881390219606694e-06,
      "loss": 0.4133,
      "step": 6320
    },
    {
      "epoch": 0.39593479384268465,
      "grad_norm": 0.8082285266961456,
      "learning_rate": 6.880450333645675e-06,
      "loss": 0.439,
      "step": 6321
    },
    {
      "epoch": 0.39599743184202696,
      "grad_norm": 0.7930596472839121,
      "learning_rate": 6.879510370284972e-06,
      "loss": 0.3932,
      "step": 6322
    },
    {
      "epoch": 0.39606006984136927,
      "grad_norm": 0.8644958153208895,
      "learning_rate": 6.878570329563273e-06,
      "loss": 0.4717,
      "step": 6323
    },
    {
      "epoch": 0.3961227078407116,
      "grad_norm": 0.819033472091712,
      "learning_rate": 6.877630211519268e-06,
      "loss": 0.4146,
      "step": 6324
    },
    {
      "epoch": 0.3961853458400539,
      "grad_norm": 1.0252521711277973,
      "learning_rate": 6.876690016191658e-06,
      "loss": 0.4358,
      "step": 6325
    },
    {
      "epoch": 0.3962479838393962,
      "grad_norm": 0.6733161201884185,
      "learning_rate": 6.875749743619138e-06,
      "loss": 0.4572,
      "step": 6326
    },
    {
      "epoch": 0.3963106218387385,
      "grad_norm": 0.9421641541324192,
      "learning_rate": 6.874809393840411e-06,
      "loss": 0.4327,
      "step": 6327
    },
    {
      "epoch": 0.39637325983808075,
      "grad_norm": 0.8977655712839151,
      "learning_rate": 6.8738689668941795e-06,
      "loss": 0.426,
      "step": 6328
    },
    {
      "epoch": 0.39643589783742306,
      "grad_norm": 0.8745066982873205,
      "learning_rate": 6.872928462819155e-06,
      "loss": 0.4383,
      "step": 6329
    },
    {
      "epoch": 0.39649853583676536,
      "grad_norm": 0.8836080852204947,
      "learning_rate": 6.871987881654047e-06,
      "loss": 0.4352,
      "step": 6330
    },
    {
      "epoch": 0.39656117383610767,
      "grad_norm": 0.8115570183919039,
      "learning_rate": 6.87104722343757e-06,
      "loss": 0.3998,
      "step": 6331
    },
    {
      "epoch": 0.39662381183545,
      "grad_norm": 0.7793007264625997,
      "learning_rate": 6.8701064882084424e-06,
      "loss": 0.4016,
      "step": 6332
    },
    {
      "epoch": 0.3966864498347923,
      "grad_norm": 0.8144537473697347,
      "learning_rate": 6.8691656760053836e-06,
      "loss": 0.4802,
      "step": 6333
    },
    {
      "epoch": 0.3967490878341346,
      "grad_norm": 0.865266738859267,
      "learning_rate": 6.868224786867119e-06,
      "loss": 0.449,
      "step": 6334
    },
    {
      "epoch": 0.3968117258334769,
      "grad_norm": 0.8224541094070261,
      "learning_rate": 6.867283820832378e-06,
      "loss": 0.3769,
      "step": 6335
    },
    {
      "epoch": 0.39687436383281915,
      "grad_norm": 0.7729915032625186,
      "learning_rate": 6.866342777939885e-06,
      "loss": 0.3832,
      "step": 6336
    },
    {
      "epoch": 0.39693700183216146,
      "grad_norm": 0.9838188515397915,
      "learning_rate": 6.865401658228378e-06,
      "loss": 0.4348,
      "step": 6337
    },
    {
      "epoch": 0.39699963983150377,
      "grad_norm": 0.8252540253493298,
      "learning_rate": 6.864460461736593e-06,
      "loss": 0.4427,
      "step": 6338
    },
    {
      "epoch": 0.3970622778308461,
      "grad_norm": 0.7817851674045069,
      "learning_rate": 6.863519188503269e-06,
      "loss": 0.4076,
      "step": 6339
    },
    {
      "epoch": 0.3971249158301884,
      "grad_norm": 0.9121907855689086,
      "learning_rate": 6.862577838567149e-06,
      "loss": 0.4148,
      "step": 6340
    },
    {
      "epoch": 0.3971875538295307,
      "grad_norm": 0.8390963161940983,
      "learning_rate": 6.86163641196698e-06,
      "loss": 0.4472,
      "step": 6341
    },
    {
      "epoch": 0.397250191828873,
      "grad_norm": 0.8281172416523275,
      "learning_rate": 6.8606949087415096e-06,
      "loss": 0.3981,
      "step": 6342
    },
    {
      "epoch": 0.3973128298282153,
      "grad_norm": 0.8601168269074425,
      "learning_rate": 6.8597533289294925e-06,
      "loss": 0.4203,
      "step": 6343
    },
    {
      "epoch": 0.39737546782755756,
      "grad_norm": 0.883680905301307,
      "learning_rate": 6.858811672569683e-06,
      "loss": 0.4325,
      "step": 6344
    },
    {
      "epoch": 0.39743810582689987,
      "grad_norm": 0.7522058421283124,
      "learning_rate": 6.85786993970084e-06,
      "loss": 0.3927,
      "step": 6345
    },
    {
      "epoch": 0.3975007438262422,
      "grad_norm": 0.8896618639639021,
      "learning_rate": 6.856928130361725e-06,
      "loss": 0.4612,
      "step": 6346
    },
    {
      "epoch": 0.3975633818255845,
      "grad_norm": 0.9247982838803341,
      "learning_rate": 6.855986244591104e-06,
      "loss": 0.396,
      "step": 6347
    },
    {
      "epoch": 0.3976260198249268,
      "grad_norm": 0.8222457926744456,
      "learning_rate": 6.855044282427746e-06,
      "loss": 0.4449,
      "step": 6348
    },
    {
      "epoch": 0.3976886578242691,
      "grad_norm": 0.829784899019049,
      "learning_rate": 6.854102243910419e-06,
      "loss": 0.4397,
      "step": 6349
    },
    {
      "epoch": 0.3977512958236114,
      "grad_norm": 0.8159934804588591,
      "learning_rate": 6.8531601290779e-06,
      "loss": 0.4316,
      "step": 6350
    },
    {
      "epoch": 0.3978139338229537,
      "grad_norm": 0.6331360625085544,
      "learning_rate": 6.852217937968965e-06,
      "loss": 0.5045,
      "step": 6351
    },
    {
      "epoch": 0.397876571822296,
      "grad_norm": 0.851968204356374,
      "learning_rate": 6.851275670622398e-06,
      "loss": 0.4185,
      "step": 6352
    },
    {
      "epoch": 0.3979392098216383,
      "grad_norm": 0.8402547032055605,
      "learning_rate": 6.8503333270769815e-06,
      "loss": 0.4351,
      "step": 6353
    },
    {
      "epoch": 0.3980018478209806,
      "grad_norm": 0.7530173020010008,
      "learning_rate": 6.8493909073714995e-06,
      "loss": 0.372,
      "step": 6354
    },
    {
      "epoch": 0.3980644858203229,
      "grad_norm": 0.8036174384252391,
      "learning_rate": 6.848448411544744e-06,
      "loss": 0.4032,
      "step": 6355
    },
    {
      "epoch": 0.3981271238196652,
      "grad_norm": 0.8072328639563744,
      "learning_rate": 6.847505839635511e-06,
      "loss": 0.406,
      "step": 6356
    },
    {
      "epoch": 0.3981897618190075,
      "grad_norm": 0.7726447413631254,
      "learning_rate": 6.846563191682594e-06,
      "loss": 0.3928,
      "step": 6357
    },
    {
      "epoch": 0.3982523998183498,
      "grad_norm": 0.8000059171658257,
      "learning_rate": 6.845620467724792e-06,
      "loss": 0.3639,
      "step": 6358
    },
    {
      "epoch": 0.3983150378176921,
      "grad_norm": 0.7798490423731174,
      "learning_rate": 6.844677667800911e-06,
      "loss": 0.4406,
      "step": 6359
    },
    {
      "epoch": 0.3983776758170344,
      "grad_norm": 0.8245188177443304,
      "learning_rate": 6.843734791949753e-06,
      "loss": 0.4148,
      "step": 6360
    },
    {
      "epoch": 0.3984403138163767,
      "grad_norm": 0.9889205777896256,
      "learning_rate": 6.84279184021013e-06,
      "loss": 0.4303,
      "step": 6361
    },
    {
      "epoch": 0.398502951815719,
      "grad_norm": 0.8244174618300955,
      "learning_rate": 6.841848812620853e-06,
      "loss": 0.4074,
      "step": 6362
    },
    {
      "epoch": 0.3985655898150613,
      "grad_norm": 0.6447318529432328,
      "learning_rate": 6.840905709220735e-06,
      "loss": 0.4728,
      "step": 6363
    },
    {
      "epoch": 0.3986282278144036,
      "grad_norm": 0.8080995609170601,
      "learning_rate": 6.839962530048599e-06,
      "loss": 0.3815,
      "step": 6364
    },
    {
      "epoch": 0.3986908658137459,
      "grad_norm": 0.8685975735584517,
      "learning_rate": 6.839019275143262e-06,
      "loss": 0.4148,
      "step": 6365
    },
    {
      "epoch": 0.3987535038130882,
      "grad_norm": 0.8404582728704616,
      "learning_rate": 6.838075944543552e-06,
      "loss": 0.4362,
      "step": 6366
    },
    {
      "epoch": 0.3988161418124305,
      "grad_norm": 0.852284694581187,
      "learning_rate": 6.837132538288292e-06,
      "loss": 0.4287,
      "step": 6367
    },
    {
      "epoch": 0.39887877981177283,
      "grad_norm": 0.7459673375411182,
      "learning_rate": 6.836189056416318e-06,
      "loss": 0.3802,
      "step": 6368
    },
    {
      "epoch": 0.39894141781111514,
      "grad_norm": 0.8104111726539314,
      "learning_rate": 6.835245498966461e-06,
      "loss": 0.3858,
      "step": 6369
    },
    {
      "epoch": 0.3990040558104574,
      "grad_norm": 0.7918591931661588,
      "learning_rate": 6.834301865977559e-06,
      "loss": 0.4001,
      "step": 6370
    },
    {
      "epoch": 0.3990666938097997,
      "grad_norm": 0.8704700925348323,
      "learning_rate": 6.833358157488451e-06,
      "loss": 0.4245,
      "step": 6371
    },
    {
      "epoch": 0.399129331809142,
      "grad_norm": 0.7935038404837981,
      "learning_rate": 6.832414373537982e-06,
      "loss": 0.4033,
      "step": 6372
    },
    {
      "epoch": 0.3991919698084843,
      "grad_norm": 0.8078686625683242,
      "learning_rate": 6.831470514164997e-06,
      "loss": 0.411,
      "step": 6373
    },
    {
      "epoch": 0.3992546078078266,
      "grad_norm": 0.8690375375337074,
      "learning_rate": 6.830526579408345e-06,
      "loss": 0.4354,
      "step": 6374
    },
    {
      "epoch": 0.39931724580716893,
      "grad_norm": 0.824421277618791,
      "learning_rate": 6.829582569306879e-06,
      "loss": 0.3927,
      "step": 6375
    },
    {
      "epoch": 0.39937988380651124,
      "grad_norm": 0.7790198260744073,
      "learning_rate": 6.828638483899456e-06,
      "loss": 0.3681,
      "step": 6376
    },
    {
      "epoch": 0.39944252180585355,
      "grad_norm": 0.838261918804857,
      "learning_rate": 6.8276943232249335e-06,
      "loss": 0.3826,
      "step": 6377
    },
    {
      "epoch": 0.3995051598051958,
      "grad_norm": 0.8904527253412109,
      "learning_rate": 6.826750087322173e-06,
      "loss": 0.4654,
      "step": 6378
    },
    {
      "epoch": 0.3995677978045381,
      "grad_norm": 0.8174222323786923,
      "learning_rate": 6.82580577623004e-06,
      "loss": 0.3929,
      "step": 6379
    },
    {
      "epoch": 0.3996304358038804,
      "grad_norm": 0.7922601476359623,
      "learning_rate": 6.824861389987402e-06,
      "loss": 0.4069,
      "step": 6380
    },
    {
      "epoch": 0.3996930738032227,
      "grad_norm": 0.8385094529216812,
      "learning_rate": 6.8239169286331304e-06,
      "loss": 0.4407,
      "step": 6381
    },
    {
      "epoch": 0.39975571180256503,
      "grad_norm": 0.7690404316400153,
      "learning_rate": 6.8229723922061e-06,
      "loss": 0.4762,
      "step": 6382
    },
    {
      "epoch": 0.39981834980190734,
      "grad_norm": 0.9757148709114507,
      "learning_rate": 6.822027780745187e-06,
      "loss": 0.4652,
      "step": 6383
    },
    {
      "epoch": 0.39988098780124964,
      "grad_norm": 0.8971417909099884,
      "learning_rate": 6.8210830942892724e-06,
      "loss": 0.4509,
      "step": 6384
    },
    {
      "epoch": 0.39994362580059195,
      "grad_norm": 0.8847282215961554,
      "learning_rate": 6.82013833287724e-06,
      "loss": 0.425,
      "step": 6385
    },
    {
      "epoch": 0.4000062637999342,
      "grad_norm": 0.7733840348360167,
      "learning_rate": 6.819193496547977e-06,
      "loss": 0.3928,
      "step": 6386
    },
    {
      "epoch": 0.4000689017992765,
      "grad_norm": 0.8303050832903448,
      "learning_rate": 6.81824858534037e-06,
      "loss": 0.4033,
      "step": 6387
    },
    {
      "epoch": 0.4001315397986188,
      "grad_norm": 0.8247018973521184,
      "learning_rate": 6.817303599293315e-06,
      "loss": 0.376,
      "step": 6388
    },
    {
      "epoch": 0.4001941777979611,
      "grad_norm": 0.8112047173034855,
      "learning_rate": 6.816358538445707e-06,
      "loss": 0.3948,
      "step": 6389
    },
    {
      "epoch": 0.40025681579730343,
      "grad_norm": 0.8008192238190006,
      "learning_rate": 6.815413402836443e-06,
      "loss": 0.4044,
      "step": 6390
    },
    {
      "epoch": 0.40031945379664574,
      "grad_norm": 0.9737781646931999,
      "learning_rate": 6.814468192504428e-06,
      "loss": 0.4173,
      "step": 6391
    },
    {
      "epoch": 0.40038209179598805,
      "grad_norm": 0.7787507184677895,
      "learning_rate": 6.813522907488562e-06,
      "loss": 0.3483,
      "step": 6392
    },
    {
      "epoch": 0.40044472979533036,
      "grad_norm": 0.8311774836184437,
      "learning_rate": 6.81257754782776e-06,
      "loss": 0.392,
      "step": 6393
    },
    {
      "epoch": 0.40050736779467266,
      "grad_norm": 0.8445187807085428,
      "learning_rate": 6.811632113560929e-06,
      "loss": 0.3932,
      "step": 6394
    },
    {
      "epoch": 0.4005700057940149,
      "grad_norm": 0.8921768850642366,
      "learning_rate": 6.810686604726984e-06,
      "loss": 0.4139,
      "step": 6395
    },
    {
      "epoch": 0.4006326437933572,
      "grad_norm": 0.8357872183427232,
      "learning_rate": 6.8097410213648415e-06,
      "loss": 0.437,
      "step": 6396
    },
    {
      "epoch": 0.40069528179269953,
      "grad_norm": 0.7858654424164156,
      "learning_rate": 6.808795363513423e-06,
      "loss": 0.4145,
      "step": 6397
    },
    {
      "epoch": 0.40075791979204184,
      "grad_norm": 0.7790712710951821,
      "learning_rate": 6.807849631211652e-06,
      "loss": 0.3932,
      "step": 6398
    },
    {
      "epoch": 0.40082055779138415,
      "grad_norm": 0.7940580314134307,
      "learning_rate": 6.806903824498454e-06,
      "loss": 0.4094,
      "step": 6399
    },
    {
      "epoch": 0.40088319579072645,
      "grad_norm": 0.8003574585258142,
      "learning_rate": 6.80595794341276e-06,
      "loss": 0.4281,
      "step": 6400
    },
    {
      "epoch": 0.40094583379006876,
      "grad_norm": 0.8401906852941367,
      "learning_rate": 6.8050119879935e-06,
      "loss": 0.4066,
      "step": 6401
    },
    {
      "epoch": 0.40100847178941107,
      "grad_norm": 0.8492971307112245,
      "learning_rate": 6.804065958279613e-06,
      "loss": 0.4518,
      "step": 6402
    },
    {
      "epoch": 0.4010711097887533,
      "grad_norm": 0.8778791219653796,
      "learning_rate": 6.803119854310037e-06,
      "loss": 0.4024,
      "step": 6403
    },
    {
      "epoch": 0.40113374778809563,
      "grad_norm": 0.896036892459301,
      "learning_rate": 6.802173676123713e-06,
      "loss": 0.4568,
      "step": 6404
    },
    {
      "epoch": 0.40119638578743794,
      "grad_norm": 0.8539718201555554,
      "learning_rate": 6.801227423759585e-06,
      "loss": 0.4686,
      "step": 6405
    },
    {
      "epoch": 0.40125902378678024,
      "grad_norm": 0.7864898407547245,
      "learning_rate": 6.800281097256601e-06,
      "loss": 0.4007,
      "step": 6406
    },
    {
      "epoch": 0.40132166178612255,
      "grad_norm": 0.7938843761586867,
      "learning_rate": 6.7993346966537145e-06,
      "loss": 0.4012,
      "step": 6407
    },
    {
      "epoch": 0.40138429978546486,
      "grad_norm": 0.8166189816253119,
      "learning_rate": 6.798388221989878e-06,
      "loss": 0.3999,
      "step": 6408
    },
    {
      "epoch": 0.40144693778480717,
      "grad_norm": 0.8081833928359938,
      "learning_rate": 6.797441673304049e-06,
      "loss": 0.3947,
      "step": 6409
    },
    {
      "epoch": 0.4015095757841495,
      "grad_norm": 0.794825330484546,
      "learning_rate": 6.7964950506351855e-06,
      "loss": 0.4549,
      "step": 6410
    },
    {
      "epoch": 0.4015722137834918,
      "grad_norm": 0.7811106266302471,
      "learning_rate": 6.795548354022252e-06,
      "loss": 0.375,
      "step": 6411
    },
    {
      "epoch": 0.40163485178283403,
      "grad_norm": 0.9687537930667443,
      "learning_rate": 6.794601583504216e-06,
      "loss": 0.4148,
      "step": 6412
    },
    {
      "epoch": 0.40169748978217634,
      "grad_norm": 0.8783834591513017,
      "learning_rate": 6.793654739120046e-06,
      "loss": 0.4626,
      "step": 6413
    },
    {
      "epoch": 0.40176012778151865,
      "grad_norm": 0.8431566021766386,
      "learning_rate": 6.792707820908714e-06,
      "loss": 0.4449,
      "step": 6414
    },
    {
      "epoch": 0.40182276578086096,
      "grad_norm": 0.8082148928689391,
      "learning_rate": 6.791760828909194e-06,
      "loss": 0.39,
      "step": 6415
    },
    {
      "epoch": 0.40188540378020327,
      "grad_norm": 0.8694915797763153,
      "learning_rate": 6.790813763160467e-06,
      "loss": 0.4202,
      "step": 6416
    },
    {
      "epoch": 0.4019480417795456,
      "grad_norm": 0.8815217959516644,
      "learning_rate": 6.789866623701513e-06,
      "loss": 0.4415,
      "step": 6417
    },
    {
      "epoch": 0.4020106797788879,
      "grad_norm": 0.8231988178860445,
      "learning_rate": 6.788919410571316e-06,
      "loss": 0.4052,
      "step": 6418
    },
    {
      "epoch": 0.4020733177782302,
      "grad_norm": 0.829999176794089,
      "learning_rate": 6.7879721238088655e-06,
      "loss": 0.3942,
      "step": 6419
    },
    {
      "epoch": 0.40213595577757244,
      "grad_norm": 0.9383274127110577,
      "learning_rate": 6.787024763453148e-06,
      "loss": 0.4134,
      "step": 6420
    },
    {
      "epoch": 0.40219859377691475,
      "grad_norm": 0.8350866181518469,
      "learning_rate": 6.786077329543161e-06,
      "loss": 0.4825,
      "step": 6421
    },
    {
      "epoch": 0.40226123177625706,
      "grad_norm": 0.8022422791862784,
      "learning_rate": 6.7851298221179e-06,
      "loss": 0.4058,
      "step": 6422
    },
    {
      "epoch": 0.40232386977559936,
      "grad_norm": 0.7874835225581366,
      "learning_rate": 6.784182241216364e-06,
      "loss": 0.3961,
      "step": 6423
    },
    {
      "epoch": 0.40238650777494167,
      "grad_norm": 0.7965507608398726,
      "learning_rate": 6.783234586877555e-06,
      "loss": 0.414,
      "step": 6424
    },
    {
      "epoch": 0.402449145774284,
      "grad_norm": 0.7694903575833083,
      "learning_rate": 6.782286859140481e-06,
      "loss": 0.3722,
      "step": 6425
    },
    {
      "epoch": 0.4025117837736263,
      "grad_norm": 0.8458940848879797,
      "learning_rate": 6.781339058044147e-06,
      "loss": 0.4017,
      "step": 6426
    },
    {
      "epoch": 0.4025744217729686,
      "grad_norm": 0.8120391312376575,
      "learning_rate": 6.780391183627569e-06,
      "loss": 0.3873,
      "step": 6427
    },
    {
      "epoch": 0.40263705977231085,
      "grad_norm": 0.6911448806797302,
      "learning_rate": 6.779443235929758e-06,
      "loss": 0.4441,
      "step": 6428
    },
    {
      "epoch": 0.40269969777165315,
      "grad_norm": 0.8098440036459258,
      "learning_rate": 6.778495214989733e-06,
      "loss": 0.4171,
      "step": 6429
    },
    {
      "epoch": 0.40276233577099546,
      "grad_norm": 0.8248336706536313,
      "learning_rate": 6.777547120846515e-06,
      "loss": 0.4377,
      "step": 6430
    },
    {
      "epoch": 0.40282497377033777,
      "grad_norm": 0.8545665351836161,
      "learning_rate": 6.7765989535391275e-06,
      "loss": 0.403,
      "step": 6431
    },
    {
      "epoch": 0.4028876117696801,
      "grad_norm": 0.9103471707414984,
      "learning_rate": 6.775650713106599e-06,
      "loss": 0.4291,
      "step": 6432
    },
    {
      "epoch": 0.4029502497690224,
      "grad_norm": 0.8841170228959101,
      "learning_rate": 6.774702399587956e-06,
      "loss": 0.4356,
      "step": 6433
    },
    {
      "epoch": 0.4030128877683647,
      "grad_norm": 0.7951623744810526,
      "learning_rate": 6.7737540130222314e-06,
      "loss": 0.4116,
      "step": 6434
    },
    {
      "epoch": 0.403075525767707,
      "grad_norm": 0.7995212955776113,
      "learning_rate": 6.772805553448464e-06,
      "loss": 0.393,
      "step": 6435
    },
    {
      "epoch": 0.4031381637670493,
      "grad_norm": 0.879075126326359,
      "learning_rate": 6.771857020905691e-06,
      "loss": 0.4477,
      "step": 6436
    },
    {
      "epoch": 0.40320080176639156,
      "grad_norm": 0.8307329455374457,
      "learning_rate": 6.770908415432955e-06,
      "loss": 0.4023,
      "step": 6437
    },
    {
      "epoch": 0.40326343976573387,
      "grad_norm": 0.8774697822640689,
      "learning_rate": 6.769959737069298e-06,
      "loss": 0.447,
      "step": 6438
    },
    {
      "epoch": 0.4033260777650762,
      "grad_norm": 0.8361553613867597,
      "learning_rate": 6.76901098585377e-06,
      "loss": 0.4003,
      "step": 6439
    },
    {
      "epoch": 0.4033887157644185,
      "grad_norm": 0.8562617240694547,
      "learning_rate": 6.7680621618254225e-06,
      "loss": 0.4165,
      "step": 6440
    },
    {
      "epoch": 0.4034513537637608,
      "grad_norm": 0.6604923094595945,
      "learning_rate": 6.767113265023309e-06,
      "loss": 0.4835,
      "step": 6441
    },
    {
      "epoch": 0.4035139917631031,
      "grad_norm": 0.8902875747607314,
      "learning_rate": 6.766164295486485e-06,
      "loss": 0.4128,
      "step": 6442
    },
    {
      "epoch": 0.4035766297624454,
      "grad_norm": 0.9104285177946264,
      "learning_rate": 6.765215253254009e-06,
      "loss": 0.4363,
      "step": 6443
    },
    {
      "epoch": 0.4036392677617877,
      "grad_norm": 0.8395230047552482,
      "learning_rate": 6.764266138364946e-06,
      "loss": 0.4209,
      "step": 6444
    },
    {
      "epoch": 0.40370190576112996,
      "grad_norm": 0.8158571737855719,
      "learning_rate": 6.763316950858363e-06,
      "loss": 0.3957,
      "step": 6445
    },
    {
      "epoch": 0.40376454376047227,
      "grad_norm": 0.8182496797048098,
      "learning_rate": 6.762367690773327e-06,
      "loss": 0.4035,
      "step": 6446
    },
    {
      "epoch": 0.4038271817598146,
      "grad_norm": 0.8140933028455337,
      "learning_rate": 6.761418358148908e-06,
      "loss": 0.4121,
      "step": 6447
    },
    {
      "epoch": 0.4038898197591569,
      "grad_norm": 0.8696609413981886,
      "learning_rate": 6.760468953024183e-06,
      "loss": 0.4126,
      "step": 6448
    },
    {
      "epoch": 0.4039524577584992,
      "grad_norm": 0.788989456378461,
      "learning_rate": 6.75951947543823e-06,
      "loss": 0.4111,
      "step": 6449
    },
    {
      "epoch": 0.4040150957578415,
      "grad_norm": 0.8372852165467597,
      "learning_rate": 6.7585699254301284e-06,
      "loss": 0.4142,
      "step": 6450
    },
    {
      "epoch": 0.4040777337571838,
      "grad_norm": 0.777927684910304,
      "learning_rate": 6.7576203030389635e-06,
      "loss": 0.4292,
      "step": 6451
    },
    {
      "epoch": 0.4041403717565261,
      "grad_norm": 0.8960890467942997,
      "learning_rate": 6.756670608303817e-06,
      "loss": 0.4329,
      "step": 6452
    },
    {
      "epoch": 0.40420300975586837,
      "grad_norm": 0.8299166689532159,
      "learning_rate": 6.755720841263784e-06,
      "loss": 0.3834,
      "step": 6453
    },
    {
      "epoch": 0.4042656477552107,
      "grad_norm": 0.7727586882732359,
      "learning_rate": 6.754771001957957e-06,
      "loss": 0.4025,
      "step": 6454
    },
    {
      "epoch": 0.404328285754553,
      "grad_norm": 0.8868280546085816,
      "learning_rate": 6.753821090425429e-06,
      "loss": 0.4381,
      "step": 6455
    },
    {
      "epoch": 0.4043909237538953,
      "grad_norm": 0.7168776164307178,
      "learning_rate": 6.752871106705299e-06,
      "loss": 0.4749,
      "step": 6456
    },
    {
      "epoch": 0.4044535617532376,
      "grad_norm": 0.8306994237463987,
      "learning_rate": 6.751921050836667e-06,
      "loss": 0.4016,
      "step": 6457
    },
    {
      "epoch": 0.4045161997525799,
      "grad_norm": 0.7986697266878768,
      "learning_rate": 6.750970922858643e-06,
      "loss": 0.4063,
      "step": 6458
    },
    {
      "epoch": 0.4045788377519222,
      "grad_norm": 0.8580199833799009,
      "learning_rate": 6.750020722810329e-06,
      "loss": 0.4052,
      "step": 6459
    },
    {
      "epoch": 0.4046414757512645,
      "grad_norm": 0.8432514360439013,
      "learning_rate": 6.749070450730837e-06,
      "loss": 0.4038,
      "step": 6460
    },
    {
      "epoch": 0.40470411375060683,
      "grad_norm": 0.8403423414266447,
      "learning_rate": 6.74812010665928e-06,
      "loss": 0.3854,
      "step": 6461
    },
    {
      "epoch": 0.4047667517499491,
      "grad_norm": 0.8262315958736968,
      "learning_rate": 6.747169690634776e-06,
      "loss": 0.4023,
      "step": 6462
    },
    {
      "epoch": 0.4048293897492914,
      "grad_norm": 0.8116316817898241,
      "learning_rate": 6.746219202696445e-06,
      "loss": 0.3882,
      "step": 6463
    },
    {
      "epoch": 0.4048920277486337,
      "grad_norm": 0.7883304304774177,
      "learning_rate": 6.7452686428834045e-06,
      "loss": 0.4263,
      "step": 6464
    },
    {
      "epoch": 0.404954665747976,
      "grad_norm": 0.6685872898162585,
      "learning_rate": 6.744318011234784e-06,
      "loss": 0.4491,
      "step": 6465
    },
    {
      "epoch": 0.4050173037473183,
      "grad_norm": 0.8336302612932646,
      "learning_rate": 6.74336730778971e-06,
      "loss": 0.4182,
      "step": 6466
    },
    {
      "epoch": 0.4050799417466606,
      "grad_norm": 0.834169317335812,
      "learning_rate": 6.742416532587315e-06,
      "loss": 0.4048,
      "step": 6467
    },
    {
      "epoch": 0.40514257974600293,
      "grad_norm": 0.858668071947745,
      "learning_rate": 6.7414656856667315e-06,
      "loss": 0.3951,
      "step": 6468
    },
    {
      "epoch": 0.40520521774534524,
      "grad_norm": 0.8323844977583631,
      "learning_rate": 6.740514767067097e-06,
      "loss": 0.3964,
      "step": 6469
    },
    {
      "epoch": 0.4052678557446875,
      "grad_norm": 0.7680100515973821,
      "learning_rate": 6.739563776827551e-06,
      "loss": 0.3818,
      "step": 6470
    },
    {
      "epoch": 0.4053304937440298,
      "grad_norm": 0.8793156584712862,
      "learning_rate": 6.738612714987238e-06,
      "loss": 0.4152,
      "step": 6471
    },
    {
      "epoch": 0.4053931317433721,
      "grad_norm": 0.8340961931989205,
      "learning_rate": 6.737661581585303e-06,
      "loss": 0.4319,
      "step": 6472
    },
    {
      "epoch": 0.4054557697427144,
      "grad_norm": 0.9287751577177514,
      "learning_rate": 6.736710376660893e-06,
      "loss": 0.4456,
      "step": 6473
    },
    {
      "epoch": 0.4055184077420567,
      "grad_norm": 0.788501455504818,
      "learning_rate": 6.735759100253164e-06,
      "loss": 0.42,
      "step": 6474
    },
    {
      "epoch": 0.405581045741399,
      "grad_norm": 0.8482168144720785,
      "learning_rate": 6.734807752401267e-06,
      "loss": 0.4245,
      "step": 6475
    },
    {
      "epoch": 0.40564368374074133,
      "grad_norm": 0.8102073336263935,
      "learning_rate": 6.7338563331443614e-06,
      "loss": 0.3972,
      "step": 6476
    },
    {
      "epoch": 0.40570632174008364,
      "grad_norm": 0.7990280007342405,
      "learning_rate": 6.732904842521607e-06,
      "loss": 0.4304,
      "step": 6477
    },
    {
      "epoch": 0.40576895973942595,
      "grad_norm": 0.8393782752493066,
      "learning_rate": 6.731953280572167e-06,
      "loss": 0.4563,
      "step": 6478
    },
    {
      "epoch": 0.4058315977387682,
      "grad_norm": 0.7966267463914509,
      "learning_rate": 6.731001647335211e-06,
      "loss": 0.3599,
      "step": 6479
    },
    {
      "epoch": 0.4058942357381105,
      "grad_norm": 0.8181905628843751,
      "learning_rate": 6.730049942849904e-06,
      "loss": 0.4261,
      "step": 6480
    },
    {
      "epoch": 0.4059568737374528,
      "grad_norm": 0.8420849004739979,
      "learning_rate": 6.72909816715542e-06,
      "loss": 0.4276,
      "step": 6481
    },
    {
      "epoch": 0.4060195117367951,
      "grad_norm": 0.9165346310224523,
      "learning_rate": 6.728146320290935e-06,
      "loss": 0.4224,
      "step": 6482
    },
    {
      "epoch": 0.40608214973613743,
      "grad_norm": 0.846173492308066,
      "learning_rate": 6.727194402295627e-06,
      "loss": 0.3841,
      "step": 6483
    },
    {
      "epoch": 0.40614478773547974,
      "grad_norm": 0.6477116128654088,
      "learning_rate": 6.7262424132086755e-06,
      "loss": 0.4799,
      "step": 6484
    },
    {
      "epoch": 0.40620742573482205,
      "grad_norm": 0.888375444727439,
      "learning_rate": 6.725290353069267e-06,
      "loss": 0.4431,
      "step": 6485
    },
    {
      "epoch": 0.40627006373416436,
      "grad_norm": 0.7548253212440292,
      "learning_rate": 6.724338221916587e-06,
      "loss": 0.4218,
      "step": 6486
    },
    {
      "epoch": 0.4063327017335066,
      "grad_norm": 0.8253157010729322,
      "learning_rate": 6.723386019789826e-06,
      "loss": 0.4275,
      "step": 6487
    },
    {
      "epoch": 0.4063953397328489,
      "grad_norm": 0.6825078230641264,
      "learning_rate": 6.722433746728176e-06,
      "loss": 0.4848,
      "step": 6488
    },
    {
      "epoch": 0.4064579777321912,
      "grad_norm": 0.855495453500215,
      "learning_rate": 6.721481402770834e-06,
      "loss": 0.4507,
      "step": 6489
    },
    {
      "epoch": 0.40652061573153353,
      "grad_norm": 0.8142081111072991,
      "learning_rate": 6.720528987956997e-06,
      "loss": 0.4127,
      "step": 6490
    },
    {
      "epoch": 0.40658325373087584,
      "grad_norm": 0.8063838244299051,
      "learning_rate": 6.7195765023258685e-06,
      "loss": 0.4235,
      "step": 6491
    },
    {
      "epoch": 0.40664589173021815,
      "grad_norm": 0.8379358320188955,
      "learning_rate": 6.7186239459166526e-06,
      "loss": 0.4103,
      "step": 6492
    },
    {
      "epoch": 0.40670852972956045,
      "grad_norm": 0.8513855355205763,
      "learning_rate": 6.717671318768555e-06,
      "loss": 0.4289,
      "step": 6493
    },
    {
      "epoch": 0.40677116772890276,
      "grad_norm": 0.8694472052693493,
      "learning_rate": 6.716718620920787e-06,
      "loss": 0.3998,
      "step": 6494
    },
    {
      "epoch": 0.406833805728245,
      "grad_norm": 0.8287992511284403,
      "learning_rate": 6.715765852412561e-06,
      "loss": 0.4094,
      "step": 6495
    },
    {
      "epoch": 0.4068964437275873,
      "grad_norm": 0.8600466718236827,
      "learning_rate": 6.714813013283097e-06,
      "loss": 0.3803,
      "step": 6496
    },
    {
      "epoch": 0.40695908172692963,
      "grad_norm": 0.8038489644159436,
      "learning_rate": 6.713860103571609e-06,
      "loss": 0.4247,
      "step": 6497
    },
    {
      "epoch": 0.40702171972627194,
      "grad_norm": 0.7978741698156248,
      "learning_rate": 6.712907123317321e-06,
      "loss": 0.3987,
      "step": 6498
    },
    {
      "epoch": 0.40708435772561424,
      "grad_norm": 0.8053436792766513,
      "learning_rate": 6.7119540725594565e-06,
      "loss": 0.3946,
      "step": 6499
    },
    {
      "epoch": 0.40714699572495655,
      "grad_norm": 0.8004156710339524,
      "learning_rate": 6.711000951337247e-06,
      "loss": 0.3436,
      "step": 6500
    },
    {
      "epoch": 0.40720963372429886,
      "grad_norm": 0.754523065565171,
      "learning_rate": 6.710047759689921e-06,
      "loss": 0.4119,
      "step": 6501
    },
    {
      "epoch": 0.40727227172364117,
      "grad_norm": 0.82064697910159,
      "learning_rate": 6.709094497656712e-06,
      "loss": 0.389,
      "step": 6502
    },
    {
      "epoch": 0.4073349097229835,
      "grad_norm": 0.8331982240785295,
      "learning_rate": 6.708141165276854e-06,
      "loss": 0.403,
      "step": 6503
    },
    {
      "epoch": 0.4073975477223257,
      "grad_norm": 0.8273622332740864,
      "learning_rate": 6.7071877625895885e-06,
      "loss": 0.3933,
      "step": 6504
    },
    {
      "epoch": 0.40746018572166803,
      "grad_norm": 0.7858888327913272,
      "learning_rate": 6.7062342896341605e-06,
      "loss": 0.4525,
      "step": 6505
    },
    {
      "epoch": 0.40752282372101034,
      "grad_norm": 0.8350637004003006,
      "learning_rate": 6.705280746449811e-06,
      "loss": 0.4385,
      "step": 6506
    },
    {
      "epoch": 0.40758546172035265,
      "grad_norm": 0.8736135168633334,
      "learning_rate": 6.7043271330757895e-06,
      "loss": 0.4594,
      "step": 6507
    },
    {
      "epoch": 0.40764809971969496,
      "grad_norm": 0.8159432814417733,
      "learning_rate": 6.703373449551346e-06,
      "loss": 0.4005,
      "step": 6508
    },
    {
      "epoch": 0.40771073771903726,
      "grad_norm": 0.7744294809482581,
      "learning_rate": 6.702419695915736e-06,
      "loss": 0.4119,
      "step": 6509
    },
    {
      "epoch": 0.40777337571837957,
      "grad_norm": 0.8433246008723598,
      "learning_rate": 6.701465872208216e-06,
      "loss": 0.4055,
      "step": 6510
    },
    {
      "epoch": 0.4078360137177219,
      "grad_norm": 0.907640237135514,
      "learning_rate": 6.700511978468044e-06,
      "loss": 0.3756,
      "step": 6511
    },
    {
      "epoch": 0.40789865171706413,
      "grad_norm": 0.7750072545646994,
      "learning_rate": 6.699558014734483e-06,
      "loss": 0.4205,
      "step": 6512
    },
    {
      "epoch": 0.40796128971640644,
      "grad_norm": 0.7942581500932295,
      "learning_rate": 6.6986039810468e-06,
      "loss": 0.3977,
      "step": 6513
    },
    {
      "epoch": 0.40802392771574875,
      "grad_norm": 0.8628353391304465,
      "learning_rate": 6.697649877444261e-06,
      "loss": 0.421,
      "step": 6514
    },
    {
      "epoch": 0.40808656571509105,
      "grad_norm": 0.7155335496295095,
      "learning_rate": 6.6966957039661384e-06,
      "loss": 0.4859,
      "step": 6515
    },
    {
      "epoch": 0.40814920371443336,
      "grad_norm": 0.9231343802475772,
      "learning_rate": 6.695741460651706e-06,
      "loss": 0.3946,
      "step": 6516
    },
    {
      "epoch": 0.40821184171377567,
      "grad_norm": 0.8177761488103796,
      "learning_rate": 6.694787147540238e-06,
      "loss": 0.4439,
      "step": 6517
    },
    {
      "epoch": 0.408274479713118,
      "grad_norm": 0.8165192141400391,
      "learning_rate": 6.693832764671019e-06,
      "loss": 0.3654,
      "step": 6518
    },
    {
      "epoch": 0.4083371177124603,
      "grad_norm": 0.7225769038457118,
      "learning_rate": 6.6928783120833286e-06,
      "loss": 0.3816,
      "step": 6519
    },
    {
      "epoch": 0.4083997557118026,
      "grad_norm": 0.8036173414941689,
      "learning_rate": 6.691923789816452e-06,
      "loss": 0.3847,
      "step": 6520
    },
    {
      "epoch": 0.40846239371114484,
      "grad_norm": 0.8671991437094351,
      "learning_rate": 6.690969197909677e-06,
      "loss": 0.4078,
      "step": 6521
    },
    {
      "epoch": 0.40852503171048715,
      "grad_norm": 0.7601463062320596,
      "learning_rate": 6.690014536402298e-06,
      "loss": 0.3876,
      "step": 6522
    },
    {
      "epoch": 0.40858766970982946,
      "grad_norm": 0.8271639625497467,
      "learning_rate": 6.689059805333606e-06,
      "loss": 0.3759,
      "step": 6523
    },
    {
      "epoch": 0.40865030770917177,
      "grad_norm": 0.8604450912080258,
      "learning_rate": 6.6881050047429e-06,
      "loss": 0.4116,
      "step": 6524
    },
    {
      "epoch": 0.4087129457085141,
      "grad_norm": 0.7993659306088696,
      "learning_rate": 6.687150134669478e-06,
      "loss": 0.407,
      "step": 6525
    },
    {
      "epoch": 0.4087755837078564,
      "grad_norm": 0.6892330424977308,
      "learning_rate": 6.686195195152643e-06,
      "loss": 0.4697,
      "step": 6526
    },
    {
      "epoch": 0.4088382217071987,
      "grad_norm": 0.7372722391467658,
      "learning_rate": 6.685240186231701e-06,
      "loss": 0.3854,
      "step": 6527
    },
    {
      "epoch": 0.408900859706541,
      "grad_norm": 0.7496992387511618,
      "learning_rate": 6.6842851079459604e-06,
      "loss": 0.3625,
      "step": 6528
    },
    {
      "epoch": 0.40896349770588325,
      "grad_norm": 0.8862423188982573,
      "learning_rate": 6.683329960334732e-06,
      "loss": 0.4274,
      "step": 6529
    },
    {
      "epoch": 0.40902613570522556,
      "grad_norm": 0.8457200925256549,
      "learning_rate": 6.682374743437328e-06,
      "loss": 0.4056,
      "step": 6530
    },
    {
      "epoch": 0.40908877370456787,
      "grad_norm": 0.7959832088146642,
      "learning_rate": 6.68141945729307e-06,
      "loss": 0.4009,
      "step": 6531
    },
    {
      "epoch": 0.4091514117039102,
      "grad_norm": 0.6353650696993817,
      "learning_rate": 6.6804641019412755e-06,
      "loss": 0.4614,
      "step": 6532
    },
    {
      "epoch": 0.4092140497032525,
      "grad_norm": 0.7186540548719682,
      "learning_rate": 6.679508677421264e-06,
      "loss": 0.3885,
      "step": 6533
    },
    {
      "epoch": 0.4092766877025948,
      "grad_norm": 0.8689173895572502,
      "learning_rate": 6.678553183772366e-06,
      "loss": 0.4135,
      "step": 6534
    },
    {
      "epoch": 0.4093393257019371,
      "grad_norm": 0.8363153840409839,
      "learning_rate": 6.677597621033906e-06,
      "loss": 0.4065,
      "step": 6535
    },
    {
      "epoch": 0.4094019637012794,
      "grad_norm": 0.7651480431347999,
      "learning_rate": 6.676641989245217e-06,
      "loss": 0.3998,
      "step": 6536
    },
    {
      "epoch": 0.40946460170062166,
      "grad_norm": 0.6126424582162525,
      "learning_rate": 6.675686288445632e-06,
      "loss": 0.44,
      "step": 6537
    },
    {
      "epoch": 0.40952723969996396,
      "grad_norm": 0.8297168550054975,
      "learning_rate": 6.6747305186744896e-06,
      "loss": 0.4028,
      "step": 6538
    },
    {
      "epoch": 0.40958987769930627,
      "grad_norm": 0.8766981925146067,
      "learning_rate": 6.673774679971128e-06,
      "loss": 0.4027,
      "step": 6539
    },
    {
      "epoch": 0.4096525156986486,
      "grad_norm": 0.8353807021173819,
      "learning_rate": 6.672818772374888e-06,
      "loss": 0.378,
      "step": 6540
    },
    {
      "epoch": 0.4097151536979909,
      "grad_norm": 0.8740407373086413,
      "learning_rate": 6.6718627959251195e-06,
      "loss": 0.4152,
      "step": 6541
    },
    {
      "epoch": 0.4097777916973332,
      "grad_norm": 0.8159934498026851,
      "learning_rate": 6.670906750661166e-06,
      "loss": 0.4316,
      "step": 6542
    },
    {
      "epoch": 0.4098404296966755,
      "grad_norm": 0.8111562976774983,
      "learning_rate": 6.669950636622381e-06,
      "loss": 0.3875,
      "step": 6543
    },
    {
      "epoch": 0.4099030676960178,
      "grad_norm": 0.7587563017696515,
      "learning_rate": 6.668994453848118e-06,
      "loss": 0.4043,
      "step": 6544
    },
    {
      "epoch": 0.4099657056953601,
      "grad_norm": 0.8250032098228864,
      "learning_rate": 6.668038202377733e-06,
      "loss": 0.3881,
      "step": 6545
    },
    {
      "epoch": 0.41002834369470237,
      "grad_norm": 0.6359517337537562,
      "learning_rate": 6.6670818822505864e-06,
      "loss": 0.4895,
      "step": 6546
    },
    {
      "epoch": 0.4100909816940447,
      "grad_norm": 0.6799599763492861,
      "learning_rate": 6.666125493506041e-06,
      "loss": 0.4704,
      "step": 6547
    },
    {
      "epoch": 0.410153619693387,
      "grad_norm": 0.8079914208967549,
      "learning_rate": 6.665169036183461e-06,
      "loss": 0.3497,
      "step": 6548
    },
    {
      "epoch": 0.4102162576927293,
      "grad_norm": 0.859060835217165,
      "learning_rate": 6.6642125103222125e-06,
      "loss": 0.3953,
      "step": 6549
    },
    {
      "epoch": 0.4102788956920716,
      "grad_norm": 0.7486008025098083,
      "learning_rate": 6.663255915961668e-06,
      "loss": 0.3923,
      "step": 6550
    },
    {
      "epoch": 0.4103415336914139,
      "grad_norm": 0.8418166017876257,
      "learning_rate": 6.662299253141203e-06,
      "loss": 0.3888,
      "step": 6551
    },
    {
      "epoch": 0.4104041716907562,
      "grad_norm": 0.8351753415256618,
      "learning_rate": 6.661342521900194e-06,
      "loss": 0.4211,
      "step": 6552
    },
    {
      "epoch": 0.4104668096900985,
      "grad_norm": 0.8833736404925802,
      "learning_rate": 6.660385722278014e-06,
      "loss": 0.4349,
      "step": 6553
    },
    {
      "epoch": 0.4105294476894408,
      "grad_norm": 0.8103596269068147,
      "learning_rate": 6.659428854314053e-06,
      "loss": 0.4273,
      "step": 6554
    },
    {
      "epoch": 0.4105920856887831,
      "grad_norm": 0.856474848652611,
      "learning_rate": 6.658471918047691e-06,
      "loss": 0.3873,
      "step": 6555
    },
    {
      "epoch": 0.4106547236881254,
      "grad_norm": 0.945832469880253,
      "learning_rate": 6.657514913518317e-06,
      "loss": 0.416,
      "step": 6556
    },
    {
      "epoch": 0.4107173616874677,
      "grad_norm": 0.8865511537513051,
      "learning_rate": 6.656557840765322e-06,
      "loss": 0.4466,
      "step": 6557
    },
    {
      "epoch": 0.41077999968681,
      "grad_norm": 0.7673041215773336,
      "learning_rate": 6.655600699828098e-06,
      "loss": 0.3981,
      "step": 6558
    },
    {
      "epoch": 0.4108426376861523,
      "grad_norm": 0.841618258339795,
      "learning_rate": 6.654643490746042e-06,
      "loss": 0.4225,
      "step": 6559
    },
    {
      "epoch": 0.4109052756854946,
      "grad_norm": 0.8082093875197354,
      "learning_rate": 6.653686213558553e-06,
      "loss": 0.404,
      "step": 6560
    },
    {
      "epoch": 0.4109679136848369,
      "grad_norm": 0.6878154222265779,
      "learning_rate": 6.652728868305034e-06,
      "loss": 0.4572,
      "step": 6561
    },
    {
      "epoch": 0.41103055168417924,
      "grad_norm": 0.7524293461092177,
      "learning_rate": 6.651771455024886e-06,
      "loss": 0.4613,
      "step": 6562
    },
    {
      "epoch": 0.4110931896835215,
      "grad_norm": 0.7006205098606554,
      "learning_rate": 6.650813973757517e-06,
      "loss": 0.4609,
      "step": 6563
    },
    {
      "epoch": 0.4111558276828638,
      "grad_norm": 0.8860707979626495,
      "learning_rate": 6.6498564245423425e-06,
      "loss": 0.4366,
      "step": 6564
    },
    {
      "epoch": 0.4112184656822061,
      "grad_norm": 0.6461656234672829,
      "learning_rate": 6.6488988074187684e-06,
      "loss": 0.4577,
      "step": 6565
    },
    {
      "epoch": 0.4112811036815484,
      "grad_norm": 0.8552384671709481,
      "learning_rate": 6.647941122426216e-06,
      "loss": 0.4104,
      "step": 6566
    },
    {
      "epoch": 0.4113437416808907,
      "grad_norm": 0.8788801506170293,
      "learning_rate": 6.646983369604099e-06,
      "loss": 0.3995,
      "step": 6567
    },
    {
      "epoch": 0.411406379680233,
      "grad_norm": 0.9223796023301072,
      "learning_rate": 6.646025548991839e-06,
      "loss": 0.4364,
      "step": 6568
    },
    {
      "epoch": 0.41146901767957533,
      "grad_norm": 0.8369504461531386,
      "learning_rate": 6.645067660628865e-06,
      "loss": 0.4004,
      "step": 6569
    },
    {
      "epoch": 0.41153165567891764,
      "grad_norm": 0.8617851286705729,
      "learning_rate": 6.6441097045546e-06,
      "loss": 0.4354,
      "step": 6570
    },
    {
      "epoch": 0.4115942936782599,
      "grad_norm": 0.8119213004063407,
      "learning_rate": 6.643151680808474e-06,
      "loss": 0.4233,
      "step": 6571
    },
    {
      "epoch": 0.4116569316776022,
      "grad_norm": 0.8031366574703883,
      "learning_rate": 6.642193589429919e-06,
      "loss": 0.4111,
      "step": 6572
    },
    {
      "epoch": 0.4117195696769445,
      "grad_norm": 0.8547198672799732,
      "learning_rate": 6.641235430458373e-06,
      "loss": 0.402,
      "step": 6573
    },
    {
      "epoch": 0.4117822076762868,
      "grad_norm": 0.8935350626092392,
      "learning_rate": 6.640277203933272e-06,
      "loss": 0.4544,
      "step": 6574
    },
    {
      "epoch": 0.4118448456756291,
      "grad_norm": 0.7880444270586193,
      "learning_rate": 6.639318909894054e-06,
      "loss": 0.4272,
      "step": 6575
    },
    {
      "epoch": 0.41190748367497143,
      "grad_norm": 0.7631015315762734,
      "learning_rate": 6.638360548380167e-06,
      "loss": 0.4104,
      "step": 6576
    },
    {
      "epoch": 0.41197012167431374,
      "grad_norm": 0.8970685460809124,
      "learning_rate": 6.6374021194310575e-06,
      "loss": 0.4193,
      "step": 6577
    },
    {
      "epoch": 0.41203275967365605,
      "grad_norm": 0.8457912436198679,
      "learning_rate": 6.636443623086172e-06,
      "loss": 0.4248,
      "step": 6578
    },
    {
      "epoch": 0.4120953976729983,
      "grad_norm": 0.8244665575029495,
      "learning_rate": 6.635485059384963e-06,
      "loss": 0.3987,
      "step": 6579
    },
    {
      "epoch": 0.4121580356723406,
      "grad_norm": 0.8133529424386704,
      "learning_rate": 6.634526428366884e-06,
      "loss": 0.4577,
      "step": 6580
    },
    {
      "epoch": 0.4122206736716829,
      "grad_norm": 0.8150247700192297,
      "learning_rate": 6.6335677300713955e-06,
      "loss": 0.3931,
      "step": 6581
    },
    {
      "epoch": 0.4122833116710252,
      "grad_norm": 0.7625002990195262,
      "learning_rate": 6.632608964537955e-06,
      "loss": 0.3901,
      "step": 6582
    },
    {
      "epoch": 0.41234594967036753,
      "grad_norm": 0.7962853625770262,
      "learning_rate": 6.631650131806026e-06,
      "loss": 0.389,
      "step": 6583
    },
    {
      "epoch": 0.41240858766970984,
      "grad_norm": 0.8685615183792181,
      "learning_rate": 6.630691231915076e-06,
      "loss": 0.4218,
      "step": 6584
    },
    {
      "epoch": 0.41247122566905214,
      "grad_norm": 0.7952718952433356,
      "learning_rate": 6.629732264904572e-06,
      "loss": 0.4418,
      "step": 6585
    },
    {
      "epoch": 0.41253386366839445,
      "grad_norm": 0.8933527350049224,
      "learning_rate": 6.628773230813986e-06,
      "loss": 0.4034,
      "step": 6586
    },
    {
      "epoch": 0.41259650166773676,
      "grad_norm": 0.8924796657719081,
      "learning_rate": 6.62781412968279e-06,
      "loss": 0.4407,
      "step": 6587
    },
    {
      "epoch": 0.412659139667079,
      "grad_norm": 0.8191553353397825,
      "learning_rate": 6.626854961550464e-06,
      "loss": 0.4196,
      "step": 6588
    },
    {
      "epoch": 0.4127217776664213,
      "grad_norm": 0.8264893117506691,
      "learning_rate": 6.625895726456485e-06,
      "loss": 0.4125,
      "step": 6589
    },
    {
      "epoch": 0.4127844156657636,
      "grad_norm": 0.794308814266276,
      "learning_rate": 6.624936424440336e-06,
      "loss": 0.3859,
      "step": 6590
    },
    {
      "epoch": 0.41284705366510593,
      "grad_norm": 0.8292881482783633,
      "learning_rate": 6.623977055541502e-06,
      "loss": 0.3947,
      "step": 6591
    },
    {
      "epoch": 0.41290969166444824,
      "grad_norm": 0.7490474373147915,
      "learning_rate": 6.6230176197994725e-06,
      "loss": 0.3901,
      "step": 6592
    },
    {
      "epoch": 0.41297232966379055,
      "grad_norm": 0.8424190442301781,
      "learning_rate": 6.622058117253734e-06,
      "loss": 0.4047,
      "step": 6593
    },
    {
      "epoch": 0.41303496766313286,
      "grad_norm": 0.8640903314025081,
      "learning_rate": 6.6210985479437835e-06,
      "loss": 0.4845,
      "step": 6594
    },
    {
      "epoch": 0.41309760566247516,
      "grad_norm": 0.90897581308804,
      "learning_rate": 6.620138911909115e-06,
      "loss": 0.4509,
      "step": 6595
    },
    {
      "epoch": 0.4131602436618174,
      "grad_norm": 0.8123616582345513,
      "learning_rate": 6.61917920918923e-06,
      "loss": 0.3726,
      "step": 6596
    },
    {
      "epoch": 0.4132228816611597,
      "grad_norm": 0.8050288663668661,
      "learning_rate": 6.618219439823628e-06,
      "loss": 0.3942,
      "step": 6597
    },
    {
      "epoch": 0.41328551966050203,
      "grad_norm": 0.7779574738659958,
      "learning_rate": 6.617259603851812e-06,
      "loss": 0.4384,
      "step": 6598
    },
    {
      "epoch": 0.41334815765984434,
      "grad_norm": 0.7653938812955126,
      "learning_rate": 6.616299701313291e-06,
      "loss": 0.366,
      "step": 6599
    },
    {
      "epoch": 0.41341079565918665,
      "grad_norm": 0.7681726046323181,
      "learning_rate": 6.615339732247576e-06,
      "loss": 0.3819,
      "step": 6600
    },
    {
      "epoch": 0.41347343365852895,
      "grad_norm": 0.7969905879855728,
      "learning_rate": 6.6143796966941755e-06,
      "loss": 0.4159,
      "step": 6601
    },
    {
      "epoch": 0.41353607165787126,
      "grad_norm": 0.8250250927143103,
      "learning_rate": 6.6134195946926085e-06,
      "loss": 0.3927,
      "step": 6602
    },
    {
      "epoch": 0.41359870965721357,
      "grad_norm": 0.8559945620348838,
      "learning_rate": 6.612459426282391e-06,
      "loss": 0.3737,
      "step": 6603
    },
    {
      "epoch": 0.4136613476565558,
      "grad_norm": 0.7645042730108389,
      "learning_rate": 6.611499191503044e-06,
      "loss": 0.4729,
      "step": 6604
    },
    {
      "epoch": 0.41372398565589813,
      "grad_norm": 0.8492755387730233,
      "learning_rate": 6.610538890394091e-06,
      "loss": 0.4218,
      "step": 6605
    },
    {
      "epoch": 0.41378662365524044,
      "grad_norm": 0.8386782541071327,
      "learning_rate": 6.60957852299506e-06,
      "loss": 0.415,
      "step": 6606
    },
    {
      "epoch": 0.41384926165458275,
      "grad_norm": 0.8249576256152265,
      "learning_rate": 6.608618089345476e-06,
      "loss": 0.3622,
      "step": 6607
    },
    {
      "epoch": 0.41391189965392505,
      "grad_norm": 0.6697942124009073,
      "learning_rate": 6.607657589484875e-06,
      "loss": 0.4636,
      "step": 6608
    },
    {
      "epoch": 0.41397453765326736,
      "grad_norm": 0.8533191325035213,
      "learning_rate": 6.606697023452787e-06,
      "loss": 0.4113,
      "step": 6609
    },
    {
      "epoch": 0.41403717565260967,
      "grad_norm": 0.8020486649152421,
      "learning_rate": 6.6057363912887515e-06,
      "loss": 0.4141,
      "step": 6610
    },
    {
      "epoch": 0.414099813651952,
      "grad_norm": 0.8425326108118011,
      "learning_rate": 6.604775693032311e-06,
      "loss": 0.4112,
      "step": 6611
    },
    {
      "epoch": 0.4141624516512943,
      "grad_norm": 0.8898310145788394,
      "learning_rate": 6.603814928723002e-06,
      "loss": 0.4106,
      "step": 6612
    },
    {
      "epoch": 0.41422508965063654,
      "grad_norm": 0.8163350569023868,
      "learning_rate": 6.602854098400374e-06,
      "loss": 0.3982,
      "step": 6613
    },
    {
      "epoch": 0.41428772764997884,
      "grad_norm": 0.8041169083321699,
      "learning_rate": 6.601893202103972e-06,
      "loss": 0.3853,
      "step": 6614
    },
    {
      "epoch": 0.41435036564932115,
      "grad_norm": 0.8460207807637927,
      "learning_rate": 6.600932239873352e-06,
      "loss": 0.3807,
      "step": 6615
    },
    {
      "epoch": 0.41441300364866346,
      "grad_norm": 0.7547539334577168,
      "learning_rate": 6.599971211748062e-06,
      "loss": 0.3987,
      "step": 6616
    },
    {
      "epoch": 0.41447564164800577,
      "grad_norm": 0.7909763860448319,
      "learning_rate": 6.599010117767662e-06,
      "loss": 0.3942,
      "step": 6617
    },
    {
      "epoch": 0.4145382796473481,
      "grad_norm": 0.841642276368199,
      "learning_rate": 6.598048957971706e-06,
      "loss": 0.4276,
      "step": 6618
    },
    {
      "epoch": 0.4146009176466904,
      "grad_norm": 0.8112070290745198,
      "learning_rate": 6.597087732399757e-06,
      "loss": 0.4063,
      "step": 6619
    },
    {
      "epoch": 0.4146635556460327,
      "grad_norm": 0.7967074351557722,
      "learning_rate": 6.596126441091384e-06,
      "loss": 0.4174,
      "step": 6620
    },
    {
      "epoch": 0.41472619364537494,
      "grad_norm": 0.8073909480459928,
      "learning_rate": 6.595165084086149e-06,
      "loss": 0.3933,
      "step": 6621
    },
    {
      "epoch": 0.41478883164471725,
      "grad_norm": 0.8377080170116291,
      "learning_rate": 6.594203661423622e-06,
      "loss": 0.4,
      "step": 6622
    },
    {
      "epoch": 0.41485146964405956,
      "grad_norm": 0.8319657288467167,
      "learning_rate": 6.5932421731433775e-06,
      "loss": 0.4004,
      "step": 6623
    },
    {
      "epoch": 0.41491410764340186,
      "grad_norm": 0.7925344407713694,
      "learning_rate": 6.59228061928499e-06,
      "loss": 0.3886,
      "step": 6624
    },
    {
      "epoch": 0.41497674564274417,
      "grad_norm": 0.8289898349898772,
      "learning_rate": 6.591318999888037e-06,
      "loss": 0.4161,
      "step": 6625
    },
    {
      "epoch": 0.4150393836420865,
      "grad_norm": 0.831438737378566,
      "learning_rate": 6.590357314992097e-06,
      "loss": 0.3819,
      "step": 6626
    },
    {
      "epoch": 0.4151020216414288,
      "grad_norm": 0.8169475167688682,
      "learning_rate": 6.589395564636755e-06,
      "loss": 0.4124,
      "step": 6627
    },
    {
      "epoch": 0.4151646596407711,
      "grad_norm": 0.7846008027523949,
      "learning_rate": 6.588433748861597e-06,
      "loss": 0.3947,
      "step": 6628
    },
    {
      "epoch": 0.4152272976401134,
      "grad_norm": 0.7694977384117211,
      "learning_rate": 6.587471867706211e-06,
      "loss": 0.4098,
      "step": 6629
    },
    {
      "epoch": 0.41528993563945565,
      "grad_norm": 0.7789298920914114,
      "learning_rate": 6.586509921210189e-06,
      "loss": 0.3768,
      "step": 6630
    },
    {
      "epoch": 0.41535257363879796,
      "grad_norm": 0.8277993725929383,
      "learning_rate": 6.5855479094131225e-06,
      "loss": 0.3829,
      "step": 6631
    },
    {
      "epoch": 0.41541521163814027,
      "grad_norm": 0.786786777251904,
      "learning_rate": 6.584585832354611e-06,
      "loss": 0.4408,
      "step": 6632
    },
    {
      "epoch": 0.4154778496374826,
      "grad_norm": 0.8199623921950224,
      "learning_rate": 6.583623690074252e-06,
      "loss": 0.4443,
      "step": 6633
    },
    {
      "epoch": 0.4155404876368249,
      "grad_norm": 0.8535936684993918,
      "learning_rate": 6.582661482611649e-06,
      "loss": 0.3911,
      "step": 6634
    },
    {
      "epoch": 0.4156031256361672,
      "grad_norm": 0.75481011012588,
      "learning_rate": 6.581699210006405e-06,
      "loss": 0.3767,
      "step": 6635
    },
    {
      "epoch": 0.4156657636355095,
      "grad_norm": 0.903726756453755,
      "learning_rate": 6.580736872298127e-06,
      "loss": 0.4703,
      "step": 6636
    },
    {
      "epoch": 0.4157284016348518,
      "grad_norm": 0.8331850422746885,
      "learning_rate": 6.5797744695264264e-06,
      "loss": 0.3937,
      "step": 6637
    },
    {
      "epoch": 0.41579103963419406,
      "grad_norm": 0.8769583700053202,
      "learning_rate": 6.578812001730917e-06,
      "loss": 0.4115,
      "step": 6638
    },
    {
      "epoch": 0.41585367763353637,
      "grad_norm": 0.7409948385985751,
      "learning_rate": 6.5778494689512115e-06,
      "loss": 0.3942,
      "step": 6639
    },
    {
      "epoch": 0.4159163156328787,
      "grad_norm": 0.8128914529760798,
      "learning_rate": 6.576886871226929e-06,
      "loss": 0.3895,
      "step": 6640
    },
    {
      "epoch": 0.415978953632221,
      "grad_norm": 0.7872484146478291,
      "learning_rate": 6.575924208597688e-06,
      "loss": 0.4065,
      "step": 6641
    },
    {
      "epoch": 0.4160415916315633,
      "grad_norm": 0.8004994832461592,
      "learning_rate": 6.574961481103117e-06,
      "loss": 0.4065,
      "step": 6642
    },
    {
      "epoch": 0.4161042296309056,
      "grad_norm": 0.9802288315602005,
      "learning_rate": 6.573998688782839e-06,
      "loss": 0.4645,
      "step": 6643
    },
    {
      "epoch": 0.4161668676302479,
      "grad_norm": 0.7961281292657971,
      "learning_rate": 6.573035831676481e-06,
      "loss": 0.373,
      "step": 6644
    },
    {
      "epoch": 0.4162295056295902,
      "grad_norm": 0.8034502965200779,
      "learning_rate": 6.572072909823677e-06,
      "loss": 0.3983,
      "step": 6645
    },
    {
      "epoch": 0.41629214362893247,
      "grad_norm": 0.812063468063844,
      "learning_rate": 6.571109923264061e-06,
      "loss": 0.4102,
      "step": 6646
    },
    {
      "epoch": 0.4163547816282748,
      "grad_norm": 0.8686097070004032,
      "learning_rate": 6.570146872037268e-06,
      "loss": 0.387,
      "step": 6647
    },
    {
      "epoch": 0.4164174196276171,
      "grad_norm": 0.8465460884104319,
      "learning_rate": 6.5691837561829384e-06,
      "loss": 0.4213,
      "step": 6648
    },
    {
      "epoch": 0.4164800576269594,
      "grad_norm": 0.7565245396169756,
      "learning_rate": 6.568220575740716e-06,
      "loss": 0.3589,
      "step": 6649
    },
    {
      "epoch": 0.4165426956263017,
      "grad_norm": 0.8298868127230542,
      "learning_rate": 6.567257330750241e-06,
      "loss": 0.4108,
      "step": 6650
    },
    {
      "epoch": 0.416605333625644,
      "grad_norm": 0.7670218058891808,
      "learning_rate": 6.566294021251165e-06,
      "loss": 0.3801,
      "step": 6651
    },
    {
      "epoch": 0.4166679716249863,
      "grad_norm": 0.8209806823025345,
      "learning_rate": 6.5653306472831355e-06,
      "loss": 0.4394,
      "step": 6652
    },
    {
      "epoch": 0.4167306096243286,
      "grad_norm": 0.9484431125121513,
      "learning_rate": 6.5643672088858066e-06,
      "loss": 0.4203,
      "step": 6653
    },
    {
      "epoch": 0.4167932476236709,
      "grad_norm": 0.7934747192691989,
      "learning_rate": 6.563403706098833e-06,
      "loss": 0.3927,
      "step": 6654
    },
    {
      "epoch": 0.4168558856230132,
      "grad_norm": 0.8035398229323963,
      "learning_rate": 6.562440138961872e-06,
      "loss": 0.4231,
      "step": 6655
    },
    {
      "epoch": 0.4169185236223555,
      "grad_norm": 0.7212033977016938,
      "learning_rate": 6.561476507514586e-06,
      "loss": 0.3819,
      "step": 6656
    },
    {
      "epoch": 0.4169811616216978,
      "grad_norm": 0.7642487038631165,
      "learning_rate": 6.5605128117966354e-06,
      "loss": 0.4855,
      "step": 6657
    },
    {
      "epoch": 0.4170437996210401,
      "grad_norm": 0.776782263392631,
      "learning_rate": 6.559549051847689e-06,
      "loss": 0.3791,
      "step": 6658
    },
    {
      "epoch": 0.4171064376203824,
      "grad_norm": 0.8473543617498092,
      "learning_rate": 6.558585227707413e-06,
      "loss": 0.4763,
      "step": 6659
    },
    {
      "epoch": 0.4171690756197247,
      "grad_norm": 0.7822853693412366,
      "learning_rate": 6.557621339415482e-06,
      "loss": 0.3751,
      "step": 6660
    },
    {
      "epoch": 0.417231713619067,
      "grad_norm": 0.7685460032527632,
      "learning_rate": 6.556657387011565e-06,
      "loss": 0.3718,
      "step": 6661
    },
    {
      "epoch": 0.41729435161840933,
      "grad_norm": 0.8436628933018139,
      "learning_rate": 6.5556933705353435e-06,
      "loss": 0.4282,
      "step": 6662
    },
    {
      "epoch": 0.4173569896177516,
      "grad_norm": 0.8680033598140477,
      "learning_rate": 6.554729290026494e-06,
      "loss": 0.3808,
      "step": 6663
    },
    {
      "epoch": 0.4174196276170939,
      "grad_norm": 0.8531717126101542,
      "learning_rate": 6.553765145524696e-06,
      "loss": 0.4343,
      "step": 6664
    },
    {
      "epoch": 0.4174822656164362,
      "grad_norm": 0.8470400880007825,
      "learning_rate": 6.5528009370696385e-06,
      "loss": 0.3718,
      "step": 6665
    },
    {
      "epoch": 0.4175449036157785,
      "grad_norm": 0.8169941893059457,
      "learning_rate": 6.551836664701004e-06,
      "loss": 0.4082,
      "step": 6666
    },
    {
      "epoch": 0.4176075416151208,
      "grad_norm": 0.8205418571643569,
      "learning_rate": 6.550872328458486e-06,
      "loss": 0.4268,
      "step": 6667
    },
    {
      "epoch": 0.4176701796144631,
      "grad_norm": 0.842641354955678,
      "learning_rate": 6.549907928381774e-06,
      "loss": 0.4168,
      "step": 6668
    },
    {
      "epoch": 0.41773281761380543,
      "grad_norm": 0.7417743896707018,
      "learning_rate": 6.548943464510566e-06,
      "loss": 0.3666,
      "step": 6669
    },
    {
      "epoch": 0.41779545561314774,
      "grad_norm": 0.7857016178387277,
      "learning_rate": 6.547978936884556e-06,
      "loss": 0.3913,
      "step": 6670
    },
    {
      "epoch": 0.41785809361249004,
      "grad_norm": 0.8807542299642095,
      "learning_rate": 6.547014345543445e-06,
      "loss": 0.3978,
      "step": 6671
    },
    {
      "epoch": 0.4179207316118323,
      "grad_norm": 0.8494029855658913,
      "learning_rate": 6.5460496905269385e-06,
      "loss": 0.4159,
      "step": 6672
    },
    {
      "epoch": 0.4179833696111746,
      "grad_norm": 0.7846088215018209,
      "learning_rate": 6.545084971874738e-06,
      "loss": 0.4163,
      "step": 6673
    },
    {
      "epoch": 0.4180460076105169,
      "grad_norm": 0.8151891317545882,
      "learning_rate": 6.544120189626553e-06,
      "loss": 0.4131,
      "step": 6674
    },
    {
      "epoch": 0.4181086456098592,
      "grad_norm": 0.7829723554522454,
      "learning_rate": 6.543155343822096e-06,
      "loss": 0.3807,
      "step": 6675
    },
    {
      "epoch": 0.4181712836092015,
      "grad_norm": 0.6678582702122686,
      "learning_rate": 6.542190434501078e-06,
      "loss": 0.4524,
      "step": 6676
    },
    {
      "epoch": 0.41823392160854383,
      "grad_norm": 0.8607596534393027,
      "learning_rate": 6.541225461703215e-06,
      "loss": 0.4164,
      "step": 6677
    },
    {
      "epoch": 0.41829655960788614,
      "grad_norm": 0.8646741176309967,
      "learning_rate": 6.540260425468225e-06,
      "loss": 0.4263,
      "step": 6678
    },
    {
      "epoch": 0.41835919760722845,
      "grad_norm": 0.8423268437141658,
      "learning_rate": 6.539295325835833e-06,
      "loss": 0.4114,
      "step": 6679
    },
    {
      "epoch": 0.4184218356065707,
      "grad_norm": 0.8285069021239342,
      "learning_rate": 6.538330162845757e-06,
      "loss": 0.4032,
      "step": 6680
    },
    {
      "epoch": 0.418484473605913,
      "grad_norm": 0.8313041535102895,
      "learning_rate": 6.537364936537728e-06,
      "loss": 0.4342,
      "step": 6681
    },
    {
      "epoch": 0.4185471116052553,
      "grad_norm": 0.8577709441235413,
      "learning_rate": 6.536399646951471e-06,
      "loss": 0.4571,
      "step": 6682
    },
    {
      "epoch": 0.4186097496045976,
      "grad_norm": 0.86394076135085,
      "learning_rate": 6.535434294126719e-06,
      "loss": 0.4614,
      "step": 6683
    },
    {
      "epoch": 0.41867238760393993,
      "grad_norm": 0.8329853212961016,
      "learning_rate": 6.534468878103208e-06,
      "loss": 0.4441,
      "step": 6684
    },
    {
      "epoch": 0.41873502560328224,
      "grad_norm": 0.8950727022668018,
      "learning_rate": 6.533503398920673e-06,
      "loss": 0.3714,
      "step": 6685
    },
    {
      "epoch": 0.41879766360262455,
      "grad_norm": 0.7961478039852888,
      "learning_rate": 6.532537856618854e-06,
      "loss": 0.3722,
      "step": 6686
    },
    {
      "epoch": 0.41886030160196686,
      "grad_norm": 0.8242524157728592,
      "learning_rate": 6.53157225123749e-06,
      "loss": 0.4439,
      "step": 6687
    },
    {
      "epoch": 0.4189229396013091,
      "grad_norm": 0.8104490481211577,
      "learning_rate": 6.53060658281633e-06,
      "loss": 0.4134,
      "step": 6688
    },
    {
      "epoch": 0.4189855776006514,
      "grad_norm": 0.7752568004721767,
      "learning_rate": 6.529640851395119e-06,
      "loss": 0.4404,
      "step": 6689
    },
    {
      "epoch": 0.4190482155999937,
      "grad_norm": 0.7615444382149605,
      "learning_rate": 6.528675057013607e-06,
      "loss": 0.4309,
      "step": 6690
    },
    {
      "epoch": 0.41911085359933603,
      "grad_norm": 0.8250653129253606,
      "learning_rate": 6.527709199711543e-06,
      "loss": 0.3957,
      "step": 6691
    },
    {
      "epoch": 0.41917349159867834,
      "grad_norm": 0.7962681638429224,
      "learning_rate": 6.526743279528686e-06,
      "loss": 0.4268,
      "step": 6692
    },
    {
      "epoch": 0.41923612959802065,
      "grad_norm": 0.8499405400908971,
      "learning_rate": 6.525777296504794e-06,
      "loss": 0.4094,
      "step": 6693
    },
    {
      "epoch": 0.41929876759736295,
      "grad_norm": 0.6319741717404288,
      "learning_rate": 6.524811250679626e-06,
      "loss": 0.478,
      "step": 6694
    },
    {
      "epoch": 0.41936140559670526,
      "grad_norm": 0.8705243752106595,
      "learning_rate": 6.5238451420929404e-06,
      "loss": 0.3978,
      "step": 6695
    },
    {
      "epoch": 0.41942404359604757,
      "grad_norm": 0.832136802771255,
      "learning_rate": 6.522878970784507e-06,
      "loss": 0.4085,
      "step": 6696
    },
    {
      "epoch": 0.4194866815953898,
      "grad_norm": 0.8287207582792625,
      "learning_rate": 6.521912736794094e-06,
      "loss": 0.412,
      "step": 6697
    },
    {
      "epoch": 0.41954931959473213,
      "grad_norm": 0.8468918388100548,
      "learning_rate": 6.5209464401614695e-06,
      "loss": 0.4,
      "step": 6698
    },
    {
      "epoch": 0.41961195759407444,
      "grad_norm": 0.7313308166560301,
      "learning_rate": 6.519980080926407e-06,
      "loss": 0.376,
      "step": 6699
    },
    {
      "epoch": 0.41967459559341674,
      "grad_norm": 0.8381248287072699,
      "learning_rate": 6.519013659128683e-06,
      "loss": 0.4312,
      "step": 6700
    },
    {
      "epoch": 0.41973723359275905,
      "grad_norm": 0.805536316994159,
      "learning_rate": 6.518047174808073e-06,
      "loss": 0.3854,
      "step": 6701
    },
    {
      "epoch": 0.41979987159210136,
      "grad_norm": 0.8935739019856328,
      "learning_rate": 6.517080628004362e-06,
      "loss": 0.4558,
      "step": 6702
    },
    {
      "epoch": 0.41986250959144367,
      "grad_norm": 0.7694798298949629,
      "learning_rate": 6.5161140187573305e-06,
      "loss": 0.3621,
      "step": 6703
    },
    {
      "epoch": 0.419925147590786,
      "grad_norm": 0.8378758362306895,
      "learning_rate": 6.515147347106763e-06,
      "loss": 0.3926,
      "step": 6704
    },
    {
      "epoch": 0.4199877855901282,
      "grad_norm": 0.7983512634660449,
      "learning_rate": 6.514180613092449e-06,
      "loss": 0.4289,
      "step": 6705
    },
    {
      "epoch": 0.42005042358947053,
      "grad_norm": 0.7806819271285,
      "learning_rate": 6.513213816754181e-06,
      "loss": 0.4008,
      "step": 6706
    },
    {
      "epoch": 0.42011306158881284,
      "grad_norm": 0.7791492805029906,
      "learning_rate": 6.512246958131753e-06,
      "loss": 0.3929,
      "step": 6707
    },
    {
      "epoch": 0.42017569958815515,
      "grad_norm": 0.8503834433618165,
      "learning_rate": 6.511280037264956e-06,
      "loss": 0.416,
      "step": 6708
    },
    {
      "epoch": 0.42023833758749746,
      "grad_norm": 0.8048085688481647,
      "learning_rate": 6.510313054193596e-06,
      "loss": 0.3941,
      "step": 6709
    },
    {
      "epoch": 0.42030097558683976,
      "grad_norm": 0.8509911466094854,
      "learning_rate": 6.509346008957467e-06,
      "loss": 0.4275,
      "step": 6710
    },
    {
      "epoch": 0.42036361358618207,
      "grad_norm": 0.9019502745418838,
      "learning_rate": 6.508378901596378e-06,
      "loss": 0.4276,
      "step": 6711
    },
    {
      "epoch": 0.4204262515855244,
      "grad_norm": 0.7857641658867366,
      "learning_rate": 6.507411732150133e-06,
      "loss": 0.3514,
      "step": 6712
    },
    {
      "epoch": 0.42048888958486663,
      "grad_norm": 0.8605203153340836,
      "learning_rate": 6.506444500658542e-06,
      "loss": 0.4152,
      "step": 6713
    },
    {
      "epoch": 0.42055152758420894,
      "grad_norm": 0.7646712339608045,
      "learning_rate": 6.505477207161415e-06,
      "loss": 0.434,
      "step": 6714
    },
    {
      "epoch": 0.42061416558355125,
      "grad_norm": 0.7732738074539308,
      "learning_rate": 6.504509851698566e-06,
      "loss": 0.4161,
      "step": 6715
    },
    {
      "epoch": 0.42067680358289355,
      "grad_norm": 0.798487593092166,
      "learning_rate": 6.503542434309814e-06,
      "loss": 0.3762,
      "step": 6716
    },
    {
      "epoch": 0.42073944158223586,
      "grad_norm": 0.834619161787785,
      "learning_rate": 6.502574955034975e-06,
      "loss": 0.4253,
      "step": 6717
    },
    {
      "epoch": 0.42080207958157817,
      "grad_norm": 0.796233418700879,
      "learning_rate": 6.5016074139138726e-06,
      "loss": 0.3945,
      "step": 6718
    },
    {
      "epoch": 0.4208647175809205,
      "grad_norm": 0.8437648229532165,
      "learning_rate": 6.50063981098633e-06,
      "loss": 0.397,
      "step": 6719
    },
    {
      "epoch": 0.4209273555802628,
      "grad_norm": 0.807248195735066,
      "learning_rate": 6.499672146292174e-06,
      "loss": 0.4235,
      "step": 6720
    },
    {
      "epoch": 0.4209899935796051,
      "grad_norm": 0.7815773740912738,
      "learning_rate": 6.4987044198712346e-06,
      "loss": 0.378,
      "step": 6721
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 0.8754172754144552,
      "learning_rate": 6.497736631763343e-06,
      "loss": 0.4169,
      "step": 6722
    },
    {
      "epoch": 0.42111526957828965,
      "grad_norm": 0.8766831604721429,
      "learning_rate": 6.496768782008334e-06,
      "loss": 0.4111,
      "step": 6723
    },
    {
      "epoch": 0.42117790757763196,
      "grad_norm": 0.9266575135224986,
      "learning_rate": 6.495800870646043e-06,
      "loss": 0.4308,
      "step": 6724
    },
    {
      "epoch": 0.42124054557697427,
      "grad_norm": 0.7853433883646528,
      "learning_rate": 6.4948328977163115e-06,
      "loss": 0.4612,
      "step": 6725
    },
    {
      "epoch": 0.4213031835763166,
      "grad_norm": 0.7837661867196688,
      "learning_rate": 6.493864863258981e-06,
      "loss": 0.4023,
      "step": 6726
    },
    {
      "epoch": 0.4213658215756589,
      "grad_norm": 0.8405947761523033,
      "learning_rate": 6.492896767313894e-06,
      "loss": 0.4093,
      "step": 6727
    },
    {
      "epoch": 0.4214284595750012,
      "grad_norm": 0.8606963924303807,
      "learning_rate": 6.491928609920899e-06,
      "loss": 0.4308,
      "step": 6728
    },
    {
      "epoch": 0.4214910975743435,
      "grad_norm": 0.7973283467571539,
      "learning_rate": 6.490960391119847e-06,
      "loss": 0.421,
      "step": 6729
    },
    {
      "epoch": 0.42155373557368575,
      "grad_norm": 0.8603609378824779,
      "learning_rate": 6.489992110950587e-06,
      "loss": 0.3904,
      "step": 6730
    },
    {
      "epoch": 0.42161637357302806,
      "grad_norm": 0.8617448009170202,
      "learning_rate": 6.4890237694529765e-06,
      "loss": 0.4309,
      "step": 6731
    },
    {
      "epoch": 0.42167901157237037,
      "grad_norm": 0.8172823598643119,
      "learning_rate": 6.4880553666668714e-06,
      "loss": 0.4211,
      "step": 6732
    },
    {
      "epoch": 0.4217416495717127,
      "grad_norm": 0.8603401171511394,
      "learning_rate": 6.487086902632129e-06,
      "loss": 0.4012,
      "step": 6733
    },
    {
      "epoch": 0.421804287571055,
      "grad_norm": 0.7830637213457762,
      "learning_rate": 6.486118377388615e-06,
      "loss": 0.3513,
      "step": 6734
    },
    {
      "epoch": 0.4218669255703973,
      "grad_norm": 0.7795247617920754,
      "learning_rate": 6.485149790976194e-06,
      "loss": 0.3536,
      "step": 6735
    },
    {
      "epoch": 0.4219295635697396,
      "grad_norm": 0.763065264628062,
      "learning_rate": 6.484181143434732e-06,
      "loss": 0.367,
      "step": 6736
    },
    {
      "epoch": 0.4219922015690819,
      "grad_norm": 0.6296739864944059,
      "learning_rate": 6.483212434804096e-06,
      "loss": 0.4631,
      "step": 6737
    },
    {
      "epoch": 0.4220548395684242,
      "grad_norm": 0.8466829214514708,
      "learning_rate": 6.482243665124161e-06,
      "loss": 0.4318,
      "step": 6738
    },
    {
      "epoch": 0.42211747756776646,
      "grad_norm": 0.800798314651756,
      "learning_rate": 6.481274834434804e-06,
      "loss": 0.3696,
      "step": 6739
    },
    {
      "epoch": 0.42218011556710877,
      "grad_norm": 0.8060366512057189,
      "learning_rate": 6.480305942775899e-06,
      "loss": 0.3947,
      "step": 6740
    },
    {
      "epoch": 0.4222427535664511,
      "grad_norm": 0.8680048424354749,
      "learning_rate": 6.479336990187329e-06,
      "loss": 0.4436,
      "step": 6741
    },
    {
      "epoch": 0.4223053915657934,
      "grad_norm": 0.8558391772047974,
      "learning_rate": 6.4783679767089695e-06,
      "loss": 0.428,
      "step": 6742
    },
    {
      "epoch": 0.4223680295651357,
      "grad_norm": 0.9385171035296522,
      "learning_rate": 6.477398902380711e-06,
      "loss": 0.4058,
      "step": 6743
    },
    {
      "epoch": 0.422430667564478,
      "grad_norm": 0.8254797332392496,
      "learning_rate": 6.47642976724244e-06,
      "loss": 0.3722,
      "step": 6744
    },
    {
      "epoch": 0.4224933055638203,
      "grad_norm": 0.8731516323984473,
      "learning_rate": 6.475460571334047e-06,
      "loss": 0.3687,
      "step": 6745
    },
    {
      "epoch": 0.4225559435631626,
      "grad_norm": 0.8107747696149792,
      "learning_rate": 6.474491314695422e-06,
      "loss": 0.4059,
      "step": 6746
    },
    {
      "epoch": 0.42261858156250487,
      "grad_norm": 0.7231021466430687,
      "learning_rate": 6.473521997366459e-06,
      "loss": 0.4869,
      "step": 6747
    },
    {
      "epoch": 0.4226812195618472,
      "grad_norm": 0.8297975642858885,
      "learning_rate": 6.472552619387059e-06,
      "loss": 0.4022,
      "step": 6748
    },
    {
      "epoch": 0.4227438575611895,
      "grad_norm": 0.7835069251809557,
      "learning_rate": 6.471583180797121e-06,
      "loss": 0.3678,
      "step": 6749
    },
    {
      "epoch": 0.4228064955605318,
      "grad_norm": 0.8374668119651049,
      "learning_rate": 6.470613681636544e-06,
      "loss": 0.3818,
      "step": 6750
    },
    {
      "epoch": 0.4228691335598741,
      "grad_norm": 0.8982962035171322,
      "learning_rate": 6.469644121945235e-06,
      "loss": 0.4542,
      "step": 6751
    },
    {
      "epoch": 0.4229317715592164,
      "grad_norm": 0.8522534909264561,
      "learning_rate": 6.4686745017631014e-06,
      "loss": 0.4203,
      "step": 6752
    },
    {
      "epoch": 0.4229944095585587,
      "grad_norm": 0.8347608017265021,
      "learning_rate": 6.467704821130054e-06,
      "loss": 0.4105,
      "step": 6753
    },
    {
      "epoch": 0.423057047557901,
      "grad_norm": 0.8457171473115924,
      "learning_rate": 6.466735080086004e-06,
      "loss": 0.42,
      "step": 6754
    },
    {
      "epoch": 0.4231196855572433,
      "grad_norm": 0.8106574648533246,
      "learning_rate": 6.465765278670863e-06,
      "loss": 0.3758,
      "step": 6755
    },
    {
      "epoch": 0.4231823235565856,
      "grad_norm": 0.8757338958232477,
      "learning_rate": 6.464795416924553e-06,
      "loss": 0.4504,
      "step": 6756
    },
    {
      "epoch": 0.4232449615559279,
      "grad_norm": 0.843722748386492,
      "learning_rate": 6.463825494886992e-06,
      "loss": 0.4287,
      "step": 6757
    },
    {
      "epoch": 0.4233075995552702,
      "grad_norm": 0.838555189905878,
      "learning_rate": 6.462855512598102e-06,
      "loss": 0.4453,
      "step": 6758
    },
    {
      "epoch": 0.4233702375546125,
      "grad_norm": 0.8078690938975001,
      "learning_rate": 6.461885470097806e-06,
      "loss": 0.3805,
      "step": 6759
    },
    {
      "epoch": 0.4234328755539548,
      "grad_norm": 0.7912267406909008,
      "learning_rate": 6.460915367426034e-06,
      "loss": 0.3726,
      "step": 6760
    },
    {
      "epoch": 0.4234955135532971,
      "grad_norm": 0.8163401788937114,
      "learning_rate": 6.459945204622714e-06,
      "loss": 0.4325,
      "step": 6761
    },
    {
      "epoch": 0.42355815155263943,
      "grad_norm": 0.8017982749291754,
      "learning_rate": 6.458974981727779e-06,
      "loss": 0.4052,
      "step": 6762
    },
    {
      "epoch": 0.42362078955198174,
      "grad_norm": 0.8487116970709371,
      "learning_rate": 6.4580046987811615e-06,
      "loss": 0.4058,
      "step": 6763
    },
    {
      "epoch": 0.423683427551324,
      "grad_norm": 0.8231096543686165,
      "learning_rate": 6.457034355822802e-06,
      "loss": 0.4066,
      "step": 6764
    },
    {
      "epoch": 0.4237460655506663,
      "grad_norm": 0.7688286095399681,
      "learning_rate": 6.456063952892636e-06,
      "loss": 0.3838,
      "step": 6765
    },
    {
      "epoch": 0.4238087035500086,
      "grad_norm": 0.7890743721591518,
      "learning_rate": 6.455093490030609e-06,
      "loss": 0.4213,
      "step": 6766
    },
    {
      "epoch": 0.4238713415493509,
      "grad_norm": 0.7836332087538952,
      "learning_rate": 6.4541229672766636e-06,
      "loss": 0.3874,
      "step": 6767
    },
    {
      "epoch": 0.4239339795486932,
      "grad_norm": 0.7878090818476671,
      "learning_rate": 6.453152384670746e-06,
      "loss": 0.3887,
      "step": 6768
    },
    {
      "epoch": 0.4239966175480355,
      "grad_norm": 0.8115361757151384,
      "learning_rate": 6.452181742252808e-06,
      "loss": 0.4091,
      "step": 6769
    },
    {
      "epoch": 0.42405925554737783,
      "grad_norm": 0.8550250842418406,
      "learning_rate": 6.451211040062798e-06,
      "loss": 0.4143,
      "step": 6770
    },
    {
      "epoch": 0.42412189354672014,
      "grad_norm": 0.7435806177576935,
      "learning_rate": 6.450240278140674e-06,
      "loss": 0.3618,
      "step": 6771
    },
    {
      "epoch": 0.4241845315460624,
      "grad_norm": 0.8613752524881251,
      "learning_rate": 6.44926945652639e-06,
      "loss": 0.4216,
      "step": 6772
    },
    {
      "epoch": 0.4242471695454047,
      "grad_norm": 0.7985722185085116,
      "learning_rate": 6.448298575259907e-06,
      "loss": 0.4121,
      "step": 6773
    },
    {
      "epoch": 0.424309807544747,
      "grad_norm": 0.7390829512933778,
      "learning_rate": 6.4473276343811845e-06,
      "loss": 0.3495,
      "step": 6774
    },
    {
      "epoch": 0.4243724455440893,
      "grad_norm": 0.8317539299913498,
      "learning_rate": 6.446356633930188e-06,
      "loss": 0.4296,
      "step": 6775
    },
    {
      "epoch": 0.4244350835434316,
      "grad_norm": 0.8599628233863941,
      "learning_rate": 6.445385573946885e-06,
      "loss": 0.4168,
      "step": 6776
    },
    {
      "epoch": 0.42449772154277393,
      "grad_norm": 0.8455315724551722,
      "learning_rate": 6.444414454471244e-06,
      "loss": 0.4025,
      "step": 6777
    },
    {
      "epoch": 0.42456035954211624,
      "grad_norm": 0.8169096136367194,
      "learning_rate": 6.443443275543236e-06,
      "loss": 0.4052,
      "step": 6778
    },
    {
      "epoch": 0.42462299754145855,
      "grad_norm": 0.855111293577577,
      "learning_rate": 6.442472037202833e-06,
      "loss": 0.3614,
      "step": 6779
    },
    {
      "epoch": 0.42468563554080085,
      "grad_norm": 0.7575838337641814,
      "learning_rate": 6.441500739490014e-06,
      "loss": 0.3723,
      "step": 6780
    },
    {
      "epoch": 0.4247482735401431,
      "grad_norm": 0.879471996012059,
      "learning_rate": 6.440529382444758e-06,
      "loss": 0.4198,
      "step": 6781
    },
    {
      "epoch": 0.4248109115394854,
      "grad_norm": 0.789957786864866,
      "learning_rate": 6.439557966107044e-06,
      "loss": 0.4231,
      "step": 6782
    },
    {
      "epoch": 0.4248735495388277,
      "grad_norm": 0.7662323265558875,
      "learning_rate": 6.438586490516858e-06,
      "loss": 0.3978,
      "step": 6783
    },
    {
      "epoch": 0.42493618753817003,
      "grad_norm": 0.7935263431252171,
      "learning_rate": 6.4376149557141844e-06,
      "loss": 0.393,
      "step": 6784
    },
    {
      "epoch": 0.42499882553751234,
      "grad_norm": 0.7464332107523234,
      "learning_rate": 6.436643361739011e-06,
      "loss": 0.3976,
      "step": 6785
    },
    {
      "epoch": 0.42506146353685464,
      "grad_norm": 0.8219186916643447,
      "learning_rate": 6.435671708631332e-06,
      "loss": 0.4607,
      "step": 6786
    },
    {
      "epoch": 0.42512410153619695,
      "grad_norm": 0.8144169119790213,
      "learning_rate": 6.43469999643114e-06,
      "loss": 0.4062,
      "step": 6787
    },
    {
      "epoch": 0.42518673953553926,
      "grad_norm": 0.7349647213071364,
      "learning_rate": 6.433728225178428e-06,
      "loss": 0.489,
      "step": 6788
    },
    {
      "epoch": 0.4252493775348815,
      "grad_norm": 0.8527304023171687,
      "learning_rate": 6.432756394913195e-06,
      "loss": 0.4042,
      "step": 6789
    },
    {
      "epoch": 0.4253120155342238,
      "grad_norm": 0.8085449331150052,
      "learning_rate": 6.431784505675446e-06,
      "loss": 0.4047,
      "step": 6790
    },
    {
      "epoch": 0.4253746535335661,
      "grad_norm": 0.7527586308085201,
      "learning_rate": 6.43081255750518e-06,
      "loss": 0.3959,
      "step": 6791
    },
    {
      "epoch": 0.42543729153290843,
      "grad_norm": 0.8160378935602867,
      "learning_rate": 6.429840550442403e-06,
      "loss": 0.3798,
      "step": 6792
    },
    {
      "epoch": 0.42549992953225074,
      "grad_norm": 0.7718486788861412,
      "learning_rate": 6.428868484527123e-06,
      "loss": 0.3906,
      "step": 6793
    },
    {
      "epoch": 0.42556256753159305,
      "grad_norm": 0.8382602636118647,
      "learning_rate": 6.427896359799351e-06,
      "loss": 0.4299,
      "step": 6794
    },
    {
      "epoch": 0.42562520553093536,
      "grad_norm": 0.8044154313444061,
      "learning_rate": 6.426924176299101e-06,
      "loss": 0.3979,
      "step": 6795
    },
    {
      "epoch": 0.42568784353027767,
      "grad_norm": 0.7854621520653662,
      "learning_rate": 6.4259519340663865e-06,
      "loss": 0.4127,
      "step": 6796
    },
    {
      "epoch": 0.4257504815296199,
      "grad_norm": 0.869275173847781,
      "learning_rate": 6.424979633141225e-06,
      "loss": 0.394,
      "step": 6797
    },
    {
      "epoch": 0.4258131195289622,
      "grad_norm": 0.8183653395313533,
      "learning_rate": 6.424007273563637e-06,
      "loss": 0.4151,
      "step": 6798
    },
    {
      "epoch": 0.42587575752830453,
      "grad_norm": 0.7972980278288899,
      "learning_rate": 6.423034855373647e-06,
      "loss": 0.3739,
      "step": 6799
    },
    {
      "epoch": 0.42593839552764684,
      "grad_norm": 0.8039696908948583,
      "learning_rate": 6.422062378611278e-06,
      "loss": 0.4406,
      "step": 6800
    },
    {
      "epoch": 0.42600103352698915,
      "grad_norm": 0.7010323482420423,
      "learning_rate": 6.421089843316557e-06,
      "loss": 0.4665,
      "step": 6801
    },
    {
      "epoch": 0.42606367152633146,
      "grad_norm": 0.8207522844579148,
      "learning_rate": 6.420117249529514e-06,
      "loss": 0.4146,
      "step": 6802
    },
    {
      "epoch": 0.42612630952567376,
      "grad_norm": 0.8641371564308263,
      "learning_rate": 6.419144597290181e-06,
      "loss": 0.4091,
      "step": 6803
    },
    {
      "epoch": 0.42618894752501607,
      "grad_norm": 0.8134636550514323,
      "learning_rate": 6.418171886638595e-06,
      "loss": 0.3875,
      "step": 6804
    },
    {
      "epoch": 0.4262515855243584,
      "grad_norm": 0.8251512145729096,
      "learning_rate": 6.4171991176147895e-06,
      "loss": 0.4301,
      "step": 6805
    },
    {
      "epoch": 0.42631422352370063,
      "grad_norm": 0.815080124374866,
      "learning_rate": 6.416226290258805e-06,
      "loss": 0.3795,
      "step": 6806
    },
    {
      "epoch": 0.42637686152304294,
      "grad_norm": 0.8388019076304123,
      "learning_rate": 6.415253404610683e-06,
      "loss": 0.4487,
      "step": 6807
    },
    {
      "epoch": 0.42643949952238525,
      "grad_norm": 0.872520275656021,
      "learning_rate": 6.414280460710469e-06,
      "loss": 0.4093,
      "step": 6808
    },
    {
      "epoch": 0.42650213752172755,
      "grad_norm": 0.6604785758176509,
      "learning_rate": 6.41330745859821e-06,
      "loss": 0.4764,
      "step": 6809
    },
    {
      "epoch": 0.42656477552106986,
      "grad_norm": 0.8160738137890386,
      "learning_rate": 6.412334398313952e-06,
      "loss": 0.4083,
      "step": 6810
    },
    {
      "epoch": 0.42662741352041217,
      "grad_norm": 0.8353375350483991,
      "learning_rate": 6.411361279897747e-06,
      "loss": 0.4181,
      "step": 6811
    },
    {
      "epoch": 0.4266900515197545,
      "grad_norm": 0.8996932460682153,
      "learning_rate": 6.410388103389653e-06,
      "loss": 0.4194,
      "step": 6812
    },
    {
      "epoch": 0.4267526895190968,
      "grad_norm": 0.8203789263701763,
      "learning_rate": 6.409414868829722e-06,
      "loss": 0.3765,
      "step": 6813
    },
    {
      "epoch": 0.42681532751843904,
      "grad_norm": 0.816674296693075,
      "learning_rate": 6.408441576258012e-06,
      "loss": 0.3906,
      "step": 6814
    },
    {
      "epoch": 0.42687796551778134,
      "grad_norm": 0.8523894563760478,
      "learning_rate": 6.407468225714586e-06,
      "loss": 0.4227,
      "step": 6815
    },
    {
      "epoch": 0.42694060351712365,
      "grad_norm": 0.8175845818865566,
      "learning_rate": 6.406494817239506e-06,
      "loss": 0.3639,
      "step": 6816
    },
    {
      "epoch": 0.42700324151646596,
      "grad_norm": 0.8427684462395966,
      "learning_rate": 6.405521350872838e-06,
      "loss": 0.4144,
      "step": 6817
    },
    {
      "epoch": 0.42706587951580827,
      "grad_norm": 0.6699215677434645,
      "learning_rate": 6.404547826654652e-06,
      "loss": 0.4933,
      "step": 6818
    },
    {
      "epoch": 0.4271285175151506,
      "grad_norm": 0.7420171122088257,
      "learning_rate": 6.403574244625016e-06,
      "loss": 0.3638,
      "step": 6819
    },
    {
      "epoch": 0.4271911555144929,
      "grad_norm": 0.7990998201458179,
      "learning_rate": 6.402600604824003e-06,
      "loss": 0.3936,
      "step": 6820
    },
    {
      "epoch": 0.4272537935138352,
      "grad_norm": 0.7781053167019569,
      "learning_rate": 6.4016269072916895e-06,
      "loss": 0.3797,
      "step": 6821
    },
    {
      "epoch": 0.42731643151317744,
      "grad_norm": 0.8778116251007051,
      "learning_rate": 6.400653152068152e-06,
      "loss": 0.4591,
      "step": 6822
    },
    {
      "epoch": 0.42737906951251975,
      "grad_norm": 0.8429921511546111,
      "learning_rate": 6.3996793391934695e-06,
      "loss": 0.3936,
      "step": 6823
    },
    {
      "epoch": 0.42744170751186206,
      "grad_norm": 0.8686375561381724,
      "learning_rate": 6.398705468707728e-06,
      "loss": 0.4717,
      "step": 6824
    },
    {
      "epoch": 0.42750434551120436,
      "grad_norm": 0.8203556898543629,
      "learning_rate": 6.397731540651009e-06,
      "loss": 0.4059,
      "step": 6825
    },
    {
      "epoch": 0.42756698351054667,
      "grad_norm": 0.6893725244594692,
      "learning_rate": 6.3967575550634e-06,
      "loss": 0.4782,
      "step": 6826
    },
    {
      "epoch": 0.427629621509889,
      "grad_norm": 0.8617134697540916,
      "learning_rate": 6.395783511984991e-06,
      "loss": 0.4253,
      "step": 6827
    },
    {
      "epoch": 0.4276922595092313,
      "grad_norm": 0.7042699991482769,
      "learning_rate": 6.394809411455874e-06,
      "loss": 0.4741,
      "step": 6828
    },
    {
      "epoch": 0.4277548975085736,
      "grad_norm": 0.8124720210867789,
      "learning_rate": 6.393835253516142e-06,
      "loss": 0.4072,
      "step": 6829
    },
    {
      "epoch": 0.4278175355079159,
      "grad_norm": 0.8010614481459831,
      "learning_rate": 6.3928610382058906e-06,
      "loss": 0.3837,
      "step": 6830
    },
    {
      "epoch": 0.42788017350725815,
      "grad_norm": 0.7996460790754579,
      "learning_rate": 6.3918867655652224e-06,
      "loss": 0.4096,
      "step": 6831
    },
    {
      "epoch": 0.42794281150660046,
      "grad_norm": 0.7510575404543337,
      "learning_rate": 6.390912435634237e-06,
      "loss": 0.3774,
      "step": 6832
    },
    {
      "epoch": 0.42800544950594277,
      "grad_norm": 0.839891465272156,
      "learning_rate": 6.389938048453037e-06,
      "loss": 0.4052,
      "step": 6833
    },
    {
      "epoch": 0.4280680875052851,
      "grad_norm": 0.8043414372359128,
      "learning_rate": 6.388963604061729e-06,
      "loss": 0.3911,
      "step": 6834
    },
    {
      "epoch": 0.4281307255046274,
      "grad_norm": 0.8552553068650348,
      "learning_rate": 6.3879891025004204e-06,
      "loss": 0.4441,
      "step": 6835
    },
    {
      "epoch": 0.4281933635039697,
      "grad_norm": 0.7725017109750117,
      "learning_rate": 6.387014543809224e-06,
      "loss": 0.367,
      "step": 6836
    },
    {
      "epoch": 0.428256001503312,
      "grad_norm": 0.7396060360829241,
      "learning_rate": 6.386039928028251e-06,
      "loss": 0.4051,
      "step": 6837
    },
    {
      "epoch": 0.4283186395026543,
      "grad_norm": 0.8063519202546503,
      "learning_rate": 6.385065255197618e-06,
      "loss": 0.4174,
      "step": 6838
    },
    {
      "epoch": 0.42838127750199656,
      "grad_norm": 0.8094947426616946,
      "learning_rate": 6.38409052535744e-06,
      "loss": 0.398,
      "step": 6839
    },
    {
      "epoch": 0.42844391550133887,
      "grad_norm": 0.7984855414116593,
      "learning_rate": 6.383115738547841e-06,
      "loss": 0.3854,
      "step": 6840
    },
    {
      "epoch": 0.4285065535006812,
      "grad_norm": 0.8372834641642993,
      "learning_rate": 6.38214089480894e-06,
      "loss": 0.3832,
      "step": 6841
    },
    {
      "epoch": 0.4285691915000235,
      "grad_norm": 0.6452798753199825,
      "learning_rate": 6.3811659941808645e-06,
      "loss": 0.488,
      "step": 6842
    },
    {
      "epoch": 0.4286318294993658,
      "grad_norm": 0.821579383853831,
      "learning_rate": 6.38019103670374e-06,
      "loss": 0.4088,
      "step": 6843
    },
    {
      "epoch": 0.4286944674987081,
      "grad_norm": 0.821364010979275,
      "learning_rate": 6.379216022417695e-06,
      "loss": 0.3914,
      "step": 6844
    },
    {
      "epoch": 0.4287571054980504,
      "grad_norm": 0.8131595881035587,
      "learning_rate": 6.378240951362864e-06,
      "loss": 0.3853,
      "step": 6845
    },
    {
      "epoch": 0.4288197434973927,
      "grad_norm": 0.8261036295654737,
      "learning_rate": 6.377265823579379e-06,
      "loss": 0.471,
      "step": 6846
    },
    {
      "epoch": 0.428882381496735,
      "grad_norm": 0.8619126360178772,
      "learning_rate": 6.376290639107379e-06,
      "loss": 0.4221,
      "step": 6847
    },
    {
      "epoch": 0.4289450194960773,
      "grad_norm": 0.8220690470384581,
      "learning_rate": 6.375315397986999e-06,
      "loss": 0.402,
      "step": 6848
    },
    {
      "epoch": 0.4290076574954196,
      "grad_norm": 0.7934688647799508,
      "learning_rate": 6.37434010025838e-06,
      "loss": 0.3998,
      "step": 6849
    },
    {
      "epoch": 0.4290702954947619,
      "grad_norm": 0.6111558884699579,
      "learning_rate": 6.373364745961671e-06,
      "loss": 0.4647,
      "step": 6850
    },
    {
      "epoch": 0.4291329334941042,
      "grad_norm": 0.8017783054392705,
      "learning_rate": 6.372389335137013e-06,
      "loss": 0.4106,
      "step": 6851
    },
    {
      "epoch": 0.4291955714934465,
      "grad_norm": 0.8100912981134643,
      "learning_rate": 6.371413867824555e-06,
      "loss": 0.4496,
      "step": 6852
    },
    {
      "epoch": 0.4292582094927888,
      "grad_norm": 0.8136127842754445,
      "learning_rate": 6.3704383440644476e-06,
      "loss": 0.4267,
      "step": 6853
    },
    {
      "epoch": 0.4293208474921311,
      "grad_norm": 0.7930948597540477,
      "learning_rate": 6.369462763896843e-06,
      "loss": 0.3993,
      "step": 6854
    },
    {
      "epoch": 0.4293834854914734,
      "grad_norm": 0.7986654288650543,
      "learning_rate": 6.368487127361898e-06,
      "loss": 0.3865,
      "step": 6855
    },
    {
      "epoch": 0.4294461234908157,
      "grad_norm": 0.7725959414791697,
      "learning_rate": 6.367511434499769e-06,
      "loss": 0.3977,
      "step": 6856
    },
    {
      "epoch": 0.429508761490158,
      "grad_norm": 0.8091137140392908,
      "learning_rate": 6.366535685350615e-06,
      "loss": 0.4037,
      "step": 6857
    },
    {
      "epoch": 0.4295713994895003,
      "grad_norm": 1.0006894929353856,
      "learning_rate": 6.365559879954597e-06,
      "loss": 0.4427,
      "step": 6858
    },
    {
      "epoch": 0.4296340374888426,
      "grad_norm": 0.8034323109692486,
      "learning_rate": 6.364584018351882e-06,
      "loss": 0.4457,
      "step": 6859
    },
    {
      "epoch": 0.4296966754881849,
      "grad_norm": 0.8365934054471423,
      "learning_rate": 6.363608100582636e-06,
      "loss": 0.4023,
      "step": 6860
    },
    {
      "epoch": 0.4297593134875272,
      "grad_norm": 0.7970076080302468,
      "learning_rate": 6.362632126687027e-06,
      "loss": 0.3708,
      "step": 6861
    },
    {
      "epoch": 0.4298219514868695,
      "grad_norm": 0.8502454814481574,
      "learning_rate": 6.3616560967052265e-06,
      "loss": 0.4471,
      "step": 6862
    },
    {
      "epoch": 0.42988458948621183,
      "grad_norm": 0.6164643886013668,
      "learning_rate": 6.360680010677409e-06,
      "loss": 0.486,
      "step": 6863
    },
    {
      "epoch": 0.4299472274855541,
      "grad_norm": 0.8867837564842136,
      "learning_rate": 6.35970386864375e-06,
      "loss": 0.4629,
      "step": 6864
    },
    {
      "epoch": 0.4300098654848964,
      "grad_norm": 0.8061588079498566,
      "learning_rate": 6.358727670644428e-06,
      "loss": 0.4263,
      "step": 6865
    },
    {
      "epoch": 0.4300725034842387,
      "grad_norm": 0.8620675415569141,
      "learning_rate": 6.357751416719621e-06,
      "loss": 0.4385,
      "step": 6866
    },
    {
      "epoch": 0.430135141483581,
      "grad_norm": 0.9180294247158842,
      "learning_rate": 6.356775106909514e-06,
      "loss": 0.3896,
      "step": 6867
    },
    {
      "epoch": 0.4301977794829233,
      "grad_norm": 0.79562525910238,
      "learning_rate": 6.355798741254292e-06,
      "loss": 0.3909,
      "step": 6868
    },
    {
      "epoch": 0.4302604174822656,
      "grad_norm": 0.8242046816411843,
      "learning_rate": 6.354822319794142e-06,
      "loss": 0.3845,
      "step": 6869
    },
    {
      "epoch": 0.43032305548160793,
      "grad_norm": 0.8787025807221915,
      "learning_rate": 6.353845842569255e-06,
      "loss": 0.4509,
      "step": 6870
    },
    {
      "epoch": 0.43038569348095024,
      "grad_norm": 0.8770793165508883,
      "learning_rate": 6.35286930961982e-06,
      "loss": 0.4058,
      "step": 6871
    },
    {
      "epoch": 0.43044833148029255,
      "grad_norm": 0.6222329904513053,
      "learning_rate": 6.351892720986034e-06,
      "loss": 0.4626,
      "step": 6872
    },
    {
      "epoch": 0.4305109694796348,
      "grad_norm": 0.9067285289361099,
      "learning_rate": 6.350916076708095e-06,
      "loss": 0.4383,
      "step": 6873
    },
    {
      "epoch": 0.4305736074789771,
      "grad_norm": 0.823867226843081,
      "learning_rate": 6.3499393768261964e-06,
      "loss": 0.3873,
      "step": 6874
    },
    {
      "epoch": 0.4306362454783194,
      "grad_norm": 0.826824670249191,
      "learning_rate": 6.3489626213805435e-06,
      "loss": 0.4464,
      "step": 6875
    },
    {
      "epoch": 0.4306988834776617,
      "grad_norm": 0.7843957546580942,
      "learning_rate": 6.347985810411339e-06,
      "loss": 0.3806,
      "step": 6876
    },
    {
      "epoch": 0.43076152147700403,
      "grad_norm": 0.7983227142607546,
      "learning_rate": 6.347008943958788e-06,
      "loss": 0.4008,
      "step": 6877
    },
    {
      "epoch": 0.43082415947634634,
      "grad_norm": 0.8079203713460428,
      "learning_rate": 6.3460320220631e-06,
      "loss": 0.3831,
      "step": 6878
    },
    {
      "epoch": 0.43088679747568864,
      "grad_norm": 0.7894108824694991,
      "learning_rate": 6.345055044764482e-06,
      "loss": 0.3788,
      "step": 6879
    },
    {
      "epoch": 0.43094943547503095,
      "grad_norm": 0.807843020615278,
      "learning_rate": 6.344078012103149e-06,
      "loss": 0.412,
      "step": 6880
    },
    {
      "epoch": 0.4310120734743732,
      "grad_norm": 0.8154127489444869,
      "learning_rate": 6.343100924119316e-06,
      "loss": 0.4271,
      "step": 6881
    },
    {
      "epoch": 0.4310747114737155,
      "grad_norm": 0.8191563304214262,
      "learning_rate": 6.342123780853201e-06,
      "loss": 0.3946,
      "step": 6882
    },
    {
      "epoch": 0.4311373494730578,
      "grad_norm": 0.7899228225966077,
      "learning_rate": 6.341146582345021e-06,
      "loss": 0.4151,
      "step": 6883
    },
    {
      "epoch": 0.4311999874724001,
      "grad_norm": 0.807639521576959,
      "learning_rate": 6.340169328635e-06,
      "loss": 0.3852,
      "step": 6884
    },
    {
      "epoch": 0.43126262547174243,
      "grad_norm": 0.7762864068279623,
      "learning_rate": 6.339192019763358e-06,
      "loss": 0.3726,
      "step": 6885
    },
    {
      "epoch": 0.43132526347108474,
      "grad_norm": 0.8092873938417777,
      "learning_rate": 6.3382146557703265e-06,
      "loss": 0.4005,
      "step": 6886
    },
    {
      "epoch": 0.43138790147042705,
      "grad_norm": 0.7730750882073855,
      "learning_rate": 6.337237236696129e-06,
      "loss": 0.3919,
      "step": 6887
    },
    {
      "epoch": 0.43145053946976936,
      "grad_norm": 0.7996718953222396,
      "learning_rate": 6.336259762581e-06,
      "loss": 0.4222,
      "step": 6888
    },
    {
      "epoch": 0.43151317746911166,
      "grad_norm": 0.8210281093750281,
      "learning_rate": 6.3352822334651706e-06,
      "loss": 0.3862,
      "step": 6889
    },
    {
      "epoch": 0.4315758154684539,
      "grad_norm": 0.8488993816574733,
      "learning_rate": 6.334304649388879e-06,
      "loss": 0.4025,
      "step": 6890
    },
    {
      "epoch": 0.4316384534677962,
      "grad_norm": 0.807758424143934,
      "learning_rate": 6.333327010392359e-06,
      "loss": 0.4327,
      "step": 6891
    },
    {
      "epoch": 0.43170109146713853,
      "grad_norm": 0.8309698540736472,
      "learning_rate": 6.332349316515849e-06,
      "loss": 0.4215,
      "step": 6892
    },
    {
      "epoch": 0.43176372946648084,
      "grad_norm": 0.8897186307960483,
      "learning_rate": 6.3313715677995965e-06,
      "loss": 0.4214,
      "step": 6893
    },
    {
      "epoch": 0.43182636746582315,
      "grad_norm": 0.7766204125373786,
      "learning_rate": 6.330393764283842e-06,
      "loss": 0.402,
      "step": 6894
    },
    {
      "epoch": 0.43188900546516545,
      "grad_norm": 0.8859295441233831,
      "learning_rate": 6.329415906008835e-06,
      "loss": 0.4148,
      "step": 6895
    },
    {
      "epoch": 0.43195164346450776,
      "grad_norm": 0.8153603185752898,
      "learning_rate": 6.328437993014821e-06,
      "loss": 0.4015,
      "step": 6896
    },
    {
      "epoch": 0.43201428146385007,
      "grad_norm": 0.8862162777959827,
      "learning_rate": 6.327460025342054e-06,
      "loss": 0.4163,
      "step": 6897
    },
    {
      "epoch": 0.4320769194631923,
      "grad_norm": 0.8074215575482968,
      "learning_rate": 6.326482003030785e-06,
      "loss": 0.4448,
      "step": 6898
    },
    {
      "epoch": 0.43213955746253463,
      "grad_norm": 0.8125179713342575,
      "learning_rate": 6.3255039261212704e-06,
      "loss": 0.3622,
      "step": 6899
    },
    {
      "epoch": 0.43220219546187694,
      "grad_norm": 0.8645627364817535,
      "learning_rate": 6.3245257946537675e-06,
      "loss": 0.4287,
      "step": 6900
    },
    {
      "epoch": 0.43226483346121924,
      "grad_norm": 0.7895499991826834,
      "learning_rate": 6.323547608668539e-06,
      "loss": 0.3676,
      "step": 6901
    },
    {
      "epoch": 0.43232747146056155,
      "grad_norm": 0.8114698551545295,
      "learning_rate": 6.322569368205844e-06,
      "loss": 0.3978,
      "step": 6902
    },
    {
      "epoch": 0.43239010945990386,
      "grad_norm": 0.8612498104368931,
      "learning_rate": 6.321591073305948e-06,
      "loss": 0.4417,
      "step": 6903
    },
    {
      "epoch": 0.43245274745924617,
      "grad_norm": 0.8199571313294881,
      "learning_rate": 6.32061272400912e-06,
      "loss": 0.4072,
      "step": 6904
    },
    {
      "epoch": 0.4325153854585885,
      "grad_norm": 0.8329089235025339,
      "learning_rate": 6.3196343203556255e-06,
      "loss": 0.4355,
      "step": 6905
    },
    {
      "epoch": 0.4325780234579307,
      "grad_norm": 0.8350092123973486,
      "learning_rate": 6.3186558623857375e-06,
      "loss": 0.4214,
      "step": 6906
    },
    {
      "epoch": 0.43264066145727303,
      "grad_norm": 0.8884874175540013,
      "learning_rate": 6.317677350139731e-06,
      "loss": 0.4297,
      "step": 6907
    },
    {
      "epoch": 0.43270329945661534,
      "grad_norm": 0.8664476039054911,
      "learning_rate": 6.316698783657877e-06,
      "loss": 0.4313,
      "step": 6908
    },
    {
      "epoch": 0.43276593745595765,
      "grad_norm": 0.8171078770190862,
      "learning_rate": 6.315720162980459e-06,
      "loss": 0.4174,
      "step": 6909
    },
    {
      "epoch": 0.43282857545529996,
      "grad_norm": 0.8284317232936551,
      "learning_rate": 6.314741488147755e-06,
      "loss": 0.4311,
      "step": 6910
    },
    {
      "epoch": 0.43289121345464227,
      "grad_norm": 0.6883341999411612,
      "learning_rate": 6.313762759200047e-06,
      "loss": 0.4734,
      "step": 6911
    },
    {
      "epoch": 0.4329538514539846,
      "grad_norm": 0.6394260903554199,
      "learning_rate": 6.312783976177619e-06,
      "loss": 0.4676,
      "step": 6912
    },
    {
      "epoch": 0.4330164894533269,
      "grad_norm": 0.8467493908634026,
      "learning_rate": 6.311805139120759e-06,
      "loss": 0.409,
      "step": 6913
    },
    {
      "epoch": 0.4330791274526692,
      "grad_norm": 0.918911263881806,
      "learning_rate": 6.310826248069757e-06,
      "loss": 0.4173,
      "step": 6914
    },
    {
      "epoch": 0.43314176545201144,
      "grad_norm": 0.6608569512664294,
      "learning_rate": 6.309847303064904e-06,
      "loss": 0.4875,
      "step": 6915
    },
    {
      "epoch": 0.43320440345135375,
      "grad_norm": 0.8249450264759965,
      "learning_rate": 6.308868304146493e-06,
      "loss": 0.4188,
      "step": 6916
    },
    {
      "epoch": 0.43326704145069606,
      "grad_norm": 0.6296622598407157,
      "learning_rate": 6.307889251354817e-06,
      "loss": 0.4466,
      "step": 6917
    },
    {
      "epoch": 0.43332967945003836,
      "grad_norm": 0.7843353322775819,
      "learning_rate": 6.306910144730178e-06,
      "loss": 0.4363,
      "step": 6918
    },
    {
      "epoch": 0.43339231744938067,
      "grad_norm": 0.8732512478339876,
      "learning_rate": 6.305930984312876e-06,
      "loss": 0.4344,
      "step": 6919
    },
    {
      "epoch": 0.433454955448723,
      "grad_norm": 0.8299659476816825,
      "learning_rate": 6.304951770143212e-06,
      "loss": 0.4511,
      "step": 6920
    },
    {
      "epoch": 0.4335175934480653,
      "grad_norm": 0.8547374292388198,
      "learning_rate": 6.30397250226149e-06,
      "loss": 0.4332,
      "step": 6921
    },
    {
      "epoch": 0.4335802314474076,
      "grad_norm": 0.8832655479612412,
      "learning_rate": 6.302993180708017e-06,
      "loss": 0.4182,
      "step": 6922
    },
    {
      "epoch": 0.43364286944674985,
      "grad_norm": 0.8426994845835769,
      "learning_rate": 6.302013805523104e-06,
      "loss": 0.3841,
      "step": 6923
    },
    {
      "epoch": 0.43370550744609215,
      "grad_norm": 0.6691197317074452,
      "learning_rate": 6.3010343767470615e-06,
      "loss": 0.4734,
      "step": 6924
    },
    {
      "epoch": 0.43376814544543446,
      "grad_norm": 0.7561611679208823,
      "learning_rate": 6.300054894420202e-06,
      "loss": 0.3934,
      "step": 6925
    },
    {
      "epoch": 0.43383078344477677,
      "grad_norm": 0.8432805015653163,
      "learning_rate": 6.29907535858284e-06,
      "loss": 0.4428,
      "step": 6926
    },
    {
      "epoch": 0.4338934214441191,
      "grad_norm": 0.7952441438238136,
      "learning_rate": 6.2980957692752985e-06,
      "loss": 0.3848,
      "step": 6927
    },
    {
      "epoch": 0.4339560594434614,
      "grad_norm": 0.7716769868661375,
      "learning_rate": 6.297116126537892e-06,
      "loss": 0.4253,
      "step": 6928
    },
    {
      "epoch": 0.4340186974428037,
      "grad_norm": 0.7986796509894714,
      "learning_rate": 6.2961364304109465e-06,
      "loss": 0.4442,
      "step": 6929
    },
    {
      "epoch": 0.434081335442146,
      "grad_norm": 0.8778382384223775,
      "learning_rate": 6.295156680934785e-06,
      "loss": 0.4374,
      "step": 6930
    },
    {
      "epoch": 0.43414397344148825,
      "grad_norm": 0.8519406376329371,
      "learning_rate": 6.294176878149732e-06,
      "loss": 0.4117,
      "step": 6931
    },
    {
      "epoch": 0.43420661144083056,
      "grad_norm": 0.7824898858398567,
      "learning_rate": 6.29319702209612e-06,
      "loss": 0.4269,
      "step": 6932
    },
    {
      "epoch": 0.43426924944017287,
      "grad_norm": 0.8432887428184046,
      "learning_rate": 6.292217112814278e-06,
      "loss": 0.4567,
      "step": 6933
    },
    {
      "epoch": 0.4343318874395152,
      "grad_norm": 0.8684506327662309,
      "learning_rate": 6.29123715034454e-06,
      "loss": 0.4006,
      "step": 6934
    },
    {
      "epoch": 0.4343945254388575,
      "grad_norm": 0.7846403166625968,
      "learning_rate": 6.29025713472724e-06,
      "loss": 0.3989,
      "step": 6935
    },
    {
      "epoch": 0.4344571634381998,
      "grad_norm": 0.8284227794559181,
      "learning_rate": 6.2892770660027195e-06,
      "loss": 0.433,
      "step": 6936
    },
    {
      "epoch": 0.4345198014375421,
      "grad_norm": 0.803561402667832,
      "learning_rate": 6.288296944211314e-06,
      "loss": 0.3867,
      "step": 6937
    },
    {
      "epoch": 0.4345824394368844,
      "grad_norm": 0.8260260787516739,
      "learning_rate": 6.287316769393368e-06,
      "loss": 0.3993,
      "step": 6938
    },
    {
      "epoch": 0.4346450774362267,
      "grad_norm": 0.7917586376613079,
      "learning_rate": 6.286336541589224e-06,
      "loss": 0.3898,
      "step": 6939
    },
    {
      "epoch": 0.43470771543556896,
      "grad_norm": 0.7938304398882511,
      "learning_rate": 6.2853562608392285e-06,
      "loss": 0.3877,
      "step": 6940
    },
    {
      "epoch": 0.43477035343491127,
      "grad_norm": 0.8058927845736918,
      "learning_rate": 6.284375927183732e-06,
      "loss": 0.4014,
      "step": 6941
    },
    {
      "epoch": 0.4348329914342536,
      "grad_norm": 0.9042809960182784,
      "learning_rate": 6.283395540663084e-06,
      "loss": 0.4275,
      "step": 6942
    },
    {
      "epoch": 0.4348956294335959,
      "grad_norm": 0.8120417589560422,
      "learning_rate": 6.282415101317637e-06,
      "loss": 0.3809,
      "step": 6943
    },
    {
      "epoch": 0.4349582674329382,
      "grad_norm": 0.8340324645112046,
      "learning_rate": 6.281434609187747e-06,
      "loss": 0.4297,
      "step": 6944
    },
    {
      "epoch": 0.4350209054322805,
      "grad_norm": 0.8020535525952452,
      "learning_rate": 6.2804540643137695e-06,
      "loss": 0.4264,
      "step": 6945
    },
    {
      "epoch": 0.4350835434316228,
      "grad_norm": 0.8501913336186879,
      "learning_rate": 6.2794734667360656e-06,
      "loss": 0.4201,
      "step": 6946
    },
    {
      "epoch": 0.4351461814309651,
      "grad_norm": 0.846692004774439,
      "learning_rate": 6.278492816494995e-06,
      "loss": 0.4368,
      "step": 6947
    },
    {
      "epoch": 0.43520881943030737,
      "grad_norm": 0.8112919494609505,
      "learning_rate": 6.277512113630926e-06,
      "loss": 0.3915,
      "step": 6948
    },
    {
      "epoch": 0.4352714574296497,
      "grad_norm": 0.7959256653546355,
      "learning_rate": 6.276531358184218e-06,
      "loss": 0.3964,
      "step": 6949
    },
    {
      "epoch": 0.435334095428992,
      "grad_norm": 0.8202473052270659,
      "learning_rate": 6.275550550195244e-06,
      "loss": 0.4123,
      "step": 6950
    },
    {
      "epoch": 0.4353967334283343,
      "grad_norm": 0.8980415437413534,
      "learning_rate": 6.274569689704373e-06,
      "loss": 0.4471,
      "step": 6951
    },
    {
      "epoch": 0.4354593714276766,
      "grad_norm": 0.8302792049004983,
      "learning_rate": 6.273588776751975e-06,
      "loss": 0.4345,
      "step": 6952
    },
    {
      "epoch": 0.4355220094270189,
      "grad_norm": 0.8365463436189223,
      "learning_rate": 6.2726078113784284e-06,
      "loss": 0.3983,
      "step": 6953
    },
    {
      "epoch": 0.4355846474263612,
      "grad_norm": 0.7340888757020818,
      "learning_rate": 6.271626793624109e-06,
      "loss": 0.3648,
      "step": 6954
    },
    {
      "epoch": 0.4356472854257035,
      "grad_norm": 0.8283953968955111,
      "learning_rate": 6.270645723529392e-06,
      "loss": 0.44,
      "step": 6955
    },
    {
      "epoch": 0.43570992342504583,
      "grad_norm": 0.8713009131873362,
      "learning_rate": 6.269664601134661e-06,
      "loss": 0.4305,
      "step": 6956
    },
    {
      "epoch": 0.4357725614243881,
      "grad_norm": 0.832871528609332,
      "learning_rate": 6.268683426480302e-06,
      "loss": 0.4386,
      "step": 6957
    },
    {
      "epoch": 0.4358351994237304,
      "grad_norm": 0.7994330509045782,
      "learning_rate": 6.267702199606697e-06,
      "loss": 0.4523,
      "step": 6958
    },
    {
      "epoch": 0.4358978374230727,
      "grad_norm": 0.7877657331228953,
      "learning_rate": 6.266720920554233e-06,
      "loss": 0.4139,
      "step": 6959
    },
    {
      "epoch": 0.435960475422415,
      "grad_norm": 0.799533192999623,
      "learning_rate": 6.265739589363301e-06,
      "loss": 0.3951,
      "step": 6960
    },
    {
      "epoch": 0.4360231134217573,
      "grad_norm": 0.825096108890505,
      "learning_rate": 6.264758206074294e-06,
      "loss": 0.4302,
      "step": 6961
    },
    {
      "epoch": 0.4360857514210996,
      "grad_norm": 0.8473080845821608,
      "learning_rate": 6.263776770727604e-06,
      "loss": 0.4238,
      "step": 6962
    },
    {
      "epoch": 0.43614838942044193,
      "grad_norm": 0.8782002921307371,
      "learning_rate": 6.2627952833636265e-06,
      "loss": 0.4138,
      "step": 6963
    },
    {
      "epoch": 0.43621102741978424,
      "grad_norm": 0.6465523935638365,
      "learning_rate": 6.2618137440227596e-06,
      "loss": 0.4434,
      "step": 6964
    },
    {
      "epoch": 0.4362736654191265,
      "grad_norm": 0.7642540240802848,
      "learning_rate": 6.2608321527454085e-06,
      "loss": 0.3892,
      "step": 6965
    },
    {
      "epoch": 0.4363363034184688,
      "grad_norm": 0.8070645748189339,
      "learning_rate": 6.25985050957197e-06,
      "loss": 0.4008,
      "step": 6966
    },
    {
      "epoch": 0.4363989414178111,
      "grad_norm": 0.8553204884324167,
      "learning_rate": 6.258868814542851e-06,
      "loss": 0.4088,
      "step": 6967
    },
    {
      "epoch": 0.4364615794171534,
      "grad_norm": 0.7567228773731739,
      "learning_rate": 6.257887067698457e-06,
      "loss": 0.3953,
      "step": 6968
    },
    {
      "epoch": 0.4365242174164957,
      "grad_norm": 0.7369655284802369,
      "learning_rate": 6.256905269079197e-06,
      "loss": 0.4071,
      "step": 6969
    },
    {
      "epoch": 0.436586855415838,
      "grad_norm": 0.8592335998322596,
      "learning_rate": 6.255923418725486e-06,
      "loss": 0.4127,
      "step": 6970
    },
    {
      "epoch": 0.43664949341518033,
      "grad_norm": 0.8159754637055912,
      "learning_rate": 6.2549415166777325e-06,
      "loss": 0.3888,
      "step": 6971
    },
    {
      "epoch": 0.43671213141452264,
      "grad_norm": 0.7041956269694497,
      "learning_rate": 6.253959562976353e-06,
      "loss": 0.4492,
      "step": 6972
    },
    {
      "epoch": 0.4367747694138649,
      "grad_norm": 0.7959781547224999,
      "learning_rate": 6.252977557661764e-06,
      "loss": 0.384,
      "step": 6973
    },
    {
      "epoch": 0.4368374074132072,
      "grad_norm": 0.8178178305160714,
      "learning_rate": 6.251995500774388e-06,
      "loss": 0.4035,
      "step": 6974
    },
    {
      "epoch": 0.4369000454125495,
      "grad_norm": 0.8002605694834637,
      "learning_rate": 6.2510133923546445e-06,
      "loss": 0.4113,
      "step": 6975
    },
    {
      "epoch": 0.4369626834118918,
      "grad_norm": 0.7840288642498756,
      "learning_rate": 6.250031232442957e-06,
      "loss": 0.4024,
      "step": 6976
    },
    {
      "epoch": 0.4370253214112341,
      "grad_norm": 0.8269678301238208,
      "learning_rate": 6.24904902107975e-06,
      "loss": 0.4068,
      "step": 6977
    },
    {
      "epoch": 0.43708795941057643,
      "grad_norm": 0.845783556342557,
      "learning_rate": 6.2480667583054564e-06,
      "loss": 0.4549,
      "step": 6978
    },
    {
      "epoch": 0.43715059740991874,
      "grad_norm": 0.7815625329079047,
      "learning_rate": 6.247084444160504e-06,
      "loss": 0.4035,
      "step": 6979
    },
    {
      "epoch": 0.43721323540926105,
      "grad_norm": 0.8086138415722025,
      "learning_rate": 6.246102078685323e-06,
      "loss": 0.4537,
      "step": 6980
    },
    {
      "epoch": 0.43727587340860335,
      "grad_norm": 0.8125912838685965,
      "learning_rate": 6.245119661920348e-06,
      "loss": 0.3804,
      "step": 6981
    },
    {
      "epoch": 0.4373385114079456,
      "grad_norm": 0.8865659442286508,
      "learning_rate": 6.244137193906018e-06,
      "loss": 0.4462,
      "step": 6982
    },
    {
      "epoch": 0.4374011494072879,
      "grad_norm": 0.8597116700804651,
      "learning_rate": 6.24315467468277e-06,
      "loss": 0.4343,
      "step": 6983
    },
    {
      "epoch": 0.4374637874066302,
      "grad_norm": 0.6679193920616371,
      "learning_rate": 6.242172104291045e-06,
      "loss": 0.4629,
      "step": 6984
    },
    {
      "epoch": 0.43752642540597253,
      "grad_norm": 0.8377059242562396,
      "learning_rate": 6.241189482771285e-06,
      "loss": 0.4053,
      "step": 6985
    },
    {
      "epoch": 0.43758906340531484,
      "grad_norm": 0.9798094281639665,
      "learning_rate": 6.2402068101639355e-06,
      "loss": 0.4088,
      "step": 6986
    },
    {
      "epoch": 0.43765170140465715,
      "grad_norm": 0.7054656418975634,
      "learning_rate": 6.239224086509444e-06,
      "loss": 0.4601,
      "step": 6987
    },
    {
      "epoch": 0.43771433940399945,
      "grad_norm": 0.6445122958381644,
      "learning_rate": 6.23824131184826e-06,
      "loss": 0.4937,
      "step": 6988
    },
    {
      "epoch": 0.43777697740334176,
      "grad_norm": 0.8420880631988714,
      "learning_rate": 6.237258486220833e-06,
      "loss": 0.4096,
      "step": 6989
    },
    {
      "epoch": 0.437839615402684,
      "grad_norm": 0.8468437778884553,
      "learning_rate": 6.236275609667616e-06,
      "loss": 0.4371,
      "step": 6990
    },
    {
      "epoch": 0.4379022534020263,
      "grad_norm": 0.8069586661467938,
      "learning_rate": 6.235292682229066e-06,
      "loss": 0.369,
      "step": 6991
    },
    {
      "epoch": 0.43796489140136863,
      "grad_norm": 0.8533864506029613,
      "learning_rate": 6.23430970394564e-06,
      "loss": 0.4356,
      "step": 6992
    },
    {
      "epoch": 0.43802752940071094,
      "grad_norm": 0.8947234670955009,
      "learning_rate": 6.233326674857798e-06,
      "loss": 0.4401,
      "step": 6993
    },
    {
      "epoch": 0.43809016740005324,
      "grad_norm": 0.8503118105477868,
      "learning_rate": 6.232343595006001e-06,
      "loss": 0.4328,
      "step": 6994
    },
    {
      "epoch": 0.43815280539939555,
      "grad_norm": 0.7796105071986964,
      "learning_rate": 6.231360464430713e-06,
      "loss": 0.3828,
      "step": 6995
    },
    {
      "epoch": 0.43821544339873786,
      "grad_norm": 0.8247537944653608,
      "learning_rate": 6.230377283172399e-06,
      "loss": 0.4237,
      "step": 6996
    },
    {
      "epoch": 0.43827808139808017,
      "grad_norm": 0.8204477576902593,
      "learning_rate": 6.229394051271529e-06,
      "loss": 0.4059,
      "step": 6997
    },
    {
      "epoch": 0.4383407193974225,
      "grad_norm": 0.8392581248420808,
      "learning_rate": 6.228410768768569e-06,
      "loss": 0.4088,
      "step": 6998
    },
    {
      "epoch": 0.4384033573967647,
      "grad_norm": 0.7814993533212238,
      "learning_rate": 6.227427435703997e-06,
      "loss": 0.4789,
      "step": 6999
    },
    {
      "epoch": 0.43846599539610703,
      "grad_norm": 0.8482480648779913,
      "learning_rate": 6.226444052118281e-06,
      "loss": 0.4342,
      "step": 7000
    },
    {
      "epoch": 0.43852863339544934,
      "grad_norm": 0.8354755741080019,
      "learning_rate": 6.225460618051903e-06,
      "loss": 0.4437,
      "step": 7001
    },
    {
      "epoch": 0.43859127139479165,
      "grad_norm": 0.845584704059052,
      "learning_rate": 6.224477133545337e-06,
      "loss": 0.4231,
      "step": 7002
    },
    {
      "epoch": 0.43865390939413396,
      "grad_norm": 0.7836456583199372,
      "learning_rate": 6.223493598639064e-06,
      "loss": 0.4167,
      "step": 7003
    },
    {
      "epoch": 0.43871654739347626,
      "grad_norm": 0.8231558048763771,
      "learning_rate": 6.2225100133735686e-06,
      "loss": 0.433,
      "step": 7004
    },
    {
      "epoch": 0.43877918539281857,
      "grad_norm": 0.7500087693226566,
      "learning_rate": 6.221526377789334e-06,
      "loss": 0.3985,
      "step": 7005
    },
    {
      "epoch": 0.4388418233921609,
      "grad_norm": 0.9385883543602057,
      "learning_rate": 6.220542691926847e-06,
      "loss": 0.4102,
      "step": 7006
    },
    {
      "epoch": 0.43890446139150313,
      "grad_norm": 0.7618135721650444,
      "learning_rate": 6.219558955826597e-06,
      "loss": 0.4008,
      "step": 7007
    },
    {
      "epoch": 0.43896709939084544,
      "grad_norm": 0.8179436682968664,
      "learning_rate": 6.218575169529073e-06,
      "loss": 0.4555,
      "step": 7008
    },
    {
      "epoch": 0.43902973739018775,
      "grad_norm": 0.9234184974136109,
      "learning_rate": 6.217591333074769e-06,
      "loss": 0.3786,
      "step": 7009
    },
    {
      "epoch": 0.43909237538953005,
      "grad_norm": 0.8426266268075493,
      "learning_rate": 6.2166074465041794e-06,
      "loss": 0.3973,
      "step": 7010
    },
    {
      "epoch": 0.43915501338887236,
      "grad_norm": 0.8772702981467944,
      "learning_rate": 6.215623509857802e-06,
      "loss": 0.424,
      "step": 7011
    },
    {
      "epoch": 0.43921765138821467,
      "grad_norm": 0.8816344303042379,
      "learning_rate": 6.2146395231761355e-06,
      "loss": 0.4163,
      "step": 7012
    },
    {
      "epoch": 0.439280289387557,
      "grad_norm": 0.8712330180288441,
      "learning_rate": 6.213655486499681e-06,
      "loss": 0.401,
      "step": 7013
    },
    {
      "epoch": 0.4393429273868993,
      "grad_norm": 0.7876667822546431,
      "learning_rate": 6.2126713998689406e-06,
      "loss": 0.3925,
      "step": 7014
    },
    {
      "epoch": 0.43940556538624154,
      "grad_norm": 0.8539965078256253,
      "learning_rate": 6.211687263324421e-06,
      "loss": 0.412,
      "step": 7015
    },
    {
      "epoch": 0.43946820338558384,
      "grad_norm": 0.791486253456644,
      "learning_rate": 6.210703076906627e-06,
      "loss": 0.3823,
      "step": 7016
    },
    {
      "epoch": 0.43953084138492615,
      "grad_norm": 0.7939741575661771,
      "learning_rate": 6.209718840656072e-06,
      "loss": 0.4042,
      "step": 7017
    },
    {
      "epoch": 0.43959347938426846,
      "grad_norm": 0.8339649302648373,
      "learning_rate": 6.208734554613263e-06,
      "loss": 0.4122,
      "step": 7018
    },
    {
      "epoch": 0.43965611738361077,
      "grad_norm": 0.8729986003978393,
      "learning_rate": 6.207750218818716e-06,
      "loss": 0.4016,
      "step": 7019
    },
    {
      "epoch": 0.4397187553829531,
      "grad_norm": 0.8269813697846571,
      "learning_rate": 6.206765833312947e-06,
      "loss": 0.4151,
      "step": 7020
    },
    {
      "epoch": 0.4397813933822954,
      "grad_norm": 0.8717408650300136,
      "learning_rate": 6.205781398136471e-06,
      "loss": 0.4381,
      "step": 7021
    },
    {
      "epoch": 0.4398440313816377,
      "grad_norm": 0.7943589701578141,
      "learning_rate": 6.2047969133298105e-06,
      "loss": 0.3937,
      "step": 7022
    },
    {
      "epoch": 0.43990666938098,
      "grad_norm": 0.8297281900664509,
      "learning_rate": 6.203812378933482e-06,
      "loss": 0.4109,
      "step": 7023
    },
    {
      "epoch": 0.43996930738032225,
      "grad_norm": 0.793808672655192,
      "learning_rate": 6.202827794988016e-06,
      "loss": 0.3723,
      "step": 7024
    },
    {
      "epoch": 0.44003194537966456,
      "grad_norm": 0.8111247260112464,
      "learning_rate": 6.201843161533934e-06,
      "loss": 0.3945,
      "step": 7025
    },
    {
      "epoch": 0.44009458337900687,
      "grad_norm": 0.8476646595251701,
      "learning_rate": 6.200858478611763e-06,
      "loss": 0.4218,
      "step": 7026
    },
    {
      "epoch": 0.4401572213783492,
      "grad_norm": 0.823637498665479,
      "learning_rate": 6.199873746262035e-06,
      "loss": 0.4175,
      "step": 7027
    },
    {
      "epoch": 0.4402198593776915,
      "grad_norm": 0.8241892241828769,
      "learning_rate": 6.198888964525282e-06,
      "loss": 0.4042,
      "step": 7028
    },
    {
      "epoch": 0.4402824973770338,
      "grad_norm": 0.8950372845408145,
      "learning_rate": 6.197904133442034e-06,
      "loss": 0.4703,
      "step": 7029
    },
    {
      "epoch": 0.4403451353763761,
      "grad_norm": 0.8075587280883637,
      "learning_rate": 6.196919253052832e-06,
      "loss": 0.4392,
      "step": 7030
    },
    {
      "epoch": 0.4404077733757184,
      "grad_norm": 0.8372637489930024,
      "learning_rate": 6.195934323398211e-06,
      "loss": 0.4404,
      "step": 7031
    },
    {
      "epoch": 0.44047041137506066,
      "grad_norm": 0.7844948901612582,
      "learning_rate": 6.19494934451871e-06,
      "loss": 0.3717,
      "step": 7032
    },
    {
      "epoch": 0.44053304937440296,
      "grad_norm": 0.7661259958228197,
      "learning_rate": 6.193964316454872e-06,
      "loss": 0.3873,
      "step": 7033
    },
    {
      "epoch": 0.44059568737374527,
      "grad_norm": 0.6654014298246277,
      "learning_rate": 6.192979239247243e-06,
      "loss": 0.4508,
      "step": 7034
    },
    {
      "epoch": 0.4406583253730876,
      "grad_norm": 0.8284603324915125,
      "learning_rate": 6.191994112936367e-06,
      "loss": 0.4442,
      "step": 7035
    },
    {
      "epoch": 0.4407209633724299,
      "grad_norm": 0.8100722072631561,
      "learning_rate": 6.191008937562791e-06,
      "loss": 0.3742,
      "step": 7036
    },
    {
      "epoch": 0.4407836013717722,
      "grad_norm": 0.9382126723477926,
      "learning_rate": 6.1900237131670645e-06,
      "loss": 0.4305,
      "step": 7037
    },
    {
      "epoch": 0.4408462393711145,
      "grad_norm": 0.7966167220074114,
      "learning_rate": 6.1890384397897445e-06,
      "loss": 0.375,
      "step": 7038
    },
    {
      "epoch": 0.4409088773704568,
      "grad_norm": 0.8044826134925255,
      "learning_rate": 6.18805311747138e-06,
      "loss": 0.4008,
      "step": 7039
    },
    {
      "epoch": 0.4409715153697991,
      "grad_norm": 0.8346667219780086,
      "learning_rate": 6.187067746252529e-06,
      "loss": 0.4731,
      "step": 7040
    },
    {
      "epoch": 0.44103415336914137,
      "grad_norm": 0.9486943430285524,
      "learning_rate": 6.186082326173748e-06,
      "loss": 0.4495,
      "step": 7041
    },
    {
      "epoch": 0.4410967913684837,
      "grad_norm": 0.9002160113070622,
      "learning_rate": 6.1850968572755985e-06,
      "loss": 0.4449,
      "step": 7042
    },
    {
      "epoch": 0.441159429367826,
      "grad_norm": 0.8917996670726691,
      "learning_rate": 6.184111339598644e-06,
      "loss": 0.4243,
      "step": 7043
    },
    {
      "epoch": 0.4412220673671683,
      "grad_norm": 0.7876695825568715,
      "learning_rate": 6.183125773183447e-06,
      "loss": 0.3852,
      "step": 7044
    },
    {
      "epoch": 0.4412847053665106,
      "grad_norm": 0.8209011410864507,
      "learning_rate": 6.182140158070572e-06,
      "loss": 0.4558,
      "step": 7045
    },
    {
      "epoch": 0.4413473433658529,
      "grad_norm": 0.8055283829149532,
      "learning_rate": 6.181154494300589e-06,
      "loss": 0.4122,
      "step": 7046
    },
    {
      "epoch": 0.4414099813651952,
      "grad_norm": 0.8386754764367852,
      "learning_rate": 6.180168781914069e-06,
      "loss": 0.3956,
      "step": 7047
    },
    {
      "epoch": 0.4414726193645375,
      "grad_norm": 0.7829288385910596,
      "learning_rate": 6.179183020951581e-06,
      "loss": 0.3571,
      "step": 7048
    },
    {
      "epoch": 0.4415352573638798,
      "grad_norm": 0.7903657141482003,
      "learning_rate": 6.178197211453703e-06,
      "loss": 0.3876,
      "step": 7049
    },
    {
      "epoch": 0.4415978953632221,
      "grad_norm": 0.8510330803816678,
      "learning_rate": 6.177211353461007e-06,
      "loss": 0.4695,
      "step": 7050
    },
    {
      "epoch": 0.4416605333625644,
      "grad_norm": 0.7514687411352243,
      "learning_rate": 6.176225447014075e-06,
      "loss": 0.3567,
      "step": 7051
    },
    {
      "epoch": 0.4417231713619067,
      "grad_norm": 0.8361808760500966,
      "learning_rate": 6.175239492153484e-06,
      "loss": 0.4143,
      "step": 7052
    },
    {
      "epoch": 0.441785809361249,
      "grad_norm": 0.8312595994143652,
      "learning_rate": 6.1742534889198195e-06,
      "loss": 0.4252,
      "step": 7053
    },
    {
      "epoch": 0.4418484473605913,
      "grad_norm": 0.8273272297968065,
      "learning_rate": 6.17326743735366e-06,
      "loss": 0.3879,
      "step": 7054
    },
    {
      "epoch": 0.4419110853599336,
      "grad_norm": 0.8957439987003355,
      "learning_rate": 6.172281337495597e-06,
      "loss": 0.427,
      "step": 7055
    },
    {
      "epoch": 0.4419737233592759,
      "grad_norm": 0.8378695546485152,
      "learning_rate": 6.171295189386217e-06,
      "loss": 0.4504,
      "step": 7056
    },
    {
      "epoch": 0.4420363613586182,
      "grad_norm": 0.8142648776410771,
      "learning_rate": 6.17030899306611e-06,
      "loss": 0.44,
      "step": 7057
    },
    {
      "epoch": 0.4420989993579605,
      "grad_norm": 0.8888622046876056,
      "learning_rate": 6.169322748575866e-06,
      "loss": 0.4228,
      "step": 7058
    },
    {
      "epoch": 0.4421616373573028,
      "grad_norm": 0.8924114498524893,
      "learning_rate": 6.168336455956081e-06,
      "loss": 0.4487,
      "step": 7059
    },
    {
      "epoch": 0.4422242753566451,
      "grad_norm": 0.7947379330436075,
      "learning_rate": 6.1673501152473506e-06,
      "loss": 0.4063,
      "step": 7060
    },
    {
      "epoch": 0.4422869133559874,
      "grad_norm": 0.8750140277668756,
      "learning_rate": 6.166363726490274e-06,
      "loss": 0.4125,
      "step": 7061
    },
    {
      "epoch": 0.4423495513553297,
      "grad_norm": 0.8652046370207572,
      "learning_rate": 6.1653772897254485e-06,
      "loss": 0.4216,
      "step": 7062
    },
    {
      "epoch": 0.442412189354672,
      "grad_norm": 0.7786271341003169,
      "learning_rate": 6.164390804993479e-06,
      "loss": 0.342,
      "step": 7063
    },
    {
      "epoch": 0.44247482735401433,
      "grad_norm": 0.8524555914423714,
      "learning_rate": 6.1634042723349655e-06,
      "loss": 0.4145,
      "step": 7064
    },
    {
      "epoch": 0.44253746535335664,
      "grad_norm": 0.6989181806113666,
      "learning_rate": 6.162417691790518e-06,
      "loss": 0.4717,
      "step": 7065
    },
    {
      "epoch": 0.4426001033526989,
      "grad_norm": 0.8671545098027121,
      "learning_rate": 6.161431063400742e-06,
      "loss": 0.4458,
      "step": 7066
    },
    {
      "epoch": 0.4426627413520412,
      "grad_norm": 0.8118286067204178,
      "learning_rate": 6.160444387206248e-06,
      "loss": 0.4067,
      "step": 7067
    },
    {
      "epoch": 0.4427253793513835,
      "grad_norm": 0.8381321645901203,
      "learning_rate": 6.159457663247647e-06,
      "loss": 0.4161,
      "step": 7068
    },
    {
      "epoch": 0.4427880173507258,
      "grad_norm": 0.8464990252102608,
      "learning_rate": 6.1584708915655545e-06,
      "loss": 0.4378,
      "step": 7069
    },
    {
      "epoch": 0.4428506553500681,
      "grad_norm": 0.7690046109671237,
      "learning_rate": 6.157484072200585e-06,
      "loss": 0.3996,
      "step": 7070
    },
    {
      "epoch": 0.44291329334941043,
      "grad_norm": 0.6721179732010419,
      "learning_rate": 6.156497205193356e-06,
      "loss": 0.4888,
      "step": 7071
    },
    {
      "epoch": 0.44297593134875274,
      "grad_norm": 0.7692889966518451,
      "learning_rate": 6.155510290584486e-06,
      "loss": 0.369,
      "step": 7072
    },
    {
      "epoch": 0.44303856934809505,
      "grad_norm": 0.8221161859308744,
      "learning_rate": 6.1545233284146015e-06,
      "loss": 0.3673,
      "step": 7073
    },
    {
      "epoch": 0.4431012073474373,
      "grad_norm": 0.7663887119948747,
      "learning_rate": 6.153536318724319e-06,
      "loss": 0.3636,
      "step": 7074
    },
    {
      "epoch": 0.4431638453467796,
      "grad_norm": 0.8082237934145556,
      "learning_rate": 6.1525492615542685e-06,
      "loss": 0.4197,
      "step": 7075
    },
    {
      "epoch": 0.4432264833461219,
      "grad_norm": 0.8311838647353378,
      "learning_rate": 6.151562156945077e-06,
      "loss": 0.3979,
      "step": 7076
    },
    {
      "epoch": 0.4432891213454642,
      "grad_norm": 0.8899660595164237,
      "learning_rate": 6.150575004937374e-06,
      "loss": 0.3904,
      "step": 7077
    },
    {
      "epoch": 0.44335175934480653,
      "grad_norm": 0.8831380240058878,
      "learning_rate": 6.149587805571787e-06,
      "loss": 0.397,
      "step": 7078
    },
    {
      "epoch": 0.44341439734414884,
      "grad_norm": 0.8375994784202399,
      "learning_rate": 6.148600558888956e-06,
      "loss": 0.3914,
      "step": 7079
    },
    {
      "epoch": 0.44347703534349114,
      "grad_norm": 0.7003251154014396,
      "learning_rate": 6.14761326492951e-06,
      "loss": 0.4788,
      "step": 7080
    },
    {
      "epoch": 0.44353967334283345,
      "grad_norm": 0.922495136605924,
      "learning_rate": 6.146625923734092e-06,
      "loss": 0.388,
      "step": 7081
    },
    {
      "epoch": 0.4436023113421757,
      "grad_norm": 0.8111315278702687,
      "learning_rate": 6.145638535343338e-06,
      "loss": 0.4237,
      "step": 7082
    },
    {
      "epoch": 0.443664949341518,
      "grad_norm": 0.8628504554053189,
      "learning_rate": 6.144651099797887e-06,
      "loss": 0.4651,
      "step": 7083
    },
    {
      "epoch": 0.4437275873408603,
      "grad_norm": 0.8315956467854672,
      "learning_rate": 6.143663617138385e-06,
      "loss": 0.3949,
      "step": 7084
    },
    {
      "epoch": 0.4437902253402026,
      "grad_norm": 0.8728054750601754,
      "learning_rate": 6.142676087405477e-06,
      "loss": 0.4237,
      "step": 7085
    },
    {
      "epoch": 0.44385286333954493,
      "grad_norm": 0.7861948733351488,
      "learning_rate": 6.14168851063981e-06,
      "loss": 0.3875,
      "step": 7086
    },
    {
      "epoch": 0.44391550133888724,
      "grad_norm": 0.8732427302054222,
      "learning_rate": 6.140700886882029e-06,
      "loss": 0.4207,
      "step": 7087
    },
    {
      "epoch": 0.44397813933822955,
      "grad_norm": 0.8371482910077233,
      "learning_rate": 6.139713216172789e-06,
      "loss": 0.4046,
      "step": 7088
    },
    {
      "epoch": 0.44404077733757186,
      "grad_norm": 0.8616797855785567,
      "learning_rate": 6.1387254985527425e-06,
      "loss": 0.4082,
      "step": 7089
    },
    {
      "epoch": 0.44410341533691416,
      "grad_norm": 0.8244726489077445,
      "learning_rate": 6.137737734062544e-06,
      "loss": 0.3983,
      "step": 7090
    },
    {
      "epoch": 0.4441660533362564,
      "grad_norm": 0.9133433565493906,
      "learning_rate": 6.1367499227428476e-06,
      "loss": 0.4181,
      "step": 7091
    },
    {
      "epoch": 0.4442286913355987,
      "grad_norm": 0.8051598304697611,
      "learning_rate": 6.135762064634313e-06,
      "loss": 0.4149,
      "step": 7092
    },
    {
      "epoch": 0.44429132933494103,
      "grad_norm": 0.6305433656807986,
      "learning_rate": 6.134774159777601e-06,
      "loss": 0.4569,
      "step": 7093
    },
    {
      "epoch": 0.44435396733428334,
      "grad_norm": 0.8353451585065521,
      "learning_rate": 6.133786208213375e-06,
      "loss": 0.3886,
      "step": 7094
    },
    {
      "epoch": 0.44441660533362565,
      "grad_norm": 0.7931094713157375,
      "learning_rate": 6.132798209982298e-06,
      "loss": 0.4214,
      "step": 7095
    },
    {
      "epoch": 0.44447924333296795,
      "grad_norm": 0.8407195364194993,
      "learning_rate": 6.1318101651250354e-06,
      "loss": 0.3608,
      "step": 7096
    },
    {
      "epoch": 0.44454188133231026,
      "grad_norm": 0.7355747914886198,
      "learning_rate": 6.130822073682257e-06,
      "loss": 0.3761,
      "step": 7097
    },
    {
      "epoch": 0.44460451933165257,
      "grad_norm": 0.8022334593950333,
      "learning_rate": 6.129833935694632e-06,
      "loss": 0.3849,
      "step": 7098
    },
    {
      "epoch": 0.4446671573309948,
      "grad_norm": 0.7807187889242819,
      "learning_rate": 6.128845751202833e-06,
      "loss": 0.3871,
      "step": 7099
    },
    {
      "epoch": 0.44472979533033713,
      "grad_norm": 0.8199217000145429,
      "learning_rate": 6.127857520247535e-06,
      "loss": 0.4195,
      "step": 7100
    },
    {
      "epoch": 0.44479243332967944,
      "grad_norm": 0.786232408065461,
      "learning_rate": 6.126869242869408e-06,
      "loss": 0.4033,
      "step": 7101
    },
    {
      "epoch": 0.44485507132902175,
      "grad_norm": 0.7573738818427667,
      "learning_rate": 6.125880919109136e-06,
      "loss": 0.387,
      "step": 7102
    },
    {
      "epoch": 0.44491770932836405,
      "grad_norm": 0.8872590705638186,
      "learning_rate": 6.1248925490073975e-06,
      "loss": 0.4173,
      "step": 7103
    },
    {
      "epoch": 0.44498034732770636,
      "grad_norm": 0.8557675893489266,
      "learning_rate": 6.123904132604872e-06,
      "loss": 0.422,
      "step": 7104
    },
    {
      "epoch": 0.44504298532704867,
      "grad_norm": 0.8935823086683431,
      "learning_rate": 6.122915669942243e-06,
      "loss": 0.416,
      "step": 7105
    },
    {
      "epoch": 0.445105623326391,
      "grad_norm": 0.8226319645255328,
      "learning_rate": 6.121927161060197e-06,
      "loss": 0.4042,
      "step": 7106
    },
    {
      "epoch": 0.4451682613257333,
      "grad_norm": 0.8455922543170097,
      "learning_rate": 6.120938605999421e-06,
      "loss": 0.3987,
      "step": 7107
    },
    {
      "epoch": 0.44523089932507554,
      "grad_norm": 0.8314268159770023,
      "learning_rate": 6.1199500048006056e-06,
      "loss": 0.4077,
      "step": 7108
    },
    {
      "epoch": 0.44529353732441784,
      "grad_norm": 0.8563341188633548,
      "learning_rate": 6.1189613575044374e-06,
      "loss": 0.3843,
      "step": 7109
    },
    {
      "epoch": 0.44535617532376015,
      "grad_norm": 0.8796905952891079,
      "learning_rate": 6.117972664151613e-06,
      "loss": 0.4186,
      "step": 7110
    },
    {
      "epoch": 0.44541881332310246,
      "grad_norm": 0.6207557589327003,
      "learning_rate": 6.116983924782827e-06,
      "loss": 0.4859,
      "step": 7111
    },
    {
      "epoch": 0.44548145132244477,
      "grad_norm": 0.8090496328075467,
      "learning_rate": 6.115995139438774e-06,
      "loss": 0.3916,
      "step": 7112
    },
    {
      "epoch": 0.4455440893217871,
      "grad_norm": 0.7784797093815878,
      "learning_rate": 6.115006308160155e-06,
      "loss": 0.3775,
      "step": 7113
    },
    {
      "epoch": 0.4456067273211294,
      "grad_norm": 0.8316654879296358,
      "learning_rate": 6.1140174309876685e-06,
      "loss": 0.4329,
      "step": 7114
    },
    {
      "epoch": 0.4456693653204717,
      "grad_norm": 0.7831812235167461,
      "learning_rate": 6.113028507962018e-06,
      "loss": 0.3742,
      "step": 7115
    },
    {
      "epoch": 0.44573200331981394,
      "grad_norm": 0.8472809942638903,
      "learning_rate": 6.112039539123909e-06,
      "loss": 0.3854,
      "step": 7116
    },
    {
      "epoch": 0.44579464131915625,
      "grad_norm": 0.8551334578736515,
      "learning_rate": 6.111050524514045e-06,
      "loss": 0.4333,
      "step": 7117
    },
    {
      "epoch": 0.44585727931849856,
      "grad_norm": 0.8285727018897218,
      "learning_rate": 6.1100614641731345e-06,
      "loss": 0.407,
      "step": 7118
    },
    {
      "epoch": 0.44591991731784086,
      "grad_norm": 0.8767133011964767,
      "learning_rate": 6.1090723581418894e-06,
      "loss": 0.4078,
      "step": 7119
    },
    {
      "epoch": 0.44598255531718317,
      "grad_norm": 0.8925871251992905,
      "learning_rate": 6.10808320646102e-06,
      "loss": 0.4299,
      "step": 7120
    },
    {
      "epoch": 0.4460451933165255,
      "grad_norm": 0.8630030257422803,
      "learning_rate": 6.10709400917124e-06,
      "loss": 0.4011,
      "step": 7121
    },
    {
      "epoch": 0.4461078313158678,
      "grad_norm": 0.8661217069914453,
      "learning_rate": 6.106104766313264e-06,
      "loss": 0.3895,
      "step": 7122
    },
    {
      "epoch": 0.4461704693152101,
      "grad_norm": 0.8219584955515756,
      "learning_rate": 6.105115477927811e-06,
      "loss": 0.3723,
      "step": 7123
    },
    {
      "epoch": 0.44623310731455235,
      "grad_norm": 0.810400339214137,
      "learning_rate": 6.1041261440556e-06,
      "loss": 0.3856,
      "step": 7124
    },
    {
      "epoch": 0.44629574531389465,
      "grad_norm": 0.7332494836448255,
      "learning_rate": 6.103136764737353e-06,
      "loss": 0.4886,
      "step": 7125
    },
    {
      "epoch": 0.44635838331323696,
      "grad_norm": 0.8229785782585796,
      "learning_rate": 6.102147340013791e-06,
      "loss": 0.3981,
      "step": 7126
    },
    {
      "epoch": 0.44642102131257927,
      "grad_norm": 0.7974142041689148,
      "learning_rate": 6.10115786992564e-06,
      "loss": 0.3944,
      "step": 7127
    },
    {
      "epoch": 0.4464836593119216,
      "grad_norm": 0.851538263913061,
      "learning_rate": 6.100168354513627e-06,
      "loss": 0.4355,
      "step": 7128
    },
    {
      "epoch": 0.4465462973112639,
      "grad_norm": 0.9040107065489676,
      "learning_rate": 6.099178793818479e-06,
      "loss": 0.4036,
      "step": 7129
    },
    {
      "epoch": 0.4466089353106062,
      "grad_norm": 0.7778193373558036,
      "learning_rate": 6.09818918788093e-06,
      "loss": 0.4137,
      "step": 7130
    },
    {
      "epoch": 0.4466715733099485,
      "grad_norm": 0.8540810300086795,
      "learning_rate": 6.0971995367417085e-06,
      "loss": 0.4098,
      "step": 7131
    },
    {
      "epoch": 0.4467342113092908,
      "grad_norm": 0.6395549141442667,
      "learning_rate": 6.096209840441552e-06,
      "loss": 0.4579,
      "step": 7132
    },
    {
      "epoch": 0.44679684930863306,
      "grad_norm": 0.8063951118329427,
      "learning_rate": 6.095220099021193e-06,
      "loss": 0.4037,
      "step": 7133
    },
    {
      "epoch": 0.44685948730797537,
      "grad_norm": 0.8346446461995014,
      "learning_rate": 6.094230312521372e-06,
      "loss": 0.4596,
      "step": 7134
    },
    {
      "epoch": 0.4469221253073177,
      "grad_norm": 0.792131783463099,
      "learning_rate": 6.093240480982828e-06,
      "loss": 0.4256,
      "step": 7135
    },
    {
      "epoch": 0.44698476330666,
      "grad_norm": 0.6466205172707459,
      "learning_rate": 6.092250604446305e-06,
      "loss": 0.4516,
      "step": 7136
    },
    {
      "epoch": 0.4470474013060023,
      "grad_norm": 0.8295242508407908,
      "learning_rate": 6.0912606829525435e-06,
      "loss": 0.4278,
      "step": 7137
    },
    {
      "epoch": 0.4471100393053446,
      "grad_norm": 0.856339321192567,
      "learning_rate": 6.090270716542288e-06,
      "loss": 0.4303,
      "step": 7138
    },
    {
      "epoch": 0.4471726773046869,
      "grad_norm": 0.7833189263819842,
      "learning_rate": 6.089280705256289e-06,
      "loss": 0.4051,
      "step": 7139
    },
    {
      "epoch": 0.4472353153040292,
      "grad_norm": 0.8268614721305938,
      "learning_rate": 6.088290649135293e-06,
      "loss": 0.3989,
      "step": 7140
    },
    {
      "epoch": 0.44729795330337146,
      "grad_norm": 0.8238451488559723,
      "learning_rate": 6.087300548220053e-06,
      "loss": 0.3845,
      "step": 7141
    },
    {
      "epoch": 0.4473605913027138,
      "grad_norm": 0.8090677457386504,
      "learning_rate": 6.086310402551321e-06,
      "loss": 0.4101,
      "step": 7142
    },
    {
      "epoch": 0.4474232293020561,
      "grad_norm": 0.9024095421453966,
      "learning_rate": 6.08532021216985e-06,
      "loss": 0.4349,
      "step": 7143
    },
    {
      "epoch": 0.4474858673013984,
      "grad_norm": 0.8508369587450704,
      "learning_rate": 6.084329977116398e-06,
      "loss": 0.4242,
      "step": 7144
    },
    {
      "epoch": 0.4475485053007407,
      "grad_norm": 0.8735107205836279,
      "learning_rate": 6.083339697431723e-06,
      "loss": 0.4154,
      "step": 7145
    },
    {
      "epoch": 0.447611143300083,
      "grad_norm": 0.8050134215608332,
      "learning_rate": 6.0823493731565855e-06,
      "loss": 0.414,
      "step": 7146
    },
    {
      "epoch": 0.4476737812994253,
      "grad_norm": 0.8322143378368125,
      "learning_rate": 6.081359004331745e-06,
      "loss": 0.392,
      "step": 7147
    },
    {
      "epoch": 0.4477364192987676,
      "grad_norm": 0.6591606008588861,
      "learning_rate": 6.080368590997968e-06,
      "loss": 0.4666,
      "step": 7148
    },
    {
      "epoch": 0.4477990572981099,
      "grad_norm": 0.8705855803952741,
      "learning_rate": 6.07937813319602e-06,
      "loss": 0.3927,
      "step": 7149
    },
    {
      "epoch": 0.4478616952974522,
      "grad_norm": 0.8191001997004829,
      "learning_rate": 6.078387630966668e-06,
      "loss": 0.4403,
      "step": 7150
    },
    {
      "epoch": 0.4479243332967945,
      "grad_norm": 0.8244242339645655,
      "learning_rate": 6.077397084350681e-06,
      "loss": 0.4073,
      "step": 7151
    },
    {
      "epoch": 0.4479869712961368,
      "grad_norm": 0.6145455597775608,
      "learning_rate": 6.076406493388829e-06,
      "loss": 0.4494,
      "step": 7152
    },
    {
      "epoch": 0.4480496092954791,
      "grad_norm": 0.8049483822742195,
      "learning_rate": 6.075415858121885e-06,
      "loss": 0.4682,
      "step": 7153
    },
    {
      "epoch": 0.4481122472948214,
      "grad_norm": 0.8521312582594768,
      "learning_rate": 6.074425178590627e-06,
      "loss": 0.4282,
      "step": 7154
    },
    {
      "epoch": 0.4481748852941637,
      "grad_norm": 0.7687532391203126,
      "learning_rate": 6.073434454835828e-06,
      "loss": 0.381,
      "step": 7155
    },
    {
      "epoch": 0.448237523293506,
      "grad_norm": 0.8563825254080027,
      "learning_rate": 6.072443686898267e-06,
      "loss": 0.3799,
      "step": 7156
    },
    {
      "epoch": 0.44830016129284833,
      "grad_norm": 0.8552363984042972,
      "learning_rate": 6.071452874818726e-06,
      "loss": 0.4109,
      "step": 7157
    },
    {
      "epoch": 0.4483627992921906,
      "grad_norm": 0.8192543934537149,
      "learning_rate": 6.070462018637986e-06,
      "loss": 0.406,
      "step": 7158
    },
    {
      "epoch": 0.4484254372915329,
      "grad_norm": 0.8203607587516769,
      "learning_rate": 6.06947111839683e-06,
      "loss": 0.3981,
      "step": 7159
    },
    {
      "epoch": 0.4484880752908752,
      "grad_norm": 0.8686874810985228,
      "learning_rate": 6.068480174136044e-06,
      "loss": 0.4245,
      "step": 7160
    },
    {
      "epoch": 0.4485507132902175,
      "grad_norm": 0.7974934551535245,
      "learning_rate": 6.067489185896416e-06,
      "loss": 0.4183,
      "step": 7161
    },
    {
      "epoch": 0.4486133512895598,
      "grad_norm": 0.7921863895147582,
      "learning_rate": 6.066498153718735e-06,
      "loss": 0.4163,
      "step": 7162
    },
    {
      "epoch": 0.4486759892889021,
      "grad_norm": 0.8706224699889564,
      "learning_rate": 6.0655070776437916e-06,
      "loss": 0.4514,
      "step": 7163
    },
    {
      "epoch": 0.44873862728824443,
      "grad_norm": 0.8429639184155217,
      "learning_rate": 6.0645159577123804e-06,
      "loss": 0.4128,
      "step": 7164
    },
    {
      "epoch": 0.44880126528758674,
      "grad_norm": 0.8816285450705315,
      "learning_rate": 6.063524793965294e-06,
      "loss": 0.4111,
      "step": 7165
    },
    {
      "epoch": 0.448863903286929,
      "grad_norm": 0.8980590371946763,
      "learning_rate": 6.062533586443329e-06,
      "loss": 0.4181,
      "step": 7166
    },
    {
      "epoch": 0.4489265412862713,
      "grad_norm": 0.8209123636597258,
      "learning_rate": 6.061542335187285e-06,
      "loss": 0.4322,
      "step": 7167
    },
    {
      "epoch": 0.4489891792856136,
      "grad_norm": 0.8402773135175436,
      "learning_rate": 6.0605510402379615e-06,
      "loss": 0.4161,
      "step": 7168
    },
    {
      "epoch": 0.4490518172849559,
      "grad_norm": 0.8017186527333146,
      "learning_rate": 6.05955970163616e-06,
      "loss": 0.3887,
      "step": 7169
    },
    {
      "epoch": 0.4491144552842982,
      "grad_norm": 0.8375699949573485,
      "learning_rate": 6.058568319422684e-06,
      "loss": 0.4274,
      "step": 7170
    },
    {
      "epoch": 0.4491770932836405,
      "grad_norm": 0.7981348982712422,
      "learning_rate": 6.057576893638342e-06,
      "loss": 0.4001,
      "step": 7171
    },
    {
      "epoch": 0.44923973128298283,
      "grad_norm": 0.7989963827478735,
      "learning_rate": 6.056585424323936e-06,
      "loss": 0.3984,
      "step": 7172
    },
    {
      "epoch": 0.44930236928232514,
      "grad_norm": 0.8511002512830053,
      "learning_rate": 6.055593911520279e-06,
      "loss": 0.4442,
      "step": 7173
    },
    {
      "epoch": 0.44936500728166745,
      "grad_norm": 0.8159146752787921,
      "learning_rate": 6.0546023552681806e-06,
      "loss": 0.3898,
      "step": 7174
    },
    {
      "epoch": 0.4494276452810097,
      "grad_norm": 0.7580487086638622,
      "learning_rate": 6.0536107556084535e-06,
      "loss": 0.3819,
      "step": 7175
    },
    {
      "epoch": 0.449490283280352,
      "grad_norm": 0.8597433479142107,
      "learning_rate": 6.052619112581913e-06,
      "loss": 0.4044,
      "step": 7176
    },
    {
      "epoch": 0.4495529212796943,
      "grad_norm": 0.8132200409486989,
      "learning_rate": 6.051627426229373e-06,
      "loss": 0.3988,
      "step": 7177
    },
    {
      "epoch": 0.4496155592790366,
      "grad_norm": 0.6936749197215721,
      "learning_rate": 6.050635696591653e-06,
      "loss": 0.4547,
      "step": 7178
    },
    {
      "epoch": 0.44967819727837893,
      "grad_norm": 0.8579318161513325,
      "learning_rate": 6.049643923709573e-06,
      "loss": 0.4027,
      "step": 7179
    },
    {
      "epoch": 0.44974083527772124,
      "grad_norm": 0.8332800318729066,
      "learning_rate": 6.048652107623954e-06,
      "loss": 0.3898,
      "step": 7180
    },
    {
      "epoch": 0.44980347327706355,
      "grad_norm": 0.7877560336325864,
      "learning_rate": 6.047660248375622e-06,
      "loss": 0.4124,
      "step": 7181
    },
    {
      "epoch": 0.44986611127640586,
      "grad_norm": 0.8476343596987411,
      "learning_rate": 6.046668346005397e-06,
      "loss": 0.4649,
      "step": 7182
    },
    {
      "epoch": 0.4499287492757481,
      "grad_norm": 0.8428252615744056,
      "learning_rate": 6.045676400554109e-06,
      "loss": 0.3902,
      "step": 7183
    },
    {
      "epoch": 0.4499913872750904,
      "grad_norm": 0.8611401604948618,
      "learning_rate": 6.044684412062587e-06,
      "loss": 0.4463,
      "step": 7184
    },
    {
      "epoch": 0.4500540252744327,
      "grad_norm": 0.8674530762534604,
      "learning_rate": 6.043692380571661e-06,
      "loss": 0.4269,
      "step": 7185
    },
    {
      "epoch": 0.45011666327377503,
      "grad_norm": 0.9301775577807422,
      "learning_rate": 6.0427003061221614e-06,
      "loss": 0.4893,
      "step": 7186
    },
    {
      "epoch": 0.45017930127311734,
      "grad_norm": 0.8653250068656798,
      "learning_rate": 6.041708188754925e-06,
      "loss": 0.4006,
      "step": 7187
    },
    {
      "epoch": 0.45024193927245965,
      "grad_norm": 0.8540147433529238,
      "learning_rate": 6.040716028510786e-06,
      "loss": 0.4451,
      "step": 7188
    },
    {
      "epoch": 0.45030457727180195,
      "grad_norm": 0.8153822526764654,
      "learning_rate": 6.039723825430581e-06,
      "loss": 0.4232,
      "step": 7189
    },
    {
      "epoch": 0.45036721527114426,
      "grad_norm": 0.8171232445376646,
      "learning_rate": 6.038731579555152e-06,
      "loss": 0.4042,
      "step": 7190
    },
    {
      "epoch": 0.4504298532704865,
      "grad_norm": 0.8573441351092969,
      "learning_rate": 6.0377392909253376e-06,
      "loss": 0.4411,
      "step": 7191
    },
    {
      "epoch": 0.4504924912698288,
      "grad_norm": 0.7627894332241419,
      "learning_rate": 6.036746959581982e-06,
      "loss": 0.3629,
      "step": 7192
    },
    {
      "epoch": 0.45055512926917113,
      "grad_norm": 0.8176446668467777,
      "learning_rate": 6.035754585565927e-06,
      "loss": 0.4233,
      "step": 7193
    },
    {
      "epoch": 0.45061776726851344,
      "grad_norm": 0.8084247839805621,
      "learning_rate": 6.034762168918024e-06,
      "loss": 0.3828,
      "step": 7194
    },
    {
      "epoch": 0.45068040526785574,
      "grad_norm": 0.8800578579238509,
      "learning_rate": 6.0337697096791144e-06,
      "loss": 0.4221,
      "step": 7195
    },
    {
      "epoch": 0.45074304326719805,
      "grad_norm": 0.9009094330588684,
      "learning_rate": 6.032777207890055e-06,
      "loss": 0.3879,
      "step": 7196
    },
    {
      "epoch": 0.45080568126654036,
      "grad_norm": 0.896902597065488,
      "learning_rate": 6.0317846635916935e-06,
      "loss": 0.4126,
      "step": 7197
    },
    {
      "epoch": 0.45086831926588267,
      "grad_norm": 0.656208554985264,
      "learning_rate": 6.030792076824883e-06,
      "loss": 0.438,
      "step": 7198
    },
    {
      "epoch": 0.450930957265225,
      "grad_norm": 0.796579891830537,
      "learning_rate": 6.029799447630478e-06,
      "loss": 0.4314,
      "step": 7199
    },
    {
      "epoch": 0.4509935952645672,
      "grad_norm": 0.8437462553481901,
      "learning_rate": 6.028806776049339e-06,
      "loss": 0.4143,
      "step": 7200
    },
    {
      "epoch": 0.45105623326390953,
      "grad_norm": 0.8018665365021894,
      "learning_rate": 6.027814062122321e-06,
      "loss": 0.4217,
      "step": 7201
    },
    {
      "epoch": 0.45111887126325184,
      "grad_norm": 0.7744697237486667,
      "learning_rate": 6.026821305890286e-06,
      "loss": 0.4085,
      "step": 7202
    },
    {
      "epoch": 0.45118150926259415,
      "grad_norm": 0.8077280285591566,
      "learning_rate": 6.025828507394095e-06,
      "loss": 0.42,
      "step": 7203
    },
    {
      "epoch": 0.45124414726193646,
      "grad_norm": 0.8247573917541463,
      "learning_rate": 6.024835666674611e-06,
      "loss": 0.4363,
      "step": 7204
    },
    {
      "epoch": 0.45130678526127876,
      "grad_norm": 0.8275986413561498,
      "learning_rate": 6.023842783772703e-06,
      "loss": 0.4266,
      "step": 7205
    },
    {
      "epoch": 0.45136942326062107,
      "grad_norm": 0.8757032081907496,
      "learning_rate": 6.0228498587292365e-06,
      "loss": 0.4479,
      "step": 7206
    },
    {
      "epoch": 0.4514320612599634,
      "grad_norm": 0.7984752285686231,
      "learning_rate": 6.021856891585077e-06,
      "loss": 0.4153,
      "step": 7207
    },
    {
      "epoch": 0.45149469925930563,
      "grad_norm": 0.9135778520380424,
      "learning_rate": 6.020863882381098e-06,
      "loss": 0.4214,
      "step": 7208
    },
    {
      "epoch": 0.45155733725864794,
      "grad_norm": 0.6417643242734742,
      "learning_rate": 6.019870831158175e-06,
      "loss": 0.4696,
      "step": 7209
    },
    {
      "epoch": 0.45161997525799025,
      "grad_norm": 0.8794297851419889,
      "learning_rate": 6.018877737957179e-06,
      "loss": 0.4395,
      "step": 7210
    },
    {
      "epoch": 0.45168261325733255,
      "grad_norm": 0.7742964076250707,
      "learning_rate": 6.017884602818985e-06,
      "loss": 0.4077,
      "step": 7211
    },
    {
      "epoch": 0.45174525125667486,
      "grad_norm": 0.915118538286134,
      "learning_rate": 6.016891425784472e-06,
      "loss": 0.4422,
      "step": 7212
    },
    {
      "epoch": 0.45180788925601717,
      "grad_norm": 0.8448738438341328,
      "learning_rate": 6.015898206894518e-06,
      "loss": 0.4185,
      "step": 7213
    },
    {
      "epoch": 0.4518705272553595,
      "grad_norm": 0.6016199551742631,
      "learning_rate": 6.014904946190008e-06,
      "loss": 0.4659,
      "step": 7214
    },
    {
      "epoch": 0.4519331652547018,
      "grad_norm": 0.8553399796461781,
      "learning_rate": 6.01391164371182e-06,
      "loss": 0.3957,
      "step": 7215
    },
    {
      "epoch": 0.4519958032540441,
      "grad_norm": 0.7266611524468352,
      "learning_rate": 6.0129182995008405e-06,
      "loss": 0.3903,
      "step": 7216
    },
    {
      "epoch": 0.45205844125338634,
      "grad_norm": 0.8740985103964959,
      "learning_rate": 6.011924913597956e-06,
      "loss": 0.4304,
      "step": 7217
    },
    {
      "epoch": 0.45212107925272865,
      "grad_norm": 0.8088478312169057,
      "learning_rate": 6.010931486044056e-06,
      "loss": 0.3928,
      "step": 7218
    },
    {
      "epoch": 0.45218371725207096,
      "grad_norm": 0.8901720030938991,
      "learning_rate": 6.009938016880027e-06,
      "loss": 0.4272,
      "step": 7219
    },
    {
      "epoch": 0.45224635525141327,
      "grad_norm": 0.9100925074105825,
      "learning_rate": 6.008944506146761e-06,
      "loss": 0.4347,
      "step": 7220
    },
    {
      "epoch": 0.4523089932507556,
      "grad_norm": 0.9586398208814862,
      "learning_rate": 6.007950953885152e-06,
      "loss": 0.4001,
      "step": 7221
    },
    {
      "epoch": 0.4523716312500979,
      "grad_norm": 0.8262855164295421,
      "learning_rate": 6.006957360136097e-06,
      "loss": 0.4282,
      "step": 7222
    },
    {
      "epoch": 0.4524342692494402,
      "grad_norm": 0.8477116858061433,
      "learning_rate": 6.005963724940488e-06,
      "loss": 0.4301,
      "step": 7223
    },
    {
      "epoch": 0.4524969072487825,
      "grad_norm": 0.8548030898981446,
      "learning_rate": 6.0049700483392256e-06,
      "loss": 0.4397,
      "step": 7224
    },
    {
      "epoch": 0.45255954524812475,
      "grad_norm": 0.8346272474980441,
      "learning_rate": 6.003976330373209e-06,
      "loss": 0.4013,
      "step": 7225
    },
    {
      "epoch": 0.45262218324746706,
      "grad_norm": 0.8489884076278897,
      "learning_rate": 6.002982571083341e-06,
      "loss": 0.4238,
      "step": 7226
    },
    {
      "epoch": 0.45268482124680937,
      "grad_norm": 0.8158716853757892,
      "learning_rate": 6.001988770510525e-06,
      "loss": 0.3594,
      "step": 7227
    },
    {
      "epoch": 0.4527474592461517,
      "grad_norm": 0.7944858110094658,
      "learning_rate": 6.000994928695665e-06,
      "loss": 0.4066,
      "step": 7228
    },
    {
      "epoch": 0.452810097245494,
      "grad_norm": 0.8653958505107483,
      "learning_rate": 6.0000010456796665e-06,
      "loss": 0.4251,
      "step": 7229
    },
    {
      "epoch": 0.4528727352448363,
      "grad_norm": 0.7772557692617131,
      "learning_rate": 5.9990071215034395e-06,
      "loss": 0.4029,
      "step": 7230
    },
    {
      "epoch": 0.4529353732441786,
      "grad_norm": 0.8123052725847562,
      "learning_rate": 5.998013156207895e-06,
      "loss": 0.3994,
      "step": 7231
    },
    {
      "epoch": 0.4529980112435209,
      "grad_norm": 0.8980272345833492,
      "learning_rate": 5.997019149833944e-06,
      "loss": 0.4497,
      "step": 7232
    },
    {
      "epoch": 0.45306064924286316,
      "grad_norm": 0.8997153538505126,
      "learning_rate": 5.996025102422498e-06,
      "loss": 0.4584,
      "step": 7233
    },
    {
      "epoch": 0.45312328724220546,
      "grad_norm": 0.7957474076371804,
      "learning_rate": 5.995031014014476e-06,
      "loss": 0.3767,
      "step": 7234
    },
    {
      "epoch": 0.45318592524154777,
      "grad_norm": 0.6991705267232449,
      "learning_rate": 5.994036884650792e-06,
      "loss": 0.4598,
      "step": 7235
    },
    {
      "epoch": 0.4532485632408901,
      "grad_norm": 0.7972065800131347,
      "learning_rate": 5.993042714372365e-06,
      "loss": 0.4249,
      "step": 7236
    },
    {
      "epoch": 0.4533112012402324,
      "grad_norm": 0.7997565009618652,
      "learning_rate": 5.992048503220116e-06,
      "loss": 0.4099,
      "step": 7237
    },
    {
      "epoch": 0.4533738392395747,
      "grad_norm": 0.9269149754130059,
      "learning_rate": 5.9910542512349655e-06,
      "loss": 0.451,
      "step": 7238
    },
    {
      "epoch": 0.453436477238917,
      "grad_norm": 0.7412081967403893,
      "learning_rate": 5.990059958457839e-06,
      "loss": 0.3982,
      "step": 7239
    },
    {
      "epoch": 0.4534991152382593,
      "grad_norm": 0.7413833629069678,
      "learning_rate": 5.989065624929661e-06,
      "loss": 0.3528,
      "step": 7240
    },
    {
      "epoch": 0.4535617532376016,
      "grad_norm": 0.7678300664366698,
      "learning_rate": 5.988071250691358e-06,
      "loss": 0.4105,
      "step": 7241
    },
    {
      "epoch": 0.45362439123694387,
      "grad_norm": 0.7832415636166005,
      "learning_rate": 5.9870768357838584e-06,
      "loss": 0.4305,
      "step": 7242
    },
    {
      "epoch": 0.4536870292362862,
      "grad_norm": 0.8231020377789262,
      "learning_rate": 5.986082380248094e-06,
      "loss": 0.4161,
      "step": 7243
    },
    {
      "epoch": 0.4537496672356285,
      "grad_norm": 0.8424602295240952,
      "learning_rate": 5.9850878841249935e-06,
      "loss": 0.4172,
      "step": 7244
    },
    {
      "epoch": 0.4538123052349708,
      "grad_norm": 0.8478914826706868,
      "learning_rate": 5.984093347455496e-06,
      "loss": 0.377,
      "step": 7245
    },
    {
      "epoch": 0.4538749432343131,
      "grad_norm": 0.8825148847709534,
      "learning_rate": 5.983098770280531e-06,
      "loss": 0.4654,
      "step": 7246
    },
    {
      "epoch": 0.4539375812336554,
      "grad_norm": 0.8256113327984209,
      "learning_rate": 5.982104152641041e-06,
      "loss": 0.3589,
      "step": 7247
    },
    {
      "epoch": 0.4540002192329977,
      "grad_norm": 0.6541957784372626,
      "learning_rate": 5.981109494577959e-06,
      "loss": 0.493,
      "step": 7248
    },
    {
      "epoch": 0.45406285723234,
      "grad_norm": 0.8049634569905252,
      "learning_rate": 5.98011479613223e-06,
      "loss": 0.4322,
      "step": 7249
    },
    {
      "epoch": 0.4541254952316823,
      "grad_norm": 0.8177960484829336,
      "learning_rate": 5.979120057344793e-06,
      "loss": 0.3814,
      "step": 7250
    },
    {
      "epoch": 0.4541881332310246,
      "grad_norm": 0.8255318684891345,
      "learning_rate": 5.978125278256593e-06,
      "loss": 0.4249,
      "step": 7251
    },
    {
      "epoch": 0.4542507712303669,
      "grad_norm": 0.8364076351267591,
      "learning_rate": 5.977130458908575e-06,
      "loss": 0.4361,
      "step": 7252
    },
    {
      "epoch": 0.4543134092297092,
      "grad_norm": 0.8503698441021725,
      "learning_rate": 5.976135599341686e-06,
      "loss": 0.3993,
      "step": 7253
    },
    {
      "epoch": 0.4543760472290515,
      "grad_norm": 0.8597839893711302,
      "learning_rate": 5.9751406995968754e-06,
      "loss": 0.3426,
      "step": 7254
    },
    {
      "epoch": 0.4544386852283938,
      "grad_norm": 0.8829433054587587,
      "learning_rate": 5.974145759715091e-06,
      "loss": 0.4138,
      "step": 7255
    },
    {
      "epoch": 0.4545013232277361,
      "grad_norm": 0.791114094366797,
      "learning_rate": 5.973150779737288e-06,
      "loss": 0.3855,
      "step": 7256
    },
    {
      "epoch": 0.45456396122707843,
      "grad_norm": 0.8037950307773267,
      "learning_rate": 5.9721557597044175e-06,
      "loss": 0.4186,
      "step": 7257
    },
    {
      "epoch": 0.45462659922642074,
      "grad_norm": 0.8445348635700054,
      "learning_rate": 5.971160699657434e-06,
      "loss": 0.4194,
      "step": 7258
    },
    {
      "epoch": 0.454689237225763,
      "grad_norm": 0.7492297485444325,
      "learning_rate": 5.970165599637298e-06,
      "loss": 0.3762,
      "step": 7259
    },
    {
      "epoch": 0.4547518752251053,
      "grad_norm": 0.8048854752931841,
      "learning_rate": 5.969170459684965e-06,
      "loss": 0.4397,
      "step": 7260
    },
    {
      "epoch": 0.4548145132244476,
      "grad_norm": 0.8344182587756327,
      "learning_rate": 5.968175279841397e-06,
      "loss": 0.4235,
      "step": 7261
    },
    {
      "epoch": 0.4548771512237899,
      "grad_norm": 0.8335477473486507,
      "learning_rate": 5.967180060147553e-06,
      "loss": 0.3836,
      "step": 7262
    },
    {
      "epoch": 0.4549397892231322,
      "grad_norm": 0.6622232972913409,
      "learning_rate": 5.966184800644399e-06,
      "loss": 0.4807,
      "step": 7263
    },
    {
      "epoch": 0.4550024272224745,
      "grad_norm": 0.7826662355061684,
      "learning_rate": 5.965189501372899e-06,
      "loss": 0.396,
      "step": 7264
    },
    {
      "epoch": 0.45506506522181683,
      "grad_norm": 0.8527757338849207,
      "learning_rate": 5.964194162374022e-06,
      "loss": 0.4214,
      "step": 7265
    },
    {
      "epoch": 0.45512770322115914,
      "grad_norm": 0.7874642856658781,
      "learning_rate": 5.963198783688733e-06,
      "loss": 0.3843,
      "step": 7266
    },
    {
      "epoch": 0.4551903412205014,
      "grad_norm": 0.8432308947115811,
      "learning_rate": 5.9622033653580015e-06,
      "loss": 0.3926,
      "step": 7267
    },
    {
      "epoch": 0.4552529792198437,
      "grad_norm": 0.676961002980568,
      "learning_rate": 5.961207907422801e-06,
      "loss": 0.4971,
      "step": 7268
    },
    {
      "epoch": 0.455315617219186,
      "grad_norm": 0.802238308978865,
      "learning_rate": 5.960212409924108e-06,
      "loss": 0.3621,
      "step": 7269
    },
    {
      "epoch": 0.4553782552185283,
      "grad_norm": 0.9103414622585043,
      "learning_rate": 5.959216872902892e-06,
      "loss": 0.4069,
      "step": 7270
    },
    {
      "epoch": 0.4554408932178706,
      "grad_norm": 0.8369955642609506,
      "learning_rate": 5.958221296400131e-06,
      "loss": 0.4036,
      "step": 7271
    },
    {
      "epoch": 0.45550353121721293,
      "grad_norm": 0.8307780593611529,
      "learning_rate": 5.957225680456803e-06,
      "loss": 0.3882,
      "step": 7272
    },
    {
      "epoch": 0.45556616921655524,
      "grad_norm": 0.7559519333460364,
      "learning_rate": 5.956230025113889e-06,
      "loss": 0.4078,
      "step": 7273
    },
    {
      "epoch": 0.45562880721589755,
      "grad_norm": 0.8005507454004357,
      "learning_rate": 5.955234330412372e-06,
      "loss": 0.3792,
      "step": 7274
    },
    {
      "epoch": 0.4556914452152398,
      "grad_norm": 0.7714666743046654,
      "learning_rate": 5.95423859639323e-06,
      "loss": 0.4011,
      "step": 7275
    },
    {
      "epoch": 0.4557540832145821,
      "grad_norm": 0.8602224262267443,
      "learning_rate": 5.9532428230974495e-06,
      "loss": 0.406,
      "step": 7276
    },
    {
      "epoch": 0.4558167212139244,
      "grad_norm": 0.7821113578712118,
      "learning_rate": 5.952247010566019e-06,
      "loss": 0.3993,
      "step": 7277
    },
    {
      "epoch": 0.4558793592132667,
      "grad_norm": 0.7968523230689563,
      "learning_rate": 5.951251158839926e-06,
      "loss": 0.3658,
      "step": 7278
    },
    {
      "epoch": 0.45594199721260903,
      "grad_norm": 0.7929512932386589,
      "learning_rate": 5.950255267960158e-06,
      "loss": 0.4239,
      "step": 7279
    },
    {
      "epoch": 0.45600463521195134,
      "grad_norm": 0.7518411673128701,
      "learning_rate": 5.949259337967706e-06,
      "loss": 0.3939,
      "step": 7280
    },
    {
      "epoch": 0.45606727321129364,
      "grad_norm": 0.8124532081326706,
      "learning_rate": 5.948263368903562e-06,
      "loss": 0.3988,
      "step": 7281
    },
    {
      "epoch": 0.45612991121063595,
      "grad_norm": 0.6312193288914961,
      "learning_rate": 5.947267360808724e-06,
      "loss": 0.4827,
      "step": 7282
    },
    {
      "epoch": 0.45619254920997826,
      "grad_norm": 1.0272610775242232,
      "learning_rate": 5.946271313724186e-06,
      "loss": 0.4219,
      "step": 7283
    },
    {
      "epoch": 0.4562551872093205,
      "grad_norm": 0.8321960873131456,
      "learning_rate": 5.945275227690943e-06,
      "loss": 0.4247,
      "step": 7284
    },
    {
      "epoch": 0.4563178252086628,
      "grad_norm": 0.8363029815161974,
      "learning_rate": 5.944279102749997e-06,
      "loss": 0.4179,
      "step": 7285
    },
    {
      "epoch": 0.4563804632080051,
      "grad_norm": 0.7978496855467927,
      "learning_rate": 5.943282938942349e-06,
      "loss": 0.3737,
      "step": 7286
    },
    {
      "epoch": 0.45644310120734743,
      "grad_norm": 0.828444359698197,
      "learning_rate": 5.9422867363089985e-06,
      "loss": 0.3995,
      "step": 7287
    },
    {
      "epoch": 0.45650573920668974,
      "grad_norm": 0.8108228439742083,
      "learning_rate": 5.941290494890952e-06,
      "loss": 0.4049,
      "step": 7288
    },
    {
      "epoch": 0.45656837720603205,
      "grad_norm": 0.8735607121477229,
      "learning_rate": 5.940294214729214e-06,
      "loss": 0.4624,
      "step": 7289
    },
    {
      "epoch": 0.45663101520537436,
      "grad_norm": 0.7471955054343452,
      "learning_rate": 5.939297895864791e-06,
      "loss": 0.3401,
      "step": 7290
    },
    {
      "epoch": 0.45669365320471667,
      "grad_norm": 0.845189451689567,
      "learning_rate": 5.9383015383386936e-06,
      "loss": 0.4352,
      "step": 7291
    },
    {
      "epoch": 0.4567562912040589,
      "grad_norm": 0.7915203261108215,
      "learning_rate": 5.937305142191929e-06,
      "loss": 0.3552,
      "step": 7292
    },
    {
      "epoch": 0.4568189292034012,
      "grad_norm": 0.8451649834964304,
      "learning_rate": 5.936308707465511e-06,
      "loss": 0.4199,
      "step": 7293
    },
    {
      "epoch": 0.45688156720274353,
      "grad_norm": 0.7848438405783399,
      "learning_rate": 5.935312234200454e-06,
      "loss": 0.3699,
      "step": 7294
    },
    {
      "epoch": 0.45694420520208584,
      "grad_norm": 0.8329741914552646,
      "learning_rate": 5.934315722437771e-06,
      "loss": 0.4172,
      "step": 7295
    },
    {
      "epoch": 0.45700684320142815,
      "grad_norm": 0.8010992069826075,
      "learning_rate": 5.93331917221848e-06,
      "loss": 0.4181,
      "step": 7296
    },
    {
      "epoch": 0.45706948120077046,
      "grad_norm": 0.838393993158111,
      "learning_rate": 5.932322583583598e-06,
      "loss": 0.4118,
      "step": 7297
    },
    {
      "epoch": 0.45713211920011276,
      "grad_norm": 0.7880807078156197,
      "learning_rate": 5.931325956574147e-06,
      "loss": 0.3923,
      "step": 7298
    },
    {
      "epoch": 0.45719475719945507,
      "grad_norm": 0.8383268761573646,
      "learning_rate": 5.930329291231144e-06,
      "loss": 0.4407,
      "step": 7299
    },
    {
      "epoch": 0.4572573951987973,
      "grad_norm": 0.8691589342035542,
      "learning_rate": 5.9293325875956165e-06,
      "loss": 0.4119,
      "step": 7300
    },
    {
      "epoch": 0.45732003319813963,
      "grad_norm": 0.8681791299524413,
      "learning_rate": 5.928335845708588e-06,
      "loss": 0.4372,
      "step": 7301
    },
    {
      "epoch": 0.45738267119748194,
      "grad_norm": 0.8399079125852288,
      "learning_rate": 5.9273390656110815e-06,
      "loss": 0.4151,
      "step": 7302
    },
    {
      "epoch": 0.45744530919682425,
      "grad_norm": 0.7838295480120805,
      "learning_rate": 5.926342247344129e-06,
      "loss": 0.3677,
      "step": 7303
    },
    {
      "epoch": 0.45750794719616655,
      "grad_norm": 0.8145316661231898,
      "learning_rate": 5.925345390948756e-06,
      "loss": 0.4128,
      "step": 7304
    },
    {
      "epoch": 0.45757058519550886,
      "grad_norm": 0.8381544553506137,
      "learning_rate": 5.924348496465997e-06,
      "loss": 0.4141,
      "step": 7305
    },
    {
      "epoch": 0.45763322319485117,
      "grad_norm": 0.8214593687780598,
      "learning_rate": 5.92335156393688e-06,
      "loss": 0.4115,
      "step": 7306
    },
    {
      "epoch": 0.4576958611941935,
      "grad_norm": 0.9767153709675364,
      "learning_rate": 5.922354593402443e-06,
      "loss": 0.3943,
      "step": 7307
    },
    {
      "epoch": 0.4577584991935358,
      "grad_norm": 0.8340497119915431,
      "learning_rate": 5.921357584903719e-06,
      "loss": 0.4285,
      "step": 7308
    },
    {
      "epoch": 0.45782113719287804,
      "grad_norm": 0.8284387745098624,
      "learning_rate": 5.920360538481746e-06,
      "loss": 0.4243,
      "step": 7309
    },
    {
      "epoch": 0.45788377519222034,
      "grad_norm": 0.7634700150856443,
      "learning_rate": 5.919363454177561e-06,
      "loss": 0.3724,
      "step": 7310
    },
    {
      "epoch": 0.45794641319156265,
      "grad_norm": 0.7980359832052474,
      "learning_rate": 5.918366332032207e-06,
      "loss": 0.3673,
      "step": 7311
    },
    {
      "epoch": 0.45800905119090496,
      "grad_norm": 0.8498059581182648,
      "learning_rate": 5.917369172086726e-06,
      "loss": 0.4229,
      "step": 7312
    },
    {
      "epoch": 0.45807168919024727,
      "grad_norm": 0.822376149540633,
      "learning_rate": 5.916371974382156e-06,
      "loss": 0.36,
      "step": 7313
    },
    {
      "epoch": 0.4581343271895896,
      "grad_norm": 0.8494242501379571,
      "learning_rate": 5.915374738959548e-06,
      "loss": 0.4503,
      "step": 7314
    },
    {
      "epoch": 0.4581969651889319,
      "grad_norm": 0.7673803978908711,
      "learning_rate": 5.914377465859944e-06,
      "loss": 0.3521,
      "step": 7315
    },
    {
      "epoch": 0.4582596031882742,
      "grad_norm": 0.8133205129927373,
      "learning_rate": 5.913380155124395e-06,
      "loss": 0.3966,
      "step": 7316
    },
    {
      "epoch": 0.45832224118761644,
      "grad_norm": 0.8335715603952599,
      "learning_rate": 5.912382806793949e-06,
      "loss": 0.4052,
      "step": 7317
    },
    {
      "epoch": 0.45838487918695875,
      "grad_norm": 0.8948647948463924,
      "learning_rate": 5.911385420909657e-06,
      "loss": 0.4605,
      "step": 7318
    },
    {
      "epoch": 0.45844751718630106,
      "grad_norm": 0.8230782313989555,
      "learning_rate": 5.910387997512573e-06,
      "loss": 0.4343,
      "step": 7319
    },
    {
      "epoch": 0.45851015518564336,
      "grad_norm": 0.8428510706525638,
      "learning_rate": 5.90939053664375e-06,
      "loss": 0.4141,
      "step": 7320
    },
    {
      "epoch": 0.45857279318498567,
      "grad_norm": 0.8557103743815296,
      "learning_rate": 5.908393038344243e-06,
      "loss": 0.4453,
      "step": 7321
    },
    {
      "epoch": 0.458635431184328,
      "grad_norm": 0.7942404996490774,
      "learning_rate": 5.90739550265511e-06,
      "loss": 0.3788,
      "step": 7322
    },
    {
      "epoch": 0.4586980691836703,
      "grad_norm": 0.8349473458142908,
      "learning_rate": 5.9063979296174086e-06,
      "loss": 0.4041,
      "step": 7323
    },
    {
      "epoch": 0.4587607071830126,
      "grad_norm": 0.7755314593346991,
      "learning_rate": 5.905400319272202e-06,
      "loss": 0.3924,
      "step": 7324
    },
    {
      "epoch": 0.4588233451823549,
      "grad_norm": 0.8210139456554161,
      "learning_rate": 5.904402671660551e-06,
      "loss": 0.3858,
      "step": 7325
    },
    {
      "epoch": 0.45888598318169715,
      "grad_norm": 0.8128185216527007,
      "learning_rate": 5.903404986823517e-06,
      "loss": 0.4171,
      "step": 7326
    },
    {
      "epoch": 0.45894862118103946,
      "grad_norm": 0.7999496764809298,
      "learning_rate": 5.9024072648021645e-06,
      "loss": 0.392,
      "step": 7327
    },
    {
      "epoch": 0.45901125918038177,
      "grad_norm": 0.8523537473306387,
      "learning_rate": 5.901409505637562e-06,
      "loss": 0.4117,
      "step": 7328
    },
    {
      "epoch": 0.4590738971797241,
      "grad_norm": 0.8620185305477255,
      "learning_rate": 5.900411709370779e-06,
      "loss": 0.4233,
      "step": 7329
    },
    {
      "epoch": 0.4591365351790664,
      "grad_norm": 0.8405955274170902,
      "learning_rate": 5.8994138760428825e-06,
      "loss": 0.4042,
      "step": 7330
    },
    {
      "epoch": 0.4591991731784087,
      "grad_norm": 0.6238118298975999,
      "learning_rate": 5.898416005694941e-06,
      "loss": 0.4447,
      "step": 7331
    },
    {
      "epoch": 0.459261811177751,
      "grad_norm": 0.7858707112381,
      "learning_rate": 5.8974180983680314e-06,
      "loss": 0.4067,
      "step": 7332
    },
    {
      "epoch": 0.4593244491770933,
      "grad_norm": 0.7398816421154825,
      "learning_rate": 5.896420154103229e-06,
      "loss": 0.3647,
      "step": 7333
    },
    {
      "epoch": 0.45938708717643556,
      "grad_norm": 0.8118826689322494,
      "learning_rate": 5.895422172941604e-06,
      "loss": 0.4133,
      "step": 7334
    },
    {
      "epoch": 0.45944972517577787,
      "grad_norm": 0.8736544930781404,
      "learning_rate": 5.894424154924237e-06,
      "loss": 0.4147,
      "step": 7335
    },
    {
      "epoch": 0.4595123631751202,
      "grad_norm": 0.8581565161244351,
      "learning_rate": 5.893426100092204e-06,
      "loss": 0.3878,
      "step": 7336
    },
    {
      "epoch": 0.4595750011744625,
      "grad_norm": 0.8585870259703227,
      "learning_rate": 5.89242800848659e-06,
      "loss": 0.4181,
      "step": 7337
    },
    {
      "epoch": 0.4596376391738048,
      "grad_norm": 0.817670804559115,
      "learning_rate": 5.891429880148471e-06,
      "loss": 0.4279,
      "step": 7338
    },
    {
      "epoch": 0.4597002771731471,
      "grad_norm": 0.7498579162682844,
      "learning_rate": 5.890431715118935e-06,
      "loss": 0.4087,
      "step": 7339
    },
    {
      "epoch": 0.4597629151724894,
      "grad_norm": 0.8598189668173543,
      "learning_rate": 5.889433513439063e-06,
      "loss": 0.4069,
      "step": 7340
    },
    {
      "epoch": 0.4598255531718317,
      "grad_norm": 0.8704018355318833,
      "learning_rate": 5.8884352751499405e-06,
      "loss": 0.4489,
      "step": 7341
    },
    {
      "epoch": 0.45988819117117397,
      "grad_norm": 0.914280208018377,
      "learning_rate": 5.887437000292659e-06,
      "loss": 0.4333,
      "step": 7342
    },
    {
      "epoch": 0.4599508291705163,
      "grad_norm": 0.8926674931602008,
      "learning_rate": 5.886438688908307e-06,
      "loss": 0.4229,
      "step": 7343
    },
    {
      "epoch": 0.4600134671698586,
      "grad_norm": 0.7667393727255714,
      "learning_rate": 5.885440341037973e-06,
      "loss": 0.3797,
      "step": 7344
    },
    {
      "epoch": 0.4600761051692009,
      "grad_norm": 0.8748169027843923,
      "learning_rate": 5.88444195672275e-06,
      "loss": 0.4194,
      "step": 7345
    },
    {
      "epoch": 0.4601387431685432,
      "grad_norm": 0.8723679585137472,
      "learning_rate": 5.883443536003733e-06,
      "loss": 0.4403,
      "step": 7346
    },
    {
      "epoch": 0.4602013811678855,
      "grad_norm": 0.8809060515290255,
      "learning_rate": 5.882445078922014e-06,
      "loss": 0.4357,
      "step": 7347
    },
    {
      "epoch": 0.4602640191672278,
      "grad_norm": 0.8825258016798011,
      "learning_rate": 5.881446585518693e-06,
      "loss": 0.4298,
      "step": 7348
    },
    {
      "epoch": 0.4603266571665701,
      "grad_norm": 0.809090792080625,
      "learning_rate": 5.880448055834868e-06,
      "loss": 0.4268,
      "step": 7349
    },
    {
      "epoch": 0.4603892951659124,
      "grad_norm": 0.831922604302968,
      "learning_rate": 5.879449489911635e-06,
      "loss": 0.4075,
      "step": 7350
    },
    {
      "epoch": 0.4604519331652547,
      "grad_norm": 0.7050610930835589,
      "learning_rate": 5.8784508877901004e-06,
      "loss": 0.4539,
      "step": 7351
    },
    {
      "epoch": 0.460514571164597,
      "grad_norm": 0.8645349657925256,
      "learning_rate": 5.877452249511363e-06,
      "loss": 0.4544,
      "step": 7352
    },
    {
      "epoch": 0.4605772091639393,
      "grad_norm": 0.8715842727312607,
      "learning_rate": 5.876453575116527e-06,
      "loss": 0.4347,
      "step": 7353
    },
    {
      "epoch": 0.4606398471632816,
      "grad_norm": 0.8965626852176053,
      "learning_rate": 5.8754548646467e-06,
      "loss": 0.4271,
      "step": 7354
    },
    {
      "epoch": 0.4607024851626239,
      "grad_norm": 0.8624845947153723,
      "learning_rate": 5.874456118142989e-06,
      "loss": 0.3979,
      "step": 7355
    },
    {
      "epoch": 0.4607651231619662,
      "grad_norm": 0.8113106068186875,
      "learning_rate": 5.873457335646503e-06,
      "loss": 0.3752,
      "step": 7356
    },
    {
      "epoch": 0.4608277611613085,
      "grad_norm": 0.8325246820847738,
      "learning_rate": 5.872458517198348e-06,
      "loss": 0.3766,
      "step": 7357
    },
    {
      "epoch": 0.46089039916065083,
      "grad_norm": 0.8410104805790336,
      "learning_rate": 5.871459662839642e-06,
      "loss": 0.4272,
      "step": 7358
    },
    {
      "epoch": 0.4609530371599931,
      "grad_norm": 0.7144333823217909,
      "learning_rate": 5.87046077261149e-06,
      "loss": 0.4655,
      "step": 7359
    },
    {
      "epoch": 0.4610156751593354,
      "grad_norm": 0.8045494538602319,
      "learning_rate": 5.869461846555014e-06,
      "loss": 0.3815,
      "step": 7360
    },
    {
      "epoch": 0.4610783131586777,
      "grad_norm": 0.8429837272819095,
      "learning_rate": 5.868462884711327e-06,
      "loss": 0.4134,
      "step": 7361
    },
    {
      "epoch": 0.46114095115802,
      "grad_norm": 0.8541785848323837,
      "learning_rate": 5.867463887121546e-06,
      "loss": 0.4192,
      "step": 7362
    },
    {
      "epoch": 0.4612035891573623,
      "grad_norm": 0.9035022451821358,
      "learning_rate": 5.866464853826791e-06,
      "loss": 0.4115,
      "step": 7363
    },
    {
      "epoch": 0.4612662271567046,
      "grad_norm": 0.87572486866157,
      "learning_rate": 5.86546578486818e-06,
      "loss": 0.4563,
      "step": 7364
    },
    {
      "epoch": 0.46132886515604693,
      "grad_norm": 0.8055881119790603,
      "learning_rate": 5.864466680286839e-06,
      "loss": 0.364,
      "step": 7365
    },
    {
      "epoch": 0.46139150315538924,
      "grad_norm": 0.7787034659412275,
      "learning_rate": 5.863467540123887e-06,
      "loss": 0.3662,
      "step": 7366
    },
    {
      "epoch": 0.46145414115473155,
      "grad_norm": 0.8243900788371756,
      "learning_rate": 5.862468364420453e-06,
      "loss": 0.4152,
      "step": 7367
    },
    {
      "epoch": 0.4615167791540738,
      "grad_norm": 0.7266384640143674,
      "learning_rate": 5.861469153217659e-06,
      "loss": 0.3883,
      "step": 7368
    },
    {
      "epoch": 0.4615794171534161,
      "grad_norm": 0.8996614565751281,
      "learning_rate": 5.860469906556636e-06,
      "loss": 0.466,
      "step": 7369
    },
    {
      "epoch": 0.4616420551527584,
      "grad_norm": 0.8237977942780469,
      "learning_rate": 5.859470624478511e-06,
      "loss": 0.4055,
      "step": 7370
    },
    {
      "epoch": 0.4617046931521007,
      "grad_norm": 0.7625333606902456,
      "learning_rate": 5.858471307024416e-06,
      "loss": 0.4246,
      "step": 7371
    },
    {
      "epoch": 0.46176733115144303,
      "grad_norm": 0.8195220619684618,
      "learning_rate": 5.857471954235483e-06,
      "loss": 0.4041,
      "step": 7372
    },
    {
      "epoch": 0.46182996915078534,
      "grad_norm": 0.8566103027716838,
      "learning_rate": 5.856472566152844e-06,
      "loss": 0.399,
      "step": 7373
    },
    {
      "epoch": 0.46189260715012764,
      "grad_norm": 0.8663553332767043,
      "learning_rate": 5.855473142817636e-06,
      "loss": 0.3985,
      "step": 7374
    },
    {
      "epoch": 0.46195524514946995,
      "grad_norm": 0.8160725943159235,
      "learning_rate": 5.854473684270995e-06,
      "loss": 0.4142,
      "step": 7375
    },
    {
      "epoch": 0.4620178831488122,
      "grad_norm": 0.8421100503058886,
      "learning_rate": 5.853474190554058e-06,
      "loss": 0.4265,
      "step": 7376
    },
    {
      "epoch": 0.4620805211481545,
      "grad_norm": 0.8287256247778623,
      "learning_rate": 5.852474661707965e-06,
      "loss": 0.3988,
      "step": 7377
    },
    {
      "epoch": 0.4621431591474968,
      "grad_norm": 0.9119647416530688,
      "learning_rate": 5.851475097773857e-06,
      "loss": 0.4331,
      "step": 7378
    },
    {
      "epoch": 0.4622057971468391,
      "grad_norm": 0.8720433269500734,
      "learning_rate": 5.850475498792876e-06,
      "loss": 0.4142,
      "step": 7379
    },
    {
      "epoch": 0.46226843514618143,
      "grad_norm": 0.808103969005756,
      "learning_rate": 5.8494758648061655e-06,
      "loss": 0.4056,
      "step": 7380
    },
    {
      "epoch": 0.46233107314552374,
      "grad_norm": 0.8971023745264679,
      "learning_rate": 5.848476195854871e-06,
      "loss": 0.3987,
      "step": 7381
    },
    {
      "epoch": 0.46239371114486605,
      "grad_norm": 0.8263179995861649,
      "learning_rate": 5.847476491980137e-06,
      "loss": 0.3832,
      "step": 7382
    },
    {
      "epoch": 0.46245634914420836,
      "grad_norm": 0.819024684154698,
      "learning_rate": 5.846476753223115e-06,
      "loss": 0.3591,
      "step": 7383
    },
    {
      "epoch": 0.4625189871435506,
      "grad_norm": 0.8008599077062348,
      "learning_rate": 5.845476979624953e-06,
      "loss": 0.3837,
      "step": 7384
    },
    {
      "epoch": 0.4625816251428929,
      "grad_norm": 0.798239931713754,
      "learning_rate": 5.8444771712268015e-06,
      "loss": 0.407,
      "step": 7385
    },
    {
      "epoch": 0.4626442631422352,
      "grad_norm": 0.8513655300037652,
      "learning_rate": 5.843477328069813e-06,
      "loss": 0.433,
      "step": 7386
    },
    {
      "epoch": 0.46270690114157753,
      "grad_norm": 0.8202401661802331,
      "learning_rate": 5.842477450195138e-06,
      "loss": 0.4248,
      "step": 7387
    },
    {
      "epoch": 0.46276953914091984,
      "grad_norm": 0.7499279211909682,
      "learning_rate": 5.841477537643938e-06,
      "loss": 0.3831,
      "step": 7388
    },
    {
      "epoch": 0.46283217714026215,
      "grad_norm": 0.8925546698337153,
      "learning_rate": 5.840477590457366e-06,
      "loss": 0.45,
      "step": 7389
    },
    {
      "epoch": 0.46289481513960445,
      "grad_norm": 0.8366760896348576,
      "learning_rate": 5.839477608676582e-06,
      "loss": 0.4387,
      "step": 7390
    },
    {
      "epoch": 0.46295745313894676,
      "grad_norm": 0.7929495638272848,
      "learning_rate": 5.838477592342741e-06,
      "loss": 0.384,
      "step": 7391
    },
    {
      "epoch": 0.46302009113828907,
      "grad_norm": 0.7907330638753441,
      "learning_rate": 5.837477541497006e-06,
      "loss": 0.4092,
      "step": 7392
    },
    {
      "epoch": 0.4630827291376313,
      "grad_norm": 0.8718516747683523,
      "learning_rate": 5.836477456180543e-06,
      "loss": 0.4153,
      "step": 7393
    },
    {
      "epoch": 0.46314536713697363,
      "grad_norm": 0.8634919499168434,
      "learning_rate": 5.83547733643451e-06,
      "loss": 0.423,
      "step": 7394
    },
    {
      "epoch": 0.46320800513631594,
      "grad_norm": 0.8195498171432036,
      "learning_rate": 5.834477182300077e-06,
      "loss": 0.4129,
      "step": 7395
    },
    {
      "epoch": 0.46327064313565824,
      "grad_norm": 0.8283967146489453,
      "learning_rate": 5.833476993818406e-06,
      "loss": 0.4193,
      "step": 7396
    },
    {
      "epoch": 0.46333328113500055,
      "grad_norm": 0.7214546980554266,
      "learning_rate": 5.832476771030669e-06,
      "loss": 0.3438,
      "step": 7397
    },
    {
      "epoch": 0.46339591913434286,
      "grad_norm": 0.7009911276719792,
      "learning_rate": 5.831476513978034e-06,
      "loss": 0.4436,
      "step": 7398
    },
    {
      "epoch": 0.46345855713368517,
      "grad_norm": 0.7777192251161787,
      "learning_rate": 5.83047622270167e-06,
      "loss": 0.3697,
      "step": 7399
    },
    {
      "epoch": 0.4635211951330275,
      "grad_norm": 0.8274355710571631,
      "learning_rate": 5.829475897242749e-06,
      "loss": 0.3899,
      "step": 7400
    },
    {
      "epoch": 0.4635838331323697,
      "grad_norm": 0.7897544376367838,
      "learning_rate": 5.828475537642448e-06,
      "loss": 0.3673,
      "step": 7401
    },
    {
      "epoch": 0.46364647113171203,
      "grad_norm": 0.8201423286933427,
      "learning_rate": 5.827475143941941e-06,
      "loss": 0.4165,
      "step": 7402
    },
    {
      "epoch": 0.46370910913105434,
      "grad_norm": 0.8619591948710202,
      "learning_rate": 5.826474716182403e-06,
      "loss": 0.4279,
      "step": 7403
    },
    {
      "epoch": 0.46377174713039665,
      "grad_norm": 0.9020847820697517,
      "learning_rate": 5.825474254405012e-06,
      "loss": 0.4429,
      "step": 7404
    },
    {
      "epoch": 0.46383438512973896,
      "grad_norm": 0.7991288882428081,
      "learning_rate": 5.824473758650947e-06,
      "loss": 0.3661,
      "step": 7405
    },
    {
      "epoch": 0.46389702312908127,
      "grad_norm": 0.8242451018719648,
      "learning_rate": 5.823473228961389e-06,
      "loss": 0.3866,
      "step": 7406
    },
    {
      "epoch": 0.4639596611284236,
      "grad_norm": 0.8346464712699988,
      "learning_rate": 5.822472665377521e-06,
      "loss": 0.4427,
      "step": 7407
    },
    {
      "epoch": 0.4640222991277659,
      "grad_norm": 0.8502275552881667,
      "learning_rate": 5.821472067940525e-06,
      "loss": 0.4459,
      "step": 7408
    },
    {
      "epoch": 0.46408493712710813,
      "grad_norm": 0.8275483928234003,
      "learning_rate": 5.820471436691586e-06,
      "loss": 0.3907,
      "step": 7409
    },
    {
      "epoch": 0.46414757512645044,
      "grad_norm": 0.7389800279344092,
      "learning_rate": 5.81947077167189e-06,
      "loss": 0.3677,
      "step": 7410
    },
    {
      "epoch": 0.46421021312579275,
      "grad_norm": 0.8771556716891337,
      "learning_rate": 5.8184700729226265e-06,
      "loss": 0.3971,
      "step": 7411
    },
    {
      "epoch": 0.46427285112513506,
      "grad_norm": 0.6664626269209791,
      "learning_rate": 5.817469340484982e-06,
      "loss": 0.4651,
      "step": 7412
    },
    {
      "epoch": 0.46433548912447736,
      "grad_norm": 0.7455795770548862,
      "learning_rate": 5.816468574400147e-06,
      "loss": 0.3814,
      "step": 7413
    },
    {
      "epoch": 0.46439812712381967,
      "grad_norm": 0.8215362728953497,
      "learning_rate": 5.815467774709314e-06,
      "loss": 0.4305,
      "step": 7414
    },
    {
      "epoch": 0.464460765123162,
      "grad_norm": 0.8271448778303871,
      "learning_rate": 5.814466941453677e-06,
      "loss": 0.4136,
      "step": 7415
    },
    {
      "epoch": 0.4645234031225043,
      "grad_norm": 0.676706790181029,
      "learning_rate": 5.813466074674429e-06,
      "loss": 0.4531,
      "step": 7416
    },
    {
      "epoch": 0.4645860411218466,
      "grad_norm": 0.8477768968248793,
      "learning_rate": 5.812465174412766e-06,
      "loss": 0.4134,
      "step": 7417
    },
    {
      "epoch": 0.46464867912118885,
      "grad_norm": 0.8263711211638451,
      "learning_rate": 5.8114642407098865e-06,
      "loss": 0.3843,
      "step": 7418
    },
    {
      "epoch": 0.46471131712053115,
      "grad_norm": 0.749129566199124,
      "learning_rate": 5.810463273606987e-06,
      "loss": 0.3697,
      "step": 7419
    },
    {
      "epoch": 0.46477395511987346,
      "grad_norm": 0.8507969964086752,
      "learning_rate": 5.80946227314527e-06,
      "loss": 0.4263,
      "step": 7420
    },
    {
      "epoch": 0.46483659311921577,
      "grad_norm": 0.7927445044684227,
      "learning_rate": 5.808461239365934e-06,
      "loss": 0.4221,
      "step": 7421
    },
    {
      "epoch": 0.4648992311185581,
      "grad_norm": 0.824096088848119,
      "learning_rate": 5.807460172310185e-06,
      "loss": 0.4188,
      "step": 7422
    },
    {
      "epoch": 0.4649618691179004,
      "grad_norm": 0.7537717968147264,
      "learning_rate": 5.806459072019225e-06,
      "loss": 0.3663,
      "step": 7423
    },
    {
      "epoch": 0.4650245071172427,
      "grad_norm": 0.853620318677933,
      "learning_rate": 5.80545793853426e-06,
      "loss": 0.3957,
      "step": 7424
    },
    {
      "epoch": 0.465087145116585,
      "grad_norm": 0.8717049055654306,
      "learning_rate": 5.804456771896497e-06,
      "loss": 0.4441,
      "step": 7425
    },
    {
      "epoch": 0.46514978311592725,
      "grad_norm": 0.8002957321497987,
      "learning_rate": 5.803455572147143e-06,
      "loss": 0.3911,
      "step": 7426
    },
    {
      "epoch": 0.46521242111526956,
      "grad_norm": 0.8380948528397048,
      "learning_rate": 5.802454339327409e-06,
      "loss": 0.4012,
      "step": 7427
    },
    {
      "epoch": 0.46527505911461187,
      "grad_norm": 0.8440693745078609,
      "learning_rate": 5.801453073478505e-06,
      "loss": 0.3898,
      "step": 7428
    },
    {
      "epoch": 0.4653376971139542,
      "grad_norm": 0.7830974693343077,
      "learning_rate": 5.800451774641645e-06,
      "loss": 0.3982,
      "step": 7429
    },
    {
      "epoch": 0.4654003351132965,
      "grad_norm": 0.8235463420977145,
      "learning_rate": 5.79945044285804e-06,
      "loss": 0.3942,
      "step": 7430
    },
    {
      "epoch": 0.4654629731126388,
      "grad_norm": 0.84601901837183,
      "learning_rate": 5.7984490781689075e-06,
      "loss": 0.4509,
      "step": 7431
    },
    {
      "epoch": 0.4655256111119811,
      "grad_norm": 0.780489882091647,
      "learning_rate": 5.7974476806154625e-06,
      "loss": 0.4245,
      "step": 7432
    },
    {
      "epoch": 0.4655882491113234,
      "grad_norm": 0.840866180882839,
      "learning_rate": 5.796446250238922e-06,
      "loss": 0.396,
      "step": 7433
    },
    {
      "epoch": 0.4656508871106657,
      "grad_norm": 0.8238289446905294,
      "learning_rate": 5.795444787080505e-06,
      "loss": 0.3944,
      "step": 7434
    },
    {
      "epoch": 0.46571352511000796,
      "grad_norm": 0.8654746891910412,
      "learning_rate": 5.794443291181435e-06,
      "loss": 0.3917,
      "step": 7435
    },
    {
      "epoch": 0.46577616310935027,
      "grad_norm": 0.834900624760926,
      "learning_rate": 5.793441762582932e-06,
      "loss": 0.3934,
      "step": 7436
    },
    {
      "epoch": 0.4658388011086926,
      "grad_norm": 0.8551059388070777,
      "learning_rate": 5.792440201326218e-06,
      "loss": 0.4625,
      "step": 7437
    },
    {
      "epoch": 0.4659014391080349,
      "grad_norm": 0.7737114942014963,
      "learning_rate": 5.791438607452519e-06,
      "loss": 0.4327,
      "step": 7438
    },
    {
      "epoch": 0.4659640771073772,
      "grad_norm": 0.8101937617228511,
      "learning_rate": 5.790436981003058e-06,
      "loss": 0.3799,
      "step": 7439
    },
    {
      "epoch": 0.4660267151067195,
      "grad_norm": 0.8747019259419067,
      "learning_rate": 5.789435322019066e-06,
      "loss": 0.3939,
      "step": 7440
    },
    {
      "epoch": 0.4660893531060618,
      "grad_norm": 0.8154530784203402,
      "learning_rate": 5.78843363054177e-06,
      "loss": 0.4049,
      "step": 7441
    },
    {
      "epoch": 0.4661519911054041,
      "grad_norm": 0.8047054682992045,
      "learning_rate": 5.787431906612397e-06,
      "loss": 0.3656,
      "step": 7442
    },
    {
      "epoch": 0.46621462910474637,
      "grad_norm": 0.8312431507581111,
      "learning_rate": 5.786430150272182e-06,
      "loss": 0.4327,
      "step": 7443
    },
    {
      "epoch": 0.4662772671040887,
      "grad_norm": 0.8050909273940338,
      "learning_rate": 5.785428361562357e-06,
      "loss": 0.417,
      "step": 7444
    },
    {
      "epoch": 0.466339905103431,
      "grad_norm": 0.765909715989697,
      "learning_rate": 5.784426540524154e-06,
      "loss": 0.3868,
      "step": 7445
    },
    {
      "epoch": 0.4664025431027733,
      "grad_norm": 0.7817451956711788,
      "learning_rate": 5.783424687198809e-06,
      "loss": 0.4034,
      "step": 7446
    },
    {
      "epoch": 0.4664651811021156,
      "grad_norm": 0.8187632305779504,
      "learning_rate": 5.782422801627557e-06,
      "loss": 0.4274,
      "step": 7447
    },
    {
      "epoch": 0.4665278191014579,
      "grad_norm": 0.786024040176266,
      "learning_rate": 5.78142088385164e-06,
      "loss": 0.383,
      "step": 7448
    },
    {
      "epoch": 0.4665904571008002,
      "grad_norm": 0.8232906135260617,
      "learning_rate": 5.780418933912293e-06,
      "loss": 0.3958,
      "step": 7449
    },
    {
      "epoch": 0.4666530951001425,
      "grad_norm": 0.8396062708107969,
      "learning_rate": 5.7794169518507585e-06,
      "loss": 0.3945,
      "step": 7450
    },
    {
      "epoch": 0.4667157330994848,
      "grad_norm": 0.825691840445722,
      "learning_rate": 5.778414937708276e-06,
      "loss": 0.4045,
      "step": 7451
    },
    {
      "epoch": 0.4667783710988271,
      "grad_norm": 0.908447591958243,
      "learning_rate": 5.777412891526091e-06,
      "loss": 0.4314,
      "step": 7452
    },
    {
      "epoch": 0.4668410090981694,
      "grad_norm": 0.7800017760278237,
      "learning_rate": 5.776410813345448e-06,
      "loss": 0.4122,
      "step": 7453
    },
    {
      "epoch": 0.4669036470975117,
      "grad_norm": 0.8213402626346648,
      "learning_rate": 5.775408703207591e-06,
      "loss": 0.393,
      "step": 7454
    },
    {
      "epoch": 0.466966285096854,
      "grad_norm": 0.832634675947275,
      "learning_rate": 5.774406561153767e-06,
      "loss": 0.3917,
      "step": 7455
    },
    {
      "epoch": 0.4670289230961963,
      "grad_norm": 0.7674389860916158,
      "learning_rate": 5.773404387225226e-06,
      "loss": 0.445,
      "step": 7456
    },
    {
      "epoch": 0.4670915610955386,
      "grad_norm": 0.8739051655898448,
      "learning_rate": 5.772402181463217e-06,
      "loss": 0.461,
      "step": 7457
    },
    {
      "epoch": 0.46715419909488093,
      "grad_norm": 0.8059657556138964,
      "learning_rate": 5.7713999439089915e-06,
      "loss": 0.4002,
      "step": 7458
    },
    {
      "epoch": 0.46721683709422324,
      "grad_norm": 0.8139310226828742,
      "learning_rate": 5.7703976746038e-06,
      "loss": 0.381,
      "step": 7459
    },
    {
      "epoch": 0.4672794750935655,
      "grad_norm": 0.960453991554637,
      "learning_rate": 5.769395373588897e-06,
      "loss": 0.4756,
      "step": 7460
    },
    {
      "epoch": 0.4673421130929078,
      "grad_norm": 0.8475941611735102,
      "learning_rate": 5.768393040905539e-06,
      "loss": 0.4265,
      "step": 7461
    },
    {
      "epoch": 0.4674047510922501,
      "grad_norm": 0.6253514780400943,
      "learning_rate": 5.76739067659498e-06,
      "loss": 0.4809,
      "step": 7462
    },
    {
      "epoch": 0.4674673890915924,
      "grad_norm": 0.8063981044182457,
      "learning_rate": 5.76638828069848e-06,
      "loss": 0.351,
      "step": 7463
    },
    {
      "epoch": 0.4675300270909347,
      "grad_norm": 0.8157475873116644,
      "learning_rate": 5.765385853257294e-06,
      "loss": 0.4035,
      "step": 7464
    },
    {
      "epoch": 0.467592665090277,
      "grad_norm": 0.840665134848864,
      "learning_rate": 5.764383394312684e-06,
      "loss": 0.3796,
      "step": 7465
    },
    {
      "epoch": 0.46765530308961933,
      "grad_norm": 0.8279326664516277,
      "learning_rate": 5.763380903905913e-06,
      "loss": 0.3995,
      "step": 7466
    },
    {
      "epoch": 0.46771794108896164,
      "grad_norm": 0.86820643607514,
      "learning_rate": 5.762378382078242e-06,
      "loss": 0.4488,
      "step": 7467
    },
    {
      "epoch": 0.4677805790883039,
      "grad_norm": 0.8415419885074327,
      "learning_rate": 5.761375828870934e-06,
      "loss": 0.4281,
      "step": 7468
    },
    {
      "epoch": 0.4678432170876462,
      "grad_norm": 0.8453737575053423,
      "learning_rate": 5.7603732443252554e-06,
      "loss": 0.4012,
      "step": 7469
    },
    {
      "epoch": 0.4679058550869885,
      "grad_norm": 0.8499282191070417,
      "learning_rate": 5.759370628482474e-06,
      "loss": 0.3729,
      "step": 7470
    },
    {
      "epoch": 0.4679684930863308,
      "grad_norm": 0.7560869044350427,
      "learning_rate": 5.758367981383858e-06,
      "loss": 0.3794,
      "step": 7471
    },
    {
      "epoch": 0.4680311310856731,
      "grad_norm": 0.8043343460175949,
      "learning_rate": 5.757365303070673e-06,
      "loss": 0.3829,
      "step": 7472
    },
    {
      "epoch": 0.46809376908501543,
      "grad_norm": 0.787883070083223,
      "learning_rate": 5.7563625935841925e-06,
      "loss": 0.4186,
      "step": 7473
    },
    {
      "epoch": 0.46815640708435774,
      "grad_norm": 0.9240505124751506,
      "learning_rate": 5.7553598529656875e-06,
      "loss": 0.3923,
      "step": 7474
    },
    {
      "epoch": 0.46821904508370005,
      "grad_norm": 0.8941126645497888,
      "learning_rate": 5.754357081256431e-06,
      "loss": 0.4202,
      "step": 7475
    },
    {
      "epoch": 0.46828168308304235,
      "grad_norm": 0.8501711034412889,
      "learning_rate": 5.753354278497696e-06,
      "loss": 0.4211,
      "step": 7476
    },
    {
      "epoch": 0.4683443210823846,
      "grad_norm": 0.905971888222946,
      "learning_rate": 5.75235144473076e-06,
      "loss": 0.4467,
      "step": 7477
    },
    {
      "epoch": 0.4684069590817269,
      "grad_norm": 0.8740673006880181,
      "learning_rate": 5.751348579996898e-06,
      "loss": 0.408,
      "step": 7478
    },
    {
      "epoch": 0.4684695970810692,
      "grad_norm": 0.8422784295749376,
      "learning_rate": 5.750345684337388e-06,
      "loss": 0.3992,
      "step": 7479
    },
    {
      "epoch": 0.46853223508041153,
      "grad_norm": 0.8525338732967783,
      "learning_rate": 5.749342757793513e-06,
      "loss": 0.407,
      "step": 7480
    },
    {
      "epoch": 0.46859487307975384,
      "grad_norm": 0.8766151161124832,
      "learning_rate": 5.74833980040655e-06,
      "loss": 0.4463,
      "step": 7481
    },
    {
      "epoch": 0.46865751107909615,
      "grad_norm": 1.011638690852574,
      "learning_rate": 5.747336812217781e-06,
      "loss": 0.4496,
      "step": 7482
    },
    {
      "epoch": 0.46872014907843845,
      "grad_norm": 0.8382143424759507,
      "learning_rate": 5.74633379326849e-06,
      "loss": 0.4055,
      "step": 7483
    },
    {
      "epoch": 0.46878278707778076,
      "grad_norm": 0.8161131203502818,
      "learning_rate": 5.745330743599963e-06,
      "loss": 0.4134,
      "step": 7484
    },
    {
      "epoch": 0.468845425077123,
      "grad_norm": 0.8886880008660026,
      "learning_rate": 5.7443276632534825e-06,
      "loss": 0.3754,
      "step": 7485
    },
    {
      "epoch": 0.4689080630764653,
      "grad_norm": 0.8601911572220059,
      "learning_rate": 5.74332455227034e-06,
      "loss": 0.406,
      "step": 7486
    },
    {
      "epoch": 0.46897070107580763,
      "grad_norm": 0.8559423270916431,
      "learning_rate": 5.742321410691819e-06,
      "loss": 0.4047,
      "step": 7487
    },
    {
      "epoch": 0.46903333907514994,
      "grad_norm": 0.9160334935675234,
      "learning_rate": 5.74131823855921e-06,
      "loss": 0.4113,
      "step": 7488
    },
    {
      "epoch": 0.46909597707449224,
      "grad_norm": 0.9119441198637808,
      "learning_rate": 5.7403150359138074e-06,
      "loss": 0.4003,
      "step": 7489
    },
    {
      "epoch": 0.46915861507383455,
      "grad_norm": 0.9377558804830229,
      "learning_rate": 5.7393118027968976e-06,
      "loss": 0.4155,
      "step": 7490
    },
    {
      "epoch": 0.46922125307317686,
      "grad_norm": 0.8355307534838727,
      "learning_rate": 5.7383085392497785e-06,
      "loss": 0.451,
      "step": 7491
    },
    {
      "epoch": 0.46928389107251917,
      "grad_norm": 0.8364585945702798,
      "learning_rate": 5.737305245313742e-06,
      "loss": 0.4018,
      "step": 7492
    },
    {
      "epoch": 0.4693465290718614,
      "grad_norm": 0.819384983642569,
      "learning_rate": 5.736301921030087e-06,
      "loss": 0.3758,
      "step": 7493
    },
    {
      "epoch": 0.4694091670712037,
      "grad_norm": 0.8448169932378978,
      "learning_rate": 5.735298566440105e-06,
      "loss": 0.4124,
      "step": 7494
    },
    {
      "epoch": 0.46947180507054603,
      "grad_norm": 0.7775313727372252,
      "learning_rate": 5.734295181585099e-06,
      "loss": 0.3773,
      "step": 7495
    },
    {
      "epoch": 0.46953444306988834,
      "grad_norm": 0.7880911846453217,
      "learning_rate": 5.733291766506368e-06,
      "loss": 0.3852,
      "step": 7496
    },
    {
      "epoch": 0.46959708106923065,
      "grad_norm": 0.8512338765585014,
      "learning_rate": 5.7322883212452106e-06,
      "loss": 0.3955,
      "step": 7497
    },
    {
      "epoch": 0.46965971906857296,
      "grad_norm": 0.9125343420732516,
      "learning_rate": 5.7312848458429296e-06,
      "loss": 0.3955,
      "step": 7498
    },
    {
      "epoch": 0.46972235706791526,
      "grad_norm": 0.6327422328500356,
      "learning_rate": 5.73028134034083e-06,
      "loss": 0.4724,
      "step": 7499
    },
    {
      "epoch": 0.46978499506725757,
      "grad_norm": 0.906407976798345,
      "learning_rate": 5.729277804780215e-06,
      "loss": 0.378,
      "step": 7500
    },
    {
      "epoch": 0.4698476330665999,
      "grad_norm": 0.8241198188226416,
      "learning_rate": 5.728274239202389e-06,
      "loss": 0.4372,
      "step": 7501
    },
    {
      "epoch": 0.46991027106594213,
      "grad_norm": 0.811804059219454,
      "learning_rate": 5.72727064364866e-06,
      "loss": 0.4073,
      "step": 7502
    },
    {
      "epoch": 0.46997290906528444,
      "grad_norm": 0.8897934545918266,
      "learning_rate": 5.726267018160337e-06,
      "loss": 0.4208,
      "step": 7503
    },
    {
      "epoch": 0.47003554706462675,
      "grad_norm": 0.899159149927829,
      "learning_rate": 5.725263362778728e-06,
      "loss": 0.4318,
      "step": 7504
    },
    {
      "epoch": 0.47009818506396905,
      "grad_norm": 0.8394961449799604,
      "learning_rate": 5.724259677545145e-06,
      "loss": 0.4229,
      "step": 7505
    },
    {
      "epoch": 0.47016082306331136,
      "grad_norm": 0.8981149463316586,
      "learning_rate": 5.723255962500899e-06,
      "loss": 0.3948,
      "step": 7506
    },
    {
      "epoch": 0.47022346106265367,
      "grad_norm": 0.898192789149829,
      "learning_rate": 5.722252217687303e-06,
      "loss": 0.3744,
      "step": 7507
    },
    {
      "epoch": 0.470286099061996,
      "grad_norm": 0.8155954022467082,
      "learning_rate": 5.721248443145673e-06,
      "loss": 0.437,
      "step": 7508
    },
    {
      "epoch": 0.4703487370613383,
      "grad_norm": 0.8230678734743162,
      "learning_rate": 5.7202446389173225e-06,
      "loss": 0.3768,
      "step": 7509
    },
    {
      "epoch": 0.47041137506068054,
      "grad_norm": 0.8072209393126867,
      "learning_rate": 5.719240805043569e-06,
      "loss": 0.4374,
      "step": 7510
    },
    {
      "epoch": 0.47047401306002284,
      "grad_norm": 0.8146648189854523,
      "learning_rate": 5.718236941565729e-06,
      "loss": 0.3973,
      "step": 7511
    },
    {
      "epoch": 0.47053665105936515,
      "grad_norm": 0.8122896378850457,
      "learning_rate": 5.7172330485251245e-06,
      "loss": 0.388,
      "step": 7512
    },
    {
      "epoch": 0.47059928905870746,
      "grad_norm": 0.8950950075091879,
      "learning_rate": 5.716229125963076e-06,
      "loss": 0.4308,
      "step": 7513
    },
    {
      "epoch": 0.47066192705804977,
      "grad_norm": 0.7935412054329256,
      "learning_rate": 5.7152251739209015e-06,
      "loss": 0.3916,
      "step": 7514
    },
    {
      "epoch": 0.4707245650573921,
      "grad_norm": 0.8525931586764125,
      "learning_rate": 5.714221192439927e-06,
      "loss": 0.4076,
      "step": 7515
    },
    {
      "epoch": 0.4707872030567344,
      "grad_norm": 0.75862719682992,
      "learning_rate": 5.7132171815614754e-06,
      "loss": 0.3347,
      "step": 7516
    },
    {
      "epoch": 0.4708498410560767,
      "grad_norm": 0.7835661373353561,
      "learning_rate": 5.712213141326872e-06,
      "loss": 0.3805,
      "step": 7517
    },
    {
      "epoch": 0.470912479055419,
      "grad_norm": 0.8160548633562025,
      "learning_rate": 5.711209071777445e-06,
      "loss": 0.3463,
      "step": 7518
    },
    {
      "epoch": 0.47097511705476125,
      "grad_norm": 0.7801286092513658,
      "learning_rate": 5.7102049729545195e-06,
      "loss": 0.3889,
      "step": 7519
    },
    {
      "epoch": 0.47103775505410356,
      "grad_norm": 0.7895926758548764,
      "learning_rate": 5.709200844899424e-06,
      "loss": 0.3574,
      "step": 7520
    },
    {
      "epoch": 0.47110039305344587,
      "grad_norm": 0.7479031594847513,
      "learning_rate": 5.708196687653492e-06,
      "loss": 0.3804,
      "step": 7521
    },
    {
      "epoch": 0.4711630310527882,
      "grad_norm": 0.8071072065774366,
      "learning_rate": 5.707192501258053e-06,
      "loss": 0.4144,
      "step": 7522
    },
    {
      "epoch": 0.4712256690521305,
      "grad_norm": 0.91543141006196,
      "learning_rate": 5.706188285754439e-06,
      "loss": 0.4232,
      "step": 7523
    },
    {
      "epoch": 0.4712883070514728,
      "grad_norm": 0.6849949104183701,
      "learning_rate": 5.705184041183984e-06,
      "loss": 0.4593,
      "step": 7524
    },
    {
      "epoch": 0.4713509450508151,
      "grad_norm": 0.875696117931053,
      "learning_rate": 5.7041797675880225e-06,
      "loss": 0.4297,
      "step": 7525
    },
    {
      "epoch": 0.4714135830501574,
      "grad_norm": 0.8862247606051495,
      "learning_rate": 5.703175465007892e-06,
      "loss": 0.4248,
      "step": 7526
    },
    {
      "epoch": 0.47147622104949966,
      "grad_norm": 0.8562235147691412,
      "learning_rate": 5.7021711334849285e-06,
      "loss": 0.4397,
      "step": 7527
    },
    {
      "epoch": 0.47153885904884196,
      "grad_norm": 0.8410575973278718,
      "learning_rate": 5.70116677306047e-06,
      "loss": 0.407,
      "step": 7528
    },
    {
      "epoch": 0.47160149704818427,
      "grad_norm": 0.8679054297821136,
      "learning_rate": 5.700162383775858e-06,
      "loss": 0.4018,
      "step": 7529
    },
    {
      "epoch": 0.4716641350475266,
      "grad_norm": 0.8696733435426353,
      "learning_rate": 5.699157965672432e-06,
      "loss": 0.4731,
      "step": 7530
    },
    {
      "epoch": 0.4717267730468689,
      "grad_norm": 0.8766392023596524,
      "learning_rate": 5.698153518791535e-06,
      "loss": 0.4305,
      "step": 7531
    },
    {
      "epoch": 0.4717894110462112,
      "grad_norm": 0.8483176937732925,
      "learning_rate": 5.697149043174508e-06,
      "loss": 0.4493,
      "step": 7532
    },
    {
      "epoch": 0.4718520490455535,
      "grad_norm": 0.8564556818093039,
      "learning_rate": 5.696144538862698e-06,
      "loss": 0.3984,
      "step": 7533
    },
    {
      "epoch": 0.4719146870448958,
      "grad_norm": 0.7983366355274739,
      "learning_rate": 5.695140005897449e-06,
      "loss": 0.4258,
      "step": 7534
    },
    {
      "epoch": 0.47197732504423806,
      "grad_norm": 0.9172386835881903,
      "learning_rate": 5.6941354443201095e-06,
      "loss": 0.398,
      "step": 7535
    },
    {
      "epoch": 0.47203996304358037,
      "grad_norm": 0.8151958987314794,
      "learning_rate": 5.6931308541720265e-06,
      "loss": 0.391,
      "step": 7536
    },
    {
      "epoch": 0.4721026010429227,
      "grad_norm": 0.8257668977189373,
      "learning_rate": 5.6921262354945475e-06,
      "loss": 0.387,
      "step": 7537
    },
    {
      "epoch": 0.472165239042265,
      "grad_norm": 0.7684606049281606,
      "learning_rate": 5.691121588329024e-06,
      "loss": 0.3636,
      "step": 7538
    },
    {
      "epoch": 0.4722278770416073,
      "grad_norm": 0.791690016480046,
      "learning_rate": 5.690116912716809e-06,
      "loss": 0.372,
      "step": 7539
    },
    {
      "epoch": 0.4722905150409496,
      "grad_norm": 0.9609689786479673,
      "learning_rate": 5.6891122086992525e-06,
      "loss": 0.4489,
      "step": 7540
    },
    {
      "epoch": 0.4723531530402919,
      "grad_norm": 0.8556145462687915,
      "learning_rate": 5.6881074763177104e-06,
      "loss": 0.4297,
      "step": 7541
    },
    {
      "epoch": 0.4724157910396342,
      "grad_norm": 0.8822301109666584,
      "learning_rate": 5.687102715613537e-06,
      "loss": 0.4157,
      "step": 7542
    },
    {
      "epoch": 0.4724784290389765,
      "grad_norm": 0.8012563768166705,
      "learning_rate": 5.686097926628087e-06,
      "loss": 0.4115,
      "step": 7543
    },
    {
      "epoch": 0.4725410670383188,
      "grad_norm": 0.8924389646451273,
      "learning_rate": 5.6850931094027204e-06,
      "loss": 0.3941,
      "step": 7544
    },
    {
      "epoch": 0.4726037050376611,
      "grad_norm": 0.7918542178952299,
      "learning_rate": 5.684088263978793e-06,
      "loss": 0.3849,
      "step": 7545
    },
    {
      "epoch": 0.4726663430370034,
      "grad_norm": 0.8173045897413244,
      "learning_rate": 5.683083390397667e-06,
      "loss": 0.4183,
      "step": 7546
    },
    {
      "epoch": 0.4727289810363457,
      "grad_norm": 0.7950459945187979,
      "learning_rate": 5.682078488700702e-06,
      "loss": 0.3688,
      "step": 7547
    },
    {
      "epoch": 0.472791619035688,
      "grad_norm": 0.8142634949909937,
      "learning_rate": 5.681073558929258e-06,
      "loss": 0.3933,
      "step": 7548
    },
    {
      "epoch": 0.4728542570350303,
      "grad_norm": 0.8449422010570827,
      "learning_rate": 5.680068601124701e-06,
      "loss": 0.4119,
      "step": 7549
    },
    {
      "epoch": 0.4729168950343726,
      "grad_norm": 0.7934776257721305,
      "learning_rate": 5.679063615328394e-06,
      "loss": 0.3866,
      "step": 7550
    },
    {
      "epoch": 0.4729795330337149,
      "grad_norm": 0.7686498442301233,
      "learning_rate": 5.678058601581704e-06,
      "loss": 0.3815,
      "step": 7551
    },
    {
      "epoch": 0.4730421710330572,
      "grad_norm": 0.9110333104074891,
      "learning_rate": 5.677053559925993e-06,
      "loss": 0.3907,
      "step": 7552
    },
    {
      "epoch": 0.4731048090323995,
      "grad_norm": 0.8094773296327528,
      "learning_rate": 5.6760484904026344e-06,
      "loss": 0.3812,
      "step": 7553
    },
    {
      "epoch": 0.4731674470317418,
      "grad_norm": 0.8052825477627191,
      "learning_rate": 5.675043393052994e-06,
      "loss": 0.3922,
      "step": 7554
    },
    {
      "epoch": 0.4732300850310841,
      "grad_norm": 0.8216854815607809,
      "learning_rate": 5.674038267918442e-06,
      "loss": 0.396,
      "step": 7555
    },
    {
      "epoch": 0.4732927230304264,
      "grad_norm": 0.8188911662781712,
      "learning_rate": 5.673033115040351e-06,
      "loss": 0.4064,
      "step": 7556
    },
    {
      "epoch": 0.4733553610297687,
      "grad_norm": 0.8872547960112821,
      "learning_rate": 5.67202793446009e-06,
      "loss": 0.4235,
      "step": 7557
    },
    {
      "epoch": 0.473417999029111,
      "grad_norm": 0.8772598761928995,
      "learning_rate": 5.671022726219034e-06,
      "loss": 0.4657,
      "step": 7558
    },
    {
      "epoch": 0.47348063702845333,
      "grad_norm": 0.8534457384386614,
      "learning_rate": 5.6700174903585605e-06,
      "loss": 0.402,
      "step": 7559
    },
    {
      "epoch": 0.4735432750277956,
      "grad_norm": 0.8839603751858535,
      "learning_rate": 5.669012226920041e-06,
      "loss": 0.4328,
      "step": 7560
    },
    {
      "epoch": 0.4736059130271379,
      "grad_norm": 0.8545964677893084,
      "learning_rate": 5.668006935944854e-06,
      "loss": 0.3814,
      "step": 7561
    },
    {
      "epoch": 0.4736685510264802,
      "grad_norm": 0.6195314373629338,
      "learning_rate": 5.667001617474378e-06,
      "loss": 0.473,
      "step": 7562
    },
    {
      "epoch": 0.4737311890258225,
      "grad_norm": 0.7966906338193845,
      "learning_rate": 5.665996271549991e-06,
      "loss": 0.4179,
      "step": 7563
    },
    {
      "epoch": 0.4737938270251648,
      "grad_norm": 0.8056807611323871,
      "learning_rate": 5.664990898213076e-06,
      "loss": 0.385,
      "step": 7564
    },
    {
      "epoch": 0.4738564650245071,
      "grad_norm": 0.8310330384940774,
      "learning_rate": 5.663985497505011e-06,
      "loss": 0.411,
      "step": 7565
    },
    {
      "epoch": 0.47391910302384943,
      "grad_norm": 0.8909645971261946,
      "learning_rate": 5.66298006946718e-06,
      "loss": 0.4009,
      "step": 7566
    },
    {
      "epoch": 0.47398174102319174,
      "grad_norm": 0.8633020684638129,
      "learning_rate": 5.661974614140965e-06,
      "loss": 0.4062,
      "step": 7567
    },
    {
      "epoch": 0.47404437902253405,
      "grad_norm": 0.8225520794666752,
      "learning_rate": 5.660969131567753e-06,
      "loss": 0.427,
      "step": 7568
    },
    {
      "epoch": 0.4741070170218763,
      "grad_norm": 0.8741019606697472,
      "learning_rate": 5.659963621788928e-06,
      "loss": 0.373,
      "step": 7569
    },
    {
      "epoch": 0.4741696550212186,
      "grad_norm": 0.8318253372718167,
      "learning_rate": 5.658958084845879e-06,
      "loss": 0.3885,
      "step": 7570
    },
    {
      "epoch": 0.4742322930205609,
      "grad_norm": 0.8097280715198402,
      "learning_rate": 5.657952520779992e-06,
      "loss": 0.4008,
      "step": 7571
    },
    {
      "epoch": 0.4742949310199032,
      "grad_norm": 0.7737064245607224,
      "learning_rate": 5.656946929632658e-06,
      "loss": 0.3795,
      "step": 7572
    },
    {
      "epoch": 0.47435756901924553,
      "grad_norm": 0.7882066462921429,
      "learning_rate": 5.6559413114452665e-06,
      "loss": 0.3554,
      "step": 7573
    },
    {
      "epoch": 0.47442020701858784,
      "grad_norm": 0.8517283428448745,
      "learning_rate": 5.654935666259209e-06,
      "loss": 0.3547,
      "step": 7574
    },
    {
      "epoch": 0.47448284501793014,
      "grad_norm": 0.8158843311108529,
      "learning_rate": 5.653929994115875e-06,
      "loss": 0.3878,
      "step": 7575
    },
    {
      "epoch": 0.47454548301727245,
      "grad_norm": 0.7894752226970219,
      "learning_rate": 5.652924295056664e-06,
      "loss": 0.4098,
      "step": 7576
    },
    {
      "epoch": 0.4746081210166147,
      "grad_norm": 0.827304195595095,
      "learning_rate": 5.651918569122967e-06,
      "loss": 0.4061,
      "step": 7577
    },
    {
      "epoch": 0.474670759015957,
      "grad_norm": 0.7659275162248118,
      "learning_rate": 5.650912816356182e-06,
      "loss": 0.3759,
      "step": 7578
    },
    {
      "epoch": 0.4747333970152993,
      "grad_norm": 0.8697249878426011,
      "learning_rate": 5.649907036797703e-06,
      "loss": 0.4255,
      "step": 7579
    },
    {
      "epoch": 0.4747960350146416,
      "grad_norm": 0.9329661310398205,
      "learning_rate": 5.648901230488929e-06,
      "loss": 0.4258,
      "step": 7580
    },
    {
      "epoch": 0.47485867301398393,
      "grad_norm": 0.9028582736129028,
      "learning_rate": 5.647895397471263e-06,
      "loss": 0.4095,
      "step": 7581
    },
    {
      "epoch": 0.47492131101332624,
      "grad_norm": 0.9338415967227579,
      "learning_rate": 5.6468895377861e-06,
      "loss": 0.3811,
      "step": 7582
    },
    {
      "epoch": 0.47498394901266855,
      "grad_norm": 0.7341528663687866,
      "learning_rate": 5.645883651474844e-06,
      "loss": 0.3691,
      "step": 7583
    },
    {
      "epoch": 0.47504658701201086,
      "grad_norm": 0.7851634215338611,
      "learning_rate": 5.644877738578896e-06,
      "loss": 0.3787,
      "step": 7584
    },
    {
      "epoch": 0.47510922501135316,
      "grad_norm": 0.8287646498025428,
      "learning_rate": 5.6438717991396615e-06,
      "loss": 0.3841,
      "step": 7585
    },
    {
      "epoch": 0.4751718630106954,
      "grad_norm": 0.8303330026498296,
      "learning_rate": 5.642865833198545e-06,
      "loss": 0.3987,
      "step": 7586
    },
    {
      "epoch": 0.4752345010100377,
      "grad_norm": 0.8264376203526039,
      "learning_rate": 5.641859840796951e-06,
      "loss": 0.3959,
      "step": 7587
    },
    {
      "epoch": 0.47529713900938003,
      "grad_norm": 0.820658604007337,
      "learning_rate": 5.6408538219762865e-06,
      "loss": 0.4103,
      "step": 7588
    },
    {
      "epoch": 0.47535977700872234,
      "grad_norm": 0.8246053308880228,
      "learning_rate": 5.639847776777958e-06,
      "loss": 0.416,
      "step": 7589
    },
    {
      "epoch": 0.47542241500806465,
      "grad_norm": 0.9077343726790641,
      "learning_rate": 5.638841705243379e-06,
      "loss": 0.4341,
      "step": 7590
    },
    {
      "epoch": 0.47548505300740695,
      "grad_norm": 0.9027597957524827,
      "learning_rate": 5.637835607413957e-06,
      "loss": 0.4344,
      "step": 7591
    },
    {
      "epoch": 0.47554769100674926,
      "grad_norm": 0.7971165239489943,
      "learning_rate": 5.636829483331101e-06,
      "loss": 0.4118,
      "step": 7592
    },
    {
      "epoch": 0.47561032900609157,
      "grad_norm": 0.843950122759513,
      "learning_rate": 5.635823333036227e-06,
      "loss": 0.4015,
      "step": 7593
    },
    {
      "epoch": 0.4756729670054338,
      "grad_norm": 0.8106151324164289,
      "learning_rate": 5.634817156570744e-06,
      "loss": 0.4323,
      "step": 7594
    },
    {
      "epoch": 0.47573560500477613,
      "grad_norm": 0.8121566911009126,
      "learning_rate": 5.633810953976072e-06,
      "loss": 0.448,
      "step": 7595
    },
    {
      "epoch": 0.47579824300411844,
      "grad_norm": 0.7802051915915843,
      "learning_rate": 5.632804725293621e-06,
      "loss": 0.4131,
      "step": 7596
    },
    {
      "epoch": 0.47586088100346075,
      "grad_norm": 0.7803232632276558,
      "learning_rate": 5.631798470564812e-06,
      "loss": 0.3679,
      "step": 7597
    },
    {
      "epoch": 0.47592351900280305,
      "grad_norm": 0.662027205524899,
      "learning_rate": 5.6307921898310595e-06,
      "loss": 0.4489,
      "step": 7598
    },
    {
      "epoch": 0.47598615700214536,
      "grad_norm": 0.803487267323282,
      "learning_rate": 5.629785883133785e-06,
      "loss": 0.4448,
      "step": 7599
    },
    {
      "epoch": 0.47604879500148767,
      "grad_norm": 0.8346573053176894,
      "learning_rate": 5.628779550514405e-06,
      "loss": 0.4034,
      "step": 7600
    },
    {
      "epoch": 0.47611143300083,
      "grad_norm": 0.8397650014527799,
      "learning_rate": 5.627773192014342e-06,
      "loss": 0.43,
      "step": 7601
    },
    {
      "epoch": 0.4761740710001722,
      "grad_norm": 0.8917604172752833,
      "learning_rate": 5.62676680767502e-06,
      "loss": 0.4379,
      "step": 7602
    },
    {
      "epoch": 0.47623670899951454,
      "grad_norm": 0.8582677359309394,
      "learning_rate": 5.625760397537857e-06,
      "loss": 0.4362,
      "step": 7603
    },
    {
      "epoch": 0.47629934699885684,
      "grad_norm": 0.8195025045401498,
      "learning_rate": 5.624753961644281e-06,
      "loss": 0.4306,
      "step": 7604
    },
    {
      "epoch": 0.47636198499819915,
      "grad_norm": 0.8505395670403835,
      "learning_rate": 5.623747500035717e-06,
      "loss": 0.4041,
      "step": 7605
    },
    {
      "epoch": 0.47642462299754146,
      "grad_norm": 0.8809195864917726,
      "learning_rate": 5.622741012753589e-06,
      "loss": 0.441,
      "step": 7606
    },
    {
      "epoch": 0.47648726099688377,
      "grad_norm": 0.7837980425967587,
      "learning_rate": 5.621734499839326e-06,
      "loss": 0.3687,
      "step": 7607
    },
    {
      "epoch": 0.4765498989962261,
      "grad_norm": 0.8453981632239935,
      "learning_rate": 5.620727961334355e-06,
      "loss": 0.4298,
      "step": 7608
    },
    {
      "epoch": 0.4766125369955684,
      "grad_norm": 0.9135642107234784,
      "learning_rate": 5.619721397280105e-06,
      "loss": 0.3978,
      "step": 7609
    },
    {
      "epoch": 0.4766751749949107,
      "grad_norm": 0.8405069345576077,
      "learning_rate": 5.618714807718009e-06,
      "loss": 0.4375,
      "step": 7610
    },
    {
      "epoch": 0.47673781299425294,
      "grad_norm": 0.8470291899186385,
      "learning_rate": 5.617708192689496e-06,
      "loss": 0.4042,
      "step": 7611
    },
    {
      "epoch": 0.47680045099359525,
      "grad_norm": 0.8029504471322804,
      "learning_rate": 5.616701552235999e-06,
      "loss": 0.4184,
      "step": 7612
    },
    {
      "epoch": 0.47686308899293756,
      "grad_norm": 0.7986567242127149,
      "learning_rate": 5.615694886398952e-06,
      "loss": 0.4182,
      "step": 7613
    },
    {
      "epoch": 0.47692572699227986,
      "grad_norm": 0.8107209264230264,
      "learning_rate": 5.614688195219788e-06,
      "loss": 0.3712,
      "step": 7614
    },
    {
      "epoch": 0.47698836499162217,
      "grad_norm": 0.8407405825590355,
      "learning_rate": 5.613681478739946e-06,
      "loss": 0.4401,
      "step": 7615
    },
    {
      "epoch": 0.4770510029909645,
      "grad_norm": 0.8835744396932966,
      "learning_rate": 5.6126747370008595e-06,
      "loss": 0.4396,
      "step": 7616
    },
    {
      "epoch": 0.4771136409903068,
      "grad_norm": 0.7991894387459769,
      "learning_rate": 5.6116679700439666e-06,
      "loss": 0.3913,
      "step": 7617
    },
    {
      "epoch": 0.4771762789896491,
      "grad_norm": 0.8631808060806841,
      "learning_rate": 5.610661177910706e-06,
      "loss": 0.3904,
      "step": 7618
    },
    {
      "epoch": 0.47723891698899135,
      "grad_norm": 0.8245094001659242,
      "learning_rate": 5.609654360642519e-06,
      "loss": 0.3914,
      "step": 7619
    },
    {
      "epoch": 0.47730155498833365,
      "grad_norm": 0.8890889612164937,
      "learning_rate": 5.608647518280845e-06,
      "loss": 0.4112,
      "step": 7620
    },
    {
      "epoch": 0.47736419298767596,
      "grad_norm": 0.8342983983923028,
      "learning_rate": 5.6076406508671264e-06,
      "loss": 0.4351,
      "step": 7621
    },
    {
      "epoch": 0.47742683098701827,
      "grad_norm": 0.7801816803049687,
      "learning_rate": 5.606633758442806e-06,
      "loss": 0.35,
      "step": 7622
    },
    {
      "epoch": 0.4774894689863606,
      "grad_norm": 0.8357731452072638,
      "learning_rate": 5.6056268410493266e-06,
      "loss": 0.4387,
      "step": 7623
    },
    {
      "epoch": 0.4775521069857029,
      "grad_norm": 0.8030100277080913,
      "learning_rate": 5.604619898728135e-06,
      "loss": 0.3992,
      "step": 7624
    },
    {
      "epoch": 0.4776147449850452,
      "grad_norm": 0.8469189887353809,
      "learning_rate": 5.603612931520677e-06,
      "loss": 0.4347,
      "step": 7625
    },
    {
      "epoch": 0.4776773829843875,
      "grad_norm": 0.8067143753730599,
      "learning_rate": 5.6026059394683965e-06,
      "loss": 0.3873,
      "step": 7626
    },
    {
      "epoch": 0.4777400209837298,
      "grad_norm": 0.8098281192212632,
      "learning_rate": 5.6015989226127445e-06,
      "loss": 0.4084,
      "step": 7627
    },
    {
      "epoch": 0.47780265898307206,
      "grad_norm": 0.8378749366191879,
      "learning_rate": 5.60059188099517e-06,
      "loss": 0.3934,
      "step": 7628
    },
    {
      "epoch": 0.47786529698241437,
      "grad_norm": 0.642917724338331,
      "learning_rate": 5.599584814657122e-06,
      "loss": 0.4817,
      "step": 7629
    },
    {
      "epoch": 0.4779279349817567,
      "grad_norm": 0.8116202767657149,
      "learning_rate": 5.598577723640049e-06,
      "loss": 0.4205,
      "step": 7630
    },
    {
      "epoch": 0.477990572981099,
      "grad_norm": 0.8951455399036353,
      "learning_rate": 5.597570607985409e-06,
      "loss": 0.3894,
      "step": 7631
    },
    {
      "epoch": 0.4780532109804413,
      "grad_norm": 0.83164507103376,
      "learning_rate": 5.59656346773465e-06,
      "loss": 0.404,
      "step": 7632
    },
    {
      "epoch": 0.4781158489797836,
      "grad_norm": 0.8185849349621243,
      "learning_rate": 5.59555630292923e-06,
      "loss": 0.396,
      "step": 7633
    },
    {
      "epoch": 0.4781784869791259,
      "grad_norm": 0.794426227250999,
      "learning_rate": 5.5945491136106e-06,
      "loss": 0.4148,
      "step": 7634
    },
    {
      "epoch": 0.4782411249784682,
      "grad_norm": 0.8938009813760087,
      "learning_rate": 5.593541899820219e-06,
      "loss": 0.421,
      "step": 7635
    },
    {
      "epoch": 0.47830376297781046,
      "grad_norm": 0.837935971043074,
      "learning_rate": 5.592534661599543e-06,
      "loss": 0.4207,
      "step": 7636
    },
    {
      "epoch": 0.4783664009771528,
      "grad_norm": 0.8638912020747327,
      "learning_rate": 5.591527398990032e-06,
      "loss": 0.4657,
      "step": 7637
    },
    {
      "epoch": 0.4784290389764951,
      "grad_norm": 0.7923710054943357,
      "learning_rate": 5.590520112033142e-06,
      "loss": 0.3879,
      "step": 7638
    },
    {
      "epoch": 0.4784916769758374,
      "grad_norm": 0.8916034751061275,
      "learning_rate": 5.589512800770334e-06,
      "loss": 0.4631,
      "step": 7639
    },
    {
      "epoch": 0.4785543149751797,
      "grad_norm": 0.8012527663750874,
      "learning_rate": 5.58850546524307e-06,
      "loss": 0.3733,
      "step": 7640
    },
    {
      "epoch": 0.478616952974522,
      "grad_norm": 0.8182043797910192,
      "learning_rate": 5.587498105492813e-06,
      "loss": 0.3911,
      "step": 7641
    },
    {
      "epoch": 0.4786795909738643,
      "grad_norm": 0.8119061202460586,
      "learning_rate": 5.586490721561024e-06,
      "loss": 0.4035,
      "step": 7642
    },
    {
      "epoch": 0.4787422289732066,
      "grad_norm": 0.7685132372805785,
      "learning_rate": 5.585483313489168e-06,
      "loss": 0.3705,
      "step": 7643
    },
    {
      "epoch": 0.47880486697254887,
      "grad_norm": 0.7803341894600284,
      "learning_rate": 5.584475881318709e-06,
      "loss": 0.3803,
      "step": 7644
    },
    {
      "epoch": 0.4788675049718912,
      "grad_norm": 0.7899273171362987,
      "learning_rate": 5.583468425091116e-06,
      "loss": 0.3944,
      "step": 7645
    },
    {
      "epoch": 0.4789301429712335,
      "grad_norm": 0.805471654194722,
      "learning_rate": 5.582460944847855e-06,
      "loss": 0.4034,
      "step": 7646
    },
    {
      "epoch": 0.4789927809705758,
      "grad_norm": 0.8766828160956219,
      "learning_rate": 5.581453440630392e-06,
      "loss": 0.4074,
      "step": 7647
    },
    {
      "epoch": 0.4790554189699181,
      "grad_norm": 0.7866047499689405,
      "learning_rate": 5.580445912480198e-06,
      "loss": 0.3717,
      "step": 7648
    },
    {
      "epoch": 0.4791180569692604,
      "grad_norm": 0.8433331624788399,
      "learning_rate": 5.579438360438742e-06,
      "loss": 0.405,
      "step": 7649
    },
    {
      "epoch": 0.4791806949686027,
      "grad_norm": 0.7931284466553121,
      "learning_rate": 5.578430784547496e-06,
      "loss": 0.3847,
      "step": 7650
    },
    {
      "epoch": 0.479243332967945,
      "grad_norm": 0.7784468269157583,
      "learning_rate": 5.577423184847932e-06,
      "loss": 0.4053,
      "step": 7651
    },
    {
      "epoch": 0.47930597096728733,
      "grad_norm": 0.8703266302868496,
      "learning_rate": 5.576415561381521e-06,
      "loss": 0.4004,
      "step": 7652
    },
    {
      "epoch": 0.4793686089666296,
      "grad_norm": 0.8173945588450079,
      "learning_rate": 5.57540791418974e-06,
      "loss": 0.3775,
      "step": 7653
    },
    {
      "epoch": 0.4794312469659719,
      "grad_norm": 0.9354349636145959,
      "learning_rate": 5.574400243314061e-06,
      "loss": 0.3936,
      "step": 7654
    },
    {
      "epoch": 0.4794938849653142,
      "grad_norm": 0.8234484683406883,
      "learning_rate": 5.573392548795963e-06,
      "loss": 0.4031,
      "step": 7655
    },
    {
      "epoch": 0.4795565229646565,
      "grad_norm": 0.8264852295127677,
      "learning_rate": 5.572384830676919e-06,
      "loss": 0.4072,
      "step": 7656
    },
    {
      "epoch": 0.4796191609639988,
      "grad_norm": 0.8801452114068201,
      "learning_rate": 5.571377088998412e-06,
      "loss": 0.4512,
      "step": 7657
    },
    {
      "epoch": 0.4796817989633411,
      "grad_norm": 0.8609611121744891,
      "learning_rate": 5.570369323801916e-06,
      "loss": 0.4317,
      "step": 7658
    },
    {
      "epoch": 0.47974443696268343,
      "grad_norm": 0.7926132259919715,
      "learning_rate": 5.569361535128913e-06,
      "loss": 0.3932,
      "step": 7659
    },
    {
      "epoch": 0.47980707496202574,
      "grad_norm": 0.7600893785937515,
      "learning_rate": 5.568353723020884e-06,
      "loss": 0.3454,
      "step": 7660
    },
    {
      "epoch": 0.479869712961368,
      "grad_norm": 0.8108845619341065,
      "learning_rate": 5.56734588751931e-06,
      "loss": 0.4065,
      "step": 7661
    },
    {
      "epoch": 0.4799323509607103,
      "grad_norm": 0.8608520300404868,
      "learning_rate": 5.5663380286656746e-06,
      "loss": 0.4613,
      "step": 7662
    },
    {
      "epoch": 0.4799949889600526,
      "grad_norm": 0.6601608866053466,
      "learning_rate": 5.565330146501459e-06,
      "loss": 0.4891,
      "step": 7663
    },
    {
      "epoch": 0.4800576269593949,
      "grad_norm": 0.7878088645701828,
      "learning_rate": 5.564322241068152e-06,
      "loss": 0.4144,
      "step": 7664
    },
    {
      "epoch": 0.4801202649587372,
      "grad_norm": 0.8105628721679198,
      "learning_rate": 5.563314312407236e-06,
      "loss": 0.4196,
      "step": 7665
    },
    {
      "epoch": 0.4801829029580795,
      "grad_norm": 0.848317468588712,
      "learning_rate": 5.562306360560198e-06,
      "loss": 0.3931,
      "step": 7666
    },
    {
      "epoch": 0.48024554095742183,
      "grad_norm": 0.8420553341128426,
      "learning_rate": 5.561298385568525e-06,
      "loss": 0.3957,
      "step": 7667
    },
    {
      "epoch": 0.48030817895676414,
      "grad_norm": 0.8198588701302505,
      "learning_rate": 5.560290387473708e-06,
      "loss": 0.3694,
      "step": 7668
    },
    {
      "epoch": 0.4803708169561064,
      "grad_norm": 0.7551902715437862,
      "learning_rate": 5.559282366317233e-06,
      "loss": 0.3527,
      "step": 7669
    },
    {
      "epoch": 0.4804334549554487,
      "grad_norm": 0.9122172897805222,
      "learning_rate": 5.5582743221405935e-06,
      "loss": 0.3966,
      "step": 7670
    },
    {
      "epoch": 0.480496092954791,
      "grad_norm": 0.88421890411236,
      "learning_rate": 5.55726625498528e-06,
      "loss": 0.4781,
      "step": 7671
    },
    {
      "epoch": 0.4805587309541333,
      "grad_norm": 0.9186521590835878,
      "learning_rate": 5.556258164892782e-06,
      "loss": 0.4359,
      "step": 7672
    },
    {
      "epoch": 0.4806213689534756,
      "grad_norm": 0.8618144556184048,
      "learning_rate": 5.555250051904594e-06,
      "loss": 0.387,
      "step": 7673
    },
    {
      "epoch": 0.48068400695281793,
      "grad_norm": 0.8073749255586788,
      "learning_rate": 5.554241916062213e-06,
      "loss": 0.3922,
      "step": 7674
    },
    {
      "epoch": 0.48074664495216024,
      "grad_norm": 0.8540921389679489,
      "learning_rate": 5.5532337574071315e-06,
      "loss": 0.4064,
      "step": 7675
    },
    {
      "epoch": 0.48080928295150255,
      "grad_norm": 0.7812973092311535,
      "learning_rate": 5.552225575980846e-06,
      "loss": 0.3763,
      "step": 7676
    },
    {
      "epoch": 0.48087192095084486,
      "grad_norm": 0.8425771607194925,
      "learning_rate": 5.551217371824852e-06,
      "loss": 0.4055,
      "step": 7677
    },
    {
      "epoch": 0.4809345589501871,
      "grad_norm": 0.8051373973061652,
      "learning_rate": 5.550209144980648e-06,
      "loss": 0.3755,
      "step": 7678
    },
    {
      "epoch": 0.4809971969495294,
      "grad_norm": 0.8709336728191635,
      "learning_rate": 5.5492008954897355e-06,
      "loss": 0.4167,
      "step": 7679
    },
    {
      "epoch": 0.4810598349488717,
      "grad_norm": 0.8765766789929856,
      "learning_rate": 5.5481926233936125e-06,
      "loss": 0.413,
      "step": 7680
    },
    {
      "epoch": 0.48112247294821403,
      "grad_norm": 0.8581232946522142,
      "learning_rate": 5.547184328733777e-06,
      "loss": 0.4002,
      "step": 7681
    },
    {
      "epoch": 0.48118511094755634,
      "grad_norm": 0.8719221086716762,
      "learning_rate": 5.546176011551733e-06,
      "loss": 0.4029,
      "step": 7682
    },
    {
      "epoch": 0.48124774894689865,
      "grad_norm": 0.7793072006804277,
      "learning_rate": 5.545167671888986e-06,
      "loss": 0.4353,
      "step": 7683
    },
    {
      "epoch": 0.48131038694624095,
      "grad_norm": 0.799857668761409,
      "learning_rate": 5.544159309787035e-06,
      "loss": 0.409,
      "step": 7684
    },
    {
      "epoch": 0.48137302494558326,
      "grad_norm": 0.845873913573082,
      "learning_rate": 5.543150925287386e-06,
      "loss": 0.4064,
      "step": 7685
    },
    {
      "epoch": 0.4814356629449255,
      "grad_norm": 0.7557915546130286,
      "learning_rate": 5.542142518431542e-06,
      "loss": 0.4091,
      "step": 7686
    },
    {
      "epoch": 0.4814983009442678,
      "grad_norm": 0.9184223916243539,
      "learning_rate": 5.541134089261013e-06,
      "loss": 0.4273,
      "step": 7687
    },
    {
      "epoch": 0.48156093894361013,
      "grad_norm": 0.8513861884386417,
      "learning_rate": 5.540125637817305e-06,
      "loss": 0.403,
      "step": 7688
    },
    {
      "epoch": 0.48162357694295244,
      "grad_norm": 0.6880930740303207,
      "learning_rate": 5.539117164141926e-06,
      "loss": 0.4452,
      "step": 7689
    },
    {
      "epoch": 0.48168621494229474,
      "grad_norm": 0.8254573122516012,
      "learning_rate": 5.538108668276383e-06,
      "loss": 0.4151,
      "step": 7690
    },
    {
      "epoch": 0.48174885294163705,
      "grad_norm": 0.7797556118799718,
      "learning_rate": 5.537100150262187e-06,
      "loss": 0.3848,
      "step": 7691
    },
    {
      "epoch": 0.48181149094097936,
      "grad_norm": 0.8967925255214368,
      "learning_rate": 5.53609161014085e-06,
      "loss": 0.3948,
      "step": 7692
    },
    {
      "epoch": 0.48187412894032167,
      "grad_norm": 0.8640583671431938,
      "learning_rate": 5.535083047953883e-06,
      "loss": 0.4092,
      "step": 7693
    },
    {
      "epoch": 0.481936766939664,
      "grad_norm": 0.7967631063589129,
      "learning_rate": 5.534074463742798e-06,
      "loss": 0.3866,
      "step": 7694
    },
    {
      "epoch": 0.4819994049390062,
      "grad_norm": 0.7641956087474686,
      "learning_rate": 5.533065857549108e-06,
      "loss": 0.367,
      "step": 7695
    },
    {
      "epoch": 0.48206204293834853,
      "grad_norm": 0.8963485574446745,
      "learning_rate": 5.53205722941433e-06,
      "loss": 0.3857,
      "step": 7696
    },
    {
      "epoch": 0.48212468093769084,
      "grad_norm": 0.944580011173828,
      "learning_rate": 5.5310485793799775e-06,
      "loss": 0.4253,
      "step": 7697
    },
    {
      "epoch": 0.48218731893703315,
      "grad_norm": 0.8510550187697165,
      "learning_rate": 5.530039907487568e-06,
      "loss": 0.475,
      "step": 7698
    },
    {
      "epoch": 0.48224995693637546,
      "grad_norm": 0.8616697217759945,
      "learning_rate": 5.529031213778615e-06,
      "loss": 0.4292,
      "step": 7699
    },
    {
      "epoch": 0.48231259493571776,
      "grad_norm": 0.7800443297089573,
      "learning_rate": 5.528022498294638e-06,
      "loss": 0.3888,
      "step": 7700
    },
    {
      "epoch": 0.48237523293506007,
      "grad_norm": 0.925188246454495,
      "learning_rate": 5.5270137610771605e-06,
      "loss": 0.4037,
      "step": 7701
    },
    {
      "epoch": 0.4824378709344024,
      "grad_norm": 0.8493890534670453,
      "learning_rate": 5.526005002167698e-06,
      "loss": 0.4391,
      "step": 7702
    },
    {
      "epoch": 0.48250050893374463,
      "grad_norm": 0.8728487454107166,
      "learning_rate": 5.52499622160777e-06,
      "loss": 0.4425,
      "step": 7703
    },
    {
      "epoch": 0.48256314693308694,
      "grad_norm": 0.8977319763805593,
      "learning_rate": 5.523987419438901e-06,
      "loss": 0.4259,
      "step": 7704
    },
    {
      "epoch": 0.48262578493242925,
      "grad_norm": 0.8801264628662789,
      "learning_rate": 5.522978595702614e-06,
      "loss": 0.43,
      "step": 7705
    },
    {
      "epoch": 0.48268842293177155,
      "grad_norm": 0.8819300031497571,
      "learning_rate": 5.5219697504404304e-06,
      "loss": 0.4026,
      "step": 7706
    },
    {
      "epoch": 0.48275106093111386,
      "grad_norm": 0.8044110553904317,
      "learning_rate": 5.520960883693873e-06,
      "loss": 0.3628,
      "step": 7707
    },
    {
      "epoch": 0.48281369893045617,
      "grad_norm": 0.8203354906400597,
      "learning_rate": 5.519951995504471e-06,
      "loss": 0.3608,
      "step": 7708
    },
    {
      "epoch": 0.4828763369297985,
      "grad_norm": 0.8516882801639779,
      "learning_rate": 5.518943085913747e-06,
      "loss": 0.4562,
      "step": 7709
    },
    {
      "epoch": 0.4829389749291408,
      "grad_norm": 0.929736840169864,
      "learning_rate": 5.517934154963232e-06,
      "loss": 0.4246,
      "step": 7710
    },
    {
      "epoch": 0.48300161292848304,
      "grad_norm": 0.7410728098883129,
      "learning_rate": 5.516925202694449e-06,
      "loss": 0.4502,
      "step": 7711
    },
    {
      "epoch": 0.48306425092782534,
      "grad_norm": 0.8116612019132464,
      "learning_rate": 5.515916229148927e-06,
      "loss": 0.4158,
      "step": 7712
    },
    {
      "epoch": 0.48312688892716765,
      "grad_norm": 0.8226922346445023,
      "learning_rate": 5.5149072343681986e-06,
      "loss": 0.3239,
      "step": 7713
    },
    {
      "epoch": 0.48318952692650996,
      "grad_norm": 0.7233908623822075,
      "learning_rate": 5.513898218393794e-06,
      "loss": 0.459,
      "step": 7714
    },
    {
      "epoch": 0.48325216492585227,
      "grad_norm": 0.7572135732354839,
      "learning_rate": 5.512889181267244e-06,
      "loss": 0.3749,
      "step": 7715
    },
    {
      "epoch": 0.4833148029251946,
      "grad_norm": 0.7930765173249704,
      "learning_rate": 5.511880123030078e-06,
      "loss": 0.3975,
      "step": 7716
    },
    {
      "epoch": 0.4833774409245369,
      "grad_norm": 0.8575700323284117,
      "learning_rate": 5.510871043723832e-06,
      "loss": 0.3986,
      "step": 7717
    },
    {
      "epoch": 0.4834400789238792,
      "grad_norm": 0.8368980199072655,
      "learning_rate": 5.509861943390038e-06,
      "loss": 0.4311,
      "step": 7718
    },
    {
      "epoch": 0.4835027169232215,
      "grad_norm": 0.8645323577754473,
      "learning_rate": 5.5088528220702335e-06,
      "loss": 0.4227,
      "step": 7719
    },
    {
      "epoch": 0.48356535492256375,
      "grad_norm": 0.7938727948552088,
      "learning_rate": 5.507843679805951e-06,
      "loss": 0.4053,
      "step": 7720
    },
    {
      "epoch": 0.48362799292190606,
      "grad_norm": 0.7922668101238655,
      "learning_rate": 5.5068345166387295e-06,
      "loss": 0.3725,
      "step": 7721
    },
    {
      "epoch": 0.48369063092124837,
      "grad_norm": 0.8000415176334377,
      "learning_rate": 5.505825332610104e-06,
      "loss": 0.3171,
      "step": 7722
    },
    {
      "epoch": 0.4837532689205907,
      "grad_norm": 0.7884556034731384,
      "learning_rate": 5.504816127761614e-06,
      "loss": 0.415,
      "step": 7723
    },
    {
      "epoch": 0.483815906919933,
      "grad_norm": 0.826169202592857,
      "learning_rate": 5.5038069021348e-06,
      "loss": 0.4258,
      "step": 7724
    },
    {
      "epoch": 0.4838785449192753,
      "grad_norm": 0.7995974515618545,
      "learning_rate": 5.5027976557711995e-06,
      "loss": 0.4181,
      "step": 7725
    },
    {
      "epoch": 0.4839411829186176,
      "grad_norm": 0.9505464077250152,
      "learning_rate": 5.501788388712353e-06,
      "loss": 0.4336,
      "step": 7726
    },
    {
      "epoch": 0.4840038209179599,
      "grad_norm": 0.8111076149969676,
      "learning_rate": 5.5007791009998046e-06,
      "loss": 0.386,
      "step": 7727
    },
    {
      "epoch": 0.48406645891730216,
      "grad_norm": 0.6387611720516625,
      "learning_rate": 5.499769792675096e-06,
      "loss": 0.4629,
      "step": 7728
    },
    {
      "epoch": 0.48412909691664446,
      "grad_norm": 0.9155316611749938,
      "learning_rate": 5.498760463779768e-06,
      "loss": 0.4209,
      "step": 7729
    },
    {
      "epoch": 0.48419173491598677,
      "grad_norm": 0.9669332673877128,
      "learning_rate": 5.497751114355369e-06,
      "loss": 0.4337,
      "step": 7730
    },
    {
      "epoch": 0.4842543729153291,
      "grad_norm": 0.9002353614593587,
      "learning_rate": 5.496741744443441e-06,
      "loss": 0.4395,
      "step": 7731
    },
    {
      "epoch": 0.4843170109146714,
      "grad_norm": 0.848119200754335,
      "learning_rate": 5.49573235408553e-06,
      "loss": 0.4471,
      "step": 7732
    },
    {
      "epoch": 0.4843796489140137,
      "grad_norm": 0.7767910106313825,
      "learning_rate": 5.494722943323183e-06,
      "loss": 0.3874,
      "step": 7733
    },
    {
      "epoch": 0.484442286913356,
      "grad_norm": 0.8015401777566884,
      "learning_rate": 5.493713512197949e-06,
      "loss": 0.3932,
      "step": 7734
    },
    {
      "epoch": 0.4845049249126983,
      "grad_norm": 0.7985747478342451,
      "learning_rate": 5.492704060751375e-06,
      "loss": 0.3811,
      "step": 7735
    },
    {
      "epoch": 0.4845675629120406,
      "grad_norm": 0.8282192876274803,
      "learning_rate": 5.4916945890250095e-06,
      "loss": 0.4375,
      "step": 7736
    },
    {
      "epoch": 0.48463020091138287,
      "grad_norm": 0.8635752072157362,
      "learning_rate": 5.490685097060404e-06,
      "loss": 0.3957,
      "step": 7737
    },
    {
      "epoch": 0.4846928389107252,
      "grad_norm": 0.9171520712613376,
      "learning_rate": 5.489675584899108e-06,
      "loss": 0.4138,
      "step": 7738
    },
    {
      "epoch": 0.4847554769100675,
      "grad_norm": 0.806206787776343,
      "learning_rate": 5.488666052582675e-06,
      "loss": 0.3979,
      "step": 7739
    },
    {
      "epoch": 0.4848181149094098,
      "grad_norm": 0.7829792049090019,
      "learning_rate": 5.4876565001526575e-06,
      "loss": 0.388,
      "step": 7740
    },
    {
      "epoch": 0.4848807529087521,
      "grad_norm": 0.7901590470444992,
      "learning_rate": 5.486646927650607e-06,
      "loss": 0.3725,
      "step": 7741
    },
    {
      "epoch": 0.4849433909080944,
      "grad_norm": 0.8124632410744235,
      "learning_rate": 5.485637335118077e-06,
      "loss": 0.4118,
      "step": 7742
    },
    {
      "epoch": 0.4850060289074367,
      "grad_norm": 0.9216434067110852,
      "learning_rate": 5.484627722596626e-06,
      "loss": 0.4141,
      "step": 7743
    },
    {
      "epoch": 0.485068666906779,
      "grad_norm": 0.8210086533155874,
      "learning_rate": 5.483618090127808e-06,
      "loss": 0.4338,
      "step": 7744
    },
    {
      "epoch": 0.4851313049061213,
      "grad_norm": 0.8900510932681304,
      "learning_rate": 5.48260843775318e-06,
      "loss": 0.4374,
      "step": 7745
    },
    {
      "epoch": 0.4851939429054636,
      "grad_norm": 0.8188941972595953,
      "learning_rate": 5.481598765514296e-06,
      "loss": 0.391,
      "step": 7746
    },
    {
      "epoch": 0.4852565809048059,
      "grad_norm": 0.819450124414884,
      "learning_rate": 5.4805890734527215e-06,
      "loss": 0.3967,
      "step": 7747
    },
    {
      "epoch": 0.4853192189041482,
      "grad_norm": 0.761281374153849,
      "learning_rate": 5.47957936161001e-06,
      "loss": 0.3873,
      "step": 7748
    },
    {
      "epoch": 0.4853818569034905,
      "grad_norm": 0.8207052763321432,
      "learning_rate": 5.478569630027724e-06,
      "loss": 0.4063,
      "step": 7749
    },
    {
      "epoch": 0.4854444949028328,
      "grad_norm": 0.8332657963462927,
      "learning_rate": 5.4775598787474196e-06,
      "loss": 0.411,
      "step": 7750
    },
    {
      "epoch": 0.4855071329021751,
      "grad_norm": 0.8215794893129449,
      "learning_rate": 5.476550107810665e-06,
      "loss": 0.3675,
      "step": 7751
    },
    {
      "epoch": 0.48556977090151743,
      "grad_norm": 0.8948148320049125,
      "learning_rate": 5.475540317259019e-06,
      "loss": 0.4162,
      "step": 7752
    },
    {
      "epoch": 0.4856324089008597,
      "grad_norm": 0.852292355224699,
      "learning_rate": 5.474530507134046e-06,
      "loss": 0.4315,
      "step": 7753
    },
    {
      "epoch": 0.485695046900202,
      "grad_norm": 0.858031268275807,
      "learning_rate": 5.473520677477308e-06,
      "loss": 0.396,
      "step": 7754
    },
    {
      "epoch": 0.4857576848995443,
      "grad_norm": 0.832032964246917,
      "learning_rate": 5.47251082833037e-06,
      "loss": 0.4229,
      "step": 7755
    },
    {
      "epoch": 0.4858203228988866,
      "grad_norm": 0.8797715616465962,
      "learning_rate": 5.471500959734801e-06,
      "loss": 0.3944,
      "step": 7756
    },
    {
      "epoch": 0.4858829608982289,
      "grad_norm": 0.8386398442424283,
      "learning_rate": 5.470491071732165e-06,
      "loss": 0.3854,
      "step": 7757
    },
    {
      "epoch": 0.4859455988975712,
      "grad_norm": 0.8751952253108982,
      "learning_rate": 5.469481164364027e-06,
      "loss": 0.3875,
      "step": 7758
    },
    {
      "epoch": 0.4860082368969135,
      "grad_norm": 0.8391006895200483,
      "learning_rate": 5.468471237671958e-06,
      "loss": 0.3737,
      "step": 7759
    },
    {
      "epoch": 0.48607087489625583,
      "grad_norm": 0.809911668904557,
      "learning_rate": 5.467461291697527e-06,
      "loss": 0.4191,
      "step": 7760
    },
    {
      "epoch": 0.48613351289559814,
      "grad_norm": 0.8073691518885828,
      "learning_rate": 5.466451326482301e-06,
      "loss": 0.3667,
      "step": 7761
    },
    {
      "epoch": 0.4861961508949404,
      "grad_norm": 0.9416004402721146,
      "learning_rate": 5.465441342067855e-06,
      "loss": 0.3958,
      "step": 7762
    },
    {
      "epoch": 0.4862587888942827,
      "grad_norm": 0.8245721100778752,
      "learning_rate": 5.464431338495753e-06,
      "loss": 0.4061,
      "step": 7763
    },
    {
      "epoch": 0.486321426893625,
      "grad_norm": 0.822583196873881,
      "learning_rate": 5.463421315807571e-06,
      "loss": 0.4426,
      "step": 7764
    },
    {
      "epoch": 0.4863840648929673,
      "grad_norm": 0.8181857902093336,
      "learning_rate": 5.462411274044884e-06,
      "loss": 0.4029,
      "step": 7765
    },
    {
      "epoch": 0.4864467028923096,
      "grad_norm": 0.8275071897583343,
      "learning_rate": 5.461401213249262e-06,
      "loss": 0.3932,
      "step": 7766
    },
    {
      "epoch": 0.48650934089165193,
      "grad_norm": 0.7581498959237205,
      "learning_rate": 5.46039113346228e-06,
      "loss": 0.4143,
      "step": 7767
    },
    {
      "epoch": 0.48657197889099424,
      "grad_norm": 0.9057280425237672,
      "learning_rate": 5.459381034725515e-06,
      "loss": 0.4615,
      "step": 7768
    },
    {
      "epoch": 0.48663461689033655,
      "grad_norm": 0.810523483213158,
      "learning_rate": 5.45837091708054e-06,
      "loss": 0.4044,
      "step": 7769
    },
    {
      "epoch": 0.4866972548896788,
      "grad_norm": 0.8346812716837084,
      "learning_rate": 5.457360780568934e-06,
      "loss": 0.4225,
      "step": 7770
    },
    {
      "epoch": 0.4867598928890211,
      "grad_norm": 0.8716370954635201,
      "learning_rate": 5.456350625232272e-06,
      "loss": 0.4125,
      "step": 7771
    },
    {
      "epoch": 0.4868225308883634,
      "grad_norm": 0.8953397982350769,
      "learning_rate": 5.455340451112134e-06,
      "loss": 0.4444,
      "step": 7772
    },
    {
      "epoch": 0.4868851688877057,
      "grad_norm": 0.8464681562568621,
      "learning_rate": 5.454330258250099e-06,
      "loss": 0.4378,
      "step": 7773
    },
    {
      "epoch": 0.48694780688704803,
      "grad_norm": 0.7974384959435178,
      "learning_rate": 5.453320046687747e-06,
      "loss": 0.3609,
      "step": 7774
    },
    {
      "epoch": 0.48701044488639034,
      "grad_norm": 0.7497932199572223,
      "learning_rate": 5.452309816466656e-06,
      "loss": 0.3558,
      "step": 7775
    },
    {
      "epoch": 0.48707308288573264,
      "grad_norm": 0.8303597583268895,
      "learning_rate": 5.45129956762841e-06,
      "loss": 0.4092,
      "step": 7776
    },
    {
      "epoch": 0.48713572088507495,
      "grad_norm": 0.8162001423251961,
      "learning_rate": 5.450289300214591e-06,
      "loss": 0.4268,
      "step": 7777
    },
    {
      "epoch": 0.4871983588844172,
      "grad_norm": 0.8735815911790556,
      "learning_rate": 5.449279014266778e-06,
      "loss": 0.386,
      "step": 7778
    },
    {
      "epoch": 0.4872609968837595,
      "grad_norm": 0.868322446134831,
      "learning_rate": 5.448268709826561e-06,
      "loss": 0.4476,
      "step": 7779
    },
    {
      "epoch": 0.4873236348831018,
      "grad_norm": 0.7916883636962564,
      "learning_rate": 5.447258386935518e-06,
      "loss": 0.3809,
      "step": 7780
    },
    {
      "epoch": 0.4873862728824441,
      "grad_norm": 0.8203766829141648,
      "learning_rate": 5.446248045635238e-06,
      "loss": 0.4153,
      "step": 7781
    },
    {
      "epoch": 0.48744891088178643,
      "grad_norm": 0.9093110519028507,
      "learning_rate": 5.445237685967304e-06,
      "loss": 0.4606,
      "step": 7782
    },
    {
      "epoch": 0.48751154888112874,
      "grad_norm": 0.833587871521891,
      "learning_rate": 5.444227307973306e-06,
      "loss": 0.3857,
      "step": 7783
    },
    {
      "epoch": 0.48757418688047105,
      "grad_norm": 0.8448530531149201,
      "learning_rate": 5.443216911694829e-06,
      "loss": 0.4112,
      "step": 7784
    },
    {
      "epoch": 0.48763682487981336,
      "grad_norm": 0.8793244908239211,
      "learning_rate": 5.442206497173461e-06,
      "loss": 0.463,
      "step": 7785
    },
    {
      "epoch": 0.48769946287915567,
      "grad_norm": 0.8464728666410072,
      "learning_rate": 5.441196064450793e-06,
      "loss": 0.3815,
      "step": 7786
    },
    {
      "epoch": 0.4877621008784979,
      "grad_norm": 0.7776044268184341,
      "learning_rate": 5.440185613568412e-06,
      "loss": 0.377,
      "step": 7787
    },
    {
      "epoch": 0.4878247388778402,
      "grad_norm": 0.8183689268710741,
      "learning_rate": 5.439175144567909e-06,
      "loss": 0.4212,
      "step": 7788
    },
    {
      "epoch": 0.48788737687718253,
      "grad_norm": 0.8160233446564648,
      "learning_rate": 5.438164657490877e-06,
      "loss": 0.3949,
      "step": 7789
    },
    {
      "epoch": 0.48795001487652484,
      "grad_norm": 0.869160858263596,
      "learning_rate": 5.437154152378905e-06,
      "loss": 0.4287,
      "step": 7790
    },
    {
      "epoch": 0.48801265287586715,
      "grad_norm": 0.8800247310376946,
      "learning_rate": 5.436143629273588e-06,
      "loss": 0.4354,
      "step": 7791
    },
    {
      "epoch": 0.48807529087520946,
      "grad_norm": 0.8507683926483326,
      "learning_rate": 5.435133088216517e-06,
      "loss": 0.3789,
      "step": 7792
    },
    {
      "epoch": 0.48813792887455176,
      "grad_norm": 0.8143354900272304,
      "learning_rate": 5.434122529249287e-06,
      "loss": 0.3891,
      "step": 7793
    },
    {
      "epoch": 0.48820056687389407,
      "grad_norm": 0.8595117462281867,
      "learning_rate": 5.433111952413496e-06,
      "loss": 0.4012,
      "step": 7794
    },
    {
      "epoch": 0.4882632048732363,
      "grad_norm": 0.8720254475095656,
      "learning_rate": 5.432101357750734e-06,
      "loss": 0.435,
      "step": 7795
    },
    {
      "epoch": 0.48832584287257863,
      "grad_norm": 0.8291313221331921,
      "learning_rate": 5.431090745302601e-06,
      "loss": 0.4101,
      "step": 7796
    },
    {
      "epoch": 0.48838848087192094,
      "grad_norm": 0.8016292701947417,
      "learning_rate": 5.430080115110692e-06,
      "loss": 0.3657,
      "step": 7797
    },
    {
      "epoch": 0.48845111887126325,
      "grad_norm": 0.839904107659988,
      "learning_rate": 5.429069467216606e-06,
      "loss": 0.3815,
      "step": 7798
    },
    {
      "epoch": 0.48851375687060555,
      "grad_norm": 0.8539687843241052,
      "learning_rate": 5.4280588016619415e-06,
      "loss": 0.4132,
      "step": 7799
    },
    {
      "epoch": 0.48857639486994786,
      "grad_norm": 0.7967949788469144,
      "learning_rate": 5.4270481184882985e-06,
      "loss": 0.4254,
      "step": 7800
    },
    {
      "epoch": 0.48863903286929017,
      "grad_norm": 0.8480529532204412,
      "learning_rate": 5.4260374177372735e-06,
      "loss": 0.4064,
      "step": 7801
    },
    {
      "epoch": 0.4887016708686325,
      "grad_norm": 0.7962109860627781,
      "learning_rate": 5.42502669945047e-06,
      "loss": 0.445,
      "step": 7802
    },
    {
      "epoch": 0.4887643088679748,
      "grad_norm": 0.6114569004983532,
      "learning_rate": 5.4240159636694886e-06,
      "loss": 0.4677,
      "step": 7803
    },
    {
      "epoch": 0.48882694686731704,
      "grad_norm": 0.7863609040109345,
      "learning_rate": 5.423005210435932e-06,
      "loss": 0.4233,
      "step": 7804
    },
    {
      "epoch": 0.48888958486665934,
      "grad_norm": 0.8592966359066769,
      "learning_rate": 5.421994439791401e-06,
      "loss": 0.4444,
      "step": 7805
    },
    {
      "epoch": 0.48895222286600165,
      "grad_norm": 0.7787910218400447,
      "learning_rate": 5.420983651777501e-06,
      "loss": 0.3976,
      "step": 7806
    },
    {
      "epoch": 0.48901486086534396,
      "grad_norm": 0.8596040888314856,
      "learning_rate": 5.419972846435838e-06,
      "loss": 0.4009,
      "step": 7807
    },
    {
      "epoch": 0.48907749886468627,
      "grad_norm": 0.8181128216708677,
      "learning_rate": 5.418962023808013e-06,
      "loss": 0.3904,
      "step": 7808
    },
    {
      "epoch": 0.4891401368640286,
      "grad_norm": 0.871837638984344,
      "learning_rate": 5.417951183935634e-06,
      "loss": 0.3967,
      "step": 7809
    },
    {
      "epoch": 0.4892027748633709,
      "grad_norm": 0.7983377523408712,
      "learning_rate": 5.416940326860305e-06,
      "loss": 0.4202,
      "step": 7810
    },
    {
      "epoch": 0.4892654128627132,
      "grad_norm": 0.8806497239697385,
      "learning_rate": 5.415929452623636e-06,
      "loss": 0.4004,
      "step": 7811
    },
    {
      "epoch": 0.48932805086205544,
      "grad_norm": 0.9073317767383656,
      "learning_rate": 5.414918561267234e-06,
      "loss": 0.4517,
      "step": 7812
    },
    {
      "epoch": 0.48939068886139775,
      "grad_norm": 0.7808474111107123,
      "learning_rate": 5.413907652832706e-06,
      "loss": 0.3627,
      "step": 7813
    },
    {
      "epoch": 0.48945332686074006,
      "grad_norm": 0.8431000876837843,
      "learning_rate": 5.412896727361663e-06,
      "loss": 0.4238,
      "step": 7814
    },
    {
      "epoch": 0.48951596486008236,
      "grad_norm": 0.6622793784676336,
      "learning_rate": 5.411885784895713e-06,
      "loss": 0.4635,
      "step": 7815
    },
    {
      "epoch": 0.48957860285942467,
      "grad_norm": 0.7699498279969255,
      "learning_rate": 5.4108748254764685e-06,
      "loss": 0.4347,
      "step": 7816
    },
    {
      "epoch": 0.489641240858767,
      "grad_norm": 0.7959289560929675,
      "learning_rate": 5.40986384914554e-06,
      "loss": 0.3887,
      "step": 7817
    },
    {
      "epoch": 0.4897038788581093,
      "grad_norm": 0.8665892572790617,
      "learning_rate": 5.408852855944539e-06,
      "loss": 0.3556,
      "step": 7818
    },
    {
      "epoch": 0.4897665168574516,
      "grad_norm": 0.8359095099015971,
      "learning_rate": 5.407841845915077e-06,
      "loss": 0.4116,
      "step": 7819
    },
    {
      "epoch": 0.48982915485679385,
      "grad_norm": 0.8660351534738039,
      "learning_rate": 5.406830819098771e-06,
      "loss": 0.4201,
      "step": 7820
    },
    {
      "epoch": 0.48989179285613615,
      "grad_norm": 0.8247001923524315,
      "learning_rate": 5.4058197755372334e-06,
      "loss": 0.4078,
      "step": 7821
    },
    {
      "epoch": 0.48995443085547846,
      "grad_norm": 0.8799007744994145,
      "learning_rate": 5.404808715272079e-06,
      "loss": 0.3797,
      "step": 7822
    },
    {
      "epoch": 0.49001706885482077,
      "grad_norm": 0.8506934694515301,
      "learning_rate": 5.403797638344919e-06,
      "loss": 0.3795,
      "step": 7823
    },
    {
      "epoch": 0.4900797068541631,
      "grad_norm": 0.8184728155438548,
      "learning_rate": 5.402786544797376e-06,
      "loss": 0.3699,
      "step": 7824
    },
    {
      "epoch": 0.4901423448535054,
      "grad_norm": 0.7681290265944334,
      "learning_rate": 5.401775434671064e-06,
      "loss": 0.349,
      "step": 7825
    },
    {
      "epoch": 0.4902049828528477,
      "grad_norm": 0.8441266430696093,
      "learning_rate": 5.400764308007601e-06,
      "loss": 0.4283,
      "step": 7826
    },
    {
      "epoch": 0.49026762085219,
      "grad_norm": 0.7981436740186872,
      "learning_rate": 5.399753164848604e-06,
      "loss": 0.392,
      "step": 7827
    },
    {
      "epoch": 0.4903302588515323,
      "grad_norm": 0.7990938495785406,
      "learning_rate": 5.398742005235691e-06,
      "loss": 0.4194,
      "step": 7828
    },
    {
      "epoch": 0.49039289685087456,
      "grad_norm": 0.887476411284897,
      "learning_rate": 5.397730829210484e-06,
      "loss": 0.4062,
      "step": 7829
    },
    {
      "epoch": 0.49045553485021687,
      "grad_norm": 0.842788637349625,
      "learning_rate": 5.396719636814604e-06,
      "loss": 0.3864,
      "step": 7830
    },
    {
      "epoch": 0.4905181728495592,
      "grad_norm": 0.8406999911505058,
      "learning_rate": 5.395708428089669e-06,
      "loss": 0.3869,
      "step": 7831
    },
    {
      "epoch": 0.4905808108489015,
      "grad_norm": 0.8557224845774545,
      "learning_rate": 5.394697203077301e-06,
      "loss": 0.4229,
      "step": 7832
    },
    {
      "epoch": 0.4906434488482438,
      "grad_norm": 0.7731646841132203,
      "learning_rate": 5.393685961819123e-06,
      "loss": 0.4141,
      "step": 7833
    },
    {
      "epoch": 0.4907060868475861,
      "grad_norm": 0.8114024183385804,
      "learning_rate": 5.39267470435676e-06,
      "loss": 0.3988,
      "step": 7834
    },
    {
      "epoch": 0.4907687248469284,
      "grad_norm": 0.8523156208942252,
      "learning_rate": 5.39166343073183e-06,
      "loss": 0.4244,
      "step": 7835
    },
    {
      "epoch": 0.4908313628462707,
      "grad_norm": 0.8470661426757569,
      "learning_rate": 5.390652140985963e-06,
      "loss": 0.4148,
      "step": 7836
    },
    {
      "epoch": 0.49089400084561297,
      "grad_norm": 0.8284440078502625,
      "learning_rate": 5.38964083516078e-06,
      "loss": 0.3933,
      "step": 7837
    },
    {
      "epoch": 0.4909566388449553,
      "grad_norm": 0.8739395705425842,
      "learning_rate": 5.388629513297908e-06,
      "loss": 0.3703,
      "step": 7838
    },
    {
      "epoch": 0.4910192768442976,
      "grad_norm": 1.0242507026505416,
      "learning_rate": 5.387618175438973e-06,
      "loss": 0.3997,
      "step": 7839
    },
    {
      "epoch": 0.4910819148436399,
      "grad_norm": 0.6648881210719019,
      "learning_rate": 5.386606821625602e-06,
      "loss": 0.4533,
      "step": 7840
    },
    {
      "epoch": 0.4911445528429822,
      "grad_norm": 0.8129163002880583,
      "learning_rate": 5.385595451899424e-06,
      "loss": 0.4157,
      "step": 7841
    },
    {
      "epoch": 0.4912071908423245,
      "grad_norm": 0.8402643074034954,
      "learning_rate": 5.3845840663020635e-06,
      "loss": 0.3892,
      "step": 7842
    },
    {
      "epoch": 0.4912698288416668,
      "grad_norm": 0.8498948508182753,
      "learning_rate": 5.3835726648751514e-06,
      "loss": 0.4158,
      "step": 7843
    },
    {
      "epoch": 0.4913324668410091,
      "grad_norm": 0.8046864053487216,
      "learning_rate": 5.382561247660319e-06,
      "loss": 0.3919,
      "step": 7844
    },
    {
      "epoch": 0.4913951048403514,
      "grad_norm": 0.8102035442754907,
      "learning_rate": 5.381549814699194e-06,
      "loss": 0.3658,
      "step": 7845
    },
    {
      "epoch": 0.4914577428396937,
      "grad_norm": 0.835587213368401,
      "learning_rate": 5.380538366033408e-06,
      "loss": 0.4012,
      "step": 7846
    },
    {
      "epoch": 0.491520380839036,
      "grad_norm": 0.926219588846347,
      "learning_rate": 5.37952690170459e-06,
      "loss": 0.4104,
      "step": 7847
    },
    {
      "epoch": 0.4915830188383783,
      "grad_norm": 0.8194331266439853,
      "learning_rate": 5.378515421754377e-06,
      "loss": 0.3866,
      "step": 7848
    },
    {
      "epoch": 0.4916456568377206,
      "grad_norm": 0.6835926797526378,
      "learning_rate": 5.3775039262243975e-06,
      "loss": 0.4826,
      "step": 7849
    },
    {
      "epoch": 0.4917082948370629,
      "grad_norm": 0.8571309467996837,
      "learning_rate": 5.376492415156288e-06,
      "loss": 0.4293,
      "step": 7850
    },
    {
      "epoch": 0.4917709328364052,
      "grad_norm": 0.8596236155413312,
      "learning_rate": 5.3754808885916785e-06,
      "loss": 0.4357,
      "step": 7851
    },
    {
      "epoch": 0.4918335708357475,
      "grad_norm": 0.8455601125330918,
      "learning_rate": 5.3744693465722075e-06,
      "loss": 0.4046,
      "step": 7852
    },
    {
      "epoch": 0.49189620883508983,
      "grad_norm": 0.7693286592471997,
      "learning_rate": 5.3734577891395076e-06,
      "loss": 0.4149,
      "step": 7853
    },
    {
      "epoch": 0.4919588468344321,
      "grad_norm": 0.8622862297643794,
      "learning_rate": 5.372446216335218e-06,
      "loss": 0.387,
      "step": 7854
    },
    {
      "epoch": 0.4920214848337744,
      "grad_norm": 0.758349678677896,
      "learning_rate": 5.371434628200971e-06,
      "loss": 0.401,
      "step": 7855
    },
    {
      "epoch": 0.4920841228331167,
      "grad_norm": 0.8849485600509556,
      "learning_rate": 5.370423024778406e-06,
      "loss": 0.4083,
      "step": 7856
    },
    {
      "epoch": 0.492146760832459,
      "grad_norm": 0.8043598312548669,
      "learning_rate": 5.36941140610916e-06,
      "loss": 0.3774,
      "step": 7857
    },
    {
      "epoch": 0.4922093988318013,
      "grad_norm": 0.8220235380744366,
      "learning_rate": 5.3683997722348735e-06,
      "loss": 0.3856,
      "step": 7858
    },
    {
      "epoch": 0.4922720368311436,
      "grad_norm": 0.7969904963749725,
      "learning_rate": 5.367388123197183e-06,
      "loss": 0.4149,
      "step": 7859
    },
    {
      "epoch": 0.49233467483048593,
      "grad_norm": 0.7920737713759081,
      "learning_rate": 5.366376459037731e-06,
      "loss": 0.397,
      "step": 7860
    },
    {
      "epoch": 0.49239731282982824,
      "grad_norm": 0.8114401972699259,
      "learning_rate": 5.365364779798152e-06,
      "loss": 0.394,
      "step": 7861
    },
    {
      "epoch": 0.4924599508291705,
      "grad_norm": 0.761833287214267,
      "learning_rate": 5.364353085520094e-06,
      "loss": 0.4012,
      "step": 7862
    },
    {
      "epoch": 0.4925225888285128,
      "grad_norm": 0.8901351018065041,
      "learning_rate": 5.363341376245195e-06,
      "loss": 0.4216,
      "step": 7863
    },
    {
      "epoch": 0.4925852268278551,
      "grad_norm": 0.7683836309605898,
      "learning_rate": 5.362329652015098e-06,
      "loss": 0.3382,
      "step": 7864
    },
    {
      "epoch": 0.4926478648271974,
      "grad_norm": 0.8053831542438471,
      "learning_rate": 5.361317912871443e-06,
      "loss": 0.4203,
      "step": 7865
    },
    {
      "epoch": 0.4927105028265397,
      "grad_norm": 0.863627552505772,
      "learning_rate": 5.360306158855877e-06,
      "loss": 0.3956,
      "step": 7866
    },
    {
      "epoch": 0.49277314082588203,
      "grad_norm": 0.8685823831818547,
      "learning_rate": 5.359294390010042e-06,
      "loss": 0.4121,
      "step": 7867
    },
    {
      "epoch": 0.49283577882522434,
      "grad_norm": 0.7673873367748734,
      "learning_rate": 5.358282606375585e-06,
      "loss": 0.388,
      "step": 7868
    },
    {
      "epoch": 0.49289841682456664,
      "grad_norm": 1.0311238123950504,
      "learning_rate": 5.357270807994148e-06,
      "loss": 0.4243,
      "step": 7869
    },
    {
      "epoch": 0.49296105482390895,
      "grad_norm": 0.5899100513273636,
      "learning_rate": 5.356258994907377e-06,
      "loss": 0.4528,
      "step": 7870
    },
    {
      "epoch": 0.4930236928232512,
      "grad_norm": 0.7986596359538951,
      "learning_rate": 5.355247167156923e-06,
      "loss": 0.3892,
      "step": 7871
    },
    {
      "epoch": 0.4930863308225935,
      "grad_norm": 0.8978114359982385,
      "learning_rate": 5.354235324784428e-06,
      "loss": 0.4201,
      "step": 7872
    },
    {
      "epoch": 0.4931489688219358,
      "grad_norm": 0.8870222247736032,
      "learning_rate": 5.353223467831541e-06,
      "loss": 0.4077,
      "step": 7873
    },
    {
      "epoch": 0.4932116068212781,
      "grad_norm": 0.835042582701501,
      "learning_rate": 5.352211596339909e-06,
      "loss": 0.392,
      "step": 7874
    },
    {
      "epoch": 0.49327424482062043,
      "grad_norm": 0.8521420935256944,
      "learning_rate": 5.351199710351185e-06,
      "loss": 0.4016,
      "step": 7875
    },
    {
      "epoch": 0.49333688281996274,
      "grad_norm": 0.790092956141432,
      "learning_rate": 5.350187809907015e-06,
      "loss": 0.4177,
      "step": 7876
    },
    {
      "epoch": 0.49339952081930505,
      "grad_norm": 0.8494269544932228,
      "learning_rate": 5.34917589504905e-06,
      "loss": 0.4231,
      "step": 7877
    },
    {
      "epoch": 0.49346215881864736,
      "grad_norm": 0.7821824369536213,
      "learning_rate": 5.348163965818939e-06,
      "loss": 0.4001,
      "step": 7878
    },
    {
      "epoch": 0.4935247968179896,
      "grad_norm": 0.7996456881929658,
      "learning_rate": 5.347152022258335e-06,
      "loss": 0.3693,
      "step": 7879
    },
    {
      "epoch": 0.4935874348173319,
      "grad_norm": 0.8805903743372407,
      "learning_rate": 5.34614006440889e-06,
      "loss": 0.4454,
      "step": 7880
    },
    {
      "epoch": 0.4936500728166742,
      "grad_norm": 0.5782300649844151,
      "learning_rate": 5.345128092312255e-06,
      "loss": 0.4661,
      "step": 7881
    },
    {
      "epoch": 0.49371271081601653,
      "grad_norm": 0.8711726444194026,
      "learning_rate": 5.344116106010085e-06,
      "loss": 0.3903,
      "step": 7882
    },
    {
      "epoch": 0.49377534881535884,
      "grad_norm": 0.8249134436194034,
      "learning_rate": 5.34310410554403e-06,
      "loss": 0.4417,
      "step": 7883
    },
    {
      "epoch": 0.49383798681470115,
      "grad_norm": 0.8117752207448842,
      "learning_rate": 5.342092090955747e-06,
      "loss": 0.4327,
      "step": 7884
    },
    {
      "epoch": 0.49390062481404345,
      "grad_norm": 0.8039528914119225,
      "learning_rate": 5.341080062286892e-06,
      "loss": 0.428,
      "step": 7885
    },
    {
      "epoch": 0.49396326281338576,
      "grad_norm": 0.8662733260885733,
      "learning_rate": 5.3400680195791165e-06,
      "loss": 0.4118,
      "step": 7886
    },
    {
      "epoch": 0.494025900812728,
      "grad_norm": 0.8270479419120977,
      "learning_rate": 5.339055962874078e-06,
      "loss": 0.3956,
      "step": 7887
    },
    {
      "epoch": 0.4940885388120703,
      "grad_norm": 0.799513815582835,
      "learning_rate": 5.338043892213434e-06,
      "loss": 0.3779,
      "step": 7888
    },
    {
      "epoch": 0.49415117681141263,
      "grad_norm": 0.7806505730769561,
      "learning_rate": 5.3370318076388405e-06,
      "loss": 0.3595,
      "step": 7889
    },
    {
      "epoch": 0.49421381481075494,
      "grad_norm": 0.9095921763509253,
      "learning_rate": 5.336019709191955e-06,
      "loss": 0.4271,
      "step": 7890
    },
    {
      "epoch": 0.49427645281009724,
      "grad_norm": 0.6893381859105598,
      "learning_rate": 5.335007596914435e-06,
      "loss": 0.4833,
      "step": 7891
    },
    {
      "epoch": 0.49433909080943955,
      "grad_norm": 0.8073637246194516,
      "learning_rate": 5.333995470847943e-06,
      "loss": 0.4166,
      "step": 7892
    },
    {
      "epoch": 0.49440172880878186,
      "grad_norm": 0.8322161369184835,
      "learning_rate": 5.332983331034133e-06,
      "loss": 0.4047,
      "step": 7893
    },
    {
      "epoch": 0.49446436680812417,
      "grad_norm": 0.8206220806994431,
      "learning_rate": 5.331971177514667e-06,
      "loss": 0.3869,
      "step": 7894
    },
    {
      "epoch": 0.4945270048074665,
      "grad_norm": 0.6488075499760424,
      "learning_rate": 5.330959010331207e-06,
      "loss": 0.4887,
      "step": 7895
    },
    {
      "epoch": 0.4945896428068087,
      "grad_norm": 0.8070859165589779,
      "learning_rate": 5.329946829525413e-06,
      "loss": 0.4046,
      "step": 7896
    },
    {
      "epoch": 0.49465228080615103,
      "grad_norm": 0.8743892645906777,
      "learning_rate": 5.328934635138945e-06,
      "loss": 0.4081,
      "step": 7897
    },
    {
      "epoch": 0.49471491880549334,
      "grad_norm": 0.8516912683525127,
      "learning_rate": 5.327922427213466e-06,
      "loss": 0.4212,
      "step": 7898
    },
    {
      "epoch": 0.49477755680483565,
      "grad_norm": 0.8319476676261982,
      "learning_rate": 5.326910205790642e-06,
      "loss": 0.424,
      "step": 7899
    },
    {
      "epoch": 0.49484019480417796,
      "grad_norm": 0.8555964868166106,
      "learning_rate": 5.325897970912131e-06,
      "loss": 0.4158,
      "step": 7900
    },
    {
      "epoch": 0.49490283280352027,
      "grad_norm": 0.864033225706373,
      "learning_rate": 5.3248857226196e-06,
      "loss": 0.4032,
      "step": 7901
    },
    {
      "epoch": 0.4949654708028626,
      "grad_norm": 0.7707699647158502,
      "learning_rate": 5.323873460954709e-06,
      "loss": 0.3582,
      "step": 7902
    },
    {
      "epoch": 0.4950281088022049,
      "grad_norm": 0.8530023822428766,
      "learning_rate": 5.3228611859591306e-06,
      "loss": 0.4086,
      "step": 7903
    },
    {
      "epoch": 0.49509074680154713,
      "grad_norm": 0.8298906351227376,
      "learning_rate": 5.321848897674523e-06,
      "loss": 0.4033,
      "step": 7904
    },
    {
      "epoch": 0.49515338480088944,
      "grad_norm": 0.8266420842412621,
      "learning_rate": 5.320836596142557e-06,
      "loss": 0.4005,
      "step": 7905
    },
    {
      "epoch": 0.49521602280023175,
      "grad_norm": 0.7812994653625422,
      "learning_rate": 5.319824281404896e-06,
      "loss": 0.4242,
      "step": 7906
    },
    {
      "epoch": 0.49527866079957406,
      "grad_norm": 0.8369635012752109,
      "learning_rate": 5.318811953503208e-06,
      "loss": 0.4095,
      "step": 7907
    },
    {
      "epoch": 0.49534129879891636,
      "grad_norm": 0.8657117774282258,
      "learning_rate": 5.317799612479161e-06,
      "loss": 0.4136,
      "step": 7908
    },
    {
      "epoch": 0.49540393679825867,
      "grad_norm": 0.8288579604049164,
      "learning_rate": 5.316787258374424e-06,
      "loss": 0.4363,
      "step": 7909
    },
    {
      "epoch": 0.495466574797601,
      "grad_norm": 0.8538797602504867,
      "learning_rate": 5.315774891230664e-06,
      "loss": 0.4208,
      "step": 7910
    },
    {
      "epoch": 0.4955292127969433,
      "grad_norm": 0.8318187547466278,
      "learning_rate": 5.31476251108955e-06,
      "loss": 0.4148,
      "step": 7911
    },
    {
      "epoch": 0.4955918507962856,
      "grad_norm": 0.7081102965052162,
      "learning_rate": 5.313750117992754e-06,
      "loss": 0.4722,
      "step": 7912
    },
    {
      "epoch": 0.49565448879562785,
      "grad_norm": 0.8283823256182651,
      "learning_rate": 5.312737711981943e-06,
      "loss": 0.3954,
      "step": 7913
    },
    {
      "epoch": 0.49571712679497015,
      "grad_norm": 0.8038997895633818,
      "learning_rate": 5.3117252930987925e-06,
      "loss": 0.3652,
      "step": 7914
    },
    {
      "epoch": 0.49577976479431246,
      "grad_norm": 0.7747548633190274,
      "learning_rate": 5.31071286138497e-06,
      "loss": 0.3998,
      "step": 7915
    },
    {
      "epoch": 0.49584240279365477,
      "grad_norm": 0.848885694076301,
      "learning_rate": 5.3097004168821475e-06,
      "loss": 0.4001,
      "step": 7916
    },
    {
      "epoch": 0.4959050407929971,
      "grad_norm": 0.9014473592385716,
      "learning_rate": 5.308687959631998e-06,
      "loss": 0.3995,
      "step": 7917
    },
    {
      "epoch": 0.4959676787923394,
      "grad_norm": 0.8442513227076458,
      "learning_rate": 5.307675489676196e-06,
      "loss": 0.4105,
      "step": 7918
    },
    {
      "epoch": 0.4960303167916817,
      "grad_norm": 0.8472701083013067,
      "learning_rate": 5.306663007056414e-06,
      "loss": 0.367,
      "step": 7919
    },
    {
      "epoch": 0.496092954791024,
      "grad_norm": 0.8038317430648945,
      "learning_rate": 5.305650511814325e-06,
      "loss": 0.3832,
      "step": 7920
    },
    {
      "epoch": 0.49615559279036625,
      "grad_norm": 0.8135051057217856,
      "learning_rate": 5.304638003991604e-06,
      "loss": 0.4016,
      "step": 7921
    },
    {
      "epoch": 0.49621823078970856,
      "grad_norm": 0.829587598439087,
      "learning_rate": 5.303625483629926e-06,
      "loss": 0.409,
      "step": 7922
    },
    {
      "epoch": 0.49628086878905087,
      "grad_norm": 0.8392312163579224,
      "learning_rate": 5.302612950770969e-06,
      "loss": 0.4207,
      "step": 7923
    },
    {
      "epoch": 0.4963435067883932,
      "grad_norm": 0.8826883802058997,
      "learning_rate": 5.301600405456405e-06,
      "loss": 0.4311,
      "step": 7924
    },
    {
      "epoch": 0.4964061447877355,
      "grad_norm": 0.7611269608598478,
      "learning_rate": 5.3005878477279115e-06,
      "loss": 0.39,
      "step": 7925
    },
    {
      "epoch": 0.4964687827870778,
      "grad_norm": 0.8507211383984858,
      "learning_rate": 5.299575277627167e-06,
      "loss": 0.4042,
      "step": 7926
    },
    {
      "epoch": 0.4965314207864201,
      "grad_norm": 0.7646915413553428,
      "learning_rate": 5.298562695195849e-06,
      "loss": 0.3777,
      "step": 7927
    },
    {
      "epoch": 0.4965940587857624,
      "grad_norm": 0.8321403105894943,
      "learning_rate": 5.297550100475636e-06,
      "loss": 0.3869,
      "step": 7928
    },
    {
      "epoch": 0.49665669678510466,
      "grad_norm": 0.9178049796167754,
      "learning_rate": 5.296537493508204e-06,
      "loss": 0.4434,
      "step": 7929
    },
    {
      "epoch": 0.49671933478444696,
      "grad_norm": 0.7988394349718981,
      "learning_rate": 5.2955248743352325e-06,
      "loss": 0.3926,
      "step": 7930
    },
    {
      "epoch": 0.49678197278378927,
      "grad_norm": 0.7514736174165068,
      "learning_rate": 5.294512242998405e-06,
      "loss": 0.3351,
      "step": 7931
    },
    {
      "epoch": 0.4968446107831316,
      "grad_norm": 0.782004623661919,
      "learning_rate": 5.293499599539399e-06,
      "loss": 0.4195,
      "step": 7932
    },
    {
      "epoch": 0.4969072487824739,
      "grad_norm": 0.8148770156390843,
      "learning_rate": 5.292486943999893e-06,
      "loss": 0.387,
      "step": 7933
    },
    {
      "epoch": 0.4969698867818162,
      "grad_norm": 0.8062887252967835,
      "learning_rate": 5.2914742764215695e-06,
      "loss": 0.3989,
      "step": 7934
    },
    {
      "epoch": 0.4970325247811585,
      "grad_norm": 0.7760871119834902,
      "learning_rate": 5.2904615968461135e-06,
      "loss": 0.3833,
      "step": 7935
    },
    {
      "epoch": 0.4970951627805008,
      "grad_norm": 0.8435145787087438,
      "learning_rate": 5.289448905315203e-06,
      "loss": 0.3965,
      "step": 7936
    },
    {
      "epoch": 0.4971578007798431,
      "grad_norm": 0.7704307295350524,
      "learning_rate": 5.288436201870522e-06,
      "loss": 0.3775,
      "step": 7937
    },
    {
      "epoch": 0.49722043877918537,
      "grad_norm": 0.8159246068451939,
      "learning_rate": 5.287423486553753e-06,
      "loss": 0.3849,
      "step": 7938
    },
    {
      "epoch": 0.4972830767785277,
      "grad_norm": 0.8544436003446759,
      "learning_rate": 5.28641075940658e-06,
      "loss": 0.4313,
      "step": 7939
    },
    {
      "epoch": 0.49734571477787,
      "grad_norm": 0.7892392216351488,
      "learning_rate": 5.285398020470688e-06,
      "loss": 0.3995,
      "step": 7940
    },
    {
      "epoch": 0.4974083527772123,
      "grad_norm": 0.8444807168901788,
      "learning_rate": 5.28438526978776e-06,
      "loss": 0.4257,
      "step": 7941
    },
    {
      "epoch": 0.4974709907765546,
      "grad_norm": 0.8263308388183,
      "learning_rate": 5.28337250739948e-06,
      "loss": 0.4291,
      "step": 7942
    },
    {
      "epoch": 0.4975336287758969,
      "grad_norm": 0.8658541577837781,
      "learning_rate": 5.282359733347538e-06,
      "loss": 0.388,
      "step": 7943
    },
    {
      "epoch": 0.4975962667752392,
      "grad_norm": 0.8114391300245896,
      "learning_rate": 5.281346947673614e-06,
      "loss": 0.3934,
      "step": 7944
    },
    {
      "epoch": 0.4976589047745815,
      "grad_norm": 0.8508272394641022,
      "learning_rate": 5.280334150419399e-06,
      "loss": 0.3635,
      "step": 7945
    },
    {
      "epoch": 0.4977215427739238,
      "grad_norm": 0.8053853064409539,
      "learning_rate": 5.279321341626576e-06,
      "loss": 0.3869,
      "step": 7946
    },
    {
      "epoch": 0.4977841807732661,
      "grad_norm": 0.8392272243803702,
      "learning_rate": 5.278308521336838e-06,
      "loss": 0.4042,
      "step": 7947
    },
    {
      "epoch": 0.4978468187726084,
      "grad_norm": 0.8203888529599408,
      "learning_rate": 5.277295689591867e-06,
      "loss": 0.3744,
      "step": 7948
    },
    {
      "epoch": 0.4979094567719507,
      "grad_norm": 0.6440666730972482,
      "learning_rate": 5.276282846433356e-06,
      "loss": 0.4639,
      "step": 7949
    },
    {
      "epoch": 0.497972094771293,
      "grad_norm": 0.7988735391111218,
      "learning_rate": 5.2752699919029916e-06,
      "loss": 0.3568,
      "step": 7950
    },
    {
      "epoch": 0.4980347327706353,
      "grad_norm": 0.8294870455593558,
      "learning_rate": 5.274257126042461e-06,
      "loss": 0.3705,
      "step": 7951
    },
    {
      "epoch": 0.4980973707699776,
      "grad_norm": 0.8754582511620503,
      "learning_rate": 5.273244248893459e-06,
      "loss": 0.3617,
      "step": 7952
    },
    {
      "epoch": 0.49816000876931993,
      "grad_norm": 0.8599521105047396,
      "learning_rate": 5.272231360497671e-06,
      "loss": 0.4236,
      "step": 7953
    },
    {
      "epoch": 0.49822264676866224,
      "grad_norm": 0.8208086732258043,
      "learning_rate": 5.271218460896791e-06,
      "loss": 0.4124,
      "step": 7954
    },
    {
      "epoch": 0.4982852847680045,
      "grad_norm": 0.8444877968827157,
      "learning_rate": 5.270205550132508e-06,
      "loss": 0.4116,
      "step": 7955
    },
    {
      "epoch": 0.4983479227673468,
      "grad_norm": 0.7994621899656763,
      "learning_rate": 5.269192628246515e-06,
      "loss": 0.393,
      "step": 7956
    },
    {
      "epoch": 0.4984105607666891,
      "grad_norm": 0.7453354700303731,
      "learning_rate": 5.268179695280502e-06,
      "loss": 0.4128,
      "step": 7957
    },
    {
      "epoch": 0.4984731987660314,
      "grad_norm": 0.6216481618109457,
      "learning_rate": 5.2671667512761644e-06,
      "loss": 0.4464,
      "step": 7958
    },
    {
      "epoch": 0.4985358367653737,
      "grad_norm": 0.8202839752861416,
      "learning_rate": 5.266153796275193e-06,
      "loss": 0.3953,
      "step": 7959
    },
    {
      "epoch": 0.498598474764716,
      "grad_norm": 0.7968070868647587,
      "learning_rate": 5.265140830319283e-06,
      "loss": 0.4577,
      "step": 7960
    },
    {
      "epoch": 0.49866111276405833,
      "grad_norm": 0.8147653442841004,
      "learning_rate": 5.264127853450127e-06,
      "loss": 0.4213,
      "step": 7961
    },
    {
      "epoch": 0.49872375076340064,
      "grad_norm": 0.81848027929116,
      "learning_rate": 5.2631148657094186e-06,
      "loss": 0.4019,
      "step": 7962
    },
    {
      "epoch": 0.4987863887627429,
      "grad_norm": 0.811596455275525,
      "learning_rate": 5.262101867138855e-06,
      "loss": 0.3965,
      "step": 7963
    },
    {
      "epoch": 0.4988490267620852,
      "grad_norm": 0.6115116410096689,
      "learning_rate": 5.2610888577801275e-06,
      "loss": 0.4718,
      "step": 7964
    },
    {
      "epoch": 0.4989116647614275,
      "grad_norm": 0.8078039146125968,
      "learning_rate": 5.260075837674936e-06,
      "loss": 0.4231,
      "step": 7965
    },
    {
      "epoch": 0.4989743027607698,
      "grad_norm": 0.7915967685217743,
      "learning_rate": 5.259062806864975e-06,
      "loss": 0.389,
      "step": 7966
    },
    {
      "epoch": 0.4990369407601121,
      "grad_norm": 0.8861880610229149,
      "learning_rate": 5.258049765391939e-06,
      "loss": 0.4373,
      "step": 7967
    },
    {
      "epoch": 0.49909957875945443,
      "grad_norm": 0.8399777850189752,
      "learning_rate": 5.257036713297527e-06,
      "loss": 0.3754,
      "step": 7968
    },
    {
      "epoch": 0.49916221675879674,
      "grad_norm": 0.8622998758205977,
      "learning_rate": 5.256023650623438e-06,
      "loss": 0.4377,
      "step": 7969
    },
    {
      "epoch": 0.49922485475813905,
      "grad_norm": 0.6751485501308481,
      "learning_rate": 5.255010577411367e-06,
      "loss": 0.4477,
      "step": 7970
    },
    {
      "epoch": 0.4992874927574813,
      "grad_norm": 0.8540396117937091,
      "learning_rate": 5.253997493703014e-06,
      "loss": 0.4349,
      "step": 7971
    },
    {
      "epoch": 0.4993501307568236,
      "grad_norm": 0.8281200389726247,
      "learning_rate": 5.252984399540075e-06,
      "loss": 0.3779,
      "step": 7972
    },
    {
      "epoch": 0.4994127687561659,
      "grad_norm": 0.8104148694047016,
      "learning_rate": 5.251971294964253e-06,
      "loss": 0.4024,
      "step": 7973
    },
    {
      "epoch": 0.4994754067555082,
      "grad_norm": 0.8357479158163168,
      "learning_rate": 5.250958180017245e-06,
      "loss": 0.3662,
      "step": 7974
    },
    {
      "epoch": 0.49953804475485053,
      "grad_norm": 0.8496535940790881,
      "learning_rate": 5.249945054740753e-06,
      "loss": 0.4898,
      "step": 7975
    },
    {
      "epoch": 0.49960068275419284,
      "grad_norm": 0.812400015028082,
      "learning_rate": 5.248931919176474e-06,
      "loss": 0.4026,
      "step": 7976
    },
    {
      "epoch": 0.49966332075353515,
      "grad_norm": 0.8226028563850549,
      "learning_rate": 5.247918773366112e-06,
      "loss": 0.3925,
      "step": 7977
    },
    {
      "epoch": 0.49972595875287745,
      "grad_norm": 0.815346627117892,
      "learning_rate": 5.246905617351368e-06,
      "loss": 0.3981,
      "step": 7978
    },
    {
      "epoch": 0.49978859675221976,
      "grad_norm": 0.7904925247117606,
      "learning_rate": 5.245892451173943e-06,
      "loss": 0.3678,
      "step": 7979
    },
    {
      "epoch": 0.499851234751562,
      "grad_norm": 0.7584586341950792,
      "learning_rate": 5.244879274875538e-06,
      "loss": 0.4021,
      "step": 7980
    },
    {
      "epoch": 0.4999138727509043,
      "grad_norm": 0.8300592137564236,
      "learning_rate": 5.243866088497855e-06,
      "loss": 0.4086,
      "step": 7981
    },
    {
      "epoch": 0.4999765107502466,
      "grad_norm": 0.8806050242063084,
      "learning_rate": 5.242852892082602e-06,
      "loss": 0.3989,
      "step": 7982
    },
    {
      "epoch": 0.500039148749589,
      "grad_norm": 0.8506439922381973,
      "learning_rate": 5.241839685671477e-06,
      "loss": 0.4501,
      "step": 7983
    },
    {
      "epoch": 0.5001017867489312,
      "grad_norm": 0.7935941142765665,
      "learning_rate": 5.240826469306187e-06,
      "loss": 0.4177,
      "step": 7984
    },
    {
      "epoch": 0.5001644247482735,
      "grad_norm": 0.7754983338674267,
      "learning_rate": 5.2398132430284324e-06,
      "loss": 0.3946,
      "step": 7985
    },
    {
      "epoch": 0.5002270627476159,
      "grad_norm": 0.8513153618286038,
      "learning_rate": 5.2388000068799215e-06,
      "loss": 0.4112,
      "step": 7986
    },
    {
      "epoch": 0.5002897007469581,
      "grad_norm": 0.8486055503098392,
      "learning_rate": 5.237786760902359e-06,
      "loss": 0.4156,
      "step": 7987
    },
    {
      "epoch": 0.5003523387463005,
      "grad_norm": 0.8049487741527228,
      "learning_rate": 5.236773505137449e-06,
      "loss": 0.4398,
      "step": 7988
    },
    {
      "epoch": 0.5004149767456427,
      "grad_norm": 0.8295204714015403,
      "learning_rate": 5.235760239626897e-06,
      "loss": 0.4072,
      "step": 7989
    },
    {
      "epoch": 0.5004776147449851,
      "grad_norm": 0.8974043095878751,
      "learning_rate": 5.234746964412409e-06,
      "loss": 0.3842,
      "step": 7990
    },
    {
      "epoch": 0.5005402527443273,
      "grad_norm": 0.8635897434161738,
      "learning_rate": 5.233733679535693e-06,
      "loss": 0.4028,
      "step": 7991
    },
    {
      "epoch": 0.5006028907436696,
      "grad_norm": 0.8727767633894371,
      "learning_rate": 5.232720385038457e-06,
      "loss": 0.427,
      "step": 7992
    },
    {
      "epoch": 0.500665528743012,
      "grad_norm": 0.8666107999003545,
      "learning_rate": 5.231707080962405e-06,
      "loss": 0.4009,
      "step": 7993
    },
    {
      "epoch": 0.5007281667423542,
      "grad_norm": 0.8749977635542127,
      "learning_rate": 5.230693767349245e-06,
      "loss": 0.3859,
      "step": 7994
    },
    {
      "epoch": 0.5007908047416966,
      "grad_norm": 0.8254774378947833,
      "learning_rate": 5.22968044424069e-06,
      "loss": 0.4366,
      "step": 7995
    },
    {
      "epoch": 0.5008534427410388,
      "grad_norm": 0.8361807198716493,
      "learning_rate": 5.228667111678444e-06,
      "loss": 0.4297,
      "step": 7996
    },
    {
      "epoch": 0.5009160807403812,
      "grad_norm": 0.8100662417850062,
      "learning_rate": 5.2276537697042175e-06,
      "loss": 0.3781,
      "step": 7997
    },
    {
      "epoch": 0.5009787187397234,
      "grad_norm": 0.8886558287824778,
      "learning_rate": 5.226640418359719e-06,
      "loss": 0.3917,
      "step": 7998
    },
    {
      "epoch": 0.5010413567390658,
      "grad_norm": 0.8162881982772808,
      "learning_rate": 5.225627057686659e-06,
      "loss": 0.381,
      "step": 7999
    },
    {
      "epoch": 0.501103994738408,
      "grad_norm": 0.8020472082690054,
      "learning_rate": 5.224613687726748e-06,
      "loss": 0.3849,
      "step": 8000
    },
    {
      "epoch": 0.5011666327377503,
      "grad_norm": 0.8306363253981661,
      "learning_rate": 5.2236003085216945e-06,
      "loss": 0.4042,
      "step": 8001
    },
    {
      "epoch": 0.5012292707370927,
      "grad_norm": 0.9002771537194435,
      "learning_rate": 5.222586920113212e-06,
      "loss": 0.4188,
      "step": 8002
    },
    {
      "epoch": 0.5012919087364349,
      "grad_norm": 0.9480919588354509,
      "learning_rate": 5.221573522543009e-06,
      "loss": 0.4185,
      "step": 8003
    },
    {
      "epoch": 0.5013545467357773,
      "grad_norm": 0.8493062797966797,
      "learning_rate": 5.220560115852801e-06,
      "loss": 0.4219,
      "step": 8004
    },
    {
      "epoch": 0.5014171847351195,
      "grad_norm": 0.8380257501779065,
      "learning_rate": 5.219546700084297e-06,
      "loss": 0.3956,
      "step": 8005
    },
    {
      "epoch": 0.5014798227344619,
      "grad_norm": 0.8379784466985123,
      "learning_rate": 5.218533275279208e-06,
      "loss": 0.3925,
      "step": 8006
    },
    {
      "epoch": 0.5015424607338042,
      "grad_norm": 0.8100619490455729,
      "learning_rate": 5.217519841479252e-06,
      "loss": 0.3812,
      "step": 8007
    },
    {
      "epoch": 0.5016050987331465,
      "grad_norm": 0.9269129404524993,
      "learning_rate": 5.2165063987261364e-06,
      "loss": 0.397,
      "step": 8008
    },
    {
      "epoch": 0.5016677367324888,
      "grad_norm": 0.8527603309041438,
      "learning_rate": 5.215492947061579e-06,
      "loss": 0.3825,
      "step": 8009
    },
    {
      "epoch": 0.501730374731831,
      "grad_norm": 0.8561274275261406,
      "learning_rate": 5.214479486527292e-06,
      "loss": 0.4132,
      "step": 8010
    },
    {
      "epoch": 0.5017930127311734,
      "grad_norm": 0.8779429906805598,
      "learning_rate": 5.213466017164988e-06,
      "loss": 0.4321,
      "step": 8011
    },
    {
      "epoch": 0.5018556507305156,
      "grad_norm": 0.8028248351682816,
      "learning_rate": 5.212452539016384e-06,
      "loss": 0.4156,
      "step": 8012
    },
    {
      "epoch": 0.501918288729858,
      "grad_norm": 0.8037583194441319,
      "learning_rate": 5.211439052123193e-06,
      "loss": 0.3672,
      "step": 8013
    },
    {
      "epoch": 0.5019809267292002,
      "grad_norm": 0.8611811905466134,
      "learning_rate": 5.210425556527133e-06,
      "loss": 0.4009,
      "step": 8014
    },
    {
      "epoch": 0.5020435647285426,
      "grad_norm": 0.857523677906572,
      "learning_rate": 5.209412052269917e-06,
      "loss": 0.3957,
      "step": 8015
    },
    {
      "epoch": 0.5021062027278849,
      "grad_norm": 0.8481795307747624,
      "learning_rate": 5.208398539393262e-06,
      "loss": 0.4023,
      "step": 8016
    },
    {
      "epoch": 0.5021688407272271,
      "grad_norm": 0.8107896379575389,
      "learning_rate": 5.207385017938884e-06,
      "loss": 0.3482,
      "step": 8017
    },
    {
      "epoch": 0.5022314787265695,
      "grad_norm": 0.8438547074022043,
      "learning_rate": 5.206371487948502e-06,
      "loss": 0.4063,
      "step": 8018
    },
    {
      "epoch": 0.5022941167259117,
      "grad_norm": 0.821161368259614,
      "learning_rate": 5.205357949463829e-06,
      "loss": 0.3826,
      "step": 8019
    },
    {
      "epoch": 0.5023567547252541,
      "grad_norm": 0.8744236883246292,
      "learning_rate": 5.204344402526586e-06,
      "loss": 0.4054,
      "step": 8020
    },
    {
      "epoch": 0.5024193927245963,
      "grad_norm": 0.8576207992013603,
      "learning_rate": 5.203330847178491e-06,
      "loss": 0.4163,
      "step": 8021
    },
    {
      "epoch": 0.5024820307239387,
      "grad_norm": 0.826617971725051,
      "learning_rate": 5.202317283461258e-06,
      "loss": 0.4063,
      "step": 8022
    },
    {
      "epoch": 0.502544668723281,
      "grad_norm": 0.8465921618698485,
      "learning_rate": 5.20130371141661e-06,
      "loss": 0.3797,
      "step": 8023
    },
    {
      "epoch": 0.5026073067226233,
      "grad_norm": 0.8187812185798657,
      "learning_rate": 5.200290131086262e-06,
      "loss": 0.4048,
      "step": 8024
    },
    {
      "epoch": 0.5026699447219656,
      "grad_norm": 0.8558601401323526,
      "learning_rate": 5.199276542511936e-06,
      "loss": 0.4512,
      "step": 8025
    },
    {
      "epoch": 0.5027325827213078,
      "grad_norm": 0.8249342012235099,
      "learning_rate": 5.19826294573535e-06,
      "loss": 0.4177,
      "step": 8026
    },
    {
      "epoch": 0.5027952207206502,
      "grad_norm": 0.7959262473278265,
      "learning_rate": 5.197249340798225e-06,
      "loss": 0.3903,
      "step": 8027
    },
    {
      "epoch": 0.5028578587199924,
      "grad_norm": 0.8393562640424916,
      "learning_rate": 5.19623572774228e-06,
      "loss": 0.3767,
      "step": 8028
    },
    {
      "epoch": 0.5029204967193348,
      "grad_norm": 0.9503802591640028,
      "learning_rate": 5.195222106609237e-06,
      "loss": 0.3921,
      "step": 8029
    },
    {
      "epoch": 0.5029831347186771,
      "grad_norm": 0.8823429411032977,
      "learning_rate": 5.194208477440815e-06,
      "loss": 0.4426,
      "step": 8030
    },
    {
      "epoch": 0.5030457727180194,
      "grad_norm": 0.8395143355612145,
      "learning_rate": 5.193194840278736e-06,
      "loss": 0.4259,
      "step": 8031
    },
    {
      "epoch": 0.5031084107173617,
      "grad_norm": 0.8882241876057857,
      "learning_rate": 5.192181195164721e-06,
      "loss": 0.4326,
      "step": 8032
    },
    {
      "epoch": 0.503171048716704,
      "grad_norm": 0.8159306140136473,
      "learning_rate": 5.191167542140494e-06,
      "loss": 0.4126,
      "step": 8033
    },
    {
      "epoch": 0.5032336867160463,
      "grad_norm": 0.664016546044879,
      "learning_rate": 5.190153881247775e-06,
      "loss": 0.4704,
      "step": 8034
    },
    {
      "epoch": 0.5032963247153885,
      "grad_norm": 0.8297950877206287,
      "learning_rate": 5.1891402125282874e-06,
      "loss": 0.3843,
      "step": 8035
    },
    {
      "epoch": 0.5033589627147309,
      "grad_norm": 0.8557950535592587,
      "learning_rate": 5.188126536023752e-06,
      "loss": 0.3668,
      "step": 8036
    },
    {
      "epoch": 0.5034216007140732,
      "grad_norm": 0.8725138012244125,
      "learning_rate": 5.187112851775893e-06,
      "loss": 0.4299,
      "step": 8037
    },
    {
      "epoch": 0.5034842387134155,
      "grad_norm": 0.7848800966670751,
      "learning_rate": 5.186099159826437e-06,
      "loss": 0.3772,
      "step": 8038
    },
    {
      "epoch": 0.5035468767127578,
      "grad_norm": 0.8771549081271235,
      "learning_rate": 5.185085460217103e-06,
      "loss": 0.4508,
      "step": 8039
    },
    {
      "epoch": 0.5036095147121001,
      "grad_norm": 0.8574122223061589,
      "learning_rate": 5.184071752989617e-06,
      "loss": 0.3817,
      "step": 8040
    },
    {
      "epoch": 0.5036721527114424,
      "grad_norm": 0.6820718466171694,
      "learning_rate": 5.1830580381857035e-06,
      "loss": 0.4745,
      "step": 8041
    },
    {
      "epoch": 0.5037347907107848,
      "grad_norm": 0.89903002884049,
      "learning_rate": 5.182044315847088e-06,
      "loss": 0.3913,
      "step": 8042
    },
    {
      "epoch": 0.503797428710127,
      "grad_norm": 0.7883772130849629,
      "learning_rate": 5.1810305860154954e-06,
      "loss": 0.3649,
      "step": 8043
    },
    {
      "epoch": 0.5038600667094693,
      "grad_norm": 0.8733375755069828,
      "learning_rate": 5.1800168487326484e-06,
      "loss": 0.3782,
      "step": 8044
    },
    {
      "epoch": 0.5039227047088116,
      "grad_norm": 0.8081392548035743,
      "learning_rate": 5.179003104040274e-06,
      "loss": 0.4168,
      "step": 8045
    },
    {
      "epoch": 0.5039853427081539,
      "grad_norm": 0.8570888128296615,
      "learning_rate": 5.1779893519801e-06,
      "loss": 0.4069,
      "step": 8046
    },
    {
      "epoch": 0.5040479807074962,
      "grad_norm": 0.8282308944934249,
      "learning_rate": 5.176975592593852e-06,
      "loss": 0.3991,
      "step": 8047
    },
    {
      "epoch": 0.5041106187068385,
      "grad_norm": 0.8454696445077061,
      "learning_rate": 5.175961825923256e-06,
      "loss": 0.4072,
      "step": 8048
    },
    {
      "epoch": 0.5041732567061809,
      "grad_norm": 0.8416495928378603,
      "learning_rate": 5.174948052010036e-06,
      "loss": 0.4255,
      "step": 8049
    },
    {
      "epoch": 0.5042358947055231,
      "grad_norm": 0.8035968177759322,
      "learning_rate": 5.173934270895924e-06,
      "loss": 0.3809,
      "step": 8050
    },
    {
      "epoch": 0.5042985327048654,
      "grad_norm": 0.7811145048950143,
      "learning_rate": 5.172920482622644e-06,
      "loss": 0.405,
      "step": 8051
    },
    {
      "epoch": 0.5043611707042077,
      "grad_norm": 0.7820981396056917,
      "learning_rate": 5.171906687231926e-06,
      "loss": 0.4173,
      "step": 8052
    },
    {
      "epoch": 0.50442380870355,
      "grad_norm": 0.9286241336374645,
      "learning_rate": 5.170892884765496e-06,
      "loss": 0.4505,
      "step": 8053
    },
    {
      "epoch": 0.5044864467028923,
      "grad_norm": 0.9315916847868658,
      "learning_rate": 5.169879075265082e-06,
      "loss": 0.4105,
      "step": 8054
    },
    {
      "epoch": 0.5045490847022346,
      "grad_norm": 0.8261978317948668,
      "learning_rate": 5.168865258772415e-06,
      "loss": 0.3585,
      "step": 8055
    },
    {
      "epoch": 0.504611722701577,
      "grad_norm": 0.8031583606995134,
      "learning_rate": 5.167851435329223e-06,
      "loss": 0.4132,
      "step": 8056
    },
    {
      "epoch": 0.5046743607009192,
      "grad_norm": 0.8174561561741606,
      "learning_rate": 5.166837604977234e-06,
      "loss": 0.4027,
      "step": 8057
    },
    {
      "epoch": 0.5047369987002616,
      "grad_norm": 0.6093826619876005,
      "learning_rate": 5.165823767758179e-06,
      "loss": 0.4428,
      "step": 8058
    },
    {
      "epoch": 0.5047996366996038,
      "grad_norm": 0.7888799784493576,
      "learning_rate": 5.164809923713786e-06,
      "loss": 0.4323,
      "step": 8059
    },
    {
      "epoch": 0.5048622746989461,
      "grad_norm": 0.8900370215194766,
      "learning_rate": 5.163796072885787e-06,
      "loss": 0.4152,
      "step": 8060
    },
    {
      "epoch": 0.5049249126982884,
      "grad_norm": 0.8364098438572981,
      "learning_rate": 5.1627822153159114e-06,
      "loss": 0.4003,
      "step": 8061
    },
    {
      "epoch": 0.5049875506976307,
      "grad_norm": 0.8749614036837053,
      "learning_rate": 5.161768351045888e-06,
      "loss": 0.4579,
      "step": 8062
    },
    {
      "epoch": 0.505050188696973,
      "grad_norm": 0.8120707595929619,
      "learning_rate": 5.160754480117449e-06,
      "loss": 0.3675,
      "step": 8063
    },
    {
      "epoch": 0.5051128266963153,
      "grad_norm": 0.8495317239529898,
      "learning_rate": 5.159740602572327e-06,
      "loss": 0.3944,
      "step": 8064
    },
    {
      "epoch": 0.5051754646956577,
      "grad_norm": 0.8364271787328391,
      "learning_rate": 5.158726718452254e-06,
      "loss": 0.4141,
      "step": 8065
    },
    {
      "epoch": 0.5052381026949999,
      "grad_norm": 0.8151853250910139,
      "learning_rate": 5.157712827798956e-06,
      "loss": 0.4367,
      "step": 8066
    },
    {
      "epoch": 0.5053007406943423,
      "grad_norm": 0.8125851261089694,
      "learning_rate": 5.156698930654171e-06,
      "loss": 0.3612,
      "step": 8067
    },
    {
      "epoch": 0.5053633786936845,
      "grad_norm": 0.836116566385592,
      "learning_rate": 5.155685027059628e-06,
      "loss": 0.3715,
      "step": 8068
    },
    {
      "epoch": 0.5054260166930268,
      "grad_norm": 0.8731375023853436,
      "learning_rate": 5.154671117057062e-06,
      "loss": 0.3901,
      "step": 8069
    },
    {
      "epoch": 0.5054886546923691,
      "grad_norm": 0.9355272388348169,
      "learning_rate": 5.1536572006882025e-06,
      "loss": 0.4074,
      "step": 8070
    },
    {
      "epoch": 0.5055512926917114,
      "grad_norm": 0.81317065166747,
      "learning_rate": 5.152643277994784e-06,
      "loss": 0.4168,
      "step": 8071
    },
    {
      "epoch": 0.5056139306910538,
      "grad_norm": 0.7794451784541079,
      "learning_rate": 5.151629349018541e-06,
      "loss": 0.409,
      "step": 8072
    },
    {
      "epoch": 0.505676568690396,
      "grad_norm": 0.8853980457621164,
      "learning_rate": 5.150615413801205e-06,
      "loss": 0.4377,
      "step": 8073
    },
    {
      "epoch": 0.5057392066897384,
      "grad_norm": 0.8364357103622432,
      "learning_rate": 5.1496014723845115e-06,
      "loss": 0.3847,
      "step": 8074
    },
    {
      "epoch": 0.5058018446890806,
      "grad_norm": 0.9111081856716149,
      "learning_rate": 5.148587524810192e-06,
      "loss": 0.4354,
      "step": 8075
    },
    {
      "epoch": 0.5058644826884229,
      "grad_norm": 0.8008319609454848,
      "learning_rate": 5.147573571119985e-06,
      "loss": 0.4053,
      "step": 8076
    },
    {
      "epoch": 0.5059271206877652,
      "grad_norm": 0.9004110357777888,
      "learning_rate": 5.146559611355619e-06,
      "loss": 0.4147,
      "step": 8077
    },
    {
      "epoch": 0.5059897586871075,
      "grad_norm": 0.87960291098056,
      "learning_rate": 5.145545645558835e-06,
      "loss": 0.4207,
      "step": 8078
    },
    {
      "epoch": 0.5060523966864499,
      "grad_norm": 0.7768895802117284,
      "learning_rate": 5.144531673771364e-06,
      "loss": 0.409,
      "step": 8079
    },
    {
      "epoch": 0.5061150346857921,
      "grad_norm": 1.2886105876281264,
      "learning_rate": 5.143517696034943e-06,
      "loss": 0.3849,
      "step": 8080
    },
    {
      "epoch": 0.5061776726851345,
      "grad_norm": 0.8154978771860472,
      "learning_rate": 5.142503712391307e-06,
      "loss": 0.424,
      "step": 8081
    },
    {
      "epoch": 0.5062403106844767,
      "grad_norm": 0.8038735797254247,
      "learning_rate": 5.141489722882191e-06,
      "loss": 0.367,
      "step": 8082
    },
    {
      "epoch": 0.5063029486838191,
      "grad_norm": 0.8357167331261031,
      "learning_rate": 5.140475727549332e-06,
      "loss": 0.3973,
      "step": 8083
    },
    {
      "epoch": 0.5063655866831613,
      "grad_norm": 0.8349711802998276,
      "learning_rate": 5.139461726434466e-06,
      "loss": 0.4304,
      "step": 8084
    },
    {
      "epoch": 0.5064282246825036,
      "grad_norm": 0.8548342644787248,
      "learning_rate": 5.1384477195793305e-06,
      "loss": 0.4479,
      "step": 8085
    },
    {
      "epoch": 0.506490862681846,
      "grad_norm": 0.83197188733113,
      "learning_rate": 5.13743370702566e-06,
      "loss": 0.4346,
      "step": 8086
    },
    {
      "epoch": 0.5065535006811882,
      "grad_norm": 0.8250549496756557,
      "learning_rate": 5.136419688815193e-06,
      "loss": 0.3859,
      "step": 8087
    },
    {
      "epoch": 0.5066161386805306,
      "grad_norm": 0.8116013772756776,
      "learning_rate": 5.135405664989666e-06,
      "loss": 0.3748,
      "step": 8088
    },
    {
      "epoch": 0.5066787766798728,
      "grad_norm": 0.84163557303631,
      "learning_rate": 5.134391635590817e-06,
      "loss": 0.4043,
      "step": 8089
    },
    {
      "epoch": 0.5067414146792152,
      "grad_norm": 0.7673559796250048,
      "learning_rate": 5.133377600660384e-06,
      "loss": 0.3762,
      "step": 8090
    },
    {
      "epoch": 0.5068040526785574,
      "grad_norm": 0.8257926747553206,
      "learning_rate": 5.132363560240103e-06,
      "loss": 0.3895,
      "step": 8091
    },
    {
      "epoch": 0.5068666906778998,
      "grad_norm": 0.7850492348659938,
      "learning_rate": 5.131349514371713e-06,
      "loss": 0.4238,
      "step": 8092
    },
    {
      "epoch": 0.506929328677242,
      "grad_norm": 0.7909102540572227,
      "learning_rate": 5.130335463096953e-06,
      "loss": 0.4208,
      "step": 8093
    },
    {
      "epoch": 0.5069919666765843,
      "grad_norm": 0.8867157074954916,
      "learning_rate": 5.129321406457562e-06,
      "loss": 0.4498,
      "step": 8094
    },
    {
      "epoch": 0.5070546046759267,
      "grad_norm": 0.8603970850642437,
      "learning_rate": 5.128307344495277e-06,
      "loss": 0.4471,
      "step": 8095
    },
    {
      "epoch": 0.5071172426752689,
      "grad_norm": 0.8225699522037913,
      "learning_rate": 5.127293277251838e-06,
      "loss": 0.3977,
      "step": 8096
    },
    {
      "epoch": 0.5071798806746113,
      "grad_norm": 0.8993782312048304,
      "learning_rate": 5.126279204768985e-06,
      "loss": 0.4245,
      "step": 8097
    },
    {
      "epoch": 0.5072425186739535,
      "grad_norm": 0.8807013207158939,
      "learning_rate": 5.125265127088457e-06,
      "loss": 0.4122,
      "step": 8098
    },
    {
      "epoch": 0.5073051566732959,
      "grad_norm": 0.8400532688221081,
      "learning_rate": 5.124251044251992e-06,
      "loss": 0.408,
      "step": 8099
    },
    {
      "epoch": 0.5073677946726382,
      "grad_norm": 0.8376484229720591,
      "learning_rate": 5.12323695630133e-06,
      "loss": 0.4198,
      "step": 8100
    },
    {
      "epoch": 0.5074304326719804,
      "grad_norm": 0.9398446725619658,
      "learning_rate": 5.1222228632782125e-06,
      "loss": 0.4253,
      "step": 8101
    },
    {
      "epoch": 0.5074930706713228,
      "grad_norm": 0.7814471523645199,
      "learning_rate": 5.12120876522438e-06,
      "loss": 0.396,
      "step": 8102
    },
    {
      "epoch": 0.507555708670665,
      "grad_norm": 0.8352154390525708,
      "learning_rate": 5.1201946621815726e-06,
      "loss": 0.4066,
      "step": 8103
    },
    {
      "epoch": 0.5076183466700074,
      "grad_norm": 0.8009129082129066,
      "learning_rate": 5.11918055419153e-06,
      "loss": 0.3641,
      "step": 8104
    },
    {
      "epoch": 0.5076809846693496,
      "grad_norm": 0.7927495789352963,
      "learning_rate": 5.1181664412959915e-06,
      "loss": 0.4111,
      "step": 8105
    },
    {
      "epoch": 0.507743622668692,
      "grad_norm": 0.8615003864246524,
      "learning_rate": 5.117152323536704e-06,
      "loss": 0.3976,
      "step": 8106
    },
    {
      "epoch": 0.5078062606680342,
      "grad_norm": 0.6234391516275531,
      "learning_rate": 5.116138200955404e-06,
      "loss": 0.4745,
      "step": 8107
    },
    {
      "epoch": 0.5078688986673766,
      "grad_norm": 0.8673996846538435,
      "learning_rate": 5.1151240735938335e-06,
      "loss": 0.4299,
      "step": 8108
    },
    {
      "epoch": 0.5079315366667189,
      "grad_norm": 0.8114425906826885,
      "learning_rate": 5.114109941493733e-06,
      "loss": 0.3912,
      "step": 8109
    },
    {
      "epoch": 0.5079941746660611,
      "grad_norm": 0.8511463312699007,
      "learning_rate": 5.11309580469685e-06,
      "loss": 0.4121,
      "step": 8110
    },
    {
      "epoch": 0.5080568126654035,
      "grad_norm": 0.9311957013962127,
      "learning_rate": 5.11208166324492e-06,
      "loss": 0.3836,
      "step": 8111
    },
    {
      "epoch": 0.5081194506647457,
      "grad_norm": 0.835675750815414,
      "learning_rate": 5.1110675171796894e-06,
      "loss": 0.397,
      "step": 8112
    },
    {
      "epoch": 0.5081820886640881,
      "grad_norm": 0.7747721306109212,
      "learning_rate": 5.110053366542897e-06,
      "loss": 0.3761,
      "step": 8113
    },
    {
      "epoch": 0.5082447266634303,
      "grad_norm": 0.8231672420442243,
      "learning_rate": 5.109039211376289e-06,
      "loss": 0.441,
      "step": 8114
    },
    {
      "epoch": 0.5083073646627727,
      "grad_norm": 0.8525045347629554,
      "learning_rate": 5.108025051721608e-06,
      "loss": 0.4036,
      "step": 8115
    },
    {
      "epoch": 0.508370002662115,
      "grad_norm": 0.8168524900142093,
      "learning_rate": 5.1070108876205945e-06,
      "loss": 0.42,
      "step": 8116
    },
    {
      "epoch": 0.5084326406614573,
      "grad_norm": 0.802805936511242,
      "learning_rate": 5.105996719114993e-06,
      "loss": 0.4165,
      "step": 8117
    },
    {
      "epoch": 0.5084952786607996,
      "grad_norm": 0.8682486002428388,
      "learning_rate": 5.104982546246546e-06,
      "loss": 0.4346,
      "step": 8118
    },
    {
      "epoch": 0.5085579166601418,
      "grad_norm": 0.8724789924733433,
      "learning_rate": 5.103968369057e-06,
      "loss": 0.4295,
      "step": 8119
    },
    {
      "epoch": 0.5086205546594842,
      "grad_norm": 0.8718102462129081,
      "learning_rate": 5.102954187588096e-06,
      "loss": 0.418,
      "step": 8120
    },
    {
      "epoch": 0.5086831926588264,
      "grad_norm": 0.8102590429913963,
      "learning_rate": 5.101940001881578e-06,
      "loss": 0.3984,
      "step": 8121
    },
    {
      "epoch": 0.5087458306581688,
      "grad_norm": 0.9130852680041658,
      "learning_rate": 5.100925811979191e-06,
      "loss": 0.4005,
      "step": 8122
    },
    {
      "epoch": 0.5088084686575111,
      "grad_norm": 0.8749224686347895,
      "learning_rate": 5.099911617922677e-06,
      "loss": 0.4371,
      "step": 8123
    },
    {
      "epoch": 0.5088711066568534,
      "grad_norm": 0.8844124138380799,
      "learning_rate": 5.0988974197537844e-06,
      "loss": 0.4015,
      "step": 8124
    },
    {
      "epoch": 0.5089337446561957,
      "grad_norm": 0.7804559421490065,
      "learning_rate": 5.097883217514255e-06,
      "loss": 0.4032,
      "step": 8125
    },
    {
      "epoch": 0.5089963826555379,
      "grad_norm": 0.8740727763685209,
      "learning_rate": 5.096869011245835e-06,
      "loss": 0.3849,
      "step": 8126
    },
    {
      "epoch": 0.5090590206548803,
      "grad_norm": 0.6334344695103602,
      "learning_rate": 5.095854800990267e-06,
      "loss": 0.4459,
      "step": 8127
    },
    {
      "epoch": 0.5091216586542225,
      "grad_norm": 0.8100561754916401,
      "learning_rate": 5.0948405867892996e-06,
      "loss": 0.398,
      "step": 8128
    },
    {
      "epoch": 0.5091842966535649,
      "grad_norm": 0.8910205252620531,
      "learning_rate": 5.093826368684675e-06,
      "loss": 0.3864,
      "step": 8129
    },
    {
      "epoch": 0.5092469346529072,
      "grad_norm": 0.850469688702369,
      "learning_rate": 5.092812146718138e-06,
      "loss": 0.4156,
      "step": 8130
    },
    {
      "epoch": 0.5093095726522495,
      "grad_norm": 0.8029490891357857,
      "learning_rate": 5.091797920931438e-06,
      "loss": 0.3941,
      "step": 8131
    },
    {
      "epoch": 0.5093722106515918,
      "grad_norm": 0.8123531563859379,
      "learning_rate": 5.090783691366318e-06,
      "loss": 0.3587,
      "step": 8132
    },
    {
      "epoch": 0.5094348486509341,
      "grad_norm": 0.860788174514449,
      "learning_rate": 5.089769458064524e-06,
      "loss": 0.3556,
      "step": 8133
    },
    {
      "epoch": 0.5094974866502764,
      "grad_norm": 0.8904094432488073,
      "learning_rate": 5.088755221067803e-06,
      "loss": 0.3506,
      "step": 8134
    },
    {
      "epoch": 0.5095601246496186,
      "grad_norm": 0.8668427724356307,
      "learning_rate": 5.0877409804179e-06,
      "loss": 0.4513,
      "step": 8135
    },
    {
      "epoch": 0.509622762648961,
      "grad_norm": 0.7714066750873813,
      "learning_rate": 5.086726736156563e-06,
      "loss": 0.3533,
      "step": 8136
    },
    {
      "epoch": 0.5096854006483033,
      "grad_norm": 0.809149306408102,
      "learning_rate": 5.085712488325536e-06,
      "loss": 0.4114,
      "step": 8137
    },
    {
      "epoch": 0.5097480386476456,
      "grad_norm": 0.820436395779765,
      "learning_rate": 5.0846982369665685e-06,
      "loss": 0.3845,
      "step": 8138
    },
    {
      "epoch": 0.5098106766469879,
      "grad_norm": 0.8558996049456354,
      "learning_rate": 5.083683982121404e-06,
      "loss": 0.4108,
      "step": 8139
    },
    {
      "epoch": 0.5098733146463302,
      "grad_norm": 0.8396233169620599,
      "learning_rate": 5.082669723831793e-06,
      "loss": 0.3787,
      "step": 8140
    },
    {
      "epoch": 0.5099359526456725,
      "grad_norm": 0.7975123818311078,
      "learning_rate": 5.08165546213948e-06,
      "loss": 0.3815,
      "step": 8141
    },
    {
      "epoch": 0.5099985906450148,
      "grad_norm": 0.7886711040506519,
      "learning_rate": 5.080641197086214e-06,
      "loss": 0.3845,
      "step": 8142
    },
    {
      "epoch": 0.5100612286443571,
      "grad_norm": 0.853168662054745,
      "learning_rate": 5.07962692871374e-06,
      "loss": 0.4329,
      "step": 8143
    },
    {
      "epoch": 0.5101238666436994,
      "grad_norm": 0.7981870658412051,
      "learning_rate": 5.0786126570638074e-06,
      "loss": 0.3701,
      "step": 8144
    },
    {
      "epoch": 0.5101865046430417,
      "grad_norm": 0.8891153115638999,
      "learning_rate": 5.077598382178164e-06,
      "loss": 0.3688,
      "step": 8145
    },
    {
      "epoch": 0.510249142642384,
      "grad_norm": 0.7963331447370106,
      "learning_rate": 5.076584104098555e-06,
      "loss": 0.3853,
      "step": 8146
    },
    {
      "epoch": 0.5103117806417263,
      "grad_norm": 0.8144980103500984,
      "learning_rate": 5.075569822866731e-06,
      "loss": 0.3645,
      "step": 8147
    },
    {
      "epoch": 0.5103744186410686,
      "grad_norm": 0.8504132906616959,
      "learning_rate": 5.0745555385244395e-06,
      "loss": 0.3928,
      "step": 8148
    },
    {
      "epoch": 0.510437056640411,
      "grad_norm": 0.8137162521695348,
      "learning_rate": 5.073541251113428e-06,
      "loss": 0.4101,
      "step": 8149
    },
    {
      "epoch": 0.5104996946397532,
      "grad_norm": 0.8632122197435231,
      "learning_rate": 5.072526960675446e-06,
      "loss": 0.3858,
      "step": 8150
    },
    {
      "epoch": 0.5105623326390956,
      "grad_norm": 0.8291133635605801,
      "learning_rate": 5.071512667252239e-06,
      "loss": 0.3919,
      "step": 8151
    },
    {
      "epoch": 0.5106249706384378,
      "grad_norm": 0.8211168111674864,
      "learning_rate": 5.0704983708855575e-06,
      "loss": 0.4086,
      "step": 8152
    },
    {
      "epoch": 0.5106876086377801,
      "grad_norm": 0.7772460819712467,
      "learning_rate": 5.0694840716171514e-06,
      "loss": 0.3867,
      "step": 8153
    },
    {
      "epoch": 0.5107502466371224,
      "grad_norm": 0.8242500236360439,
      "learning_rate": 5.068469769488767e-06,
      "loss": 0.4106,
      "step": 8154
    },
    {
      "epoch": 0.5108128846364647,
      "grad_norm": 0.8718000425989151,
      "learning_rate": 5.067455464542154e-06,
      "loss": 0.3872,
      "step": 8155
    },
    {
      "epoch": 0.510875522635807,
      "grad_norm": 0.8212499481755572,
      "learning_rate": 5.066441156819061e-06,
      "loss": 0.3712,
      "step": 8156
    },
    {
      "epoch": 0.5109381606351493,
      "grad_norm": 0.8123176945494615,
      "learning_rate": 5.065426846361239e-06,
      "loss": 0.394,
      "step": 8157
    },
    {
      "epoch": 0.5110007986344917,
      "grad_norm": 0.8155692014466943,
      "learning_rate": 5.064412533210436e-06,
      "loss": 0.3838,
      "step": 8158
    },
    {
      "epoch": 0.5110634366338339,
      "grad_norm": 0.7717409647903027,
      "learning_rate": 5.063398217408401e-06,
      "loss": 0.4015,
      "step": 8159
    },
    {
      "epoch": 0.5111260746331762,
      "grad_norm": 0.7916534049254282,
      "learning_rate": 5.062383898996881e-06,
      "loss": 0.4542,
      "step": 8160
    },
    {
      "epoch": 0.5111887126325185,
      "grad_norm": 0.8447602763841638,
      "learning_rate": 5.061369578017631e-06,
      "loss": 0.3978,
      "step": 8161
    },
    {
      "epoch": 0.5112513506318608,
      "grad_norm": 0.8442638416411274,
      "learning_rate": 5.060355254512398e-06,
      "loss": 0.3851,
      "step": 8162
    },
    {
      "epoch": 0.5113139886312031,
      "grad_norm": 0.8376393770252213,
      "learning_rate": 5.059340928522931e-06,
      "loss": 0.4026,
      "step": 8163
    },
    {
      "epoch": 0.5113766266305454,
      "grad_norm": 0.9284692312567296,
      "learning_rate": 5.058326600090979e-06,
      "loss": 0.4269,
      "step": 8164
    },
    {
      "epoch": 0.5114392646298878,
      "grad_norm": 0.759287628163956,
      "learning_rate": 5.0573122692582944e-06,
      "loss": 0.3937,
      "step": 8165
    },
    {
      "epoch": 0.51150190262923,
      "grad_norm": 0.7953324350140922,
      "learning_rate": 5.056297936066626e-06,
      "loss": 0.4346,
      "step": 8166
    },
    {
      "epoch": 0.5115645406285724,
      "grad_norm": 0.7853856801386242,
      "learning_rate": 5.055283600557724e-06,
      "loss": 0.419,
      "step": 8167
    },
    {
      "epoch": 0.5116271786279146,
      "grad_norm": 0.8257065895466549,
      "learning_rate": 5.054269262773339e-06,
      "loss": 0.3682,
      "step": 8168
    },
    {
      "epoch": 0.5116898166272569,
      "grad_norm": 0.8418582609388429,
      "learning_rate": 5.053254922755221e-06,
      "loss": 0.4184,
      "step": 8169
    },
    {
      "epoch": 0.5117524546265992,
      "grad_norm": 0.7940881567032659,
      "learning_rate": 5.05224058054512e-06,
      "loss": 0.4226,
      "step": 8170
    },
    {
      "epoch": 0.5118150926259415,
      "grad_norm": 0.9060713627497136,
      "learning_rate": 5.051226236184789e-06,
      "loss": 0.4397,
      "step": 8171
    },
    {
      "epoch": 0.5118777306252839,
      "grad_norm": 0.8187363233897705,
      "learning_rate": 5.050211889715975e-06,
      "loss": 0.3751,
      "step": 8172
    },
    {
      "epoch": 0.5119403686246261,
      "grad_norm": 0.7893762629117332,
      "learning_rate": 5.049197541180429e-06,
      "loss": 0.3599,
      "step": 8173
    },
    {
      "epoch": 0.5120030066239685,
      "grad_norm": 0.8780007790526597,
      "learning_rate": 5.048183190619904e-06,
      "loss": 0.4165,
      "step": 8174
    },
    {
      "epoch": 0.5120656446233107,
      "grad_norm": 0.6004435776961486,
      "learning_rate": 5.04716883807615e-06,
      "loss": 0.4752,
      "step": 8175
    },
    {
      "epoch": 0.5121282826226531,
      "grad_norm": 0.8325553783440437,
      "learning_rate": 5.04615448359092e-06,
      "loss": 0.4573,
      "step": 8176
    },
    {
      "epoch": 0.5121909206219953,
      "grad_norm": 0.8325072703192636,
      "learning_rate": 5.045140127205961e-06,
      "loss": 0.4183,
      "step": 8177
    },
    {
      "epoch": 0.5122535586213376,
      "grad_norm": 0.793268313924038,
      "learning_rate": 5.044125768963025e-06,
      "loss": 0.4082,
      "step": 8178
    },
    {
      "epoch": 0.51231619662068,
      "grad_norm": 0.830256471234238,
      "learning_rate": 5.043111408903865e-06,
      "loss": 0.4293,
      "step": 8179
    },
    {
      "epoch": 0.5123788346200222,
      "grad_norm": 0.7990623442745683,
      "learning_rate": 5.042097047070232e-06,
      "loss": 0.3673,
      "step": 8180
    },
    {
      "epoch": 0.5124414726193646,
      "grad_norm": 0.8125581975866385,
      "learning_rate": 5.041082683503876e-06,
      "loss": 0.4071,
      "step": 8181
    },
    {
      "epoch": 0.5125041106187068,
      "grad_norm": 0.7503788488776666,
      "learning_rate": 5.04006831824655e-06,
      "loss": 0.3525,
      "step": 8182
    },
    {
      "epoch": 0.5125667486180492,
      "grad_norm": 0.7972272407324117,
      "learning_rate": 5.039053951340004e-06,
      "loss": 0.4058,
      "step": 8183
    },
    {
      "epoch": 0.5126293866173914,
      "grad_norm": 0.8427175982523937,
      "learning_rate": 5.038039582825991e-06,
      "loss": 0.3419,
      "step": 8184
    },
    {
      "epoch": 0.5126920246167337,
      "grad_norm": 0.809310962822573,
      "learning_rate": 5.037025212746261e-06,
      "loss": 0.4036,
      "step": 8185
    },
    {
      "epoch": 0.512754662616076,
      "grad_norm": 0.7925747545242319,
      "learning_rate": 5.036010841142565e-06,
      "loss": 0.3735,
      "step": 8186
    },
    {
      "epoch": 0.5128173006154183,
      "grad_norm": 0.8506693738981296,
      "learning_rate": 5.034996468056658e-06,
      "loss": 0.4267,
      "step": 8187
    },
    {
      "epoch": 0.5128799386147607,
      "grad_norm": 0.8242075153267425,
      "learning_rate": 5.03398209353029e-06,
      "loss": 0.3769,
      "step": 8188
    },
    {
      "epoch": 0.5129425766141029,
      "grad_norm": 0.7851733855253988,
      "learning_rate": 5.032967717605211e-06,
      "loss": 0.3929,
      "step": 8189
    },
    {
      "epoch": 0.5130052146134453,
      "grad_norm": 0.593507373843471,
      "learning_rate": 5.031953340323176e-06,
      "loss": 0.4359,
      "step": 8190
    },
    {
      "epoch": 0.5130678526127875,
      "grad_norm": 0.8742871694148957,
      "learning_rate": 5.030938961725937e-06,
      "loss": 0.4573,
      "step": 8191
    },
    {
      "epoch": 0.5131304906121299,
      "grad_norm": 0.7764141209221231,
      "learning_rate": 5.029924581855241e-06,
      "loss": 0.3808,
      "step": 8192
    },
    {
      "epoch": 0.5131931286114721,
      "grad_norm": 0.8474582411315787,
      "learning_rate": 5.028910200752845e-06,
      "loss": 0.3876,
      "step": 8193
    },
    {
      "epoch": 0.5132557666108144,
      "grad_norm": 0.8155711101107981,
      "learning_rate": 5.027895818460501e-06,
      "loss": 0.4103,
      "step": 8194
    },
    {
      "epoch": 0.5133184046101568,
      "grad_norm": 0.7849134527702236,
      "learning_rate": 5.026881435019959e-06,
      "loss": 0.4048,
      "step": 8195
    },
    {
      "epoch": 0.513381042609499,
      "grad_norm": 1.147885338187495,
      "learning_rate": 5.025867050472973e-06,
      "loss": 0.37,
      "step": 8196
    },
    {
      "epoch": 0.5134436806088414,
      "grad_norm": 0.8158199460703229,
      "learning_rate": 5.024852664861293e-06,
      "loss": 0.3596,
      "step": 8197
    },
    {
      "epoch": 0.5135063186081836,
      "grad_norm": 0.8838762368246317,
      "learning_rate": 5.023838278226674e-06,
      "loss": 0.4684,
      "step": 8198
    },
    {
      "epoch": 0.513568956607526,
      "grad_norm": 0.8026550697076904,
      "learning_rate": 5.022823890610866e-06,
      "loss": 0.3802,
      "step": 8199
    },
    {
      "epoch": 0.5136315946068682,
      "grad_norm": 0.7817695767275606,
      "learning_rate": 5.021809502055624e-06,
      "loss": 0.3999,
      "step": 8200
    },
    {
      "epoch": 0.5136942326062106,
      "grad_norm": 0.7985235231883355,
      "learning_rate": 5.0207951126026965e-06,
      "loss": 0.3915,
      "step": 8201
    },
    {
      "epoch": 0.5137568706055529,
      "grad_norm": 0.8446644384406377,
      "learning_rate": 5.01978072229384e-06,
      "loss": 0.447,
      "step": 8202
    },
    {
      "epoch": 0.5138195086048951,
      "grad_norm": 0.7712975427400917,
      "learning_rate": 5.018766331170804e-06,
      "loss": 0.3877,
      "step": 8203
    },
    {
      "epoch": 0.5138821466042375,
      "grad_norm": 0.8704864603668108,
      "learning_rate": 5.017751939275344e-06,
      "loss": 0.4311,
      "step": 8204
    },
    {
      "epoch": 0.5139447846035797,
      "grad_norm": 0.876030647378853,
      "learning_rate": 5.016737546649211e-06,
      "loss": 0.4032,
      "step": 8205
    },
    {
      "epoch": 0.5140074226029221,
      "grad_norm": 0.8380114615861824,
      "learning_rate": 5.015723153334156e-06,
      "loss": 0.4364,
      "step": 8206
    },
    {
      "epoch": 0.5140700606022643,
      "grad_norm": 0.9084707108747004,
      "learning_rate": 5.014708759371933e-06,
      "loss": 0.4539,
      "step": 8207
    },
    {
      "epoch": 0.5141326986016067,
      "grad_norm": 0.687212550392713,
      "learning_rate": 5.0136943648042945e-06,
      "loss": 0.4527,
      "step": 8208
    },
    {
      "epoch": 0.514195336600949,
      "grad_norm": 0.837037220347326,
      "learning_rate": 5.012679969672996e-06,
      "loss": 0.3772,
      "step": 8209
    },
    {
      "epoch": 0.5142579746002912,
      "grad_norm": 0.8254889530425422,
      "learning_rate": 5.011665574019785e-06,
      "loss": 0.3781,
      "step": 8210
    },
    {
      "epoch": 0.5143206125996336,
      "grad_norm": 0.8519098050777713,
      "learning_rate": 5.010651177886418e-06,
      "loss": 0.4309,
      "step": 8211
    },
    {
      "epoch": 0.5143832505989758,
      "grad_norm": 0.8600762942264881,
      "learning_rate": 5.009636781314645e-06,
      "loss": 0.3941,
      "step": 8212
    },
    {
      "epoch": 0.5144458885983182,
      "grad_norm": 0.9065036575879774,
      "learning_rate": 5.008622384346223e-06,
      "loss": 0.4462,
      "step": 8213
    },
    {
      "epoch": 0.5145085265976604,
      "grad_norm": 0.8614397836200416,
      "learning_rate": 5.0076079870229e-06,
      "loss": 0.3942,
      "step": 8214
    },
    {
      "epoch": 0.5145711645970028,
      "grad_norm": 0.8170226229385308,
      "learning_rate": 5.006593589386432e-06,
      "loss": 0.4099,
      "step": 8215
    },
    {
      "epoch": 0.5146338025963451,
      "grad_norm": 0.9140444540926199,
      "learning_rate": 5.005579191478569e-06,
      "loss": 0.4493,
      "step": 8216
    },
    {
      "epoch": 0.5146964405956874,
      "grad_norm": 0.7996468026087599,
      "learning_rate": 5.004564793341067e-06,
      "loss": 0.3996,
      "step": 8217
    },
    {
      "epoch": 0.5147590785950297,
      "grad_norm": 0.8302413625404509,
      "learning_rate": 5.003550395015677e-06,
      "loss": 0.4069,
      "step": 8218
    },
    {
      "epoch": 0.5148217165943719,
      "grad_norm": 0.8799563870785911,
      "learning_rate": 5.002535996544151e-06,
      "loss": 0.4107,
      "step": 8219
    },
    {
      "epoch": 0.5148843545937143,
      "grad_norm": 0.6025268732464654,
      "learning_rate": 5.001521597968243e-06,
      "loss": 0.4449,
      "step": 8220
    },
    {
      "epoch": 0.5149469925930565,
      "grad_norm": 0.838228644515971,
      "learning_rate": 5.000507199329707e-06,
      "loss": 0.4103,
      "step": 8221
    },
    {
      "epoch": 0.5150096305923989,
      "grad_norm": 0.9011588226587794,
      "learning_rate": 4.999492800670295e-06,
      "loss": 0.3787,
      "step": 8222
    },
    {
      "epoch": 0.5150722685917412,
      "grad_norm": 0.9211265470495409,
      "learning_rate": 4.9984784020317585e-06,
      "loss": 0.3875,
      "step": 8223
    },
    {
      "epoch": 0.5151349065910835,
      "grad_norm": 0.8336764394111199,
      "learning_rate": 4.997464003455851e-06,
      "loss": 0.3745,
      "step": 8224
    },
    {
      "epoch": 0.5151975445904258,
      "grad_norm": 0.8568850954855144,
      "learning_rate": 4.996449604984326e-06,
      "loss": 0.4333,
      "step": 8225
    },
    {
      "epoch": 0.5152601825897681,
      "grad_norm": 0.8192524118220557,
      "learning_rate": 4.995435206658936e-06,
      "loss": 0.4029,
      "step": 8226
    },
    {
      "epoch": 0.5153228205891104,
      "grad_norm": 0.9186246423189697,
      "learning_rate": 4.994420808521432e-06,
      "loss": 0.4355,
      "step": 8227
    },
    {
      "epoch": 0.5153854585884526,
      "grad_norm": 0.842169111792994,
      "learning_rate": 4.993406410613571e-06,
      "loss": 0.4247,
      "step": 8228
    },
    {
      "epoch": 0.515448096587795,
      "grad_norm": 0.8151011012291627,
      "learning_rate": 4.992392012977101e-06,
      "loss": 0.433,
      "step": 8229
    },
    {
      "epoch": 0.5155107345871373,
      "grad_norm": 0.6835126937199044,
      "learning_rate": 4.991377615653779e-06,
      "loss": 0.4605,
      "step": 8230
    },
    {
      "epoch": 0.5155733725864796,
      "grad_norm": 0.8292155460650013,
      "learning_rate": 4.990363218685356e-06,
      "loss": 0.4249,
      "step": 8231
    },
    {
      "epoch": 0.5156360105858219,
      "grad_norm": 0.9072666740665797,
      "learning_rate": 4.989348822113584e-06,
      "loss": 0.3906,
      "step": 8232
    },
    {
      "epoch": 0.5156986485851642,
      "grad_norm": 0.8459255112664072,
      "learning_rate": 4.988334425980217e-06,
      "loss": 0.3923,
      "step": 8233
    },
    {
      "epoch": 0.5157612865845065,
      "grad_norm": 0.8106019197128435,
      "learning_rate": 4.987320030327007e-06,
      "loss": 0.3827,
      "step": 8234
    },
    {
      "epoch": 0.5158239245838488,
      "grad_norm": 0.8885156916622667,
      "learning_rate": 4.986305635195707e-06,
      "loss": 0.469,
      "step": 8235
    },
    {
      "epoch": 0.5158865625831911,
      "grad_norm": 0.9150318129800625,
      "learning_rate": 4.985291240628068e-06,
      "loss": 0.4355,
      "step": 8236
    },
    {
      "epoch": 0.5159492005825334,
      "grad_norm": 0.827923641617823,
      "learning_rate": 4.984276846665846e-06,
      "loss": 0.4019,
      "step": 8237
    },
    {
      "epoch": 0.5160118385818757,
      "grad_norm": 0.827646843631745,
      "learning_rate": 4.983262453350791e-06,
      "loss": 0.4291,
      "step": 8238
    },
    {
      "epoch": 0.516074476581218,
      "grad_norm": 0.7756214692605201,
      "learning_rate": 4.982248060724658e-06,
      "loss": 0.3976,
      "step": 8239
    },
    {
      "epoch": 0.5161371145805603,
      "grad_norm": 0.7763333087568015,
      "learning_rate": 4.981233668829197e-06,
      "loss": 0.4072,
      "step": 8240
    },
    {
      "epoch": 0.5161997525799026,
      "grad_norm": 0.8084699117763976,
      "learning_rate": 4.980219277706162e-06,
      "loss": 0.4768,
      "step": 8241
    },
    {
      "epoch": 0.516262390579245,
      "grad_norm": 0.8752254011324899,
      "learning_rate": 4.979204887397305e-06,
      "loss": 0.4099,
      "step": 8242
    },
    {
      "epoch": 0.5163250285785872,
      "grad_norm": 0.8828522677098437,
      "learning_rate": 4.978190497944379e-06,
      "loss": 0.4244,
      "step": 8243
    },
    {
      "epoch": 0.5163876665779294,
      "grad_norm": 0.7901172044364118,
      "learning_rate": 4.977176109389137e-06,
      "loss": 0.373,
      "step": 8244
    },
    {
      "epoch": 0.5164503045772718,
      "grad_norm": 0.6143202859095989,
      "learning_rate": 4.976161721773327e-06,
      "loss": 0.4346,
      "step": 8245
    },
    {
      "epoch": 0.5165129425766141,
      "grad_norm": 0.6291407398036827,
      "learning_rate": 4.975147335138707e-06,
      "loss": 0.4865,
      "step": 8246
    },
    {
      "epoch": 0.5165755805759564,
      "grad_norm": 0.8643768448732758,
      "learning_rate": 4.9741329495270285e-06,
      "loss": 0.3993,
      "step": 8247
    },
    {
      "epoch": 0.5166382185752987,
      "grad_norm": 0.8057230066816489,
      "learning_rate": 4.973118564980042e-06,
      "loss": 0.4196,
      "step": 8248
    },
    {
      "epoch": 0.516700856574641,
      "grad_norm": 0.81420651305386,
      "learning_rate": 4.972104181539501e-06,
      "loss": 0.4012,
      "step": 8249
    },
    {
      "epoch": 0.5167634945739833,
      "grad_norm": 0.7611102643474783,
      "learning_rate": 4.971089799247155e-06,
      "loss": 0.3851,
      "step": 8250
    },
    {
      "epoch": 0.5168261325733257,
      "grad_norm": 0.8042271439111526,
      "learning_rate": 4.97007541814476e-06,
      "loss": 0.3891,
      "step": 8251
    },
    {
      "epoch": 0.5168887705726679,
      "grad_norm": 0.9401912685682359,
      "learning_rate": 4.969061038274067e-06,
      "loss": 0.4212,
      "step": 8252
    },
    {
      "epoch": 0.5169514085720102,
      "grad_norm": 0.8215498967807255,
      "learning_rate": 4.968046659676825e-06,
      "loss": 0.42,
      "step": 8253
    },
    {
      "epoch": 0.5170140465713525,
      "grad_norm": 0.8352947932288015,
      "learning_rate": 4.967032282394789e-06,
      "loss": 0.4099,
      "step": 8254
    },
    {
      "epoch": 0.5170766845706948,
      "grad_norm": 0.7974978969511005,
      "learning_rate": 4.966017906469711e-06,
      "loss": 0.3793,
      "step": 8255
    },
    {
      "epoch": 0.5171393225700371,
      "grad_norm": 0.6616749435423456,
      "learning_rate": 4.965003531943342e-06,
      "loss": 0.4612,
      "step": 8256
    },
    {
      "epoch": 0.5172019605693794,
      "grad_norm": 0.832357808452164,
      "learning_rate": 4.9639891588574355e-06,
      "loss": 0.3959,
      "step": 8257
    },
    {
      "epoch": 0.5172645985687218,
      "grad_norm": 0.8091187360376992,
      "learning_rate": 4.96297478725374e-06,
      "loss": 0.3975,
      "step": 8258
    },
    {
      "epoch": 0.517327236568064,
      "grad_norm": 0.815399737765134,
      "learning_rate": 4.96196041717401e-06,
      "loss": 0.401,
      "step": 8259
    },
    {
      "epoch": 0.5173898745674064,
      "grad_norm": 0.79173461999385,
      "learning_rate": 4.9609460486599975e-06,
      "loss": 0.3727,
      "step": 8260
    },
    {
      "epoch": 0.5174525125667486,
      "grad_norm": 0.889254010581241,
      "learning_rate": 4.959931681753452e-06,
      "loss": 0.4452,
      "step": 8261
    },
    {
      "epoch": 0.5175151505660909,
      "grad_norm": 0.8605392512603198,
      "learning_rate": 4.958917316496125e-06,
      "loss": 0.384,
      "step": 8262
    },
    {
      "epoch": 0.5175777885654332,
      "grad_norm": 0.8456699344476227,
      "learning_rate": 4.957902952929767e-06,
      "loss": 0.4113,
      "step": 8263
    },
    {
      "epoch": 0.5176404265647755,
      "grad_norm": 0.8433476854099257,
      "learning_rate": 4.9568885910961346e-06,
      "loss": 0.4131,
      "step": 8264
    },
    {
      "epoch": 0.5177030645641179,
      "grad_norm": 0.8340628488471051,
      "learning_rate": 4.955874231036976e-06,
      "loss": 0.4,
      "step": 8265
    },
    {
      "epoch": 0.5177657025634601,
      "grad_norm": 0.6931030114538771,
      "learning_rate": 4.954859872794041e-06,
      "loss": 0.4578,
      "step": 8266
    },
    {
      "epoch": 0.5178283405628025,
      "grad_norm": 0.8011190885217403,
      "learning_rate": 4.953845516409082e-06,
      "loss": 0.4212,
      "step": 8267
    },
    {
      "epoch": 0.5178909785621447,
      "grad_norm": 0.8512431700461776,
      "learning_rate": 4.9528311619238505e-06,
      "loss": 0.4343,
      "step": 8268
    },
    {
      "epoch": 0.517953616561487,
      "grad_norm": 0.7682427802294428,
      "learning_rate": 4.951816809380098e-06,
      "loss": 0.3841,
      "step": 8269
    },
    {
      "epoch": 0.5180162545608293,
      "grad_norm": 0.6146662680002114,
      "learning_rate": 4.950802458819573e-06,
      "loss": 0.4445,
      "step": 8270
    },
    {
      "epoch": 0.5180788925601716,
      "grad_norm": 0.8509936467588098,
      "learning_rate": 4.949788110284028e-06,
      "loss": 0.4005,
      "step": 8271
    },
    {
      "epoch": 0.518141530559514,
      "grad_norm": 0.8364949678758272,
      "learning_rate": 4.948773763815215e-06,
      "loss": 0.4072,
      "step": 8272
    },
    {
      "epoch": 0.5182041685588562,
      "grad_norm": 0.831107904886415,
      "learning_rate": 4.94775941945488e-06,
      "loss": 0.4135,
      "step": 8273
    },
    {
      "epoch": 0.5182668065581986,
      "grad_norm": 0.8261837197946177,
      "learning_rate": 4.94674507724478e-06,
      "loss": 0.3757,
      "step": 8274
    },
    {
      "epoch": 0.5183294445575408,
      "grad_norm": 0.8143575860522001,
      "learning_rate": 4.9457307372266626e-06,
      "loss": 0.3793,
      "step": 8275
    },
    {
      "epoch": 0.5183920825568832,
      "grad_norm": 0.9110989653488255,
      "learning_rate": 4.944716399442277e-06,
      "loss": 0.4336,
      "step": 8276
    },
    {
      "epoch": 0.5184547205562254,
      "grad_norm": 0.8356966059248633,
      "learning_rate": 4.943702063933375e-06,
      "loss": 0.3502,
      "step": 8277
    },
    {
      "epoch": 0.5185173585555677,
      "grad_norm": 0.7350057760555081,
      "learning_rate": 4.942687730741707e-06,
      "loss": 0.4641,
      "step": 8278
    },
    {
      "epoch": 0.51857999655491,
      "grad_norm": 0.8851145211257363,
      "learning_rate": 4.9416733999090224e-06,
      "loss": 0.4187,
      "step": 8279
    },
    {
      "epoch": 0.5186426345542523,
      "grad_norm": 0.8371013148767406,
      "learning_rate": 4.940659071477071e-06,
      "loss": 0.3954,
      "step": 8280
    },
    {
      "epoch": 0.5187052725535947,
      "grad_norm": 0.7936648583949438,
      "learning_rate": 4.939644745487605e-06,
      "loss": 0.3954,
      "step": 8281
    },
    {
      "epoch": 0.5187679105529369,
      "grad_norm": 0.7015726917456452,
      "learning_rate": 4.938630421982369e-06,
      "loss": 0.4615,
      "step": 8282
    },
    {
      "epoch": 0.5188305485522793,
      "grad_norm": 0.8435321880558735,
      "learning_rate": 4.937616101003119e-06,
      "loss": 0.3821,
      "step": 8283
    },
    {
      "epoch": 0.5188931865516215,
      "grad_norm": 0.8485357330768095,
      "learning_rate": 4.9366017825916e-06,
      "loss": 0.4147,
      "step": 8284
    },
    {
      "epoch": 0.5189558245509639,
      "grad_norm": 0.8265796330554315,
      "learning_rate": 4.935587466789565e-06,
      "loss": 0.3552,
      "step": 8285
    },
    {
      "epoch": 0.5190184625503061,
      "grad_norm": 0.7675391020301982,
      "learning_rate": 4.9345731536387624e-06,
      "loss": 0.4196,
      "step": 8286
    },
    {
      "epoch": 0.5190811005496484,
      "grad_norm": 0.8926290526553952,
      "learning_rate": 4.93355884318094e-06,
      "loss": 0.4004,
      "step": 8287
    },
    {
      "epoch": 0.5191437385489908,
      "grad_norm": 0.804698253728434,
      "learning_rate": 4.932544535457848e-06,
      "loss": 0.4282,
      "step": 8288
    },
    {
      "epoch": 0.519206376548333,
      "grad_norm": 0.92177242013466,
      "learning_rate": 4.931530230511236e-06,
      "loss": 0.4409,
      "step": 8289
    },
    {
      "epoch": 0.5192690145476754,
      "grad_norm": 0.8215425522240162,
      "learning_rate": 4.930515928382852e-06,
      "loss": 0.3989,
      "step": 8290
    },
    {
      "epoch": 0.5193316525470176,
      "grad_norm": 0.8331119500621229,
      "learning_rate": 4.9295016291144425e-06,
      "loss": 0.4129,
      "step": 8291
    },
    {
      "epoch": 0.51939429054636,
      "grad_norm": 0.8321720908750989,
      "learning_rate": 4.928487332747762e-06,
      "loss": 0.3826,
      "step": 8292
    },
    {
      "epoch": 0.5194569285457022,
      "grad_norm": 0.9172825962983301,
      "learning_rate": 4.927473039324556e-06,
      "loss": 0.4167,
      "step": 8293
    },
    {
      "epoch": 0.5195195665450445,
      "grad_norm": 0.8981800452771517,
      "learning_rate": 4.9264587488865725e-06,
      "loss": 0.3965,
      "step": 8294
    },
    {
      "epoch": 0.5195822045443869,
      "grad_norm": 0.7825633850341278,
      "learning_rate": 4.925444461475562e-06,
      "loss": 0.423,
      "step": 8295
    },
    {
      "epoch": 0.5196448425437291,
      "grad_norm": 0.8813045588858088,
      "learning_rate": 4.92443017713327e-06,
      "loss": 0.428,
      "step": 8296
    },
    {
      "epoch": 0.5197074805430715,
      "grad_norm": 0.8532770034252974,
      "learning_rate": 4.923415895901446e-06,
      "loss": 0.4306,
      "step": 8297
    },
    {
      "epoch": 0.5197701185424137,
      "grad_norm": 0.86377767924659,
      "learning_rate": 4.922401617821839e-06,
      "loss": 0.4418,
      "step": 8298
    },
    {
      "epoch": 0.5198327565417561,
      "grad_norm": 0.8148973039925508,
      "learning_rate": 4.921387342936195e-06,
      "loss": 0.4157,
      "step": 8299
    },
    {
      "epoch": 0.5198953945410983,
      "grad_norm": 0.821509069593858,
      "learning_rate": 4.9203730712862605e-06,
      "loss": 0.3929,
      "step": 8300
    },
    {
      "epoch": 0.5199580325404407,
      "grad_norm": 0.6552489831018905,
      "learning_rate": 4.919358802913788e-06,
      "loss": 0.4901,
      "step": 8301
    },
    {
      "epoch": 0.520020670539783,
      "grad_norm": 0.8431769067183542,
      "learning_rate": 4.91834453786052e-06,
      "loss": 0.3825,
      "step": 8302
    },
    {
      "epoch": 0.5200833085391252,
      "grad_norm": 0.7598506183319799,
      "learning_rate": 4.917330276168208e-06,
      "loss": 0.345,
      "step": 8303
    },
    {
      "epoch": 0.5201459465384676,
      "grad_norm": 0.8376127005486502,
      "learning_rate": 4.9163160178785974e-06,
      "loss": 0.4103,
      "step": 8304
    },
    {
      "epoch": 0.5202085845378098,
      "grad_norm": 0.9141167011673258,
      "learning_rate": 4.915301763033433e-06,
      "loss": 0.4194,
      "step": 8305
    },
    {
      "epoch": 0.5202712225371522,
      "grad_norm": 0.9280675885667682,
      "learning_rate": 4.914287511674465e-06,
      "loss": 0.4336,
      "step": 8306
    },
    {
      "epoch": 0.5203338605364944,
      "grad_norm": 0.8239301464585392,
      "learning_rate": 4.913273263843439e-06,
      "loss": 0.3831,
      "step": 8307
    },
    {
      "epoch": 0.5203964985358368,
      "grad_norm": 0.8569234667200174,
      "learning_rate": 4.912259019582102e-06,
      "loss": 0.433,
      "step": 8308
    },
    {
      "epoch": 0.5204591365351791,
      "grad_norm": 0.8347503872170327,
      "learning_rate": 4.911244778932198e-06,
      "loss": 0.4203,
      "step": 8309
    },
    {
      "epoch": 0.5205217745345214,
      "grad_norm": 0.8991425219628698,
      "learning_rate": 4.9102305419354765e-06,
      "loss": 0.4311,
      "step": 8310
    },
    {
      "epoch": 0.5205844125338637,
      "grad_norm": 0.8520573512286624,
      "learning_rate": 4.909216308633683e-06,
      "loss": 0.3989,
      "step": 8311
    },
    {
      "epoch": 0.5206470505332059,
      "grad_norm": 0.8185562911718917,
      "learning_rate": 4.908202079068563e-06,
      "loss": 0.4184,
      "step": 8312
    },
    {
      "epoch": 0.5207096885325483,
      "grad_norm": 0.822184273964514,
      "learning_rate": 4.907187853281863e-06,
      "loss": 0.3981,
      "step": 8313
    },
    {
      "epoch": 0.5207723265318905,
      "grad_norm": 0.6846611508335989,
      "learning_rate": 4.906173631315327e-06,
      "loss": 0.4446,
      "step": 8314
    },
    {
      "epoch": 0.5208349645312329,
      "grad_norm": 0.8314560880297882,
      "learning_rate": 4.905159413210703e-06,
      "loss": 0.4268,
      "step": 8315
    },
    {
      "epoch": 0.5208976025305752,
      "grad_norm": 0.7675838948863013,
      "learning_rate": 4.9041451990097345e-06,
      "loss": 0.4026,
      "step": 8316
    },
    {
      "epoch": 0.5209602405299175,
      "grad_norm": 0.812425978302691,
      "learning_rate": 4.903130988754167e-06,
      "loss": 0.4223,
      "step": 8317
    },
    {
      "epoch": 0.5210228785292598,
      "grad_norm": 0.855245647063735,
      "learning_rate": 4.9021167824857464e-06,
      "loss": 0.4256,
      "step": 8318
    },
    {
      "epoch": 0.521085516528602,
      "grad_norm": 0.8599840599051471,
      "learning_rate": 4.9011025802462155e-06,
      "loss": 0.3878,
      "step": 8319
    },
    {
      "epoch": 0.5211481545279444,
      "grad_norm": 0.76629488607917,
      "learning_rate": 4.9000883820773226e-06,
      "loss": 0.3649,
      "step": 8320
    },
    {
      "epoch": 0.5212107925272866,
      "grad_norm": 0.867576202566668,
      "learning_rate": 4.899074188020811e-06,
      "loss": 0.3971,
      "step": 8321
    },
    {
      "epoch": 0.521273430526629,
      "grad_norm": 0.9380380379149199,
      "learning_rate": 4.898059998118424e-06,
      "loss": 0.3881,
      "step": 8322
    },
    {
      "epoch": 0.5213360685259713,
      "grad_norm": 0.7625159470075953,
      "learning_rate": 4.897045812411906e-06,
      "loss": 0.4065,
      "step": 8323
    },
    {
      "epoch": 0.5213987065253136,
      "grad_norm": 0.8074871144965239,
      "learning_rate": 4.896031630943002e-06,
      "loss": 0.3877,
      "step": 8324
    },
    {
      "epoch": 0.5214613445246559,
      "grad_norm": 0.8814120407625147,
      "learning_rate": 4.895017453753455e-06,
      "loss": 0.401,
      "step": 8325
    },
    {
      "epoch": 0.5215239825239982,
      "grad_norm": 0.8814988126299924,
      "learning_rate": 4.894003280885009e-06,
      "loss": 0.3974,
      "step": 8326
    },
    {
      "epoch": 0.5215866205233405,
      "grad_norm": 0.8479427046925739,
      "learning_rate": 4.892989112379408e-06,
      "loss": 0.4268,
      "step": 8327
    },
    {
      "epoch": 0.5216492585226827,
      "grad_norm": 0.776645153653262,
      "learning_rate": 4.891974948278393e-06,
      "loss": 0.3646,
      "step": 8328
    },
    {
      "epoch": 0.5217118965220251,
      "grad_norm": 0.8367775185278502,
      "learning_rate": 4.890960788623711e-06,
      "loss": 0.4411,
      "step": 8329
    },
    {
      "epoch": 0.5217745345213674,
      "grad_norm": 0.8313669719625093,
      "learning_rate": 4.8899466334571035e-06,
      "loss": 0.3983,
      "step": 8330
    },
    {
      "epoch": 0.5218371725207097,
      "grad_norm": 0.915134588890553,
      "learning_rate": 4.888932482820312e-06,
      "loss": 0.4074,
      "step": 8331
    },
    {
      "epoch": 0.521899810520052,
      "grad_norm": 0.8038796411409596,
      "learning_rate": 4.8879183367550814e-06,
      "loss": 0.378,
      "step": 8332
    },
    {
      "epoch": 0.5219624485193943,
      "grad_norm": 0.7570974113717254,
      "learning_rate": 4.886904195303153e-06,
      "loss": 0.3512,
      "step": 8333
    },
    {
      "epoch": 0.5220250865187366,
      "grad_norm": 0.6263168362328521,
      "learning_rate": 4.885890058506268e-06,
      "loss": 0.4615,
      "step": 8334
    },
    {
      "epoch": 0.5220877245180789,
      "grad_norm": 0.8440334048885821,
      "learning_rate": 4.884875926406168e-06,
      "loss": 0.3962,
      "step": 8335
    },
    {
      "epoch": 0.5221503625174212,
      "grad_norm": 0.8512598014101018,
      "learning_rate": 4.883861799044599e-06,
      "loss": 0.3968,
      "step": 8336
    },
    {
      "epoch": 0.5222130005167634,
      "grad_norm": 0.8203551914400148,
      "learning_rate": 4.882847676463296e-06,
      "loss": 0.3833,
      "step": 8337
    },
    {
      "epoch": 0.5222756385161058,
      "grad_norm": 0.8578814583859293,
      "learning_rate": 4.8818335587040084e-06,
      "loss": 0.3848,
      "step": 8338
    },
    {
      "epoch": 0.5223382765154481,
      "grad_norm": 0.8954215095449285,
      "learning_rate": 4.880819445808473e-06,
      "loss": 0.4081,
      "step": 8339
    },
    {
      "epoch": 0.5224009145147904,
      "grad_norm": 0.8462433468545525,
      "learning_rate": 4.879805337818429e-06,
      "loss": 0.4469,
      "step": 8340
    },
    {
      "epoch": 0.5224635525141327,
      "grad_norm": 0.8855930314455385,
      "learning_rate": 4.878791234775621e-06,
      "loss": 0.3918,
      "step": 8341
    },
    {
      "epoch": 0.522526190513475,
      "grad_norm": 0.637816447769661,
      "learning_rate": 4.877777136721789e-06,
      "loss": 0.4472,
      "step": 8342
    },
    {
      "epoch": 0.5225888285128173,
      "grad_norm": 0.8485499781466452,
      "learning_rate": 4.876763043698671e-06,
      "loss": 0.4197,
      "step": 8343
    },
    {
      "epoch": 0.5226514665121597,
      "grad_norm": 0.8103629571524302,
      "learning_rate": 4.875748955748011e-06,
      "loss": 0.3888,
      "step": 8344
    },
    {
      "epoch": 0.5227141045115019,
      "grad_norm": 0.8350342649271675,
      "learning_rate": 4.874734872911546e-06,
      "loss": 0.403,
      "step": 8345
    },
    {
      "epoch": 0.5227767425108442,
      "grad_norm": 0.8583710748960465,
      "learning_rate": 4.873720795231015e-06,
      "loss": 0.4177,
      "step": 8346
    },
    {
      "epoch": 0.5228393805101865,
      "grad_norm": 0.8991801394828847,
      "learning_rate": 4.872706722748163e-06,
      "loss": 0.4165,
      "step": 8347
    },
    {
      "epoch": 0.5229020185095288,
      "grad_norm": 0.9054270943299326,
      "learning_rate": 4.871692655504723e-06,
      "loss": 0.4151,
      "step": 8348
    },
    {
      "epoch": 0.5229646565088711,
      "grad_norm": 0.8285499819690326,
      "learning_rate": 4.870678593542439e-06,
      "loss": 0.363,
      "step": 8349
    },
    {
      "epoch": 0.5230272945082134,
      "grad_norm": 0.8202270380093121,
      "learning_rate": 4.8696645369030475e-06,
      "loss": 0.3918,
      "step": 8350
    },
    {
      "epoch": 0.5230899325075558,
      "grad_norm": 0.8239767160615856,
      "learning_rate": 4.868650485628289e-06,
      "loss": 0.3904,
      "step": 8351
    },
    {
      "epoch": 0.523152570506898,
      "grad_norm": 0.878423312327017,
      "learning_rate": 4.867636439759898e-06,
      "loss": 0.3678,
      "step": 8352
    },
    {
      "epoch": 0.5232152085062403,
      "grad_norm": 0.8106257963076834,
      "learning_rate": 4.866622399339619e-06,
      "loss": 0.3839,
      "step": 8353
    },
    {
      "epoch": 0.5232778465055826,
      "grad_norm": 0.7658372441309625,
      "learning_rate": 4.865608364409185e-06,
      "loss": 0.3588,
      "step": 8354
    },
    {
      "epoch": 0.5233404845049249,
      "grad_norm": 0.8693178757647078,
      "learning_rate": 4.864594335010335e-06,
      "loss": 0.3902,
      "step": 8355
    },
    {
      "epoch": 0.5234031225042672,
      "grad_norm": 0.8420626133106874,
      "learning_rate": 4.863580311184809e-06,
      "loss": 0.4018,
      "step": 8356
    },
    {
      "epoch": 0.5234657605036095,
      "grad_norm": 0.7665215866853986,
      "learning_rate": 4.86256629297434e-06,
      "loss": 0.393,
      "step": 8357
    },
    {
      "epoch": 0.5235283985029519,
      "grad_norm": 0.7988679749310039,
      "learning_rate": 4.861552280420671e-06,
      "loss": 0.4178,
      "step": 8358
    },
    {
      "epoch": 0.5235910365022941,
      "grad_norm": 0.8234980148189301,
      "learning_rate": 4.860538273565535e-06,
      "loss": 0.3868,
      "step": 8359
    },
    {
      "epoch": 0.5236536745016365,
      "grad_norm": 0.8298855336775456,
      "learning_rate": 4.8595242724506696e-06,
      "loss": 0.4128,
      "step": 8360
    },
    {
      "epoch": 0.5237163125009787,
      "grad_norm": 0.82806868008111,
      "learning_rate": 4.8585102771178105e-06,
      "loss": 0.4025,
      "step": 8361
    },
    {
      "epoch": 0.523778950500321,
      "grad_norm": 0.8574148350554026,
      "learning_rate": 4.857496287608695e-06,
      "loss": 0.4641,
      "step": 8362
    },
    {
      "epoch": 0.5238415884996633,
      "grad_norm": 0.6205821496586041,
      "learning_rate": 4.856482303965059e-06,
      "loss": 0.4504,
      "step": 8363
    },
    {
      "epoch": 0.5239042264990056,
      "grad_norm": 0.7758440027640646,
      "learning_rate": 4.855468326228638e-06,
      "loss": 0.3613,
      "step": 8364
    },
    {
      "epoch": 0.523966864498348,
      "grad_norm": 0.7930971167718887,
      "learning_rate": 4.854454354441166e-06,
      "loss": 0.3809,
      "step": 8365
    },
    {
      "epoch": 0.5240295024976902,
      "grad_norm": 0.8431568107795171,
      "learning_rate": 4.853440388644381e-06,
      "loss": 0.4421,
      "step": 8366
    },
    {
      "epoch": 0.5240921404970326,
      "grad_norm": 0.7635144192730712,
      "learning_rate": 4.852426428880017e-06,
      "loss": 0.3557,
      "step": 8367
    },
    {
      "epoch": 0.5241547784963748,
      "grad_norm": 0.8704897877294172,
      "learning_rate": 4.851412475189809e-06,
      "loss": 0.4198,
      "step": 8368
    },
    {
      "epoch": 0.5242174164957172,
      "grad_norm": 0.7997707379065948,
      "learning_rate": 4.850398527615489e-06,
      "loss": 0.4208,
      "step": 8369
    },
    {
      "epoch": 0.5242800544950594,
      "grad_norm": 0.8076608914069535,
      "learning_rate": 4.8493845861987956e-06,
      "loss": 0.3619,
      "step": 8370
    },
    {
      "epoch": 0.5243426924944017,
      "grad_norm": 0.7952971870742692,
      "learning_rate": 4.848370650981461e-06,
      "loss": 0.3845,
      "step": 8371
    },
    {
      "epoch": 0.524405330493744,
      "grad_norm": 0.7619663335362562,
      "learning_rate": 4.8473567220052184e-06,
      "loss": 0.3706,
      "step": 8372
    },
    {
      "epoch": 0.5244679684930863,
      "grad_norm": 0.7589487960416016,
      "learning_rate": 4.846342799311799e-06,
      "loss": 0.3397,
      "step": 8373
    },
    {
      "epoch": 0.5245306064924287,
      "grad_norm": 0.7976099123610207,
      "learning_rate": 4.845328882942939e-06,
      "loss": 0.3711,
      "step": 8374
    },
    {
      "epoch": 0.5245932444917709,
      "grad_norm": 0.8516831085533494,
      "learning_rate": 4.844314972940372e-06,
      "loss": 0.4008,
      "step": 8375
    },
    {
      "epoch": 0.5246558824911133,
      "grad_norm": 0.8498487127884362,
      "learning_rate": 4.84330106934583e-06,
      "loss": 0.4362,
      "step": 8376
    },
    {
      "epoch": 0.5247185204904555,
      "grad_norm": 0.7411643772093817,
      "learning_rate": 4.842287172201045e-06,
      "loss": 0.4563,
      "step": 8377
    },
    {
      "epoch": 0.5247811584897978,
      "grad_norm": 0.767939072266038,
      "learning_rate": 4.841273281547748e-06,
      "loss": 0.3418,
      "step": 8378
    },
    {
      "epoch": 0.5248437964891401,
      "grad_norm": 0.8784258580136556,
      "learning_rate": 4.840259397427674e-06,
      "loss": 0.4156,
      "step": 8379
    },
    {
      "epoch": 0.5249064344884824,
      "grad_norm": 0.8400903470707345,
      "learning_rate": 4.839245519882552e-06,
      "loss": 0.37,
      "step": 8380
    },
    {
      "epoch": 0.5249690724878248,
      "grad_norm": 0.8529783698934237,
      "learning_rate": 4.838231648954115e-06,
      "loss": 0.4434,
      "step": 8381
    },
    {
      "epoch": 0.525031710487167,
      "grad_norm": 0.8438851814813682,
      "learning_rate": 4.837217784684091e-06,
      "loss": 0.3965,
      "step": 8382
    },
    {
      "epoch": 0.5250943484865094,
      "grad_norm": 0.6585478982526103,
      "learning_rate": 4.836203927114214e-06,
      "loss": 0.4836,
      "step": 8383
    },
    {
      "epoch": 0.5251569864858516,
      "grad_norm": 0.8378215686517626,
      "learning_rate": 4.835190076286214e-06,
      "loss": 0.4103,
      "step": 8384
    },
    {
      "epoch": 0.525219624485194,
      "grad_norm": 0.6713805220985095,
      "learning_rate": 4.834176232241822e-06,
      "loss": 0.4676,
      "step": 8385
    },
    {
      "epoch": 0.5252822624845362,
      "grad_norm": 0.8483044385091858,
      "learning_rate": 4.833162395022767e-06,
      "loss": 0.4208,
      "step": 8386
    },
    {
      "epoch": 0.5253449004838785,
      "grad_norm": 0.8757765509062551,
      "learning_rate": 4.8321485646707775e-06,
      "loss": 0.4557,
      "step": 8387
    },
    {
      "epoch": 0.5254075384832209,
      "grad_norm": 0.8168022779286629,
      "learning_rate": 4.831134741227586e-06,
      "loss": 0.3649,
      "step": 8388
    },
    {
      "epoch": 0.5254701764825631,
      "grad_norm": 0.8520543878770334,
      "learning_rate": 4.8301209247349195e-06,
      "loss": 0.4038,
      "step": 8389
    },
    {
      "epoch": 0.5255328144819055,
      "grad_norm": 0.8088179102228152,
      "learning_rate": 4.829107115234506e-06,
      "loss": 0.4141,
      "step": 8390
    },
    {
      "epoch": 0.5255954524812477,
      "grad_norm": 0.824519619482851,
      "learning_rate": 4.828093312768077e-06,
      "loss": 0.4433,
      "step": 8391
    },
    {
      "epoch": 0.5256580904805901,
      "grad_norm": 0.769653672651507,
      "learning_rate": 4.827079517377356e-06,
      "loss": 0.4202,
      "step": 8392
    },
    {
      "epoch": 0.5257207284799323,
      "grad_norm": 0.613339692648587,
      "learning_rate": 4.8260657291040766e-06,
      "loss": 0.4209,
      "step": 8393
    },
    {
      "epoch": 0.5257833664792747,
      "grad_norm": 0.8058754390742849,
      "learning_rate": 4.825051947989965e-06,
      "loss": 0.3868,
      "step": 8394
    },
    {
      "epoch": 0.525846004478617,
      "grad_norm": 0.8610861032962648,
      "learning_rate": 4.824038174076746e-06,
      "loss": 0.3967,
      "step": 8395
    },
    {
      "epoch": 0.5259086424779592,
      "grad_norm": 0.8807194228280499,
      "learning_rate": 4.823024407406149e-06,
      "loss": 0.3864,
      "step": 8396
    },
    {
      "epoch": 0.5259712804773016,
      "grad_norm": 0.8154321281134747,
      "learning_rate": 4.8220106480199004e-06,
      "loss": 0.4204,
      "step": 8397
    },
    {
      "epoch": 0.5260339184766438,
      "grad_norm": 0.8595353287879933,
      "learning_rate": 4.820996895959727e-06,
      "loss": 0.4284,
      "step": 8398
    },
    {
      "epoch": 0.5260965564759862,
      "grad_norm": 0.8161517946080806,
      "learning_rate": 4.819983151267353e-06,
      "loss": 0.4042,
      "step": 8399
    },
    {
      "epoch": 0.5261591944753284,
      "grad_norm": 0.8782843650398903,
      "learning_rate": 4.818969413984508e-06,
      "loss": 0.4385,
      "step": 8400
    },
    {
      "epoch": 0.5262218324746708,
      "grad_norm": 0.7992976706980636,
      "learning_rate": 4.817955684152915e-06,
      "loss": 0.394,
      "step": 8401
    },
    {
      "epoch": 0.526284470474013,
      "grad_norm": 0.9025715185615782,
      "learning_rate": 4.816941961814297e-06,
      "loss": 0.3686,
      "step": 8402
    },
    {
      "epoch": 0.5263471084733553,
      "grad_norm": 0.8329998690616642,
      "learning_rate": 4.815928247010385e-06,
      "loss": 0.3483,
      "step": 8403
    },
    {
      "epoch": 0.5264097464726977,
      "grad_norm": 0.7593134202013102,
      "learning_rate": 4.814914539782899e-06,
      "loss": 0.3771,
      "step": 8404
    },
    {
      "epoch": 0.5264723844720399,
      "grad_norm": 0.7833125207390207,
      "learning_rate": 4.813900840173566e-06,
      "loss": 0.3697,
      "step": 8405
    },
    {
      "epoch": 0.5265350224713823,
      "grad_norm": 0.8700357118395365,
      "learning_rate": 4.8128871482241085e-06,
      "loss": 0.4255,
      "step": 8406
    },
    {
      "epoch": 0.5265976604707245,
      "grad_norm": 0.6562720523143896,
      "learning_rate": 4.81187346397625e-06,
      "loss": 0.4698,
      "step": 8407
    },
    {
      "epoch": 0.5266602984700669,
      "grad_norm": 0.8886634614829774,
      "learning_rate": 4.810859787471716e-06,
      "loss": 0.3977,
      "step": 8408
    },
    {
      "epoch": 0.5267229364694092,
      "grad_norm": 0.6232919283719757,
      "learning_rate": 4.809846118752228e-06,
      "loss": 0.4957,
      "step": 8409
    },
    {
      "epoch": 0.5267855744687515,
      "grad_norm": 0.8469086421446168,
      "learning_rate": 4.808832457859509e-06,
      "loss": 0.4632,
      "step": 8410
    },
    {
      "epoch": 0.5268482124680938,
      "grad_norm": 0.8539336772120959,
      "learning_rate": 4.8078188048352795e-06,
      "loss": 0.4137,
      "step": 8411
    },
    {
      "epoch": 0.526910850467436,
      "grad_norm": 0.8493410391721196,
      "learning_rate": 4.806805159721265e-06,
      "loss": 0.4114,
      "step": 8412
    },
    {
      "epoch": 0.5269734884667784,
      "grad_norm": 0.8020180782102472,
      "learning_rate": 4.805791522559186e-06,
      "loss": 0.4283,
      "step": 8413
    },
    {
      "epoch": 0.5270361264661206,
      "grad_norm": 0.80091785377821,
      "learning_rate": 4.8047778933907646e-06,
      "loss": 0.3736,
      "step": 8414
    },
    {
      "epoch": 0.527098764465463,
      "grad_norm": 0.8328861586184464,
      "learning_rate": 4.803764272257721e-06,
      "loss": 0.4252,
      "step": 8415
    },
    {
      "epoch": 0.5271614024648053,
      "grad_norm": 0.840321941462567,
      "learning_rate": 4.802750659201776e-06,
      "loss": 0.3816,
      "step": 8416
    },
    {
      "epoch": 0.5272240404641476,
      "grad_norm": 0.8306058943916537,
      "learning_rate": 4.801737054264651e-06,
      "loss": 0.4174,
      "step": 8417
    },
    {
      "epoch": 0.5272866784634899,
      "grad_norm": 0.8690476615057071,
      "learning_rate": 4.800723457488067e-06,
      "loss": 0.4211,
      "step": 8418
    },
    {
      "epoch": 0.5273493164628322,
      "grad_norm": 0.8715311147400883,
      "learning_rate": 4.799709868913741e-06,
      "loss": 0.4095,
      "step": 8419
    },
    {
      "epoch": 0.5274119544621745,
      "grad_norm": 1.0188360859693413,
      "learning_rate": 4.798696288583392e-06,
      "loss": 0.4267,
      "step": 8420
    },
    {
      "epoch": 0.5274745924615167,
      "grad_norm": 0.8830083148672181,
      "learning_rate": 4.797682716538742e-06,
      "loss": 0.4214,
      "step": 8421
    },
    {
      "epoch": 0.5275372304608591,
      "grad_norm": 0.7945811223409723,
      "learning_rate": 4.79666915282151e-06,
      "loss": 0.3927,
      "step": 8422
    },
    {
      "epoch": 0.5275998684602013,
      "grad_norm": 0.8166124943007212,
      "learning_rate": 4.795655597473415e-06,
      "loss": 0.4072,
      "step": 8423
    },
    {
      "epoch": 0.5276625064595437,
      "grad_norm": 0.8414101612901311,
      "learning_rate": 4.7946420505361715e-06,
      "loss": 0.3996,
      "step": 8424
    },
    {
      "epoch": 0.527725144458886,
      "grad_norm": 0.7988905908725723,
      "learning_rate": 4.793628512051499e-06,
      "loss": 0.4292,
      "step": 8425
    },
    {
      "epoch": 0.5277877824582283,
      "grad_norm": 0.8137057765568944,
      "learning_rate": 4.792614982061117e-06,
      "loss": 0.3568,
      "step": 8426
    },
    {
      "epoch": 0.5278504204575706,
      "grad_norm": 0.7594865046103919,
      "learning_rate": 4.7916014606067405e-06,
      "loss": 0.3481,
      "step": 8427
    },
    {
      "epoch": 0.5279130584569128,
      "grad_norm": 0.7931012516542959,
      "learning_rate": 4.790587947730085e-06,
      "loss": 0.4143,
      "step": 8428
    },
    {
      "epoch": 0.5279756964562552,
      "grad_norm": 0.817557305999515,
      "learning_rate": 4.789574443472868e-06,
      "loss": 0.4016,
      "step": 8429
    },
    {
      "epoch": 0.5280383344555974,
      "grad_norm": 0.830356892882505,
      "learning_rate": 4.788560947876807e-06,
      "loss": 0.3847,
      "step": 8430
    },
    {
      "epoch": 0.5281009724549398,
      "grad_norm": 0.8419739154870107,
      "learning_rate": 4.787547460983618e-06,
      "loss": 0.3853,
      "step": 8431
    },
    {
      "epoch": 0.5281636104542821,
      "grad_norm": 0.8124343556354402,
      "learning_rate": 4.786533982835014e-06,
      "loss": 0.4008,
      "step": 8432
    },
    {
      "epoch": 0.5282262484536244,
      "grad_norm": 0.8252887119894482,
      "learning_rate": 4.78552051347271e-06,
      "loss": 0.3993,
      "step": 8433
    },
    {
      "epoch": 0.5282888864529667,
      "grad_norm": 0.8287322864271958,
      "learning_rate": 4.784507052938422e-06,
      "loss": 0.3783,
      "step": 8434
    },
    {
      "epoch": 0.528351524452309,
      "grad_norm": 0.8220961313624732,
      "learning_rate": 4.783493601273865e-06,
      "loss": 0.3875,
      "step": 8435
    },
    {
      "epoch": 0.5284141624516513,
      "grad_norm": 0.8370707852547616,
      "learning_rate": 4.7824801585207505e-06,
      "loss": 0.4561,
      "step": 8436
    },
    {
      "epoch": 0.5284768004509935,
      "grad_norm": 0.7848331301212559,
      "learning_rate": 4.781466724720793e-06,
      "loss": 0.3643,
      "step": 8437
    },
    {
      "epoch": 0.5285394384503359,
      "grad_norm": 0.8391149280381097,
      "learning_rate": 4.780453299915705e-06,
      "loss": 0.4171,
      "step": 8438
    },
    {
      "epoch": 0.5286020764496782,
      "grad_norm": 0.8038354919374762,
      "learning_rate": 4.7794398841472e-06,
      "loss": 0.4385,
      "step": 8439
    },
    {
      "epoch": 0.5286647144490205,
      "grad_norm": 0.864626195795564,
      "learning_rate": 4.778426477456991e-06,
      "loss": 0.4631,
      "step": 8440
    },
    {
      "epoch": 0.5287273524483628,
      "grad_norm": 0.6790903458323172,
      "learning_rate": 4.77741307988679e-06,
      "loss": 0.4833,
      "step": 8441
    },
    {
      "epoch": 0.5287899904477051,
      "grad_norm": 0.7875130070270306,
      "learning_rate": 4.776399691478306e-06,
      "loss": 0.4072,
      "step": 8442
    },
    {
      "epoch": 0.5288526284470474,
      "grad_norm": 0.9486887186176891,
      "learning_rate": 4.775386312273255e-06,
      "loss": 0.398,
      "step": 8443
    },
    {
      "epoch": 0.5289152664463898,
      "grad_norm": 0.6207010556105994,
      "learning_rate": 4.774372942313344e-06,
      "loss": 0.4428,
      "step": 8444
    },
    {
      "epoch": 0.528977904445732,
      "grad_norm": 0.8197421655477164,
      "learning_rate": 4.773359581640284e-06,
      "loss": 0.3998,
      "step": 8445
    },
    {
      "epoch": 0.5290405424450743,
      "grad_norm": 0.8400927241396039,
      "learning_rate": 4.772346230295785e-06,
      "loss": 0.3696,
      "step": 8446
    },
    {
      "epoch": 0.5291031804444166,
      "grad_norm": 0.8796538692129006,
      "learning_rate": 4.7713328883215585e-06,
      "loss": 0.4643,
      "step": 8447
    },
    {
      "epoch": 0.5291658184437589,
      "grad_norm": 0.841138301736903,
      "learning_rate": 4.770319555759311e-06,
      "loss": 0.4141,
      "step": 8448
    },
    {
      "epoch": 0.5292284564431012,
      "grad_norm": 0.8296306323086969,
      "learning_rate": 4.769306232650755e-06,
      "loss": 0.4451,
      "step": 8449
    },
    {
      "epoch": 0.5292910944424435,
      "grad_norm": 0.8606735315280005,
      "learning_rate": 4.768292919037597e-06,
      "loss": 0.3994,
      "step": 8450
    },
    {
      "epoch": 0.5293537324417859,
      "grad_norm": 0.801435858625443,
      "learning_rate": 4.767279614961545e-06,
      "loss": 0.3903,
      "step": 8451
    },
    {
      "epoch": 0.5294163704411281,
      "grad_norm": 0.8658539509683838,
      "learning_rate": 4.766266320464308e-06,
      "loss": 0.4287,
      "step": 8452
    },
    {
      "epoch": 0.5294790084404705,
      "grad_norm": 0.9145566626841278,
      "learning_rate": 4.765253035587593e-06,
      "loss": 0.377,
      "step": 8453
    },
    {
      "epoch": 0.5295416464398127,
      "grad_norm": 0.8012393258632249,
      "learning_rate": 4.764239760373105e-06,
      "loss": 0.4118,
      "step": 8454
    },
    {
      "epoch": 0.529604284439155,
      "grad_norm": 0.7742160869930997,
      "learning_rate": 4.763226494862553e-06,
      "loss": 0.4201,
      "step": 8455
    },
    {
      "epoch": 0.5296669224384973,
      "grad_norm": 0.7999674862224556,
      "learning_rate": 4.762213239097644e-06,
      "loss": 0.3937,
      "step": 8456
    },
    {
      "epoch": 0.5297295604378396,
      "grad_norm": 0.7650003994916776,
      "learning_rate": 4.7611999931200785e-06,
      "loss": 0.3563,
      "step": 8457
    },
    {
      "epoch": 0.529792198437182,
      "grad_norm": 0.8057518028500243,
      "learning_rate": 4.760186756971568e-06,
      "loss": 0.3673,
      "step": 8458
    },
    {
      "epoch": 0.5298548364365242,
      "grad_norm": 0.8129428632307455,
      "learning_rate": 4.7591735306938144e-06,
      "loss": 0.3701,
      "step": 8459
    },
    {
      "epoch": 0.5299174744358666,
      "grad_norm": 0.7528753831259157,
      "learning_rate": 4.758160314328524e-06,
      "loss": 0.37,
      "step": 8460
    },
    {
      "epoch": 0.5299801124352088,
      "grad_norm": 0.6615284763400032,
      "learning_rate": 4.7571471079174e-06,
      "loss": 0.4589,
      "step": 8461
    },
    {
      "epoch": 0.5300427504345511,
      "grad_norm": 0.7614224348371016,
      "learning_rate": 4.756133911502146e-06,
      "loss": 0.4006,
      "step": 8462
    },
    {
      "epoch": 0.5301053884338934,
      "grad_norm": 0.6181027303970573,
      "learning_rate": 4.755120725124464e-06,
      "loss": 0.4719,
      "step": 8463
    },
    {
      "epoch": 0.5301680264332357,
      "grad_norm": 0.849906428039547,
      "learning_rate": 4.75410754882606e-06,
      "loss": 0.3694,
      "step": 8464
    },
    {
      "epoch": 0.530230664432578,
      "grad_norm": 0.8643811872880157,
      "learning_rate": 4.7530943826486346e-06,
      "loss": 0.3804,
      "step": 8465
    },
    {
      "epoch": 0.5302933024319203,
      "grad_norm": 0.8120786903020153,
      "learning_rate": 4.752081226633888e-06,
      "loss": 0.3458,
      "step": 8466
    },
    {
      "epoch": 0.5303559404312627,
      "grad_norm": 0.8465476308347545,
      "learning_rate": 4.751068080823527e-06,
      "loss": 0.3908,
      "step": 8467
    },
    {
      "epoch": 0.5304185784306049,
      "grad_norm": 0.8099323582478126,
      "learning_rate": 4.750054945259249e-06,
      "loss": 0.4,
      "step": 8468
    },
    {
      "epoch": 0.5304812164299473,
      "grad_norm": 0.8281894615413593,
      "learning_rate": 4.7490418199827555e-06,
      "loss": 0.4138,
      "step": 8469
    },
    {
      "epoch": 0.5305438544292895,
      "grad_norm": 0.9554040639268596,
      "learning_rate": 4.7480287050357485e-06,
      "loss": 0.435,
      "step": 8470
    },
    {
      "epoch": 0.5306064924286318,
      "grad_norm": 0.8673062663612796,
      "learning_rate": 4.7470156004599265e-06,
      "loss": 0.4083,
      "step": 8471
    },
    {
      "epoch": 0.5306691304279741,
      "grad_norm": 0.8297521844871348,
      "learning_rate": 4.746002506296989e-06,
      "loss": 0.3802,
      "step": 8472
    },
    {
      "epoch": 0.5307317684273164,
      "grad_norm": 0.8786653351890761,
      "learning_rate": 4.744989422588635e-06,
      "loss": 0.4429,
      "step": 8473
    },
    {
      "epoch": 0.5307944064266588,
      "grad_norm": 0.8881565578152587,
      "learning_rate": 4.743976349376565e-06,
      "loss": 0.3979,
      "step": 8474
    },
    {
      "epoch": 0.530857044426001,
      "grad_norm": 0.5796710474482555,
      "learning_rate": 4.7429632867024726e-06,
      "loss": 0.4474,
      "step": 8475
    },
    {
      "epoch": 0.5309196824253434,
      "grad_norm": 0.5745348610497721,
      "learning_rate": 4.741950234608062e-06,
      "loss": 0.4194,
      "step": 8476
    },
    {
      "epoch": 0.5309823204246856,
      "grad_norm": 0.6522677862473089,
      "learning_rate": 4.740937193135027e-06,
      "loss": 0.4738,
      "step": 8477
    },
    {
      "epoch": 0.531044958424028,
      "grad_norm": 0.7983832235709819,
      "learning_rate": 4.739924162325065e-06,
      "loss": 0.3913,
      "step": 8478
    },
    {
      "epoch": 0.5311075964233702,
      "grad_norm": 0.8652145885129,
      "learning_rate": 4.738911142219874e-06,
      "loss": 0.4402,
      "step": 8479
    },
    {
      "epoch": 0.5311702344227125,
      "grad_norm": 0.8504741645729315,
      "learning_rate": 4.7378981328611475e-06,
      "loss": 0.4218,
      "step": 8480
    },
    {
      "epoch": 0.5312328724220549,
      "grad_norm": 0.8223584744509833,
      "learning_rate": 4.736885134290583e-06,
      "loss": 0.4096,
      "step": 8481
    },
    {
      "epoch": 0.5312955104213971,
      "grad_norm": 0.8442517782000695,
      "learning_rate": 4.735872146549875e-06,
      "loss": 0.4036,
      "step": 8482
    },
    {
      "epoch": 0.5313581484207395,
      "grad_norm": 0.7685115511404648,
      "learning_rate": 4.7348591696807195e-06,
      "loss": 0.4152,
      "step": 8483
    },
    {
      "epoch": 0.5314207864200817,
      "grad_norm": 0.803529806272926,
      "learning_rate": 4.733846203724807e-06,
      "loss": 0.3706,
      "step": 8484
    },
    {
      "epoch": 0.5314834244194241,
      "grad_norm": 0.841338412328239,
      "learning_rate": 4.7328332487238355e-06,
      "loss": 0.4363,
      "step": 8485
    },
    {
      "epoch": 0.5315460624187663,
      "grad_norm": 0.890131064158318,
      "learning_rate": 4.731820304719498e-06,
      "loss": 0.3979,
      "step": 8486
    },
    {
      "epoch": 0.5316087004181086,
      "grad_norm": 0.8611482922366105,
      "learning_rate": 4.730807371753486e-06,
      "loss": 0.4074,
      "step": 8487
    },
    {
      "epoch": 0.531671338417451,
      "grad_norm": 0.8577886363225563,
      "learning_rate": 4.7297944498674936e-06,
      "loss": 0.3846,
      "step": 8488
    },
    {
      "epoch": 0.5317339764167932,
      "grad_norm": 0.7759674483181839,
      "learning_rate": 4.7287815391032105e-06,
      "loss": 0.3243,
      "step": 8489
    },
    {
      "epoch": 0.5317966144161356,
      "grad_norm": 0.8136479178717942,
      "learning_rate": 4.7277686395023305e-06,
      "loss": 0.3946,
      "step": 8490
    },
    {
      "epoch": 0.5318592524154778,
      "grad_norm": 0.8747044702606179,
      "learning_rate": 4.7267557511065435e-06,
      "loss": 0.3983,
      "step": 8491
    },
    {
      "epoch": 0.5319218904148202,
      "grad_norm": 0.8751716196325963,
      "learning_rate": 4.72574287395754e-06,
      "loss": 0.4328,
      "step": 8492
    },
    {
      "epoch": 0.5319845284141624,
      "grad_norm": 0.8759550027225119,
      "learning_rate": 4.724730008097011e-06,
      "loss": 0.4111,
      "step": 8493
    },
    {
      "epoch": 0.5320471664135048,
      "grad_norm": 0.8384794958249635,
      "learning_rate": 4.723717153566644e-06,
      "loss": 0.4291,
      "step": 8494
    },
    {
      "epoch": 0.532109804412847,
      "grad_norm": 0.8356331514591465,
      "learning_rate": 4.722704310408133e-06,
      "loss": 0.3656,
      "step": 8495
    },
    {
      "epoch": 0.5321724424121893,
      "grad_norm": 0.8606498983335911,
      "learning_rate": 4.721691478663163e-06,
      "loss": 0.4007,
      "step": 8496
    },
    {
      "epoch": 0.5322350804115317,
      "grad_norm": 0.8088110542643416,
      "learning_rate": 4.7206786583734245e-06,
      "loss": 0.3973,
      "step": 8497
    },
    {
      "epoch": 0.5322977184108739,
      "grad_norm": 0.8033528692020342,
      "learning_rate": 4.719665849580603e-06,
      "loss": 0.3459,
      "step": 8498
    },
    {
      "epoch": 0.5323603564102163,
      "grad_norm": 0.8347800701705566,
      "learning_rate": 4.718653052326388e-06,
      "loss": 0.4127,
      "step": 8499
    },
    {
      "epoch": 0.5324229944095585,
      "grad_norm": 0.9039422005179815,
      "learning_rate": 4.717640266652466e-06,
      "loss": 0.3958,
      "step": 8500
    },
    {
      "epoch": 0.5324856324089009,
      "grad_norm": 0.7663363530811628,
      "learning_rate": 4.716627492600521e-06,
      "loss": 0.3727,
      "step": 8501
    },
    {
      "epoch": 0.5325482704082432,
      "grad_norm": 0.809727258048179,
      "learning_rate": 4.715614730212243e-06,
      "loss": 0.3705,
      "step": 8502
    },
    {
      "epoch": 0.5326109084075855,
      "grad_norm": 0.8692739906803444,
      "learning_rate": 4.714601979529313e-06,
      "loss": 0.4113,
      "step": 8503
    },
    {
      "epoch": 0.5326735464069278,
      "grad_norm": 0.8675514476511111,
      "learning_rate": 4.71358924059342e-06,
      "loss": 0.3915,
      "step": 8504
    },
    {
      "epoch": 0.53273618440627,
      "grad_norm": 0.657091522970408,
      "learning_rate": 4.7125765134462485e-06,
      "loss": 0.4834,
      "step": 8505
    },
    {
      "epoch": 0.5327988224056124,
      "grad_norm": 0.7704490734692838,
      "learning_rate": 4.711563798129478e-06,
      "loss": 0.3626,
      "step": 8506
    },
    {
      "epoch": 0.5328614604049546,
      "grad_norm": 0.8007594500292604,
      "learning_rate": 4.7105510946847974e-06,
      "loss": 0.426,
      "step": 8507
    },
    {
      "epoch": 0.532924098404297,
      "grad_norm": 0.7650632316873688,
      "learning_rate": 4.709538403153887e-06,
      "loss": 0.3874,
      "step": 8508
    },
    {
      "epoch": 0.5329867364036392,
      "grad_norm": 0.8301906676764672,
      "learning_rate": 4.708525723578431e-06,
      "loss": 0.4306,
      "step": 8509
    },
    {
      "epoch": 0.5330493744029816,
      "grad_norm": 0.794793305462976,
      "learning_rate": 4.707513056000109e-06,
      "loss": 0.3669,
      "step": 8510
    },
    {
      "epoch": 0.5331120124023239,
      "grad_norm": 0.8134689306565951,
      "learning_rate": 4.706500400460604e-06,
      "loss": 0.3918,
      "step": 8511
    },
    {
      "epoch": 0.5331746504016661,
      "grad_norm": 0.9975015062398807,
      "learning_rate": 4.705487757001595e-06,
      "loss": 0.3928,
      "step": 8512
    },
    {
      "epoch": 0.5332372884010085,
      "grad_norm": 0.8972716492508751,
      "learning_rate": 4.704475125664767e-06,
      "loss": 0.4149,
      "step": 8513
    },
    {
      "epoch": 0.5332999264003507,
      "grad_norm": 0.8665096069662614,
      "learning_rate": 4.703462506491798e-06,
      "loss": 0.3856,
      "step": 8514
    },
    {
      "epoch": 0.5333625643996931,
      "grad_norm": 0.8345556206871755,
      "learning_rate": 4.702449899524366e-06,
      "loss": 0.4024,
      "step": 8515
    },
    {
      "epoch": 0.5334252023990353,
      "grad_norm": 0.8624409010333672,
      "learning_rate": 4.701437304804152e-06,
      "loss": 0.4372,
      "step": 8516
    },
    {
      "epoch": 0.5334878403983777,
      "grad_norm": 0.6652151144200709,
      "learning_rate": 4.700424722372835e-06,
      "loss": 0.47,
      "step": 8517
    },
    {
      "epoch": 0.53355047839772,
      "grad_norm": 0.8623183143235869,
      "learning_rate": 4.69941215227209e-06,
      "loss": 0.3895,
      "step": 8518
    },
    {
      "epoch": 0.5336131163970623,
      "grad_norm": 0.7974306567907623,
      "learning_rate": 4.6983995945435976e-06,
      "loss": 0.4115,
      "step": 8519
    },
    {
      "epoch": 0.5336757543964046,
      "grad_norm": 0.7886421710119246,
      "learning_rate": 4.697387049229034e-06,
      "loss": 0.3794,
      "step": 8520
    },
    {
      "epoch": 0.5337383923957468,
      "grad_norm": 0.6299261353891272,
      "learning_rate": 4.696374516370074e-06,
      "loss": 0.4495,
      "step": 8521
    },
    {
      "epoch": 0.5338010303950892,
      "grad_norm": 0.8313265786297619,
      "learning_rate": 4.695361996008397e-06,
      "loss": 0.4009,
      "step": 8522
    },
    {
      "epoch": 0.5338636683944314,
      "grad_norm": 0.8795218247214486,
      "learning_rate": 4.6943494881856764e-06,
      "loss": 0.4062,
      "step": 8523
    },
    {
      "epoch": 0.5339263063937738,
      "grad_norm": 0.7656828706028218,
      "learning_rate": 4.693336992943587e-06,
      "loss": 0.3787,
      "step": 8524
    },
    {
      "epoch": 0.5339889443931161,
      "grad_norm": 0.8047435289493533,
      "learning_rate": 4.692324510323806e-06,
      "loss": 0.4094,
      "step": 8525
    },
    {
      "epoch": 0.5340515823924584,
      "grad_norm": 0.8915335408823764,
      "learning_rate": 4.6913120403680036e-06,
      "loss": 0.4361,
      "step": 8526
    },
    {
      "epoch": 0.5341142203918007,
      "grad_norm": 0.7550587117335418,
      "learning_rate": 4.690299583117854e-06,
      "loss": 0.4126,
      "step": 8527
    },
    {
      "epoch": 0.534176858391143,
      "grad_norm": 0.7759318564331061,
      "learning_rate": 4.689287138615032e-06,
      "loss": 0.3902,
      "step": 8528
    },
    {
      "epoch": 0.5342394963904853,
      "grad_norm": 0.8334389834941682,
      "learning_rate": 4.68827470690121e-06,
      "loss": 0.4086,
      "step": 8529
    },
    {
      "epoch": 0.5343021343898275,
      "grad_norm": 0.8007653907531018,
      "learning_rate": 4.687262288018057e-06,
      "loss": 0.3612,
      "step": 8530
    },
    {
      "epoch": 0.5343647723891699,
      "grad_norm": 0.8615092163036691,
      "learning_rate": 4.686249882007247e-06,
      "loss": 0.4211,
      "step": 8531
    },
    {
      "epoch": 0.5344274103885122,
      "grad_norm": 0.7861744972181965,
      "learning_rate": 4.68523748891045e-06,
      "loss": 0.3958,
      "step": 8532
    },
    {
      "epoch": 0.5344900483878545,
      "grad_norm": 0.7377644576785397,
      "learning_rate": 4.684225108769337e-06,
      "loss": 0.3818,
      "step": 8533
    },
    {
      "epoch": 0.5345526863871968,
      "grad_norm": 0.7748792933025885,
      "learning_rate": 4.683212741625578e-06,
      "loss": 0.3854,
      "step": 8534
    },
    {
      "epoch": 0.5346153243865391,
      "grad_norm": 0.6239464191512842,
      "learning_rate": 4.68220038752084e-06,
      "loss": 0.4657,
      "step": 8535
    },
    {
      "epoch": 0.5346779623858814,
      "grad_norm": 0.8526356080617827,
      "learning_rate": 4.681188046496793e-06,
      "loss": 0.4076,
      "step": 8536
    },
    {
      "epoch": 0.5347406003852236,
      "grad_norm": 0.8976844267340167,
      "learning_rate": 4.680175718595107e-06,
      "loss": 0.4112,
      "step": 8537
    },
    {
      "epoch": 0.534803238384566,
      "grad_norm": 0.8030213829075548,
      "learning_rate": 4.679163403857446e-06,
      "loss": 0.4038,
      "step": 8538
    },
    {
      "epoch": 0.5348658763839083,
      "grad_norm": 0.7935074874497828,
      "learning_rate": 4.678151102325479e-06,
      "loss": 0.379,
      "step": 8539
    },
    {
      "epoch": 0.5349285143832506,
      "grad_norm": 0.917849386906571,
      "learning_rate": 4.677138814040871e-06,
      "loss": 0.4084,
      "step": 8540
    },
    {
      "epoch": 0.5349911523825929,
      "grad_norm": 0.857618743617297,
      "learning_rate": 4.676126539045291e-06,
      "loss": 0.4389,
      "step": 8541
    },
    {
      "epoch": 0.5350537903819352,
      "grad_norm": 0.7669309557540673,
      "learning_rate": 4.675114277380403e-06,
      "loss": 0.3754,
      "step": 8542
    },
    {
      "epoch": 0.5351164283812775,
      "grad_norm": 0.8718933104051194,
      "learning_rate": 4.6741020290878715e-06,
      "loss": 0.4003,
      "step": 8543
    },
    {
      "epoch": 0.5351790663806199,
      "grad_norm": 0.8354789370777383,
      "learning_rate": 4.67308979420936e-06,
      "loss": 0.3817,
      "step": 8544
    },
    {
      "epoch": 0.5352417043799621,
      "grad_norm": 0.8226377288165075,
      "learning_rate": 4.6720775727865346e-06,
      "loss": 0.3832,
      "step": 8545
    },
    {
      "epoch": 0.5353043423793044,
      "grad_norm": 0.8926365704161758,
      "learning_rate": 4.671065364861057e-06,
      "loss": 0.4471,
      "step": 8546
    },
    {
      "epoch": 0.5353669803786467,
      "grad_norm": 0.7509759440909816,
      "learning_rate": 4.67005317047459e-06,
      "loss": 0.3614,
      "step": 8547
    },
    {
      "epoch": 0.535429618377989,
      "grad_norm": 0.8579693102480402,
      "learning_rate": 4.669040989668794e-06,
      "loss": 0.4106,
      "step": 8548
    },
    {
      "epoch": 0.5354922563773313,
      "grad_norm": 0.7684435886907219,
      "learning_rate": 4.668028822485332e-06,
      "loss": 0.4237,
      "step": 8549
    },
    {
      "epoch": 0.5355548943766736,
      "grad_norm": 0.8139181867172096,
      "learning_rate": 4.667016668965868e-06,
      "loss": 0.3616,
      "step": 8550
    },
    {
      "epoch": 0.535617532376016,
      "grad_norm": 0.8166356933554751,
      "learning_rate": 4.666004529152059e-06,
      "loss": 0.3781,
      "step": 8551
    },
    {
      "epoch": 0.5356801703753582,
      "grad_norm": 0.8482998427779693,
      "learning_rate": 4.6649924030855655e-06,
      "loss": 0.4002,
      "step": 8552
    },
    {
      "epoch": 0.5357428083747006,
      "grad_norm": 0.8131137198362869,
      "learning_rate": 4.663980290808046e-06,
      "loss": 0.3703,
      "step": 8553
    },
    {
      "epoch": 0.5358054463740428,
      "grad_norm": 0.7765144377823517,
      "learning_rate": 4.662968192361161e-06,
      "loss": 0.3832,
      "step": 8554
    },
    {
      "epoch": 0.5358680843733851,
      "grad_norm": 0.8498136547141185,
      "learning_rate": 4.661956107786568e-06,
      "loss": 0.4226,
      "step": 8555
    },
    {
      "epoch": 0.5359307223727274,
      "grad_norm": 0.814803841522404,
      "learning_rate": 4.660944037125924e-06,
      "loss": 0.3919,
      "step": 8556
    },
    {
      "epoch": 0.5359933603720697,
      "grad_norm": 0.7763152395018666,
      "learning_rate": 4.659931980420885e-06,
      "loss": 0.3954,
      "step": 8557
    },
    {
      "epoch": 0.536055998371412,
      "grad_norm": 0.9032277520218136,
      "learning_rate": 4.658919937713109e-06,
      "loss": 0.4019,
      "step": 8558
    },
    {
      "epoch": 0.5361186363707543,
      "grad_norm": 0.9020945295785122,
      "learning_rate": 4.657907909044253e-06,
      "loss": 0.4094,
      "step": 8559
    },
    {
      "epoch": 0.5361812743700967,
      "grad_norm": 0.8292600981727662,
      "learning_rate": 4.6568958944559705e-06,
      "loss": 0.389,
      "step": 8560
    },
    {
      "epoch": 0.5362439123694389,
      "grad_norm": 0.8171194737431698,
      "learning_rate": 4.655883893989918e-06,
      "loss": 0.3727,
      "step": 8561
    },
    {
      "epoch": 0.5363065503687813,
      "grad_norm": 0.8181240508414599,
      "learning_rate": 4.654871907687746e-06,
      "loss": 0.3951,
      "step": 8562
    },
    {
      "epoch": 0.5363691883681235,
      "grad_norm": 0.8547815329356324,
      "learning_rate": 4.653859935591112e-06,
      "loss": 0.4356,
      "step": 8563
    },
    {
      "epoch": 0.5364318263674658,
      "grad_norm": 0.8780003507940628,
      "learning_rate": 4.652847977741667e-06,
      "loss": 0.3929,
      "step": 8564
    },
    {
      "epoch": 0.5364944643668081,
      "grad_norm": 0.8265740243799823,
      "learning_rate": 4.651836034181063e-06,
      "loss": 0.3996,
      "step": 8565
    },
    {
      "epoch": 0.5365571023661504,
      "grad_norm": 0.881561027878183,
      "learning_rate": 4.650824104950953e-06,
      "loss": 0.4226,
      "step": 8566
    },
    {
      "epoch": 0.5366197403654928,
      "grad_norm": 0.8365895144774037,
      "learning_rate": 4.649812190092985e-06,
      "loss": 0.4065,
      "step": 8567
    },
    {
      "epoch": 0.536682378364835,
      "grad_norm": 0.9169495859664819,
      "learning_rate": 4.648800289648815e-06,
      "loss": 0.4281,
      "step": 8568
    },
    {
      "epoch": 0.5367450163641774,
      "grad_norm": 0.8334993207763901,
      "learning_rate": 4.647788403660091e-06,
      "loss": 0.3852,
      "step": 8569
    },
    {
      "epoch": 0.5368076543635196,
      "grad_norm": 0.853030406626701,
      "learning_rate": 4.64677653216846e-06,
      "loss": 0.4066,
      "step": 8570
    },
    {
      "epoch": 0.5368702923628619,
      "grad_norm": 0.7452566029148869,
      "learning_rate": 4.645764675215574e-06,
      "loss": 0.3704,
      "step": 8571
    },
    {
      "epoch": 0.5369329303622042,
      "grad_norm": 0.8207728637478179,
      "learning_rate": 4.644752832843079e-06,
      "loss": 0.3742,
      "step": 8572
    },
    {
      "epoch": 0.5369955683615465,
      "grad_norm": 0.823992810637124,
      "learning_rate": 4.6437410050926236e-06,
      "loss": 0.4291,
      "step": 8573
    },
    {
      "epoch": 0.5370582063608889,
      "grad_norm": 0.9205656961402713,
      "learning_rate": 4.6427291920058535e-06,
      "loss": 0.3816,
      "step": 8574
    },
    {
      "epoch": 0.5371208443602311,
      "grad_norm": 0.7452103296375718,
      "learning_rate": 4.641717393624417e-06,
      "loss": 0.4049,
      "step": 8575
    },
    {
      "epoch": 0.5371834823595735,
      "grad_norm": 0.7358737639620359,
      "learning_rate": 4.640705609989957e-06,
      "loss": 0.4484,
      "step": 8576
    },
    {
      "epoch": 0.5372461203589157,
      "grad_norm": 0.8596648756980765,
      "learning_rate": 4.639693841144123e-06,
      "loss": 0.452,
      "step": 8577
    },
    {
      "epoch": 0.5373087583582581,
      "grad_norm": 0.8951858991581522,
      "learning_rate": 4.6386820871285585e-06,
      "loss": 0.4011,
      "step": 8578
    },
    {
      "epoch": 0.5373713963576003,
      "grad_norm": 0.7812270883013765,
      "learning_rate": 4.637670347984904e-06,
      "loss": 0.3585,
      "step": 8579
    },
    {
      "epoch": 0.5374340343569426,
      "grad_norm": 0.7892978790689239,
      "learning_rate": 4.636658623754806e-06,
      "loss": 0.3786,
      "step": 8580
    },
    {
      "epoch": 0.537496672356285,
      "grad_norm": 0.8490821482587102,
      "learning_rate": 4.6356469144799075e-06,
      "loss": 0.414,
      "step": 8581
    },
    {
      "epoch": 0.5375593103556272,
      "grad_norm": 0.7886319867892835,
      "learning_rate": 4.634635220201849e-06,
      "loss": 0.3619,
      "step": 8582
    },
    {
      "epoch": 0.5376219483549696,
      "grad_norm": 0.7866299135705869,
      "learning_rate": 4.633623540962272e-06,
      "loss": 0.4185,
      "step": 8583
    },
    {
      "epoch": 0.5376845863543118,
      "grad_norm": 0.6613530593009469,
      "learning_rate": 4.6326118768028185e-06,
      "loss": 0.479,
      "step": 8584
    },
    {
      "epoch": 0.5377472243536542,
      "grad_norm": 0.8724711337171289,
      "learning_rate": 4.631600227765129e-06,
      "loss": 0.4192,
      "step": 8585
    },
    {
      "epoch": 0.5378098623529964,
      "grad_norm": 0.8071383468556116,
      "learning_rate": 4.6305885938908404e-06,
      "loss": 0.3949,
      "step": 8586
    },
    {
      "epoch": 0.5378725003523388,
      "grad_norm": 0.8015165545687636,
      "learning_rate": 4.629576975221595e-06,
      "loss": 0.3861,
      "step": 8587
    },
    {
      "epoch": 0.537935138351681,
      "grad_norm": 0.8758734129316289,
      "learning_rate": 4.62856537179903e-06,
      "loss": 0.4287,
      "step": 8588
    },
    {
      "epoch": 0.5379977763510233,
      "grad_norm": 0.8081228425484251,
      "learning_rate": 4.627553783664785e-06,
      "loss": 0.4079,
      "step": 8589
    },
    {
      "epoch": 0.5380604143503657,
      "grad_norm": 0.7753723334877602,
      "learning_rate": 4.626542210860493e-06,
      "loss": 0.3559,
      "step": 8590
    },
    {
      "epoch": 0.5381230523497079,
      "grad_norm": 0.8332003906941802,
      "learning_rate": 4.625530653427794e-06,
      "loss": 0.4282,
      "step": 8591
    },
    {
      "epoch": 0.5381856903490503,
      "grad_norm": 0.77725966562478,
      "learning_rate": 4.624519111408322e-06,
      "loss": 0.3819,
      "step": 8592
    },
    {
      "epoch": 0.5382483283483925,
      "grad_norm": 0.8010264169383214,
      "learning_rate": 4.623507584843715e-06,
      "loss": 0.3947,
      "step": 8593
    },
    {
      "epoch": 0.5383109663477349,
      "grad_norm": 0.8052209415365797,
      "learning_rate": 4.622496073775605e-06,
      "loss": 0.3988,
      "step": 8594
    },
    {
      "epoch": 0.5383736043470772,
      "grad_norm": 0.8848100025075215,
      "learning_rate": 4.621484578245624e-06,
      "loss": 0.4131,
      "step": 8595
    },
    {
      "epoch": 0.5384362423464194,
      "grad_norm": 0.8253895407070881,
      "learning_rate": 4.620473098295409e-06,
      "loss": 0.385,
      "step": 8596
    },
    {
      "epoch": 0.5384988803457618,
      "grad_norm": 0.8045665862357558,
      "learning_rate": 4.619461633966594e-06,
      "loss": 0.3867,
      "step": 8597
    },
    {
      "epoch": 0.538561518345104,
      "grad_norm": 0.8315847778567144,
      "learning_rate": 4.618450185300808e-06,
      "loss": 0.3857,
      "step": 8598
    },
    {
      "epoch": 0.5386241563444464,
      "grad_norm": 0.8731585884412407,
      "learning_rate": 4.617438752339683e-06,
      "loss": 0.4493,
      "step": 8599
    },
    {
      "epoch": 0.5386867943437886,
      "grad_norm": 0.8514981147959592,
      "learning_rate": 4.616427335124849e-06,
      "loss": 0.4058,
      "step": 8600
    },
    {
      "epoch": 0.538749432343131,
      "grad_norm": 0.8863130777631777,
      "learning_rate": 4.615415933697939e-06,
      "loss": 0.399,
      "step": 8601
    },
    {
      "epoch": 0.5388120703424732,
      "grad_norm": 0.8769595733675947,
      "learning_rate": 4.6144045481005795e-06,
      "loss": 0.4263,
      "step": 8602
    },
    {
      "epoch": 0.5388747083418156,
      "grad_norm": 0.9247867768786829,
      "learning_rate": 4.6133931783744e-06,
      "loss": 0.4412,
      "step": 8603
    },
    {
      "epoch": 0.5389373463411579,
      "grad_norm": 0.8280747952883598,
      "learning_rate": 4.612381824561028e-06,
      "loss": 0.3959,
      "step": 8604
    },
    {
      "epoch": 0.5389999843405001,
      "grad_norm": 0.8047901820033639,
      "learning_rate": 4.611370486702092e-06,
      "loss": 0.3655,
      "step": 8605
    },
    {
      "epoch": 0.5390626223398425,
      "grad_norm": 0.8120344163829782,
      "learning_rate": 4.610359164839221e-06,
      "loss": 0.3861,
      "step": 8606
    },
    {
      "epoch": 0.5391252603391847,
      "grad_norm": 0.8302240187869443,
      "learning_rate": 4.60934785901404e-06,
      "loss": 0.3776,
      "step": 8607
    },
    {
      "epoch": 0.5391878983385271,
      "grad_norm": 0.8224426673396226,
      "learning_rate": 4.608336569268171e-06,
      "loss": 0.422,
      "step": 8608
    },
    {
      "epoch": 0.5392505363378693,
      "grad_norm": 0.7592416007342319,
      "learning_rate": 4.607325295643243e-06,
      "loss": 0.3774,
      "step": 8609
    },
    {
      "epoch": 0.5393131743372117,
      "grad_norm": 0.847112958027747,
      "learning_rate": 4.606314038180878e-06,
      "loss": 0.4368,
      "step": 8610
    },
    {
      "epoch": 0.539375812336554,
      "grad_norm": 0.8000871252012979,
      "learning_rate": 4.605302796922701e-06,
      "loss": 0.3833,
      "step": 8611
    },
    {
      "epoch": 0.5394384503358963,
      "grad_norm": 0.8318448451076517,
      "learning_rate": 4.6042915719103336e-06,
      "loss": 0.3964,
      "step": 8612
    },
    {
      "epoch": 0.5395010883352386,
      "grad_norm": 0.7600354999714688,
      "learning_rate": 4.603280363185397e-06,
      "loss": 0.3511,
      "step": 8613
    },
    {
      "epoch": 0.5395637263345808,
      "grad_norm": 0.8103888192305806,
      "learning_rate": 4.602269170789515e-06,
      "loss": 0.3629,
      "step": 8614
    },
    {
      "epoch": 0.5396263643339232,
      "grad_norm": 0.8245287900289039,
      "learning_rate": 4.601257994764309e-06,
      "loss": 0.3631,
      "step": 8615
    },
    {
      "epoch": 0.5396890023332654,
      "grad_norm": 0.8399591736297621,
      "learning_rate": 4.600246835151399e-06,
      "loss": 0.4082,
      "step": 8616
    },
    {
      "epoch": 0.5397516403326078,
      "grad_norm": 0.9099926198030557,
      "learning_rate": 4.599235691992401e-06,
      "loss": 0.452,
      "step": 8617
    },
    {
      "epoch": 0.5398142783319501,
      "grad_norm": 0.8838386037275572,
      "learning_rate": 4.598224565328938e-06,
      "loss": 0.4362,
      "step": 8618
    },
    {
      "epoch": 0.5398769163312924,
      "grad_norm": 0.8586504368342553,
      "learning_rate": 4.5972134552026255e-06,
      "loss": 0.4362,
      "step": 8619
    },
    {
      "epoch": 0.5399395543306347,
      "grad_norm": 0.8613257333022379,
      "learning_rate": 4.596202361655083e-06,
      "loss": 0.4086,
      "step": 8620
    },
    {
      "epoch": 0.5400021923299769,
      "grad_norm": 0.8578673873342019,
      "learning_rate": 4.595191284727925e-06,
      "loss": 0.3817,
      "step": 8621
    },
    {
      "epoch": 0.5400648303293193,
      "grad_norm": 0.8354815538930884,
      "learning_rate": 4.594180224462767e-06,
      "loss": 0.3736,
      "step": 8622
    },
    {
      "epoch": 0.5401274683286615,
      "grad_norm": 0.8251973649942925,
      "learning_rate": 4.5931691809012294e-06,
      "loss": 0.4023,
      "step": 8623
    },
    {
      "epoch": 0.5401901063280039,
      "grad_norm": 0.8183495101240186,
      "learning_rate": 4.592158154084923e-06,
      "loss": 0.3928,
      "step": 8624
    },
    {
      "epoch": 0.5402527443273462,
      "grad_norm": 0.7651899046276451,
      "learning_rate": 4.591147144055464e-06,
      "loss": 0.3778,
      "step": 8625
    },
    {
      "epoch": 0.5403153823266885,
      "grad_norm": 0.7885767017023846,
      "learning_rate": 4.590136150854462e-06,
      "loss": 0.3544,
      "step": 8626
    },
    {
      "epoch": 0.5403780203260308,
      "grad_norm": 0.8419296695730727,
      "learning_rate": 4.589125174523533e-06,
      "loss": 0.398,
      "step": 8627
    },
    {
      "epoch": 0.5404406583253731,
      "grad_norm": 0.8168406971301244,
      "learning_rate": 4.588114215104289e-06,
      "loss": 0.4088,
      "step": 8628
    },
    {
      "epoch": 0.5405032963247154,
      "grad_norm": 0.815804311826966,
      "learning_rate": 4.587103272638339e-06,
      "loss": 0.3895,
      "step": 8629
    },
    {
      "epoch": 0.5405659343240576,
      "grad_norm": 0.8018184675947149,
      "learning_rate": 4.586092347167296e-06,
      "loss": 0.3616,
      "step": 8630
    },
    {
      "epoch": 0.5406285723234,
      "grad_norm": 0.8628797355372565,
      "learning_rate": 4.585081438732769e-06,
      "loss": 0.4076,
      "step": 8631
    },
    {
      "epoch": 0.5406912103227423,
      "grad_norm": 0.7427170815601744,
      "learning_rate": 4.5840705473763645e-06,
      "loss": 0.3519,
      "step": 8632
    },
    {
      "epoch": 0.5407538483220846,
      "grad_norm": 0.7645466312216643,
      "learning_rate": 4.583059673139696e-06,
      "loss": 0.3796,
      "step": 8633
    },
    {
      "epoch": 0.5408164863214269,
      "grad_norm": 0.8360453518880107,
      "learning_rate": 4.582048816064367e-06,
      "loss": 0.3919,
      "step": 8634
    },
    {
      "epoch": 0.5408791243207692,
      "grad_norm": 0.8049227798919629,
      "learning_rate": 4.581037976191988e-06,
      "loss": 0.4014,
      "step": 8635
    },
    {
      "epoch": 0.5409417623201115,
      "grad_norm": 0.8514753489589875,
      "learning_rate": 4.580027153564164e-06,
      "loss": 0.4261,
      "step": 8636
    },
    {
      "epoch": 0.5410044003194538,
      "grad_norm": 0.6091840274831328,
      "learning_rate": 4.5790163482225e-06,
      "loss": 0.4512,
      "step": 8637
    },
    {
      "epoch": 0.5410670383187961,
      "grad_norm": 0.7716727883392713,
      "learning_rate": 4.5780055602085995e-06,
      "loss": 0.3965,
      "step": 8638
    },
    {
      "epoch": 0.5411296763181384,
      "grad_norm": 0.8743585227800618,
      "learning_rate": 4.57699478956407e-06,
      "loss": 0.4063,
      "step": 8639
    },
    {
      "epoch": 0.5411923143174807,
      "grad_norm": 0.8289444642992058,
      "learning_rate": 4.575984036330514e-06,
      "loss": 0.4463,
      "step": 8640
    },
    {
      "epoch": 0.541254952316823,
      "grad_norm": 0.8541505830537509,
      "learning_rate": 4.574973300549531e-06,
      "loss": 0.3802,
      "step": 8641
    },
    {
      "epoch": 0.5413175903161653,
      "grad_norm": 0.8139541388282158,
      "learning_rate": 4.573962582262727e-06,
      "loss": 0.3652,
      "step": 8642
    },
    {
      "epoch": 0.5413802283155076,
      "grad_norm": 0.8869633255862771,
      "learning_rate": 4.572951881511703e-06,
      "loss": 0.4124,
      "step": 8643
    },
    {
      "epoch": 0.54144286631485,
      "grad_norm": 1.0166344685242523,
      "learning_rate": 4.571941198338059e-06,
      "loss": 0.4098,
      "step": 8644
    },
    {
      "epoch": 0.5415055043141922,
      "grad_norm": 0.7647582203903737,
      "learning_rate": 4.570930532783395e-06,
      "loss": 0.422,
      "step": 8645
    },
    {
      "epoch": 0.5415681423135345,
      "grad_norm": 0.8116078983050721,
      "learning_rate": 4.56991988488931e-06,
      "loss": 0.4085,
      "step": 8646
    },
    {
      "epoch": 0.5416307803128768,
      "grad_norm": 0.8699580690698412,
      "learning_rate": 4.568909254697401e-06,
      "loss": 0.3593,
      "step": 8647
    },
    {
      "epoch": 0.5416934183122191,
      "grad_norm": 0.8468522791304863,
      "learning_rate": 4.567898642249268e-06,
      "loss": 0.4067,
      "step": 8648
    },
    {
      "epoch": 0.5417560563115614,
      "grad_norm": 0.8194267656329062,
      "learning_rate": 4.5668880475865074e-06,
      "loss": 0.3992,
      "step": 8649
    },
    {
      "epoch": 0.5418186943109037,
      "grad_norm": 0.8714582716115331,
      "learning_rate": 4.565877470750712e-06,
      "loss": 0.4208,
      "step": 8650
    },
    {
      "epoch": 0.541881332310246,
      "grad_norm": 0.8084668805087262,
      "learning_rate": 4.564866911783484e-06,
      "loss": 0.3719,
      "step": 8651
    },
    {
      "epoch": 0.5419439703095883,
      "grad_norm": 0.8234704420274026,
      "learning_rate": 4.563856370726413e-06,
      "loss": 0.3818,
      "step": 8652
    },
    {
      "epoch": 0.5420066083089307,
      "grad_norm": 0.8325303019058051,
      "learning_rate": 4.562845847621096e-06,
      "loss": 0.4147,
      "step": 8653
    },
    {
      "epoch": 0.5420692463082729,
      "grad_norm": 0.8659306993011628,
      "learning_rate": 4.561835342509126e-06,
      "loss": 0.4233,
      "step": 8654
    },
    {
      "epoch": 0.5421318843076152,
      "grad_norm": 0.8160510977057552,
      "learning_rate": 4.560824855432092e-06,
      "loss": 0.3946,
      "step": 8655
    },
    {
      "epoch": 0.5421945223069575,
      "grad_norm": 0.8209282441017852,
      "learning_rate": 4.559814386431591e-06,
      "loss": 0.369,
      "step": 8656
    },
    {
      "epoch": 0.5422571603062998,
      "grad_norm": 0.8180717869808087,
      "learning_rate": 4.55880393554921e-06,
      "loss": 0.3992,
      "step": 8657
    },
    {
      "epoch": 0.5423197983056421,
      "grad_norm": 0.8393114174532597,
      "learning_rate": 4.557793502826542e-06,
      "loss": 0.4002,
      "step": 8658
    },
    {
      "epoch": 0.5423824363049844,
      "grad_norm": 0.7892594922254912,
      "learning_rate": 4.556783088305172e-06,
      "loss": 0.3962,
      "step": 8659
    },
    {
      "epoch": 0.5424450743043268,
      "grad_norm": 0.9296505174570873,
      "learning_rate": 4.555772692026694e-06,
      "loss": 0.3744,
      "step": 8660
    },
    {
      "epoch": 0.542507712303669,
      "grad_norm": 0.8180178533451575,
      "learning_rate": 4.554762314032697e-06,
      "loss": 0.4235,
      "step": 8661
    },
    {
      "epoch": 0.5425703503030114,
      "grad_norm": 0.8773522202944793,
      "learning_rate": 4.553751954364764e-06,
      "loss": 0.4044,
      "step": 8662
    },
    {
      "epoch": 0.5426329883023536,
      "grad_norm": 0.8572873404528825,
      "learning_rate": 4.552741613064484e-06,
      "loss": 0.4066,
      "step": 8663
    },
    {
      "epoch": 0.5426956263016959,
      "grad_norm": 0.8183542124555219,
      "learning_rate": 4.551731290173441e-06,
      "loss": 0.4067,
      "step": 8664
    },
    {
      "epoch": 0.5427582643010382,
      "grad_norm": 0.8855322071903567,
      "learning_rate": 4.550720985733223e-06,
      "loss": 0.4221,
      "step": 8665
    },
    {
      "epoch": 0.5428209023003805,
      "grad_norm": 0.8563454968304164,
      "learning_rate": 4.549710699785412e-06,
      "loss": 0.3789,
      "step": 8666
    },
    {
      "epoch": 0.5428835402997229,
      "grad_norm": 0.8497230836807781,
      "learning_rate": 4.5487004323715926e-06,
      "loss": 0.4103,
      "step": 8667
    },
    {
      "epoch": 0.5429461782990651,
      "grad_norm": 0.9074384666512223,
      "learning_rate": 4.5476901835333455e-06,
      "loss": 0.3896,
      "step": 8668
    },
    {
      "epoch": 0.5430088162984075,
      "grad_norm": 0.9128556546408672,
      "learning_rate": 4.546679953312253e-06,
      "loss": 0.4159,
      "step": 8669
    },
    {
      "epoch": 0.5430714542977497,
      "grad_norm": 0.7722160476202012,
      "learning_rate": 4.545669741749901e-06,
      "loss": 0.3669,
      "step": 8670
    },
    {
      "epoch": 0.5431340922970921,
      "grad_norm": 0.8229033332554118,
      "learning_rate": 4.544659548887867e-06,
      "loss": 0.3931,
      "step": 8671
    },
    {
      "epoch": 0.5431967302964343,
      "grad_norm": 0.7861989102161862,
      "learning_rate": 4.543649374767729e-06,
      "loss": 0.3683,
      "step": 8672
    },
    {
      "epoch": 0.5432593682957766,
      "grad_norm": 0.8266903732747362,
      "learning_rate": 4.542639219431067e-06,
      "loss": 0.3567,
      "step": 8673
    },
    {
      "epoch": 0.543322006295119,
      "grad_norm": 0.8021200967062938,
      "learning_rate": 4.541629082919461e-06,
      "loss": 0.3705,
      "step": 8674
    },
    {
      "epoch": 0.5433846442944612,
      "grad_norm": 0.8551118188032625,
      "learning_rate": 4.540618965274487e-06,
      "loss": 0.4076,
      "step": 8675
    },
    {
      "epoch": 0.5434472822938036,
      "grad_norm": 0.6751752733764843,
      "learning_rate": 4.5396088665377205e-06,
      "loss": 0.476,
      "step": 8676
    },
    {
      "epoch": 0.5435099202931458,
      "grad_norm": 0.7984399198367409,
      "learning_rate": 4.538598786750739e-06,
      "loss": 0.4066,
      "step": 8677
    },
    {
      "epoch": 0.5435725582924882,
      "grad_norm": 0.8356154994630994,
      "learning_rate": 4.537588725955117e-06,
      "loss": 0.4255,
      "step": 8678
    },
    {
      "epoch": 0.5436351962918304,
      "grad_norm": 0.7608267430054005,
      "learning_rate": 4.536578684192429e-06,
      "loss": 0.3734,
      "step": 8679
    },
    {
      "epoch": 0.5436978342911727,
      "grad_norm": 0.8516350651978298,
      "learning_rate": 4.535568661504249e-06,
      "loss": 0.3968,
      "step": 8680
    },
    {
      "epoch": 0.543760472290515,
      "grad_norm": 0.8201126000276872,
      "learning_rate": 4.534558657932148e-06,
      "loss": 0.4031,
      "step": 8681
    },
    {
      "epoch": 0.5438231102898573,
      "grad_norm": 0.8843213812247163,
      "learning_rate": 4.5335486735176995e-06,
      "loss": 0.3961,
      "step": 8682
    },
    {
      "epoch": 0.5438857482891997,
      "grad_norm": 0.8395661200317318,
      "learning_rate": 4.532538708302475e-06,
      "loss": 0.4163,
      "step": 8683
    },
    {
      "epoch": 0.5439483862885419,
      "grad_norm": 0.8111918120104595,
      "learning_rate": 4.531528762328044e-06,
      "loss": 0.4038,
      "step": 8684
    },
    {
      "epoch": 0.5440110242878843,
      "grad_norm": 0.8891191016934314,
      "learning_rate": 4.530518835635974e-06,
      "loss": 0.3981,
      "step": 8685
    },
    {
      "epoch": 0.5440736622872265,
      "grad_norm": 0.8759219434550524,
      "learning_rate": 4.5295089282678375e-06,
      "loss": 0.3591,
      "step": 8686
    },
    {
      "epoch": 0.5441363002865689,
      "grad_norm": 0.826977552948902,
      "learning_rate": 4.528499040265199e-06,
      "loss": 0.4064,
      "step": 8687
    },
    {
      "epoch": 0.5441989382859111,
      "grad_norm": 0.8676448647771443,
      "learning_rate": 4.52748917166963e-06,
      "loss": 0.4398,
      "step": 8688
    },
    {
      "epoch": 0.5442615762852534,
      "grad_norm": 0.8796027814567974,
      "learning_rate": 4.526479322522694e-06,
      "loss": 0.4048,
      "step": 8689
    },
    {
      "epoch": 0.5443242142845958,
      "grad_norm": 0.714108980385441,
      "learning_rate": 4.525469492865956e-06,
      "loss": 0.4666,
      "step": 8690
    },
    {
      "epoch": 0.544386852283938,
      "grad_norm": 0.8524017786867305,
      "learning_rate": 4.5244596827409824e-06,
      "loss": 0.4013,
      "step": 8691
    },
    {
      "epoch": 0.5444494902832804,
      "grad_norm": 0.9007358271880064,
      "learning_rate": 4.523449892189336e-06,
      "loss": 0.4231,
      "step": 8692
    },
    {
      "epoch": 0.5445121282826226,
      "grad_norm": 0.8134751362843943,
      "learning_rate": 4.522440121252581e-06,
      "loss": 0.3707,
      "step": 8693
    },
    {
      "epoch": 0.544574766281965,
      "grad_norm": 0.7985911566517242,
      "learning_rate": 4.521430369972279e-06,
      "loss": 0.3876,
      "step": 8694
    },
    {
      "epoch": 0.5446374042813072,
      "grad_norm": 0.7915629605426515,
      "learning_rate": 4.520420638389992e-06,
      "loss": 0.4001,
      "step": 8695
    },
    {
      "epoch": 0.5447000422806496,
      "grad_norm": 0.8311040898597415,
      "learning_rate": 4.519410926547279e-06,
      "loss": 0.4003,
      "step": 8696
    },
    {
      "epoch": 0.5447626802799919,
      "grad_norm": 0.8002243994691046,
      "learning_rate": 4.518401234485703e-06,
      "loss": 0.3773,
      "step": 8697
    },
    {
      "epoch": 0.5448253182793341,
      "grad_norm": 0.8579145823377928,
      "learning_rate": 4.517391562246823e-06,
      "loss": 0.4315,
      "step": 8698
    },
    {
      "epoch": 0.5448879562786765,
      "grad_norm": 0.8001699805475285,
      "learning_rate": 4.516381909872193e-06,
      "loss": 0.3441,
      "step": 8699
    },
    {
      "epoch": 0.5449505942780187,
      "grad_norm": 0.8400720033006611,
      "learning_rate": 4.515372277403375e-06,
      "loss": 0.4262,
      "step": 8700
    },
    {
      "epoch": 0.5450132322773611,
      "grad_norm": 0.7657622810937764,
      "learning_rate": 4.514362664881924e-06,
      "loss": 0.3808,
      "step": 8701
    },
    {
      "epoch": 0.5450758702767033,
      "grad_norm": 0.8701869964318742,
      "learning_rate": 4.513353072349395e-06,
      "loss": 0.4247,
      "step": 8702
    },
    {
      "epoch": 0.5451385082760457,
      "grad_norm": 0.8422714699085098,
      "learning_rate": 4.512343499847345e-06,
      "loss": 0.4153,
      "step": 8703
    },
    {
      "epoch": 0.545201146275388,
      "grad_norm": 0.7949842710408792,
      "learning_rate": 4.511333947417327e-06,
      "loss": 0.3728,
      "step": 8704
    },
    {
      "epoch": 0.5452637842747302,
      "grad_norm": 0.8573995286008472,
      "learning_rate": 4.510324415100892e-06,
      "loss": 0.4522,
      "step": 8705
    },
    {
      "epoch": 0.5453264222740726,
      "grad_norm": 0.7838455481062667,
      "learning_rate": 4.509314902939597e-06,
      "loss": 0.4005,
      "step": 8706
    },
    {
      "epoch": 0.5453890602734148,
      "grad_norm": 0.8039399073888147,
      "learning_rate": 4.508305410974991e-06,
      "loss": 0.3748,
      "step": 8707
    },
    {
      "epoch": 0.5454516982727572,
      "grad_norm": 0.6359995722686133,
      "learning_rate": 4.507295939248627e-06,
      "loss": 0.4906,
      "step": 8708
    },
    {
      "epoch": 0.5455143362720994,
      "grad_norm": 0.7944841590378993,
      "learning_rate": 4.506286487802053e-06,
      "loss": 0.3773,
      "step": 8709
    },
    {
      "epoch": 0.5455769742714418,
      "grad_norm": 0.795086125180082,
      "learning_rate": 4.505277056676819e-06,
      "loss": 0.4048,
      "step": 8710
    },
    {
      "epoch": 0.5456396122707841,
      "grad_norm": 0.5886969868773282,
      "learning_rate": 4.504267645914472e-06,
      "loss": 0.4526,
      "step": 8711
    },
    {
      "epoch": 0.5457022502701264,
      "grad_norm": 0.8303443441273884,
      "learning_rate": 4.503258255556561e-06,
      "loss": 0.4021,
      "step": 8712
    },
    {
      "epoch": 0.5457648882694687,
      "grad_norm": 0.8200314616331927,
      "learning_rate": 4.502248885644634e-06,
      "loss": 0.3879,
      "step": 8713
    },
    {
      "epoch": 0.5458275262688109,
      "grad_norm": 0.8537446081599446,
      "learning_rate": 4.501239536220233e-06,
      "loss": 0.3904,
      "step": 8714
    },
    {
      "epoch": 0.5458901642681533,
      "grad_norm": 0.8205096783883978,
      "learning_rate": 4.5002302073249054e-06,
      "loss": 0.3742,
      "step": 8715
    },
    {
      "epoch": 0.5459528022674955,
      "grad_norm": 0.8376322835488214,
      "learning_rate": 4.499220899000195e-06,
      "loss": 0.3802,
      "step": 8716
    },
    {
      "epoch": 0.5460154402668379,
      "grad_norm": 0.8419011686963737,
      "learning_rate": 4.498211611287648e-06,
      "loss": 0.3904,
      "step": 8717
    },
    {
      "epoch": 0.5460780782661802,
      "grad_norm": 0.8342793856574692,
      "learning_rate": 4.497202344228803e-06,
      "loss": 0.4093,
      "step": 8718
    },
    {
      "epoch": 0.5461407162655225,
      "grad_norm": 0.6525500755173509,
      "learning_rate": 4.496193097865201e-06,
      "loss": 0.4583,
      "step": 8719
    },
    {
      "epoch": 0.5462033542648648,
      "grad_norm": 0.665222221007658,
      "learning_rate": 4.495183872238387e-06,
      "loss": 0.4668,
      "step": 8720
    },
    {
      "epoch": 0.5462659922642071,
      "grad_norm": 0.7392075911745356,
      "learning_rate": 4.4941746673898975e-06,
      "loss": 0.3919,
      "step": 8721
    },
    {
      "epoch": 0.5463286302635494,
      "grad_norm": 0.7795778012694874,
      "learning_rate": 4.493165483361274e-06,
      "loss": 0.3929,
      "step": 8722
    },
    {
      "epoch": 0.5463912682628916,
      "grad_norm": 0.7683705829475411,
      "learning_rate": 4.492156320194051e-06,
      "loss": 0.3888,
      "step": 8723
    },
    {
      "epoch": 0.546453906262234,
      "grad_norm": 0.8068147943991193,
      "learning_rate": 4.491147177929768e-06,
      "loss": 0.3937,
      "step": 8724
    },
    {
      "epoch": 0.5465165442615763,
      "grad_norm": 0.8205050749355604,
      "learning_rate": 4.490138056609962e-06,
      "loss": 0.3665,
      "step": 8725
    },
    {
      "epoch": 0.5465791822609186,
      "grad_norm": 0.7971698351116298,
      "learning_rate": 4.4891289562761695e-06,
      "loss": 0.3861,
      "step": 8726
    },
    {
      "epoch": 0.5466418202602609,
      "grad_norm": 0.8617131531755979,
      "learning_rate": 4.488119876969924e-06,
      "loss": 0.4268,
      "step": 8727
    },
    {
      "epoch": 0.5467044582596032,
      "grad_norm": 0.8649220392379688,
      "learning_rate": 4.487110818732759e-06,
      "loss": 0.4105,
      "step": 8728
    },
    {
      "epoch": 0.5467670962589455,
      "grad_norm": 0.8138600350671747,
      "learning_rate": 4.486101781606208e-06,
      "loss": 0.418,
      "step": 8729
    },
    {
      "epoch": 0.5468297342582877,
      "grad_norm": 0.7570525710096044,
      "learning_rate": 4.485092765631802e-06,
      "loss": 0.3878,
      "step": 8730
    },
    {
      "epoch": 0.5468923722576301,
      "grad_norm": 0.8990989684801048,
      "learning_rate": 4.484083770851075e-06,
      "loss": 0.4004,
      "step": 8731
    },
    {
      "epoch": 0.5469550102569724,
      "grad_norm": 0.7831740976983047,
      "learning_rate": 4.483074797305555e-06,
      "loss": 0.4157,
      "step": 8732
    },
    {
      "epoch": 0.5470176482563147,
      "grad_norm": 1.008093142520931,
      "learning_rate": 4.482065845036769e-06,
      "loss": 0.4394,
      "step": 8733
    },
    {
      "epoch": 0.547080286255657,
      "grad_norm": 0.6615527975185659,
      "learning_rate": 4.481056914086253e-06,
      "loss": 0.4787,
      "step": 8734
    },
    {
      "epoch": 0.5471429242549993,
      "grad_norm": 0.8499117728578817,
      "learning_rate": 4.48004800449553e-06,
      "loss": 0.3865,
      "step": 8735
    },
    {
      "epoch": 0.5472055622543416,
      "grad_norm": 0.9478365683252198,
      "learning_rate": 4.479039116306128e-06,
      "loss": 0.3941,
      "step": 8736
    },
    {
      "epoch": 0.547268200253684,
      "grad_norm": 0.8859089286349159,
      "learning_rate": 4.478030249559571e-06,
      "loss": 0.4668,
      "step": 8737
    },
    {
      "epoch": 0.5473308382530262,
      "grad_norm": 0.7594792551181221,
      "learning_rate": 4.477021404297388e-06,
      "loss": 0.3485,
      "step": 8738
    },
    {
      "epoch": 0.5473934762523684,
      "grad_norm": 0.8613535868814068,
      "learning_rate": 4.4760125805611e-06,
      "loss": 0.4086,
      "step": 8739
    },
    {
      "epoch": 0.5474561142517108,
      "grad_norm": 0.9092184394120338,
      "learning_rate": 4.475003778392231e-06,
      "loss": 0.4601,
      "step": 8740
    },
    {
      "epoch": 0.5475187522510531,
      "grad_norm": 0.9131311681127091,
      "learning_rate": 4.473994997832305e-06,
      "loss": 0.402,
      "step": 8741
    },
    {
      "epoch": 0.5475813902503954,
      "grad_norm": 0.8522904467865263,
      "learning_rate": 4.4729862389228395e-06,
      "loss": 0.3842,
      "step": 8742
    },
    {
      "epoch": 0.5476440282497377,
      "grad_norm": 0.83011499957043,
      "learning_rate": 4.471977501705361e-06,
      "loss": 0.3762,
      "step": 8743
    },
    {
      "epoch": 0.54770666624908,
      "grad_norm": 0.906469301906311,
      "learning_rate": 4.4709687862213866e-06,
      "loss": 0.3912,
      "step": 8744
    },
    {
      "epoch": 0.5477693042484223,
      "grad_norm": 0.842820391037041,
      "learning_rate": 4.469960092512434e-06,
      "loss": 0.4,
      "step": 8745
    },
    {
      "epoch": 0.5478319422477647,
      "grad_norm": 0.8599749344806793,
      "learning_rate": 4.468951420620023e-06,
      "loss": 0.4259,
      "step": 8746
    },
    {
      "epoch": 0.5478945802471069,
      "grad_norm": 0.8493718361881636,
      "learning_rate": 4.4679427705856716e-06,
      "loss": 0.3832,
      "step": 8747
    },
    {
      "epoch": 0.5479572182464492,
      "grad_norm": 0.7999333141781302,
      "learning_rate": 4.466934142450893e-06,
      "loss": 0.4297,
      "step": 8748
    },
    {
      "epoch": 0.5480198562457915,
      "grad_norm": 0.8718383415980234,
      "learning_rate": 4.4659255362572034e-06,
      "loss": 0.4225,
      "step": 8749
    },
    {
      "epoch": 0.5480824942451338,
      "grad_norm": 0.8085488540749443,
      "learning_rate": 4.464916952046119e-06,
      "loss": 0.4084,
      "step": 8750
    },
    {
      "epoch": 0.5481451322444761,
      "grad_norm": 0.8161412682035433,
      "learning_rate": 4.46390838985915e-06,
      "loss": 0.3929,
      "step": 8751
    },
    {
      "epoch": 0.5482077702438184,
      "grad_norm": 0.6189875968843953,
      "learning_rate": 4.462899849737814e-06,
      "loss": 0.4238,
      "step": 8752
    },
    {
      "epoch": 0.5482704082431608,
      "grad_norm": 0.8599309357599648,
      "learning_rate": 4.461891331723619e-06,
      "loss": 0.3957,
      "step": 8753
    },
    {
      "epoch": 0.548333046242503,
      "grad_norm": 0.8704519170320036,
      "learning_rate": 4.460882835858076e-06,
      "loss": 0.3666,
      "step": 8754
    },
    {
      "epoch": 0.5483956842418453,
      "grad_norm": 1.1039026884354772,
      "learning_rate": 4.459874362182696e-06,
      "loss": 0.4495,
      "step": 8755
    },
    {
      "epoch": 0.5484583222411876,
      "grad_norm": 0.9369652866182835,
      "learning_rate": 4.458865910738988e-06,
      "loss": 0.4188,
      "step": 8756
    },
    {
      "epoch": 0.5485209602405299,
      "grad_norm": 0.8634630691605202,
      "learning_rate": 4.4578574815684594e-06,
      "loss": 0.4308,
      "step": 8757
    },
    {
      "epoch": 0.5485835982398722,
      "grad_norm": 0.6180715008543457,
      "learning_rate": 4.456849074712617e-06,
      "loss": 0.454,
      "step": 8758
    },
    {
      "epoch": 0.5486462362392145,
      "grad_norm": 0.8462215663666632,
      "learning_rate": 4.4558406902129676e-06,
      "loss": 0.4028,
      "step": 8759
    },
    {
      "epoch": 0.5487088742385569,
      "grad_norm": 0.7990003863561067,
      "learning_rate": 4.454832328111017e-06,
      "loss": 0.4056,
      "step": 8760
    },
    {
      "epoch": 0.5487715122378991,
      "grad_norm": 0.8417851641851641,
      "learning_rate": 4.453823988448267e-06,
      "loss": 0.3871,
      "step": 8761
    },
    {
      "epoch": 0.5488341502372415,
      "grad_norm": 0.7965953607893494,
      "learning_rate": 4.452815671266224e-06,
      "loss": 0.4232,
      "step": 8762
    },
    {
      "epoch": 0.5488967882365837,
      "grad_norm": 0.8254643297115859,
      "learning_rate": 4.45180737660639e-06,
      "loss": 0.4092,
      "step": 8763
    },
    {
      "epoch": 0.548959426235926,
      "grad_norm": 0.8589993484463168,
      "learning_rate": 4.450799104510266e-06,
      "loss": 0.4521,
      "step": 8764
    },
    {
      "epoch": 0.5490220642352683,
      "grad_norm": 0.8171111604888377,
      "learning_rate": 4.449790855019353e-06,
      "loss": 0.3989,
      "step": 8765
    },
    {
      "epoch": 0.5490847022346106,
      "grad_norm": 0.7662102279502758,
      "learning_rate": 4.44878262817515e-06,
      "loss": 0.3804,
      "step": 8766
    },
    {
      "epoch": 0.549147340233953,
      "grad_norm": 0.8112324928141962,
      "learning_rate": 4.447774424019157e-06,
      "loss": 0.4252,
      "step": 8767
    },
    {
      "epoch": 0.5492099782332952,
      "grad_norm": 0.784753264871901,
      "learning_rate": 4.446766242592872e-06,
      "loss": 0.3772,
      "step": 8768
    },
    {
      "epoch": 0.5492726162326376,
      "grad_norm": 0.8477352082175696,
      "learning_rate": 4.44575808393779e-06,
      "loss": 0.4144,
      "step": 8769
    },
    {
      "epoch": 0.5493352542319798,
      "grad_norm": 0.82156319511374,
      "learning_rate": 4.444749948095407e-06,
      "loss": 0.4084,
      "step": 8770
    },
    {
      "epoch": 0.5493978922313222,
      "grad_norm": 0.7868513018952876,
      "learning_rate": 4.443741835107219e-06,
      "loss": 0.3804,
      "step": 8771
    },
    {
      "epoch": 0.5494605302306644,
      "grad_norm": 0.7609558785458117,
      "learning_rate": 4.442733745014722e-06,
      "loss": 0.3492,
      "step": 8772
    },
    {
      "epoch": 0.5495231682300067,
      "grad_norm": 0.8504765806639322,
      "learning_rate": 4.441725677859407e-06,
      "loss": 0.4286,
      "step": 8773
    },
    {
      "epoch": 0.549585806229349,
      "grad_norm": 0.8651755673633245,
      "learning_rate": 4.440717633682768e-06,
      "loss": 0.3964,
      "step": 8774
    },
    {
      "epoch": 0.5496484442286913,
      "grad_norm": 0.7729948871953441,
      "learning_rate": 4.4397096125262935e-06,
      "loss": 0.3818,
      "step": 8775
    },
    {
      "epoch": 0.5497110822280337,
      "grad_norm": 0.7730306680175655,
      "learning_rate": 4.438701614431476e-06,
      "loss": 0.3512,
      "step": 8776
    },
    {
      "epoch": 0.5497737202273759,
      "grad_norm": 0.8406537252115255,
      "learning_rate": 4.437693639439804e-06,
      "loss": 0.4084,
      "step": 8777
    },
    {
      "epoch": 0.5498363582267183,
      "grad_norm": 0.8434581406798797,
      "learning_rate": 4.436685687592766e-06,
      "loss": 0.3984,
      "step": 8778
    },
    {
      "epoch": 0.5498989962260605,
      "grad_norm": 0.8671994137802351,
      "learning_rate": 4.435677758931849e-06,
      "loss": 0.4175,
      "step": 8779
    },
    {
      "epoch": 0.5499616342254029,
      "grad_norm": 0.8286847190288255,
      "learning_rate": 4.434669853498541e-06,
      "loss": 0.3822,
      "step": 8780
    },
    {
      "epoch": 0.5500242722247451,
      "grad_norm": 0.8895303619791296,
      "learning_rate": 4.433661971334326e-06,
      "loss": 0.4056,
      "step": 8781
    },
    {
      "epoch": 0.5500869102240874,
      "grad_norm": 0.7174618594689168,
      "learning_rate": 4.432654112480691e-06,
      "loss": 0.3562,
      "step": 8782
    },
    {
      "epoch": 0.5501495482234298,
      "grad_norm": 0.8523831987463145,
      "learning_rate": 4.431646276979118e-06,
      "loss": 0.3855,
      "step": 8783
    },
    {
      "epoch": 0.550212186222772,
      "grad_norm": 0.8730891780325573,
      "learning_rate": 4.430638464871088e-06,
      "loss": 0.3982,
      "step": 8784
    },
    {
      "epoch": 0.5502748242221144,
      "grad_norm": 0.7642049438637409,
      "learning_rate": 4.429630676198085e-06,
      "loss": 0.409,
      "step": 8785
    },
    {
      "epoch": 0.5503374622214566,
      "grad_norm": 0.8611154384369291,
      "learning_rate": 4.428622911001591e-06,
      "loss": 0.4003,
      "step": 8786
    },
    {
      "epoch": 0.550400100220799,
      "grad_norm": 0.8101619530545775,
      "learning_rate": 4.427615169323082e-06,
      "loss": 0.4178,
      "step": 8787
    },
    {
      "epoch": 0.5504627382201412,
      "grad_norm": 0.858638762563344,
      "learning_rate": 4.4266074512040385e-06,
      "loss": 0.4239,
      "step": 8788
    },
    {
      "epoch": 0.5505253762194835,
      "grad_norm": 0.9115879557326895,
      "learning_rate": 4.425599756685939e-06,
      "loss": 0.44,
      "step": 8789
    },
    {
      "epoch": 0.5505880142188259,
      "grad_norm": 0.9257866497022995,
      "learning_rate": 4.4245920858102615e-06,
      "loss": 0.4175,
      "step": 8790
    },
    {
      "epoch": 0.5506506522181681,
      "grad_norm": 0.6778372789655936,
      "learning_rate": 4.4235844386184806e-06,
      "loss": 0.4594,
      "step": 8791
    },
    {
      "epoch": 0.5507132902175105,
      "grad_norm": 0.775576758657971,
      "learning_rate": 4.42257681515207e-06,
      "loss": 0.4469,
      "step": 8792
    },
    {
      "epoch": 0.5507759282168527,
      "grad_norm": 0.8009381239420793,
      "learning_rate": 4.421569215452506e-06,
      "loss": 0.3761,
      "step": 8793
    },
    {
      "epoch": 0.5508385662161951,
      "grad_norm": 0.8610251038604615,
      "learning_rate": 4.42056163956126e-06,
      "loss": 0.4102,
      "step": 8794
    },
    {
      "epoch": 0.5509012042155373,
      "grad_norm": 0.7581608704002927,
      "learning_rate": 4.419554087519805e-06,
      "loss": 0.3713,
      "step": 8795
    },
    {
      "epoch": 0.5509638422148797,
      "grad_norm": 0.8064117551188036,
      "learning_rate": 4.41854655936961e-06,
      "loss": 0.3927,
      "step": 8796
    },
    {
      "epoch": 0.551026480214222,
      "grad_norm": 0.8158863347466037,
      "learning_rate": 4.417539055152145e-06,
      "loss": 0.3602,
      "step": 8797
    },
    {
      "epoch": 0.5510891182135642,
      "grad_norm": 0.857207174921926,
      "learning_rate": 4.416531574908884e-06,
      "loss": 0.435,
      "step": 8798
    },
    {
      "epoch": 0.5511517562129066,
      "grad_norm": 0.8122475783704319,
      "learning_rate": 4.4155241186812906e-06,
      "loss": 0.3759,
      "step": 8799
    },
    {
      "epoch": 0.5512143942122488,
      "grad_norm": 0.806960575421135,
      "learning_rate": 4.414516686510833e-06,
      "loss": 0.395,
      "step": 8800
    },
    {
      "epoch": 0.5512770322115912,
      "grad_norm": 0.8842976469449171,
      "learning_rate": 4.4135092784389775e-06,
      "loss": 0.393,
      "step": 8801
    },
    {
      "epoch": 0.5513396702109334,
      "grad_norm": 0.8121133691128163,
      "learning_rate": 4.412501894507188e-06,
      "loss": 0.4093,
      "step": 8802
    },
    {
      "epoch": 0.5514023082102758,
      "grad_norm": 0.8671912793404551,
      "learning_rate": 4.411494534756931e-06,
      "loss": 0.4658,
      "step": 8803
    },
    {
      "epoch": 0.5514649462096181,
      "grad_norm": 0.846800354933488,
      "learning_rate": 4.410487199229667e-06,
      "loss": 0.4482,
      "step": 8804
    },
    {
      "epoch": 0.5515275842089604,
      "grad_norm": 0.84561616375093,
      "learning_rate": 4.40947988796686e-06,
      "loss": 0.3622,
      "step": 8805
    },
    {
      "epoch": 0.5515902222083027,
      "grad_norm": 0.6915392549651753,
      "learning_rate": 4.408472601009971e-06,
      "loss": 0.4559,
      "step": 8806
    },
    {
      "epoch": 0.5516528602076449,
      "grad_norm": 0.7923789968265642,
      "learning_rate": 4.407465338400457e-06,
      "loss": 0.3782,
      "step": 8807
    },
    {
      "epoch": 0.5517154982069873,
      "grad_norm": 0.863038959872379,
      "learning_rate": 4.406458100179781e-06,
      "loss": 0.4386,
      "step": 8808
    },
    {
      "epoch": 0.5517781362063295,
      "grad_norm": 0.8470063366982831,
      "learning_rate": 4.405450886389401e-06,
      "loss": 0.4342,
      "step": 8809
    },
    {
      "epoch": 0.5518407742056719,
      "grad_norm": 0.7858528930620371,
      "learning_rate": 4.4044436970707714e-06,
      "loss": 0.4226,
      "step": 8810
    },
    {
      "epoch": 0.5519034122050142,
      "grad_norm": 0.7548632596406363,
      "learning_rate": 4.403436532265351e-06,
      "loss": 0.3823,
      "step": 8811
    },
    {
      "epoch": 0.5519660502043565,
      "grad_norm": 0.614726727433017,
      "learning_rate": 4.402429392014593e-06,
      "loss": 0.4582,
      "step": 8812
    },
    {
      "epoch": 0.5520286882036988,
      "grad_norm": 0.8228313708530004,
      "learning_rate": 4.4014222763599516e-06,
      "loss": 0.3828,
      "step": 8813
    },
    {
      "epoch": 0.552091326203041,
      "grad_norm": 0.8159195601419343,
      "learning_rate": 4.400415185342881e-06,
      "loss": 0.3887,
      "step": 8814
    },
    {
      "epoch": 0.5521539642023834,
      "grad_norm": 0.8069859853286084,
      "learning_rate": 4.399408119004833e-06,
      "loss": 0.4061,
      "step": 8815
    },
    {
      "epoch": 0.5522166022017256,
      "grad_norm": 0.6667763572874912,
      "learning_rate": 4.398401077387256e-06,
      "loss": 0.4655,
      "step": 8816
    },
    {
      "epoch": 0.552279240201068,
      "grad_norm": 0.8482454626720464,
      "learning_rate": 4.397394060531605e-06,
      "loss": 0.4102,
      "step": 8817
    },
    {
      "epoch": 0.5523418782004103,
      "grad_norm": 0.7687748588907338,
      "learning_rate": 4.396387068479325e-06,
      "loss": 0.4094,
      "step": 8818
    },
    {
      "epoch": 0.5524045161997526,
      "grad_norm": 0.8258935600410918,
      "learning_rate": 4.395380101271866e-06,
      "loss": 0.4251,
      "step": 8819
    },
    {
      "epoch": 0.5524671541990949,
      "grad_norm": 0.8363685152289603,
      "learning_rate": 4.394373158950674e-06,
      "loss": 0.4312,
      "step": 8820
    },
    {
      "epoch": 0.5525297921984372,
      "grad_norm": 0.8509015673016521,
      "learning_rate": 4.393366241557197e-06,
      "loss": 0.4151,
      "step": 8821
    },
    {
      "epoch": 0.5525924301977795,
      "grad_norm": 0.8707405310547339,
      "learning_rate": 4.392359349132874e-06,
      "loss": 0.4011,
      "step": 8822
    },
    {
      "epoch": 0.5526550681971217,
      "grad_norm": 0.8265185552655989,
      "learning_rate": 4.391352481719156e-06,
      "loss": 0.4015,
      "step": 8823
    },
    {
      "epoch": 0.5527177061964641,
      "grad_norm": 0.8417618277481987,
      "learning_rate": 4.390345639357484e-06,
      "loss": 0.3801,
      "step": 8824
    },
    {
      "epoch": 0.5527803441958064,
      "grad_norm": 0.8461606397584447,
      "learning_rate": 4.389338822089294e-06,
      "loss": 0.377,
      "step": 8825
    },
    {
      "epoch": 0.5528429821951487,
      "grad_norm": 0.8363292649202372,
      "learning_rate": 4.388332029956035e-06,
      "loss": 0.4183,
      "step": 8826
    },
    {
      "epoch": 0.552905620194491,
      "grad_norm": 0.8378407044956516,
      "learning_rate": 4.387325262999141e-06,
      "loss": 0.3675,
      "step": 8827
    },
    {
      "epoch": 0.5529682581938333,
      "grad_norm": 0.8534021685930547,
      "learning_rate": 4.386318521260056e-06,
      "loss": 0.4481,
      "step": 8828
    },
    {
      "epoch": 0.5530308961931756,
      "grad_norm": 0.8560692195674516,
      "learning_rate": 4.385311804780213e-06,
      "loss": 0.4328,
      "step": 8829
    },
    {
      "epoch": 0.5530935341925179,
      "grad_norm": 0.8524958861893095,
      "learning_rate": 4.3843051136010486e-06,
      "loss": 0.4285,
      "step": 8830
    },
    {
      "epoch": 0.5531561721918602,
      "grad_norm": 0.8721633438256171,
      "learning_rate": 4.3832984477640025e-06,
      "loss": 0.4009,
      "step": 8831
    },
    {
      "epoch": 0.5532188101912024,
      "grad_norm": 0.8312988998727714,
      "learning_rate": 4.382291807310506e-06,
      "loss": 0.3877,
      "step": 8832
    },
    {
      "epoch": 0.5532814481905448,
      "grad_norm": 0.7987481141487346,
      "learning_rate": 4.381285192281994e-06,
      "loss": 0.3888,
      "step": 8833
    },
    {
      "epoch": 0.5533440861898871,
      "grad_norm": 0.8953325808712677,
      "learning_rate": 4.380278602719895e-06,
      "loss": 0.3842,
      "step": 8834
    },
    {
      "epoch": 0.5534067241892294,
      "grad_norm": 0.8204572795839753,
      "learning_rate": 4.379272038665646e-06,
      "loss": 0.3832,
      "step": 8835
    },
    {
      "epoch": 0.5534693621885717,
      "grad_norm": 0.8469574353111705,
      "learning_rate": 4.378265500160675e-06,
      "loss": 0.3732,
      "step": 8836
    },
    {
      "epoch": 0.553532000187914,
      "grad_norm": 0.8257813581919773,
      "learning_rate": 4.377258987246412e-06,
      "loss": 0.4066,
      "step": 8837
    },
    {
      "epoch": 0.5535946381872563,
      "grad_norm": 0.8797394670334177,
      "learning_rate": 4.376252499964285e-06,
      "loss": 0.421,
      "step": 8838
    },
    {
      "epoch": 0.5536572761865985,
      "grad_norm": 0.620202706154314,
      "learning_rate": 4.3752460383557195e-06,
      "loss": 0.441,
      "step": 8839
    },
    {
      "epoch": 0.5537199141859409,
      "grad_norm": 0.9287154011994591,
      "learning_rate": 4.374239602462144e-06,
      "loss": 0.4299,
      "step": 8840
    },
    {
      "epoch": 0.5537825521852832,
      "grad_norm": 0.8367555653620719,
      "learning_rate": 4.373233192324983e-06,
      "loss": 0.3846,
      "step": 8841
    },
    {
      "epoch": 0.5538451901846255,
      "grad_norm": 0.84926903706584,
      "learning_rate": 4.37222680798566e-06,
      "loss": 0.4077,
      "step": 8842
    },
    {
      "epoch": 0.5539078281839678,
      "grad_norm": 0.8237091325153442,
      "learning_rate": 4.371220449485596e-06,
      "loss": 0.4311,
      "step": 8843
    },
    {
      "epoch": 0.5539704661833101,
      "grad_norm": 0.7486922831732652,
      "learning_rate": 4.370214116866216e-06,
      "loss": 0.3897,
      "step": 8844
    },
    {
      "epoch": 0.5540331041826524,
      "grad_norm": 0.8474162529441699,
      "learning_rate": 4.3692078101689405e-06,
      "loss": 0.4032,
      "step": 8845
    },
    {
      "epoch": 0.5540957421819948,
      "grad_norm": 0.7991510967950041,
      "learning_rate": 4.368201529435189e-06,
      "loss": 0.3711,
      "step": 8846
    },
    {
      "epoch": 0.554158380181337,
      "grad_norm": 0.8121861409051884,
      "learning_rate": 4.3671952747063805e-06,
      "loss": 0.4227,
      "step": 8847
    },
    {
      "epoch": 0.5542210181806793,
      "grad_norm": 0.9007755424515778,
      "learning_rate": 4.36618904602393e-06,
      "loss": 0.4157,
      "step": 8848
    },
    {
      "epoch": 0.5542836561800216,
      "grad_norm": 0.8425458447368818,
      "learning_rate": 4.365182843429257e-06,
      "loss": 0.429,
      "step": 8849
    },
    {
      "epoch": 0.5543462941793639,
      "grad_norm": 0.8424287113443708,
      "learning_rate": 4.3641766669637765e-06,
      "loss": 0.3748,
      "step": 8850
    },
    {
      "epoch": 0.5544089321787062,
      "grad_norm": 0.6787925690033931,
      "learning_rate": 4.363170516668901e-06,
      "loss": 0.4604,
      "step": 8851
    },
    {
      "epoch": 0.5544715701780485,
      "grad_norm": 0.8056913629317355,
      "learning_rate": 4.362164392586046e-06,
      "loss": 0.4099,
      "step": 8852
    },
    {
      "epoch": 0.5545342081773909,
      "grad_norm": 0.9170905949242301,
      "learning_rate": 4.361158294756621e-06,
      "loss": 0.3978,
      "step": 8853
    },
    {
      "epoch": 0.5545968461767331,
      "grad_norm": 0.8803360549457399,
      "learning_rate": 4.3601522232220414e-06,
      "loss": 0.4232,
      "step": 8854
    },
    {
      "epoch": 0.5546594841760755,
      "grad_norm": 0.8894763821011104,
      "learning_rate": 4.359146178023715e-06,
      "loss": 0.4149,
      "step": 8855
    },
    {
      "epoch": 0.5547221221754177,
      "grad_norm": 0.8484053009108277,
      "learning_rate": 4.35814015920305e-06,
      "loss": 0.4223,
      "step": 8856
    },
    {
      "epoch": 0.55478476017476,
      "grad_norm": 0.7807841464093069,
      "learning_rate": 4.3571341668014565e-06,
      "loss": 0.3932,
      "step": 8857
    },
    {
      "epoch": 0.5548473981741023,
      "grad_norm": 0.8539629400610177,
      "learning_rate": 4.35612820086034e-06,
      "loss": 0.4263,
      "step": 8858
    },
    {
      "epoch": 0.5549100361734446,
      "grad_norm": 0.625455880717446,
      "learning_rate": 4.355122261421106e-06,
      "loss": 0.4527,
      "step": 8859
    },
    {
      "epoch": 0.554972674172787,
      "grad_norm": 0.8673560558867687,
      "learning_rate": 4.354116348525158e-06,
      "loss": 0.4293,
      "step": 8860
    },
    {
      "epoch": 0.5550353121721292,
      "grad_norm": 0.8450604396131685,
      "learning_rate": 4.353110462213903e-06,
      "loss": 0.416,
      "step": 8861
    },
    {
      "epoch": 0.5550979501714716,
      "grad_norm": 0.785718698890534,
      "learning_rate": 4.352104602528738e-06,
      "loss": 0.3662,
      "step": 8862
    },
    {
      "epoch": 0.5551605881708138,
      "grad_norm": 0.8004545942476894,
      "learning_rate": 4.351098769511071e-06,
      "loss": 0.4069,
      "step": 8863
    },
    {
      "epoch": 0.5552232261701561,
      "grad_norm": 0.83837965711179,
      "learning_rate": 4.350092963202299e-06,
      "loss": 0.3838,
      "step": 8864
    },
    {
      "epoch": 0.5552858641694984,
      "grad_norm": 0.8043836788642341,
      "learning_rate": 4.34908718364382e-06,
      "loss": 0.4006,
      "step": 8865
    },
    {
      "epoch": 0.5553485021688407,
      "grad_norm": 0.8401974257919825,
      "learning_rate": 4.348081430877034e-06,
      "loss": 0.3756,
      "step": 8866
    },
    {
      "epoch": 0.555411140168183,
      "grad_norm": 0.8030993033489529,
      "learning_rate": 4.3470757049433374e-06,
      "loss": 0.395,
      "step": 8867
    },
    {
      "epoch": 0.5554737781675253,
      "grad_norm": 0.8329897737397362,
      "learning_rate": 4.346070005884126e-06,
      "loss": 0.4035,
      "step": 8868
    },
    {
      "epoch": 0.5555364161668677,
      "grad_norm": 0.747957268107343,
      "learning_rate": 4.345064333740794e-06,
      "loss": 0.3626,
      "step": 8869
    },
    {
      "epoch": 0.5555990541662099,
      "grad_norm": 0.8587767417211878,
      "learning_rate": 4.344058688554737e-06,
      "loss": 0.383,
      "step": 8870
    },
    {
      "epoch": 0.5556616921655523,
      "grad_norm": 0.888416706393778,
      "learning_rate": 4.343053070367342e-06,
      "loss": 0.4435,
      "step": 8871
    },
    {
      "epoch": 0.5557243301648945,
      "grad_norm": 0.8337335876001202,
      "learning_rate": 4.342047479220008e-06,
      "loss": 0.4195,
      "step": 8872
    },
    {
      "epoch": 0.5557869681642368,
      "grad_norm": 0.8310951030616605,
      "learning_rate": 4.341041915154123e-06,
      "loss": 0.3925,
      "step": 8873
    },
    {
      "epoch": 0.5558496061635791,
      "grad_norm": 0.797903168227397,
      "learning_rate": 4.3400363782110726e-06,
      "loss": 0.3623,
      "step": 8874
    },
    {
      "epoch": 0.5559122441629214,
      "grad_norm": 0.8915419499591158,
      "learning_rate": 4.339030868432249e-06,
      "loss": 0.4086,
      "step": 8875
    },
    {
      "epoch": 0.5559748821622638,
      "grad_norm": 0.8592020431000599,
      "learning_rate": 4.338025385859036e-06,
      "loss": 0.424,
      "step": 8876
    },
    {
      "epoch": 0.556037520161606,
      "grad_norm": 0.8720707072291513,
      "learning_rate": 4.337019930532822e-06,
      "loss": 0.3964,
      "step": 8877
    },
    {
      "epoch": 0.5561001581609484,
      "grad_norm": 0.8524034376029885,
      "learning_rate": 4.336014502494991e-06,
      "loss": 0.3893,
      "step": 8878
    },
    {
      "epoch": 0.5561627961602906,
      "grad_norm": 1.01460841446399,
      "learning_rate": 4.3350091017869265e-06,
      "loss": 0.4146,
      "step": 8879
    },
    {
      "epoch": 0.556225434159633,
      "grad_norm": 0.8752960827196645,
      "learning_rate": 4.334003728450008e-06,
      "loss": 0.4031,
      "step": 8880
    },
    {
      "epoch": 0.5562880721589752,
      "grad_norm": 0.8220877241363093,
      "learning_rate": 4.332998382525622e-06,
      "loss": 0.3741,
      "step": 8881
    },
    {
      "epoch": 0.5563507101583175,
      "grad_norm": 0.8287098983566771,
      "learning_rate": 4.331993064055146e-06,
      "loss": 0.4025,
      "step": 8882
    },
    {
      "epoch": 0.5564133481576599,
      "grad_norm": 0.8566036948241738,
      "learning_rate": 4.33098777307996e-06,
      "loss": 0.3973,
      "step": 8883
    },
    {
      "epoch": 0.5564759861570021,
      "grad_norm": 0.9072245507454003,
      "learning_rate": 4.329982509641442e-06,
      "loss": 0.4405,
      "step": 8884
    },
    {
      "epoch": 0.5565386241563445,
      "grad_norm": 0.7878760315139149,
      "learning_rate": 4.328977273780967e-06,
      "loss": 0.377,
      "step": 8885
    },
    {
      "epoch": 0.5566012621556867,
      "grad_norm": 0.8418760407582627,
      "learning_rate": 4.327972065539912e-06,
      "loss": 0.3698,
      "step": 8886
    },
    {
      "epoch": 0.5566639001550291,
      "grad_norm": 0.8258474278302255,
      "learning_rate": 4.326966884959653e-06,
      "loss": 0.3573,
      "step": 8887
    },
    {
      "epoch": 0.5567265381543713,
      "grad_norm": 0.8081291380361365,
      "learning_rate": 4.32596173208156e-06,
      "loss": 0.3855,
      "step": 8888
    },
    {
      "epoch": 0.5567891761537137,
      "grad_norm": 0.8640484230613181,
      "learning_rate": 4.324956606947007e-06,
      "loss": 0.3983,
      "step": 8889
    },
    {
      "epoch": 0.556851814153056,
      "grad_norm": 0.8501599580320645,
      "learning_rate": 4.323951509597366e-06,
      "loss": 0.4128,
      "step": 8890
    },
    {
      "epoch": 0.5569144521523982,
      "grad_norm": 0.9396763308155353,
      "learning_rate": 4.322946440074007e-06,
      "loss": 0.423,
      "step": 8891
    },
    {
      "epoch": 0.5569770901517406,
      "grad_norm": 0.9068851056933501,
      "learning_rate": 4.321941398418298e-06,
      "loss": 0.3906,
      "step": 8892
    },
    {
      "epoch": 0.5570397281510828,
      "grad_norm": 0.8031898678527488,
      "learning_rate": 4.320936384671607e-06,
      "loss": 0.364,
      "step": 8893
    },
    {
      "epoch": 0.5571023661504252,
      "grad_norm": 0.8605307725117637,
      "learning_rate": 4.319931398875301e-06,
      "loss": 0.4021,
      "step": 8894
    },
    {
      "epoch": 0.5571650041497674,
      "grad_norm": 0.8297527264787296,
      "learning_rate": 4.318926441070743e-06,
      "loss": 0.4354,
      "step": 8895
    },
    {
      "epoch": 0.5572276421491098,
      "grad_norm": 0.8801912287912068,
      "learning_rate": 4.3179215112993e-06,
      "loss": 0.4266,
      "step": 8896
    },
    {
      "epoch": 0.557290280148452,
      "grad_norm": 0.7916290741802737,
      "learning_rate": 4.316916609602335e-06,
      "loss": 0.401,
      "step": 8897
    },
    {
      "epoch": 0.5573529181477943,
      "grad_norm": 0.846472520389119,
      "learning_rate": 4.315911736021208e-06,
      "loss": 0.4161,
      "step": 8898
    },
    {
      "epoch": 0.5574155561471367,
      "grad_norm": 0.847841319452227,
      "learning_rate": 4.314906890597281e-06,
      "loss": 0.3904,
      "step": 8899
    },
    {
      "epoch": 0.5574781941464789,
      "grad_norm": 0.850044801842975,
      "learning_rate": 4.313902073371913e-06,
      "loss": 0.4157,
      "step": 8900
    },
    {
      "epoch": 0.5575408321458213,
      "grad_norm": 0.8544290406696509,
      "learning_rate": 4.312897284386464e-06,
      "loss": 0.4227,
      "step": 8901
    },
    {
      "epoch": 0.5576034701451635,
      "grad_norm": 0.8633695750405825,
      "learning_rate": 4.311892523682292e-06,
      "loss": 0.3918,
      "step": 8902
    },
    {
      "epoch": 0.5576661081445059,
      "grad_norm": 0.7787183327341837,
      "learning_rate": 4.310887791300748e-06,
      "loss": 0.3771,
      "step": 8903
    },
    {
      "epoch": 0.5577287461438482,
      "grad_norm": 0.8298417415018248,
      "learning_rate": 4.309883087283193e-06,
      "loss": 0.3851,
      "step": 8904
    },
    {
      "epoch": 0.5577913841431905,
      "grad_norm": 0.8630839124505457,
      "learning_rate": 4.308878411670978e-06,
      "loss": 0.4375,
      "step": 8905
    },
    {
      "epoch": 0.5578540221425328,
      "grad_norm": 0.8228516126025475,
      "learning_rate": 4.307873764505456e-06,
      "loss": 0.399,
      "step": 8906
    },
    {
      "epoch": 0.557916660141875,
      "grad_norm": 0.7636333513305394,
      "learning_rate": 4.306869145827977e-06,
      "loss": 0.3549,
      "step": 8907
    },
    {
      "epoch": 0.5579792981412174,
      "grad_norm": 0.8376629621758838,
      "learning_rate": 4.3058645556798905e-06,
      "loss": 0.4137,
      "step": 8908
    },
    {
      "epoch": 0.5580419361405596,
      "grad_norm": 0.7346850812716104,
      "learning_rate": 4.304859994102551e-06,
      "loss": 0.345,
      "step": 8909
    },
    {
      "epoch": 0.558104574139902,
      "grad_norm": 0.811745904410283,
      "learning_rate": 4.303855461137303e-06,
      "loss": 0.3908,
      "step": 8910
    },
    {
      "epoch": 0.5581672121392443,
      "grad_norm": 0.8501930243715233,
      "learning_rate": 4.302850956825493e-06,
      "loss": 0.4041,
      "step": 8911
    },
    {
      "epoch": 0.5582298501385866,
      "grad_norm": 0.7745382756901013,
      "learning_rate": 4.301846481208466e-06,
      "loss": 0.4003,
      "step": 8912
    },
    {
      "epoch": 0.5582924881379289,
      "grad_norm": 0.8245631729582189,
      "learning_rate": 4.300842034327569e-06,
      "loss": 0.3518,
      "step": 8913
    },
    {
      "epoch": 0.5583551261372712,
      "grad_norm": 0.8290656252150235,
      "learning_rate": 4.299837616224144e-06,
      "loss": 0.4037,
      "step": 8914
    },
    {
      "epoch": 0.5584177641366135,
      "grad_norm": 0.7911768064451823,
      "learning_rate": 4.298833226939531e-06,
      "loss": 0.3622,
      "step": 8915
    },
    {
      "epoch": 0.5584804021359557,
      "grad_norm": 0.8056336827607878,
      "learning_rate": 4.297828866515073e-06,
      "loss": 0.4149,
      "step": 8916
    },
    {
      "epoch": 0.5585430401352981,
      "grad_norm": 0.8573964978222457,
      "learning_rate": 4.296824534992108e-06,
      "loss": 0.4077,
      "step": 8917
    },
    {
      "epoch": 0.5586056781346403,
      "grad_norm": 0.8403239127561999,
      "learning_rate": 4.295820232411978e-06,
      "loss": 0.4097,
      "step": 8918
    },
    {
      "epoch": 0.5586683161339827,
      "grad_norm": 0.8164003771207038,
      "learning_rate": 4.294815958816017e-06,
      "loss": 0.414,
      "step": 8919
    },
    {
      "epoch": 0.558730954133325,
      "grad_norm": 0.8296211319289342,
      "learning_rate": 4.293811714245562e-06,
      "loss": 0.4419,
      "step": 8920
    },
    {
      "epoch": 0.5587935921326673,
      "grad_norm": 0.8126781383546376,
      "learning_rate": 4.292807498741948e-06,
      "loss": 0.4017,
      "step": 8921
    },
    {
      "epoch": 0.5588562301320096,
      "grad_norm": 0.7901943852430922,
      "learning_rate": 4.291803312346509e-06,
      "loss": 0.4245,
      "step": 8922
    },
    {
      "epoch": 0.5589188681313518,
      "grad_norm": 0.7828384404203131,
      "learning_rate": 4.290799155100577e-06,
      "loss": 0.4264,
      "step": 8923
    },
    {
      "epoch": 0.5589815061306942,
      "grad_norm": 0.7898492068696185,
      "learning_rate": 4.289795027045482e-06,
      "loss": 0.38,
      "step": 8924
    },
    {
      "epoch": 0.5590441441300364,
      "grad_norm": 0.7774481174283694,
      "learning_rate": 4.288790928222557e-06,
      "loss": 0.367,
      "step": 8925
    },
    {
      "epoch": 0.5591067821293788,
      "grad_norm": 0.8609254414132054,
      "learning_rate": 4.287786858673127e-06,
      "loss": 0.442,
      "step": 8926
    },
    {
      "epoch": 0.5591694201287211,
      "grad_norm": 0.9310645388924375,
      "learning_rate": 4.286782818438525e-06,
      "loss": 0.4586,
      "step": 8927
    },
    {
      "epoch": 0.5592320581280634,
      "grad_norm": 0.7942532033157346,
      "learning_rate": 4.285778807560074e-06,
      "loss": 0.3785,
      "step": 8928
    },
    {
      "epoch": 0.5592946961274057,
      "grad_norm": 0.6626434236879376,
      "learning_rate": 4.284774826079099e-06,
      "loss": 0.4642,
      "step": 8929
    },
    {
      "epoch": 0.559357334126748,
      "grad_norm": 0.8061063490734179,
      "learning_rate": 4.283770874036925e-06,
      "loss": 0.4261,
      "step": 8930
    },
    {
      "epoch": 0.5594199721260903,
      "grad_norm": 0.8549865513228989,
      "learning_rate": 4.282766951474876e-06,
      "loss": 0.3729,
      "step": 8931
    },
    {
      "epoch": 0.5594826101254325,
      "grad_norm": 0.8119693671084341,
      "learning_rate": 4.281763058434272e-06,
      "loss": 0.3927,
      "step": 8932
    },
    {
      "epoch": 0.5595452481247749,
      "grad_norm": 0.7926083624532693,
      "learning_rate": 4.280759194956433e-06,
      "loss": 0.3873,
      "step": 8933
    },
    {
      "epoch": 0.5596078861241172,
      "grad_norm": 0.8580109084862262,
      "learning_rate": 4.27975536108268e-06,
      "loss": 0.3941,
      "step": 8934
    },
    {
      "epoch": 0.5596705241234595,
      "grad_norm": 0.9118540385135707,
      "learning_rate": 4.278751556854327e-06,
      "loss": 0.4185,
      "step": 8935
    },
    {
      "epoch": 0.5597331621228018,
      "grad_norm": 0.8838124795049036,
      "learning_rate": 4.277747782312697e-06,
      "loss": 0.4155,
      "step": 8936
    },
    {
      "epoch": 0.5597958001221441,
      "grad_norm": 0.8494957262935877,
      "learning_rate": 4.276744037499102e-06,
      "loss": 0.4135,
      "step": 8937
    },
    {
      "epoch": 0.5598584381214864,
      "grad_norm": 0.6493866667079219,
      "learning_rate": 4.275740322454856e-06,
      "loss": 0.4303,
      "step": 8938
    },
    {
      "epoch": 0.5599210761208288,
      "grad_norm": 0.8495557728171185,
      "learning_rate": 4.2747366372212735e-06,
      "loss": 0.4072,
      "step": 8939
    },
    {
      "epoch": 0.559983714120171,
      "grad_norm": 0.7905104268841443,
      "learning_rate": 4.273732981839666e-06,
      "loss": 0.3739,
      "step": 8940
    },
    {
      "epoch": 0.5600463521195133,
      "grad_norm": 0.7873868246250928,
      "learning_rate": 4.272729356351342e-06,
      "loss": 0.3773,
      "step": 8941
    },
    {
      "epoch": 0.5601089901188556,
      "grad_norm": 0.841889091546919,
      "learning_rate": 4.271725760797614e-06,
      "loss": 0.3765,
      "step": 8942
    },
    {
      "epoch": 0.5601716281181979,
      "grad_norm": 0.9225513394890702,
      "learning_rate": 4.270722195219789e-06,
      "loss": 0.3766,
      "step": 8943
    },
    {
      "epoch": 0.5602342661175402,
      "grad_norm": 0.812170579006237,
      "learning_rate": 4.269718659659174e-06,
      "loss": 0.3744,
      "step": 8944
    },
    {
      "epoch": 0.5602969041168825,
      "grad_norm": 0.8788981492227589,
      "learning_rate": 4.268715154157071e-06,
      "loss": 0.4068,
      "step": 8945
    },
    {
      "epoch": 0.5603595421162249,
      "grad_norm": 0.9122530202080029,
      "learning_rate": 4.26771167875479e-06,
      "loss": 0.4042,
      "step": 8946
    },
    {
      "epoch": 0.5604221801155671,
      "grad_norm": 0.8871327684502051,
      "learning_rate": 4.266708233493633e-06,
      "loss": 0.4023,
      "step": 8947
    },
    {
      "epoch": 0.5604848181149094,
      "grad_norm": 0.7926108400389431,
      "learning_rate": 4.265704818414902e-06,
      "loss": 0.3887,
      "step": 8948
    },
    {
      "epoch": 0.5605474561142517,
      "grad_norm": 0.8587792016791875,
      "learning_rate": 4.264701433559897e-06,
      "loss": 0.3927,
      "step": 8949
    },
    {
      "epoch": 0.560610094113594,
      "grad_norm": 0.8213471993670857,
      "learning_rate": 4.263698078969916e-06,
      "loss": 0.4079,
      "step": 8950
    },
    {
      "epoch": 0.5606727321129363,
      "grad_norm": 0.8346638117294238,
      "learning_rate": 4.262694754686259e-06,
      "loss": 0.403,
      "step": 8951
    },
    {
      "epoch": 0.5607353701122786,
      "grad_norm": 0.8079861238483157,
      "learning_rate": 4.261691460750223e-06,
      "loss": 0.3864,
      "step": 8952
    },
    {
      "epoch": 0.560798008111621,
      "grad_norm": 0.7619154083263758,
      "learning_rate": 4.260688197203105e-06,
      "loss": 0.363,
      "step": 8953
    },
    {
      "epoch": 0.5608606461109632,
      "grad_norm": 0.8414542425518661,
      "learning_rate": 4.259684964086195e-06,
      "loss": 0.3502,
      "step": 8954
    },
    {
      "epoch": 0.5609232841103056,
      "grad_norm": 0.8555485472104805,
      "learning_rate": 4.25868176144079e-06,
      "loss": 0.3787,
      "step": 8955
    },
    {
      "epoch": 0.5609859221096478,
      "grad_norm": 0.8837702150128155,
      "learning_rate": 4.257678589308183e-06,
      "loss": 0.3966,
      "step": 8956
    },
    {
      "epoch": 0.5610485601089901,
      "grad_norm": 0.8625706178132027,
      "learning_rate": 4.256675447729662e-06,
      "loss": 0.4097,
      "step": 8957
    },
    {
      "epoch": 0.5611111981083324,
      "grad_norm": 0.8222954908086152,
      "learning_rate": 4.255672336746518e-06,
      "loss": 0.391,
      "step": 8958
    },
    {
      "epoch": 0.5611738361076747,
      "grad_norm": 0.8254212553859952,
      "learning_rate": 4.254669256400039e-06,
      "loss": 0.4032,
      "step": 8959
    },
    {
      "epoch": 0.561236474107017,
      "grad_norm": 0.7719832568013405,
      "learning_rate": 4.253666206731511e-06,
      "loss": 0.3982,
      "step": 8960
    },
    {
      "epoch": 0.5612991121063593,
      "grad_norm": 0.8306324234249766,
      "learning_rate": 4.252663187782221e-06,
      "loss": 0.4145,
      "step": 8961
    },
    {
      "epoch": 0.5613617501057017,
      "grad_norm": 0.8021509220655104,
      "learning_rate": 4.251660199593453e-06,
      "loss": 0.4272,
      "step": 8962
    },
    {
      "epoch": 0.5614243881050439,
      "grad_norm": 0.6660566686626523,
      "learning_rate": 4.250657242206488e-06,
      "loss": 0.4592,
      "step": 8963
    },
    {
      "epoch": 0.5614870261043863,
      "grad_norm": 0.8240112126934138,
      "learning_rate": 4.2496543156626116e-06,
      "loss": 0.3915,
      "step": 8964
    },
    {
      "epoch": 0.5615496641037285,
      "grad_norm": 0.7484139308207889,
      "learning_rate": 4.2486514200031036e-06,
      "loss": 0.3551,
      "step": 8965
    },
    {
      "epoch": 0.5616123021030708,
      "grad_norm": 0.778466893847344,
      "learning_rate": 4.247648555269242e-06,
      "loss": 0.3711,
      "step": 8966
    },
    {
      "epoch": 0.5616749401024131,
      "grad_norm": 0.8997811158176471,
      "learning_rate": 4.246645721502305e-06,
      "loss": 0.4609,
      "step": 8967
    },
    {
      "epoch": 0.5617375781017554,
      "grad_norm": 0.7933645603526323,
      "learning_rate": 4.245642918743571e-06,
      "loss": 0.3768,
      "step": 8968
    },
    {
      "epoch": 0.5618002161010978,
      "grad_norm": 0.8322255082564521,
      "learning_rate": 4.244640147034314e-06,
      "loss": 0.3551,
      "step": 8969
    },
    {
      "epoch": 0.56186285410044,
      "grad_norm": 0.9180942519647391,
      "learning_rate": 4.243637406415809e-06,
      "loss": 0.3938,
      "step": 8970
    },
    {
      "epoch": 0.5619254920997824,
      "grad_norm": 0.8221396703547958,
      "learning_rate": 4.2426346969293285e-06,
      "loss": 0.4209,
      "step": 8971
    },
    {
      "epoch": 0.5619881300991246,
      "grad_norm": 0.832977165663261,
      "learning_rate": 4.241632018616142e-06,
      "loss": 0.3818,
      "step": 8972
    },
    {
      "epoch": 0.5620507680984669,
      "grad_norm": 0.8333768814030588,
      "learning_rate": 4.240629371517525e-06,
      "loss": 0.4273,
      "step": 8973
    },
    {
      "epoch": 0.5621134060978092,
      "grad_norm": 0.8886107956841365,
      "learning_rate": 4.2396267556747445e-06,
      "loss": 0.4073,
      "step": 8974
    },
    {
      "epoch": 0.5621760440971515,
      "grad_norm": 0.8950602280476848,
      "learning_rate": 4.2386241711290676e-06,
      "loss": 0.4647,
      "step": 8975
    },
    {
      "epoch": 0.5622386820964939,
      "grad_norm": 0.8607882373135369,
      "learning_rate": 4.23762161792176e-06,
      "loss": 0.3915,
      "step": 8976
    },
    {
      "epoch": 0.5623013200958361,
      "grad_norm": 0.8772104684856662,
      "learning_rate": 4.236619096094089e-06,
      "loss": 0.3539,
      "step": 8977
    },
    {
      "epoch": 0.5623639580951785,
      "grad_norm": 0.8806165296870622,
      "learning_rate": 4.235616605687318e-06,
      "loss": 0.4085,
      "step": 8978
    },
    {
      "epoch": 0.5624265960945207,
      "grad_norm": 0.8075084667479057,
      "learning_rate": 4.234614146742709e-06,
      "loss": 0.3936,
      "step": 8979
    },
    {
      "epoch": 0.5624892340938631,
      "grad_norm": 0.8771080909165959,
      "learning_rate": 4.233611719301523e-06,
      "loss": 0.4192,
      "step": 8980
    },
    {
      "epoch": 0.5625518720932053,
      "grad_norm": 0.6660268974239746,
      "learning_rate": 4.232609323405022e-06,
      "loss": 0.4463,
      "step": 8981
    },
    {
      "epoch": 0.5626145100925476,
      "grad_norm": 0.877757904616518,
      "learning_rate": 4.231606959094462e-06,
      "loss": 0.4488,
      "step": 8982
    },
    {
      "epoch": 0.56267714809189,
      "grad_norm": 0.8662825842123826,
      "learning_rate": 4.2306046264111035e-06,
      "loss": 0.3776,
      "step": 8983
    },
    {
      "epoch": 0.5627397860912322,
      "grad_norm": 0.850716440429454,
      "learning_rate": 4.229602325396201e-06,
      "loss": 0.3676,
      "step": 8984
    },
    {
      "epoch": 0.5628024240905746,
      "grad_norm": 0.851763877062701,
      "learning_rate": 4.22860005609101e-06,
      "loss": 0.4445,
      "step": 8985
    },
    {
      "epoch": 0.5628650620899168,
      "grad_norm": 0.8247387772461785,
      "learning_rate": 4.227597818536784e-06,
      "loss": 0.3949,
      "step": 8986
    },
    {
      "epoch": 0.5629277000892592,
      "grad_norm": 0.800712448025856,
      "learning_rate": 4.226595612774775e-06,
      "loss": 0.3963,
      "step": 8987
    },
    {
      "epoch": 0.5629903380886014,
      "grad_norm": 0.8677203342342891,
      "learning_rate": 4.225593438846234e-06,
      "loss": 0.3836,
      "step": 8988
    },
    {
      "epoch": 0.5630529760879438,
      "grad_norm": 0.841857837905872,
      "learning_rate": 4.224591296792411e-06,
      "loss": 0.3848,
      "step": 8989
    },
    {
      "epoch": 0.563115614087286,
      "grad_norm": 0.6001350412563271,
      "learning_rate": 4.223589186654555e-06,
      "loss": 0.444,
      "step": 8990
    },
    {
      "epoch": 0.5631782520866283,
      "grad_norm": 0.6469290450157653,
      "learning_rate": 4.222587108473909e-06,
      "loss": 0.4546,
      "step": 8991
    },
    {
      "epoch": 0.5632408900859707,
      "grad_norm": 0.821100092864716,
      "learning_rate": 4.2215850622917255e-06,
      "loss": 0.3858,
      "step": 8992
    },
    {
      "epoch": 0.5633035280853129,
      "grad_norm": 0.828132127755665,
      "learning_rate": 4.220583048149243e-06,
      "loss": 0.4099,
      "step": 8993
    },
    {
      "epoch": 0.5633661660846553,
      "grad_norm": 0.817283646307675,
      "learning_rate": 4.2195810660877085e-06,
      "loss": 0.3707,
      "step": 8994
    },
    {
      "epoch": 0.5634288040839975,
      "grad_norm": 0.6799324986259885,
      "learning_rate": 4.218579116148362e-06,
      "loss": 0.4599,
      "step": 8995
    },
    {
      "epoch": 0.5634914420833399,
      "grad_norm": 0.898000152829016,
      "learning_rate": 4.217577198372444e-06,
      "loss": 0.4441,
      "step": 8996
    },
    {
      "epoch": 0.5635540800826822,
      "grad_norm": 0.7956594657001702,
      "learning_rate": 4.216575312801193e-06,
      "loss": 0.3727,
      "step": 8997
    },
    {
      "epoch": 0.5636167180820245,
      "grad_norm": 0.5909556095479785,
      "learning_rate": 4.215573459475848e-06,
      "loss": 0.4384,
      "step": 8998
    },
    {
      "epoch": 0.5636793560813668,
      "grad_norm": 0.8292295452850293,
      "learning_rate": 4.214571638437646e-06,
      "loss": 0.3551,
      "step": 8999
    },
    {
      "epoch": 0.563741994080709,
      "grad_norm": 0.7967709276382278,
      "learning_rate": 4.213569849727818e-06,
      "loss": 0.3688,
      "step": 9000
    },
    {
      "epoch": 0.5638046320800514,
      "grad_norm": 0.8482496193766773,
      "learning_rate": 4.2125680933876034e-06,
      "loss": 0.3682,
      "step": 9001
    },
    {
      "epoch": 0.5638672700793936,
      "grad_norm": 0.7377651788576776,
      "learning_rate": 4.211566369458231e-06,
      "loss": 0.3694,
      "step": 9002
    },
    {
      "epoch": 0.563929908078736,
      "grad_norm": 0.8906190227863341,
      "learning_rate": 4.210564677980935e-06,
      "loss": 0.4394,
      "step": 9003
    },
    {
      "epoch": 0.5639925460780782,
      "grad_norm": 0.8775181552274232,
      "learning_rate": 4.209563018996943e-06,
      "loss": 0.4078,
      "step": 9004
    },
    {
      "epoch": 0.5640551840774206,
      "grad_norm": 0.8543795693568778,
      "learning_rate": 4.208561392547483e-06,
      "loss": 0.3709,
      "step": 9005
    },
    {
      "epoch": 0.5641178220767629,
      "grad_norm": 0.7725985391771746,
      "learning_rate": 4.207559798673784e-06,
      "loss": 0.4106,
      "step": 9006
    },
    {
      "epoch": 0.5641804600761051,
      "grad_norm": 0.8558869271120437,
      "learning_rate": 4.20655823741707e-06,
      "loss": 0.3926,
      "step": 9007
    },
    {
      "epoch": 0.5642430980754475,
      "grad_norm": 0.8076914515337426,
      "learning_rate": 4.205556708818567e-06,
      "loss": 0.3934,
      "step": 9008
    },
    {
      "epoch": 0.5643057360747897,
      "grad_norm": 0.8159237182253515,
      "learning_rate": 4.204555212919494e-06,
      "loss": 0.4248,
      "step": 9009
    },
    {
      "epoch": 0.5643683740741321,
      "grad_norm": 0.856631519307626,
      "learning_rate": 4.203553749761079e-06,
      "loss": 0.3734,
      "step": 9010
    },
    {
      "epoch": 0.5644310120734743,
      "grad_norm": 0.7821386615030638,
      "learning_rate": 4.202552319384539e-06,
      "loss": 0.4017,
      "step": 9011
    },
    {
      "epoch": 0.5644936500728167,
      "grad_norm": 0.854476566645593,
      "learning_rate": 4.201550921831094e-06,
      "loss": 0.384,
      "step": 9012
    },
    {
      "epoch": 0.564556288072159,
      "grad_norm": 0.7655070762257632,
      "learning_rate": 4.200549557141962e-06,
      "loss": 0.3777,
      "step": 9013
    },
    {
      "epoch": 0.5646189260715013,
      "grad_norm": 0.7721076374414685,
      "learning_rate": 4.199548225358357e-06,
      "loss": 0.3771,
      "step": 9014
    },
    {
      "epoch": 0.5646815640708436,
      "grad_norm": 0.8094753013057944,
      "learning_rate": 4.198546926521496e-06,
      "loss": 0.4074,
      "step": 9015
    },
    {
      "epoch": 0.5647442020701858,
      "grad_norm": 0.7940037027024056,
      "learning_rate": 4.197545660672593e-06,
      "loss": 0.3591,
      "step": 9016
    },
    {
      "epoch": 0.5648068400695282,
      "grad_norm": 0.8109560740933962,
      "learning_rate": 4.1965444278528596e-06,
      "loss": 0.4219,
      "step": 9017
    },
    {
      "epoch": 0.5648694780688704,
      "grad_norm": 0.8160078696197399,
      "learning_rate": 4.1955432281035045e-06,
      "loss": 0.4022,
      "step": 9018
    },
    {
      "epoch": 0.5649321160682128,
      "grad_norm": 0.7826220139099365,
      "learning_rate": 4.1945420614657405e-06,
      "loss": 0.3541,
      "step": 9019
    },
    {
      "epoch": 0.5649947540675551,
      "grad_norm": 0.8975085830502799,
      "learning_rate": 4.193540927980776e-06,
      "loss": 0.3481,
      "step": 9020
    },
    {
      "epoch": 0.5650573920668974,
      "grad_norm": 0.8975207007780642,
      "learning_rate": 4.192539827689816e-06,
      "loss": 0.4366,
      "step": 9021
    },
    {
      "epoch": 0.5651200300662397,
      "grad_norm": 0.7731238195339272,
      "learning_rate": 4.1915387606340664e-06,
      "loss": 0.3587,
      "step": 9022
    },
    {
      "epoch": 0.565182668065582,
      "grad_norm": 0.7835964017016582,
      "learning_rate": 4.190537726854731e-06,
      "loss": 0.3828,
      "step": 9023
    },
    {
      "epoch": 0.5652453060649243,
      "grad_norm": 0.7911796777236286,
      "learning_rate": 4.189536726393014e-06,
      "loss": 0.3733,
      "step": 9024
    },
    {
      "epoch": 0.5653079440642665,
      "grad_norm": 0.8859474615576641,
      "learning_rate": 4.188535759290116e-06,
      "loss": 0.3768,
      "step": 9025
    },
    {
      "epoch": 0.5653705820636089,
      "grad_norm": 0.6337596044941708,
      "learning_rate": 4.187534825587236e-06,
      "loss": 0.4516,
      "step": 9026
    },
    {
      "epoch": 0.5654332200629512,
      "grad_norm": 0.8181425502357973,
      "learning_rate": 4.186533925325573e-06,
      "loss": 0.3875,
      "step": 9027
    },
    {
      "epoch": 0.5654958580622935,
      "grad_norm": 0.8649233531374786,
      "learning_rate": 4.185533058546324e-06,
      "loss": 0.425,
      "step": 9028
    },
    {
      "epoch": 0.5655584960616358,
      "grad_norm": 0.866278382519741,
      "learning_rate": 4.184532225290687e-06,
      "loss": 0.4014,
      "step": 9029
    },
    {
      "epoch": 0.5656211340609781,
      "grad_norm": 0.6514685518883522,
      "learning_rate": 4.183531425599855e-06,
      "loss": 0.4545,
      "step": 9030
    },
    {
      "epoch": 0.5656837720603204,
      "grad_norm": 0.853948094248871,
      "learning_rate": 4.1825306595150195e-06,
      "loss": 0.4195,
      "step": 9031
    },
    {
      "epoch": 0.5657464100596626,
      "grad_norm": 0.7213422080767339,
      "learning_rate": 4.181529927077376e-06,
      "loss": 0.4705,
      "step": 9032
    },
    {
      "epoch": 0.565809048059005,
      "grad_norm": 0.880970898417182,
      "learning_rate": 4.1805292283281115e-06,
      "loss": 0.3881,
      "step": 9033
    },
    {
      "epoch": 0.5658716860583473,
      "grad_norm": 0.7685229676712284,
      "learning_rate": 4.179528563308416e-06,
      "loss": 0.3602,
      "step": 9034
    },
    {
      "epoch": 0.5659343240576896,
      "grad_norm": 0.8829839253910727,
      "learning_rate": 4.178527932059477e-06,
      "loss": 0.3786,
      "step": 9035
    },
    {
      "epoch": 0.5659969620570319,
      "grad_norm": 0.8326659347015134,
      "learning_rate": 4.177527334622481e-06,
      "loss": 0.4167,
      "step": 9036
    },
    {
      "epoch": 0.5660596000563742,
      "grad_norm": 0.8441953255129491,
      "learning_rate": 4.176526771038611e-06,
      "loss": 0.3838,
      "step": 9037
    },
    {
      "epoch": 0.5661222380557165,
      "grad_norm": 0.6124373263320664,
      "learning_rate": 4.175526241349054e-06,
      "loss": 0.4366,
      "step": 9038
    },
    {
      "epoch": 0.5661848760550588,
      "grad_norm": 0.8336924001943044,
      "learning_rate": 4.174525745594989e-06,
      "loss": 0.4292,
      "step": 9039
    },
    {
      "epoch": 0.5662475140544011,
      "grad_norm": 0.832521388875729,
      "learning_rate": 4.173525283817598e-06,
      "loss": 0.3882,
      "step": 9040
    },
    {
      "epoch": 0.5663101520537434,
      "grad_norm": 0.818232123544067,
      "learning_rate": 4.17252485605806e-06,
      "loss": 0.4007,
      "step": 9041
    },
    {
      "epoch": 0.5663727900530857,
      "grad_norm": 0.8706118572510553,
      "learning_rate": 4.171524462357553e-06,
      "loss": 0.4,
      "step": 9042
    },
    {
      "epoch": 0.566435428052428,
      "grad_norm": 0.8395603424772847,
      "learning_rate": 4.170524102757252e-06,
      "loss": 0.4259,
      "step": 9043
    },
    {
      "epoch": 0.5664980660517703,
      "grad_norm": 0.8894226498664206,
      "learning_rate": 4.1695237772983324e-06,
      "loss": 0.4091,
      "step": 9044
    },
    {
      "epoch": 0.5665607040511126,
      "grad_norm": 0.7729933431528359,
      "learning_rate": 4.168523486021969e-06,
      "loss": 0.3887,
      "step": 9045
    },
    {
      "epoch": 0.566623342050455,
      "grad_norm": 0.7735134998404896,
      "learning_rate": 4.167523228969331e-06,
      "loss": 0.4094,
      "step": 9046
    },
    {
      "epoch": 0.5666859800497972,
      "grad_norm": 0.8222126195372413,
      "learning_rate": 4.166523006181594e-06,
      "loss": 0.3815,
      "step": 9047
    },
    {
      "epoch": 0.5667486180491396,
      "grad_norm": 0.8349738730174896,
      "learning_rate": 4.165522817699925e-06,
      "loss": 0.397,
      "step": 9048
    },
    {
      "epoch": 0.5668112560484818,
      "grad_norm": 0.7237469638758185,
      "learning_rate": 4.1645226635654905e-06,
      "loss": 0.4951,
      "step": 9049
    },
    {
      "epoch": 0.5668738940478241,
      "grad_norm": 0.8571360589361725,
      "learning_rate": 4.16352254381946e-06,
      "loss": 0.4312,
      "step": 9050
    },
    {
      "epoch": 0.5669365320471664,
      "grad_norm": 0.8519413415472317,
      "learning_rate": 4.162522458502995e-06,
      "loss": 0.3568,
      "step": 9051
    },
    {
      "epoch": 0.5669991700465087,
      "grad_norm": 0.8272698106484035,
      "learning_rate": 4.161522407657261e-06,
      "loss": 0.4036,
      "step": 9052
    },
    {
      "epoch": 0.567061808045851,
      "grad_norm": 0.8947769148412213,
      "learning_rate": 4.1605223913234216e-06,
      "loss": 0.3979,
      "step": 9053
    },
    {
      "epoch": 0.5671244460451933,
      "grad_norm": 0.8236390446586193,
      "learning_rate": 4.159522409542636e-06,
      "loss": 0.414,
      "step": 9054
    },
    {
      "epoch": 0.5671870840445357,
      "grad_norm": 0.7952101978079202,
      "learning_rate": 4.158522462356062e-06,
      "loss": 0.4193,
      "step": 9055
    },
    {
      "epoch": 0.5672497220438779,
      "grad_norm": 0.8397563493239961,
      "learning_rate": 4.157522549804862e-06,
      "loss": 0.3819,
      "step": 9056
    },
    {
      "epoch": 0.5673123600432202,
      "grad_norm": 0.7948840343630229,
      "learning_rate": 4.156522671930188e-06,
      "loss": 0.3972,
      "step": 9057
    },
    {
      "epoch": 0.5673749980425625,
      "grad_norm": 0.8460380660428998,
      "learning_rate": 4.155522828773199e-06,
      "loss": 0.3651,
      "step": 9058
    },
    {
      "epoch": 0.5674376360419048,
      "grad_norm": 0.8023750216264258,
      "learning_rate": 4.154523020375048e-06,
      "loss": 0.4116,
      "step": 9059
    },
    {
      "epoch": 0.5675002740412471,
      "grad_norm": 0.8437878151502904,
      "learning_rate": 4.153523246776887e-06,
      "loss": 0.3898,
      "step": 9060
    },
    {
      "epoch": 0.5675629120405894,
      "grad_norm": 0.8579137805658927,
      "learning_rate": 4.1525235080198635e-06,
      "loss": 0.4303,
      "step": 9061
    },
    {
      "epoch": 0.5676255500399318,
      "grad_norm": 0.7832550765003957,
      "learning_rate": 4.151523804145131e-06,
      "loss": 0.3948,
      "step": 9062
    },
    {
      "epoch": 0.567688188039274,
      "grad_norm": 0.7600768275962893,
      "learning_rate": 4.150524135193837e-06,
      "loss": 0.3423,
      "step": 9063
    },
    {
      "epoch": 0.5677508260386164,
      "grad_norm": 0.8549728652754164,
      "learning_rate": 4.149524501207125e-06,
      "loss": 0.4116,
      "step": 9064
    },
    {
      "epoch": 0.5678134640379586,
      "grad_norm": 0.8036085812873484,
      "learning_rate": 4.148524902226144e-06,
      "loss": 0.3679,
      "step": 9065
    },
    {
      "epoch": 0.5678761020373009,
      "grad_norm": 0.8663898434516495,
      "learning_rate": 4.147525338292036e-06,
      "loss": 0.3962,
      "step": 9066
    },
    {
      "epoch": 0.5679387400366432,
      "grad_norm": 0.8380988591348376,
      "learning_rate": 4.1465258094459425e-06,
      "loss": 0.4088,
      "step": 9067
    },
    {
      "epoch": 0.5680013780359855,
      "grad_norm": 0.8245446411571812,
      "learning_rate": 4.145526315729007e-06,
      "loss": 0.4009,
      "step": 9068
    },
    {
      "epoch": 0.5680640160353279,
      "grad_norm": 0.8100490276999892,
      "learning_rate": 4.144526857182366e-06,
      "loss": 0.3983,
      "step": 9069
    },
    {
      "epoch": 0.5681266540346701,
      "grad_norm": 0.8542611820269308,
      "learning_rate": 4.143527433847157e-06,
      "loss": 0.3883,
      "step": 9070
    },
    {
      "epoch": 0.5681892920340125,
      "grad_norm": 0.7983340179748324,
      "learning_rate": 4.14252804576452e-06,
      "loss": 0.3754,
      "step": 9071
    },
    {
      "epoch": 0.5682519300333547,
      "grad_norm": 0.8356768794490791,
      "learning_rate": 4.141528692975587e-06,
      "loss": 0.3775,
      "step": 9072
    },
    {
      "epoch": 0.5683145680326971,
      "grad_norm": 0.8528837294151288,
      "learning_rate": 4.140529375521491e-06,
      "loss": 0.4212,
      "step": 9073
    },
    {
      "epoch": 0.5683772060320393,
      "grad_norm": 0.8363384425608786,
      "learning_rate": 4.1395300934433656e-06,
      "loss": 0.3796,
      "step": 9074
    },
    {
      "epoch": 0.5684398440313816,
      "grad_norm": 0.7433462227702758,
      "learning_rate": 4.138530846782341e-06,
      "loss": 0.3768,
      "step": 9075
    },
    {
      "epoch": 0.568502482030724,
      "grad_norm": 0.8391322834916494,
      "learning_rate": 4.137531635579549e-06,
      "loss": 0.3861,
      "step": 9076
    },
    {
      "epoch": 0.5685651200300662,
      "grad_norm": 0.8418169835785805,
      "learning_rate": 4.1365324598761135e-06,
      "loss": 0.3715,
      "step": 9077
    },
    {
      "epoch": 0.5686277580294086,
      "grad_norm": 0.7707820178781668,
      "learning_rate": 4.135533319713162e-06,
      "loss": 0.4013,
      "step": 9078
    },
    {
      "epoch": 0.5686903960287508,
      "grad_norm": 0.7760127053359359,
      "learning_rate": 4.1345342151318205e-06,
      "loss": 0.3205,
      "step": 9079
    },
    {
      "epoch": 0.5687530340280932,
      "grad_norm": 0.8387316036597797,
      "learning_rate": 4.133535146173211e-06,
      "loss": 0.3727,
      "step": 9080
    },
    {
      "epoch": 0.5688156720274354,
      "grad_norm": 0.8686479074979002,
      "learning_rate": 4.132536112878456e-06,
      "loss": 0.4625,
      "step": 9081
    },
    {
      "epoch": 0.5688783100267777,
      "grad_norm": 0.842810310745741,
      "learning_rate": 4.131537115288675e-06,
      "loss": 0.3698,
      "step": 9082
    },
    {
      "epoch": 0.56894094802612,
      "grad_norm": 0.7978947723640736,
      "learning_rate": 4.130538153444985e-06,
      "loss": 0.39,
      "step": 9083
    },
    {
      "epoch": 0.5690035860254623,
      "grad_norm": 0.8500707292062027,
      "learning_rate": 4.1295392273885096e-06,
      "loss": 0.4232,
      "step": 9084
    },
    {
      "epoch": 0.5690662240248047,
      "grad_norm": 0.7850457175377245,
      "learning_rate": 4.128540337160361e-06,
      "loss": 0.4191,
      "step": 9085
    },
    {
      "epoch": 0.5691288620241469,
      "grad_norm": 0.6320252014628114,
      "learning_rate": 4.127541482801653e-06,
      "loss": 0.4737,
      "step": 9086
    },
    {
      "epoch": 0.5691915000234893,
      "grad_norm": 0.7969737250381412,
      "learning_rate": 4.126542664353499e-06,
      "loss": 0.3799,
      "step": 9087
    },
    {
      "epoch": 0.5692541380228315,
      "grad_norm": 0.8447488851934101,
      "learning_rate": 4.125543881857012e-06,
      "loss": 0.4339,
      "step": 9088
    },
    {
      "epoch": 0.5693167760221739,
      "grad_norm": 0.7612103290100898,
      "learning_rate": 4.124545135353301e-06,
      "loss": 0.3856,
      "step": 9089
    },
    {
      "epoch": 0.5693794140215162,
      "grad_norm": 0.7677657292072165,
      "learning_rate": 4.123546424883474e-06,
      "loss": 0.3313,
      "step": 9090
    },
    {
      "epoch": 0.5694420520208584,
      "grad_norm": 0.7915922756938965,
      "learning_rate": 4.1225477504886395e-06,
      "loss": 0.3884,
      "step": 9091
    },
    {
      "epoch": 0.5695046900202008,
      "grad_norm": 0.8477467874011984,
      "learning_rate": 4.1215491122099e-06,
      "loss": 0.3963,
      "step": 9092
    },
    {
      "epoch": 0.569567328019543,
      "grad_norm": 0.8909494090389064,
      "learning_rate": 4.120550510088365e-06,
      "loss": 0.3983,
      "step": 9093
    },
    {
      "epoch": 0.5696299660188854,
      "grad_norm": 0.7506584312509169,
      "learning_rate": 4.119551944165135e-06,
      "loss": 0.3362,
      "step": 9094
    },
    {
      "epoch": 0.5696926040182276,
      "grad_norm": 0.8117717410206902,
      "learning_rate": 4.118553414481309e-06,
      "loss": 0.3733,
      "step": 9095
    },
    {
      "epoch": 0.56975524201757,
      "grad_norm": 0.8334762792058173,
      "learning_rate": 4.117554921077987e-06,
      "loss": 0.3807,
      "step": 9096
    },
    {
      "epoch": 0.5698178800169122,
      "grad_norm": 0.8691593095065586,
      "learning_rate": 4.11655646399627e-06,
      "loss": 0.3957,
      "step": 9097
    },
    {
      "epoch": 0.5698805180162546,
      "grad_norm": 0.8003169486579127,
      "learning_rate": 4.115558043277252e-06,
      "loss": 0.411,
      "step": 9098
    },
    {
      "epoch": 0.5699431560155969,
      "grad_norm": 0.8566646408629202,
      "learning_rate": 4.114559658962028e-06,
      "loss": 0.4136,
      "step": 9099
    },
    {
      "epoch": 0.5700057940149391,
      "grad_norm": 0.8552949510609552,
      "learning_rate": 4.113561311091695e-06,
      "loss": 0.3689,
      "step": 9100
    },
    {
      "epoch": 0.5700684320142815,
      "grad_norm": 0.8127725579498769,
      "learning_rate": 4.11256299970734e-06,
      "loss": 0.3745,
      "step": 9101
    },
    {
      "epoch": 0.5701310700136237,
      "grad_norm": 0.8694494527447885,
      "learning_rate": 4.1115647248500595e-06,
      "loss": 0.3912,
      "step": 9102
    },
    {
      "epoch": 0.5701937080129661,
      "grad_norm": 0.8432338307499968,
      "learning_rate": 4.11056648656094e-06,
      "loss": 0.3574,
      "step": 9103
    },
    {
      "epoch": 0.5702563460123083,
      "grad_norm": 0.8625076836253832,
      "learning_rate": 4.109568284881067e-06,
      "loss": 0.4126,
      "step": 9104
    },
    {
      "epoch": 0.5703189840116507,
      "grad_norm": 0.6011501683715904,
      "learning_rate": 4.1085701198515295e-06,
      "loss": 0.4702,
      "step": 9105
    },
    {
      "epoch": 0.570381622010993,
      "grad_norm": 0.9075888175278651,
      "learning_rate": 4.107571991513413e-06,
      "loss": 0.4207,
      "step": 9106
    },
    {
      "epoch": 0.5704442600103353,
      "grad_norm": 0.7703548262389723,
      "learning_rate": 4.1065738999077975e-06,
      "loss": 0.372,
      "step": 9107
    },
    {
      "epoch": 0.5705068980096776,
      "grad_norm": 0.8333869238160954,
      "learning_rate": 4.105575845075765e-06,
      "loss": 0.3927,
      "step": 9108
    },
    {
      "epoch": 0.5705695360090198,
      "grad_norm": 0.8489385747619631,
      "learning_rate": 4.1045778270583985e-06,
      "loss": 0.4103,
      "step": 9109
    },
    {
      "epoch": 0.5706321740083622,
      "grad_norm": 0.850541482650302,
      "learning_rate": 4.103579845896772e-06,
      "loss": 0.4389,
      "step": 9110
    },
    {
      "epoch": 0.5706948120077044,
      "grad_norm": 0.7811265950549899,
      "learning_rate": 4.1025819016319685e-06,
      "loss": 0.3325,
      "step": 9111
    },
    {
      "epoch": 0.5707574500070468,
      "grad_norm": 0.8899452736010368,
      "learning_rate": 4.10158399430506e-06,
      "loss": 0.3838,
      "step": 9112
    },
    {
      "epoch": 0.5708200880063891,
      "grad_norm": 0.8619038703622535,
      "learning_rate": 4.10058612395712e-06,
      "loss": 0.4105,
      "step": 9113
    },
    {
      "epoch": 0.5708827260057314,
      "grad_norm": 0.8849689200318144,
      "learning_rate": 4.099588290629223e-06,
      "loss": 0.3847,
      "step": 9114
    },
    {
      "epoch": 0.5709453640050737,
      "grad_norm": 0.8598218951672996,
      "learning_rate": 4.098590494362439e-06,
      "loss": 0.4169,
      "step": 9115
    },
    {
      "epoch": 0.5710080020044159,
      "grad_norm": 0.8036572851030126,
      "learning_rate": 4.097592735197836e-06,
      "loss": 0.4084,
      "step": 9116
    },
    {
      "epoch": 0.5710706400037583,
      "grad_norm": 0.7708364195021155,
      "learning_rate": 4.0965950131764856e-06,
      "loss": 0.3732,
      "step": 9117
    },
    {
      "epoch": 0.5711332780031005,
      "grad_norm": 0.8066837253536556,
      "learning_rate": 4.0955973283394525e-06,
      "loss": 0.4102,
      "step": 9118
    },
    {
      "epoch": 0.5711959160024429,
      "grad_norm": 0.8612498546816416,
      "learning_rate": 4.0945996807278e-06,
      "loss": 0.3965,
      "step": 9119
    },
    {
      "epoch": 0.5712585540017852,
      "grad_norm": 0.8166241814633945,
      "learning_rate": 4.093602070382591e-06,
      "loss": 0.4477,
      "step": 9120
    },
    {
      "epoch": 0.5713211920011275,
      "grad_norm": 0.8452090557351012,
      "learning_rate": 4.09260449734489e-06,
      "loss": 0.4117,
      "step": 9121
    },
    {
      "epoch": 0.5713838300004698,
      "grad_norm": 0.5964505608076732,
      "learning_rate": 4.091606961655758e-06,
      "loss": 0.4421,
      "step": 9122
    },
    {
      "epoch": 0.5714464679998121,
      "grad_norm": 0.8448504174931037,
      "learning_rate": 4.0906094633562514e-06,
      "loss": 0.4041,
      "step": 9123
    },
    {
      "epoch": 0.5715091059991544,
      "grad_norm": 0.6834294409940257,
      "learning_rate": 4.089612002487428e-06,
      "loss": 0.4484,
      "step": 9124
    },
    {
      "epoch": 0.5715717439984966,
      "grad_norm": 0.86043878808204,
      "learning_rate": 4.0886145790903434e-06,
      "loss": 0.3897,
      "step": 9125
    },
    {
      "epoch": 0.571634381997839,
      "grad_norm": 0.8723999140072135,
      "learning_rate": 4.087617193206052e-06,
      "loss": 0.396,
      "step": 9126
    },
    {
      "epoch": 0.5716970199971813,
      "grad_norm": 0.8757522407016729,
      "learning_rate": 4.086619844875606e-06,
      "loss": 0.4177,
      "step": 9127
    },
    {
      "epoch": 0.5717596579965236,
      "grad_norm": 0.7974246579156059,
      "learning_rate": 4.085622534140058e-06,
      "loss": 0.3682,
      "step": 9128
    },
    {
      "epoch": 0.5718222959958659,
      "grad_norm": 0.795497902845987,
      "learning_rate": 4.084625261040453e-06,
      "loss": 0.3971,
      "step": 9129
    },
    {
      "epoch": 0.5718849339952082,
      "grad_norm": 0.9050040920148305,
      "learning_rate": 4.083628025617844e-06,
      "loss": 0.389,
      "step": 9130
    },
    {
      "epoch": 0.5719475719945505,
      "grad_norm": 0.7773381721287426,
      "learning_rate": 4.082630827913277e-06,
      "loss": 0.3326,
      "step": 9131
    },
    {
      "epoch": 0.5720102099938928,
      "grad_norm": 0.7965304655290741,
      "learning_rate": 4.081633667967794e-06,
      "loss": 0.4,
      "step": 9132
    },
    {
      "epoch": 0.5720728479932351,
      "grad_norm": 0.8316061627240057,
      "learning_rate": 4.0806365458224405e-06,
      "loss": 0.4034,
      "step": 9133
    },
    {
      "epoch": 0.5721354859925774,
      "grad_norm": 0.9614309771006826,
      "learning_rate": 4.079639461518256e-06,
      "loss": 0.3859,
      "step": 9134
    },
    {
      "epoch": 0.5721981239919197,
      "grad_norm": 0.7476738969074131,
      "learning_rate": 4.0786424150962835e-06,
      "loss": 0.488,
      "step": 9135
    },
    {
      "epoch": 0.572260761991262,
      "grad_norm": 0.8113406132758085,
      "learning_rate": 4.07764540659756e-06,
      "loss": 0.4238,
      "step": 9136
    },
    {
      "epoch": 0.5723233999906043,
      "grad_norm": 0.8258008910533977,
      "learning_rate": 4.076648436063122e-06,
      "loss": 0.4049,
      "step": 9137
    },
    {
      "epoch": 0.5723860379899466,
      "grad_norm": 0.8118071566118852,
      "learning_rate": 4.075651503534005e-06,
      "loss": 0.3785,
      "step": 9138
    },
    {
      "epoch": 0.572448675989289,
      "grad_norm": 0.8485712404739856,
      "learning_rate": 4.074654609051244e-06,
      "loss": 0.362,
      "step": 9139
    },
    {
      "epoch": 0.5725113139886312,
      "grad_norm": 0.632136280150685,
      "learning_rate": 4.0736577526558726e-06,
      "loss": 0.4512,
      "step": 9140
    },
    {
      "epoch": 0.5725739519879735,
      "grad_norm": 0.8068292255794987,
      "learning_rate": 4.072660934388919e-06,
      "loss": 0.39,
      "step": 9141
    },
    {
      "epoch": 0.5726365899873158,
      "grad_norm": 0.7751875132791606,
      "learning_rate": 4.071664154291414e-06,
      "loss": 0.3906,
      "step": 9142
    },
    {
      "epoch": 0.5726992279866581,
      "grad_norm": 0.8109721128272229,
      "learning_rate": 4.070667412404384e-06,
      "loss": 0.4207,
      "step": 9143
    },
    {
      "epoch": 0.5727618659860004,
      "grad_norm": 0.858758950351563,
      "learning_rate": 4.069670708768857e-06,
      "loss": 0.3737,
      "step": 9144
    },
    {
      "epoch": 0.5728245039853427,
      "grad_norm": 0.8317566574833665,
      "learning_rate": 4.0686740434258565e-06,
      "loss": 0.417,
      "step": 9145
    },
    {
      "epoch": 0.572887141984685,
      "grad_norm": 0.7682195439278361,
      "learning_rate": 4.067677416416403e-06,
      "loss": 0.3365,
      "step": 9146
    },
    {
      "epoch": 0.5729497799840273,
      "grad_norm": 0.7684683145023714,
      "learning_rate": 4.0666808277815204e-06,
      "loss": 0.3616,
      "step": 9147
    },
    {
      "epoch": 0.5730124179833697,
      "grad_norm": 0.7825411226663413,
      "learning_rate": 4.0656842775622295e-06,
      "loss": 0.3607,
      "step": 9148
    },
    {
      "epoch": 0.5730750559827119,
      "grad_norm": 0.9153194023590385,
      "learning_rate": 4.064687765799547e-06,
      "loss": 0.4249,
      "step": 9149
    },
    {
      "epoch": 0.5731376939820542,
      "grad_norm": 0.8647722208545455,
      "learning_rate": 4.0636912925344896e-06,
      "loss": 0.3792,
      "step": 9150
    },
    {
      "epoch": 0.5732003319813965,
      "grad_norm": 0.9074863238402099,
      "learning_rate": 4.0626948578080715e-06,
      "loss": 0.3773,
      "step": 9151
    },
    {
      "epoch": 0.5732629699807388,
      "grad_norm": 0.8140366181148173,
      "learning_rate": 4.061698461661309e-06,
      "loss": 0.3744,
      "step": 9152
    },
    {
      "epoch": 0.5733256079800811,
      "grad_norm": 0.804121468648543,
      "learning_rate": 4.060702104135211e-06,
      "loss": 0.3654,
      "step": 9153
    },
    {
      "epoch": 0.5733882459794234,
      "grad_norm": 0.89817180940552,
      "learning_rate": 4.059705785270789e-06,
      "loss": 0.3828,
      "step": 9154
    },
    {
      "epoch": 0.5734508839787658,
      "grad_norm": 0.8238490122570967,
      "learning_rate": 4.058709505109049e-06,
      "loss": 0.3831,
      "step": 9155
    },
    {
      "epoch": 0.573513521978108,
      "grad_norm": 0.8420778198134564,
      "learning_rate": 4.0577132636910015e-06,
      "loss": 0.3884,
      "step": 9156
    },
    {
      "epoch": 0.5735761599774504,
      "grad_norm": 0.7665435335550538,
      "learning_rate": 4.056717061057652e-06,
      "loss": 0.3926,
      "step": 9157
    },
    {
      "epoch": 0.5736387979767926,
      "grad_norm": 0.8242961659018726,
      "learning_rate": 4.055720897250004e-06,
      "loss": 0.3542,
      "step": 9158
    },
    {
      "epoch": 0.5737014359761349,
      "grad_norm": 0.7517851855879413,
      "learning_rate": 4.054724772309059e-06,
      "loss": 0.3145,
      "step": 9159
    },
    {
      "epoch": 0.5737640739754772,
      "grad_norm": 0.8664357639243848,
      "learning_rate": 4.053728686275816e-06,
      "loss": 0.4256,
      "step": 9160
    },
    {
      "epoch": 0.5738267119748195,
      "grad_norm": 0.7850426801057693,
      "learning_rate": 4.0527326391912775e-06,
      "loss": 0.3678,
      "step": 9161
    },
    {
      "epoch": 0.5738893499741619,
      "grad_norm": 0.8109427116579194,
      "learning_rate": 4.05173663109644e-06,
      "loss": 0.3965,
      "step": 9162
    },
    {
      "epoch": 0.5739519879735041,
      "grad_norm": 0.800967103439767,
      "learning_rate": 4.050740662032296e-06,
      "loss": 0.392,
      "step": 9163
    },
    {
      "epoch": 0.5740146259728465,
      "grad_norm": 0.6681850145486236,
      "learning_rate": 4.049744732039846e-06,
      "loss": 0.4728,
      "step": 9164
    },
    {
      "epoch": 0.5740772639721887,
      "grad_norm": 0.852330474679865,
      "learning_rate": 4.048748841160077e-06,
      "loss": 0.4609,
      "step": 9165
    },
    {
      "epoch": 0.574139901971531,
      "grad_norm": 0.9490688758138467,
      "learning_rate": 4.047752989433982e-06,
      "loss": 0.4314,
      "step": 9166
    },
    {
      "epoch": 0.5742025399708733,
      "grad_norm": 0.7970848990394845,
      "learning_rate": 4.046757176902551e-06,
      "loss": 0.3785,
      "step": 9167
    },
    {
      "epoch": 0.5742651779702156,
      "grad_norm": 0.8085289193499289,
      "learning_rate": 4.045761403606771e-06,
      "loss": 0.4298,
      "step": 9168
    },
    {
      "epoch": 0.574327815969558,
      "grad_norm": 0.7786448481327419,
      "learning_rate": 4.0447656695876305e-06,
      "loss": 0.3546,
      "step": 9169
    },
    {
      "epoch": 0.5743904539689002,
      "grad_norm": 0.8649307575884547,
      "learning_rate": 4.043769974886112e-06,
      "loss": 0.3775,
      "step": 9170
    },
    {
      "epoch": 0.5744530919682426,
      "grad_norm": 0.8419033788865952,
      "learning_rate": 4.042774319543198e-06,
      "loss": 0.3825,
      "step": 9171
    },
    {
      "epoch": 0.5745157299675848,
      "grad_norm": 0.8753520295033514,
      "learning_rate": 4.041778703599872e-06,
      "loss": 0.4281,
      "step": 9172
    },
    {
      "epoch": 0.5745783679669272,
      "grad_norm": 0.8176706830461352,
      "learning_rate": 4.040783127097111e-06,
      "loss": 0.3752,
      "step": 9173
    },
    {
      "epoch": 0.5746410059662694,
      "grad_norm": 0.9387307620558478,
      "learning_rate": 4.039787590075896e-06,
      "loss": 0.4506,
      "step": 9174
    },
    {
      "epoch": 0.5747036439656117,
      "grad_norm": 0.8747175693828758,
      "learning_rate": 4.038792092577199e-06,
      "loss": 0.4019,
      "step": 9175
    },
    {
      "epoch": 0.574766281964954,
      "grad_norm": 0.7941789306273972,
      "learning_rate": 4.037796634641999e-06,
      "loss": 0.358,
      "step": 9176
    },
    {
      "epoch": 0.5748289199642963,
      "grad_norm": 0.8614534299011919,
      "learning_rate": 4.036801216311269e-06,
      "loss": 0.4354,
      "step": 9177
    },
    {
      "epoch": 0.5748915579636387,
      "grad_norm": 0.8091169575999202,
      "learning_rate": 4.0358058376259805e-06,
      "loss": 0.3955,
      "step": 9178
    },
    {
      "epoch": 0.5749541959629809,
      "grad_norm": 0.7948647450137211,
      "learning_rate": 4.034810498627102e-06,
      "loss": 0.3785,
      "step": 9179
    },
    {
      "epoch": 0.5750168339623233,
      "grad_norm": 0.8459991456719577,
      "learning_rate": 4.033815199355603e-06,
      "loss": 0.4345,
      "step": 9180
    },
    {
      "epoch": 0.5750794719616655,
      "grad_norm": 0.6854488938440206,
      "learning_rate": 4.032819939852448e-06,
      "loss": 0.4523,
      "step": 9181
    },
    {
      "epoch": 0.5751421099610079,
      "grad_norm": 0.9043972998223937,
      "learning_rate": 4.031824720158606e-06,
      "loss": 0.4093,
      "step": 9182
    },
    {
      "epoch": 0.5752047479603501,
      "grad_norm": 0.7853719161491862,
      "learning_rate": 4.030829540315037e-06,
      "loss": 0.3681,
      "step": 9183
    },
    {
      "epoch": 0.5752673859596924,
      "grad_norm": 0.8140304312546152,
      "learning_rate": 4.029834400362702e-06,
      "loss": 0.3647,
      "step": 9184
    },
    {
      "epoch": 0.5753300239590348,
      "grad_norm": 0.8097900857884539,
      "learning_rate": 4.0288393003425665e-06,
      "loss": 0.3508,
      "step": 9185
    },
    {
      "epoch": 0.575392661958377,
      "grad_norm": 0.8177217170752902,
      "learning_rate": 4.027844240295583e-06,
      "loss": 0.3772,
      "step": 9186
    },
    {
      "epoch": 0.5754552999577194,
      "grad_norm": 0.8278724041765154,
      "learning_rate": 4.026849220262714e-06,
      "loss": 0.3662,
      "step": 9187
    },
    {
      "epoch": 0.5755179379570616,
      "grad_norm": 0.8110314069523847,
      "learning_rate": 4.02585424028491e-06,
      "loss": 0.4001,
      "step": 9188
    },
    {
      "epoch": 0.575580575956404,
      "grad_norm": 0.8029426833593647,
      "learning_rate": 4.024859300403126e-06,
      "loss": 0.388,
      "step": 9189
    },
    {
      "epoch": 0.5756432139557462,
      "grad_norm": 0.8503303905464353,
      "learning_rate": 4.023864400658315e-06,
      "loss": 0.3886,
      "step": 9190
    },
    {
      "epoch": 0.5757058519550885,
      "grad_norm": 0.8336715075515339,
      "learning_rate": 4.022869541091427e-06,
      "loss": 0.3904,
      "step": 9191
    },
    {
      "epoch": 0.5757684899544309,
      "grad_norm": 0.8134556661121898,
      "learning_rate": 4.021874721743409e-06,
      "loss": 0.3996,
      "step": 9192
    },
    {
      "epoch": 0.5758311279537731,
      "grad_norm": 0.8252010604876214,
      "learning_rate": 4.020879942655207e-06,
      "loss": 0.4122,
      "step": 9193
    },
    {
      "epoch": 0.5758937659531155,
      "grad_norm": 0.8165824677710315,
      "learning_rate": 4.019885203867771e-06,
      "loss": 0.3848,
      "step": 9194
    },
    {
      "epoch": 0.5759564039524577,
      "grad_norm": 0.7798561297551468,
      "learning_rate": 4.018890505422041e-06,
      "loss": 0.3136,
      "step": 9195
    },
    {
      "epoch": 0.5760190419518001,
      "grad_norm": 0.893284577755628,
      "learning_rate": 4.017895847358961e-06,
      "loss": 0.4025,
      "step": 9196
    },
    {
      "epoch": 0.5760816799511423,
      "grad_norm": 0.8475868783520006,
      "learning_rate": 4.0169012297194695e-06,
      "loss": 0.3325,
      "step": 9197
    },
    {
      "epoch": 0.5761443179504847,
      "grad_norm": 0.8287359863432083,
      "learning_rate": 4.015906652544506e-06,
      "loss": 0.3863,
      "step": 9198
    },
    {
      "epoch": 0.576206955949827,
      "grad_norm": 0.8660351053074302,
      "learning_rate": 4.014912115875007e-06,
      "loss": 0.4206,
      "step": 9199
    },
    {
      "epoch": 0.5762695939491692,
      "grad_norm": 0.7780169127365633,
      "learning_rate": 4.013917619751909e-06,
      "loss": 0.3649,
      "step": 9200
    },
    {
      "epoch": 0.5763322319485116,
      "grad_norm": 0.5912279374491998,
      "learning_rate": 4.012923164216143e-06,
      "loss": 0.4963,
      "step": 9201
    },
    {
      "epoch": 0.5763948699478538,
      "grad_norm": 0.8177633594714069,
      "learning_rate": 4.011928749308643e-06,
      "loss": 0.3691,
      "step": 9202
    },
    {
      "epoch": 0.5764575079471962,
      "grad_norm": 0.8928041790797343,
      "learning_rate": 4.01093437507034e-06,
      "loss": 0.3839,
      "step": 9203
    },
    {
      "epoch": 0.5765201459465384,
      "grad_norm": 0.657605542993357,
      "learning_rate": 4.0099400415421615e-06,
      "loss": 0.4584,
      "step": 9204
    },
    {
      "epoch": 0.5765827839458808,
      "grad_norm": 0.8673239501699737,
      "learning_rate": 4.008945748765036e-06,
      "loss": 0.3903,
      "step": 9205
    },
    {
      "epoch": 0.5766454219452231,
      "grad_norm": 0.7991635542310801,
      "learning_rate": 4.007951496779885e-06,
      "loss": 0.4071,
      "step": 9206
    },
    {
      "epoch": 0.5767080599445654,
      "grad_norm": 0.8411673907542685,
      "learning_rate": 4.006957285627637e-06,
      "loss": 0.4198,
      "step": 9207
    },
    {
      "epoch": 0.5767706979439077,
      "grad_norm": 0.8699973667754853,
      "learning_rate": 4.00596311534921e-06,
      "loss": 0.4603,
      "step": 9208
    },
    {
      "epoch": 0.5768333359432499,
      "grad_norm": 0.851217733290639,
      "learning_rate": 4.0049689859855265e-06,
      "loss": 0.3907,
      "step": 9209
    },
    {
      "epoch": 0.5768959739425923,
      "grad_norm": 0.8644449031769051,
      "learning_rate": 4.003974897577503e-06,
      "loss": 0.4057,
      "step": 9210
    },
    {
      "epoch": 0.5769586119419345,
      "grad_norm": 0.8211445414996701,
      "learning_rate": 4.0029808501660585e-06,
      "loss": 0.4077,
      "step": 9211
    },
    {
      "epoch": 0.5770212499412769,
      "grad_norm": 0.9162899407744532,
      "learning_rate": 4.001986843792105e-06,
      "loss": 0.4309,
      "step": 9212
    },
    {
      "epoch": 0.5770838879406192,
      "grad_norm": 0.7675235184205325,
      "learning_rate": 4.000992878496561e-06,
      "loss": 0.3949,
      "step": 9213
    },
    {
      "epoch": 0.5771465259399615,
      "grad_norm": 0.8529829192305607,
      "learning_rate": 3.999998954320335e-06,
      "loss": 0.3905,
      "step": 9214
    },
    {
      "epoch": 0.5772091639393038,
      "grad_norm": 0.8438605563346278,
      "learning_rate": 3.999005071304337e-06,
      "loss": 0.4113,
      "step": 9215
    },
    {
      "epoch": 0.5772718019386461,
      "grad_norm": 0.7478300066971886,
      "learning_rate": 3.998011229489477e-06,
      "loss": 0.3415,
      "step": 9216
    },
    {
      "epoch": 0.5773344399379884,
      "grad_norm": 0.8353708886966609,
      "learning_rate": 3.997017428916661e-06,
      "loss": 0.3796,
      "step": 9217
    },
    {
      "epoch": 0.5773970779373306,
      "grad_norm": 0.8898195221793773,
      "learning_rate": 3.996023669626793e-06,
      "loss": 0.4033,
      "step": 9218
    },
    {
      "epoch": 0.577459715936673,
      "grad_norm": 0.6206744381760402,
      "learning_rate": 3.995029951660777e-06,
      "loss": 0.4822,
      "step": 9219
    },
    {
      "epoch": 0.5775223539360153,
      "grad_norm": 0.8952898542299293,
      "learning_rate": 3.994036275059515e-06,
      "loss": 0.398,
      "step": 9220
    },
    {
      "epoch": 0.5775849919353576,
      "grad_norm": 0.786520142072733,
      "learning_rate": 3.993042639863904e-06,
      "loss": 0.3635,
      "step": 9221
    },
    {
      "epoch": 0.5776476299346999,
      "grad_norm": 0.8544735172837111,
      "learning_rate": 3.992049046114848e-06,
      "loss": 0.4366,
      "step": 9222
    },
    {
      "epoch": 0.5777102679340422,
      "grad_norm": 0.7997455442068655,
      "learning_rate": 3.99105549385324e-06,
      "loss": 0.385,
      "step": 9223
    },
    {
      "epoch": 0.5777729059333845,
      "grad_norm": 0.6339778106361418,
      "learning_rate": 3.9900619831199746e-06,
      "loss": 0.4576,
      "step": 9224
    },
    {
      "epoch": 0.5778355439327267,
      "grad_norm": 0.9316430841104023,
      "learning_rate": 3.989068513955946e-06,
      "loss": 0.401,
      "step": 9225
    },
    {
      "epoch": 0.5778981819320691,
      "grad_norm": 0.8239781497651938,
      "learning_rate": 3.9880750864020455e-06,
      "loss": 0.3768,
      "step": 9226
    },
    {
      "epoch": 0.5779608199314114,
      "grad_norm": 0.8453066420838193,
      "learning_rate": 3.98708170049916e-06,
      "loss": 0.3806,
      "step": 9227
    },
    {
      "epoch": 0.5780234579307537,
      "grad_norm": 0.8432989797099439,
      "learning_rate": 3.986088356288182e-06,
      "loss": 0.3921,
      "step": 9228
    },
    {
      "epoch": 0.578086095930096,
      "grad_norm": 0.8546790695498535,
      "learning_rate": 3.9850950538099955e-06,
      "loss": 0.4177,
      "step": 9229
    },
    {
      "epoch": 0.5781487339294383,
      "grad_norm": 0.7548129751501416,
      "learning_rate": 3.984101793105482e-06,
      "loss": 0.3557,
      "step": 9230
    },
    {
      "epoch": 0.5782113719287806,
      "grad_norm": 0.8128059314363433,
      "learning_rate": 3.983108574215529e-06,
      "loss": 0.3824,
      "step": 9231
    },
    {
      "epoch": 0.578274009928123,
      "grad_norm": 0.8330050370678508,
      "learning_rate": 3.982115397181016e-06,
      "loss": 0.3779,
      "step": 9232
    },
    {
      "epoch": 0.5783366479274652,
      "grad_norm": 0.9030929860579232,
      "learning_rate": 3.981122262042822e-06,
      "loss": 0.3937,
      "step": 9233
    },
    {
      "epoch": 0.5783992859268074,
      "grad_norm": 0.8798571439047934,
      "learning_rate": 3.980129168841826e-06,
      "loss": 0.4539,
      "step": 9234
    },
    {
      "epoch": 0.5784619239261498,
      "grad_norm": 0.8144838660040709,
      "learning_rate": 3.979136117618903e-06,
      "loss": 0.336,
      "step": 9235
    },
    {
      "epoch": 0.5785245619254921,
      "grad_norm": 0.9789005180445544,
      "learning_rate": 3.978143108414924e-06,
      "loss": 0.4184,
      "step": 9236
    },
    {
      "epoch": 0.5785871999248344,
      "grad_norm": 0.8770774414513642,
      "learning_rate": 3.977150141270767e-06,
      "loss": 0.3718,
      "step": 9237
    },
    {
      "epoch": 0.5786498379241767,
      "grad_norm": 0.7991531317722628,
      "learning_rate": 3.9761572162272996e-06,
      "loss": 0.3997,
      "step": 9238
    },
    {
      "epoch": 0.578712475923519,
      "grad_norm": 0.7674747504133311,
      "learning_rate": 3.975164333325388e-06,
      "loss": 0.3676,
      "step": 9239
    },
    {
      "epoch": 0.5787751139228613,
      "grad_norm": 0.92333420387007,
      "learning_rate": 3.974171492605907e-06,
      "loss": 0.3914,
      "step": 9240
    },
    {
      "epoch": 0.5788377519222037,
      "grad_norm": 0.8341282815611117,
      "learning_rate": 3.973178694109715e-06,
      "loss": 0.3898,
      "step": 9241
    },
    {
      "epoch": 0.5789003899215459,
      "grad_norm": 0.8214588907912201,
      "learning_rate": 3.97218593787768e-06,
      "loss": 0.3704,
      "step": 9242
    },
    {
      "epoch": 0.5789630279208882,
      "grad_norm": 0.8077983870110159,
      "learning_rate": 3.971193223950663e-06,
      "loss": 0.4068,
      "step": 9243
    },
    {
      "epoch": 0.5790256659202305,
      "grad_norm": 0.9098281741350787,
      "learning_rate": 3.970200552369523e-06,
      "loss": 0.4849,
      "step": 9244
    },
    {
      "epoch": 0.5790883039195728,
      "grad_norm": 0.8239003356089123,
      "learning_rate": 3.969207923175119e-06,
      "loss": 0.3714,
      "step": 9245
    },
    {
      "epoch": 0.5791509419189151,
      "grad_norm": 0.8674315676105785,
      "learning_rate": 3.968215336408309e-06,
      "loss": 0.4281,
      "step": 9246
    },
    {
      "epoch": 0.5792135799182574,
      "grad_norm": 0.8675790462703468,
      "learning_rate": 3.967222792109948e-06,
      "loss": 0.4027,
      "step": 9247
    },
    {
      "epoch": 0.5792762179175998,
      "grad_norm": 0.8260809178675771,
      "learning_rate": 3.966230290320886e-06,
      "loss": 0.3745,
      "step": 9248
    },
    {
      "epoch": 0.579338855916942,
      "grad_norm": 0.9272543112883376,
      "learning_rate": 3.965237831081979e-06,
      "loss": 0.3968,
      "step": 9249
    },
    {
      "epoch": 0.5794014939162843,
      "grad_norm": 0.8177814410122711,
      "learning_rate": 3.964245414434073e-06,
      "loss": 0.3953,
      "step": 9250
    },
    {
      "epoch": 0.5794641319156266,
      "grad_norm": 0.6259555474605812,
      "learning_rate": 3.963253040418021e-06,
      "loss": 0.4508,
      "step": 9251
    },
    {
      "epoch": 0.5795267699149689,
      "grad_norm": 0.7856692861431003,
      "learning_rate": 3.962260709074665e-06,
      "loss": 0.3793,
      "step": 9252
    },
    {
      "epoch": 0.5795894079143112,
      "grad_norm": 0.80354142506726,
      "learning_rate": 3.961268420444849e-06,
      "loss": 0.3722,
      "step": 9253
    },
    {
      "epoch": 0.5796520459136535,
      "grad_norm": 0.8238630929051528,
      "learning_rate": 3.96027617456942e-06,
      "loss": 0.3693,
      "step": 9254
    },
    {
      "epoch": 0.5797146839129959,
      "grad_norm": 0.5586446140127471,
      "learning_rate": 3.9592839714892165e-06,
      "loss": 0.4433,
      "step": 9255
    },
    {
      "epoch": 0.5797773219123381,
      "grad_norm": 0.8267607426883095,
      "learning_rate": 3.958291811245077e-06,
      "loss": 0.3916,
      "step": 9256
    },
    {
      "epoch": 0.5798399599116805,
      "grad_norm": 0.8298631232595794,
      "learning_rate": 3.95729969387784e-06,
      "loss": 0.408,
      "step": 9257
    },
    {
      "epoch": 0.5799025979110227,
      "grad_norm": 0.8604404822413042,
      "learning_rate": 3.95630761942834e-06,
      "loss": 0.3885,
      "step": 9258
    },
    {
      "epoch": 0.579965235910365,
      "grad_norm": 0.8203183744763,
      "learning_rate": 3.955315587937414e-06,
      "loss": 0.3786,
      "step": 9259
    },
    {
      "epoch": 0.5800278739097073,
      "grad_norm": 0.7745435895231673,
      "learning_rate": 3.954323599445892e-06,
      "loss": 0.3629,
      "step": 9260
    },
    {
      "epoch": 0.5800905119090496,
      "grad_norm": 0.8373591706865864,
      "learning_rate": 3.953331653994605e-06,
      "loss": 0.3976,
      "step": 9261
    },
    {
      "epoch": 0.580153149908392,
      "grad_norm": 0.8591924190489527,
      "learning_rate": 3.95233975162438e-06,
      "loss": 0.4227,
      "step": 9262
    },
    {
      "epoch": 0.5802157879077342,
      "grad_norm": 0.8160342275817565,
      "learning_rate": 3.9513478923760465e-06,
      "loss": 0.4177,
      "step": 9263
    },
    {
      "epoch": 0.5802784259070766,
      "grad_norm": 0.7596658056017551,
      "learning_rate": 3.950356076290429e-06,
      "loss": 0.3768,
      "step": 9264
    },
    {
      "epoch": 0.5803410639064188,
      "grad_norm": 0.8097210265492545,
      "learning_rate": 3.949364303408349e-06,
      "loss": 0.4038,
      "step": 9265
    },
    {
      "epoch": 0.5804037019057612,
      "grad_norm": 0.92545007526652,
      "learning_rate": 3.948372573770629e-06,
      "loss": 0.4339,
      "step": 9266
    },
    {
      "epoch": 0.5804663399051034,
      "grad_norm": 0.8784464696031571,
      "learning_rate": 3.947380887418088e-06,
      "loss": 0.437,
      "step": 9267
    },
    {
      "epoch": 0.5805289779044457,
      "grad_norm": 0.6265999998234161,
      "learning_rate": 3.946389244391547e-06,
      "loss": 0.4442,
      "step": 9268
    },
    {
      "epoch": 0.580591615903788,
      "grad_norm": 0.802599968007027,
      "learning_rate": 3.94539764473182e-06,
      "loss": 0.3622,
      "step": 9269
    },
    {
      "epoch": 0.5806542539031303,
      "grad_norm": 0.8918104314527737,
      "learning_rate": 3.944406088479722e-06,
      "loss": 0.3994,
      "step": 9270
    },
    {
      "epoch": 0.5807168919024727,
      "grad_norm": 0.8156285249974992,
      "learning_rate": 3.943414575676065e-06,
      "loss": 0.4177,
      "step": 9271
    },
    {
      "epoch": 0.5807795299018149,
      "grad_norm": 0.8677083331491727,
      "learning_rate": 3.9424231063616605e-06,
      "loss": 0.4132,
      "step": 9272
    },
    {
      "epoch": 0.5808421679011573,
      "grad_norm": 0.8282407707027899,
      "learning_rate": 3.941431680577317e-06,
      "loss": 0.4359,
      "step": 9273
    },
    {
      "epoch": 0.5809048059004995,
      "grad_norm": 0.778541881147215,
      "learning_rate": 3.940440298363841e-06,
      "loss": 0.3799,
      "step": 9274
    },
    {
      "epoch": 0.5809674438998418,
      "grad_norm": 0.777723888531033,
      "learning_rate": 3.93944895976204e-06,
      "loss": 0.357,
      "step": 9275
    },
    {
      "epoch": 0.5810300818991841,
      "grad_norm": 0.8201687387315164,
      "learning_rate": 3.938457664812716e-06,
      "loss": 0.4294,
      "step": 9276
    },
    {
      "epoch": 0.5810927198985264,
      "grad_norm": 0.8350961237257022,
      "learning_rate": 3.937466413556672e-06,
      "loss": 0.4114,
      "step": 9277
    },
    {
      "epoch": 0.5811553578978688,
      "grad_norm": 0.9037376881131411,
      "learning_rate": 3.9364752060347076e-06,
      "loss": 0.3883,
      "step": 9278
    },
    {
      "epoch": 0.581217995897211,
      "grad_norm": 0.8853764770222549,
      "learning_rate": 3.93548404228762e-06,
      "loss": 0.4249,
      "step": 9279
    },
    {
      "epoch": 0.5812806338965534,
      "grad_norm": 0.8554272757271584,
      "learning_rate": 3.934492922356209e-06,
      "loss": 0.3849,
      "step": 9280
    },
    {
      "epoch": 0.5813432718958956,
      "grad_norm": 0.8276789267551391,
      "learning_rate": 3.9335018462812664e-06,
      "loss": 0.3491,
      "step": 9281
    },
    {
      "epoch": 0.581405909895238,
      "grad_norm": 0.8975645003371701,
      "learning_rate": 3.932510814103586e-06,
      "loss": 0.41,
      "step": 9282
    },
    {
      "epoch": 0.5814685478945802,
      "grad_norm": 0.8530568035616807,
      "learning_rate": 3.931519825863957e-06,
      "loss": 0.4173,
      "step": 9283
    },
    {
      "epoch": 0.5815311858939225,
      "grad_norm": 0.8716744091834964,
      "learning_rate": 3.930528881603172e-06,
      "loss": 0.4471,
      "step": 9284
    },
    {
      "epoch": 0.5815938238932649,
      "grad_norm": 0.7762953082333861,
      "learning_rate": 3.929537981362015e-06,
      "loss": 0.365,
      "step": 9285
    },
    {
      "epoch": 0.5816564618926071,
      "grad_norm": 0.839493384468355,
      "learning_rate": 3.928547125181275e-06,
      "loss": 0.4274,
      "step": 9286
    },
    {
      "epoch": 0.5817190998919495,
      "grad_norm": 0.8058894974040506,
      "learning_rate": 3.9275563131017334e-06,
      "loss": 0.3846,
      "step": 9287
    },
    {
      "epoch": 0.5817817378912917,
      "grad_norm": 0.8009062455816796,
      "learning_rate": 3.9265655451641724e-06,
      "loss": 0.3404,
      "step": 9288
    },
    {
      "epoch": 0.5818443758906341,
      "grad_norm": 0.8552428125558591,
      "learning_rate": 3.925574821409375e-06,
      "loss": 0.3933,
      "step": 9289
    },
    {
      "epoch": 0.5819070138899763,
      "grad_norm": 0.8279765235464153,
      "learning_rate": 3.924584141878116e-06,
      "loss": 0.4252,
      "step": 9290
    },
    {
      "epoch": 0.5819696518893187,
      "grad_norm": 0.7773568752483974,
      "learning_rate": 3.923593506611173e-06,
      "loss": 0.378,
      "step": 9291
    },
    {
      "epoch": 0.582032289888661,
      "grad_norm": 0.8801324223611535,
      "learning_rate": 3.922602915649322e-06,
      "loss": 0.4037,
      "step": 9292
    },
    {
      "epoch": 0.5820949278880032,
      "grad_norm": 0.9235182981754939,
      "learning_rate": 3.921612369033334e-06,
      "loss": 0.4245,
      "step": 9293
    },
    {
      "epoch": 0.5821575658873456,
      "grad_norm": 0.8521363865322845,
      "learning_rate": 3.920621866803982e-06,
      "loss": 0.4154,
      "step": 9294
    },
    {
      "epoch": 0.5822202038866878,
      "grad_norm": 0.8611008783283619,
      "learning_rate": 3.9196314090020325e-06,
      "loss": 0.4138,
      "step": 9295
    },
    {
      "epoch": 0.5822828418860302,
      "grad_norm": 0.8789688638349739,
      "learning_rate": 3.918640995668256e-06,
      "loss": 0.4207,
      "step": 9296
    },
    {
      "epoch": 0.5823454798853724,
      "grad_norm": 0.6282608950144507,
      "learning_rate": 3.917650626843417e-06,
      "loss": 0.4694,
      "step": 9297
    },
    {
      "epoch": 0.5824081178847148,
      "grad_norm": 0.6816585784911534,
      "learning_rate": 3.916660302568279e-06,
      "loss": 0.4788,
      "step": 9298
    },
    {
      "epoch": 0.5824707558840571,
      "grad_norm": 0.8015324597002215,
      "learning_rate": 3.915670022883604e-06,
      "loss": 0.361,
      "step": 9299
    },
    {
      "epoch": 0.5825333938833994,
      "grad_norm": 0.8691927060512609,
      "learning_rate": 3.914679787830151e-06,
      "loss": 0.4013,
      "step": 9300
    },
    {
      "epoch": 0.5825960318827417,
      "grad_norm": 0.7914541580831037,
      "learning_rate": 3.913689597448681e-06,
      "loss": 0.3733,
      "step": 9301
    },
    {
      "epoch": 0.5826586698820839,
      "grad_norm": 0.818992061393021,
      "learning_rate": 3.912699451779948e-06,
      "loss": 0.3621,
      "step": 9302
    },
    {
      "epoch": 0.5827213078814263,
      "grad_norm": 0.9075483649747602,
      "learning_rate": 3.911709350864709e-06,
      "loss": 0.415,
      "step": 9303
    },
    {
      "epoch": 0.5827839458807685,
      "grad_norm": 0.867859144184351,
      "learning_rate": 3.910719294743712e-06,
      "loss": 0.4325,
      "step": 9304
    },
    {
      "epoch": 0.5828465838801109,
      "grad_norm": 0.8586174585317268,
      "learning_rate": 3.9097292834577125e-06,
      "loss": 0.3913,
      "step": 9305
    },
    {
      "epoch": 0.5829092218794532,
      "grad_norm": 0.805833348667351,
      "learning_rate": 3.908739317047458e-06,
      "loss": 0.4179,
      "step": 9306
    },
    {
      "epoch": 0.5829718598787955,
      "grad_norm": 0.8813991176067705,
      "learning_rate": 3.9077493955536966e-06,
      "loss": 0.398,
      "step": 9307
    },
    {
      "epoch": 0.5830344978781378,
      "grad_norm": 0.8448686073032198,
      "learning_rate": 3.9067595190171726e-06,
      "loss": 0.4151,
      "step": 9308
    },
    {
      "epoch": 0.58309713587748,
      "grad_norm": 0.7510801614900123,
      "learning_rate": 3.9057696874786285e-06,
      "loss": 0.3482,
      "step": 9309
    },
    {
      "epoch": 0.5831597738768224,
      "grad_norm": 0.9134795153402595,
      "learning_rate": 3.904779900978809e-06,
      "loss": 0.4569,
      "step": 9310
    },
    {
      "epoch": 0.5832224118761646,
      "grad_norm": 0.7845623296342462,
      "learning_rate": 3.9037901595584516e-06,
      "loss": 0.3518,
      "step": 9311
    },
    {
      "epoch": 0.583285049875507,
      "grad_norm": 0.8495250437174282,
      "learning_rate": 3.902800463258293e-06,
      "loss": 0.4039,
      "step": 9312
    },
    {
      "epoch": 0.5833476878748493,
      "grad_norm": 0.787846399161558,
      "learning_rate": 3.901810812119072e-06,
      "loss": 0.3753,
      "step": 9313
    },
    {
      "epoch": 0.5834103258741916,
      "grad_norm": 0.6538709845382875,
      "learning_rate": 3.900821206181521e-06,
      "loss": 0.4629,
      "step": 9314
    },
    {
      "epoch": 0.5834729638735339,
      "grad_norm": 0.8719786226726617,
      "learning_rate": 3.899831645486375e-06,
      "loss": 0.4252,
      "step": 9315
    },
    {
      "epoch": 0.5835356018728762,
      "grad_norm": 0.8894246290680687,
      "learning_rate": 3.898842130074362e-06,
      "loss": 0.4171,
      "step": 9316
    },
    {
      "epoch": 0.5835982398722185,
      "grad_norm": 0.8402913199931554,
      "learning_rate": 3.8978526599862105e-06,
      "loss": 0.3594,
      "step": 9317
    },
    {
      "epoch": 0.5836608778715607,
      "grad_norm": 0.8844876600616342,
      "learning_rate": 3.896863235262649e-06,
      "loss": 0.4341,
      "step": 9318
    },
    {
      "epoch": 0.5837235158709031,
      "grad_norm": 0.7419016657531234,
      "learning_rate": 3.895873855944402e-06,
      "loss": 0.3704,
      "step": 9319
    },
    {
      "epoch": 0.5837861538702454,
      "grad_norm": 0.9004846238439643,
      "learning_rate": 3.894884522072191e-06,
      "loss": 0.4255,
      "step": 9320
    },
    {
      "epoch": 0.5838487918695877,
      "grad_norm": 0.8684989580514672,
      "learning_rate": 3.893895233686738e-06,
      "loss": 0.3962,
      "step": 9321
    },
    {
      "epoch": 0.58391142986893,
      "grad_norm": 0.7791255525094425,
      "learning_rate": 3.892905990828762e-06,
      "loss": 0.3463,
      "step": 9322
    },
    {
      "epoch": 0.5839740678682723,
      "grad_norm": 0.8693009886339688,
      "learning_rate": 3.891916793538981e-06,
      "loss": 0.4023,
      "step": 9323
    },
    {
      "epoch": 0.5840367058676146,
      "grad_norm": 0.8408933278463772,
      "learning_rate": 3.890927641858112e-06,
      "loss": 0.3854,
      "step": 9324
    },
    {
      "epoch": 0.5840993438669569,
      "grad_norm": 0.8214476378487731,
      "learning_rate": 3.889938535826866e-06,
      "loss": 0.3979,
      "step": 9325
    },
    {
      "epoch": 0.5841619818662992,
      "grad_norm": 0.8620911139859612,
      "learning_rate": 3.8889494754859565e-06,
      "loss": 0.4213,
      "step": 9326
    },
    {
      "epoch": 0.5842246198656414,
      "grad_norm": 0.8825802488102054,
      "learning_rate": 3.887960460876093e-06,
      "loss": 0.4068,
      "step": 9327
    },
    {
      "epoch": 0.5842872578649838,
      "grad_norm": 0.8021179045535233,
      "learning_rate": 3.886971492037983e-06,
      "loss": 0.3736,
      "step": 9328
    },
    {
      "epoch": 0.5843498958643261,
      "grad_norm": 0.784759849622292,
      "learning_rate": 3.885982569012333e-06,
      "loss": 0.4105,
      "step": 9329
    },
    {
      "epoch": 0.5844125338636684,
      "grad_norm": 0.8259056792577519,
      "learning_rate": 3.884993691839847e-06,
      "loss": 0.3972,
      "step": 9330
    },
    {
      "epoch": 0.5844751718630107,
      "grad_norm": 0.7830608463860602,
      "learning_rate": 3.884004860561226e-06,
      "loss": 0.3604,
      "step": 9331
    },
    {
      "epoch": 0.584537809862353,
      "grad_norm": 0.7290004166956301,
      "learning_rate": 3.883016075217174e-06,
      "loss": 0.4608,
      "step": 9332
    },
    {
      "epoch": 0.5846004478616953,
      "grad_norm": 0.7531086859005183,
      "learning_rate": 3.882027335848388e-06,
      "loss": 0.3763,
      "step": 9333
    },
    {
      "epoch": 0.5846630858610375,
      "grad_norm": 0.8486345869770495,
      "learning_rate": 3.881038642495564e-06,
      "loss": 0.372,
      "step": 9334
    },
    {
      "epoch": 0.5847257238603799,
      "grad_norm": 0.845436544218654,
      "learning_rate": 3.880049995199396e-06,
      "loss": 0.4167,
      "step": 9335
    },
    {
      "epoch": 0.5847883618597222,
      "grad_norm": 0.8324889171093961,
      "learning_rate": 3.87906139400058e-06,
      "loss": 0.4048,
      "step": 9336
    },
    {
      "epoch": 0.5848509998590645,
      "grad_norm": 0.9372624365438007,
      "learning_rate": 3.878072838939805e-06,
      "loss": 0.4491,
      "step": 9337
    },
    {
      "epoch": 0.5849136378584068,
      "grad_norm": 0.8336248746849739,
      "learning_rate": 3.8770843300577586e-06,
      "loss": 0.3827,
      "step": 9338
    },
    {
      "epoch": 0.5849762758577491,
      "grad_norm": 0.8107530571196853,
      "learning_rate": 3.87609586739513e-06,
      "loss": 0.3933,
      "step": 9339
    },
    {
      "epoch": 0.5850389138570914,
      "grad_norm": 0.9297394636485009,
      "learning_rate": 3.875107450992605e-06,
      "loss": 0.3948,
      "step": 9340
    },
    {
      "epoch": 0.5851015518564338,
      "grad_norm": 0.8255109450370091,
      "learning_rate": 3.874119080890864e-06,
      "loss": 0.4229,
      "step": 9341
    },
    {
      "epoch": 0.585164189855776,
      "grad_norm": 0.6476857689668699,
      "learning_rate": 3.873130757130593e-06,
      "loss": 0.4636,
      "step": 9342
    },
    {
      "epoch": 0.5852268278551183,
      "grad_norm": 0.906715683396556,
      "learning_rate": 3.872142479752467e-06,
      "loss": 0.3915,
      "step": 9343
    },
    {
      "epoch": 0.5852894658544606,
      "grad_norm": 0.807713895444904,
      "learning_rate": 3.871154248797168e-06,
      "loss": 0.4197,
      "step": 9344
    },
    {
      "epoch": 0.5853521038538029,
      "grad_norm": 0.7569420926502558,
      "learning_rate": 3.870166064305369e-06,
      "loss": 0.385,
      "step": 9345
    },
    {
      "epoch": 0.5854147418531452,
      "grad_norm": 0.7878914944621,
      "learning_rate": 3.869177926317745e-06,
      "loss": 0.3751,
      "step": 9346
    },
    {
      "epoch": 0.5854773798524875,
      "grad_norm": 0.7262646916769039,
      "learning_rate": 3.868189834874965e-06,
      "loss": 0.3495,
      "step": 9347
    },
    {
      "epoch": 0.5855400178518299,
      "grad_norm": 0.8586260584178836,
      "learning_rate": 3.867201790017705e-06,
      "loss": 0.415,
      "step": 9348
    },
    {
      "epoch": 0.5856026558511721,
      "grad_norm": 0.8658001800215857,
      "learning_rate": 3.866213791786628e-06,
      "loss": 0.4182,
      "step": 9349
    },
    {
      "epoch": 0.5856652938505145,
      "grad_norm": 0.6438488097369989,
      "learning_rate": 3.8652258402224e-06,
      "loss": 0.4558,
      "step": 9350
    },
    {
      "epoch": 0.5857279318498567,
      "grad_norm": 0.8122270185601821,
      "learning_rate": 3.864237935365689e-06,
      "loss": 0.3788,
      "step": 9351
    },
    {
      "epoch": 0.585790569849199,
      "grad_norm": 0.8282963676854748,
      "learning_rate": 3.863250077257154e-06,
      "loss": 0.4012,
      "step": 9352
    },
    {
      "epoch": 0.5858532078485413,
      "grad_norm": 0.8018151977044892,
      "learning_rate": 3.862262265937459e-06,
      "loss": 0.4195,
      "step": 9353
    },
    {
      "epoch": 0.5859158458478836,
      "grad_norm": 0.8134350297005,
      "learning_rate": 3.861274501447259e-06,
      "loss": 0.3919,
      "step": 9354
    },
    {
      "epoch": 0.585978483847226,
      "grad_norm": 0.8015340198991988,
      "learning_rate": 3.8602867838272125e-06,
      "loss": 0.4006,
      "step": 9355
    },
    {
      "epoch": 0.5860411218465682,
      "grad_norm": 0.8063261016013364,
      "learning_rate": 3.859299113117972e-06,
      "loss": 0.3692,
      "step": 9356
    },
    {
      "epoch": 0.5861037598459106,
      "grad_norm": 0.8604890553832818,
      "learning_rate": 3.858311489360194e-06,
      "loss": 0.3806,
      "step": 9357
    },
    {
      "epoch": 0.5861663978452528,
      "grad_norm": 0.8276882461566658,
      "learning_rate": 3.857323912594526e-06,
      "loss": 0.42,
      "step": 9358
    },
    {
      "epoch": 0.5862290358445951,
      "grad_norm": 0.8875958914920968,
      "learning_rate": 3.856336382861615e-06,
      "loss": 0.4251,
      "step": 9359
    },
    {
      "epoch": 0.5862916738439374,
      "grad_norm": 0.8187553642230909,
      "learning_rate": 3.855348900202114e-06,
      "loss": 0.3945,
      "step": 9360
    },
    {
      "epoch": 0.5863543118432797,
      "grad_norm": 0.8392129085295704,
      "learning_rate": 3.854361464656664e-06,
      "loss": 0.4225,
      "step": 9361
    },
    {
      "epoch": 0.586416949842622,
      "grad_norm": 0.8542464415506619,
      "learning_rate": 3.853374076265909e-06,
      "loss": 0.44,
      "step": 9362
    },
    {
      "epoch": 0.5864795878419643,
      "grad_norm": 0.862512467438548,
      "learning_rate": 3.8523867350704905e-06,
      "loss": 0.4136,
      "step": 9363
    },
    {
      "epoch": 0.5865422258413067,
      "grad_norm": 0.8653755132308424,
      "learning_rate": 3.851399441111045e-06,
      "loss": 0.4185,
      "step": 9364
    },
    {
      "epoch": 0.5866048638406489,
      "grad_norm": 0.7698889389169881,
      "learning_rate": 3.850412194428214e-06,
      "loss": 0.377,
      "step": 9365
    },
    {
      "epoch": 0.5866675018399913,
      "grad_norm": 0.8326812059072358,
      "learning_rate": 3.84942499506263e-06,
      "loss": 0.4419,
      "step": 9366
    },
    {
      "epoch": 0.5867301398393335,
      "grad_norm": 0.8072115248736348,
      "learning_rate": 3.848437843054926e-06,
      "loss": 0.3882,
      "step": 9367
    },
    {
      "epoch": 0.5867927778386758,
      "grad_norm": 0.8380537409945225,
      "learning_rate": 3.847450738445732e-06,
      "loss": 0.4258,
      "step": 9368
    },
    {
      "epoch": 0.5868554158380181,
      "grad_norm": 0.8785212089093404,
      "learning_rate": 3.8464636812756815e-06,
      "loss": 0.4029,
      "step": 9369
    },
    {
      "epoch": 0.5869180538373604,
      "grad_norm": 0.8813111288095716,
      "learning_rate": 3.845476671585401e-06,
      "loss": 0.3884,
      "step": 9370
    },
    {
      "epoch": 0.5869806918367028,
      "grad_norm": 0.8711885279961111,
      "learning_rate": 3.844489709415515e-06,
      "loss": 0.4115,
      "step": 9371
    },
    {
      "epoch": 0.587043329836045,
      "grad_norm": 0.7809072747849007,
      "learning_rate": 3.843502794806646e-06,
      "loss": 0.3756,
      "step": 9372
    },
    {
      "epoch": 0.5871059678353874,
      "grad_norm": 0.8684762498169585,
      "learning_rate": 3.842515927799416e-06,
      "loss": 0.403,
      "step": 9373
    },
    {
      "epoch": 0.5871686058347296,
      "grad_norm": 0.6343339872513408,
      "learning_rate": 3.841529108434447e-06,
      "loss": 0.4399,
      "step": 9374
    },
    {
      "epoch": 0.587231243834072,
      "grad_norm": 0.6630056334176522,
      "learning_rate": 3.840542336752354e-06,
      "loss": 0.4582,
      "step": 9375
    },
    {
      "epoch": 0.5872938818334142,
      "grad_norm": 0.8451919223223922,
      "learning_rate": 3.839555612793754e-06,
      "loss": 0.4124,
      "step": 9376
    },
    {
      "epoch": 0.5873565198327565,
      "grad_norm": 0.85522463291276,
      "learning_rate": 3.838568936599259e-06,
      "loss": 0.3966,
      "step": 9377
    },
    {
      "epoch": 0.5874191578320989,
      "grad_norm": 0.8517464918918133,
      "learning_rate": 3.837582308209482e-06,
      "loss": 0.3822,
      "step": 9378
    },
    {
      "epoch": 0.5874817958314411,
      "grad_norm": 0.8286514661347697,
      "learning_rate": 3.8365957276650345e-06,
      "loss": 0.3894,
      "step": 9379
    },
    {
      "epoch": 0.5875444338307835,
      "grad_norm": 0.8702069601791402,
      "learning_rate": 3.835609195006524e-06,
      "loss": 0.4083,
      "step": 9380
    },
    {
      "epoch": 0.5876070718301257,
      "grad_norm": 0.8915064556995311,
      "learning_rate": 3.834622710274553e-06,
      "loss": 0.41,
      "step": 9381
    },
    {
      "epoch": 0.5876697098294681,
      "grad_norm": 0.8599247448170824,
      "learning_rate": 3.833636273509727e-06,
      "loss": 0.3705,
      "step": 9382
    },
    {
      "epoch": 0.5877323478288103,
      "grad_norm": 0.8991745462157105,
      "learning_rate": 3.83264988475265e-06,
      "loss": 0.4234,
      "step": 9383
    },
    {
      "epoch": 0.5877949858281526,
      "grad_norm": 0.8961705048164459,
      "learning_rate": 3.831663544043921e-06,
      "loss": 0.4227,
      "step": 9384
    },
    {
      "epoch": 0.587857623827495,
      "grad_norm": 0.8119049060895035,
      "learning_rate": 3.830677251424136e-06,
      "loss": 0.4317,
      "step": 9385
    },
    {
      "epoch": 0.5879202618268372,
      "grad_norm": 0.8527311111500927,
      "learning_rate": 3.829691006933893e-06,
      "loss": 0.4064,
      "step": 9386
    },
    {
      "epoch": 0.5879828998261796,
      "grad_norm": 0.8418570182052006,
      "learning_rate": 3.8287048106137835e-06,
      "loss": 0.3654,
      "step": 9387
    },
    {
      "epoch": 0.5880455378255218,
      "grad_norm": 0.8018741776908334,
      "learning_rate": 3.827718662504403e-06,
      "loss": 0.4092,
      "step": 9388
    },
    {
      "epoch": 0.5881081758248642,
      "grad_norm": 0.8335119329178756,
      "learning_rate": 3.826732562646341e-06,
      "loss": 0.3922,
      "step": 9389
    },
    {
      "epoch": 0.5881708138242064,
      "grad_norm": 0.9528785943954262,
      "learning_rate": 3.825746511080183e-06,
      "loss": 0.3827,
      "step": 9390
    },
    {
      "epoch": 0.5882334518235488,
      "grad_norm": 0.7955921925813335,
      "learning_rate": 3.8247605078465165e-06,
      "loss": 0.3779,
      "step": 9391
    },
    {
      "epoch": 0.588296089822891,
      "grad_norm": 0.8329017083586222,
      "learning_rate": 3.8237745529859274e-06,
      "loss": 0.4212,
      "step": 9392
    },
    {
      "epoch": 0.5883587278222333,
      "grad_norm": 0.8454336257652281,
      "learning_rate": 3.822788646538995e-06,
      "loss": 0.3435,
      "step": 9393
    },
    {
      "epoch": 0.5884213658215757,
      "grad_norm": 0.8954181779617465,
      "learning_rate": 3.8218027885463e-06,
      "loss": 0.4095,
      "step": 9394
    },
    {
      "epoch": 0.5884840038209179,
      "grad_norm": 0.810641968222367,
      "learning_rate": 3.82081697904842e-06,
      "loss": 0.4359,
      "step": 9395
    },
    {
      "epoch": 0.5885466418202603,
      "grad_norm": 0.8479033286735888,
      "learning_rate": 3.819831218085932e-06,
      "loss": 0.3793,
      "step": 9396
    },
    {
      "epoch": 0.5886092798196025,
      "grad_norm": 0.803098092205834,
      "learning_rate": 3.818845505699412e-06,
      "loss": 0.4045,
      "step": 9397
    },
    {
      "epoch": 0.5886719178189449,
      "grad_norm": 0.8440707514322338,
      "learning_rate": 3.81785984192943e-06,
      "loss": 0.4003,
      "step": 9398
    },
    {
      "epoch": 0.5887345558182872,
      "grad_norm": 0.8574758527256114,
      "learning_rate": 3.816874226816555e-06,
      "loss": 0.4031,
      "step": 9399
    },
    {
      "epoch": 0.5887971938176295,
      "grad_norm": 0.8389969002974337,
      "learning_rate": 3.815888660401357e-06,
      "loss": 0.4007,
      "step": 9400
    },
    {
      "epoch": 0.5888598318169718,
      "grad_norm": 0.8656389501753936,
      "learning_rate": 3.814903142724402e-06,
      "loss": 0.4043,
      "step": 9401
    },
    {
      "epoch": 0.588922469816314,
      "grad_norm": 0.615461160666068,
      "learning_rate": 3.8139176738262528e-06,
      "loss": 0.4292,
      "step": 9402
    },
    {
      "epoch": 0.5889851078156564,
      "grad_norm": 0.8570907382598272,
      "learning_rate": 3.8129322537474735e-06,
      "loss": 0.4203,
      "step": 9403
    },
    {
      "epoch": 0.5890477458149986,
      "grad_norm": 0.8335337512055533,
      "learning_rate": 3.8119468825286226e-06,
      "loss": 0.3955,
      "step": 9404
    },
    {
      "epoch": 0.589110383814341,
      "grad_norm": 0.8660857342628715,
      "learning_rate": 3.8109615602102568e-06,
      "loss": 0.427,
      "step": 9405
    },
    {
      "epoch": 0.5891730218136833,
      "grad_norm": 0.927337389938096,
      "learning_rate": 3.8099762868329355e-06,
      "loss": 0.4314,
      "step": 9406
    },
    {
      "epoch": 0.5892356598130256,
      "grad_norm": 0.7684532403403537,
      "learning_rate": 3.80899106243721e-06,
      "loss": 0.4046,
      "step": 9407
    },
    {
      "epoch": 0.5892982978123679,
      "grad_norm": 0.8568659791856101,
      "learning_rate": 3.808005887063635e-06,
      "loss": 0.4249,
      "step": 9408
    },
    {
      "epoch": 0.5893609358117102,
      "grad_norm": 0.773109381483739,
      "learning_rate": 3.8070207607527587e-06,
      "loss": 0.3687,
      "step": 9409
    },
    {
      "epoch": 0.5894235738110525,
      "grad_norm": 0.8388772470365053,
      "learning_rate": 3.806035683545129e-06,
      "loss": 0.4274,
      "step": 9410
    },
    {
      "epoch": 0.5894862118103947,
      "grad_norm": 0.7978302254720402,
      "learning_rate": 3.8050506554812918e-06,
      "loss": 0.3555,
      "step": 9411
    },
    {
      "epoch": 0.5895488498097371,
      "grad_norm": 0.8068977918272396,
      "learning_rate": 3.8040656766017912e-06,
      "loss": 0.3888,
      "step": 9412
    },
    {
      "epoch": 0.5896114878090793,
      "grad_norm": 0.8442984057532069,
      "learning_rate": 3.8030807469471708e-06,
      "loss": 0.399,
      "step": 9413
    },
    {
      "epoch": 0.5896741258084217,
      "grad_norm": 0.8727898303265742,
      "learning_rate": 3.802095866557966e-06,
      "loss": 0.3885,
      "step": 9414
    },
    {
      "epoch": 0.589736763807764,
      "grad_norm": 0.8542466752485053,
      "learning_rate": 3.80111103547472e-06,
      "loss": 0.3975,
      "step": 9415
    },
    {
      "epoch": 0.5897994018071063,
      "grad_norm": 0.8036711571473193,
      "learning_rate": 3.8001262537379657e-06,
      "loss": 0.3964,
      "step": 9416
    },
    {
      "epoch": 0.5898620398064486,
      "grad_norm": 0.8759750572261364,
      "learning_rate": 3.7991415213882375e-06,
      "loss": 0.3932,
      "step": 9417
    },
    {
      "epoch": 0.5899246778057908,
      "grad_norm": 0.8490069813335078,
      "learning_rate": 3.7981568384660683e-06,
      "loss": 0.4113,
      "step": 9418
    },
    {
      "epoch": 0.5899873158051332,
      "grad_norm": 0.7782619269059683,
      "learning_rate": 3.7971722050119865e-06,
      "loss": 0.4068,
      "step": 9419
    },
    {
      "epoch": 0.5900499538044754,
      "grad_norm": 0.7937355369682404,
      "learning_rate": 3.7961876210665184e-06,
      "loss": 0.4144,
      "step": 9420
    },
    {
      "epoch": 0.5901125918038178,
      "grad_norm": 0.8231844523835266,
      "learning_rate": 3.795203086670193e-06,
      "loss": 0.4028,
      "step": 9421
    },
    {
      "epoch": 0.5901752298031601,
      "grad_norm": 0.8670031125717438,
      "learning_rate": 3.7942186018635315e-06,
      "loss": 0.4506,
      "step": 9422
    },
    {
      "epoch": 0.5902378678025024,
      "grad_norm": 0.8146831374836344,
      "learning_rate": 3.7932341666870543e-06,
      "loss": 0.3922,
      "step": 9423
    },
    {
      "epoch": 0.5903005058018447,
      "grad_norm": 0.878460543920206,
      "learning_rate": 3.792249781181285e-06,
      "loss": 0.4155,
      "step": 9424
    },
    {
      "epoch": 0.590363143801187,
      "grad_norm": 0.8845606164384965,
      "learning_rate": 3.7912654453867374e-06,
      "loss": 0.4409,
      "step": 9425
    },
    {
      "epoch": 0.5904257818005293,
      "grad_norm": 0.8121926677040058,
      "learning_rate": 3.79028115934393e-06,
      "loss": 0.3983,
      "step": 9426
    },
    {
      "epoch": 0.5904884197998715,
      "grad_norm": 0.8383762772025947,
      "learning_rate": 3.789296923093374e-06,
      "loss": 0.4131,
      "step": 9427
    },
    {
      "epoch": 0.5905510577992139,
      "grad_norm": 0.8057120647070957,
      "learning_rate": 3.788312736675581e-06,
      "loss": 0.426,
      "step": 9428
    },
    {
      "epoch": 0.5906136957985562,
      "grad_norm": 0.88289965673992,
      "learning_rate": 3.7873286001310615e-06,
      "loss": 0.4285,
      "step": 9429
    },
    {
      "epoch": 0.5906763337978985,
      "grad_norm": 0.7946113602197927,
      "learning_rate": 3.7863445135003217e-06,
      "loss": 0.392,
      "step": 9430
    },
    {
      "epoch": 0.5907389717972408,
      "grad_norm": 0.8315627646564079,
      "learning_rate": 3.785360476823867e-06,
      "loss": 0.3492,
      "step": 9431
    },
    {
      "epoch": 0.5908016097965831,
      "grad_norm": 0.8648266102349383,
      "learning_rate": 3.7843764901421997e-06,
      "loss": 0.3857,
      "step": 9432
    },
    {
      "epoch": 0.5908642477959254,
      "grad_norm": 0.8826094877142557,
      "learning_rate": 3.7833925534958205e-06,
      "loss": 0.3947,
      "step": 9433
    },
    {
      "epoch": 0.5909268857952678,
      "grad_norm": 0.846168718539482,
      "learning_rate": 3.7824086669252317e-06,
      "loss": 0.4024,
      "step": 9434
    },
    {
      "epoch": 0.59098952379461,
      "grad_norm": 0.846798840516969,
      "learning_rate": 3.7814248304709283e-06,
      "loss": 0.3825,
      "step": 9435
    },
    {
      "epoch": 0.5910521617939523,
      "grad_norm": 0.8312879662819238,
      "learning_rate": 3.780441044173405e-06,
      "loss": 0.4246,
      "step": 9436
    },
    {
      "epoch": 0.5911147997932946,
      "grad_norm": 0.6405909542743241,
      "learning_rate": 3.7794573080731533e-06,
      "loss": 0.473,
      "step": 9437
    },
    {
      "epoch": 0.5911774377926369,
      "grad_norm": 0.9000794865656011,
      "learning_rate": 3.778473622210667e-06,
      "loss": 0.3813,
      "step": 9438
    },
    {
      "epoch": 0.5912400757919792,
      "grad_norm": 0.8117413430873354,
      "learning_rate": 3.777489986626433e-06,
      "loss": 0.3857,
      "step": 9439
    },
    {
      "epoch": 0.5913027137913215,
      "grad_norm": 0.8206051051630182,
      "learning_rate": 3.7765064013609377e-06,
      "loss": 0.3896,
      "step": 9440
    },
    {
      "epoch": 0.5913653517906639,
      "grad_norm": 0.8119922727633184,
      "learning_rate": 3.775522866454665e-06,
      "loss": 0.3758,
      "step": 9441
    },
    {
      "epoch": 0.5914279897900061,
      "grad_norm": 0.8342288915893228,
      "learning_rate": 3.774539381948098e-06,
      "loss": 0.4037,
      "step": 9442
    },
    {
      "epoch": 0.5914906277893484,
      "grad_norm": 0.8183122179919767,
      "learning_rate": 3.7735559478817187e-06,
      "loss": 0.3542,
      "step": 9443
    },
    {
      "epoch": 0.5915532657886907,
      "grad_norm": 0.8807031281507498,
      "learning_rate": 3.7725725642960047e-06,
      "loss": 0.3898,
      "step": 9444
    },
    {
      "epoch": 0.591615903788033,
      "grad_norm": 0.8890342034589305,
      "learning_rate": 3.771589231231432e-06,
      "loss": 0.4258,
      "step": 9445
    },
    {
      "epoch": 0.5916785417873753,
      "grad_norm": 0.627356124502783,
      "learning_rate": 3.770605948728473e-06,
      "loss": 0.4454,
      "step": 9446
    },
    {
      "epoch": 0.5917411797867176,
      "grad_norm": 0.7950358842777119,
      "learning_rate": 3.769622716827602e-06,
      "loss": 0.3671,
      "step": 9447
    },
    {
      "epoch": 0.59180381778606,
      "grad_norm": 0.8939115243339973,
      "learning_rate": 3.7686395355692893e-06,
      "loss": 0.3932,
      "step": 9448
    },
    {
      "epoch": 0.5918664557854022,
      "grad_norm": 0.848168790136014,
      "learning_rate": 3.7676564049940006e-06,
      "loss": 0.4312,
      "step": 9449
    },
    {
      "epoch": 0.5919290937847446,
      "grad_norm": 0.8731012605389756,
      "learning_rate": 3.7666733251422037e-06,
      "loss": 0.4041,
      "step": 9450
    },
    {
      "epoch": 0.5919917317840868,
      "grad_norm": 0.8399871538934409,
      "learning_rate": 3.7656902960543594e-06,
      "loss": 0.3793,
      "step": 9451
    },
    {
      "epoch": 0.5920543697834291,
      "grad_norm": 0.7756633826687063,
      "learning_rate": 3.764707317770935e-06,
      "loss": 0.3746,
      "step": 9452
    },
    {
      "epoch": 0.5921170077827714,
      "grad_norm": 0.8503247446723915,
      "learning_rate": 3.763724390332385e-06,
      "loss": 0.3713,
      "step": 9453
    },
    {
      "epoch": 0.5921796457821137,
      "grad_norm": 0.7750107331050015,
      "learning_rate": 3.7627415137791684e-06,
      "loss": 0.3659,
      "step": 9454
    },
    {
      "epoch": 0.592242283781456,
      "grad_norm": 0.8757175808585556,
      "learning_rate": 3.761758688151742e-06,
      "loss": 0.4447,
      "step": 9455
    },
    {
      "epoch": 0.5923049217807983,
      "grad_norm": 0.8050960732478746,
      "learning_rate": 3.7607759134905577e-06,
      "loss": 0.3914,
      "step": 9456
    },
    {
      "epoch": 0.5923675597801407,
      "grad_norm": 0.8849930608928874,
      "learning_rate": 3.759793189836066e-06,
      "loss": 0.3915,
      "step": 9457
    },
    {
      "epoch": 0.5924301977794829,
      "grad_norm": 0.7605465464536162,
      "learning_rate": 3.7588105172287165e-06,
      "loss": 0.3573,
      "step": 9458
    },
    {
      "epoch": 0.5924928357788253,
      "grad_norm": 0.6268822566713679,
      "learning_rate": 3.7578278957089576e-06,
      "loss": 0.4567,
      "step": 9459
    },
    {
      "epoch": 0.5925554737781675,
      "grad_norm": 0.8704499683119844,
      "learning_rate": 3.7568453253172305e-06,
      "loss": 0.4375,
      "step": 9460
    },
    {
      "epoch": 0.5926181117775098,
      "grad_norm": 0.8160474499310038,
      "learning_rate": 3.755862806093983e-06,
      "loss": 0.3982,
      "step": 9461
    },
    {
      "epoch": 0.5926807497768521,
      "grad_norm": 0.8267396002405445,
      "learning_rate": 3.7548803380796527e-06,
      "loss": 0.4075,
      "step": 9462
    },
    {
      "epoch": 0.5927433877761944,
      "grad_norm": 0.8397192672001312,
      "learning_rate": 3.753897921314679e-06,
      "loss": 0.4105,
      "step": 9463
    },
    {
      "epoch": 0.5928060257755368,
      "grad_norm": 0.760634142549383,
      "learning_rate": 3.752915555839498e-06,
      "loss": 0.3645,
      "step": 9464
    },
    {
      "epoch": 0.592868663774879,
      "grad_norm": 0.8868627640653154,
      "learning_rate": 3.7519332416945444e-06,
      "loss": 0.4089,
      "step": 9465
    },
    {
      "epoch": 0.5929313017742214,
      "grad_norm": 0.84923344775496,
      "learning_rate": 3.7509509789202505e-06,
      "loss": 0.3669,
      "step": 9466
    },
    {
      "epoch": 0.5929939397735636,
      "grad_norm": 0.6040476325027667,
      "learning_rate": 3.7499687675570453e-06,
      "loss": 0.4483,
      "step": 9467
    },
    {
      "epoch": 0.5930565777729059,
      "grad_norm": 0.8468092969315549,
      "learning_rate": 3.7489866076453585e-06,
      "loss": 0.4021,
      "step": 9468
    },
    {
      "epoch": 0.5931192157722482,
      "grad_norm": 0.8008068149367267,
      "learning_rate": 3.7480044992256126e-06,
      "loss": 0.3632,
      "step": 9469
    },
    {
      "epoch": 0.5931818537715905,
      "grad_norm": 0.8332358433011937,
      "learning_rate": 3.747022442338236e-06,
      "loss": 0.431,
      "step": 9470
    },
    {
      "epoch": 0.5932444917709329,
      "grad_norm": 0.7669645437007613,
      "learning_rate": 3.746040437023649e-06,
      "loss": 0.3879,
      "step": 9471
    },
    {
      "epoch": 0.5933071297702751,
      "grad_norm": 0.8515004784850266,
      "learning_rate": 3.745058483322269e-06,
      "loss": 0.4092,
      "step": 9472
    },
    {
      "epoch": 0.5933697677696175,
      "grad_norm": 0.9022660484934328,
      "learning_rate": 3.7440765812745157e-06,
      "loss": 0.4117,
      "step": 9473
    },
    {
      "epoch": 0.5934324057689597,
      "grad_norm": 0.8388213361983982,
      "learning_rate": 3.743094730920803e-06,
      "loss": 0.3953,
      "step": 9474
    },
    {
      "epoch": 0.5934950437683021,
      "grad_norm": 0.9402203610280374,
      "learning_rate": 3.742112932301544e-06,
      "loss": 0.4668,
      "step": 9475
    },
    {
      "epoch": 0.5935576817676443,
      "grad_norm": 0.8576448348935242,
      "learning_rate": 3.7411311854571513e-06,
      "loss": 0.4222,
      "step": 9476
    },
    {
      "epoch": 0.5936203197669866,
      "grad_norm": 0.8281108180985431,
      "learning_rate": 3.740149490428032e-06,
      "loss": 0.3769,
      "step": 9477
    },
    {
      "epoch": 0.593682957766329,
      "grad_norm": 0.7889098866517851,
      "learning_rate": 3.7391678472545945e-06,
      "loss": 0.3916,
      "step": 9478
    },
    {
      "epoch": 0.5937455957656712,
      "grad_norm": 0.8502482253384477,
      "learning_rate": 3.7381862559772396e-06,
      "loss": 0.3606,
      "step": 9479
    },
    {
      "epoch": 0.5938082337650136,
      "grad_norm": 0.772656109352512,
      "learning_rate": 3.737204716636374e-06,
      "loss": 0.3448,
      "step": 9480
    },
    {
      "epoch": 0.5938708717643558,
      "grad_norm": 0.9104187566525882,
      "learning_rate": 3.7362232292723975e-06,
      "loss": 0.4373,
      "step": 9481
    },
    {
      "epoch": 0.5939335097636982,
      "grad_norm": 0.8634733526289926,
      "learning_rate": 3.735241793925707e-06,
      "loss": 0.4477,
      "step": 9482
    },
    {
      "epoch": 0.5939961477630404,
      "grad_norm": 0.7584551463849788,
      "learning_rate": 3.7342604106366996e-06,
      "loss": 0.3401,
      "step": 9483
    },
    {
      "epoch": 0.5940587857623828,
      "grad_norm": 0.8303498776980013,
      "learning_rate": 3.733279079445768e-06,
      "loss": 0.3808,
      "step": 9484
    },
    {
      "epoch": 0.594121423761725,
      "grad_norm": 0.6337941723253169,
      "learning_rate": 3.7322978003933052e-06,
      "loss": 0.4644,
      "step": 9485
    },
    {
      "epoch": 0.5941840617610673,
      "grad_norm": 0.8258789887628346,
      "learning_rate": 3.7313165735197e-06,
      "loss": 0.4049,
      "step": 9486
    },
    {
      "epoch": 0.5942466997604097,
      "grad_norm": 0.8139026906025005,
      "learning_rate": 3.730335398865339e-06,
      "loss": 0.3783,
      "step": 9487
    },
    {
      "epoch": 0.5943093377597519,
      "grad_norm": 1.0233225130584365,
      "learning_rate": 3.7293542764706094e-06,
      "loss": 0.4068,
      "step": 9488
    },
    {
      "epoch": 0.5943719757590943,
      "grad_norm": 0.7989597572381435,
      "learning_rate": 3.7283732063758926e-06,
      "loss": 0.3809,
      "step": 9489
    },
    {
      "epoch": 0.5944346137584365,
      "grad_norm": 0.8275141213857731,
      "learning_rate": 3.7273921886215724e-06,
      "loss": 0.3988,
      "step": 9490
    },
    {
      "epoch": 0.5944972517577789,
      "grad_norm": 0.8285886262625258,
      "learning_rate": 3.7264112232480254e-06,
      "loss": 0.3736,
      "step": 9491
    },
    {
      "epoch": 0.5945598897571212,
      "grad_norm": 0.8390167389170006,
      "learning_rate": 3.7254303102956285e-06,
      "loss": 0.398,
      "step": 9492
    },
    {
      "epoch": 0.5946225277564634,
      "grad_norm": 0.7933910312881017,
      "learning_rate": 3.7244494498047575e-06,
      "loss": 0.3674,
      "step": 9493
    },
    {
      "epoch": 0.5946851657558058,
      "grad_norm": 0.8436520187118405,
      "learning_rate": 3.7234686418157834e-06,
      "loss": 0.3821,
      "step": 9494
    },
    {
      "epoch": 0.594747803755148,
      "grad_norm": 0.8230577918998445,
      "learning_rate": 3.722487886369077e-06,
      "loss": 0.3656,
      "step": 9495
    },
    {
      "epoch": 0.5948104417544904,
      "grad_norm": 0.7794557369419104,
      "learning_rate": 3.7215071835050055e-06,
      "loss": 0.3815,
      "step": 9496
    },
    {
      "epoch": 0.5948730797538326,
      "grad_norm": 0.8890008307668446,
      "learning_rate": 3.720526533263936e-06,
      "loss": 0.4152,
      "step": 9497
    },
    {
      "epoch": 0.594935717753175,
      "grad_norm": 0.7717513542816439,
      "learning_rate": 3.7195459356862314e-06,
      "loss": 0.3647,
      "step": 9498
    },
    {
      "epoch": 0.5949983557525172,
      "grad_norm": 0.8575249711820822,
      "learning_rate": 3.718565390812255e-06,
      "loss": 0.408,
      "step": 9499
    },
    {
      "epoch": 0.5950609937518596,
      "grad_norm": 0.8193440222927083,
      "learning_rate": 3.717584898682365e-06,
      "loss": 0.3657,
      "step": 9500
    },
    {
      "epoch": 0.5951236317512019,
      "grad_norm": 0.8833132776292795,
      "learning_rate": 3.7166044593369166e-06,
      "loss": 0.396,
      "step": 9501
    },
    {
      "epoch": 0.5951862697505441,
      "grad_norm": 0.8370452784796697,
      "learning_rate": 3.715624072816269e-06,
      "loss": 0.4113,
      "step": 9502
    },
    {
      "epoch": 0.5952489077498865,
      "grad_norm": 0.8340777373402584,
      "learning_rate": 3.7146437391607727e-06,
      "loss": 0.3977,
      "step": 9503
    },
    {
      "epoch": 0.5953115457492287,
      "grad_norm": 0.8870956818192562,
      "learning_rate": 3.7136634584107787e-06,
      "loss": 0.4298,
      "step": 9504
    },
    {
      "epoch": 0.5953741837485711,
      "grad_norm": 0.7956984755005726,
      "learning_rate": 3.712683230606634e-06,
      "loss": 0.3898,
      "step": 9505
    },
    {
      "epoch": 0.5954368217479133,
      "grad_norm": 0.828677090603777,
      "learning_rate": 3.7117030557886857e-06,
      "loss": 0.4182,
      "step": 9506
    },
    {
      "epoch": 0.5954994597472557,
      "grad_norm": 0.7821305625638973,
      "learning_rate": 3.7107229339972817e-06,
      "loss": 0.379,
      "step": 9507
    },
    {
      "epoch": 0.595562097746598,
      "grad_norm": 0.8581747992376428,
      "learning_rate": 3.7097428652727596e-06,
      "loss": 0.4136,
      "step": 9508
    },
    {
      "epoch": 0.5956247357459403,
      "grad_norm": 0.8144620069815539,
      "learning_rate": 3.708762849655462e-06,
      "loss": 0.4002,
      "step": 9509
    },
    {
      "epoch": 0.5956873737452826,
      "grad_norm": 0.786103301405273,
      "learning_rate": 3.707782887185723e-06,
      "loss": 0.4003,
      "step": 9510
    },
    {
      "epoch": 0.5957500117446248,
      "grad_norm": 0.6403882062705168,
      "learning_rate": 3.7068029779038815e-06,
      "loss": 0.4308,
      "step": 9511
    },
    {
      "epoch": 0.5958126497439672,
      "grad_norm": 0.8425261952745727,
      "learning_rate": 3.70582312185027e-06,
      "loss": 0.3928,
      "step": 9512
    },
    {
      "epoch": 0.5958752877433094,
      "grad_norm": 0.8424423674101679,
      "learning_rate": 3.704843319065218e-06,
      "loss": 0.4174,
      "step": 9513
    },
    {
      "epoch": 0.5959379257426518,
      "grad_norm": 0.8693277191337955,
      "learning_rate": 3.7038635695890556e-06,
      "loss": 0.4516,
      "step": 9514
    },
    {
      "epoch": 0.5960005637419941,
      "grad_norm": 0.7891615282858704,
      "learning_rate": 3.7028838734621077e-06,
      "loss": 0.3558,
      "step": 9515
    },
    {
      "epoch": 0.5960632017413364,
      "grad_norm": 0.8972776921305003,
      "learning_rate": 3.7019042307247023e-06,
      "loss": 0.4088,
      "step": 9516
    },
    {
      "epoch": 0.5961258397406787,
      "grad_norm": 0.8228242525462338,
      "learning_rate": 3.7009246414171594e-06,
      "loss": 0.3926,
      "step": 9517
    },
    {
      "epoch": 0.596188477740021,
      "grad_norm": 0.9367546038982905,
      "learning_rate": 3.6999451055797987e-06,
      "loss": 0.4157,
      "step": 9518
    },
    {
      "epoch": 0.5962511157393633,
      "grad_norm": 0.8423935749449137,
      "learning_rate": 3.69896562325294e-06,
      "loss": 0.4024,
      "step": 9519
    },
    {
      "epoch": 0.5963137537387055,
      "grad_norm": 0.861903403552497,
      "learning_rate": 3.6979861944768974e-06,
      "loss": 0.377,
      "step": 9520
    },
    {
      "epoch": 0.5963763917380479,
      "grad_norm": 0.7902364947374011,
      "learning_rate": 3.697006819291985e-06,
      "loss": 0.3707,
      "step": 9521
    },
    {
      "epoch": 0.5964390297373902,
      "grad_norm": 0.793768571930897,
      "learning_rate": 3.6960274977385123e-06,
      "loss": 0.378,
      "step": 9522
    },
    {
      "epoch": 0.5965016677367325,
      "grad_norm": 0.8371608646844813,
      "learning_rate": 3.6950482298567913e-06,
      "loss": 0.403,
      "step": 9523
    },
    {
      "epoch": 0.5965643057360748,
      "grad_norm": 0.7931060162781939,
      "learning_rate": 3.6940690156871274e-06,
      "loss": 0.4132,
      "step": 9524
    },
    {
      "epoch": 0.5966269437354171,
      "grad_norm": 0.8701771980986978,
      "learning_rate": 3.6930898552698223e-06,
      "loss": 0.4017,
      "step": 9525
    },
    {
      "epoch": 0.5966895817347594,
      "grad_norm": 0.8403652495239872,
      "learning_rate": 3.692110748645184e-06,
      "loss": 0.4108,
      "step": 9526
    },
    {
      "epoch": 0.5967522197341016,
      "grad_norm": 0.8319355590419495,
      "learning_rate": 3.6911316958535094e-06,
      "loss": 0.4117,
      "step": 9527
    },
    {
      "epoch": 0.596814857733444,
      "grad_norm": 0.8175857936130638,
      "learning_rate": 3.6901526969350975e-06,
      "loss": 0.4001,
      "step": 9528
    },
    {
      "epoch": 0.5968774957327863,
      "grad_norm": 0.6005760820867894,
      "learning_rate": 3.689173751930244e-06,
      "loss": 0.4772,
      "step": 9529
    },
    {
      "epoch": 0.5969401337321286,
      "grad_norm": 0.8715194893183594,
      "learning_rate": 3.688194860879242e-06,
      "loss": 0.4,
      "step": 9530
    },
    {
      "epoch": 0.5970027717314709,
      "grad_norm": 0.6588809133179446,
      "learning_rate": 3.687216023822382e-06,
      "loss": 0.4438,
      "step": 9531
    },
    {
      "epoch": 0.5970654097308132,
      "grad_norm": 0.6240061007302489,
      "learning_rate": 3.6862372407999556e-06,
      "loss": 0.4496,
      "step": 9532
    },
    {
      "epoch": 0.5971280477301555,
      "grad_norm": 0.8883927051904333,
      "learning_rate": 3.6852585118522478e-06,
      "loss": 0.4145,
      "step": 9533
    },
    {
      "epoch": 0.5971906857294978,
      "grad_norm": 0.6704239600954294,
      "learning_rate": 3.6842798370195415e-06,
      "loss": 0.4575,
      "step": 9534
    },
    {
      "epoch": 0.5972533237288401,
      "grad_norm": 0.8218919228067874,
      "learning_rate": 3.6833012163421235e-06,
      "loss": 0.3719,
      "step": 9535
    },
    {
      "epoch": 0.5973159617281824,
      "grad_norm": 0.8017756892051985,
      "learning_rate": 3.682322649860271e-06,
      "loss": 0.3934,
      "step": 9536
    },
    {
      "epoch": 0.5973785997275247,
      "grad_norm": 0.8862374378723851,
      "learning_rate": 3.6813441376142638e-06,
      "loss": 0.4384,
      "step": 9537
    },
    {
      "epoch": 0.597441237726867,
      "grad_norm": 0.921630330863919,
      "learning_rate": 3.6803656796443766e-06,
      "loss": 0.4205,
      "step": 9538
    },
    {
      "epoch": 0.5975038757262093,
      "grad_norm": 0.8866821112381152,
      "learning_rate": 3.679387275990882e-06,
      "loss": 0.4035,
      "step": 9539
    },
    {
      "epoch": 0.5975665137255516,
      "grad_norm": 0.7721333598036609,
      "learning_rate": 3.678408926694053e-06,
      "loss": 0.3682,
      "step": 9540
    },
    {
      "epoch": 0.597629151724894,
      "grad_norm": 0.8077549466257694,
      "learning_rate": 3.6774306317941577e-06,
      "loss": 0.358,
      "step": 9541
    },
    {
      "epoch": 0.5976917897242362,
      "grad_norm": 0.8577152519304525,
      "learning_rate": 3.6764523913314635e-06,
      "loss": 0.3646,
      "step": 9542
    },
    {
      "epoch": 0.5977544277235786,
      "grad_norm": 0.8318066595261245,
      "learning_rate": 3.675474205346233e-06,
      "loss": 0.3751,
      "step": 9543
    },
    {
      "epoch": 0.5978170657229208,
      "grad_norm": 0.615956177021101,
      "learning_rate": 3.67449607387873e-06,
      "loss": 0.4506,
      "step": 9544
    },
    {
      "epoch": 0.5978797037222631,
      "grad_norm": 0.6298886932727799,
      "learning_rate": 3.673517996969216e-06,
      "loss": 0.4698,
      "step": 9545
    },
    {
      "epoch": 0.5979423417216054,
      "grad_norm": 0.8663668682324432,
      "learning_rate": 3.672539974657948e-06,
      "loss": 0.4367,
      "step": 9546
    },
    {
      "epoch": 0.5980049797209477,
      "grad_norm": 0.8916734013180277,
      "learning_rate": 3.6715620069851804e-06,
      "loss": 0.4007,
      "step": 9547
    },
    {
      "epoch": 0.59806761772029,
      "grad_norm": 0.78975835548435,
      "learning_rate": 3.6705840939911665e-06,
      "loss": 0.3841,
      "step": 9548
    },
    {
      "epoch": 0.5981302557196323,
      "grad_norm": 0.8155362605955118,
      "learning_rate": 3.669606235716159e-06,
      "loss": 0.4167,
      "step": 9549
    },
    {
      "epoch": 0.5981928937189747,
      "grad_norm": 0.8787309027916234,
      "learning_rate": 3.6686284322004056e-06,
      "loss": 0.4213,
      "step": 9550
    },
    {
      "epoch": 0.5982555317183169,
      "grad_norm": 0.8230223901511561,
      "learning_rate": 3.667650683484153e-06,
      "loss": 0.4094,
      "step": 9551
    },
    {
      "epoch": 0.5983181697176592,
      "grad_norm": 0.7967650696436859,
      "learning_rate": 3.666672989607644e-06,
      "loss": 0.3965,
      "step": 9552
    },
    {
      "epoch": 0.5983808077170015,
      "grad_norm": 0.8456189579865719,
      "learning_rate": 3.6656953506111226e-06,
      "loss": 0.3986,
      "step": 9553
    },
    {
      "epoch": 0.5984434457163438,
      "grad_norm": 0.833354623484281,
      "learning_rate": 3.6647177665348294e-06,
      "loss": 0.406,
      "step": 9554
    },
    {
      "epoch": 0.5985060837156861,
      "grad_norm": 0.8513875893534262,
      "learning_rate": 3.6637402374190013e-06,
      "loss": 0.4485,
      "step": 9555
    },
    {
      "epoch": 0.5985687217150284,
      "grad_norm": 0.7705486947057313,
      "learning_rate": 3.662762763303872e-06,
      "loss": 0.3886,
      "step": 9556
    },
    {
      "epoch": 0.5986313597143708,
      "grad_norm": 0.8835269357940436,
      "learning_rate": 3.661785344229676e-06,
      "loss": 0.3766,
      "step": 9557
    },
    {
      "epoch": 0.598693997713713,
      "grad_norm": 0.8692560382718213,
      "learning_rate": 3.6608079802366435e-06,
      "loss": 0.4248,
      "step": 9558
    },
    {
      "epoch": 0.5987566357130554,
      "grad_norm": 0.891332880557128,
      "learning_rate": 3.6598306713650035e-06,
      "loss": 0.4002,
      "step": 9559
    },
    {
      "epoch": 0.5988192737123976,
      "grad_norm": 0.8886426178934678,
      "learning_rate": 3.6588534176549806e-06,
      "loss": 0.412,
      "step": 9560
    },
    {
      "epoch": 0.5988819117117399,
      "grad_norm": 0.8217234745649474,
      "learning_rate": 3.657876219146801e-06,
      "loss": 0.3961,
      "step": 9561
    },
    {
      "epoch": 0.5989445497110822,
      "grad_norm": 0.8619632633168278,
      "learning_rate": 3.6568990758806833e-06,
      "loss": 0.3946,
      "step": 9562
    },
    {
      "epoch": 0.5990071877104245,
      "grad_norm": 0.854135534415068,
      "learning_rate": 3.655921987896851e-06,
      "loss": 0.3925,
      "step": 9563
    },
    {
      "epoch": 0.5990698257097669,
      "grad_norm": 0.8486975708022418,
      "learning_rate": 3.6549449552355197e-06,
      "loss": 0.3989,
      "step": 9564
    },
    {
      "epoch": 0.5991324637091091,
      "grad_norm": 0.8211524165616707,
      "learning_rate": 3.653967977936902e-06,
      "loss": 0.3778,
      "step": 9565
    },
    {
      "epoch": 0.5991951017084515,
      "grad_norm": 0.8136029636581908,
      "learning_rate": 3.652991056041213e-06,
      "loss": 0.3586,
      "step": 9566
    },
    {
      "epoch": 0.5992577397077937,
      "grad_norm": 0.8283909139487015,
      "learning_rate": 3.652014189588663e-06,
      "loss": 0.4039,
      "step": 9567
    },
    {
      "epoch": 0.5993203777071361,
      "grad_norm": 0.8568707250421048,
      "learning_rate": 3.6510373786194586e-06,
      "loss": 0.4078,
      "step": 9568
    },
    {
      "epoch": 0.5993830157064783,
      "grad_norm": 0.8455894458273573,
      "learning_rate": 3.6500606231738052e-06,
      "loss": 0.4011,
      "step": 9569
    },
    {
      "epoch": 0.5994456537058206,
      "grad_norm": 0.8834199370739313,
      "learning_rate": 3.6490839232919085e-06,
      "loss": 0.4195,
      "step": 9570
    },
    {
      "epoch": 0.599508291705163,
      "grad_norm": 0.8852088892837442,
      "learning_rate": 3.6481072790139647e-06,
      "loss": 0.3729,
      "step": 9571
    },
    {
      "epoch": 0.5995709297045052,
      "grad_norm": 0.8068068019733186,
      "learning_rate": 3.6471306903801797e-06,
      "loss": 0.3814,
      "step": 9572
    },
    {
      "epoch": 0.5996335677038476,
      "grad_norm": 0.8852697432682156,
      "learning_rate": 3.6461541574307467e-06,
      "loss": 0.4125,
      "step": 9573
    },
    {
      "epoch": 0.5996962057031898,
      "grad_norm": 0.8064310450742124,
      "learning_rate": 3.645177680205858e-06,
      "loss": 0.4157,
      "step": 9574
    },
    {
      "epoch": 0.5997588437025322,
      "grad_norm": 0.8147988973628107,
      "learning_rate": 3.6442012587457086e-06,
      "loss": 0.3481,
      "step": 9575
    },
    {
      "epoch": 0.5998214817018744,
      "grad_norm": 0.9183682649586963,
      "learning_rate": 3.6432248930904878e-06,
      "loss": 0.3835,
      "step": 9576
    },
    {
      "epoch": 0.5998841197012167,
      "grad_norm": 0.7870756721571569,
      "learning_rate": 3.6422485832803804e-06,
      "loss": 0.3744,
      "step": 9577
    },
    {
      "epoch": 0.599946757700559,
      "grad_norm": 0.6304989667842099,
      "learning_rate": 3.641272329355575e-06,
      "loss": 0.4614,
      "step": 9578
    },
    {
      "epoch": 0.6000093956999013,
      "grad_norm": 0.8638603811887173,
      "learning_rate": 3.6402961313562525e-06,
      "loss": 0.3796,
      "step": 9579
    },
    {
      "epoch": 0.6000720336992437,
      "grad_norm": 0.8886887615702299,
      "learning_rate": 3.6393199893225912e-06,
      "loss": 0.366,
      "step": 9580
    },
    {
      "epoch": 0.6001346716985859,
      "grad_norm": 0.8042618091287869,
      "learning_rate": 3.6383439032947743e-06,
      "loss": 0.4051,
      "step": 9581
    },
    {
      "epoch": 0.6001973096979283,
      "grad_norm": 0.8516525896948317,
      "learning_rate": 3.637367873312975e-06,
      "loss": 0.3609,
      "step": 9582
    },
    {
      "epoch": 0.6002599476972705,
      "grad_norm": 0.8043546614597497,
      "learning_rate": 3.6363918994173653e-06,
      "loss": 0.384,
      "step": 9583
    },
    {
      "epoch": 0.6003225856966129,
      "grad_norm": 0.7979289646080456,
      "learning_rate": 3.6354159816481193e-06,
      "loss": 0.3604,
      "step": 9584
    },
    {
      "epoch": 0.6003852236959552,
      "grad_norm": 0.8580939208378414,
      "learning_rate": 3.634440120045405e-06,
      "loss": 0.4003,
      "step": 9585
    },
    {
      "epoch": 0.6004478616952974,
      "grad_norm": 0.7846405047787695,
      "learning_rate": 3.6334643146493875e-06,
      "loss": 0.3723,
      "step": 9586
    },
    {
      "epoch": 0.6005104996946398,
      "grad_norm": 0.7960274327288812,
      "learning_rate": 3.632488565500234e-06,
      "loss": 0.3935,
      "step": 9587
    },
    {
      "epoch": 0.600573137693982,
      "grad_norm": 0.8665987951593725,
      "learning_rate": 3.631512872638104e-06,
      "loss": 0.3979,
      "step": 9588
    },
    {
      "epoch": 0.6006357756933244,
      "grad_norm": 0.8217486749044349,
      "learning_rate": 3.630537236103158e-06,
      "loss": 0.3887,
      "step": 9589
    },
    {
      "epoch": 0.6006984136926666,
      "grad_norm": 0.8543184415357866,
      "learning_rate": 3.629561655935554e-06,
      "loss": 0.3914,
      "step": 9590
    },
    {
      "epoch": 0.600761051692009,
      "grad_norm": 0.8223504747090821,
      "learning_rate": 3.6285861321754455e-06,
      "loss": 0.3727,
      "step": 9591
    },
    {
      "epoch": 0.6008236896913512,
      "grad_norm": 0.7973731334375687,
      "learning_rate": 3.6276106648629884e-06,
      "loss": 0.3505,
      "step": 9592
    },
    {
      "epoch": 0.6008863276906936,
      "grad_norm": 0.8238963511537256,
      "learning_rate": 3.6266352540383303e-06,
      "loss": 0.3854,
      "step": 9593
    },
    {
      "epoch": 0.6009489656900359,
      "grad_norm": 0.8468260436768466,
      "learning_rate": 3.6256598997416203e-06,
      "loss": 0.4066,
      "step": 9594
    },
    {
      "epoch": 0.6010116036893781,
      "grad_norm": 0.8554000358730725,
      "learning_rate": 3.6246846020130033e-06,
      "loss": 0.3912,
      "step": 9595
    },
    {
      "epoch": 0.6010742416887205,
      "grad_norm": 0.8107039907994752,
      "learning_rate": 3.6237093608926234e-06,
      "loss": 0.416,
      "step": 9596
    },
    {
      "epoch": 0.6011368796880627,
      "grad_norm": 0.7814688809840568,
      "learning_rate": 3.6227341764206225e-06,
      "loss": 0.3757,
      "step": 9597
    },
    {
      "epoch": 0.6011995176874051,
      "grad_norm": 0.8732152938828355,
      "learning_rate": 3.6217590486371364e-06,
      "loss": 0.3822,
      "step": 9598
    },
    {
      "epoch": 0.6012621556867473,
      "grad_norm": 0.834926256365086,
      "learning_rate": 3.620783977582305e-06,
      "loss": 0.365,
      "step": 9599
    },
    {
      "epoch": 0.6013247936860897,
      "grad_norm": 0.9084849503960782,
      "learning_rate": 3.619808963296261e-06,
      "loss": 0.4147,
      "step": 9600
    },
    {
      "epoch": 0.601387431685432,
      "grad_norm": 0.7114398305689339,
      "learning_rate": 3.6188340058191367e-06,
      "loss": 0.3706,
      "step": 9601
    },
    {
      "epoch": 0.6014500696847742,
      "grad_norm": 0.8453936973535966,
      "learning_rate": 3.617859105191061e-06,
      "loss": 0.3918,
      "step": 9602
    },
    {
      "epoch": 0.6015127076841166,
      "grad_norm": 0.8162154463793851,
      "learning_rate": 3.6168842614521606e-06,
      "loss": 0.3697,
      "step": 9603
    },
    {
      "epoch": 0.6015753456834588,
      "grad_norm": 0.6506604899101996,
      "learning_rate": 3.6159094746425616e-06,
      "loss": 0.4512,
      "step": 9604
    },
    {
      "epoch": 0.6016379836828012,
      "grad_norm": 0.7982046872804865,
      "learning_rate": 3.614934744802385e-06,
      "loss": 0.3936,
      "step": 9605
    },
    {
      "epoch": 0.6017006216821434,
      "grad_norm": 0.7179636280204259,
      "learning_rate": 3.6139600719717515e-06,
      "loss": 0.4594,
      "step": 9606
    },
    {
      "epoch": 0.6017632596814858,
      "grad_norm": 0.8679618139448028,
      "learning_rate": 3.6129854561907786e-06,
      "loss": 0.4035,
      "step": 9607
    },
    {
      "epoch": 0.6018258976808281,
      "grad_norm": 0.8924998507843402,
      "learning_rate": 3.6120108974995804e-06,
      "loss": 0.3991,
      "step": 9608
    },
    {
      "epoch": 0.6018885356801704,
      "grad_norm": 0.8347072823109402,
      "learning_rate": 3.611036395938272e-06,
      "loss": 0.425,
      "step": 9609
    },
    {
      "epoch": 0.6019511736795127,
      "grad_norm": 0.9209942459325431,
      "learning_rate": 3.6100619515469644e-06,
      "loss": 0.4199,
      "step": 9610
    },
    {
      "epoch": 0.6020138116788549,
      "grad_norm": 0.8166827203686606,
      "learning_rate": 3.6090875643657646e-06,
      "loss": 0.3668,
      "step": 9611
    },
    {
      "epoch": 0.6020764496781973,
      "grad_norm": 0.8336717128636475,
      "learning_rate": 3.608113234434778e-06,
      "loss": 0.3979,
      "step": 9612
    },
    {
      "epoch": 0.6021390876775395,
      "grad_norm": 0.8618165610151592,
      "learning_rate": 3.60713896179411e-06,
      "loss": 0.3838,
      "step": 9613
    },
    {
      "epoch": 0.6022017256768819,
      "grad_norm": 0.7921176816242823,
      "learning_rate": 3.6061647464838613e-06,
      "loss": 0.3595,
      "step": 9614
    },
    {
      "epoch": 0.6022643636762242,
      "grad_norm": 0.842328117078862,
      "learning_rate": 3.6051905885441297e-06,
      "loss": 0.3892,
      "step": 9615
    },
    {
      "epoch": 0.6023270016755665,
      "grad_norm": 0.8438838737828993,
      "learning_rate": 3.6042164880150114e-06,
      "loss": 0.469,
      "step": 9616
    },
    {
      "epoch": 0.6023896396749088,
      "grad_norm": 0.8167333640402633,
      "learning_rate": 3.6032424449366006e-06,
      "loss": 0.3813,
      "step": 9617
    },
    {
      "epoch": 0.6024522776742511,
      "grad_norm": 0.8352752535267886,
      "learning_rate": 3.6022684593489922e-06,
      "loss": 0.364,
      "step": 9618
    },
    {
      "epoch": 0.6025149156735934,
      "grad_norm": 0.7169465918563471,
      "learning_rate": 3.6012945312922733e-06,
      "loss": 0.4594,
      "step": 9619
    },
    {
      "epoch": 0.6025775536729356,
      "grad_norm": 0.8154730263712073,
      "learning_rate": 3.6003206608065313e-06,
      "loss": 0.4266,
      "step": 9620
    },
    {
      "epoch": 0.602640191672278,
      "grad_norm": 0.8764435463371122,
      "learning_rate": 3.5993468479318495e-06,
      "loss": 0.3729,
      "step": 9621
    },
    {
      "epoch": 0.6027028296716203,
      "grad_norm": 0.7756398272231071,
      "learning_rate": 3.5983730927083126e-06,
      "loss": 0.3618,
      "step": 9622
    },
    {
      "epoch": 0.6027654676709626,
      "grad_norm": 0.863466457340737,
      "learning_rate": 3.5973993951759987e-06,
      "loss": 0.3977,
      "step": 9623
    },
    {
      "epoch": 0.6028281056703049,
      "grad_norm": 0.8073139402399383,
      "learning_rate": 3.596425755374986e-06,
      "loss": 0.3562,
      "step": 9624
    },
    {
      "epoch": 0.6028907436696472,
      "grad_norm": 0.8104859733718505,
      "learning_rate": 3.5954521733453503e-06,
      "loss": 0.3919,
      "step": 9625
    },
    {
      "epoch": 0.6029533816689895,
      "grad_norm": 0.7762998907569248,
      "learning_rate": 3.5944786491271617e-06,
      "loss": 0.3843,
      "step": 9626
    },
    {
      "epoch": 0.6030160196683318,
      "grad_norm": 0.8544914970410309,
      "learning_rate": 3.593505182760495e-06,
      "loss": 0.3956,
      "step": 9627
    },
    {
      "epoch": 0.6030786576676741,
      "grad_norm": 0.8208390876517021,
      "learning_rate": 3.592531774285416e-06,
      "loss": 0.3355,
      "step": 9628
    },
    {
      "epoch": 0.6031412956670164,
      "grad_norm": 0.7844438831544356,
      "learning_rate": 3.5915584237419887e-06,
      "loss": 0.3923,
      "step": 9629
    },
    {
      "epoch": 0.6032039336663587,
      "grad_norm": 0.8347378194490674,
      "learning_rate": 3.59058513117028e-06,
      "loss": 0.3614,
      "step": 9630
    },
    {
      "epoch": 0.603266571665701,
      "grad_norm": 0.8543432083166389,
      "learning_rate": 3.589611896610349e-06,
      "loss": 0.4087,
      "step": 9631
    },
    {
      "epoch": 0.6033292096650433,
      "grad_norm": 0.9123922950967838,
      "learning_rate": 3.588638720102253e-06,
      "loss": 0.4152,
      "step": 9632
    },
    {
      "epoch": 0.6033918476643856,
      "grad_norm": 0.7922271917990966,
      "learning_rate": 3.587665601686049e-06,
      "loss": 0.3913,
      "step": 9633
    },
    {
      "epoch": 0.603454485663728,
      "grad_norm": 0.8407933360886621,
      "learning_rate": 3.586692541401792e-06,
      "loss": 0.3838,
      "step": 9634
    },
    {
      "epoch": 0.6035171236630702,
      "grad_norm": 0.8209278792319589,
      "learning_rate": 3.5857195392895303e-06,
      "loss": 0.4568,
      "step": 9635
    },
    {
      "epoch": 0.6035797616624125,
      "grad_norm": 0.8622343159893129,
      "learning_rate": 3.584746595389317e-06,
      "loss": 0.4463,
      "step": 9636
    },
    {
      "epoch": 0.6036423996617548,
      "grad_norm": 0.8120149555028301,
      "learning_rate": 3.5837737097411963e-06,
      "loss": 0.4269,
      "step": 9637
    },
    {
      "epoch": 0.6037050376610971,
      "grad_norm": 0.8840096458509763,
      "learning_rate": 3.5828008823852118e-06,
      "loss": 0.3872,
      "step": 9638
    },
    {
      "epoch": 0.6037676756604394,
      "grad_norm": 0.8224846539805668,
      "learning_rate": 3.581828113361407e-06,
      "loss": 0.3854,
      "step": 9639
    },
    {
      "epoch": 0.6038303136597817,
      "grad_norm": 0.9157380847166438,
      "learning_rate": 3.5808554027098196e-06,
      "loss": 0.4212,
      "step": 9640
    },
    {
      "epoch": 0.603892951659124,
      "grad_norm": 0.5842337134352109,
      "learning_rate": 3.579882750470488e-06,
      "loss": 0.4696,
      "step": 9641
    },
    {
      "epoch": 0.6039555896584663,
      "grad_norm": 0.8123910131847504,
      "learning_rate": 3.578910156683445e-06,
      "loss": 0.3557,
      "step": 9642
    },
    {
      "epoch": 0.6040182276578087,
      "grad_norm": 0.8072377170760772,
      "learning_rate": 3.577937621388724e-06,
      "loss": 0.3874,
      "step": 9643
    },
    {
      "epoch": 0.6040808656571509,
      "grad_norm": 0.8252856511342065,
      "learning_rate": 3.5769651446263533e-06,
      "loss": 0.3974,
      "step": 9644
    },
    {
      "epoch": 0.6041435036564932,
      "grad_norm": 0.9085029999267213,
      "learning_rate": 3.5759927264363624e-06,
      "loss": 0.4052,
      "step": 9645
    },
    {
      "epoch": 0.6042061416558355,
      "grad_norm": 0.8151278442976209,
      "learning_rate": 3.5750203668587763e-06,
      "loss": 0.3736,
      "step": 9646
    },
    {
      "epoch": 0.6042687796551778,
      "grad_norm": 0.8446415651455651,
      "learning_rate": 3.574048065933615e-06,
      "loss": 0.423,
      "step": 9647
    },
    {
      "epoch": 0.6043314176545201,
      "grad_norm": 0.8439053864206462,
      "learning_rate": 3.573075823700901e-06,
      "loss": 0.3885,
      "step": 9648
    },
    {
      "epoch": 0.6043940556538624,
      "grad_norm": 0.7770111906514254,
      "learning_rate": 3.5721036402006504e-06,
      "loss": 0.3725,
      "step": 9649
    },
    {
      "epoch": 0.6044566936532048,
      "grad_norm": 0.8115748031712515,
      "learning_rate": 3.5711315154728787e-06,
      "loss": 0.3957,
      "step": 9650
    },
    {
      "epoch": 0.604519331652547,
      "grad_norm": 0.9089311651471786,
      "learning_rate": 3.5701594495575996e-06,
      "loss": 0.3869,
      "step": 9651
    },
    {
      "epoch": 0.6045819696518894,
      "grad_norm": 0.8725336598712466,
      "learning_rate": 3.569187442494823e-06,
      "loss": 0.3534,
      "step": 9652
    },
    {
      "epoch": 0.6046446076512316,
      "grad_norm": 0.8574064844448988,
      "learning_rate": 3.5682154943245573e-06,
      "loss": 0.3853,
      "step": 9653
    },
    {
      "epoch": 0.6047072456505739,
      "grad_norm": 0.865206634551699,
      "learning_rate": 3.567243605086805e-06,
      "loss": 0.4021,
      "step": 9654
    },
    {
      "epoch": 0.6047698836499162,
      "grad_norm": 0.8094246797801979,
      "learning_rate": 3.5662717748215724e-06,
      "loss": 0.3956,
      "step": 9655
    },
    {
      "epoch": 0.6048325216492585,
      "grad_norm": 1.0485827203990103,
      "learning_rate": 3.565300003568862e-06,
      "loss": 0.4187,
      "step": 9656
    },
    {
      "epoch": 0.6048951596486009,
      "grad_norm": 0.8166640186620533,
      "learning_rate": 3.5643282913686687e-06,
      "loss": 0.3742,
      "step": 9657
    },
    {
      "epoch": 0.6049577976479431,
      "grad_norm": 0.81365600960788,
      "learning_rate": 3.5633566382609896e-06,
      "loss": 0.4164,
      "step": 9658
    },
    {
      "epoch": 0.6050204356472855,
      "grad_norm": 0.826857984067531,
      "learning_rate": 3.5623850442858176e-06,
      "loss": 0.4417,
      "step": 9659
    },
    {
      "epoch": 0.6050830736466277,
      "grad_norm": 0.794450685166045,
      "learning_rate": 3.5614135094831443e-06,
      "loss": 0.3676,
      "step": 9660
    },
    {
      "epoch": 0.60514571164597,
      "grad_norm": 0.6659454984540912,
      "learning_rate": 3.560442033892958e-06,
      "loss": 0.4595,
      "step": 9661
    },
    {
      "epoch": 0.6052083496453123,
      "grad_norm": 0.6705442107669021,
      "learning_rate": 3.559470617555244e-06,
      "loss": 0.4439,
      "step": 9662
    },
    {
      "epoch": 0.6052709876446546,
      "grad_norm": 0.8225150113328279,
      "learning_rate": 3.5584992605099864e-06,
      "loss": 0.3872,
      "step": 9663
    },
    {
      "epoch": 0.605333625643997,
      "grad_norm": 0.8666271635703626,
      "learning_rate": 3.557527962797167e-06,
      "loss": 0.3891,
      "step": 9664
    },
    {
      "epoch": 0.6053962636433392,
      "grad_norm": 0.8608107915849893,
      "learning_rate": 3.5565567244567656e-06,
      "loss": 0.3947,
      "step": 9665
    },
    {
      "epoch": 0.6054589016426816,
      "grad_norm": 0.7958299310308541,
      "learning_rate": 3.5555855455287576e-06,
      "loss": 0.3673,
      "step": 9666
    },
    {
      "epoch": 0.6055215396420238,
      "grad_norm": 0.8091326280492803,
      "learning_rate": 3.5546144260531164e-06,
      "loss": 0.4318,
      "step": 9667
    },
    {
      "epoch": 0.6055841776413662,
      "grad_norm": 0.7996416482459321,
      "learning_rate": 3.553643366069812e-06,
      "loss": 0.4065,
      "step": 9668
    },
    {
      "epoch": 0.6056468156407084,
      "grad_norm": 0.8275159691540213,
      "learning_rate": 3.5526723656188168e-06,
      "loss": 0.4621,
      "step": 9669
    },
    {
      "epoch": 0.6057094536400507,
      "grad_norm": 0.8608102874422148,
      "learning_rate": 3.551701424740096e-06,
      "loss": 0.4262,
      "step": 9670
    },
    {
      "epoch": 0.605772091639393,
      "grad_norm": 0.8071048780718468,
      "learning_rate": 3.550730543473612e-06,
      "loss": 0.392,
      "step": 9671
    },
    {
      "epoch": 0.6058347296387353,
      "grad_norm": 0.9783096324787565,
      "learning_rate": 3.549759721859327e-06,
      "loss": 0.4378,
      "step": 9672
    },
    {
      "epoch": 0.6058973676380777,
      "grad_norm": 0.8873433301721682,
      "learning_rate": 3.5487889599372022e-06,
      "loss": 0.3781,
      "step": 9673
    },
    {
      "epoch": 0.6059600056374199,
      "grad_norm": 0.9146620930280432,
      "learning_rate": 3.547818257747194e-06,
      "loss": 0.4011,
      "step": 9674
    },
    {
      "epoch": 0.6060226436367623,
      "grad_norm": 0.8611045308575306,
      "learning_rate": 3.5468476153292554e-06,
      "loss": 0.4009,
      "step": 9675
    },
    {
      "epoch": 0.6060852816361045,
      "grad_norm": 0.8390794536313446,
      "learning_rate": 3.5458770327233377e-06,
      "loss": 0.3869,
      "step": 9676
    },
    {
      "epoch": 0.6061479196354469,
      "grad_norm": 0.8065591163643271,
      "learning_rate": 3.5449065099693923e-06,
      "loss": 0.3869,
      "step": 9677
    },
    {
      "epoch": 0.6062105576347891,
      "grad_norm": 0.8088138973754808,
      "learning_rate": 3.5439360471073653e-06,
      "loss": 0.3627,
      "step": 9678
    },
    {
      "epoch": 0.6062731956341314,
      "grad_norm": 0.8489917489202126,
      "learning_rate": 3.542965644177201e-06,
      "loss": 0.3984,
      "step": 9679
    },
    {
      "epoch": 0.6063358336334738,
      "grad_norm": 0.8439624694958294,
      "learning_rate": 3.541995301218839e-06,
      "loss": 0.4102,
      "step": 9680
    },
    {
      "epoch": 0.606398471632816,
      "grad_norm": 0.7483815627521515,
      "learning_rate": 3.541025018272221e-06,
      "loss": 0.3472,
      "step": 9681
    },
    {
      "epoch": 0.6064611096321584,
      "grad_norm": 0.8514469763952743,
      "learning_rate": 3.540054795377286e-06,
      "loss": 0.3899,
      "step": 9682
    },
    {
      "epoch": 0.6065237476315006,
      "grad_norm": 0.8301024584142481,
      "learning_rate": 3.5390846325739668e-06,
      "loss": 0.4417,
      "step": 9683
    },
    {
      "epoch": 0.606586385630843,
      "grad_norm": 0.821005871923224,
      "learning_rate": 3.5381145299021944e-06,
      "loss": 0.4291,
      "step": 9684
    },
    {
      "epoch": 0.6066490236301852,
      "grad_norm": 0.84178640778587,
      "learning_rate": 3.5371444874018995e-06,
      "loss": 0.4027,
      "step": 9685
    },
    {
      "epoch": 0.6067116616295275,
      "grad_norm": 0.741312662342523,
      "learning_rate": 3.5361745051130093e-06,
      "loss": 0.3744,
      "step": 9686
    },
    {
      "epoch": 0.6067742996288699,
      "grad_norm": 0.8310305992765353,
      "learning_rate": 3.535204583075449e-06,
      "loss": 0.4324,
      "step": 9687
    },
    {
      "epoch": 0.6068369376282121,
      "grad_norm": 0.8775238484418909,
      "learning_rate": 3.534234721329138e-06,
      "loss": 0.3909,
      "step": 9688
    },
    {
      "epoch": 0.6068995756275545,
      "grad_norm": 0.8926100381242794,
      "learning_rate": 3.533264919913999e-06,
      "loss": 0.4157,
      "step": 9689
    },
    {
      "epoch": 0.6069622136268967,
      "grad_norm": 0.8708117373108819,
      "learning_rate": 3.532295178869946e-06,
      "loss": 0.4008,
      "step": 9690
    },
    {
      "epoch": 0.6070248516262391,
      "grad_norm": 0.783375585200903,
      "learning_rate": 3.531325498236898e-06,
      "loss": 0.3504,
      "step": 9691
    },
    {
      "epoch": 0.6070874896255813,
      "grad_norm": 0.8302752484203094,
      "learning_rate": 3.5303558780547653e-06,
      "loss": 0.398,
      "step": 9692
    },
    {
      "epoch": 0.6071501276249237,
      "grad_norm": 0.8277044068845071,
      "learning_rate": 3.529386318363458e-06,
      "loss": 0.3979,
      "step": 9693
    },
    {
      "epoch": 0.607212765624266,
      "grad_norm": 0.8066586863539971,
      "learning_rate": 3.528416819202881e-06,
      "loss": 0.3977,
      "step": 9694
    },
    {
      "epoch": 0.6072754036236082,
      "grad_norm": 0.8712630731101312,
      "learning_rate": 3.5274473806129417e-06,
      "loss": 0.4479,
      "step": 9695
    },
    {
      "epoch": 0.6073380416229506,
      "grad_norm": 0.8449047043372487,
      "learning_rate": 3.526478002633542e-06,
      "loss": 0.3775,
      "step": 9696
    },
    {
      "epoch": 0.6074006796222928,
      "grad_norm": 0.8321718492149762,
      "learning_rate": 3.5255086853045804e-06,
      "loss": 0.4157,
      "step": 9697
    },
    {
      "epoch": 0.6074633176216352,
      "grad_norm": 0.8289154611929592,
      "learning_rate": 3.5245394286659557e-06,
      "loss": 0.4036,
      "step": 9698
    },
    {
      "epoch": 0.6075259556209774,
      "grad_norm": 0.8126482990241076,
      "learning_rate": 3.5235702327575617e-06,
      "loss": 0.3979,
      "step": 9699
    },
    {
      "epoch": 0.6075885936203198,
      "grad_norm": 0.901266378226136,
      "learning_rate": 3.522601097619289e-06,
      "loss": 0.4001,
      "step": 9700
    },
    {
      "epoch": 0.6076512316196621,
      "grad_norm": 0.8097399479566938,
      "learning_rate": 3.5216320232910317e-06,
      "loss": 0.3812,
      "step": 9701
    },
    {
      "epoch": 0.6077138696190044,
      "grad_norm": 0.8471664499750191,
      "learning_rate": 3.520663009812674e-06,
      "loss": 0.4252,
      "step": 9702
    },
    {
      "epoch": 0.6077765076183467,
      "grad_norm": 0.779155122475151,
      "learning_rate": 3.519694057224102e-06,
      "loss": 0.3852,
      "step": 9703
    },
    {
      "epoch": 0.6078391456176889,
      "grad_norm": 0.7889769241469368,
      "learning_rate": 3.518725165565197e-06,
      "loss": 0.3958,
      "step": 9704
    },
    {
      "epoch": 0.6079017836170313,
      "grad_norm": 0.8423295741698041,
      "learning_rate": 3.517756334875839e-06,
      "loss": 0.3957,
      "step": 9705
    },
    {
      "epoch": 0.6079644216163735,
      "grad_norm": 0.813358978200895,
      "learning_rate": 3.5167875651959056e-06,
      "loss": 0.3738,
      "step": 9706
    },
    {
      "epoch": 0.6080270596157159,
      "grad_norm": 0.7812316642271944,
      "learning_rate": 3.5158188565652718e-06,
      "loss": 0.3311,
      "step": 9707
    },
    {
      "epoch": 0.6080896976150582,
      "grad_norm": 0.8900615245155519,
      "learning_rate": 3.514850209023809e-06,
      "loss": 0.4282,
      "step": 9708
    },
    {
      "epoch": 0.6081523356144005,
      "grad_norm": 0.9395384790120399,
      "learning_rate": 3.513881622611385e-06,
      "loss": 0.3772,
      "step": 9709
    },
    {
      "epoch": 0.6082149736137428,
      "grad_norm": 0.8054092091032726,
      "learning_rate": 3.5129130973678717e-06,
      "loss": 0.4279,
      "step": 9710
    },
    {
      "epoch": 0.608277611613085,
      "grad_norm": 0.7878079690640529,
      "learning_rate": 3.51194463333313e-06,
      "loss": 0.3728,
      "step": 9711
    },
    {
      "epoch": 0.6083402496124274,
      "grad_norm": 0.9903094437931644,
      "learning_rate": 3.5109762305470248e-06,
      "loss": 0.4343,
      "step": 9712
    },
    {
      "epoch": 0.6084028876117696,
      "grad_norm": 0.6949756107440873,
      "learning_rate": 3.5100078890494138e-06,
      "loss": 0.4598,
      "step": 9713
    },
    {
      "epoch": 0.608465525611112,
      "grad_norm": 0.8115056853472925,
      "learning_rate": 3.5090396088801544e-06,
      "loss": 0.3642,
      "step": 9714
    },
    {
      "epoch": 0.6085281636104543,
      "grad_norm": 0.8034507825936942,
      "learning_rate": 3.508071390079102e-06,
      "loss": 0.4007,
      "step": 9715
    },
    {
      "epoch": 0.6085908016097966,
      "grad_norm": 0.8394936564828512,
      "learning_rate": 3.507103232686108e-06,
      "loss": 0.3964,
      "step": 9716
    },
    {
      "epoch": 0.6086534396091389,
      "grad_norm": 0.8416980489230802,
      "learning_rate": 3.5061351367410223e-06,
      "loss": 0.3744,
      "step": 9717
    },
    {
      "epoch": 0.6087160776084812,
      "grad_norm": 0.8060512105231278,
      "learning_rate": 3.5051671022836893e-06,
      "loss": 0.3544,
      "step": 9718
    },
    {
      "epoch": 0.6087787156078235,
      "grad_norm": 0.793205456931171,
      "learning_rate": 3.504199129353957e-06,
      "loss": 0.3778,
      "step": 9719
    },
    {
      "epoch": 0.6088413536071657,
      "grad_norm": 0.9596707205192185,
      "learning_rate": 3.503231217991667e-06,
      "loss": 0.4431,
      "step": 9720
    },
    {
      "epoch": 0.6089039916065081,
      "grad_norm": 0.7997533123595434,
      "learning_rate": 3.502263368236658e-06,
      "loss": 0.3877,
      "step": 9721
    },
    {
      "epoch": 0.6089666296058504,
      "grad_norm": 0.7958846859366928,
      "learning_rate": 3.5012955801287675e-06,
      "loss": 0.3412,
      "step": 9722
    },
    {
      "epoch": 0.6090292676051927,
      "grad_norm": 0.8387820941012577,
      "learning_rate": 3.5003278537078268e-06,
      "loss": 0.3649,
      "step": 9723
    },
    {
      "epoch": 0.609091905604535,
      "grad_norm": 0.8814341978136442,
      "learning_rate": 3.4993601890136718e-06,
      "loss": 0.397,
      "step": 9724
    },
    {
      "epoch": 0.6091545436038773,
      "grad_norm": 0.7882195860513179,
      "learning_rate": 3.49839258608613e-06,
      "loss": 0.3774,
      "step": 9725
    },
    {
      "epoch": 0.6092171816032196,
      "grad_norm": 0.8018303317534629,
      "learning_rate": 3.4974250449650276e-06,
      "loss": 0.4038,
      "step": 9726
    },
    {
      "epoch": 0.609279819602562,
      "grad_norm": 0.7836940251136513,
      "learning_rate": 3.496457565690188e-06,
      "loss": 0.3601,
      "step": 9727
    },
    {
      "epoch": 0.6093424576019042,
      "grad_norm": 0.8911560696852069,
      "learning_rate": 3.4954901483014337e-06,
      "loss": 0.4104,
      "step": 9728
    },
    {
      "epoch": 0.6094050956012464,
      "grad_norm": 0.796598012138569,
      "learning_rate": 3.494522792838586e-06,
      "loss": 0.334,
      "step": 9729
    },
    {
      "epoch": 0.6094677336005888,
      "grad_norm": 0.827075507452092,
      "learning_rate": 3.49355549934146e-06,
      "loss": 0.3617,
      "step": 9730
    },
    {
      "epoch": 0.6095303715999311,
      "grad_norm": 0.7846067967712754,
      "learning_rate": 3.492588267849869e-06,
      "loss": 0.3688,
      "step": 9731
    },
    {
      "epoch": 0.6095930095992734,
      "grad_norm": 0.8239232324498017,
      "learning_rate": 3.491621098403623e-06,
      "loss": 0.4023,
      "step": 9732
    },
    {
      "epoch": 0.6096556475986157,
      "grad_norm": 0.7741889871765422,
      "learning_rate": 3.490653991042534e-06,
      "loss": 0.3865,
      "step": 9733
    },
    {
      "epoch": 0.609718285597958,
      "grad_norm": 0.8413030309448518,
      "learning_rate": 3.4896869458064073e-06,
      "loss": 0.3691,
      "step": 9734
    },
    {
      "epoch": 0.6097809235973003,
      "grad_norm": 0.8381235624660623,
      "learning_rate": 3.488719962735045e-06,
      "loss": 0.3937,
      "step": 9735
    },
    {
      "epoch": 0.6098435615966427,
      "grad_norm": 0.8722923140820434,
      "learning_rate": 3.487753041868249e-06,
      "loss": 0.3834,
      "step": 9736
    },
    {
      "epoch": 0.6099061995959849,
      "grad_norm": 0.75835529682703,
      "learning_rate": 3.4867861832458193e-06,
      "loss": 0.3982,
      "step": 9737
    },
    {
      "epoch": 0.6099688375953272,
      "grad_norm": 0.8436492909756026,
      "learning_rate": 3.4858193869075517e-06,
      "loss": 0.3969,
      "step": 9738
    },
    {
      "epoch": 0.6100314755946695,
      "grad_norm": 0.83079520168407,
      "learning_rate": 3.484852652893239e-06,
      "loss": 0.3997,
      "step": 9739
    },
    {
      "epoch": 0.6100941135940118,
      "grad_norm": 0.8939683514050346,
      "learning_rate": 3.4838859812426716e-06,
      "loss": 0.4672,
      "step": 9740
    },
    {
      "epoch": 0.6101567515933541,
      "grad_norm": 0.8964803948579952,
      "learning_rate": 3.48291937199564e-06,
      "loss": 0.3674,
      "step": 9741
    },
    {
      "epoch": 0.6102193895926964,
      "grad_norm": 0.8051854408046961,
      "learning_rate": 3.4819528251919277e-06,
      "loss": 0.4205,
      "step": 9742
    },
    {
      "epoch": 0.6102820275920388,
      "grad_norm": 0.8953496634927135,
      "learning_rate": 3.4809863408713197e-06,
      "loss": 0.4459,
      "step": 9743
    },
    {
      "epoch": 0.610344665591381,
      "grad_norm": 0.8976732073609286,
      "learning_rate": 3.4800199190735945e-06,
      "loss": 0.413,
      "step": 9744
    },
    {
      "epoch": 0.6104073035907233,
      "grad_norm": 0.9099482947022447,
      "learning_rate": 3.479053559838532e-06,
      "loss": 0.4087,
      "step": 9745
    },
    {
      "epoch": 0.6104699415900656,
      "grad_norm": 0.8701921808527087,
      "learning_rate": 3.478087263205906e-06,
      "loss": 0.3799,
      "step": 9746
    },
    {
      "epoch": 0.6105325795894079,
      "grad_norm": 0.6090944986162163,
      "learning_rate": 3.4771210292154933e-06,
      "loss": 0.4405,
      "step": 9747
    },
    {
      "epoch": 0.6105952175887502,
      "grad_norm": 0.8407902343623279,
      "learning_rate": 3.4761548579070604e-06,
      "loss": 0.3877,
      "step": 9748
    },
    {
      "epoch": 0.6106578555880925,
      "grad_norm": 0.7980849310418586,
      "learning_rate": 3.475188749320376e-06,
      "loss": 0.4005,
      "step": 9749
    },
    {
      "epoch": 0.6107204935874349,
      "grad_norm": 0.8463628861542035,
      "learning_rate": 3.474222703495207e-06,
      "loss": 0.4081,
      "step": 9750
    },
    {
      "epoch": 0.6107831315867771,
      "grad_norm": 0.809815687031824,
      "learning_rate": 3.473256720471314e-06,
      "loss": 0.3634,
      "step": 9751
    },
    {
      "epoch": 0.6108457695861195,
      "grad_norm": 0.8405927264955733,
      "learning_rate": 3.4722908002884586e-06,
      "loss": 0.4007,
      "step": 9752
    },
    {
      "epoch": 0.6109084075854617,
      "grad_norm": 0.9073940825472115,
      "learning_rate": 3.471324942986396e-06,
      "loss": 0.4463,
      "step": 9753
    },
    {
      "epoch": 0.610971045584804,
      "grad_norm": 0.7741343752390233,
      "learning_rate": 3.470359148604884e-06,
      "loss": 0.383,
      "step": 9754
    },
    {
      "epoch": 0.6110336835841463,
      "grad_norm": 0.9289339404935111,
      "learning_rate": 3.46939341718367e-06,
      "loss": 0.4107,
      "step": 9755
    },
    {
      "epoch": 0.6110963215834886,
      "grad_norm": 0.7692906162614732,
      "learning_rate": 3.46842774876251e-06,
      "loss": 0.3656,
      "step": 9756
    },
    {
      "epoch": 0.611158959582831,
      "grad_norm": 0.7994154726786582,
      "learning_rate": 3.4674621433811486e-06,
      "loss": 0.352,
      "step": 9757
    },
    {
      "epoch": 0.6112215975821732,
      "grad_norm": 0.8514407688802049,
      "learning_rate": 3.4664966010793276e-06,
      "loss": 0.3551,
      "step": 9758
    },
    {
      "epoch": 0.6112842355815156,
      "grad_norm": 0.6383338215697982,
      "learning_rate": 3.465531121896793e-06,
      "loss": 0.4681,
      "step": 9759
    },
    {
      "epoch": 0.6113468735808578,
      "grad_norm": 0.9173296376843815,
      "learning_rate": 3.464565705873282e-06,
      "loss": 0.4176,
      "step": 9760
    },
    {
      "epoch": 0.6114095115802002,
      "grad_norm": 0.772579411865578,
      "learning_rate": 3.463600353048531e-06,
      "loss": 0.377,
      "step": 9761
    },
    {
      "epoch": 0.6114721495795424,
      "grad_norm": 0.8112864470059851,
      "learning_rate": 3.4626350634622745e-06,
      "loss": 0.4,
      "step": 9762
    },
    {
      "epoch": 0.6115347875788847,
      "grad_norm": 0.8270597228309834,
      "learning_rate": 3.461669837154245e-06,
      "loss": 0.3935,
      "step": 9763
    },
    {
      "epoch": 0.611597425578227,
      "grad_norm": 0.8346252141836049,
      "learning_rate": 3.460704674164168e-06,
      "loss": 0.3841,
      "step": 9764
    },
    {
      "epoch": 0.6116600635775693,
      "grad_norm": 0.861275895043666,
      "learning_rate": 3.459739574531775e-06,
      "loss": 0.3823,
      "step": 9765
    },
    {
      "epoch": 0.6117227015769117,
      "grad_norm": 0.6054865180271968,
      "learning_rate": 3.458774538296785e-06,
      "loss": 0.4659,
      "step": 9766
    },
    {
      "epoch": 0.6117853395762539,
      "grad_norm": 0.9211374929087534,
      "learning_rate": 3.457809565498923e-06,
      "loss": 0.4127,
      "step": 9767
    },
    {
      "epoch": 0.6118479775755963,
      "grad_norm": 0.8506401059648963,
      "learning_rate": 3.4568446561779056e-06,
      "loss": 0.4438,
      "step": 9768
    },
    {
      "epoch": 0.6119106155749385,
      "grad_norm": 0.7798453847900142,
      "learning_rate": 3.455879810373448e-06,
      "loss": 0.3523,
      "step": 9769
    },
    {
      "epoch": 0.6119732535742808,
      "grad_norm": 0.823302625321148,
      "learning_rate": 3.4549150281252635e-06,
      "loss": 0.4141,
      "step": 9770
    },
    {
      "epoch": 0.6120358915736231,
      "grad_norm": 0.8569296759908261,
      "learning_rate": 3.453950309473064e-06,
      "loss": 0.4123,
      "step": 9771
    },
    {
      "epoch": 0.6120985295729654,
      "grad_norm": 0.8316559765606508,
      "learning_rate": 3.4529856544565563e-06,
      "loss": 0.3956,
      "step": 9772
    },
    {
      "epoch": 0.6121611675723078,
      "grad_norm": 0.7800614913423736,
      "learning_rate": 3.452021063115445e-06,
      "loss": 0.3948,
      "step": 9773
    },
    {
      "epoch": 0.61222380557165,
      "grad_norm": 0.768378489870306,
      "learning_rate": 3.4510565354894354e-06,
      "loss": 0.3969,
      "step": 9774
    },
    {
      "epoch": 0.6122864435709924,
      "grad_norm": 0.8153828900095746,
      "learning_rate": 3.4500920716182256e-06,
      "loss": 0.4374,
      "step": 9775
    },
    {
      "epoch": 0.6123490815703346,
      "grad_norm": 0.818873972060771,
      "learning_rate": 3.4491276715415143e-06,
      "loss": 0.4021,
      "step": 9776
    },
    {
      "epoch": 0.612411719569677,
      "grad_norm": 0.8403618231351219,
      "learning_rate": 3.4481633352989967e-06,
      "loss": 0.405,
      "step": 9777
    },
    {
      "epoch": 0.6124743575690192,
      "grad_norm": 0.7867535802352226,
      "learning_rate": 3.447199062930363e-06,
      "loss": 0.3726,
      "step": 9778
    },
    {
      "epoch": 0.6125369955683615,
      "grad_norm": 0.8195034074801414,
      "learning_rate": 3.4462348544753055e-06,
      "loss": 0.3938,
      "step": 9779
    },
    {
      "epoch": 0.6125996335677039,
      "grad_norm": 0.8191474642654354,
      "learning_rate": 3.4452707099735087e-06,
      "loss": 0.3793,
      "step": 9780
    },
    {
      "epoch": 0.6126622715670461,
      "grad_norm": 0.8160393267360601,
      "learning_rate": 3.4443066294646586e-06,
      "loss": 0.3821,
      "step": 9781
    },
    {
      "epoch": 0.6127249095663885,
      "grad_norm": 0.8263534860563171,
      "learning_rate": 3.4433426129884346e-06,
      "loss": 0.3805,
      "step": 9782
    },
    {
      "epoch": 0.6127875475657307,
      "grad_norm": 0.7435804736681602,
      "learning_rate": 3.4423786605845196e-06,
      "loss": 0.3274,
      "step": 9783
    },
    {
      "epoch": 0.6128501855650731,
      "grad_norm": 0.8565792561244718,
      "learning_rate": 3.4414147722925868e-06,
      "loss": 0.3921,
      "step": 9784
    },
    {
      "epoch": 0.6129128235644153,
      "grad_norm": 0.7861873231027915,
      "learning_rate": 3.4404509481523116e-06,
      "loss": 0.3331,
      "step": 9785
    },
    {
      "epoch": 0.6129754615637577,
      "grad_norm": 0.9022709359352509,
      "learning_rate": 3.439487188203366e-06,
      "loss": 0.4186,
      "step": 9786
    },
    {
      "epoch": 0.6130380995631,
      "grad_norm": 0.8605052890970081,
      "learning_rate": 3.4385234924854162e-06,
      "loss": 0.35,
      "step": 9787
    },
    {
      "epoch": 0.6131007375624422,
      "grad_norm": 0.8799264511403584,
      "learning_rate": 3.43755986103813e-06,
      "loss": 0.3995,
      "step": 9788
    },
    {
      "epoch": 0.6131633755617846,
      "grad_norm": 0.8117983528180653,
      "learning_rate": 3.43659629390117e-06,
      "loss": 0.3752,
      "step": 9789
    },
    {
      "epoch": 0.6132260135611268,
      "grad_norm": 0.8338358434617744,
      "learning_rate": 3.4356327911141964e-06,
      "loss": 0.3622,
      "step": 9790
    },
    {
      "epoch": 0.6132886515604692,
      "grad_norm": 0.8434852293544977,
      "learning_rate": 3.434669352716866e-06,
      "loss": 0.3838,
      "step": 9791
    },
    {
      "epoch": 0.6133512895598114,
      "grad_norm": 0.8421216084795262,
      "learning_rate": 3.4337059787488357e-06,
      "loss": 0.3999,
      "step": 9792
    },
    {
      "epoch": 0.6134139275591538,
      "grad_norm": 0.748796404298811,
      "learning_rate": 3.4327426692497594e-06,
      "loss": 0.3896,
      "step": 9793
    },
    {
      "epoch": 0.6134765655584961,
      "grad_norm": 0.9601961324964248,
      "learning_rate": 3.431779424259286e-06,
      "loss": 0.4153,
      "step": 9794
    },
    {
      "epoch": 0.6135392035578383,
      "grad_norm": 0.8540534781725708,
      "learning_rate": 3.430816243817063e-06,
      "loss": 0.3949,
      "step": 9795
    },
    {
      "epoch": 0.6136018415571807,
      "grad_norm": 0.8832530690395711,
      "learning_rate": 3.429853127962733e-06,
      "loss": 0.4061,
      "step": 9796
    },
    {
      "epoch": 0.6136644795565229,
      "grad_norm": 0.885884383925136,
      "learning_rate": 3.4288900767359407e-06,
      "loss": 0.4101,
      "step": 9797
    },
    {
      "epoch": 0.6137271175558653,
      "grad_norm": 0.8450733179760356,
      "learning_rate": 3.4279270901763244e-06,
      "loss": 0.3763,
      "step": 9798
    },
    {
      "epoch": 0.6137897555552075,
      "grad_norm": 0.8476878656920313,
      "learning_rate": 3.42696416832352e-06,
      "loss": 0.4205,
      "step": 9799
    },
    {
      "epoch": 0.6138523935545499,
      "grad_norm": 0.6192656166685508,
      "learning_rate": 3.4260013112171637e-06,
      "loss": 0.4575,
      "step": 9800
    },
    {
      "epoch": 0.6139150315538922,
      "grad_norm": 0.813232081996365,
      "learning_rate": 3.4250385188968834e-06,
      "loss": 0.3875,
      "step": 9801
    },
    {
      "epoch": 0.6139776695532345,
      "grad_norm": 0.8972722056870552,
      "learning_rate": 3.4240757914023113e-06,
      "loss": 0.4396,
      "step": 9802
    },
    {
      "epoch": 0.6140403075525768,
      "grad_norm": 0.8330601475518499,
      "learning_rate": 3.4231131287730735e-06,
      "loss": 0.3695,
      "step": 9803
    },
    {
      "epoch": 0.614102945551919,
      "grad_norm": 0.9572819233708229,
      "learning_rate": 3.42215053104879e-06,
      "loss": 0.4265,
      "step": 9804
    },
    {
      "epoch": 0.6141655835512614,
      "grad_norm": 0.8455706801053025,
      "learning_rate": 3.421187998269085e-06,
      "loss": 0.3605,
      "step": 9805
    },
    {
      "epoch": 0.6142282215506036,
      "grad_norm": 0.850183752106454,
      "learning_rate": 3.4202255304735744e-06,
      "loss": 0.3847,
      "step": 9806
    },
    {
      "epoch": 0.614290859549946,
      "grad_norm": 0.839822619250771,
      "learning_rate": 3.419263127701875e-06,
      "loss": 0.3652,
      "step": 9807
    },
    {
      "epoch": 0.6143534975492883,
      "grad_norm": 0.8893791385669508,
      "learning_rate": 3.4183007899935973e-06,
      "loss": 0.373,
      "step": 9808
    },
    {
      "epoch": 0.6144161355486306,
      "grad_norm": 0.7949443255504632,
      "learning_rate": 3.4173385173883533e-06,
      "loss": 0.3932,
      "step": 9809
    },
    {
      "epoch": 0.6144787735479729,
      "grad_norm": 0.8698944418121313,
      "learning_rate": 3.416376309925748e-06,
      "loss": 0.3688,
      "step": 9810
    },
    {
      "epoch": 0.6145414115473152,
      "grad_norm": 0.8660315789379274,
      "learning_rate": 3.41541416764539e-06,
      "loss": 0.3976,
      "step": 9811
    },
    {
      "epoch": 0.6146040495466575,
      "grad_norm": 0.928227206441588,
      "learning_rate": 3.4144520905868784e-06,
      "loss": 0.4751,
      "step": 9812
    },
    {
      "epoch": 0.6146666875459997,
      "grad_norm": 0.897221648515523,
      "learning_rate": 3.4134900787898123e-06,
      "loss": 0.3865,
      "step": 9813
    },
    {
      "epoch": 0.6147293255453421,
      "grad_norm": 0.7725722913758315,
      "learning_rate": 3.4125281322937903e-06,
      "loss": 0.3469,
      "step": 9814
    },
    {
      "epoch": 0.6147919635446844,
      "grad_norm": 0.8474026198136682,
      "learning_rate": 3.411566251138405e-06,
      "loss": 0.3785,
      "step": 9815
    },
    {
      "epoch": 0.6148546015440267,
      "grad_norm": 0.8273386137884132,
      "learning_rate": 3.4106044353632472e-06,
      "loss": 0.3655,
      "step": 9816
    },
    {
      "epoch": 0.614917239543369,
      "grad_norm": 0.8414422434669443,
      "learning_rate": 3.409642685007905e-06,
      "loss": 0.4147,
      "step": 9817
    },
    {
      "epoch": 0.6149798775427113,
      "grad_norm": 0.8101820259974948,
      "learning_rate": 3.4086810001119665e-06,
      "loss": 0.385,
      "step": 9818
    },
    {
      "epoch": 0.6150425155420536,
      "grad_norm": 0.6749869991724371,
      "learning_rate": 3.4077193807150106e-06,
      "loss": 0.4567,
      "step": 9819
    },
    {
      "epoch": 0.6151051535413958,
      "grad_norm": 0.8689103163520406,
      "learning_rate": 3.4067578268566225e-06,
      "loss": 0.3922,
      "step": 9820
    },
    {
      "epoch": 0.6151677915407382,
      "grad_norm": 0.9591017886200244,
      "learning_rate": 3.4057963385763794e-06,
      "loss": 0.4403,
      "step": 9821
    },
    {
      "epoch": 0.6152304295400804,
      "grad_norm": 0.8271424137490421,
      "learning_rate": 3.4048349159138525e-06,
      "loss": 0.391,
      "step": 9822
    },
    {
      "epoch": 0.6152930675394228,
      "grad_norm": 0.8195170470142639,
      "learning_rate": 3.403873558908618e-06,
      "loss": 0.3856,
      "step": 9823
    },
    {
      "epoch": 0.6153557055387651,
      "grad_norm": 0.8742728750940196,
      "learning_rate": 3.4029122676002433e-06,
      "loss": 0.424,
      "step": 9824
    },
    {
      "epoch": 0.6154183435381074,
      "grad_norm": 0.8263266881430765,
      "learning_rate": 3.4019510420282963e-06,
      "loss": 0.3658,
      "step": 9825
    },
    {
      "epoch": 0.6154809815374497,
      "grad_norm": 0.8543237134524705,
      "learning_rate": 3.4009898822323416e-06,
      "loss": 0.3842,
      "step": 9826
    },
    {
      "epoch": 0.615543619536792,
      "grad_norm": 0.8503644826461033,
      "learning_rate": 3.40002878825194e-06,
      "loss": 0.4107,
      "step": 9827
    },
    {
      "epoch": 0.6156062575361343,
      "grad_norm": 0.8186109639502599,
      "learning_rate": 3.3990677601266485e-06,
      "loss": 0.387,
      "step": 9828
    },
    {
      "epoch": 0.6156688955354765,
      "grad_norm": 0.8431883508435931,
      "learning_rate": 3.398106797896027e-06,
      "loss": 0.4021,
      "step": 9829
    },
    {
      "epoch": 0.6157315335348189,
      "grad_norm": 0.7914854700200612,
      "learning_rate": 3.3971459015996265e-06,
      "loss": 0.3888,
      "step": 9830
    },
    {
      "epoch": 0.6157941715341612,
      "grad_norm": 0.8638886214791687,
      "learning_rate": 3.3961850712769988e-06,
      "loss": 0.3768,
      "step": 9831
    },
    {
      "epoch": 0.6158568095335035,
      "grad_norm": 0.8657727411900771,
      "learning_rate": 3.3952243069676915e-06,
      "loss": 0.3773,
      "step": 9832
    },
    {
      "epoch": 0.6159194475328458,
      "grad_norm": 0.8291463856658744,
      "learning_rate": 3.394263608711249e-06,
      "loss": 0.3811,
      "step": 9833
    },
    {
      "epoch": 0.6159820855321881,
      "grad_norm": 0.7933157196735786,
      "learning_rate": 3.393302976547214e-06,
      "loss": 0.3631,
      "step": 9834
    },
    {
      "epoch": 0.6160447235315304,
      "grad_norm": 0.8632717677115246,
      "learning_rate": 3.392342410515128e-06,
      "loss": 0.4322,
      "step": 9835
    },
    {
      "epoch": 0.6161073615308728,
      "grad_norm": 0.9330233290942278,
      "learning_rate": 3.3913819106545266e-06,
      "loss": 0.3922,
      "step": 9836
    },
    {
      "epoch": 0.616169999530215,
      "grad_norm": 0.8143436198484681,
      "learning_rate": 3.3904214770049438e-06,
      "loss": 0.3739,
      "step": 9837
    },
    {
      "epoch": 0.6162326375295573,
      "grad_norm": 0.8653744572845068,
      "learning_rate": 3.3894611096059095e-06,
      "loss": 0.3887,
      "step": 9838
    },
    {
      "epoch": 0.6162952755288996,
      "grad_norm": 0.8883004829777138,
      "learning_rate": 3.388500808496956e-06,
      "loss": 0.3915,
      "step": 9839
    },
    {
      "epoch": 0.6163579135282419,
      "grad_norm": 0.8001032084514366,
      "learning_rate": 3.3875405737176104e-06,
      "loss": 0.3631,
      "step": 9840
    },
    {
      "epoch": 0.6164205515275842,
      "grad_norm": 0.8573803294308272,
      "learning_rate": 3.3865804053073936e-06,
      "loss": 0.4069,
      "step": 9841
    },
    {
      "epoch": 0.6164831895269265,
      "grad_norm": 0.7910096375913158,
      "learning_rate": 3.3856203033058266e-06,
      "loss": 0.4124,
      "step": 9842
    },
    {
      "epoch": 0.6165458275262689,
      "grad_norm": 0.8369616395904045,
      "learning_rate": 3.384660267752426e-06,
      "loss": 0.4107,
      "step": 9843
    },
    {
      "epoch": 0.6166084655256111,
      "grad_norm": 0.8370660730477724,
      "learning_rate": 3.3837002986867103e-06,
      "loss": 0.3986,
      "step": 9844
    },
    {
      "epoch": 0.6166711035249535,
      "grad_norm": 0.8574289878464181,
      "learning_rate": 3.38274039614819e-06,
      "loss": 0.4094,
      "step": 9845
    },
    {
      "epoch": 0.6167337415242957,
      "grad_norm": 0.7461478505018856,
      "learning_rate": 3.381780560176374e-06,
      "loss": 0.3653,
      "step": 9846
    },
    {
      "epoch": 0.616796379523638,
      "grad_norm": 0.8200672930525423,
      "learning_rate": 3.380820790810771e-06,
      "loss": 0.3734,
      "step": 9847
    },
    {
      "epoch": 0.6168590175229803,
      "grad_norm": 0.8265763979899611,
      "learning_rate": 3.3798610880908844e-06,
      "loss": 0.3842,
      "step": 9848
    },
    {
      "epoch": 0.6169216555223226,
      "grad_norm": 0.8405813155367117,
      "learning_rate": 3.3789014520562173e-06,
      "loss": 0.4053,
      "step": 9849
    },
    {
      "epoch": 0.616984293521665,
      "grad_norm": 0.8772164237560036,
      "learning_rate": 3.3779418827462673e-06,
      "loss": 0.3836,
      "step": 9850
    },
    {
      "epoch": 0.6170469315210072,
      "grad_norm": 0.8184853950065766,
      "learning_rate": 3.3769823802005296e-06,
      "loss": 0.3788,
      "step": 9851
    },
    {
      "epoch": 0.6171095695203496,
      "grad_norm": 0.8256907054701123,
      "learning_rate": 3.376022944458499e-06,
      "loss": 0.3726,
      "step": 9852
    },
    {
      "epoch": 0.6171722075196918,
      "grad_norm": 0.8819572439533047,
      "learning_rate": 3.3750635755596662e-06,
      "loss": 0.4013,
      "step": 9853
    },
    {
      "epoch": 0.6172348455190341,
      "grad_norm": 0.8161390232724338,
      "learning_rate": 3.3741042735435175e-06,
      "loss": 0.4034,
      "step": 9854
    },
    {
      "epoch": 0.6172974835183764,
      "grad_norm": 0.8990926843818193,
      "learning_rate": 3.373145038449538e-06,
      "loss": 0.4158,
      "step": 9855
    },
    {
      "epoch": 0.6173601215177187,
      "grad_norm": 0.845699986176942,
      "learning_rate": 3.3721858703172095e-06,
      "loss": 0.3974,
      "step": 9856
    },
    {
      "epoch": 0.617422759517061,
      "grad_norm": 0.8967685593296006,
      "learning_rate": 3.3712267691860144e-06,
      "loss": 0.4273,
      "step": 9857
    },
    {
      "epoch": 0.6174853975164033,
      "grad_norm": 0.858282957107384,
      "learning_rate": 3.370267735095428e-06,
      "loss": 0.41,
      "step": 9858
    },
    {
      "epoch": 0.6175480355157457,
      "grad_norm": 0.7982285223854865,
      "learning_rate": 3.3693087680849245e-06,
      "loss": 0.3713,
      "step": 9859
    },
    {
      "epoch": 0.6176106735150879,
      "grad_norm": 0.8446681725164991,
      "learning_rate": 3.368349868193974e-06,
      "loss": 0.3636,
      "step": 9860
    },
    {
      "epoch": 0.6176733115144303,
      "grad_norm": 0.8486723909698393,
      "learning_rate": 3.3673910354620467e-06,
      "loss": 0.3955,
      "step": 9861
    },
    {
      "epoch": 0.6177359495137725,
      "grad_norm": 0.8192960363856093,
      "learning_rate": 3.3664322699286066e-06,
      "loss": 0.3668,
      "step": 9862
    },
    {
      "epoch": 0.6177985875131148,
      "grad_norm": 0.7973238609798827,
      "learning_rate": 3.365473571633117e-06,
      "loss": 0.3725,
      "step": 9863
    },
    {
      "epoch": 0.6178612255124571,
      "grad_norm": 0.8399492613833552,
      "learning_rate": 3.3645149406150396e-06,
      "loss": 0.4073,
      "step": 9864
    },
    {
      "epoch": 0.6179238635117994,
      "grad_norm": 0.8213249134769567,
      "learning_rate": 3.363556376913829e-06,
      "loss": 0.3762,
      "step": 9865
    },
    {
      "epoch": 0.6179865015111418,
      "grad_norm": 0.8162755310062095,
      "learning_rate": 3.362597880568943e-06,
      "loss": 0.385,
      "step": 9866
    },
    {
      "epoch": 0.618049139510484,
      "grad_norm": 0.7960218024942821,
      "learning_rate": 3.3616394516198327e-06,
      "loss": 0.3862,
      "step": 9867
    },
    {
      "epoch": 0.6181117775098264,
      "grad_norm": 0.8536182542037762,
      "learning_rate": 3.360681090105946e-06,
      "loss": 0.3762,
      "step": 9868
    },
    {
      "epoch": 0.6181744155091686,
      "grad_norm": 0.6395996224597477,
      "learning_rate": 3.35972279606673e-06,
      "loss": 0.4411,
      "step": 9869
    },
    {
      "epoch": 0.618237053508511,
      "grad_norm": 0.6226802049427427,
      "learning_rate": 3.358764569541628e-06,
      "loss": 0.4721,
      "step": 9870
    },
    {
      "epoch": 0.6182996915078532,
      "grad_norm": 0.9139862678085126,
      "learning_rate": 3.357806410570082e-06,
      "loss": 0.4363,
      "step": 9871
    },
    {
      "epoch": 0.6183623295071955,
      "grad_norm": 0.9574181443844422,
      "learning_rate": 3.3568483191915275e-06,
      "loss": 0.4384,
      "step": 9872
    },
    {
      "epoch": 0.6184249675065379,
      "grad_norm": 0.8833154305570581,
      "learning_rate": 3.355890295445402e-06,
      "loss": 0.3622,
      "step": 9873
    },
    {
      "epoch": 0.6184876055058801,
      "grad_norm": 0.9316008924005139,
      "learning_rate": 3.3549323393711376e-06,
      "loss": 0.4401,
      "step": 9874
    },
    {
      "epoch": 0.6185502435052225,
      "grad_norm": 0.8546987274602134,
      "learning_rate": 3.3539744510081605e-06,
      "loss": 0.3805,
      "step": 9875
    },
    {
      "epoch": 0.6186128815045647,
      "grad_norm": 0.8425493027908019,
      "learning_rate": 3.3530166303959035e-06,
      "loss": 0.4087,
      "step": 9876
    },
    {
      "epoch": 0.6186755195039071,
      "grad_norm": 0.8738511495886717,
      "learning_rate": 3.3520588775737863e-06,
      "loss": 0.3778,
      "step": 9877
    },
    {
      "epoch": 0.6187381575032493,
      "grad_norm": 0.7576205690570722,
      "learning_rate": 3.351101192581232e-06,
      "loss": 0.3642,
      "step": 9878
    },
    {
      "epoch": 0.6188007955025916,
      "grad_norm": 0.8685669070624663,
      "learning_rate": 3.3501435754576596e-06,
      "loss": 0.4387,
      "step": 9879
    },
    {
      "epoch": 0.618863433501934,
      "grad_norm": 0.9078648775024661,
      "learning_rate": 3.3491860262424835e-06,
      "loss": 0.4138,
      "step": 9880
    },
    {
      "epoch": 0.6189260715012762,
      "grad_norm": 0.8404686276862937,
      "learning_rate": 3.3482285449751164e-06,
      "loss": 0.3849,
      "step": 9881
    },
    {
      "epoch": 0.6189887095006186,
      "grad_norm": 0.8435397974269538,
      "learning_rate": 3.347271131694969e-06,
      "loss": 0.4142,
      "step": 9882
    },
    {
      "epoch": 0.6190513474999608,
      "grad_norm": 0.8141369389217802,
      "learning_rate": 3.3463137864414493e-06,
      "loss": 0.4115,
      "step": 9883
    },
    {
      "epoch": 0.6191139854993032,
      "grad_norm": 0.7876462654840778,
      "learning_rate": 3.3453565092539586e-06,
      "loss": 0.3589,
      "step": 9884
    },
    {
      "epoch": 0.6191766234986454,
      "grad_norm": 0.8612849686777218,
      "learning_rate": 3.3443993001719033e-06,
      "loss": 0.3755,
      "step": 9885
    },
    {
      "epoch": 0.6192392614979878,
      "grad_norm": 0.8848200634263593,
      "learning_rate": 3.3434421592346787e-06,
      "loss": 0.4262,
      "step": 9886
    },
    {
      "epoch": 0.61930189949733,
      "grad_norm": 0.8326777078869274,
      "learning_rate": 3.3424850864816844e-06,
      "loss": 0.396,
      "step": 9887
    },
    {
      "epoch": 0.6193645374966723,
      "grad_norm": 0.8525793608778848,
      "learning_rate": 3.341528081952311e-06,
      "loss": 0.3877,
      "step": 9888
    },
    {
      "epoch": 0.6194271754960147,
      "grad_norm": 0.9076697227494595,
      "learning_rate": 3.3405711456859485e-06,
      "loss": 0.4054,
      "step": 9889
    },
    {
      "epoch": 0.6194898134953569,
      "grad_norm": 0.8361560390713397,
      "learning_rate": 3.3396142777219865e-06,
      "loss": 0.3594,
      "step": 9890
    },
    {
      "epoch": 0.6195524514946993,
      "grad_norm": 0.8121466919961073,
      "learning_rate": 3.3386574780998093e-06,
      "loss": 0.3324,
      "step": 9891
    },
    {
      "epoch": 0.6196150894940415,
      "grad_norm": 0.6002024281240804,
      "learning_rate": 3.3377007468587984e-06,
      "loss": 0.4407,
      "step": 9892
    },
    {
      "epoch": 0.6196777274933839,
      "grad_norm": 1.0469240291032438,
      "learning_rate": 3.336744084038331e-06,
      "loss": 0.3911,
      "step": 9893
    },
    {
      "epoch": 0.6197403654927262,
      "grad_norm": 0.8298508992020505,
      "learning_rate": 3.3357874896777887e-06,
      "loss": 0.4004,
      "step": 9894
    },
    {
      "epoch": 0.6198030034920685,
      "grad_norm": 0.867944761988687,
      "learning_rate": 3.334830963816541e-06,
      "loss": 0.3982,
      "step": 9895
    },
    {
      "epoch": 0.6198656414914108,
      "grad_norm": 0.6739956677249181,
      "learning_rate": 3.3338745064939604e-06,
      "loss": 0.4718,
      "step": 9896
    },
    {
      "epoch": 0.619928279490753,
      "grad_norm": 0.9056997535597704,
      "learning_rate": 3.3329181177494152e-06,
      "loss": 0.4035,
      "step": 9897
    },
    {
      "epoch": 0.6199909174900954,
      "grad_norm": 0.8248899974567109,
      "learning_rate": 3.331961797622268e-06,
      "loss": 0.3883,
      "step": 9898
    },
    {
      "epoch": 0.6200535554894376,
      "grad_norm": 0.8507858081009424,
      "learning_rate": 3.3310055461518844e-06,
      "loss": 0.3995,
      "step": 9899
    },
    {
      "epoch": 0.62011619348878,
      "grad_norm": 0.7582421280198478,
      "learning_rate": 3.330049363377621e-06,
      "loss": 0.3758,
      "step": 9900
    },
    {
      "epoch": 0.6201788314881223,
      "grad_norm": 0.8200717536133159,
      "learning_rate": 3.3290932493388373e-06,
      "loss": 0.3568,
      "step": 9901
    },
    {
      "epoch": 0.6202414694874646,
      "grad_norm": 0.8375968243119839,
      "learning_rate": 3.3281372040748826e-06,
      "loss": 0.4157,
      "step": 9902
    },
    {
      "epoch": 0.6203041074868069,
      "grad_norm": 0.8299064769994459,
      "learning_rate": 3.327181227625112e-06,
      "loss": 0.4091,
      "step": 9903
    },
    {
      "epoch": 0.6203667454861491,
      "grad_norm": 0.7782152287015951,
      "learning_rate": 3.326225320028874e-06,
      "loss": 0.4155,
      "step": 9904
    },
    {
      "epoch": 0.6204293834854915,
      "grad_norm": 0.8648267533317593,
      "learning_rate": 3.325269481325512e-06,
      "loss": 0.3739,
      "step": 9905
    },
    {
      "epoch": 0.6204920214848337,
      "grad_norm": 0.8009629785552789,
      "learning_rate": 3.3243137115543695e-06,
      "loss": 0.3819,
      "step": 9906
    },
    {
      "epoch": 0.6205546594841761,
      "grad_norm": 0.8224993986157951,
      "learning_rate": 3.323358010754784e-06,
      "loss": 0.4102,
      "step": 9907
    },
    {
      "epoch": 0.6206172974835183,
      "grad_norm": 0.8990799109399378,
      "learning_rate": 3.3224023789660954e-06,
      "loss": 0.3962,
      "step": 9908
    },
    {
      "epoch": 0.6206799354828607,
      "grad_norm": 0.9317754570969504,
      "learning_rate": 3.3214468162276364e-06,
      "loss": 0.4368,
      "step": 9909
    },
    {
      "epoch": 0.620742573482203,
      "grad_norm": 0.8845314067880615,
      "learning_rate": 3.3204913225787373e-06,
      "loss": 0.4181,
      "step": 9910
    },
    {
      "epoch": 0.6208052114815453,
      "grad_norm": 0.8509462147307928,
      "learning_rate": 3.3195358980587266e-06,
      "loss": 0.4081,
      "step": 9911
    },
    {
      "epoch": 0.6208678494808876,
      "grad_norm": 0.8647013022615758,
      "learning_rate": 3.3185805427069297e-06,
      "loss": 0.4085,
      "step": 9912
    },
    {
      "epoch": 0.6209304874802298,
      "grad_norm": 0.9032703720767141,
      "learning_rate": 3.3176252565626716e-06,
      "loss": 0.3753,
      "step": 9913
    },
    {
      "epoch": 0.6209931254795722,
      "grad_norm": 0.7801637228323545,
      "learning_rate": 3.3166700396652696e-06,
      "loss": 0.3826,
      "step": 9914
    },
    {
      "epoch": 0.6210557634789144,
      "grad_norm": 0.80946870756844,
      "learning_rate": 3.3157148920540404e-06,
      "loss": 0.4132,
      "step": 9915
    },
    {
      "epoch": 0.6211184014782568,
      "grad_norm": 0.819089550133005,
      "learning_rate": 3.3147598137683e-06,
      "loss": 0.4148,
      "step": 9916
    },
    {
      "epoch": 0.6211810394775991,
      "grad_norm": 0.9011478000663169,
      "learning_rate": 3.313804804847359e-06,
      "loss": 0.4302,
      "step": 9917
    },
    {
      "epoch": 0.6212436774769414,
      "grad_norm": 0.7973767786695912,
      "learning_rate": 3.3128498653305248e-06,
      "loss": 0.3596,
      "step": 9918
    },
    {
      "epoch": 0.6213063154762837,
      "grad_norm": 0.8391138271437536,
      "learning_rate": 3.311894995257102e-06,
      "loss": 0.3737,
      "step": 9919
    },
    {
      "epoch": 0.621368953475626,
      "grad_norm": 0.835202713029005,
      "learning_rate": 3.3109401946663954e-06,
      "loss": 0.4093,
      "step": 9920
    },
    {
      "epoch": 0.6214315914749683,
      "grad_norm": 0.8575829583422815,
      "learning_rate": 3.3099854635977024e-06,
      "loss": 0.4477,
      "step": 9921
    },
    {
      "epoch": 0.6214942294743105,
      "grad_norm": 0.8803012010216477,
      "learning_rate": 3.3090308020903226e-06,
      "loss": 0.3879,
      "step": 9922
    },
    {
      "epoch": 0.6215568674736529,
      "grad_norm": 0.7033472755183396,
      "learning_rate": 3.30807621018355e-06,
      "loss": 0.4683,
      "step": 9923
    },
    {
      "epoch": 0.6216195054729952,
      "grad_norm": 0.8030275886443429,
      "learning_rate": 3.3071216879166727e-06,
      "loss": 0.3959,
      "step": 9924
    },
    {
      "epoch": 0.6216821434723375,
      "grad_norm": 0.8719853820186455,
      "learning_rate": 3.3061672353289826e-06,
      "loss": 0.4067,
      "step": 9925
    },
    {
      "epoch": 0.6217447814716798,
      "grad_norm": 0.849615826305834,
      "learning_rate": 3.3052128524597633e-06,
      "loss": 0.3997,
      "step": 9926
    },
    {
      "epoch": 0.6218074194710221,
      "grad_norm": 0.8684141413183983,
      "learning_rate": 3.3042585393482973e-06,
      "loss": 0.4033,
      "step": 9927
    },
    {
      "epoch": 0.6218700574703644,
      "grad_norm": 0.8313589052574375,
      "learning_rate": 3.3033042960338636e-06,
      "loss": 0.4102,
      "step": 9928
    },
    {
      "epoch": 0.6219326954697066,
      "grad_norm": 0.8157447087843034,
      "learning_rate": 3.302350122555741e-06,
      "loss": 0.4171,
      "step": 9929
    },
    {
      "epoch": 0.621995333469049,
      "grad_norm": 0.7554451495436163,
      "learning_rate": 3.301396018953201e-06,
      "loss": 0.3575,
      "step": 9930
    },
    {
      "epoch": 0.6220579714683913,
      "grad_norm": 0.8746985056419716,
      "learning_rate": 3.3004419852655172e-06,
      "loss": 0.3866,
      "step": 9931
    },
    {
      "epoch": 0.6221206094677336,
      "grad_norm": 0.7733269984661096,
      "learning_rate": 3.2994880215319574e-06,
      "loss": 0.3675,
      "step": 9932
    },
    {
      "epoch": 0.6221832474670759,
      "grad_norm": 0.8529501986640472,
      "learning_rate": 3.298534127791785e-06,
      "loss": 0.3632,
      "step": 9933
    },
    {
      "epoch": 0.6222458854664182,
      "grad_norm": 0.7816473527738388,
      "learning_rate": 3.2975803040842646e-06,
      "loss": 0.3918,
      "step": 9934
    },
    {
      "epoch": 0.6223085234657605,
      "grad_norm": 0.8253066268017523,
      "learning_rate": 3.296626550448655e-06,
      "loss": 0.3884,
      "step": 9935
    },
    {
      "epoch": 0.6223711614651029,
      "grad_norm": 0.8751479878499975,
      "learning_rate": 3.295672866924212e-06,
      "loss": 0.4094,
      "step": 9936
    },
    {
      "epoch": 0.6224337994644451,
      "grad_norm": 0.8264341610135324,
      "learning_rate": 3.294719253550191e-06,
      "loss": 0.3846,
      "step": 9937
    },
    {
      "epoch": 0.6224964374637874,
      "grad_norm": 0.7979877710012544,
      "learning_rate": 3.293765710365842e-06,
      "loss": 0.3728,
      "step": 9938
    },
    {
      "epoch": 0.6225590754631297,
      "grad_norm": 0.9268623079455228,
      "learning_rate": 3.292812237410411e-06,
      "loss": 0.4059,
      "step": 9939
    },
    {
      "epoch": 0.622621713462472,
      "grad_norm": 0.8048249847738604,
      "learning_rate": 3.2918588347231473e-06,
      "loss": 0.3863,
      "step": 9940
    },
    {
      "epoch": 0.6226843514618143,
      "grad_norm": 0.8258769914528391,
      "learning_rate": 3.29090550234329e-06,
      "loss": 0.3632,
      "step": 9941
    },
    {
      "epoch": 0.6227469894611566,
      "grad_norm": 0.8740371750251013,
      "learning_rate": 3.2899522403100804e-06,
      "loss": 0.4008,
      "step": 9942
    },
    {
      "epoch": 0.622809627460499,
      "grad_norm": 0.8947994085635457,
      "learning_rate": 3.288999048662754e-06,
      "loss": 0.4434,
      "step": 9943
    },
    {
      "epoch": 0.6228722654598412,
      "grad_norm": 0.7433949303631712,
      "learning_rate": 3.288045927440544e-06,
      "loss": 0.3944,
      "step": 9944
    },
    {
      "epoch": 0.6229349034591836,
      "grad_norm": 0.8262809351320275,
      "learning_rate": 3.2870928766826804e-06,
      "loss": 0.3995,
      "step": 9945
    },
    {
      "epoch": 0.6229975414585258,
      "grad_norm": 0.8328271139496463,
      "learning_rate": 3.286139896428393e-06,
      "loss": 0.3929,
      "step": 9946
    },
    {
      "epoch": 0.6230601794578681,
      "grad_norm": 0.7783285505600054,
      "learning_rate": 3.285186986716906e-06,
      "loss": 0.3586,
      "step": 9947
    },
    {
      "epoch": 0.6231228174572104,
      "grad_norm": 0.8237485749632816,
      "learning_rate": 3.2842341475874382e-06,
      "loss": 0.3935,
      "step": 9948
    },
    {
      "epoch": 0.6231854554565527,
      "grad_norm": 0.8613241241485031,
      "learning_rate": 3.2832813790792144e-06,
      "loss": 0.3897,
      "step": 9949
    },
    {
      "epoch": 0.623248093455895,
      "grad_norm": 0.8624373955203666,
      "learning_rate": 3.282328681231446e-06,
      "loss": 0.4365,
      "step": 9950
    },
    {
      "epoch": 0.6233107314552373,
      "grad_norm": 0.8351256034788757,
      "learning_rate": 3.2813760540833495e-06,
      "loss": 0.4025,
      "step": 9951
    },
    {
      "epoch": 0.6233733694545797,
      "grad_norm": 0.753724186534271,
      "learning_rate": 3.2804234976741327e-06,
      "loss": 0.3739,
      "step": 9952
    },
    {
      "epoch": 0.6234360074539219,
      "grad_norm": 0.8051227012942032,
      "learning_rate": 3.279471012043004e-06,
      "loss": 0.4116,
      "step": 9953
    },
    {
      "epoch": 0.6234986454532643,
      "grad_norm": 0.8971359002699296,
      "learning_rate": 3.2785185972291678e-06,
      "loss": 0.3906,
      "step": 9954
    },
    {
      "epoch": 0.6235612834526065,
      "grad_norm": 0.7702972195537033,
      "learning_rate": 3.2775662532718256e-06,
      "loss": 0.3729,
      "step": 9955
    },
    {
      "epoch": 0.6236239214519488,
      "grad_norm": 0.798035564520532,
      "learning_rate": 3.2766139802101764e-06,
      "loss": 0.3586,
      "step": 9956
    },
    {
      "epoch": 0.6236865594512911,
      "grad_norm": 0.7802297347969629,
      "learning_rate": 3.275661778083414e-06,
      "loss": 0.4036,
      "step": 9957
    },
    {
      "epoch": 0.6237491974506334,
      "grad_norm": 0.8361457181555709,
      "learning_rate": 3.2747096469307344e-06,
      "loss": 0.3755,
      "step": 9958
    },
    {
      "epoch": 0.6238118354499758,
      "grad_norm": 0.8481412740949169,
      "learning_rate": 3.2737575867913253e-06,
      "loss": 0.3785,
      "step": 9959
    },
    {
      "epoch": 0.623874473449318,
      "grad_norm": 0.8037302300298711,
      "learning_rate": 3.272805597704375e-06,
      "loss": 0.39,
      "step": 9960
    },
    {
      "epoch": 0.6239371114486604,
      "grad_norm": 0.8369197784693977,
      "learning_rate": 3.271853679709067e-06,
      "loss": 0.4209,
      "step": 9961
    },
    {
      "epoch": 0.6239997494480026,
      "grad_norm": 0.8856967295384879,
      "learning_rate": 3.2709018328445814e-06,
      "loss": 0.4099,
      "step": 9962
    },
    {
      "epoch": 0.6240623874473449,
      "grad_norm": 0.8538743687395097,
      "learning_rate": 3.269950057150098e-06,
      "loss": 0.3985,
      "step": 9963
    },
    {
      "epoch": 0.6241250254466872,
      "grad_norm": 0.8324557625424475,
      "learning_rate": 3.2689983526647916e-06,
      "loss": 0.3868,
      "step": 9964
    },
    {
      "epoch": 0.6241876634460295,
      "grad_norm": 0.803612900933842,
      "learning_rate": 3.2680467194278343e-06,
      "loss": 0.3674,
      "step": 9965
    },
    {
      "epoch": 0.6242503014453719,
      "grad_norm": 0.8883773999271743,
      "learning_rate": 3.267095157478395e-06,
      "loss": 0.428,
      "step": 9966
    },
    {
      "epoch": 0.6243129394447141,
      "grad_norm": 0.840831869906111,
      "learning_rate": 3.266143666855639e-06,
      "loss": 0.3561,
      "step": 9967
    },
    {
      "epoch": 0.6243755774440565,
      "grad_norm": 0.8397482963592986,
      "learning_rate": 3.2651922475987337e-06,
      "loss": 0.386,
      "step": 9968
    },
    {
      "epoch": 0.6244382154433987,
      "grad_norm": 0.8073222685632787,
      "learning_rate": 3.2642408997468373e-06,
      "loss": 0.3773,
      "step": 9969
    },
    {
      "epoch": 0.6245008534427411,
      "grad_norm": 0.8678350896789233,
      "learning_rate": 3.2632896233391074e-06,
      "loss": 0.4219,
      "step": 9970
    },
    {
      "epoch": 0.6245634914420833,
      "grad_norm": 0.8474451205517592,
      "learning_rate": 3.2623384184146988e-06,
      "loss": 0.3983,
      "step": 9971
    },
    {
      "epoch": 0.6246261294414256,
      "grad_norm": 0.8153013217827841,
      "learning_rate": 3.2613872850127637e-06,
      "loss": 0.4125,
      "step": 9972
    },
    {
      "epoch": 0.624688767440768,
      "grad_norm": 0.8348835465659808,
      "learning_rate": 3.2604362231724508e-06,
      "loss": 0.365,
      "step": 9973
    },
    {
      "epoch": 0.6247514054401102,
      "grad_norm": 0.8419164836010276,
      "learning_rate": 3.2594852329329052e-06,
      "loss": 0.3999,
      "step": 9974
    },
    {
      "epoch": 0.6248140434394526,
      "grad_norm": 0.795945215411622,
      "learning_rate": 3.258534314333271e-06,
      "loss": 0.4089,
      "step": 9975
    },
    {
      "epoch": 0.6248766814387948,
      "grad_norm": 0.8393596581538777,
      "learning_rate": 3.257583467412686e-06,
      "loss": 0.3952,
      "step": 9976
    },
    {
      "epoch": 0.6249393194381372,
      "grad_norm": 0.8670695985278849,
      "learning_rate": 3.256632692210291e-06,
      "loss": 0.4019,
      "step": 9977
    },
    {
      "epoch": 0.6250019574374794,
      "grad_norm": 0.818870970429117,
      "learning_rate": 3.255681988765217e-06,
      "loss": 0.3867,
      "step": 9978
    },
    {
      "epoch": 0.6250645954368218,
      "grad_norm": 0.805176230963566,
      "learning_rate": 3.254731357116597e-06,
      "loss": 0.3882,
      "step": 9979
    },
    {
      "epoch": 0.625127233436164,
      "grad_norm": 0.8186074330642693,
      "learning_rate": 3.2537807973035573e-06,
      "loss": 0.3864,
      "step": 9980
    },
    {
      "epoch": 0.6251898714355063,
      "grad_norm": 0.8432707306664072,
      "learning_rate": 3.2528303093652247e-06,
      "loss": 0.4313,
      "step": 9981
    },
    {
      "epoch": 0.6252525094348487,
      "grad_norm": 0.819881900904387,
      "learning_rate": 3.2518798933407215e-06,
      "loss": 0.4169,
      "step": 9982
    },
    {
      "epoch": 0.6253151474341909,
      "grad_norm": 0.8586405351498242,
      "learning_rate": 3.2509295492691646e-06,
      "loss": 0.418,
      "step": 9983
    },
    {
      "epoch": 0.6253777854335333,
      "grad_norm": 0.8395543139350511,
      "learning_rate": 3.249979277189673e-06,
      "loss": 0.3661,
      "step": 9984
    },
    {
      "epoch": 0.6254404234328755,
      "grad_norm": 0.8194059574922934,
      "learning_rate": 3.249029077141358e-06,
      "loss": 0.3547,
      "step": 9985
    },
    {
      "epoch": 0.6255030614322179,
      "grad_norm": 0.8159147476690534,
      "learning_rate": 3.2480789491633326e-06,
      "loss": 0.3928,
      "step": 9986
    },
    {
      "epoch": 0.6255656994315602,
      "grad_norm": 0.8173003611855921,
      "learning_rate": 3.247128893294703e-06,
      "loss": 0.4084,
      "step": 9987
    },
    {
      "epoch": 0.6256283374309024,
      "grad_norm": 0.8098595470619774,
      "learning_rate": 3.246178909574572e-06,
      "loss": 0.4028,
      "step": 9988
    },
    {
      "epoch": 0.6256909754302448,
      "grad_norm": 1.3909637587397286,
      "learning_rate": 3.245228998042045e-06,
      "loss": 0.3851,
      "step": 9989
    },
    {
      "epoch": 0.625753613429587,
      "grad_norm": 0.8695215772656758,
      "learning_rate": 3.2442791587362165e-06,
      "loss": 0.4241,
      "step": 9990
    },
    {
      "epoch": 0.6258162514289294,
      "grad_norm": 0.8365707649620259,
      "learning_rate": 3.243329391696185e-06,
      "loss": 0.374,
      "step": 9991
    },
    {
      "epoch": 0.6258788894282716,
      "grad_norm": 0.8367945720431932,
      "learning_rate": 3.24237969696104e-06,
      "loss": 0.3668,
      "step": 9992
    },
    {
      "epoch": 0.625941527427614,
      "grad_norm": 0.663890711764318,
      "learning_rate": 3.241430074569874e-06,
      "loss": 0.4553,
      "step": 9993
    },
    {
      "epoch": 0.6260041654269562,
      "grad_norm": 0.8827493724834891,
      "learning_rate": 3.240480524561771e-06,
      "loss": 0.3752,
      "step": 9994
    },
    {
      "epoch": 0.6260668034262986,
      "grad_norm": 0.7983413724237716,
      "learning_rate": 3.2395310469758178e-06,
      "loss": 0.3962,
      "step": 9995
    },
    {
      "epoch": 0.6261294414256409,
      "grad_norm": 0.8025552525628159,
      "learning_rate": 3.2385816418510928e-06,
      "loss": 0.3733,
      "step": 9996
    },
    {
      "epoch": 0.6261920794249831,
      "grad_norm": 0.7848572104930112,
      "learning_rate": 3.237632309226675e-06,
      "loss": 0.3872,
      "step": 9997
    },
    {
      "epoch": 0.6262547174243255,
      "grad_norm": 0.7931330575632385,
      "learning_rate": 3.236683049141638e-06,
      "loss": 0.3684,
      "step": 9998
    },
    {
      "epoch": 0.6263173554236677,
      "grad_norm": 0.8403972693133809,
      "learning_rate": 3.235733861635054e-06,
      "loss": 0.4314,
      "step": 9999
    },
    {
      "epoch": 0.6263799934230101,
      "grad_norm": 0.8131115553381607,
      "learning_rate": 3.2347847467459915e-06,
      "loss": 0.3653,
      "step": 10000
    },
    {
      "epoch": 0.6264426314223523,
      "grad_norm": 0.8708853283747959,
      "learning_rate": 3.2338357045135173e-06,
      "loss": 0.3996,
      "step": 10001
    },
    {
      "epoch": 0.6265052694216947,
      "grad_norm": 0.897739341697837,
      "learning_rate": 3.2328867349766933e-06,
      "loss": 0.4167,
      "step": 10002
    },
    {
      "epoch": 0.626567907421037,
      "grad_norm": 0.781492272360584,
      "learning_rate": 3.231937838174577e-06,
      "loss": 0.3458,
      "step": 10003
    },
    {
      "epoch": 0.6266305454203793,
      "grad_norm": 0.8726647341175183,
      "learning_rate": 3.2309890141462295e-06,
      "loss": 0.4042,
      "step": 10004
    },
    {
      "epoch": 0.6266931834197216,
      "grad_norm": 0.8448343330142105,
      "learning_rate": 3.230040262930702e-06,
      "loss": 0.4064,
      "step": 10005
    },
    {
      "epoch": 0.6267558214190638,
      "grad_norm": 0.8273460188594023,
      "learning_rate": 3.2290915845670467e-06,
      "loss": 0.3857,
      "step": 10006
    },
    {
      "epoch": 0.6268184594184062,
      "grad_norm": 0.8608878504480427,
      "learning_rate": 3.2281429790943097e-06,
      "loss": 0.3707,
      "step": 10007
    },
    {
      "epoch": 0.6268810974177484,
      "grad_norm": 0.8569418949830816,
      "learning_rate": 3.2271944465515376e-06,
      "loss": 0.4024,
      "step": 10008
    },
    {
      "epoch": 0.6269437354170908,
      "grad_norm": 0.7836666432738681,
      "learning_rate": 3.2262459869777694e-06,
      "loss": 0.3708,
      "step": 10009
    },
    {
      "epoch": 0.6270063734164331,
      "grad_norm": 0.8057241369532389,
      "learning_rate": 3.2252976004120473e-06,
      "loss": 0.3971,
      "step": 10010
    },
    {
      "epoch": 0.6270690114157754,
      "grad_norm": 0.8395015288530249,
      "learning_rate": 3.2243492868934047e-06,
      "loss": 0.4105,
      "step": 10011
    },
    {
      "epoch": 0.6271316494151177,
      "grad_norm": 0.8888444614752293,
      "learning_rate": 3.223401046460875e-06,
      "loss": 0.3659,
      "step": 10012
    },
    {
      "epoch": 0.6271942874144599,
      "grad_norm": 0.8207625910131067,
      "learning_rate": 3.222452879153486e-06,
      "loss": 0.3601,
      "step": 10013
    },
    {
      "epoch": 0.6272569254138023,
      "grad_norm": 0.818162612087328,
      "learning_rate": 3.2215047850102676e-06,
      "loss": 0.4276,
      "step": 10014
    },
    {
      "epoch": 0.6273195634131445,
      "grad_norm": 0.8330615292213903,
      "learning_rate": 3.220556764070244e-06,
      "loss": 0.3593,
      "step": 10015
    },
    {
      "epoch": 0.6273822014124869,
      "grad_norm": 0.8137811358748187,
      "learning_rate": 3.219608816372433e-06,
      "loss": 0.3997,
      "step": 10016
    },
    {
      "epoch": 0.6274448394118292,
      "grad_norm": 0.8934690105316516,
      "learning_rate": 3.2186609419558545e-06,
      "loss": 0.4001,
      "step": 10017
    },
    {
      "epoch": 0.6275074774111715,
      "grad_norm": 0.6002135827085645,
      "learning_rate": 3.2177131408595207e-06,
      "loss": 0.4601,
      "step": 10018
    },
    {
      "epoch": 0.6275701154105138,
      "grad_norm": 0.8680181193387471,
      "learning_rate": 3.2167654131224463e-06,
      "loss": 0.4079,
      "step": 10019
    },
    {
      "epoch": 0.6276327534098561,
      "grad_norm": 0.8320900122617565,
      "learning_rate": 3.215817758783638e-06,
      "loss": 0.3847,
      "step": 10020
    },
    {
      "epoch": 0.6276953914091984,
      "grad_norm": 0.8263119520112634,
      "learning_rate": 3.214870177882101e-06,
      "loss": 0.3809,
      "step": 10021
    },
    {
      "epoch": 0.6277580294085406,
      "grad_norm": 0.868465694041472,
      "learning_rate": 3.2139226704568393e-06,
      "loss": 0.4073,
      "step": 10022
    },
    {
      "epoch": 0.627820667407883,
      "grad_norm": 0.8666924864844658,
      "learning_rate": 3.212975236546852e-06,
      "loss": 0.3856,
      "step": 10023
    },
    {
      "epoch": 0.6278833054072253,
      "grad_norm": 0.81066449646964,
      "learning_rate": 3.212027876191136e-06,
      "loss": 0.377,
      "step": 10024
    },
    {
      "epoch": 0.6279459434065676,
      "grad_norm": 0.806559401387833,
      "learning_rate": 3.211080589428685e-06,
      "loss": 0.3913,
      "step": 10025
    },
    {
      "epoch": 0.6280085814059099,
      "grad_norm": 0.8122360006004732,
      "learning_rate": 3.2101333762984876e-06,
      "loss": 0.3442,
      "step": 10026
    },
    {
      "epoch": 0.6280712194052522,
      "grad_norm": 0.8875487110092369,
      "learning_rate": 3.2091862368395343e-06,
      "loss": 0.4329,
      "step": 10027
    },
    {
      "epoch": 0.6281338574045945,
      "grad_norm": 0.8913116916447883,
      "learning_rate": 3.2082391710908074e-06,
      "loss": 0.4294,
      "step": 10028
    },
    {
      "epoch": 0.6281964954039368,
      "grad_norm": 0.8102169239167812,
      "learning_rate": 3.207292179091289e-06,
      "loss": 0.3819,
      "step": 10029
    },
    {
      "epoch": 0.6282591334032791,
      "grad_norm": 0.8577155098845842,
      "learning_rate": 3.2063452608799562e-06,
      "loss": 0.4338,
      "step": 10030
    },
    {
      "epoch": 0.6283217714026214,
      "grad_norm": 0.861169274382156,
      "learning_rate": 3.205398416495784e-06,
      "loss": 0.399,
      "step": 10031
    },
    {
      "epoch": 0.6283844094019637,
      "grad_norm": 0.7993526768507209,
      "learning_rate": 3.204451645977748e-06,
      "loss": 0.415,
      "step": 10032
    },
    {
      "epoch": 0.628447047401306,
      "grad_norm": 0.7988604988768119,
      "learning_rate": 3.203504949364816e-06,
      "loss": 0.3637,
      "step": 10033
    },
    {
      "epoch": 0.6285096854006483,
      "grad_norm": 0.854117918534581,
      "learning_rate": 3.2025583266959535e-06,
      "loss": 0.385,
      "step": 10034
    },
    {
      "epoch": 0.6285723233999906,
      "grad_norm": 0.8578574271345212,
      "learning_rate": 3.201611778010123e-06,
      "loss": 0.4029,
      "step": 10035
    },
    {
      "epoch": 0.628634961399333,
      "grad_norm": 0.8186812791387317,
      "learning_rate": 3.2006653033462863e-06,
      "loss": 0.396,
      "step": 10036
    },
    {
      "epoch": 0.6286975993986752,
      "grad_norm": 0.7895222773560193,
      "learning_rate": 3.1997189027434e-06,
      "loss": 0.3557,
      "step": 10037
    },
    {
      "epoch": 0.6287602373980175,
      "grad_norm": 0.7701570939647804,
      "learning_rate": 3.1987725762404174e-06,
      "loss": 0.3419,
      "step": 10038
    },
    {
      "epoch": 0.6288228753973598,
      "grad_norm": 0.8111331728361865,
      "learning_rate": 3.1978263238762895e-06,
      "loss": 0.3742,
      "step": 10039
    },
    {
      "epoch": 0.6288855133967021,
      "grad_norm": 0.8100016275470417,
      "learning_rate": 3.1968801456899635e-06,
      "loss": 0.3983,
      "step": 10040
    },
    {
      "epoch": 0.6289481513960444,
      "grad_norm": 0.8169217699514902,
      "learning_rate": 3.1959340417203866e-06,
      "loss": 0.3827,
      "step": 10041
    },
    {
      "epoch": 0.6290107893953867,
      "grad_norm": 0.9048469396308211,
      "learning_rate": 3.1949880120065e-06,
      "loss": 0.4274,
      "step": 10042
    },
    {
      "epoch": 0.629073427394729,
      "grad_norm": 0.863317576187418,
      "learning_rate": 3.194042056587242e-06,
      "loss": 0.4067,
      "step": 10043
    },
    {
      "epoch": 0.6291360653940713,
      "grad_norm": 0.5966553504590368,
      "learning_rate": 3.193096175501547e-06,
      "loss": 0.4519,
      "step": 10044
    },
    {
      "epoch": 0.6291987033934137,
      "grad_norm": 0.8777923251154279,
      "learning_rate": 3.1921503687883497e-06,
      "loss": 0.4158,
      "step": 10045
    },
    {
      "epoch": 0.6292613413927559,
      "grad_norm": 0.7965466372500477,
      "learning_rate": 3.191204636486579e-06,
      "loss": 0.3738,
      "step": 10046
    },
    {
      "epoch": 0.6293239793920982,
      "grad_norm": 0.7817715989415831,
      "learning_rate": 3.19025897863516e-06,
      "loss": 0.3524,
      "step": 10047
    },
    {
      "epoch": 0.6293866173914405,
      "grad_norm": 0.9221802956226562,
      "learning_rate": 3.1893133952730183e-06,
      "loss": 0.3983,
      "step": 10048
    },
    {
      "epoch": 0.6294492553907828,
      "grad_norm": 0.8407534235818618,
      "learning_rate": 3.188367886439071e-06,
      "loss": 0.3804,
      "step": 10049
    },
    {
      "epoch": 0.6295118933901251,
      "grad_norm": 0.8685526043731063,
      "learning_rate": 3.1874224521722406e-06,
      "loss": 0.4135,
      "step": 10050
    },
    {
      "epoch": 0.6295745313894674,
      "grad_norm": 0.8564520792219998,
      "learning_rate": 3.186477092511438e-06,
      "loss": 0.3979,
      "step": 10051
    },
    {
      "epoch": 0.6296371693888098,
      "grad_norm": 0.8346828906033581,
      "learning_rate": 3.1855318074955737e-06,
      "loss": 0.4196,
      "step": 10052
    },
    {
      "epoch": 0.629699807388152,
      "grad_norm": 0.8476365833644792,
      "learning_rate": 3.184586597163558e-06,
      "loss": 0.3836,
      "step": 10053
    },
    {
      "epoch": 0.6297624453874944,
      "grad_norm": 0.8202879309028138,
      "learning_rate": 3.1836414615542952e-06,
      "loss": 0.3999,
      "step": 10054
    },
    {
      "epoch": 0.6298250833868366,
      "grad_norm": 0.8828361525409305,
      "learning_rate": 3.1826964007066873e-06,
      "loss": 0.4437,
      "step": 10055
    },
    {
      "epoch": 0.6298877213861789,
      "grad_norm": 0.936863695710271,
      "learning_rate": 3.181751414659631e-06,
      "loss": 0.4236,
      "step": 10056
    },
    {
      "epoch": 0.6299503593855212,
      "grad_norm": 0.8829332067685569,
      "learning_rate": 3.180806503452026e-06,
      "loss": 0.3902,
      "step": 10057
    },
    {
      "epoch": 0.6300129973848635,
      "grad_norm": 0.8243841156625323,
      "learning_rate": 3.1798616671227624e-06,
      "loss": 0.3932,
      "step": 10058
    },
    {
      "epoch": 0.6300756353842059,
      "grad_norm": 0.7578670124463665,
      "learning_rate": 3.1789169057107284e-06,
      "loss": 0.3645,
      "step": 10059
    },
    {
      "epoch": 0.6301382733835481,
      "grad_norm": 0.8594334902761426,
      "learning_rate": 3.1779722192548146e-06,
      "loss": 0.3998,
      "step": 10060
    },
    {
      "epoch": 0.6302009113828905,
      "grad_norm": 0.9004578201014023,
      "learning_rate": 3.1770276077939012e-06,
      "loss": 0.3666,
      "step": 10061
    },
    {
      "epoch": 0.6302635493822327,
      "grad_norm": 0.8557628958278738,
      "learning_rate": 3.1760830713668712e-06,
      "loss": 0.3809,
      "step": 10062
    },
    {
      "epoch": 0.6303261873815751,
      "grad_norm": 0.8878073186010379,
      "learning_rate": 3.1751386100126004e-06,
      "loss": 0.3969,
      "step": 10063
    },
    {
      "epoch": 0.6303888253809173,
      "grad_norm": 0.8327486960546945,
      "learning_rate": 3.1741942237699618e-06,
      "loss": 0.3928,
      "step": 10064
    },
    {
      "epoch": 0.6304514633802596,
      "grad_norm": 0.7826557034735965,
      "learning_rate": 3.173249912677829e-06,
      "loss": 0.3995,
      "step": 10065
    },
    {
      "epoch": 0.630514101379602,
      "grad_norm": 0.7820268381859058,
      "learning_rate": 3.1723056767750685e-06,
      "loss": 0.3918,
      "step": 10066
    },
    {
      "epoch": 0.6305767393789442,
      "grad_norm": 0.8159875262850627,
      "learning_rate": 3.171361516100546e-06,
      "loss": 0.4158,
      "step": 10067
    },
    {
      "epoch": 0.6306393773782866,
      "grad_norm": 0.9185687458384924,
      "learning_rate": 3.1704174306931212e-06,
      "loss": 0.4285,
      "step": 10068
    },
    {
      "epoch": 0.6307020153776288,
      "grad_norm": 0.6496418723581682,
      "learning_rate": 3.1694734205916568e-06,
      "loss": 0.4429,
      "step": 10069
    },
    {
      "epoch": 0.6307646533769712,
      "grad_norm": 0.8977202138072483,
      "learning_rate": 3.168529485835004e-06,
      "loss": 0.4256,
      "step": 10070
    },
    {
      "epoch": 0.6308272913763134,
      "grad_norm": 0.8608825050125894,
      "learning_rate": 3.1675856264620198e-06,
      "loss": 0.4011,
      "step": 10071
    },
    {
      "epoch": 0.6308899293756557,
      "grad_norm": 0.8918371587631131,
      "learning_rate": 3.16664184251155e-06,
      "loss": 0.429,
      "step": 10072
    },
    {
      "epoch": 0.630952567374998,
      "grad_norm": 0.8332667662644055,
      "learning_rate": 3.1656981340224424e-06,
      "loss": 0.4085,
      "step": 10073
    },
    {
      "epoch": 0.6310152053743403,
      "grad_norm": 0.8966365026181222,
      "learning_rate": 3.16475450103354e-06,
      "loss": 0.4058,
      "step": 10074
    },
    {
      "epoch": 0.6310778433736827,
      "grad_norm": 0.5949851228967236,
      "learning_rate": 3.1638109435836837e-06,
      "loss": 0.4409,
      "step": 10075
    },
    {
      "epoch": 0.6311404813730249,
      "grad_norm": 0.7873986043444872,
      "learning_rate": 3.16286746171171e-06,
      "loss": 0.3719,
      "step": 10076
    },
    {
      "epoch": 0.6312031193723673,
      "grad_norm": 0.8117930487127004,
      "learning_rate": 3.1619240554564503e-06,
      "loss": 0.3863,
      "step": 10077
    },
    {
      "epoch": 0.6312657573717095,
      "grad_norm": 0.7963511245868238,
      "learning_rate": 3.160980724856738e-06,
      "loss": 0.3616,
      "step": 10078
    },
    {
      "epoch": 0.6313283953710519,
      "grad_norm": 0.8499713468180391,
      "learning_rate": 3.1600374699514023e-06,
      "loss": 0.3864,
      "step": 10079
    },
    {
      "epoch": 0.6313910333703942,
      "grad_norm": 0.8643275210855316,
      "learning_rate": 3.1590942907792654e-06,
      "loss": 0.3909,
      "step": 10080
    },
    {
      "epoch": 0.6314536713697364,
      "grad_norm": 0.8172178229074062,
      "learning_rate": 3.1581511873791495e-06,
      "loss": 0.4068,
      "step": 10081
    },
    {
      "epoch": 0.6315163093690788,
      "grad_norm": 0.8742128222748927,
      "learning_rate": 3.1572081597898714e-06,
      "loss": 0.4245,
      "step": 10082
    },
    {
      "epoch": 0.631578947368421,
      "grad_norm": 0.812115779353254,
      "learning_rate": 3.1562652080502483e-06,
      "loss": 0.3762,
      "step": 10083
    },
    {
      "epoch": 0.6316415853677634,
      "grad_norm": 0.6562375820679582,
      "learning_rate": 3.1553223321990915e-06,
      "loss": 0.4693,
      "step": 10084
    },
    {
      "epoch": 0.6317042233671056,
      "grad_norm": 0.817189336971027,
      "learning_rate": 3.154379532275209e-06,
      "loss": 0.384,
      "step": 10085
    },
    {
      "epoch": 0.631766861366448,
      "grad_norm": 0.65270300915403,
      "learning_rate": 3.1534368083174072e-06,
      "loss": 0.4532,
      "step": 10086
    },
    {
      "epoch": 0.6318294993657902,
      "grad_norm": 0.8687342304284852,
      "learning_rate": 3.15249416036449e-06,
      "loss": 0.411,
      "step": 10087
    },
    {
      "epoch": 0.6318921373651326,
      "grad_norm": 0.8594235047181206,
      "learning_rate": 3.1515515884552562e-06,
      "loss": 0.3836,
      "step": 10088
    },
    {
      "epoch": 0.6319547753644749,
      "grad_norm": 0.918977647967224,
      "learning_rate": 3.1506090926285026e-06,
      "loss": 0.4192,
      "step": 10089
    },
    {
      "epoch": 0.6320174133638171,
      "grad_norm": 0.8250795527758451,
      "learning_rate": 3.1496666729230206e-06,
      "loss": 0.4103,
      "step": 10090
    },
    {
      "epoch": 0.6320800513631595,
      "grad_norm": 0.8146944769177754,
      "learning_rate": 3.1487243293776027e-06,
      "loss": 0.3801,
      "step": 10091
    },
    {
      "epoch": 0.6321426893625017,
      "grad_norm": 0.8511479036631459,
      "learning_rate": 3.147782062031036e-06,
      "loss": 0.3718,
      "step": 10092
    },
    {
      "epoch": 0.6322053273618441,
      "grad_norm": 0.8653882607749145,
      "learning_rate": 3.1468398709221026e-06,
      "loss": 0.3946,
      "step": 10093
    },
    {
      "epoch": 0.6322679653611863,
      "grad_norm": 0.8293914299611647,
      "learning_rate": 3.145897756089583e-06,
      "loss": 0.4142,
      "step": 10094
    },
    {
      "epoch": 0.6323306033605287,
      "grad_norm": 0.7714493445151747,
      "learning_rate": 3.1449557175722556e-06,
      "loss": 0.3742,
      "step": 10095
    },
    {
      "epoch": 0.632393241359871,
      "grad_norm": 0.8210482695314857,
      "learning_rate": 3.1440137554088957e-06,
      "loss": 0.3978,
      "step": 10096
    },
    {
      "epoch": 0.6324558793592132,
      "grad_norm": 0.6312033095845699,
      "learning_rate": 3.1430718696382757e-06,
      "loss": 0.4646,
      "step": 10097
    },
    {
      "epoch": 0.6325185173585556,
      "grad_norm": 0.8176760248075898,
      "learning_rate": 3.142130060299161e-06,
      "loss": 0.344,
      "step": 10098
    },
    {
      "epoch": 0.6325811553578978,
      "grad_norm": 0.8545090759730612,
      "learning_rate": 3.141188327430318e-06,
      "loss": 0.4063,
      "step": 10099
    },
    {
      "epoch": 0.6326437933572402,
      "grad_norm": 0.8094477209303108,
      "learning_rate": 3.1402466710705087e-06,
      "loss": 0.3943,
      "step": 10100
    },
    {
      "epoch": 0.6327064313565824,
      "grad_norm": 0.8393830809184062,
      "learning_rate": 3.139305091258492e-06,
      "loss": 0.3951,
      "step": 10101
    },
    {
      "epoch": 0.6327690693559248,
      "grad_norm": 0.8299325756579354,
      "learning_rate": 3.138363588033023e-06,
      "loss": 0.3507,
      "step": 10102
    },
    {
      "epoch": 0.6328317073552671,
      "grad_norm": 0.6482010844011756,
      "learning_rate": 3.1374221614328525e-06,
      "loss": 0.4719,
      "step": 10103
    },
    {
      "epoch": 0.6328943453546094,
      "grad_norm": 0.7452811658556635,
      "learning_rate": 3.136480811496734e-06,
      "loss": 0.3768,
      "step": 10104
    },
    {
      "epoch": 0.6329569833539517,
      "grad_norm": 0.822923133000225,
      "learning_rate": 3.135539538263408e-06,
      "loss": 0.3324,
      "step": 10105
    },
    {
      "epoch": 0.6330196213532939,
      "grad_norm": 0.8699950973202858,
      "learning_rate": 3.134598341771623e-06,
      "loss": 0.4102,
      "step": 10106
    },
    {
      "epoch": 0.6330822593526363,
      "grad_norm": 0.8627909319810553,
      "learning_rate": 3.133657222060117e-06,
      "loss": 0.4233,
      "step": 10107
    },
    {
      "epoch": 0.6331448973519785,
      "grad_norm": 0.8408188348786505,
      "learning_rate": 3.132716179167624e-06,
      "loss": 0.3719,
      "step": 10108
    },
    {
      "epoch": 0.6332075353513209,
      "grad_norm": 0.8513108010489995,
      "learning_rate": 3.1317752131328817e-06,
      "loss": 0.426,
      "step": 10109
    },
    {
      "epoch": 0.6332701733506632,
      "grad_norm": 0.9282812919096408,
      "learning_rate": 3.130834323994618e-06,
      "loss": 0.4292,
      "step": 10110
    },
    {
      "epoch": 0.6333328113500055,
      "grad_norm": 0.8570684450355165,
      "learning_rate": 3.1298935117915597e-06,
      "loss": 0.3869,
      "step": 10111
    },
    {
      "epoch": 0.6333954493493478,
      "grad_norm": 0.8573332670252863,
      "learning_rate": 3.128952776562432e-06,
      "loss": 0.3744,
      "step": 10112
    },
    {
      "epoch": 0.6334580873486901,
      "grad_norm": 0.8857345117561461,
      "learning_rate": 3.1280121183459554e-06,
      "loss": 0.4103,
      "step": 10113
    },
    {
      "epoch": 0.6335207253480324,
      "grad_norm": 0.8606946999288179,
      "learning_rate": 3.127071537180846e-06,
      "loss": 0.3927,
      "step": 10114
    },
    {
      "epoch": 0.6335833633473746,
      "grad_norm": 0.8538680217288657,
      "learning_rate": 3.1261310331058214e-06,
      "loss": 0.3878,
      "step": 10115
    },
    {
      "epoch": 0.633646001346717,
      "grad_norm": 0.867028664126616,
      "learning_rate": 3.1251906061595905e-06,
      "loss": 0.3753,
      "step": 10116
    },
    {
      "epoch": 0.6337086393460593,
      "grad_norm": 0.8692211755909209,
      "learning_rate": 3.124250256380863e-06,
      "loss": 0.3851,
      "step": 10117
    },
    {
      "epoch": 0.6337712773454016,
      "grad_norm": 0.776317513181448,
      "learning_rate": 3.1233099838083436e-06,
      "loss": 0.3796,
      "step": 10118
    },
    {
      "epoch": 0.6338339153447439,
      "grad_norm": 0.9916166809884424,
      "learning_rate": 3.1223697884807323e-06,
      "loss": 0.3967,
      "step": 10119
    },
    {
      "epoch": 0.6338965533440862,
      "grad_norm": 0.8293383293314004,
      "learning_rate": 3.12142967043673e-06,
      "loss": 0.3976,
      "step": 10120
    },
    {
      "epoch": 0.6339591913434285,
      "grad_norm": 0.8736564337442472,
      "learning_rate": 3.1204896297150305e-06,
      "loss": 0.3743,
      "step": 10121
    },
    {
      "epoch": 0.6340218293427707,
      "grad_norm": 0.8572098473673238,
      "learning_rate": 3.1195496663543273e-06,
      "loss": 0.3737,
      "step": 10122
    },
    {
      "epoch": 0.6340844673421131,
      "grad_norm": 0.7968741248763062,
      "learning_rate": 3.118609780393307e-06,
      "loss": 0.3797,
      "step": 10123
    },
    {
      "epoch": 0.6341471053414554,
      "grad_norm": 0.7880267100178604,
      "learning_rate": 3.1176699718706586e-06,
      "loss": 0.3454,
      "step": 10124
    },
    {
      "epoch": 0.6342097433407977,
      "grad_norm": 0.8344006795597237,
      "learning_rate": 3.116730240825063e-06,
      "loss": 0.3708,
      "step": 10125
    },
    {
      "epoch": 0.63427238134014,
      "grad_norm": 0.8829717368556941,
      "learning_rate": 3.1157905872952017e-06,
      "loss": 0.4022,
      "step": 10126
    },
    {
      "epoch": 0.6343350193394823,
      "grad_norm": 0.8151762369248666,
      "learning_rate": 3.1148510113197493e-06,
      "loss": 0.3806,
      "step": 10127
    },
    {
      "epoch": 0.6343976573388246,
      "grad_norm": 0.8292154264055333,
      "learning_rate": 3.1139115129373785e-06,
      "loss": 0.3722,
      "step": 10128
    },
    {
      "epoch": 0.634460295338167,
      "grad_norm": 0.8563956080613868,
      "learning_rate": 3.1129720921867603e-06,
      "loss": 0.397,
      "step": 10129
    },
    {
      "epoch": 0.6345229333375092,
      "grad_norm": 0.8887833010728619,
      "learning_rate": 3.1120327491065607e-06,
      "loss": 0.4221,
      "step": 10130
    },
    {
      "epoch": 0.6345855713368515,
      "grad_norm": 0.9160778547876105,
      "learning_rate": 3.111093483735444e-06,
      "loss": 0.3987,
      "step": 10131
    },
    {
      "epoch": 0.6346482093361938,
      "grad_norm": 0.8161620613434719,
      "learning_rate": 3.110154296112068e-06,
      "loss": 0.3929,
      "step": 10132
    },
    {
      "epoch": 0.6347108473355361,
      "grad_norm": 0.8563702515480907,
      "learning_rate": 3.109215186275094e-06,
      "loss": 0.363,
      "step": 10133
    },
    {
      "epoch": 0.6347734853348784,
      "grad_norm": 0.8527330524004936,
      "learning_rate": 3.1082761542631736e-06,
      "loss": 0.3842,
      "step": 10134
    },
    {
      "epoch": 0.6348361233342207,
      "grad_norm": 0.9355372399768268,
      "learning_rate": 3.107337200114958e-06,
      "loss": 0.4189,
      "step": 10135
    },
    {
      "epoch": 0.634898761333563,
      "grad_norm": 0.8610878906906931,
      "learning_rate": 3.1063983238690943e-06,
      "loss": 0.3921,
      "step": 10136
    },
    {
      "epoch": 0.6349613993329053,
      "grad_norm": 0.9186969064026246,
      "learning_rate": 3.1054595255642266e-06,
      "loss": 0.425,
      "step": 10137
    },
    {
      "epoch": 0.6350240373322477,
      "grad_norm": 0.8258968837406847,
      "learning_rate": 3.1045208052389976e-06,
      "loss": 0.4187,
      "step": 10138
    },
    {
      "epoch": 0.6350866753315899,
      "grad_norm": 0.8750973125729861,
      "learning_rate": 3.103582162932044e-06,
      "loss": 0.4175,
      "step": 10139
    },
    {
      "epoch": 0.6351493133309322,
      "grad_norm": 0.8935285350108169,
      "learning_rate": 3.102643598682e-06,
      "loss": 0.3978,
      "step": 10140
    },
    {
      "epoch": 0.6352119513302745,
      "grad_norm": 0.863520941664485,
      "learning_rate": 3.101705112527497e-06,
      "loss": 0.3948,
      "step": 10141
    },
    {
      "epoch": 0.6352745893296168,
      "grad_norm": 0.7915669552393123,
      "learning_rate": 3.100766704507163e-06,
      "loss": 0.3874,
      "step": 10142
    },
    {
      "epoch": 0.6353372273289591,
      "grad_norm": 0.8687714014165596,
      "learning_rate": 3.099828374659626e-06,
      "loss": 0.4323,
      "step": 10143
    },
    {
      "epoch": 0.6353998653283014,
      "grad_norm": 0.8153647665392386,
      "learning_rate": 3.098890123023506e-06,
      "loss": 0.3739,
      "step": 10144
    },
    {
      "epoch": 0.6354625033276438,
      "grad_norm": 0.8228468909055324,
      "learning_rate": 3.097951949637421e-06,
      "loss": 0.3951,
      "step": 10145
    },
    {
      "epoch": 0.635525141326986,
      "grad_norm": 0.8206275120264467,
      "learning_rate": 3.097013854539986e-06,
      "loss": 0.3824,
      "step": 10146
    },
    {
      "epoch": 0.6355877793263283,
      "grad_norm": 0.8385255876178358,
      "learning_rate": 3.0960758377698147e-06,
      "loss": 0.4036,
      "step": 10147
    },
    {
      "epoch": 0.6356504173256706,
      "grad_norm": 0.8120078358477439,
      "learning_rate": 3.0951378993655156e-06,
      "loss": 0.3904,
      "step": 10148
    },
    {
      "epoch": 0.6357130553250129,
      "grad_norm": 0.9634750338404504,
      "learning_rate": 3.0942000393656923e-06,
      "loss": 0.3886,
      "step": 10149
    },
    {
      "epoch": 0.6357756933243552,
      "grad_norm": 0.6075010455549793,
      "learning_rate": 3.093262257808951e-06,
      "loss": 0.4395,
      "step": 10150
    },
    {
      "epoch": 0.6358383313236975,
      "grad_norm": 0.8728127375422928,
      "learning_rate": 3.092324554733887e-06,
      "loss": 0.4108,
      "step": 10151
    },
    {
      "epoch": 0.6359009693230399,
      "grad_norm": 0.8329886809715121,
      "learning_rate": 3.0913869301791e-06,
      "loss": 0.3705,
      "step": 10152
    },
    {
      "epoch": 0.6359636073223821,
      "grad_norm": 0.8047235107840978,
      "learning_rate": 3.0904493841831815e-06,
      "loss": 0.3931,
      "step": 10153
    },
    {
      "epoch": 0.6360262453217245,
      "grad_norm": 0.83202698516249,
      "learning_rate": 3.089511916784721e-06,
      "loss": 0.401,
      "step": 10154
    },
    {
      "epoch": 0.6360888833210667,
      "grad_norm": 0.8537680927700864,
      "learning_rate": 3.0885745280223035e-06,
      "loss": 0.4454,
      "step": 10155
    },
    {
      "epoch": 0.636151521320409,
      "grad_norm": 0.8434581589099183,
      "learning_rate": 3.0876372179345138e-06,
      "loss": 0.3623,
      "step": 10156
    },
    {
      "epoch": 0.6362141593197513,
      "grad_norm": 0.8085295402698722,
      "learning_rate": 3.0866999865599317e-06,
      "loss": 0.3892,
      "step": 10157
    },
    {
      "epoch": 0.6362767973190936,
      "grad_norm": 0.8288799077801371,
      "learning_rate": 3.0857628339371313e-06,
      "loss": 0.4035,
      "step": 10158
    },
    {
      "epoch": 0.636339435318436,
      "grad_norm": 0.8667876815121858,
      "learning_rate": 3.0848257601046907e-06,
      "loss": 0.3631,
      "step": 10159
    },
    {
      "epoch": 0.6364020733177782,
      "grad_norm": 0.7466847392964676,
      "learning_rate": 3.083888765101174e-06,
      "loss": 0.3354,
      "step": 10160
    },
    {
      "epoch": 0.6364647113171206,
      "grad_norm": 0.91162342921726,
      "learning_rate": 3.082951848965153e-06,
      "loss": 0.4476,
      "step": 10161
    },
    {
      "epoch": 0.6365273493164628,
      "grad_norm": 0.8319374867333426,
      "learning_rate": 3.082015011735191e-06,
      "loss": 0.366,
      "step": 10162
    },
    {
      "epoch": 0.6365899873158052,
      "grad_norm": 0.8556993880960331,
      "learning_rate": 3.081078253449845e-06,
      "loss": 0.4229,
      "step": 10163
    },
    {
      "epoch": 0.6366526253151474,
      "grad_norm": 0.8827044386602175,
      "learning_rate": 3.0801415741476757e-06,
      "loss": 0.412,
      "step": 10164
    },
    {
      "epoch": 0.6367152633144897,
      "grad_norm": 0.8404787404792975,
      "learning_rate": 3.0792049738672358e-06,
      "loss": 0.3965,
      "step": 10165
    },
    {
      "epoch": 0.636777901313832,
      "grad_norm": 0.6059610141413021,
      "learning_rate": 3.0782684526470753e-06,
      "loss": 0.456,
      "step": 10166
    },
    {
      "epoch": 0.6368405393131743,
      "grad_norm": 0.7947946918997177,
      "learning_rate": 3.077332010525741e-06,
      "loss": 0.3598,
      "step": 10167
    },
    {
      "epoch": 0.6369031773125167,
      "grad_norm": 0.8604210119028428,
      "learning_rate": 3.0763956475417795e-06,
      "loss": 0.42,
      "step": 10168
    },
    {
      "epoch": 0.6369658153118589,
      "grad_norm": 0.8456807787072534,
      "learning_rate": 3.0754593637337276e-06,
      "loss": 0.3829,
      "step": 10169
    },
    {
      "epoch": 0.6370284533112013,
      "grad_norm": 0.8276554712237506,
      "learning_rate": 3.0745231591401283e-06,
      "loss": 0.3749,
      "step": 10170
    },
    {
      "epoch": 0.6370910913105435,
      "grad_norm": 0.8130443046245603,
      "learning_rate": 3.0735870337995122e-06,
      "loss": 0.3869,
      "step": 10171
    },
    {
      "epoch": 0.6371537293098859,
      "grad_norm": 0.8193633808357939,
      "learning_rate": 3.0726509877504106e-06,
      "loss": 0.3811,
      "step": 10172
    },
    {
      "epoch": 0.6372163673092281,
      "grad_norm": 0.8970015467150805,
      "learning_rate": 3.071715021031353e-06,
      "loss": 0.3901,
      "step": 10173
    },
    {
      "epoch": 0.6372790053085704,
      "grad_norm": 0.8503761006171168,
      "learning_rate": 3.0707791336808636e-06,
      "loss": 0.4044,
      "step": 10174
    },
    {
      "epoch": 0.6373416433079128,
      "grad_norm": 0.8480174247491218,
      "learning_rate": 3.0698433257374618e-06,
      "loss": 0.4097,
      "step": 10175
    },
    {
      "epoch": 0.637404281307255,
      "grad_norm": 0.7551723417216355,
      "learning_rate": 3.0689075972396674e-06,
      "loss": 0.381,
      "step": 10176
    },
    {
      "epoch": 0.6374669193065974,
      "grad_norm": 0.8717668954014587,
      "learning_rate": 3.0679719482259956e-06,
      "loss": 0.4062,
      "step": 10177
    },
    {
      "epoch": 0.6375295573059396,
      "grad_norm": 0.8093850164310228,
      "learning_rate": 3.0670363787349545e-06,
      "loss": 0.3632,
      "step": 10178
    },
    {
      "epoch": 0.637592195305282,
      "grad_norm": 0.7809645827650716,
      "learning_rate": 3.066100888805057e-06,
      "loss": 0.3794,
      "step": 10179
    },
    {
      "epoch": 0.6376548333046242,
      "grad_norm": 0.8592304401112142,
      "learning_rate": 3.0651654784748062e-06,
      "loss": 0.3938,
      "step": 10180
    },
    {
      "epoch": 0.6377174713039665,
      "grad_norm": 0.8806990160534268,
      "learning_rate": 3.064230147782702e-06,
      "loss": 0.4228,
      "step": 10181
    },
    {
      "epoch": 0.6377801093033089,
      "grad_norm": 0.7601102239837214,
      "learning_rate": 3.0632948967672455e-06,
      "loss": 0.3453,
      "step": 10182
    },
    {
      "epoch": 0.6378427473026511,
      "grad_norm": 0.8740076252137299,
      "learning_rate": 3.0623597254669303e-06,
      "loss": 0.4086,
      "step": 10183
    },
    {
      "epoch": 0.6379053853019935,
      "grad_norm": 0.9390072404268757,
      "learning_rate": 3.0614246339202474e-06,
      "loss": 0.4041,
      "step": 10184
    },
    {
      "epoch": 0.6379680233013357,
      "grad_norm": 0.9376314916272918,
      "learning_rate": 3.0604896221656877e-06,
      "loss": 0.4499,
      "step": 10185
    },
    {
      "epoch": 0.6380306613006781,
      "grad_norm": 0.8305228041417172,
      "learning_rate": 3.0595546902417352e-06,
      "loss": 0.3809,
      "step": 10186
    },
    {
      "epoch": 0.6380932993000203,
      "grad_norm": 0.6875189329131757,
      "learning_rate": 3.0586198381868707e-06,
      "loss": 0.4483,
      "step": 10187
    },
    {
      "epoch": 0.6381559372993627,
      "grad_norm": 0.8852400692734131,
      "learning_rate": 3.0576850660395727e-06,
      "loss": 0.4251,
      "step": 10188
    },
    {
      "epoch": 0.638218575298705,
      "grad_norm": 0.833732456921471,
      "learning_rate": 3.05675037383832e-06,
      "loss": 0.3991,
      "step": 10189
    },
    {
      "epoch": 0.6382812132980472,
      "grad_norm": 0.8290947633853082,
      "learning_rate": 3.0558157616215823e-06,
      "loss": 0.3686,
      "step": 10190
    },
    {
      "epoch": 0.6383438512973896,
      "grad_norm": 0.8259933468603977,
      "learning_rate": 3.0548812294278285e-06,
      "loss": 0.3974,
      "step": 10191
    },
    {
      "epoch": 0.6384064892967318,
      "grad_norm": 0.8497706151403738,
      "learning_rate": 3.053946777295525e-06,
      "loss": 0.423,
      "step": 10192
    },
    {
      "epoch": 0.6384691272960742,
      "grad_norm": 0.6514798825360523,
      "learning_rate": 3.0530124052631323e-06,
      "loss": 0.4402,
      "step": 10193
    },
    {
      "epoch": 0.6385317652954164,
      "grad_norm": 0.8700521855968011,
      "learning_rate": 3.052078113369111e-06,
      "loss": 0.3955,
      "step": 10194
    },
    {
      "epoch": 0.6385944032947588,
      "grad_norm": 0.5571768141074077,
      "learning_rate": 3.051143901651916e-06,
      "loss": 0.4506,
      "step": 10195
    },
    {
      "epoch": 0.6386570412941011,
      "grad_norm": 0.8202136874803999,
      "learning_rate": 3.05020977015e-06,
      "loss": 0.3804,
      "step": 10196
    },
    {
      "epoch": 0.6387196792934434,
      "grad_norm": 0.618696206763964,
      "learning_rate": 3.0492757189018105e-06,
      "loss": 0.4186,
      "step": 10197
    },
    {
      "epoch": 0.6387823172927857,
      "grad_norm": 0.8012566614477595,
      "learning_rate": 3.048341747945795e-06,
      "loss": 0.3768,
      "step": 10198
    },
    {
      "epoch": 0.6388449552921279,
      "grad_norm": 0.8735894933965009,
      "learning_rate": 3.047407857320396e-06,
      "loss": 0.4151,
      "step": 10199
    },
    {
      "epoch": 0.6389075932914703,
      "grad_norm": 0.8492221903800766,
      "learning_rate": 3.0464740470640526e-06,
      "loss": 0.4049,
      "step": 10200
    },
    {
      "epoch": 0.6389702312908125,
      "grad_norm": 0.8974610351503725,
      "learning_rate": 3.045540317215199e-06,
      "loss": 0.4275,
      "step": 10201
    },
    {
      "epoch": 0.6390328692901549,
      "grad_norm": 0.7706889370358103,
      "learning_rate": 3.0446066678122688e-06,
      "loss": 0.3693,
      "step": 10202
    },
    {
      "epoch": 0.6390955072894972,
      "grad_norm": 0.9060423603262582,
      "learning_rate": 3.043673098893691e-06,
      "loss": 0.4184,
      "step": 10203
    },
    {
      "epoch": 0.6391581452888395,
      "grad_norm": 0.8288596014263737,
      "learning_rate": 3.0427396104978928e-06,
      "loss": 0.3634,
      "step": 10204
    },
    {
      "epoch": 0.6392207832881818,
      "grad_norm": 0.823938608798319,
      "learning_rate": 3.0418062026632942e-06,
      "loss": 0.3861,
      "step": 10205
    },
    {
      "epoch": 0.639283421287524,
      "grad_norm": 0.8659463115225638,
      "learning_rate": 3.0408728754283144e-06,
      "loss": 0.4026,
      "step": 10206
    },
    {
      "epoch": 0.6393460592868664,
      "grad_norm": 0.8111319010151798,
      "learning_rate": 3.039939628831373e-06,
      "loss": 0.3781,
      "step": 10207
    },
    {
      "epoch": 0.6394086972862086,
      "grad_norm": 0.8875240495465351,
      "learning_rate": 3.0390064629108797e-06,
      "loss": 0.4019,
      "step": 10208
    },
    {
      "epoch": 0.639471335285551,
      "grad_norm": 0.8373148879419275,
      "learning_rate": 3.038073377705245e-06,
      "loss": 0.3671,
      "step": 10209
    },
    {
      "epoch": 0.6395339732848933,
      "grad_norm": 0.8697757604930156,
      "learning_rate": 3.0371403732528727e-06,
      "loss": 0.4189,
      "step": 10210
    },
    {
      "epoch": 0.6395966112842356,
      "grad_norm": 0.8198972035096292,
      "learning_rate": 3.0362074495921684e-06,
      "loss": 0.3698,
      "step": 10211
    },
    {
      "epoch": 0.6396592492835779,
      "grad_norm": 0.7943369967877525,
      "learning_rate": 3.0352746067615296e-06,
      "loss": 0.4118,
      "step": 10212
    },
    {
      "epoch": 0.6397218872829202,
      "grad_norm": 0.9104902390300931,
      "learning_rate": 3.0343418447993535e-06,
      "loss": 0.4146,
      "step": 10213
    },
    {
      "epoch": 0.6397845252822625,
      "grad_norm": 0.8507042517698316,
      "learning_rate": 3.03340916374403e-06,
      "loss": 0.4159,
      "step": 10214
    },
    {
      "epoch": 0.6398471632816047,
      "grad_norm": 0.8684223392559914,
      "learning_rate": 3.032476563633949e-06,
      "loss": 0.433,
      "step": 10215
    },
    {
      "epoch": 0.6399098012809471,
      "grad_norm": 0.7748748302389251,
      "learning_rate": 3.0315440445075004e-06,
      "loss": 0.376,
      "step": 10216
    },
    {
      "epoch": 0.6399724392802894,
      "grad_norm": 0.8287748683338101,
      "learning_rate": 3.0306116064030642e-06,
      "loss": 0.3435,
      "step": 10217
    },
    {
      "epoch": 0.6400350772796317,
      "grad_norm": 0.8170381355859425,
      "learning_rate": 3.0296792493590204e-06,
      "loss": 0.3804,
      "step": 10218
    },
    {
      "epoch": 0.640097715278974,
      "grad_norm": 0.7872027261340864,
      "learning_rate": 3.028746973413743e-06,
      "loss": 0.3816,
      "step": 10219
    },
    {
      "epoch": 0.6401603532783163,
      "grad_norm": 0.8509654855802199,
      "learning_rate": 3.0278147786056066e-06,
      "loss": 0.4352,
      "step": 10220
    },
    {
      "epoch": 0.6402229912776586,
      "grad_norm": 0.8614318987503798,
      "learning_rate": 3.026882664972981e-06,
      "loss": 0.3841,
      "step": 10221
    },
    {
      "epoch": 0.640285629277001,
      "grad_norm": 0.6106593986566886,
      "learning_rate": 3.0259506325542293e-06,
      "loss": 0.4534,
      "step": 10222
    },
    {
      "epoch": 0.6403482672763432,
      "grad_norm": 0.8502064654406232,
      "learning_rate": 3.0250186813877168e-06,
      "loss": 0.4053,
      "step": 10223
    },
    {
      "epoch": 0.6404109052756854,
      "grad_norm": 0.8478353167976594,
      "learning_rate": 3.0240868115118007e-06,
      "loss": 0.4068,
      "step": 10224
    },
    {
      "epoch": 0.6404735432750278,
      "grad_norm": 0.8054464306230693,
      "learning_rate": 3.02315502296484e-06,
      "loss": 0.3839,
      "step": 10225
    },
    {
      "epoch": 0.6405361812743701,
      "grad_norm": 0.8122833801971849,
      "learning_rate": 3.0222233157851847e-06,
      "loss": 0.3682,
      "step": 10226
    },
    {
      "epoch": 0.6405988192737124,
      "grad_norm": 0.8743379113091581,
      "learning_rate": 3.0212916900111846e-06,
      "loss": 0.3979,
      "step": 10227
    },
    {
      "epoch": 0.6406614572730547,
      "grad_norm": 0.852205265903732,
      "learning_rate": 3.0203601456811865e-06,
      "loss": 0.3866,
      "step": 10228
    },
    {
      "epoch": 0.640724095272397,
      "grad_norm": 0.6411043661920018,
      "learning_rate": 3.0194286828335324e-06,
      "loss": 0.4538,
      "step": 10229
    },
    {
      "epoch": 0.6407867332717393,
      "grad_norm": 0.781394022072226,
      "learning_rate": 3.018497301506561e-06,
      "loss": 0.3522,
      "step": 10230
    },
    {
      "epoch": 0.6408493712710815,
      "grad_norm": 0.8756165132589925,
      "learning_rate": 3.0175660017386076e-06,
      "loss": 0.4311,
      "step": 10231
    },
    {
      "epoch": 0.6409120092704239,
      "grad_norm": 0.856223373565613,
      "learning_rate": 3.0166347835680073e-06,
      "loss": 0.3902,
      "step": 10232
    },
    {
      "epoch": 0.6409746472697662,
      "grad_norm": 0.8845656275464507,
      "learning_rate": 3.0157036470330864e-06,
      "loss": 0.4271,
      "step": 10233
    },
    {
      "epoch": 0.6410372852691085,
      "grad_norm": 0.8616938703306906,
      "learning_rate": 3.0147725921721705e-06,
      "loss": 0.3971,
      "step": 10234
    },
    {
      "epoch": 0.6410999232684508,
      "grad_norm": 0.8096566021778046,
      "learning_rate": 3.013841619023584e-06,
      "loss": 0.3778,
      "step": 10235
    },
    {
      "epoch": 0.6411625612677931,
      "grad_norm": 0.6474775949506898,
      "learning_rate": 3.0129107276256453e-06,
      "loss": 0.4489,
      "step": 10236
    },
    {
      "epoch": 0.6412251992671354,
      "grad_norm": 0.8065594888975145,
      "learning_rate": 3.0119799180166704e-06,
      "loss": 0.4193,
      "step": 10237
    },
    {
      "epoch": 0.6412878372664778,
      "grad_norm": 0.9121186361982992,
      "learning_rate": 3.0110491902349716e-06,
      "loss": 0.3722,
      "step": 10238
    },
    {
      "epoch": 0.64135047526582,
      "grad_norm": 0.8370006868985697,
      "learning_rate": 3.0101185443188564e-06,
      "loss": 0.3574,
      "step": 10239
    },
    {
      "epoch": 0.6414131132651623,
      "grad_norm": 0.8291413182967841,
      "learning_rate": 3.0091879803066305e-06,
      "loss": 0.3824,
      "step": 10240
    },
    {
      "epoch": 0.6414757512645046,
      "grad_norm": 0.8137008719024821,
      "learning_rate": 3.008257498236598e-06,
      "loss": 0.36,
      "step": 10241
    },
    {
      "epoch": 0.6415383892638469,
      "grad_norm": 0.848797315942417,
      "learning_rate": 3.0073270981470566e-06,
      "loss": 0.4721,
      "step": 10242
    },
    {
      "epoch": 0.6416010272631892,
      "grad_norm": 0.9386103922984361,
      "learning_rate": 3.0063967800762993e-06,
      "loss": 0.4341,
      "step": 10243
    },
    {
      "epoch": 0.6416636652625315,
      "grad_norm": 0.8071919842227091,
      "learning_rate": 3.005466544062623e-06,
      "loss": 0.4243,
      "step": 10244
    },
    {
      "epoch": 0.6417263032618739,
      "grad_norm": 0.8393665216047513,
      "learning_rate": 3.004536390144313e-06,
      "loss": 0.3961,
      "step": 10245
    },
    {
      "epoch": 0.6417889412612161,
      "grad_norm": 0.8497316644048435,
      "learning_rate": 3.0036063183596563e-06,
      "loss": 0.3877,
      "step": 10246
    },
    {
      "epoch": 0.6418515792605585,
      "grad_norm": 0.781170171249849,
      "learning_rate": 3.002676328746934e-06,
      "loss": 0.3647,
      "step": 10247
    },
    {
      "epoch": 0.6419142172599007,
      "grad_norm": 0.8494703007875176,
      "learning_rate": 3.001746421344424e-06,
      "loss": 0.384,
      "step": 10248
    },
    {
      "epoch": 0.641976855259243,
      "grad_norm": 0.8927099262988704,
      "learning_rate": 3.0008165961904035e-06,
      "loss": 0.4076,
      "step": 10249
    },
    {
      "epoch": 0.6420394932585853,
      "grad_norm": 0.8574776770335985,
      "learning_rate": 2.999886853323143e-06,
      "loss": 0.4291,
      "step": 10250
    },
    {
      "epoch": 0.6421021312579276,
      "grad_norm": 0.7970548661309891,
      "learning_rate": 2.9989571927809103e-06,
      "loss": 0.4134,
      "step": 10251
    },
    {
      "epoch": 0.64216476925727,
      "grad_norm": 0.7576989250739656,
      "learning_rate": 2.998027614601969e-06,
      "loss": 0.3404,
      "step": 10252
    },
    {
      "epoch": 0.6422274072566122,
      "grad_norm": 0.8543811159948124,
      "learning_rate": 2.9970981188245834e-06,
      "loss": 0.3718,
      "step": 10253
    },
    {
      "epoch": 0.6422900452559546,
      "grad_norm": 0.8522154150798915,
      "learning_rate": 2.9961687054870127e-06,
      "loss": 0.3484,
      "step": 10254
    },
    {
      "epoch": 0.6423526832552968,
      "grad_norm": 0.8780909610561823,
      "learning_rate": 2.9952393746275096e-06,
      "loss": 0.3666,
      "step": 10255
    },
    {
      "epoch": 0.6424153212546392,
      "grad_norm": 0.8252485115581059,
      "learning_rate": 2.9943101262843256e-06,
      "loss": 0.3792,
      "step": 10256
    },
    {
      "epoch": 0.6424779592539814,
      "grad_norm": 0.8496886147729128,
      "learning_rate": 2.9933809604957086e-06,
      "loss": 0.4048,
      "step": 10257
    },
    {
      "epoch": 0.6425405972533237,
      "grad_norm": 0.8393120334782985,
      "learning_rate": 2.9924518772999047e-06,
      "loss": 0.4129,
      "step": 10258
    },
    {
      "epoch": 0.642603235252666,
      "grad_norm": 0.8393416019209339,
      "learning_rate": 2.991522876735154e-06,
      "loss": 0.3729,
      "step": 10259
    },
    {
      "epoch": 0.6426658732520083,
      "grad_norm": 0.8225849753187521,
      "learning_rate": 2.9905939588396935e-06,
      "loss": 0.3523,
      "step": 10260
    },
    {
      "epoch": 0.6427285112513507,
      "grad_norm": 0.774592759946924,
      "learning_rate": 2.9896651236517586e-06,
      "loss": 0.3716,
      "step": 10261
    },
    {
      "epoch": 0.6427911492506929,
      "grad_norm": 0.9099907599973424,
      "learning_rate": 2.9887363712095807e-06,
      "loss": 0.4174,
      "step": 10262
    },
    {
      "epoch": 0.6428537872500353,
      "grad_norm": 0.8464504575992209,
      "learning_rate": 2.9878077015513874e-06,
      "loss": 0.4146,
      "step": 10263
    },
    {
      "epoch": 0.6429164252493775,
      "grad_norm": 0.8723521932922328,
      "learning_rate": 2.986879114715403e-06,
      "loss": 0.3962,
      "step": 10264
    },
    {
      "epoch": 0.6429790632487198,
      "grad_norm": 0.8489252520227913,
      "learning_rate": 2.9859506107398478e-06,
      "loss": 0.46,
      "step": 10265
    },
    {
      "epoch": 0.6430417012480621,
      "grad_norm": 0.9120070548478395,
      "learning_rate": 2.9850221896629385e-06,
      "loss": 0.4255,
      "step": 10266
    },
    {
      "epoch": 0.6431043392474044,
      "grad_norm": 0.9055479845802242,
      "learning_rate": 2.9840938515228906e-06,
      "loss": 0.4014,
      "step": 10267
    },
    {
      "epoch": 0.6431669772467468,
      "grad_norm": 0.8102025509213342,
      "learning_rate": 2.9831655963579144e-06,
      "loss": 0.4171,
      "step": 10268
    },
    {
      "epoch": 0.643229615246089,
      "grad_norm": 0.8557794590223856,
      "learning_rate": 2.9822374242062146e-06,
      "loss": 0.3735,
      "step": 10269
    },
    {
      "epoch": 0.6432922532454314,
      "grad_norm": 0.891417691823911,
      "learning_rate": 2.981309335105998e-06,
      "loss": 0.4229,
      "step": 10270
    },
    {
      "epoch": 0.6433548912447736,
      "grad_norm": 0.8168693356465079,
      "learning_rate": 2.980381329095463e-06,
      "loss": 0.3733,
      "step": 10271
    },
    {
      "epoch": 0.643417529244116,
      "grad_norm": 0.8949759401918692,
      "learning_rate": 2.9794534062128088e-06,
      "loss": 0.4081,
      "step": 10272
    },
    {
      "epoch": 0.6434801672434582,
      "grad_norm": 0.8656555314200102,
      "learning_rate": 2.9785255664962276e-06,
      "loss": 0.373,
      "step": 10273
    },
    {
      "epoch": 0.6435428052428005,
      "grad_norm": 0.8631045384608343,
      "learning_rate": 2.977597809983908e-06,
      "loss": 0.3808,
      "step": 10274
    },
    {
      "epoch": 0.6436054432421429,
      "grad_norm": 0.7972156564728105,
      "learning_rate": 2.9766701367140394e-06,
      "loss": 0.3979,
      "step": 10275
    },
    {
      "epoch": 0.6436680812414851,
      "grad_norm": 0.7969857469671194,
      "learning_rate": 2.9757425467248024e-06,
      "loss": 0.3898,
      "step": 10276
    },
    {
      "epoch": 0.6437307192408275,
      "grad_norm": 0.7993441197206472,
      "learning_rate": 2.974815040054379e-06,
      "loss": 0.4062,
      "step": 10277
    },
    {
      "epoch": 0.6437933572401697,
      "grad_norm": 0.7304738659173694,
      "learning_rate": 2.973887616740943e-06,
      "loss": 0.3451,
      "step": 10278
    },
    {
      "epoch": 0.6438559952395121,
      "grad_norm": 0.9049949035000983,
      "learning_rate": 2.9729602768226702e-06,
      "loss": 0.4217,
      "step": 10279
    },
    {
      "epoch": 0.6439186332388543,
      "grad_norm": 0.8726994574626912,
      "learning_rate": 2.9720330203377256e-06,
      "loss": 0.345,
      "step": 10280
    },
    {
      "epoch": 0.6439812712381967,
      "grad_norm": 1.1129882886445541,
      "learning_rate": 2.9711058473242814e-06,
      "loss": 0.4143,
      "step": 10281
    },
    {
      "epoch": 0.644043909237539,
      "grad_norm": 0.81683238055874,
      "learning_rate": 2.9701787578204965e-06,
      "loss": 0.3503,
      "step": 10282
    },
    {
      "epoch": 0.6441065472368812,
      "grad_norm": 0.7424718657324446,
      "learning_rate": 2.96925175186453e-06,
      "loss": 0.3668,
      "step": 10283
    },
    {
      "epoch": 0.6441691852362236,
      "grad_norm": 0.8393890094020249,
      "learning_rate": 2.9683248294945387e-06,
      "loss": 0.3722,
      "step": 10284
    },
    {
      "epoch": 0.6442318232355658,
      "grad_norm": 0.8023857287996329,
      "learning_rate": 2.9673979907486753e-06,
      "loss": 0.3517,
      "step": 10285
    },
    {
      "epoch": 0.6442944612349082,
      "grad_norm": 0.7749960276712824,
      "learning_rate": 2.9664712356650867e-06,
      "loss": 0.3603,
      "step": 10286
    },
    {
      "epoch": 0.6443570992342504,
      "grad_norm": 0.8248568614110336,
      "learning_rate": 2.96554456428192e-06,
      "loss": 0.4021,
      "step": 10287
    },
    {
      "epoch": 0.6444197372335928,
      "grad_norm": 0.9134601352400841,
      "learning_rate": 2.9646179766373173e-06,
      "loss": 0.4385,
      "step": 10288
    },
    {
      "epoch": 0.6444823752329351,
      "grad_norm": 0.6827649474643953,
      "learning_rate": 2.9636914727694143e-06,
      "loss": 0.456,
      "step": 10289
    },
    {
      "epoch": 0.6445450132322773,
      "grad_norm": 0.8693053195639877,
      "learning_rate": 2.96276505271635e-06,
      "loss": 0.402,
      "step": 10290
    },
    {
      "epoch": 0.6446076512316197,
      "grad_norm": 0.6781345810487694,
      "learning_rate": 2.961838716516253e-06,
      "loss": 0.4312,
      "step": 10291
    },
    {
      "epoch": 0.6446702892309619,
      "grad_norm": 0.962944011488053,
      "learning_rate": 2.960912464207254e-06,
      "loss": 0.4521,
      "step": 10292
    },
    {
      "epoch": 0.6447329272303043,
      "grad_norm": 0.9026021508867224,
      "learning_rate": 2.9599862958274766e-06,
      "loss": 0.3917,
      "step": 10293
    },
    {
      "epoch": 0.6447955652296465,
      "grad_norm": 0.8855419089219417,
      "learning_rate": 2.9590602114150423e-06,
      "loss": 0.4005,
      "step": 10294
    },
    {
      "epoch": 0.6448582032289889,
      "grad_norm": 0.8110215222370898,
      "learning_rate": 2.958134211008067e-06,
      "loss": 0.4069,
      "step": 10295
    },
    {
      "epoch": 0.6449208412283312,
      "grad_norm": 0.8611409868941502,
      "learning_rate": 2.957208294644667e-06,
      "loss": 0.4168,
      "step": 10296
    },
    {
      "epoch": 0.6449834792276735,
      "grad_norm": 0.8315556443618934,
      "learning_rate": 2.9562824623629537e-06,
      "loss": 0.4167,
      "step": 10297
    },
    {
      "epoch": 0.6450461172270158,
      "grad_norm": 0.8429850727787845,
      "learning_rate": 2.955356714201031e-06,
      "loss": 0.3842,
      "step": 10298
    },
    {
      "epoch": 0.645108755226358,
      "grad_norm": 0.8080620988163791,
      "learning_rate": 2.9544310501970074e-06,
      "loss": 0.3842,
      "step": 10299
    },
    {
      "epoch": 0.6451713932257004,
      "grad_norm": 0.8733440553156026,
      "learning_rate": 2.953505470388981e-06,
      "loss": 0.4157,
      "step": 10300
    },
    {
      "epoch": 0.6452340312250426,
      "grad_norm": 0.8806759358769563,
      "learning_rate": 2.95257997481505e-06,
      "loss": 0.3879,
      "step": 10301
    },
    {
      "epoch": 0.645296669224385,
      "grad_norm": 0.8258119587047822,
      "learning_rate": 2.9516545635133076e-06,
      "loss": 0.4152,
      "step": 10302
    },
    {
      "epoch": 0.6453593072237273,
      "grad_norm": 0.7848702807496213,
      "learning_rate": 2.950729236521843e-06,
      "loss": 0.4331,
      "step": 10303
    },
    {
      "epoch": 0.6454219452230696,
      "grad_norm": 0.7956511194903353,
      "learning_rate": 2.9498039938787426e-06,
      "loss": 0.3808,
      "step": 10304
    },
    {
      "epoch": 0.6454845832224119,
      "grad_norm": 0.879989382318073,
      "learning_rate": 2.9488788356220917e-06,
      "loss": 0.4216,
      "step": 10305
    },
    {
      "epoch": 0.6455472212217542,
      "grad_norm": 0.8857677088891731,
      "learning_rate": 2.9479537617899678e-06,
      "loss": 0.4363,
      "step": 10306
    },
    {
      "epoch": 0.6456098592210965,
      "grad_norm": 0.7971509826507288,
      "learning_rate": 2.9470287724204465e-06,
      "loss": 0.4065,
      "step": 10307
    },
    {
      "epoch": 0.6456724972204387,
      "grad_norm": 0.7968974309148572,
      "learning_rate": 2.946103867551604e-06,
      "loss": 0.3939,
      "step": 10308
    },
    {
      "epoch": 0.6457351352197811,
      "grad_norm": 0.6619685586891969,
      "learning_rate": 2.9451790472215064e-06,
      "loss": 0.4598,
      "step": 10309
    },
    {
      "epoch": 0.6457977732191233,
      "grad_norm": 0.8044010442450343,
      "learning_rate": 2.944254311468222e-06,
      "loss": 0.4132,
      "step": 10310
    },
    {
      "epoch": 0.6458604112184657,
      "grad_norm": 0.909642161256398,
      "learning_rate": 2.943329660329811e-06,
      "loss": 0.3797,
      "step": 10311
    },
    {
      "epoch": 0.645923049217808,
      "grad_norm": 0.8221438122054281,
      "learning_rate": 2.9424050938443327e-06,
      "loss": 0.3669,
      "step": 10312
    },
    {
      "epoch": 0.6459856872171503,
      "grad_norm": 0.8488889132447285,
      "learning_rate": 2.941480612049843e-06,
      "loss": 0.3947,
      "step": 10313
    },
    {
      "epoch": 0.6460483252164926,
      "grad_norm": 0.8524910115824746,
      "learning_rate": 2.9405562149843938e-06,
      "loss": 0.4013,
      "step": 10314
    },
    {
      "epoch": 0.6461109632158348,
      "grad_norm": 0.8844814306399492,
      "learning_rate": 2.9396319026860324e-06,
      "loss": 0.3921,
      "step": 10315
    },
    {
      "epoch": 0.6461736012151772,
      "grad_norm": 0.7854301659094172,
      "learning_rate": 2.938707675192803e-06,
      "loss": 0.3591,
      "step": 10316
    },
    {
      "epoch": 0.6462362392145194,
      "grad_norm": 0.8504965802041341,
      "learning_rate": 2.937783532542749e-06,
      "loss": 0.3955,
      "step": 10317
    },
    {
      "epoch": 0.6462988772138618,
      "grad_norm": 0.8086124692608985,
      "learning_rate": 2.936859474773909e-06,
      "loss": 0.3891,
      "step": 10318
    },
    {
      "epoch": 0.6463615152132041,
      "grad_norm": 0.794207150205077,
      "learning_rate": 2.935935501924315e-06,
      "loss": 0.373,
      "step": 10319
    },
    {
      "epoch": 0.6464241532125464,
      "grad_norm": 0.8875678962695084,
      "learning_rate": 2.9350116140319995e-06,
      "loss": 0.4041,
      "step": 10320
    },
    {
      "epoch": 0.6464867912118887,
      "grad_norm": 0.9006211715410856,
      "learning_rate": 2.934087811134988e-06,
      "loss": 0.3771,
      "step": 10321
    },
    {
      "epoch": 0.646549429211231,
      "grad_norm": 0.8415219035225022,
      "learning_rate": 2.933164093271307e-06,
      "loss": 0.4228,
      "step": 10322
    },
    {
      "epoch": 0.6466120672105733,
      "grad_norm": 0.8285426493375389,
      "learning_rate": 2.9322404604789754e-06,
      "loss": 0.3666,
      "step": 10323
    },
    {
      "epoch": 0.6466747052099155,
      "grad_norm": 0.8290085358984696,
      "learning_rate": 2.93131691279601e-06,
      "loss": 0.3676,
      "step": 10324
    },
    {
      "epoch": 0.6467373432092579,
      "grad_norm": 0.8481468063310242,
      "learning_rate": 2.9303934502604238e-06,
      "loss": 0.4214,
      "step": 10325
    },
    {
      "epoch": 0.6467999812086002,
      "grad_norm": 0.8448333199855925,
      "learning_rate": 2.929470072910226e-06,
      "loss": 0.3844,
      "step": 10326
    },
    {
      "epoch": 0.6468626192079425,
      "grad_norm": 0.8320634043379745,
      "learning_rate": 2.9285467807834266e-06,
      "loss": 0.4009,
      "step": 10327
    },
    {
      "epoch": 0.6469252572072848,
      "grad_norm": 0.8274107996728112,
      "learning_rate": 2.9276235739180257e-06,
      "loss": 0.3857,
      "step": 10328
    },
    {
      "epoch": 0.6469878952066271,
      "grad_norm": 0.9178922706531921,
      "learning_rate": 2.9267004523520237e-06,
      "loss": 0.4052,
      "step": 10329
    },
    {
      "epoch": 0.6470505332059694,
      "grad_norm": 0.8405711063514392,
      "learning_rate": 2.9257774161234146e-06,
      "loss": 0.4043,
      "step": 10330
    },
    {
      "epoch": 0.6471131712053118,
      "grad_norm": 0.8040466906975114,
      "learning_rate": 2.9248544652701934e-06,
      "loss": 0.3949,
      "step": 10331
    },
    {
      "epoch": 0.647175809204654,
      "grad_norm": 0.8752371576569937,
      "learning_rate": 2.9239315998303473e-06,
      "loss": 0.4544,
      "step": 10332
    },
    {
      "epoch": 0.6472384472039963,
      "grad_norm": 0.8908212459330261,
      "learning_rate": 2.9230088198418607e-06,
      "loss": 0.4076,
      "step": 10333
    },
    {
      "epoch": 0.6473010852033386,
      "grad_norm": 0.8178052810762054,
      "learning_rate": 2.922086125342718e-06,
      "loss": 0.3991,
      "step": 10334
    },
    {
      "epoch": 0.6473637232026809,
      "grad_norm": 0.9519684117357654,
      "learning_rate": 2.921163516370894e-06,
      "loss": 0.4422,
      "step": 10335
    },
    {
      "epoch": 0.6474263612020232,
      "grad_norm": 0.803030541899218,
      "learning_rate": 2.920240992964367e-06,
      "loss": 0.3689,
      "step": 10336
    },
    {
      "epoch": 0.6474889992013655,
      "grad_norm": 0.8716630519113805,
      "learning_rate": 2.919318555161107e-06,
      "loss": 0.3886,
      "step": 10337
    },
    {
      "epoch": 0.6475516372007079,
      "grad_norm": 0.8907214819312874,
      "learning_rate": 2.918396202999082e-06,
      "loss": 0.4277,
      "step": 10338
    },
    {
      "epoch": 0.6476142752000501,
      "grad_norm": 0.8768206537179998,
      "learning_rate": 2.917473936516255e-06,
      "loss": 0.4065,
      "step": 10339
    },
    {
      "epoch": 0.6476769131993924,
      "grad_norm": 0.867673632825933,
      "learning_rate": 2.9165517557505853e-06,
      "loss": 0.3917,
      "step": 10340
    },
    {
      "epoch": 0.6477395511987347,
      "grad_norm": 0.9264809550224065,
      "learning_rate": 2.9156296607400346e-06,
      "loss": 0.3989,
      "step": 10341
    },
    {
      "epoch": 0.647802189198077,
      "grad_norm": 0.8386319277767555,
      "learning_rate": 2.914707651522554e-06,
      "loss": 0.3914,
      "step": 10342
    },
    {
      "epoch": 0.6478648271974193,
      "grad_norm": 0.9494762896867374,
      "learning_rate": 2.9137857281360926e-06,
      "loss": 0.399,
      "step": 10343
    },
    {
      "epoch": 0.6479274651967616,
      "grad_norm": 0.8515455839039108,
      "learning_rate": 2.9128638906185964e-06,
      "loss": 0.4027,
      "step": 10344
    },
    {
      "epoch": 0.647990103196104,
      "grad_norm": 0.8992388241834988,
      "learning_rate": 2.9119421390080118e-06,
      "loss": 0.38,
      "step": 10345
    },
    {
      "epoch": 0.6480527411954462,
      "grad_norm": 0.8310640961699087,
      "learning_rate": 2.9110204733422765e-06,
      "loss": 0.4095,
      "step": 10346
    },
    {
      "epoch": 0.6481153791947886,
      "grad_norm": 0.8142870392487495,
      "learning_rate": 2.9100988936593254e-06,
      "loss": 0.3992,
      "step": 10347
    },
    {
      "epoch": 0.6481780171941308,
      "grad_norm": 0.8067688999593547,
      "learning_rate": 2.9091773999970928e-06,
      "loss": 0.3879,
      "step": 10348
    },
    {
      "epoch": 0.6482406551934731,
      "grad_norm": 0.8056179737525891,
      "learning_rate": 2.9082559923935038e-06,
      "loss": 0.3619,
      "step": 10349
    },
    {
      "epoch": 0.6483032931928154,
      "grad_norm": 0.8395649785630387,
      "learning_rate": 2.9073346708864893e-06,
      "loss": 0.4126,
      "step": 10350
    },
    {
      "epoch": 0.6483659311921577,
      "grad_norm": 0.8628716660130868,
      "learning_rate": 2.9064134355139673e-06,
      "loss": 0.4228,
      "step": 10351
    },
    {
      "epoch": 0.6484285691915,
      "grad_norm": 0.8260663583274955,
      "learning_rate": 2.9054922863138568e-06,
      "loss": 0.393,
      "step": 10352
    },
    {
      "epoch": 0.6484912071908423,
      "grad_norm": 0.8344115524617474,
      "learning_rate": 2.9045712233240714e-06,
      "loss": 0.4147,
      "step": 10353
    },
    {
      "epoch": 0.6485538451901847,
      "grad_norm": 0.8358296308118945,
      "learning_rate": 2.9036502465825244e-06,
      "loss": 0.4077,
      "step": 10354
    },
    {
      "epoch": 0.6486164831895269,
      "grad_norm": 0.8437963542236553,
      "learning_rate": 2.902729356127123e-06,
      "loss": 0.4231,
      "step": 10355
    },
    {
      "epoch": 0.6486791211888693,
      "grad_norm": 0.8912944601692091,
      "learning_rate": 2.9018085519957708e-06,
      "loss": 0.4127,
      "step": 10356
    },
    {
      "epoch": 0.6487417591882115,
      "grad_norm": 0.8699787871602025,
      "learning_rate": 2.900887834226366e-06,
      "loss": 0.4006,
      "step": 10357
    },
    {
      "epoch": 0.6488043971875538,
      "grad_norm": 0.8935710333830624,
      "learning_rate": 2.8999672028568086e-06,
      "loss": 0.3532,
      "step": 10358
    },
    {
      "epoch": 0.6488670351868961,
      "grad_norm": 0.6267816201418333,
      "learning_rate": 2.899046657924992e-06,
      "loss": 0.4515,
      "step": 10359
    },
    {
      "epoch": 0.6489296731862384,
      "grad_norm": 0.8567879348347045,
      "learning_rate": 2.898126199468804e-06,
      "loss": 0.4048,
      "step": 10360
    },
    {
      "epoch": 0.6489923111855808,
      "grad_norm": 0.859383708238034,
      "learning_rate": 2.897205827526132e-06,
      "loss": 0.3708,
      "step": 10361
    },
    {
      "epoch": 0.649054949184923,
      "grad_norm": 0.8013163292822556,
      "learning_rate": 2.896285542134857e-06,
      "loss": 0.3576,
      "step": 10362
    },
    {
      "epoch": 0.6491175871842654,
      "grad_norm": 0.8417205897410457,
      "learning_rate": 2.895365343332861e-06,
      "loss": 0.3396,
      "step": 10363
    },
    {
      "epoch": 0.6491802251836076,
      "grad_norm": 0.8073575119958157,
      "learning_rate": 2.8944452311580183e-06,
      "loss": 0.4036,
      "step": 10364
    },
    {
      "epoch": 0.64924286318295,
      "grad_norm": 0.8393851353054099,
      "learning_rate": 2.893525205648201e-06,
      "loss": 0.3954,
      "step": 10365
    },
    {
      "epoch": 0.6493055011822922,
      "grad_norm": 0.8291041905643626,
      "learning_rate": 2.8926052668412747e-06,
      "loss": 0.4253,
      "step": 10366
    },
    {
      "epoch": 0.6493681391816345,
      "grad_norm": 0.8698599590773702,
      "learning_rate": 2.8916854147751093e-06,
      "loss": 0.4225,
      "step": 10367
    },
    {
      "epoch": 0.6494307771809769,
      "grad_norm": 0.8228742987130873,
      "learning_rate": 2.890765649487563e-06,
      "loss": 0.3943,
      "step": 10368
    },
    {
      "epoch": 0.6494934151803191,
      "grad_norm": 0.8298622697754173,
      "learning_rate": 2.8898459710164937e-06,
      "loss": 0.3916,
      "step": 10369
    },
    {
      "epoch": 0.6495560531796615,
      "grad_norm": 0.8073854430803125,
      "learning_rate": 2.888926379399757e-06,
      "loss": 0.3638,
      "step": 10370
    },
    {
      "epoch": 0.6496186911790037,
      "grad_norm": 0.7897633186127557,
      "learning_rate": 2.888006874675202e-06,
      "loss": 0.3596,
      "step": 10371
    },
    {
      "epoch": 0.6496813291783461,
      "grad_norm": 0.8706973642402919,
      "learning_rate": 2.887087456880674e-06,
      "loss": 0.3939,
      "step": 10372
    },
    {
      "epoch": 0.6497439671776883,
      "grad_norm": 0.8344013214617296,
      "learning_rate": 2.8861681260540197e-06,
      "loss": 0.3496,
      "step": 10373
    },
    {
      "epoch": 0.6498066051770306,
      "grad_norm": 0.8819693736223284,
      "learning_rate": 2.885248882233079e-06,
      "loss": 0.418,
      "step": 10374
    },
    {
      "epoch": 0.649869243176373,
      "grad_norm": 0.7792810147012744,
      "learning_rate": 2.8843297254556836e-06,
      "loss": 0.3597,
      "step": 10375
    },
    {
      "epoch": 0.6499318811757152,
      "grad_norm": 0.8799978896618733,
      "learning_rate": 2.883410655759672e-06,
      "loss": 0.4493,
      "step": 10376
    },
    {
      "epoch": 0.6499945191750576,
      "grad_norm": 0.8033233341260583,
      "learning_rate": 2.882491673182871e-06,
      "loss": 0.3808,
      "step": 10377
    },
    {
      "epoch": 0.6500571571743998,
      "grad_norm": 0.8405588796915817,
      "learning_rate": 2.8815727777631054e-06,
      "loss": 0.3785,
      "step": 10378
    },
    {
      "epoch": 0.6501197951737422,
      "grad_norm": 0.7803805301132385,
      "learning_rate": 2.8806539695381975e-06,
      "loss": 0.3251,
      "step": 10379
    },
    {
      "epoch": 0.6501824331730844,
      "grad_norm": 0.8642376898156906,
      "learning_rate": 2.879735248545965e-06,
      "loss": 0.4144,
      "step": 10380
    },
    {
      "epoch": 0.6502450711724268,
      "grad_norm": 1.0126000152465713,
      "learning_rate": 2.878816614824222e-06,
      "loss": 0.3264,
      "step": 10381
    },
    {
      "epoch": 0.650307709171769,
      "grad_norm": 0.6358940282060264,
      "learning_rate": 2.877898068410783e-06,
      "loss": 0.4509,
      "step": 10382
    },
    {
      "epoch": 0.6503703471711113,
      "grad_norm": 0.7740573458112713,
      "learning_rate": 2.8769796093434505e-06,
      "loss": 0.3658,
      "step": 10383
    },
    {
      "epoch": 0.6504329851704537,
      "grad_norm": 0.9487106090636596,
      "learning_rate": 2.8760612376600337e-06,
      "loss": 0.4128,
      "step": 10384
    },
    {
      "epoch": 0.6504956231697959,
      "grad_norm": 0.6394488448321326,
      "learning_rate": 2.8751429533983304e-06,
      "loss": 0.4488,
      "step": 10385
    },
    {
      "epoch": 0.6505582611691383,
      "grad_norm": 0.8369254197177118,
      "learning_rate": 2.874224756596138e-06,
      "loss": 0.3536,
      "step": 10386
    },
    {
      "epoch": 0.6506208991684805,
      "grad_norm": 0.897369570145749,
      "learning_rate": 2.8733066472912486e-06,
      "loss": 0.4022,
      "step": 10387
    },
    {
      "epoch": 0.6506835371678229,
      "grad_norm": 0.8640380918730703,
      "learning_rate": 2.872388625521453e-06,
      "loss": 0.3967,
      "step": 10388
    },
    {
      "epoch": 0.6507461751671652,
      "grad_norm": 0.8331931280319146,
      "learning_rate": 2.871470691324536e-06,
      "loss": 0.369,
      "step": 10389
    },
    {
      "epoch": 0.6508088131665075,
      "grad_norm": 0.8364466715540126,
      "learning_rate": 2.870552844738279e-06,
      "loss": 0.357,
      "step": 10390
    },
    {
      "epoch": 0.6508714511658498,
      "grad_norm": 0.8133391605243996,
      "learning_rate": 2.8696350858004636e-06,
      "loss": 0.3882,
      "step": 10391
    },
    {
      "epoch": 0.650934089165192,
      "grad_norm": 0.6465287628990616,
      "learning_rate": 2.8687174145488617e-06,
      "loss": 0.4519,
      "step": 10392
    },
    {
      "epoch": 0.6509967271645344,
      "grad_norm": 0.8403809766936473,
      "learning_rate": 2.8677998310212485e-06,
      "loss": 0.4101,
      "step": 10393
    },
    {
      "epoch": 0.6510593651638766,
      "grad_norm": 0.7949436606825384,
      "learning_rate": 2.8668823352553898e-06,
      "loss": 0.4307,
      "step": 10394
    },
    {
      "epoch": 0.651122003163219,
      "grad_norm": 0.8883898961441539,
      "learning_rate": 2.8659649272890505e-06,
      "loss": 0.3878,
      "step": 10395
    },
    {
      "epoch": 0.6511846411625613,
      "grad_norm": 0.8819889338710134,
      "learning_rate": 2.8650476071599907e-06,
      "loss": 0.4286,
      "step": 10396
    },
    {
      "epoch": 0.6512472791619036,
      "grad_norm": 0.8291253881621348,
      "learning_rate": 2.864130374905968e-06,
      "loss": 0.3827,
      "step": 10397
    },
    {
      "epoch": 0.6513099171612459,
      "grad_norm": 0.8601341467449632,
      "learning_rate": 2.863213230564735e-06,
      "loss": 0.3998,
      "step": 10398
    },
    {
      "epoch": 0.6513725551605881,
      "grad_norm": 0.8063093706024058,
      "learning_rate": 2.8622961741740403e-06,
      "loss": 0.3818,
      "step": 10399
    },
    {
      "epoch": 0.6514351931599305,
      "grad_norm": 0.8476001235240168,
      "learning_rate": 2.861379205771634e-06,
      "loss": 0.4063,
      "step": 10400
    },
    {
      "epoch": 0.6514978311592727,
      "grad_norm": 0.896570138044474,
      "learning_rate": 2.8604623253952535e-06,
      "loss": 0.3905,
      "step": 10401
    },
    {
      "epoch": 0.6515604691586151,
      "grad_norm": 0.8462058990737387,
      "learning_rate": 2.859545533082644e-06,
      "loss": 0.3894,
      "step": 10402
    },
    {
      "epoch": 0.6516231071579573,
      "grad_norm": 0.8743977530242601,
      "learning_rate": 2.8586288288715368e-06,
      "loss": 0.3649,
      "step": 10403
    },
    {
      "epoch": 0.6516857451572997,
      "grad_norm": 0.8215147644248464,
      "learning_rate": 2.857712212799665e-06,
      "loss": 0.3713,
      "step": 10404
    },
    {
      "epoch": 0.651748383156642,
      "grad_norm": 0.8670843449997193,
      "learning_rate": 2.856795684904756e-06,
      "loss": 0.4197,
      "step": 10405
    },
    {
      "epoch": 0.6518110211559843,
      "grad_norm": 0.8431192775188515,
      "learning_rate": 2.855879245224535e-06,
      "loss": 0.3799,
      "step": 10406
    },
    {
      "epoch": 0.6518736591553266,
      "grad_norm": 0.8326989497353712,
      "learning_rate": 2.8549628937967215e-06,
      "loss": 0.3801,
      "step": 10407
    },
    {
      "epoch": 0.6519362971546688,
      "grad_norm": 0.8432863265575351,
      "learning_rate": 2.854046630659032e-06,
      "loss": 0.4044,
      "step": 10408
    },
    {
      "epoch": 0.6519989351540112,
      "grad_norm": 0.9192246783144232,
      "learning_rate": 2.8531304558491813e-06,
      "loss": 0.4233,
      "step": 10409
    },
    {
      "epoch": 0.6520615731533534,
      "grad_norm": 0.836515415001787,
      "learning_rate": 2.852214369404881e-06,
      "loss": 0.334,
      "step": 10410
    },
    {
      "epoch": 0.6521242111526958,
      "grad_norm": 0.9402295327671001,
      "learning_rate": 2.8512983713638365e-06,
      "loss": 0.3859,
      "step": 10411
    },
    {
      "epoch": 0.6521868491520381,
      "grad_norm": 0.8673302445144948,
      "learning_rate": 2.8503824617637494e-06,
      "loss": 0.4184,
      "step": 10412
    },
    {
      "epoch": 0.6522494871513804,
      "grad_norm": 0.8932502576670382,
      "learning_rate": 2.8494666406423187e-06,
      "loss": 0.4239,
      "step": 10413
    },
    {
      "epoch": 0.6523121251507227,
      "grad_norm": 0.8152511115895374,
      "learning_rate": 2.8485509080372414e-06,
      "loss": 0.3968,
      "step": 10414
    },
    {
      "epoch": 0.652374763150065,
      "grad_norm": 0.8068816374269449,
      "learning_rate": 2.8476352639862074e-06,
      "loss": 0.3852,
      "step": 10415
    },
    {
      "epoch": 0.6524374011494073,
      "grad_norm": 0.9048735704041232,
      "learning_rate": 2.8467197085269037e-06,
      "loss": 0.419,
      "step": 10416
    },
    {
      "epoch": 0.6525000391487495,
      "grad_norm": 0.6278350465927872,
      "learning_rate": 2.8458042416970183e-06,
      "loss": 0.4472,
      "step": 10417
    },
    {
      "epoch": 0.6525626771480919,
      "grad_norm": 0.8398318901052955,
      "learning_rate": 2.844888863534228e-06,
      "loss": 0.4051,
      "step": 10418
    },
    {
      "epoch": 0.6526253151474342,
      "grad_norm": 0.9012059278610793,
      "learning_rate": 2.843973574076214e-06,
      "loss": 0.4196,
      "step": 10419
    },
    {
      "epoch": 0.6526879531467765,
      "grad_norm": 0.8498325539587697,
      "learning_rate": 2.843058373360649e-06,
      "loss": 0.37,
      "step": 10420
    },
    {
      "epoch": 0.6527505911461188,
      "grad_norm": 0.8858974356403049,
      "learning_rate": 2.8421432614252004e-06,
      "loss": 0.4287,
      "step": 10421
    },
    {
      "epoch": 0.6528132291454611,
      "grad_norm": 0.82391750660526,
      "learning_rate": 2.8412282383075362e-06,
      "loss": 0.395,
      "step": 10422
    },
    {
      "epoch": 0.6528758671448034,
      "grad_norm": 0.8389989858640459,
      "learning_rate": 2.8403133040453185e-06,
      "loss": 0.3806,
      "step": 10423
    },
    {
      "epoch": 0.6529385051441456,
      "grad_norm": 0.8120858542936563,
      "learning_rate": 2.839398458676206e-06,
      "loss": 0.4255,
      "step": 10424
    },
    {
      "epoch": 0.653001143143488,
      "grad_norm": 0.8300560437668864,
      "learning_rate": 2.838483702237852e-06,
      "loss": 0.3713,
      "step": 10425
    },
    {
      "epoch": 0.6530637811428303,
      "grad_norm": 0.8726164344858178,
      "learning_rate": 2.8375690347679114e-06,
      "loss": 0.3903,
      "step": 10426
    },
    {
      "epoch": 0.6531264191421726,
      "grad_norm": 0.7860442326831735,
      "learning_rate": 2.8366544563040288e-06,
      "loss": 0.3769,
      "step": 10427
    },
    {
      "epoch": 0.6531890571415149,
      "grad_norm": 0.9018672309171418,
      "learning_rate": 2.835739966883853e-06,
      "loss": 0.388,
      "step": 10428
    },
    {
      "epoch": 0.6532516951408572,
      "grad_norm": 0.8599167739367177,
      "learning_rate": 2.834825566545022e-06,
      "loss": 0.3662,
      "step": 10429
    },
    {
      "epoch": 0.6533143331401995,
      "grad_norm": 0.9040832249469196,
      "learning_rate": 2.8339112553251712e-06,
      "loss": 0.4325,
      "step": 10430
    },
    {
      "epoch": 0.6533769711395419,
      "grad_norm": 0.789926999174696,
      "learning_rate": 2.832997033261937e-06,
      "loss": 0.3945,
      "step": 10431
    },
    {
      "epoch": 0.6534396091388841,
      "grad_norm": 0.763476073936599,
      "learning_rate": 2.8320829003929455e-06,
      "loss": 0.3583,
      "step": 10432
    },
    {
      "epoch": 0.6535022471382264,
      "grad_norm": 0.7939676141690398,
      "learning_rate": 2.831168856755825e-06,
      "loss": 0.419,
      "step": 10433
    },
    {
      "epoch": 0.6535648851375687,
      "grad_norm": 0.785675178988405,
      "learning_rate": 2.8302549023881955e-06,
      "loss": 0.3541,
      "step": 10434
    },
    {
      "epoch": 0.653627523136911,
      "grad_norm": 0.7706955621109115,
      "learning_rate": 2.8293410373276775e-06,
      "loss": 0.3672,
      "step": 10435
    },
    {
      "epoch": 0.6536901611362533,
      "grad_norm": 0.8580449312516804,
      "learning_rate": 2.8284272616118847e-06,
      "loss": 0.4169,
      "step": 10436
    },
    {
      "epoch": 0.6537527991355956,
      "grad_norm": 0.8366227083282205,
      "learning_rate": 2.82751357527843e-06,
      "loss": 0.3887,
      "step": 10437
    },
    {
      "epoch": 0.653815437134938,
      "grad_norm": 0.9447327512416109,
      "learning_rate": 2.82659997836492e-06,
      "loss": 0.3762,
      "step": 10438
    },
    {
      "epoch": 0.6538780751342802,
      "grad_norm": 0.9230915395198712,
      "learning_rate": 2.825686470908958e-06,
      "loss": 0.4167,
      "step": 10439
    },
    {
      "epoch": 0.6539407131336226,
      "grad_norm": 0.7953174303597588,
      "learning_rate": 2.8247730529481447e-06,
      "loss": 0.3828,
      "step": 10440
    },
    {
      "epoch": 0.6540033511329648,
      "grad_norm": 0.8492289788678702,
      "learning_rate": 2.823859724520076e-06,
      "loss": 0.3492,
      "step": 10441
    },
    {
      "epoch": 0.6540659891323071,
      "grad_norm": 0.8224368972877308,
      "learning_rate": 2.8229464856623434e-06,
      "loss": 0.4022,
      "step": 10442
    },
    {
      "epoch": 0.6541286271316494,
      "grad_norm": 0.8659729911840525,
      "learning_rate": 2.82203333641254e-06,
      "loss": 0.4202,
      "step": 10443
    },
    {
      "epoch": 0.6541912651309917,
      "grad_norm": 0.8972835251848739,
      "learning_rate": 2.821120276808248e-06,
      "loss": 0.4158,
      "step": 10444
    },
    {
      "epoch": 0.654253903130334,
      "grad_norm": 0.7540810571712666,
      "learning_rate": 2.8202073068870474e-06,
      "loss": 0.3406,
      "step": 10445
    },
    {
      "epoch": 0.6543165411296763,
      "grad_norm": 0.8975113017570621,
      "learning_rate": 2.8192944266865212e-06,
      "loss": 0.3906,
      "step": 10446
    },
    {
      "epoch": 0.6543791791290187,
      "grad_norm": 0.8675089979319727,
      "learning_rate": 2.8183816362442414e-06,
      "loss": 0.4214,
      "step": 10447
    },
    {
      "epoch": 0.6544418171283609,
      "grad_norm": 0.8181750002384726,
      "learning_rate": 2.817468935597778e-06,
      "loss": 0.4209,
      "step": 10448
    },
    {
      "epoch": 0.6545044551277032,
      "grad_norm": 0.8861150242341866,
      "learning_rate": 2.8165563247846983e-06,
      "loss": 0.4276,
      "step": 10449
    },
    {
      "epoch": 0.6545670931270455,
      "grad_norm": 0.8714412160240209,
      "learning_rate": 2.815643803842566e-06,
      "loss": 0.3685,
      "step": 10450
    },
    {
      "epoch": 0.6546297311263878,
      "grad_norm": 0.7414584897244966,
      "learning_rate": 2.8147313728089374e-06,
      "loss": 0.3594,
      "step": 10451
    },
    {
      "epoch": 0.6546923691257301,
      "grad_norm": 0.9315725875473181,
      "learning_rate": 2.8138190317213737e-06,
      "loss": 0.4223,
      "step": 10452
    },
    {
      "epoch": 0.6547550071250724,
      "grad_norm": 0.7882347550047597,
      "learning_rate": 2.812906780617424e-06,
      "loss": 0.378,
      "step": 10453
    },
    {
      "epoch": 0.6548176451244148,
      "grad_norm": 0.898681617417943,
      "learning_rate": 2.8119946195346375e-06,
      "loss": 0.4041,
      "step": 10454
    },
    {
      "epoch": 0.654880283123757,
      "grad_norm": 0.8693102352823863,
      "learning_rate": 2.8110825485105566e-06,
      "loss": 0.396,
      "step": 10455
    },
    {
      "epoch": 0.6549429211230994,
      "grad_norm": 0.7748172730440683,
      "learning_rate": 2.8101705675827263e-06,
      "loss": 0.377,
      "step": 10456
    },
    {
      "epoch": 0.6550055591224416,
      "grad_norm": 0.8336362379233505,
      "learning_rate": 2.8092586767886817e-06,
      "loss": 0.4149,
      "step": 10457
    },
    {
      "epoch": 0.6550681971217839,
      "grad_norm": 0.8618077682341168,
      "learning_rate": 2.8083468761659567e-06,
      "loss": 0.3698,
      "step": 10458
    },
    {
      "epoch": 0.6551308351211262,
      "grad_norm": 0.8211277854607744,
      "learning_rate": 2.8074351657520805e-06,
      "loss": 0.3693,
      "step": 10459
    },
    {
      "epoch": 0.6551934731204685,
      "grad_norm": 0.8146099831940025,
      "learning_rate": 2.8065235455845784e-06,
      "loss": 0.3597,
      "step": 10460
    },
    {
      "epoch": 0.6552561111198109,
      "grad_norm": 0.7824374980987565,
      "learning_rate": 2.8056120157009765e-06,
      "loss": 0.3841,
      "step": 10461
    },
    {
      "epoch": 0.6553187491191531,
      "grad_norm": 0.9037893403367803,
      "learning_rate": 2.8047005761387914e-06,
      "loss": 0.4394,
      "step": 10462
    },
    {
      "epoch": 0.6553813871184955,
      "grad_norm": 0.8873896728965475,
      "learning_rate": 2.803789226935537e-06,
      "loss": 0.4176,
      "step": 10463
    },
    {
      "epoch": 0.6554440251178377,
      "grad_norm": 0.8461357340046244,
      "learning_rate": 2.8028779681287244e-06,
      "loss": 0.3885,
      "step": 10464
    },
    {
      "epoch": 0.6555066631171801,
      "grad_norm": 0.822568675239481,
      "learning_rate": 2.8019667997558646e-06,
      "loss": 0.3742,
      "step": 10465
    },
    {
      "epoch": 0.6555693011165223,
      "grad_norm": 0.8854268255230161,
      "learning_rate": 2.8010557218544593e-06,
      "loss": 0.3959,
      "step": 10466
    },
    {
      "epoch": 0.6556319391158646,
      "grad_norm": 0.8115569344770054,
      "learning_rate": 2.800144734462008e-06,
      "loss": 0.3756,
      "step": 10467
    },
    {
      "epoch": 0.655694577115207,
      "grad_norm": 0.6235798428320724,
      "learning_rate": 2.7992338376160065e-06,
      "loss": 0.4683,
      "step": 10468
    },
    {
      "epoch": 0.6557572151145492,
      "grad_norm": 0.8249861650459941,
      "learning_rate": 2.7983230313539507e-06,
      "loss": 0.3856,
      "step": 10469
    },
    {
      "epoch": 0.6558198531138916,
      "grad_norm": 0.843061284957409,
      "learning_rate": 2.797412315713327e-06,
      "loss": 0.4036,
      "step": 10470
    },
    {
      "epoch": 0.6558824911132338,
      "grad_norm": 0.8697665540943362,
      "learning_rate": 2.796501690731621e-06,
      "loss": 0.3816,
      "step": 10471
    },
    {
      "epoch": 0.6559451291125762,
      "grad_norm": 0.8811405648175704,
      "learning_rate": 2.7955911564463157e-06,
      "loss": 0.3788,
      "step": 10472
    },
    {
      "epoch": 0.6560077671119184,
      "grad_norm": 0.8852399762296699,
      "learning_rate": 2.7946807128948844e-06,
      "loss": 0.3996,
      "step": 10473
    },
    {
      "epoch": 0.6560704051112608,
      "grad_norm": 0.8568141789280022,
      "learning_rate": 2.793770360114807e-06,
      "loss": 0.3817,
      "step": 10474
    },
    {
      "epoch": 0.656133043110603,
      "grad_norm": 0.8301428765710761,
      "learning_rate": 2.7928600981435503e-06,
      "loss": 0.3944,
      "step": 10475
    },
    {
      "epoch": 0.6561956811099453,
      "grad_norm": 0.8528294216043608,
      "learning_rate": 2.7919499270185825e-06,
      "loss": 0.3783,
      "step": 10476
    },
    {
      "epoch": 0.6562583191092877,
      "grad_norm": 0.8093223996443621,
      "learning_rate": 2.7910398467773623e-06,
      "loss": 0.4067,
      "step": 10477
    },
    {
      "epoch": 0.6563209571086299,
      "grad_norm": 0.8760548633035322,
      "learning_rate": 2.7901298574573554e-06,
      "loss": 0.4344,
      "step": 10478
    },
    {
      "epoch": 0.6563835951079723,
      "grad_norm": 0.8001644296913684,
      "learning_rate": 2.789219959096013e-06,
      "loss": 0.3699,
      "step": 10479
    },
    {
      "epoch": 0.6564462331073145,
      "grad_norm": 1.0290098066345696,
      "learning_rate": 2.7883101517307876e-06,
      "loss": 0.3872,
      "step": 10480
    },
    {
      "epoch": 0.6565088711066569,
      "grad_norm": 0.8729315527646886,
      "learning_rate": 2.7874004353991268e-06,
      "loss": 0.4006,
      "step": 10481
    },
    {
      "epoch": 0.6565715091059992,
      "grad_norm": 0.8973349695715365,
      "learning_rate": 2.7864908101384723e-06,
      "loss": 0.3986,
      "step": 10482
    },
    {
      "epoch": 0.6566341471053414,
      "grad_norm": 0.8731882037364518,
      "learning_rate": 2.7855812759862695e-06,
      "loss": 0.4167,
      "step": 10483
    },
    {
      "epoch": 0.6566967851046838,
      "grad_norm": 0.9195960588044376,
      "learning_rate": 2.784671832979952e-06,
      "loss": 0.3645,
      "step": 10484
    },
    {
      "epoch": 0.656759423104026,
      "grad_norm": 0.5794271468969137,
      "learning_rate": 2.7837624811569536e-06,
      "loss": 0.439,
      "step": 10485
    },
    {
      "epoch": 0.6568220611033684,
      "grad_norm": 0.8523861024068181,
      "learning_rate": 2.7828532205547005e-06,
      "loss": 0.4312,
      "step": 10486
    },
    {
      "epoch": 0.6568846991027106,
      "grad_norm": 0.8893370121647297,
      "learning_rate": 2.781944051210622e-06,
      "loss": 0.4264,
      "step": 10487
    },
    {
      "epoch": 0.656947337102053,
      "grad_norm": 0.8877196817427271,
      "learning_rate": 2.781034973162139e-06,
      "loss": 0.4119,
      "step": 10488
    },
    {
      "epoch": 0.6570099751013952,
      "grad_norm": 0.8859792750214247,
      "learning_rate": 2.780125986446668e-06,
      "loss": 0.3847,
      "step": 10489
    },
    {
      "epoch": 0.6570726131007376,
      "grad_norm": 0.8043505653036723,
      "learning_rate": 2.7792170911016226e-06,
      "loss": 0.3578,
      "step": 10490
    },
    {
      "epoch": 0.6571352511000799,
      "grad_norm": 0.8199554973030083,
      "learning_rate": 2.7783082871644127e-06,
      "loss": 0.4033,
      "step": 10491
    },
    {
      "epoch": 0.6571978890994221,
      "grad_norm": 0.6983037552330591,
      "learning_rate": 2.777399574672448e-06,
      "loss": 0.4637,
      "step": 10492
    },
    {
      "epoch": 0.6572605270987645,
      "grad_norm": 0.8010046337123049,
      "learning_rate": 2.776490953663129e-06,
      "loss": 0.364,
      "step": 10493
    },
    {
      "epoch": 0.6573231650981067,
      "grad_norm": 0.906547839826394,
      "learning_rate": 2.775582424173854e-06,
      "loss": 0.422,
      "step": 10494
    },
    {
      "epoch": 0.6573858030974491,
      "grad_norm": 0.6376351667438945,
      "learning_rate": 2.7746739862420204e-06,
      "loss": 0.4796,
      "step": 10495
    },
    {
      "epoch": 0.6574484410967913,
      "grad_norm": 0.8532615923187502,
      "learning_rate": 2.7737656399050183e-06,
      "loss": 0.3793,
      "step": 10496
    },
    {
      "epoch": 0.6575110790961337,
      "grad_norm": 0.875471421914361,
      "learning_rate": 2.772857385200236e-06,
      "loss": 0.377,
      "step": 10497
    },
    {
      "epoch": 0.657573717095476,
      "grad_norm": 0.9256858643508727,
      "learning_rate": 2.771949222165057e-06,
      "loss": 0.3769,
      "step": 10498
    },
    {
      "epoch": 0.6576363550948183,
      "grad_norm": 0.9125184672019782,
      "learning_rate": 2.7710411508368624e-06,
      "loss": 0.3773,
      "step": 10499
    },
    {
      "epoch": 0.6576989930941606,
      "grad_norm": 0.848082196410417,
      "learning_rate": 2.7701331712530266e-06,
      "loss": 0.3903,
      "step": 10500
    },
    {
      "epoch": 0.6577616310935028,
      "grad_norm": 0.7932083697750576,
      "learning_rate": 2.7692252834509227e-06,
      "loss": 0.3901,
      "step": 10501
    },
    {
      "epoch": 0.6578242690928452,
      "grad_norm": 0.6589029834508733,
      "learning_rate": 2.7683174874679207e-06,
      "loss": 0.4515,
      "step": 10502
    },
    {
      "epoch": 0.6578869070921874,
      "grad_norm": 0.8237624160155903,
      "learning_rate": 2.767409783341384e-06,
      "loss": 0.4176,
      "step": 10503
    },
    {
      "epoch": 0.6579495450915298,
      "grad_norm": 0.7896899257036095,
      "learning_rate": 2.766502171108677e-06,
      "loss": 0.355,
      "step": 10504
    },
    {
      "epoch": 0.6580121830908721,
      "grad_norm": 0.8695192033307263,
      "learning_rate": 2.765594650807155e-06,
      "loss": 0.388,
      "step": 10505
    },
    {
      "epoch": 0.6580748210902144,
      "grad_norm": 0.8557599087706516,
      "learning_rate": 2.7646872224741716e-06,
      "loss": 0.3996,
      "step": 10506
    },
    {
      "epoch": 0.6581374590895567,
      "grad_norm": 0.893409557056463,
      "learning_rate": 2.7637798861470777e-06,
      "loss": 0.4003,
      "step": 10507
    },
    {
      "epoch": 0.6582000970888989,
      "grad_norm": 0.9105818470024488,
      "learning_rate": 2.762872641863218e-06,
      "loss": 0.4352,
      "step": 10508
    },
    {
      "epoch": 0.6582627350882413,
      "grad_norm": 0.9250197017180433,
      "learning_rate": 2.7619654896599357e-06,
      "loss": 0.4033,
      "step": 10509
    },
    {
      "epoch": 0.6583253730875835,
      "grad_norm": 0.7174034241127208,
      "learning_rate": 2.761058429574568e-06,
      "loss": 0.3553,
      "step": 10510
    },
    {
      "epoch": 0.6583880110869259,
      "grad_norm": 0.802499997191021,
      "learning_rate": 2.7601514616444526e-06,
      "loss": 0.3653,
      "step": 10511
    },
    {
      "epoch": 0.6584506490862682,
      "grad_norm": 0.8694793004881942,
      "learning_rate": 2.759244585906916e-06,
      "loss": 0.3924,
      "step": 10512
    },
    {
      "epoch": 0.6585132870856105,
      "grad_norm": 0.8265276286817648,
      "learning_rate": 2.7583378023992914e-06,
      "loss": 0.3698,
      "step": 10513
    },
    {
      "epoch": 0.6585759250849528,
      "grad_norm": 0.8161750343480915,
      "learning_rate": 2.7574311111588982e-06,
      "loss": 0.3678,
      "step": 10514
    },
    {
      "epoch": 0.6586385630842951,
      "grad_norm": 0.8102956315409197,
      "learning_rate": 2.7565245122230566e-06,
      "loss": 0.3803,
      "step": 10515
    },
    {
      "epoch": 0.6587012010836374,
      "grad_norm": 0.9506141470957025,
      "learning_rate": 2.755618005629083e-06,
      "loss": 0.4194,
      "step": 10516
    },
    {
      "epoch": 0.6587638390829796,
      "grad_norm": 0.8131104540383892,
      "learning_rate": 2.754711591414288e-06,
      "loss": 0.3883,
      "step": 10517
    },
    {
      "epoch": 0.658826477082322,
      "grad_norm": 0.810085299833827,
      "learning_rate": 2.753805269615981e-06,
      "loss": 0.3535,
      "step": 10518
    },
    {
      "epoch": 0.6588891150816643,
      "grad_norm": 0.8061575767062655,
      "learning_rate": 2.7528990402714637e-06,
      "loss": 0.4028,
      "step": 10519
    },
    {
      "epoch": 0.6589517530810066,
      "grad_norm": 0.8448042912300225,
      "learning_rate": 2.7519929034180403e-06,
      "loss": 0.3732,
      "step": 10520
    },
    {
      "epoch": 0.6590143910803489,
      "grad_norm": 0.8670627012717862,
      "learning_rate": 2.7510868590930073e-06,
      "loss": 0.3614,
      "step": 10521
    },
    {
      "epoch": 0.6590770290796912,
      "grad_norm": 0.8590302245285453,
      "learning_rate": 2.7501809073336567e-06,
      "loss": 0.4143,
      "step": 10522
    },
    {
      "epoch": 0.6591396670790335,
      "grad_norm": 0.8188386568777282,
      "learning_rate": 2.7492750481772774e-06,
      "loss": 0.3777,
      "step": 10523
    },
    {
      "epoch": 0.6592023050783758,
      "grad_norm": 0.8943682546878242,
      "learning_rate": 2.748369281661155e-06,
      "loss": 0.393,
      "step": 10524
    },
    {
      "epoch": 0.6592649430777181,
      "grad_norm": 0.7384619430408738,
      "learning_rate": 2.747463607822571e-06,
      "loss": 0.3776,
      "step": 10525
    },
    {
      "epoch": 0.6593275810770604,
      "grad_norm": 0.8396966693354714,
      "learning_rate": 2.7465580266988027e-06,
      "loss": 0.4079,
      "step": 10526
    },
    {
      "epoch": 0.6593902190764027,
      "grad_norm": 0.7633615293457282,
      "learning_rate": 2.7456525383271227e-06,
      "loss": 0.3961,
      "step": 10527
    },
    {
      "epoch": 0.659452857075745,
      "grad_norm": 0.803665882773251,
      "learning_rate": 2.744747142744805e-06,
      "loss": 0.3931,
      "step": 10528
    },
    {
      "epoch": 0.6595154950750873,
      "grad_norm": 0.9018625869067244,
      "learning_rate": 2.74384183998911e-06,
      "loss": 0.4096,
      "step": 10529
    },
    {
      "epoch": 0.6595781330744296,
      "grad_norm": 0.8294406211138323,
      "learning_rate": 2.742936630097307e-06,
      "loss": 0.3866,
      "step": 10530
    },
    {
      "epoch": 0.659640771073772,
      "grad_norm": 0.8581959918850026,
      "learning_rate": 2.7420315131066504e-06,
      "loss": 0.3778,
      "step": 10531
    },
    {
      "epoch": 0.6597034090731142,
      "grad_norm": 0.8367833979549189,
      "learning_rate": 2.7411264890543966e-06,
      "loss": 0.3966,
      "step": 10532
    },
    {
      "epoch": 0.6597660470724565,
      "grad_norm": 0.8213657630150504,
      "learning_rate": 2.740221557977795e-06,
      "loss": 0.3866,
      "step": 10533
    },
    {
      "epoch": 0.6598286850717988,
      "grad_norm": 0.8337861443385182,
      "learning_rate": 2.739316719914094e-06,
      "loss": 0.4179,
      "step": 10534
    },
    {
      "epoch": 0.6598913230711411,
      "grad_norm": 0.8744381535718171,
      "learning_rate": 2.7384119749005366e-06,
      "loss": 0.4217,
      "step": 10535
    },
    {
      "epoch": 0.6599539610704834,
      "grad_norm": 0.8368926056603122,
      "learning_rate": 2.73750732297436e-06,
      "loss": 0.3692,
      "step": 10536
    },
    {
      "epoch": 0.6600165990698257,
      "grad_norm": 0.9177925624499972,
      "learning_rate": 2.7366027641728032e-06,
      "loss": 0.4137,
      "step": 10537
    },
    {
      "epoch": 0.660079237069168,
      "grad_norm": 0.7933454033427582,
      "learning_rate": 2.7356982985330953e-06,
      "loss": 0.3656,
      "step": 10538
    },
    {
      "epoch": 0.6601418750685103,
      "grad_norm": 0.8107925197255438,
      "learning_rate": 2.734793926092468e-06,
      "loss": 0.3628,
      "step": 10539
    },
    {
      "epoch": 0.6602045130678527,
      "grad_norm": 0.834629669218541,
      "learning_rate": 2.7338896468881425e-06,
      "loss": 0.4011,
      "step": 10540
    },
    {
      "epoch": 0.6602671510671949,
      "grad_norm": 0.8180125868046948,
      "learning_rate": 2.7329854609573405e-06,
      "loss": 0.4464,
      "step": 10541
    },
    {
      "epoch": 0.6603297890665372,
      "grad_norm": 0.8366481658583663,
      "learning_rate": 2.732081368337277e-06,
      "loss": 0.3823,
      "step": 10542
    },
    {
      "epoch": 0.6603924270658795,
      "grad_norm": 0.8059377659570938,
      "learning_rate": 2.7311773690651654e-06,
      "loss": 0.3762,
      "step": 10543
    },
    {
      "epoch": 0.6604550650652218,
      "grad_norm": 0.8470430606423485,
      "learning_rate": 2.730273463178215e-06,
      "loss": 0.3783,
      "step": 10544
    },
    {
      "epoch": 0.6605177030645641,
      "grad_norm": 0.783093410140835,
      "learning_rate": 2.729369650713627e-06,
      "loss": 0.3693,
      "step": 10545
    },
    {
      "epoch": 0.6605803410639064,
      "grad_norm": 0.8520242260210891,
      "learning_rate": 2.7284659317086083e-06,
      "loss": 0.4111,
      "step": 10546
    },
    {
      "epoch": 0.6606429790632488,
      "grad_norm": 0.8264447457527572,
      "learning_rate": 2.727562306200352e-06,
      "loss": 0.3744,
      "step": 10547
    },
    {
      "epoch": 0.660705617062591,
      "grad_norm": 0.9328623063216254,
      "learning_rate": 2.7266587742260543e-06,
      "loss": 0.384,
      "step": 10548
    },
    {
      "epoch": 0.6607682550619334,
      "grad_norm": 0.8886507548437687,
      "learning_rate": 2.725755335822903e-06,
      "loss": 0.406,
      "step": 10549
    },
    {
      "epoch": 0.6608308930612756,
      "grad_norm": 0.8534968708877394,
      "learning_rate": 2.724851991028085e-06,
      "loss": 0.4321,
      "step": 10550
    },
    {
      "epoch": 0.6608935310606179,
      "grad_norm": 0.8404263686467499,
      "learning_rate": 2.723948739878781e-06,
      "loss": 0.3565,
      "step": 10551
    },
    {
      "epoch": 0.6609561690599602,
      "grad_norm": 0.8431324916171505,
      "learning_rate": 2.723045582412169e-06,
      "loss": 0.4067,
      "step": 10552
    },
    {
      "epoch": 0.6610188070593025,
      "grad_norm": 0.7710335206681765,
      "learning_rate": 2.7221425186654228e-06,
      "loss": 0.3689,
      "step": 10553
    },
    {
      "epoch": 0.6610814450586449,
      "grad_norm": 0.8847690098468955,
      "learning_rate": 2.7212395486757144e-06,
      "loss": 0.4072,
      "step": 10554
    },
    {
      "epoch": 0.6611440830579871,
      "grad_norm": 0.8619659069107781,
      "learning_rate": 2.720336672480209e-06,
      "loss": 0.3864,
      "step": 10555
    },
    {
      "epoch": 0.6612067210573295,
      "grad_norm": 0.6766013911322136,
      "learning_rate": 2.7194338901160673e-06,
      "loss": 0.4462,
      "step": 10556
    },
    {
      "epoch": 0.6612693590566717,
      "grad_norm": 0.8175288777127528,
      "learning_rate": 2.7185312016204524e-06,
      "loss": 0.4087,
      "step": 10557
    },
    {
      "epoch": 0.661331997056014,
      "grad_norm": 0.8295655114850395,
      "learning_rate": 2.7176286070305158e-06,
      "loss": 0.3938,
      "step": 10558
    },
    {
      "epoch": 0.6613946350553563,
      "grad_norm": 0.8171957499995495,
      "learning_rate": 2.7167261063834105e-06,
      "loss": 0.3776,
      "step": 10559
    },
    {
      "epoch": 0.6614572730546986,
      "grad_norm": 0.8515607205138644,
      "learning_rate": 2.715823699716281e-06,
      "loss": 0.3985,
      "step": 10560
    },
    {
      "epoch": 0.661519911054041,
      "grad_norm": 0.7758217919607128,
      "learning_rate": 2.7149213870662738e-06,
      "loss": 0.3748,
      "step": 10561
    },
    {
      "epoch": 0.6615825490533832,
      "grad_norm": 0.9619279418412371,
      "learning_rate": 2.7140191684705233e-06,
      "loss": 0.432,
      "step": 10562
    },
    {
      "epoch": 0.6616451870527256,
      "grad_norm": 0.8445117755896003,
      "learning_rate": 2.7131170439661703e-06,
      "loss": 0.4059,
      "step": 10563
    },
    {
      "epoch": 0.6617078250520678,
      "grad_norm": 0.8594997113126612,
      "learning_rate": 2.7122150135903445e-06,
      "loss": 0.4076,
      "step": 10564
    },
    {
      "epoch": 0.6617704630514102,
      "grad_norm": 0.8884557954442183,
      "learning_rate": 2.711313077380171e-06,
      "loss": 0.4046,
      "step": 10565
    },
    {
      "epoch": 0.6618331010507524,
      "grad_norm": 0.8489197568254601,
      "learning_rate": 2.7104112353727785e-06,
      "loss": 0.405,
      "step": 10566
    },
    {
      "epoch": 0.6618957390500947,
      "grad_norm": 0.8468240103970458,
      "learning_rate": 2.709509487605284e-06,
      "loss": 0.4011,
      "step": 10567
    },
    {
      "epoch": 0.661958377049437,
      "grad_norm": 0.6342086440247348,
      "learning_rate": 2.7086078341148043e-06,
      "loss": 0.4308,
      "step": 10568
    },
    {
      "epoch": 0.6620210150487793,
      "grad_norm": 1.0097317932022372,
      "learning_rate": 2.7077062749384515e-06,
      "loss": 0.4149,
      "step": 10569
    },
    {
      "epoch": 0.6620836530481217,
      "grad_norm": 0.8475632724966082,
      "learning_rate": 2.706804810113334e-06,
      "loss": 0.4072,
      "step": 10570
    },
    {
      "epoch": 0.6621462910474639,
      "grad_norm": 0.8701569218210408,
      "learning_rate": 2.7059034396765537e-06,
      "loss": 0.3884,
      "step": 10571
    },
    {
      "epoch": 0.6622089290468063,
      "grad_norm": 0.8222236176108152,
      "learning_rate": 2.7050021636652163e-06,
      "loss": 0.3986,
      "step": 10572
    },
    {
      "epoch": 0.6622715670461485,
      "grad_norm": 0.8667029208814021,
      "learning_rate": 2.7041009821164154e-06,
      "loss": 0.4086,
      "step": 10573
    },
    {
      "epoch": 0.6623342050454909,
      "grad_norm": 0.6017389148564084,
      "learning_rate": 2.703199895067242e-06,
      "loss": 0.4598,
      "step": 10574
    },
    {
      "epoch": 0.6623968430448332,
      "grad_norm": 0.8126763791506332,
      "learning_rate": 2.7022989025547896e-06,
      "loss": 0.3667,
      "step": 10575
    },
    {
      "epoch": 0.6624594810441754,
      "grad_norm": 0.9530524813019465,
      "learning_rate": 2.7013980046161407e-06,
      "loss": 0.4037,
      "step": 10576
    },
    {
      "epoch": 0.6625221190435178,
      "grad_norm": 0.5799334926510245,
      "learning_rate": 2.7004972012883767e-06,
      "loss": 0.4675,
      "step": 10577
    },
    {
      "epoch": 0.66258475704286,
      "grad_norm": 0.8394396479903493,
      "learning_rate": 2.699596492608575e-06,
      "loss": 0.3629,
      "step": 10578
    },
    {
      "epoch": 0.6626473950422024,
      "grad_norm": 0.80381113652779,
      "learning_rate": 2.6986958786138064e-06,
      "loss": 0.3937,
      "step": 10579
    },
    {
      "epoch": 0.6627100330415446,
      "grad_norm": 0.7873035916427787,
      "learning_rate": 2.6977953593411442e-06,
      "loss": 0.3772,
      "step": 10580
    },
    {
      "epoch": 0.662772671040887,
      "grad_norm": 0.8752793353244589,
      "learning_rate": 2.6968949348276524e-06,
      "loss": 0.3781,
      "step": 10581
    },
    {
      "epoch": 0.6628353090402292,
      "grad_norm": 0.6464450114716772,
      "learning_rate": 2.695994605110393e-06,
      "loss": 0.4513,
      "step": 10582
    },
    {
      "epoch": 0.6628979470395716,
      "grad_norm": 0.834569172986903,
      "learning_rate": 2.6950943702264207e-06,
      "loss": 0.3521,
      "step": 10583
    },
    {
      "epoch": 0.6629605850389139,
      "grad_norm": 0.8416570715685402,
      "learning_rate": 2.694194230212794e-06,
      "loss": 0.3672,
      "step": 10584
    },
    {
      "epoch": 0.6630232230382561,
      "grad_norm": 0.8417448566458916,
      "learning_rate": 2.693294185106562e-06,
      "loss": 0.354,
      "step": 10585
    },
    {
      "epoch": 0.6630858610375985,
      "grad_norm": 0.8075919763134194,
      "learning_rate": 2.6923942349447677e-06,
      "loss": 0.3651,
      "step": 10586
    },
    {
      "epoch": 0.6631484990369407,
      "grad_norm": 0.8548873062861493,
      "learning_rate": 2.691494379764456e-06,
      "loss": 0.4434,
      "step": 10587
    },
    {
      "epoch": 0.6632111370362831,
      "grad_norm": 0.8950966787052406,
      "learning_rate": 2.690594619602662e-06,
      "loss": 0.3847,
      "step": 10588
    },
    {
      "epoch": 0.6632737750356253,
      "grad_norm": 0.8330852079889811,
      "learning_rate": 2.689694954496424e-06,
      "loss": 0.3932,
      "step": 10589
    },
    {
      "epoch": 0.6633364130349677,
      "grad_norm": 0.8413903491299921,
      "learning_rate": 2.688795384482771e-06,
      "loss": 0.4313,
      "step": 10590
    },
    {
      "epoch": 0.66339905103431,
      "grad_norm": 0.7999669852664408,
      "learning_rate": 2.687895909598728e-06,
      "loss": 0.3815,
      "step": 10591
    },
    {
      "epoch": 0.6634616890336522,
      "grad_norm": 0.8511676982858343,
      "learning_rate": 2.6869965298813194e-06,
      "loss": 0.4407,
      "step": 10592
    },
    {
      "epoch": 0.6635243270329946,
      "grad_norm": 0.7709921354444276,
      "learning_rate": 2.6860972453675603e-06,
      "loss": 0.3959,
      "step": 10593
    },
    {
      "epoch": 0.6635869650323368,
      "grad_norm": 0.8261454011823598,
      "learning_rate": 2.6851980560944703e-06,
      "loss": 0.3781,
      "step": 10594
    },
    {
      "epoch": 0.6636496030316792,
      "grad_norm": 0.8516701775117562,
      "learning_rate": 2.6842989620990577e-06,
      "loss": 0.4121,
      "step": 10595
    },
    {
      "epoch": 0.6637122410310214,
      "grad_norm": 0.9253871625103057,
      "learning_rate": 2.68339996341833e-06,
      "loss": 0.4024,
      "step": 10596
    },
    {
      "epoch": 0.6637748790303638,
      "grad_norm": 0.8112542013911264,
      "learning_rate": 2.6825010600892876e-06,
      "loss": 0.4093,
      "step": 10597
    },
    {
      "epoch": 0.6638375170297061,
      "grad_norm": 0.801155577692176,
      "learning_rate": 2.681602252148934e-06,
      "loss": 0.3685,
      "step": 10598
    },
    {
      "epoch": 0.6639001550290484,
      "grad_norm": 0.8200075984022431,
      "learning_rate": 2.680703539634262e-06,
      "loss": 0.4245,
      "step": 10599
    },
    {
      "epoch": 0.6639627930283907,
      "grad_norm": 0.9231068314324931,
      "learning_rate": 2.679804922582262e-06,
      "loss": 0.3967,
      "step": 10600
    },
    {
      "epoch": 0.6640254310277329,
      "grad_norm": 0.6062107412042328,
      "learning_rate": 2.6789064010299224e-06,
      "loss": 0.4481,
      "step": 10601
    },
    {
      "epoch": 0.6640880690270753,
      "grad_norm": 0.8402437300848702,
      "learning_rate": 2.6780079750142253e-06,
      "loss": 0.4136,
      "step": 10602
    },
    {
      "epoch": 0.6641507070264175,
      "grad_norm": 0.9418817337825358,
      "learning_rate": 2.6771096445721513e-06,
      "loss": 0.4273,
      "step": 10603
    },
    {
      "epoch": 0.6642133450257599,
      "grad_norm": 0.8556702093036436,
      "learning_rate": 2.6762114097406766e-06,
      "loss": 0.4175,
      "step": 10604
    },
    {
      "epoch": 0.6642759830251022,
      "grad_norm": 0.8388002181862593,
      "learning_rate": 2.6753132705567684e-06,
      "loss": 0.3858,
      "step": 10605
    },
    {
      "epoch": 0.6643386210244445,
      "grad_norm": 0.7772250655925604,
      "learning_rate": 2.674415227057401e-06,
      "loss": 0.3717,
      "step": 10606
    },
    {
      "epoch": 0.6644012590237868,
      "grad_norm": 0.8581132949980658,
      "learning_rate": 2.673517279279534e-06,
      "loss": 0.3886,
      "step": 10607
    },
    {
      "epoch": 0.6644638970231291,
      "grad_norm": 0.8401760053573407,
      "learning_rate": 2.672619427260127e-06,
      "loss": 0.4219,
      "step": 10608
    },
    {
      "epoch": 0.6645265350224714,
      "grad_norm": 0.795440395922846,
      "learning_rate": 2.6717216710361367e-06,
      "loss": 0.3762,
      "step": 10609
    },
    {
      "epoch": 0.6645891730218136,
      "grad_norm": 0.9114303403187887,
      "learning_rate": 2.6708240106445145e-06,
      "loss": 0.4309,
      "step": 10610
    },
    {
      "epoch": 0.664651811021156,
      "grad_norm": 0.9150278786085277,
      "learning_rate": 2.669926446122206e-06,
      "loss": 0.4261,
      "step": 10611
    },
    {
      "epoch": 0.6647144490204983,
      "grad_norm": 0.8825711360061459,
      "learning_rate": 2.66902897750616e-06,
      "loss": 0.39,
      "step": 10612
    },
    {
      "epoch": 0.6647770870198406,
      "grad_norm": 0.7849367288761202,
      "learning_rate": 2.6681316048333138e-06,
      "loss": 0.3798,
      "step": 10613
    },
    {
      "epoch": 0.6648397250191829,
      "grad_norm": 0.7837228912999019,
      "learning_rate": 2.6672343281406016e-06,
      "loss": 0.3844,
      "step": 10614
    },
    {
      "epoch": 0.6649023630185252,
      "grad_norm": 0.8858909605274421,
      "learning_rate": 2.6663371474649592e-06,
      "loss": 0.3987,
      "step": 10615
    },
    {
      "epoch": 0.6649650010178675,
      "grad_norm": 1.0132521876618958,
      "learning_rate": 2.665440062843313e-06,
      "loss": 0.4225,
      "step": 10616
    },
    {
      "epoch": 0.6650276390172097,
      "grad_norm": 0.8437265022293478,
      "learning_rate": 2.6645430743125876e-06,
      "loss": 0.3711,
      "step": 10617
    },
    {
      "epoch": 0.6650902770165521,
      "grad_norm": 0.8608168045761188,
      "learning_rate": 2.6636461819097027e-06,
      "loss": 0.4647,
      "step": 10618
    },
    {
      "epoch": 0.6651529150158944,
      "grad_norm": 0.8359356169005911,
      "learning_rate": 2.662749385671574e-06,
      "loss": 0.4302,
      "step": 10619
    },
    {
      "epoch": 0.6652155530152367,
      "grad_norm": 0.8540187691810339,
      "learning_rate": 2.661852685635113e-06,
      "loss": 0.3824,
      "step": 10620
    },
    {
      "epoch": 0.665278191014579,
      "grad_norm": 0.9036093374109347,
      "learning_rate": 2.660956081837231e-06,
      "loss": 0.4013,
      "step": 10621
    },
    {
      "epoch": 0.6653408290139213,
      "grad_norm": 0.8204087811028371,
      "learning_rate": 2.660059574314831e-06,
      "loss": 0.3938,
      "step": 10622
    },
    {
      "epoch": 0.6654034670132636,
      "grad_norm": 0.8862260146032699,
      "learning_rate": 2.6591631631048108e-06,
      "loss": 0.4154,
      "step": 10623
    },
    {
      "epoch": 0.665466105012606,
      "grad_norm": 0.8308021456024344,
      "learning_rate": 2.658266848244072e-06,
      "loss": 0.3871,
      "step": 10624
    },
    {
      "epoch": 0.6655287430119482,
      "grad_norm": 0.9382020684349474,
      "learning_rate": 2.6573706297695047e-06,
      "loss": 0.4072,
      "step": 10625
    },
    {
      "epoch": 0.6655913810112905,
      "grad_norm": 0.8727588030985728,
      "learning_rate": 2.656474507717996e-06,
      "loss": 0.4202,
      "step": 10626
    },
    {
      "epoch": 0.6656540190106328,
      "grad_norm": 0.9100666500676278,
      "learning_rate": 2.655578482126432e-06,
      "loss": 0.4265,
      "step": 10627
    },
    {
      "epoch": 0.6657166570099751,
      "grad_norm": 0.8015641110420767,
      "learning_rate": 2.654682553031693e-06,
      "loss": 0.3535,
      "step": 10628
    },
    {
      "epoch": 0.6657792950093174,
      "grad_norm": 0.7885341821397119,
      "learning_rate": 2.653786720470653e-06,
      "loss": 0.4117,
      "step": 10629
    },
    {
      "epoch": 0.6658419330086597,
      "grad_norm": 0.813900909932111,
      "learning_rate": 2.6528909844801893e-06,
      "loss": 0.4102,
      "step": 10630
    },
    {
      "epoch": 0.665904571008002,
      "grad_norm": 0.8075637000411765,
      "learning_rate": 2.6519953450971663e-06,
      "loss": 0.3781,
      "step": 10631
    },
    {
      "epoch": 0.6659672090073443,
      "grad_norm": 0.8840144108030318,
      "learning_rate": 2.6510998023584527e-06,
      "loss": 0.3958,
      "step": 10632
    },
    {
      "epoch": 0.6660298470066867,
      "grad_norm": 0.8167060442379384,
      "learning_rate": 2.6502043563009074e-06,
      "loss": 0.3854,
      "step": 10633
    },
    {
      "epoch": 0.6660924850060289,
      "grad_norm": 0.9230818945522841,
      "learning_rate": 2.649309006961387e-06,
      "loss": 0.3979,
      "step": 10634
    },
    {
      "epoch": 0.6661551230053712,
      "grad_norm": 0.957866674791101,
      "learning_rate": 2.648413754376744e-06,
      "loss": 0.471,
      "step": 10635
    },
    {
      "epoch": 0.6662177610047135,
      "grad_norm": 0.794939748809353,
      "learning_rate": 2.647518598583827e-06,
      "loss": 0.3882,
      "step": 10636
    },
    {
      "epoch": 0.6662803990040558,
      "grad_norm": 0.8935799626132236,
      "learning_rate": 2.6466235396194815e-06,
      "loss": 0.4043,
      "step": 10637
    },
    {
      "epoch": 0.6663430370033981,
      "grad_norm": 0.8448842978234916,
      "learning_rate": 2.6457285775205455e-06,
      "loss": 0.387,
      "step": 10638
    },
    {
      "epoch": 0.6664056750027404,
      "grad_norm": 0.8369656130004922,
      "learning_rate": 2.644833712323861e-06,
      "loss": 0.3814,
      "step": 10639
    },
    {
      "epoch": 0.6664683130020828,
      "grad_norm": 0.8021444601910567,
      "learning_rate": 2.6439389440662554e-06,
      "loss": 0.3843,
      "step": 10640
    },
    {
      "epoch": 0.666530951001425,
      "grad_norm": 0.8119608335996713,
      "learning_rate": 2.643044272784563e-06,
      "loss": 0.4072,
      "step": 10641
    },
    {
      "epoch": 0.6665935890007673,
      "grad_norm": 0.7825587706306486,
      "learning_rate": 2.6421496985156048e-06,
      "loss": 0.3631,
      "step": 10642
    },
    {
      "epoch": 0.6666562270001096,
      "grad_norm": 0.8158661365218994,
      "learning_rate": 2.641255221296203e-06,
      "loss": 0.3637,
      "step": 10643
    },
    {
      "epoch": 0.6667188649994519,
      "grad_norm": 0.8466405478533939,
      "learning_rate": 2.6403608411631744e-06,
      "loss": 0.4062,
      "step": 10644
    },
    {
      "epoch": 0.6667815029987942,
      "grad_norm": 0.890123554783813,
      "learning_rate": 2.6394665581533307e-06,
      "loss": 0.3749,
      "step": 10645
    },
    {
      "epoch": 0.6668441409981365,
      "grad_norm": 0.6248831007575749,
      "learning_rate": 2.6385723723034828e-06,
      "loss": 0.4402,
      "step": 10646
    },
    {
      "epoch": 0.6669067789974789,
      "grad_norm": 0.6835137522607952,
      "learning_rate": 2.637678283650431e-06,
      "loss": 0.4544,
      "step": 10647
    },
    {
      "epoch": 0.6669694169968211,
      "grad_norm": 0.9026128698519981,
      "learning_rate": 2.6367842922309805e-06,
      "loss": 0.3999,
      "step": 10648
    },
    {
      "epoch": 0.6670320549961635,
      "grad_norm": 0.7525337443421984,
      "learning_rate": 2.6358903980819294e-06,
      "loss": 0.3504,
      "step": 10649
    },
    {
      "epoch": 0.6670946929955057,
      "grad_norm": 0.7965820258277311,
      "learning_rate": 2.6349966012400674e-06,
      "loss": 0.3838,
      "step": 10650
    },
    {
      "epoch": 0.667157330994848,
      "grad_norm": 0.8882726506493914,
      "learning_rate": 2.634102901742185e-06,
      "loss": 0.3923,
      "step": 10651
    },
    {
      "epoch": 0.6672199689941903,
      "grad_norm": 0.899076796131552,
      "learning_rate": 2.633209299625066e-06,
      "loss": 0.4016,
      "step": 10652
    },
    {
      "epoch": 0.6672826069935326,
      "grad_norm": 0.8654493915775263,
      "learning_rate": 2.632315794925491e-06,
      "loss": 0.3556,
      "step": 10653
    },
    {
      "epoch": 0.667345244992875,
      "grad_norm": 0.8602233268870063,
      "learning_rate": 2.631422387680238e-06,
      "loss": 0.3616,
      "step": 10654
    },
    {
      "epoch": 0.6674078829922172,
      "grad_norm": 0.8535453381079506,
      "learning_rate": 2.6305290779260793e-06,
      "loss": 0.4174,
      "step": 10655
    },
    {
      "epoch": 0.6674705209915596,
      "grad_norm": 0.8682138679104213,
      "learning_rate": 2.629635865699781e-06,
      "loss": 0.4137,
      "step": 10656
    },
    {
      "epoch": 0.6675331589909018,
      "grad_norm": 0.7868585996292452,
      "learning_rate": 2.6287427510381107e-06,
      "loss": 0.3822,
      "step": 10657
    },
    {
      "epoch": 0.6675957969902442,
      "grad_norm": 0.9239727116970551,
      "learning_rate": 2.627849733977831e-06,
      "loss": 0.4059,
      "step": 10658
    },
    {
      "epoch": 0.6676584349895864,
      "grad_norm": 0.8159073722856902,
      "learning_rate": 2.6269568145556966e-06,
      "loss": 0.3691,
      "step": 10659
    },
    {
      "epoch": 0.6677210729889287,
      "grad_norm": 0.8299321012667853,
      "learning_rate": 2.62606399280846e-06,
      "loss": 0.3839,
      "step": 10660
    },
    {
      "epoch": 0.667783710988271,
      "grad_norm": 0.7769095789420725,
      "learning_rate": 2.62517126877287e-06,
      "loss": 0.3404,
      "step": 10661
    },
    {
      "epoch": 0.6678463489876133,
      "grad_norm": 0.8334908653802141,
      "learning_rate": 2.6242786424856716e-06,
      "loss": 0.3766,
      "step": 10662
    },
    {
      "epoch": 0.6679089869869557,
      "grad_norm": 0.8608658187109715,
      "learning_rate": 2.623386113983605e-06,
      "loss": 0.3982,
      "step": 10663
    },
    {
      "epoch": 0.6679716249862979,
      "grad_norm": 0.8331475853086834,
      "learning_rate": 2.622493683303405e-06,
      "loss": 0.3816,
      "step": 10664
    },
    {
      "epoch": 0.6680342629856403,
      "grad_norm": 0.755745772230954,
      "learning_rate": 2.6216013504818085e-06,
      "loss": 0.3647,
      "step": 10665
    },
    {
      "epoch": 0.6680969009849825,
      "grad_norm": 0.7931091988789316,
      "learning_rate": 2.6207091155555396e-06,
      "loss": 0.3543,
      "step": 10666
    },
    {
      "epoch": 0.6681595389843248,
      "grad_norm": 0.8408217688354258,
      "learning_rate": 2.619816978561327e-06,
      "loss": 0.3739,
      "step": 10667
    },
    {
      "epoch": 0.6682221769836671,
      "grad_norm": 0.8403639121717686,
      "learning_rate": 2.61892493953589e-06,
      "loss": 0.4043,
      "step": 10668
    },
    {
      "epoch": 0.6682848149830094,
      "grad_norm": 0.8802322674976492,
      "learning_rate": 2.618032998515944e-06,
      "loss": 0.3767,
      "step": 10669
    },
    {
      "epoch": 0.6683474529823518,
      "grad_norm": 0.8865497194012776,
      "learning_rate": 2.617141155538202e-06,
      "loss": 0.4031,
      "step": 10670
    },
    {
      "epoch": 0.668410090981694,
      "grad_norm": 0.8293657267713671,
      "learning_rate": 2.616249410639372e-06,
      "loss": 0.3767,
      "step": 10671
    },
    {
      "epoch": 0.6684727289810364,
      "grad_norm": 0.8644855345878866,
      "learning_rate": 2.615357763856159e-06,
      "loss": 0.4131,
      "step": 10672
    },
    {
      "epoch": 0.6685353669803786,
      "grad_norm": 0.9048535143508941,
      "learning_rate": 2.6144662152252613e-06,
      "loss": 0.4476,
      "step": 10673
    },
    {
      "epoch": 0.668598004979721,
      "grad_norm": 0.84324778568926,
      "learning_rate": 2.6135747647833786e-06,
      "loss": 0.4179,
      "step": 10674
    },
    {
      "epoch": 0.6686606429790632,
      "grad_norm": 0.8740625779797676,
      "learning_rate": 2.6126834125671995e-06,
      "loss": 0.3998,
      "step": 10675
    },
    {
      "epoch": 0.6687232809784055,
      "grad_norm": 0.8236544387418826,
      "learning_rate": 2.6117921586134165e-06,
      "loss": 0.3754,
      "step": 10676
    },
    {
      "epoch": 0.6687859189777479,
      "grad_norm": 0.7774866959183975,
      "learning_rate": 2.610901002958711e-06,
      "loss": 0.3472,
      "step": 10677
    },
    {
      "epoch": 0.6688485569770901,
      "grad_norm": 0.593816115673158,
      "learning_rate": 2.6100099456397644e-06,
      "loss": 0.4568,
      "step": 10678
    },
    {
      "epoch": 0.6689111949764325,
      "grad_norm": 0.8098778865193274,
      "learning_rate": 2.609118986693252e-06,
      "loss": 0.3971,
      "step": 10679
    },
    {
      "epoch": 0.6689738329757747,
      "grad_norm": 0.8969598311286378,
      "learning_rate": 2.608228126155845e-06,
      "loss": 0.3811,
      "step": 10680
    },
    {
      "epoch": 0.6690364709751171,
      "grad_norm": 0.8697771725772051,
      "learning_rate": 2.607337364064213e-06,
      "loss": 0.388,
      "step": 10681
    },
    {
      "epoch": 0.6690991089744593,
      "grad_norm": 0.583792182793882,
      "learning_rate": 2.6064467004550178e-06,
      "loss": 0.4567,
      "step": 10682
    },
    {
      "epoch": 0.6691617469738017,
      "grad_norm": 0.8518745815325568,
      "learning_rate": 2.6055561353649216e-06,
      "loss": 0.4115,
      "step": 10683
    },
    {
      "epoch": 0.669224384973144,
      "grad_norm": 0.8914765416263372,
      "learning_rate": 2.6046656688305795e-06,
      "loss": 0.4056,
      "step": 10684
    },
    {
      "epoch": 0.6692870229724862,
      "grad_norm": 0.8311113244619405,
      "learning_rate": 2.6037753008886415e-06,
      "loss": 0.3789,
      "step": 10685
    },
    {
      "epoch": 0.6693496609718286,
      "grad_norm": 0.8751924455313296,
      "learning_rate": 2.602885031575758e-06,
      "loss": 0.3941,
      "step": 10686
    },
    {
      "epoch": 0.6694122989711708,
      "grad_norm": 0.859956094224656,
      "learning_rate": 2.6019948609285724e-06,
      "loss": 0.3991,
      "step": 10687
    },
    {
      "epoch": 0.6694749369705132,
      "grad_norm": 0.8778073192113568,
      "learning_rate": 2.601104788983723e-06,
      "loss": 0.3997,
      "step": 10688
    },
    {
      "epoch": 0.6695375749698554,
      "grad_norm": 0.8531956432928838,
      "learning_rate": 2.6002148157778458e-06,
      "loss": 0.4262,
      "step": 10689
    },
    {
      "epoch": 0.6696002129691978,
      "grad_norm": 0.8254857281477964,
      "learning_rate": 2.5993249413475702e-06,
      "loss": 0.3946,
      "step": 10690
    },
    {
      "epoch": 0.6696628509685401,
      "grad_norm": 0.7745020143648099,
      "learning_rate": 2.5984351657295274e-06,
      "loss": 0.358,
      "step": 10691
    },
    {
      "epoch": 0.6697254889678824,
      "grad_norm": 0.8388475867440472,
      "learning_rate": 2.5975454889603392e-06,
      "loss": 0.3709,
      "step": 10692
    },
    {
      "epoch": 0.6697881269672247,
      "grad_norm": 0.8624695669324024,
      "learning_rate": 2.596655911076624e-06,
      "loss": 0.4255,
      "step": 10693
    },
    {
      "epoch": 0.6698507649665669,
      "grad_norm": 0.9245339500477394,
      "learning_rate": 2.5957664321149966e-06,
      "loss": 0.4135,
      "step": 10694
    },
    {
      "epoch": 0.6699134029659093,
      "grad_norm": 0.8138449234905999,
      "learning_rate": 2.594877052112071e-06,
      "loss": 0.3966,
      "step": 10695
    },
    {
      "epoch": 0.6699760409652515,
      "grad_norm": 0.8675268239885912,
      "learning_rate": 2.5939877711044513e-06,
      "loss": 0.3956,
      "step": 10696
    },
    {
      "epoch": 0.6700386789645939,
      "grad_norm": 0.8564459536150926,
      "learning_rate": 2.593098589128743e-06,
      "loss": 0.3927,
      "step": 10697
    },
    {
      "epoch": 0.6701013169639362,
      "grad_norm": 0.8247550633650648,
      "learning_rate": 2.5922095062215426e-06,
      "loss": 0.3498,
      "step": 10698
    },
    {
      "epoch": 0.6701639549632785,
      "grad_norm": 0.8811524930022312,
      "learning_rate": 2.5913205224194437e-06,
      "loss": 0.3929,
      "step": 10699
    },
    {
      "epoch": 0.6702265929626208,
      "grad_norm": 0.7734706816935365,
      "learning_rate": 2.5904316377590416e-06,
      "loss": 0.3351,
      "step": 10700
    },
    {
      "epoch": 0.670289230961963,
      "grad_norm": 0.8385084337880891,
      "learning_rate": 2.5895428522769204e-06,
      "loss": 0.3752,
      "step": 10701
    },
    {
      "epoch": 0.6703518689613054,
      "grad_norm": 0.8178245750000868,
      "learning_rate": 2.5886541660096624e-06,
      "loss": 0.3911,
      "step": 10702
    },
    {
      "epoch": 0.6704145069606476,
      "grad_norm": 0.8764153628977895,
      "learning_rate": 2.587765578993845e-06,
      "loss": 0.4101,
      "step": 10703
    },
    {
      "epoch": 0.67047714495999,
      "grad_norm": 0.8604546646200988,
      "learning_rate": 2.5868770912660456e-06,
      "loss": 0.3629,
      "step": 10704
    },
    {
      "epoch": 0.6705397829593323,
      "grad_norm": 0.8075245222800462,
      "learning_rate": 2.5859887028628337e-06,
      "loss": 0.3869,
      "step": 10705
    },
    {
      "epoch": 0.6706024209586746,
      "grad_norm": 0.9514977720281678,
      "learning_rate": 2.5851004138207747e-06,
      "loss": 0.3915,
      "step": 10706
    },
    {
      "epoch": 0.6706650589580169,
      "grad_norm": 0.8040394884027999,
      "learning_rate": 2.584212224176429e-06,
      "loss": 0.3602,
      "step": 10707
    },
    {
      "epoch": 0.6707276969573592,
      "grad_norm": 0.8652738387190789,
      "learning_rate": 2.583324133966358e-06,
      "loss": 0.3549,
      "step": 10708
    },
    {
      "epoch": 0.6707903349567015,
      "grad_norm": 0.6270474573896456,
      "learning_rate": 2.582436143227114e-06,
      "loss": 0.4411,
      "step": 10709
    },
    {
      "epoch": 0.6708529729560437,
      "grad_norm": 0.8446094654331299,
      "learning_rate": 2.581548251995247e-06,
      "loss": 0.4173,
      "step": 10710
    },
    {
      "epoch": 0.6709156109553861,
      "grad_norm": 0.8493896520558735,
      "learning_rate": 2.580660460307304e-06,
      "loss": 0.3487,
      "step": 10711
    },
    {
      "epoch": 0.6709782489547284,
      "grad_norm": 0.889849484861892,
      "learning_rate": 2.579772768199823e-06,
      "loss": 0.4253,
      "step": 10712
    },
    {
      "epoch": 0.6710408869540707,
      "grad_norm": 0.8064636139504762,
      "learning_rate": 2.5788851757093457e-06,
      "loss": 0.3883,
      "step": 10713
    },
    {
      "epoch": 0.671103524953413,
      "grad_norm": 0.8795587906476979,
      "learning_rate": 2.577997682872404e-06,
      "loss": 0.4085,
      "step": 10714
    },
    {
      "epoch": 0.6711661629527553,
      "grad_norm": 0.8645616763583072,
      "learning_rate": 2.577110289725527e-06,
      "loss": 0.3885,
      "step": 10715
    },
    {
      "epoch": 0.6712288009520976,
      "grad_norm": 0.8592897144151562,
      "learning_rate": 2.5762229963052393e-06,
      "loss": 0.3529,
      "step": 10716
    },
    {
      "epoch": 0.67129143895144,
      "grad_norm": 0.8162050781732761,
      "learning_rate": 2.575335802648064e-06,
      "loss": 0.3757,
      "step": 10717
    },
    {
      "epoch": 0.6713540769507822,
      "grad_norm": 0.7421994649708512,
      "learning_rate": 2.574448708790518e-06,
      "loss": 0.3678,
      "step": 10718
    },
    {
      "epoch": 0.6714167149501244,
      "grad_norm": 0.876534348910957,
      "learning_rate": 2.5735617147691123e-06,
      "loss": 0.4388,
      "step": 10719
    },
    {
      "epoch": 0.6714793529494668,
      "grad_norm": 0.886610446090802,
      "learning_rate": 2.572674820620357e-06,
      "loss": 0.4109,
      "step": 10720
    },
    {
      "epoch": 0.6715419909488091,
      "grad_norm": 0.8290731192768549,
      "learning_rate": 2.571788026380757e-06,
      "loss": 0.3895,
      "step": 10721
    },
    {
      "epoch": 0.6716046289481514,
      "grad_norm": 0.8364320159018006,
      "learning_rate": 2.5709013320868103e-06,
      "loss": 0.3946,
      "step": 10722
    },
    {
      "epoch": 0.6716672669474937,
      "grad_norm": 0.8432750164878831,
      "learning_rate": 2.5700147377750173e-06,
      "loss": 0.3907,
      "step": 10723
    },
    {
      "epoch": 0.671729904946836,
      "grad_norm": 0.8400358105219738,
      "learning_rate": 2.569128243481869e-06,
      "loss": 0.3776,
      "step": 10724
    },
    {
      "epoch": 0.6717925429461783,
      "grad_norm": 0.5428813801186494,
      "learning_rate": 2.5682418492438514e-06,
      "loss": 0.4256,
      "step": 10725
    },
    {
      "epoch": 0.6718551809455205,
      "grad_norm": 0.8030071865289986,
      "learning_rate": 2.567355555097453e-06,
      "loss": 0.3786,
      "step": 10726
    },
    {
      "epoch": 0.6719178189448629,
      "grad_norm": 0.7707166441989749,
      "learning_rate": 2.5664693610791515e-06,
      "loss": 0.3922,
      "step": 10727
    },
    {
      "epoch": 0.6719804569442052,
      "grad_norm": 0.892654029568887,
      "learning_rate": 2.565583267225422e-06,
      "loss": 0.4242,
      "step": 10728
    },
    {
      "epoch": 0.6720430949435475,
      "grad_norm": 0.8028247303219977,
      "learning_rate": 2.5646972735727378e-06,
      "loss": 0.3869,
      "step": 10729
    },
    {
      "epoch": 0.6721057329428898,
      "grad_norm": 0.8872107015294696,
      "learning_rate": 2.563811380157565e-06,
      "loss": 0.4042,
      "step": 10730
    },
    {
      "epoch": 0.6721683709422321,
      "grad_norm": 0.8971235263024111,
      "learning_rate": 2.562925587016367e-06,
      "loss": 0.4159,
      "step": 10731
    },
    {
      "epoch": 0.6722310089415744,
      "grad_norm": 0.7924067646747152,
      "learning_rate": 2.5620398941856055e-06,
      "loss": 0.3732,
      "step": 10732
    },
    {
      "epoch": 0.6722936469409168,
      "grad_norm": 0.841714177396116,
      "learning_rate": 2.5611543017017328e-06,
      "loss": 0.4158,
      "step": 10733
    },
    {
      "epoch": 0.672356284940259,
      "grad_norm": 0.7815945760751002,
      "learning_rate": 2.5602688096012033e-06,
      "loss": 0.3725,
      "step": 10734
    },
    {
      "epoch": 0.6724189229396013,
      "grad_norm": 0.8706718635735867,
      "learning_rate": 2.559383417920463e-06,
      "loss": 0.4056,
      "step": 10735
    },
    {
      "epoch": 0.6724815609389436,
      "grad_norm": 0.8205270255949842,
      "learning_rate": 2.558498126695954e-06,
      "loss": 0.3887,
      "step": 10736
    },
    {
      "epoch": 0.6725441989382859,
      "grad_norm": 0.8169579533569875,
      "learning_rate": 2.557612935964115e-06,
      "loss": 0.4028,
      "step": 10737
    },
    {
      "epoch": 0.6726068369376282,
      "grad_norm": 0.8685144825128706,
      "learning_rate": 2.556727845761381e-06,
      "loss": 0.4066,
      "step": 10738
    },
    {
      "epoch": 0.6726694749369705,
      "grad_norm": 0.877074533196129,
      "learning_rate": 2.555842856124182e-06,
      "loss": 0.404,
      "step": 10739
    },
    {
      "epoch": 0.6727321129363129,
      "grad_norm": 0.7958350558644522,
      "learning_rate": 2.5549579670889433e-06,
      "loss": 0.3936,
      "step": 10740
    },
    {
      "epoch": 0.6727947509356551,
      "grad_norm": 0.79808732601827,
      "learning_rate": 2.554073178692091e-06,
      "loss": 0.3735,
      "step": 10741
    },
    {
      "epoch": 0.6728573889349975,
      "grad_norm": 0.8400980980127353,
      "learning_rate": 2.5531884909700377e-06,
      "loss": 0.378,
      "step": 10742
    },
    {
      "epoch": 0.6729200269343397,
      "grad_norm": 0.7848487052065268,
      "learning_rate": 2.5523039039592025e-06,
      "loss": 0.3919,
      "step": 10743
    },
    {
      "epoch": 0.672982664933682,
      "grad_norm": 0.8574123954108184,
      "learning_rate": 2.551419417695993e-06,
      "loss": 0.4141,
      "step": 10744
    },
    {
      "epoch": 0.6730453029330243,
      "grad_norm": 0.8577679628855537,
      "learning_rate": 2.550535032216815e-06,
      "loss": 0.404,
      "step": 10745
    },
    {
      "epoch": 0.6731079409323666,
      "grad_norm": 0.7880892789891811,
      "learning_rate": 2.549650747558069e-06,
      "loss": 0.3913,
      "step": 10746
    },
    {
      "epoch": 0.673170578931709,
      "grad_norm": 0.8893753533833559,
      "learning_rate": 2.548766563756154e-06,
      "loss": 0.3875,
      "step": 10747
    },
    {
      "epoch": 0.6732332169310512,
      "grad_norm": 0.8780644920673786,
      "learning_rate": 2.5478824808474613e-06,
      "loss": 0.4392,
      "step": 10748
    },
    {
      "epoch": 0.6732958549303936,
      "grad_norm": 0.6127433578572419,
      "learning_rate": 2.5469984988683792e-06,
      "loss": 0.4428,
      "step": 10749
    },
    {
      "epoch": 0.6733584929297358,
      "grad_norm": 0.8303817458605341,
      "learning_rate": 2.5461146178552966e-06,
      "loss": 0.3885,
      "step": 10750
    },
    {
      "epoch": 0.6734211309290781,
      "grad_norm": 0.8206208944194407,
      "learning_rate": 2.545230837844589e-06,
      "loss": 0.3794,
      "step": 10751
    },
    {
      "epoch": 0.6734837689284204,
      "grad_norm": 0.7963244199442371,
      "learning_rate": 2.544347158872638e-06,
      "loss": 0.3825,
      "step": 10752
    },
    {
      "epoch": 0.6735464069277627,
      "grad_norm": 0.9492123392774926,
      "learning_rate": 2.543463580975814e-06,
      "loss": 0.3914,
      "step": 10753
    },
    {
      "epoch": 0.673609044927105,
      "grad_norm": 0.8574946142819554,
      "learning_rate": 2.5425801041904853e-06,
      "loss": 0.3828,
      "step": 10754
    },
    {
      "epoch": 0.6736716829264473,
      "grad_norm": 0.8428136066042107,
      "learning_rate": 2.541696728553015e-06,
      "loss": 0.3672,
      "step": 10755
    },
    {
      "epoch": 0.6737343209257897,
      "grad_norm": 0.8159117503969284,
      "learning_rate": 2.5408134540997633e-06,
      "loss": 0.3607,
      "step": 10756
    },
    {
      "epoch": 0.6737969589251319,
      "grad_norm": 0.9176953020683433,
      "learning_rate": 2.5399302808670866e-06,
      "loss": 0.4117,
      "step": 10757
    },
    {
      "epoch": 0.6738595969244743,
      "grad_norm": 0.8849556536773068,
      "learning_rate": 2.5390472088913344e-06,
      "loss": 0.4176,
      "step": 10758
    },
    {
      "epoch": 0.6739222349238165,
      "grad_norm": 0.8775490175786548,
      "learning_rate": 2.5381642382088556e-06,
      "loss": 0.3786,
      "step": 10759
    },
    {
      "epoch": 0.6739848729231588,
      "grad_norm": 0.7914848050719703,
      "learning_rate": 2.5372813688559963e-06,
      "loss": 0.375,
      "step": 10760
    },
    {
      "epoch": 0.6740475109225011,
      "grad_norm": 0.7769295840167323,
      "learning_rate": 2.536398600869092e-06,
      "loss": 0.3478,
      "step": 10761
    },
    {
      "epoch": 0.6741101489218434,
      "grad_norm": 0.8047067503875575,
      "learning_rate": 2.535515934284479e-06,
      "loss": 0.3795,
      "step": 10762
    },
    {
      "epoch": 0.6741727869211858,
      "grad_norm": 0.9588042738231642,
      "learning_rate": 2.5346333691384883e-06,
      "loss": 0.3872,
      "step": 10763
    },
    {
      "epoch": 0.674235424920528,
      "grad_norm": 0.8673552843192767,
      "learning_rate": 2.5337509054674448e-06,
      "loss": 0.4286,
      "step": 10764
    },
    {
      "epoch": 0.6742980629198704,
      "grad_norm": 0.8430265844531897,
      "learning_rate": 2.5328685433076723e-06,
      "loss": 0.3664,
      "step": 10765
    },
    {
      "epoch": 0.6743607009192126,
      "grad_norm": 0.7334282199021954,
      "learning_rate": 2.531986282695488e-06,
      "loss": 0.3401,
      "step": 10766
    },
    {
      "epoch": 0.674423338918555,
      "grad_norm": 0.9005148770257352,
      "learning_rate": 2.5311041236672045e-06,
      "loss": 0.3857,
      "step": 10767
    },
    {
      "epoch": 0.6744859769178972,
      "grad_norm": 0.8439616951596272,
      "learning_rate": 2.5302220662591338e-06,
      "loss": 0.3533,
      "step": 10768
    },
    {
      "epoch": 0.6745486149172395,
      "grad_norm": 0.8572661615093986,
      "learning_rate": 2.5293401105075833e-06,
      "loss": 0.3622,
      "step": 10769
    },
    {
      "epoch": 0.6746112529165819,
      "grad_norm": 0.8248808082167839,
      "learning_rate": 2.5284582564488523e-06,
      "loss": 0.3604,
      "step": 10770
    },
    {
      "epoch": 0.6746738909159241,
      "grad_norm": 0.5981757001779028,
      "learning_rate": 2.5275765041192377e-06,
      "loss": 0.4651,
      "step": 10771
    },
    {
      "epoch": 0.6747365289152665,
      "grad_norm": 0.8091282121284623,
      "learning_rate": 2.526694853555034e-06,
      "loss": 0.3522,
      "step": 10772
    },
    {
      "epoch": 0.6747991669146087,
      "grad_norm": 0.870267650601535,
      "learning_rate": 2.525813304792528e-06,
      "loss": 0.3944,
      "step": 10773
    },
    {
      "epoch": 0.6748618049139511,
      "grad_norm": 0.9130349549404958,
      "learning_rate": 2.524931857868007e-06,
      "loss": 0.3825,
      "step": 10774
    },
    {
      "epoch": 0.6749244429132933,
      "grad_norm": 0.8435989803405092,
      "learning_rate": 2.5240505128177474e-06,
      "loss": 0.4068,
      "step": 10775
    },
    {
      "epoch": 0.6749870809126356,
      "grad_norm": 0.8048443388948254,
      "learning_rate": 2.52316926967803e-06,
      "loss": 0.3393,
      "step": 10776
    },
    {
      "epoch": 0.675049718911978,
      "grad_norm": 0.7797309877631976,
      "learning_rate": 2.5222881284851234e-06,
      "loss": 0.3713,
      "step": 10777
    },
    {
      "epoch": 0.6751123569113202,
      "grad_norm": 0.8629000467717709,
      "learning_rate": 2.5214070892752996e-06,
      "loss": 0.3869,
      "step": 10778
    },
    {
      "epoch": 0.6751749949106626,
      "grad_norm": 0.8424331060292389,
      "learning_rate": 2.52052615208482e-06,
      "loss": 0.3672,
      "step": 10779
    },
    {
      "epoch": 0.6752376329100048,
      "grad_norm": 0.8390574938875655,
      "learning_rate": 2.519645316949944e-06,
      "loss": 0.3986,
      "step": 10780
    },
    {
      "epoch": 0.6753002709093472,
      "grad_norm": 0.8627121781396145,
      "learning_rate": 2.5187645839069274e-06,
      "loss": 0.4424,
      "step": 10781
    },
    {
      "epoch": 0.6753629089086894,
      "grad_norm": 0.6610000969192578,
      "learning_rate": 2.5178839529920207e-06,
      "loss": 0.4538,
      "step": 10782
    },
    {
      "epoch": 0.6754255469080318,
      "grad_norm": 0.8147243881532583,
      "learning_rate": 2.517003424241471e-06,
      "loss": 0.3987,
      "step": 10783
    },
    {
      "epoch": 0.6754881849073741,
      "grad_norm": 0.8152209209768609,
      "learning_rate": 2.5161229976915196e-06,
      "loss": 0.3735,
      "step": 10784
    },
    {
      "epoch": 0.6755508229067163,
      "grad_norm": 0.8078338142214772,
      "learning_rate": 2.5152426733784083e-06,
      "loss": 0.3767,
      "step": 10785
    },
    {
      "epoch": 0.6756134609060587,
      "grad_norm": 0.8663405914080429,
      "learning_rate": 2.5143624513383683e-06,
      "loss": 0.3908,
      "step": 10786
    },
    {
      "epoch": 0.6756760989054009,
      "grad_norm": 0.8301393516687318,
      "learning_rate": 2.5134823316076328e-06,
      "loss": 0.3772,
      "step": 10787
    },
    {
      "epoch": 0.6757387369047433,
      "grad_norm": 0.9383968413695656,
      "learning_rate": 2.5126023142224266e-06,
      "loss": 0.4172,
      "step": 10788
    },
    {
      "epoch": 0.6758013749040855,
      "grad_norm": 0.8856415051079414,
      "learning_rate": 2.5117223992189704e-06,
      "loss": 0.4266,
      "step": 10789
    },
    {
      "epoch": 0.6758640129034279,
      "grad_norm": 0.8806111794919843,
      "learning_rate": 2.510842586633482e-06,
      "loss": 0.4343,
      "step": 10790
    },
    {
      "epoch": 0.6759266509027702,
      "grad_norm": 0.9202710732792349,
      "learning_rate": 2.509962876502175e-06,
      "loss": 0.4064,
      "step": 10791
    },
    {
      "epoch": 0.6759892889021125,
      "grad_norm": 0.96947254629041,
      "learning_rate": 2.509083268861256e-06,
      "loss": 0.4069,
      "step": 10792
    },
    {
      "epoch": 0.6760519269014548,
      "grad_norm": 0.8396737872297189,
      "learning_rate": 2.5082037637469337e-06,
      "loss": 0.3651,
      "step": 10793
    },
    {
      "epoch": 0.676114564900797,
      "grad_norm": 0.8575889772477862,
      "learning_rate": 2.507324361195408e-06,
      "loss": 0.3818,
      "step": 10794
    },
    {
      "epoch": 0.6761772029001394,
      "grad_norm": 0.671050166426404,
      "learning_rate": 2.5064450612428716e-06,
      "loss": 0.4582,
      "step": 10795
    },
    {
      "epoch": 0.6762398408994816,
      "grad_norm": 0.6400454360017604,
      "learning_rate": 2.505565863925521e-06,
      "loss": 0.4665,
      "step": 10796
    },
    {
      "epoch": 0.676302478898824,
      "grad_norm": 0.8566058682532637,
      "learning_rate": 2.504686769279542e-06,
      "loss": 0.3877,
      "step": 10797
    },
    {
      "epoch": 0.6763651168981663,
      "grad_norm": 0.8889441340956707,
      "learning_rate": 2.5038077773411194e-06,
      "loss": 0.3894,
      "step": 10798
    },
    {
      "epoch": 0.6764277548975086,
      "grad_norm": 0.8138552263739688,
      "learning_rate": 2.502928888146432e-06,
      "loss": 0.3718,
      "step": 10799
    },
    {
      "epoch": 0.6764903928968509,
      "grad_norm": 0.8752402370097337,
      "learning_rate": 2.5020501017316555e-06,
      "loss": 0.401,
      "step": 10800
    },
    {
      "epoch": 0.6765530308961932,
      "grad_norm": 0.8183395673152927,
      "learning_rate": 2.501171418132958e-06,
      "loss": 0.36,
      "step": 10801
    },
    {
      "epoch": 0.6766156688955355,
      "grad_norm": 0.9086791897950403,
      "learning_rate": 2.500292837386511e-06,
      "loss": 0.3982,
      "step": 10802
    },
    {
      "epoch": 0.6766783068948777,
      "grad_norm": 0.8248196691656104,
      "learning_rate": 2.4994143595284743e-06,
      "loss": 0.3946,
      "step": 10803
    },
    {
      "epoch": 0.6767409448942201,
      "grad_norm": 0.8630610874549276,
      "learning_rate": 2.4985359845950046e-06,
      "loss": 0.4007,
      "step": 10804
    },
    {
      "epoch": 0.6768035828935623,
      "grad_norm": 0.8318710535256063,
      "learning_rate": 2.4976577126222606e-06,
      "loss": 0.3582,
      "step": 10805
    },
    {
      "epoch": 0.6768662208929047,
      "grad_norm": 0.8790474312651967,
      "learning_rate": 2.4967795436463895e-06,
      "loss": 0.342,
      "step": 10806
    },
    {
      "epoch": 0.676928858892247,
      "grad_norm": 0.9068569737661667,
      "learning_rate": 2.495901477703537e-06,
      "loss": 0.4143,
      "step": 10807
    },
    {
      "epoch": 0.6769914968915893,
      "grad_norm": 0.8125131248094245,
      "learning_rate": 2.495023514829844e-06,
      "loss": 0.3735,
      "step": 10808
    },
    {
      "epoch": 0.6770541348909316,
      "grad_norm": 0.8320985783334076,
      "learning_rate": 2.4941456550614485e-06,
      "loss": 0.3824,
      "step": 10809
    },
    {
      "epoch": 0.6771167728902738,
      "grad_norm": 0.8551305623488387,
      "learning_rate": 2.493267898434481e-06,
      "loss": 0.4052,
      "step": 10810
    },
    {
      "epoch": 0.6771794108896162,
      "grad_norm": 0.8725580248859187,
      "learning_rate": 2.492390244985074e-06,
      "loss": 0.4051,
      "step": 10811
    },
    {
      "epoch": 0.6772420488889584,
      "grad_norm": 0.8144793648981948,
      "learning_rate": 2.4915126947493496e-06,
      "loss": 0.3684,
      "step": 10812
    },
    {
      "epoch": 0.6773046868883008,
      "grad_norm": 0.7706573038020891,
      "learning_rate": 2.4906352477634287e-06,
      "loss": 0.3679,
      "step": 10813
    },
    {
      "epoch": 0.6773673248876431,
      "grad_norm": 0.8349434621904186,
      "learning_rate": 2.4897579040634246e-06,
      "loss": 0.3716,
      "step": 10814
    },
    {
      "epoch": 0.6774299628869854,
      "grad_norm": 0.8606512870200662,
      "learning_rate": 2.4888806636854536e-06,
      "loss": 0.3854,
      "step": 10815
    },
    {
      "epoch": 0.6774926008863277,
      "grad_norm": 0.7314160286778257,
      "learning_rate": 2.48800352666562e-06,
      "loss": 0.3462,
      "step": 10816
    },
    {
      "epoch": 0.67755523888567,
      "grad_norm": 0.7913200801156506,
      "learning_rate": 2.4871264930400275e-06,
      "loss": 0.36,
      "step": 10817
    },
    {
      "epoch": 0.6776178768850123,
      "grad_norm": 0.8882209560996701,
      "learning_rate": 2.486249562844773e-06,
      "loss": 0.4134,
      "step": 10818
    },
    {
      "epoch": 0.6776805148843545,
      "grad_norm": 0.8950935884998076,
      "learning_rate": 2.485372736115955e-06,
      "loss": 0.4362,
      "step": 10819
    },
    {
      "epoch": 0.6777431528836969,
      "grad_norm": 0.8719037498723706,
      "learning_rate": 2.4844960128896625e-06,
      "loss": 0.4018,
      "step": 10820
    },
    {
      "epoch": 0.6778057908830392,
      "grad_norm": 0.8903671988188038,
      "learning_rate": 2.48361939320198e-06,
      "loss": 0.4469,
      "step": 10821
    },
    {
      "epoch": 0.6778684288823815,
      "grad_norm": 0.7844147512964819,
      "learning_rate": 2.482742877088991e-06,
      "loss": 0.3627,
      "step": 10822
    },
    {
      "epoch": 0.6779310668817238,
      "grad_norm": 0.796158502845056,
      "learning_rate": 2.48186646458677e-06,
      "loss": 0.3602,
      "step": 10823
    },
    {
      "epoch": 0.6779937048810661,
      "grad_norm": 0.9258049921028516,
      "learning_rate": 2.480990155731395e-06,
      "loss": 0.4395,
      "step": 10824
    },
    {
      "epoch": 0.6780563428804084,
      "grad_norm": 0.8696161930803189,
      "learning_rate": 2.4801139505589327e-06,
      "loss": 0.3938,
      "step": 10825
    },
    {
      "epoch": 0.6781189808797508,
      "grad_norm": 0.8211082544609482,
      "learning_rate": 2.479237849105448e-06,
      "loss": 0.4098,
      "step": 10826
    },
    {
      "epoch": 0.678181618879093,
      "grad_norm": 0.8158446180695702,
      "learning_rate": 2.4783618514069994e-06,
      "loss": 0.403,
      "step": 10827
    },
    {
      "epoch": 0.6782442568784353,
      "grad_norm": 0.7608475771450403,
      "learning_rate": 2.4774859574996467e-06,
      "loss": 0.3655,
      "step": 10828
    },
    {
      "epoch": 0.6783068948777776,
      "grad_norm": 0.8947193760506698,
      "learning_rate": 2.4766101674194404e-06,
      "loss": 0.3666,
      "step": 10829
    },
    {
      "epoch": 0.6783695328771199,
      "grad_norm": 0.9871457967135265,
      "learning_rate": 2.475734481202428e-06,
      "loss": 0.4224,
      "step": 10830
    },
    {
      "epoch": 0.6784321708764622,
      "grad_norm": 0.9148316616069587,
      "learning_rate": 2.4748588988846526e-06,
      "loss": 0.4104,
      "step": 10831
    },
    {
      "epoch": 0.6784948088758045,
      "grad_norm": 0.7798319508892482,
      "learning_rate": 2.4739834205021524e-06,
      "loss": 0.3445,
      "step": 10832
    },
    {
      "epoch": 0.6785574468751469,
      "grad_norm": 0.8659819269190102,
      "learning_rate": 2.4731080460909644e-06,
      "loss": 0.3513,
      "step": 10833
    },
    {
      "epoch": 0.6786200848744891,
      "grad_norm": 0.8941020051077083,
      "learning_rate": 2.472232775687119e-06,
      "loss": 0.4044,
      "step": 10834
    },
    {
      "epoch": 0.6786827228738314,
      "grad_norm": 0.8591172334657698,
      "learning_rate": 2.4713576093266413e-06,
      "loss": 0.3901,
      "step": 10835
    },
    {
      "epoch": 0.6787453608731737,
      "grad_norm": 0.9017943400310083,
      "learning_rate": 2.470482547045552e-06,
      "loss": 0.3909,
      "step": 10836
    },
    {
      "epoch": 0.678807998872516,
      "grad_norm": 0.8630941692820906,
      "learning_rate": 2.4696075888798724e-06,
      "loss": 0.3931,
      "step": 10837
    },
    {
      "epoch": 0.6788706368718583,
      "grad_norm": 0.8422429667702894,
      "learning_rate": 2.4687327348656144e-06,
      "loss": 0.4135,
      "step": 10838
    },
    {
      "epoch": 0.6789332748712006,
      "grad_norm": 0.8462642784374209,
      "learning_rate": 2.467857985038786e-06,
      "loss": 0.3834,
      "step": 10839
    },
    {
      "epoch": 0.678995912870543,
      "grad_norm": 0.8190071754943492,
      "learning_rate": 2.4669833394353936e-06,
      "loss": 0.3628,
      "step": 10840
    },
    {
      "epoch": 0.6790585508698852,
      "grad_norm": 0.5774231239284378,
      "learning_rate": 2.4661087980914354e-06,
      "loss": 0.4617,
      "step": 10841
    },
    {
      "epoch": 0.6791211888692276,
      "grad_norm": 0.9779019887041442,
      "learning_rate": 2.465234361042911e-06,
      "loss": 0.4272,
      "step": 10842
    },
    {
      "epoch": 0.6791838268685698,
      "grad_norm": 0.8843630414697256,
      "learning_rate": 2.4643600283258107e-06,
      "loss": 0.4059,
      "step": 10843
    },
    {
      "epoch": 0.6792464648679121,
      "grad_norm": 0.9400910775252487,
      "learning_rate": 2.463485799976121e-06,
      "loss": 0.4232,
      "step": 10844
    },
    {
      "epoch": 0.6793091028672544,
      "grad_norm": 0.8244832459035792,
      "learning_rate": 2.4626116760298274e-06,
      "loss": 0.3984,
      "step": 10845
    },
    {
      "epoch": 0.6793717408665967,
      "grad_norm": 0.844896274080171,
      "learning_rate": 2.461737656522909e-06,
      "loss": 0.3834,
      "step": 10846
    },
    {
      "epoch": 0.679434378865939,
      "grad_norm": 0.794479970444011,
      "learning_rate": 2.4608637414913393e-06,
      "loss": 0.3443,
      "step": 10847
    },
    {
      "epoch": 0.6794970168652813,
      "grad_norm": 0.8283705363779441,
      "learning_rate": 2.4599899309710897e-06,
      "loss": 0.3426,
      "step": 10848
    },
    {
      "epoch": 0.6795596548646237,
      "grad_norm": 0.6400450598513979,
      "learning_rate": 2.4591162249981253e-06,
      "loss": 0.4392,
      "step": 10849
    },
    {
      "epoch": 0.6796222928639659,
      "grad_norm": 0.8554999911341441,
      "learning_rate": 2.4582426236084074e-06,
      "loss": 0.3818,
      "step": 10850
    },
    {
      "epoch": 0.6796849308633083,
      "grad_norm": 0.8504907948021365,
      "learning_rate": 2.457369126837897e-06,
      "loss": 0.3767,
      "step": 10851
    },
    {
      "epoch": 0.6797475688626505,
      "grad_norm": 0.6117339965921168,
      "learning_rate": 2.456495734722544e-06,
      "loss": 0.4376,
      "step": 10852
    },
    {
      "epoch": 0.6798102068619928,
      "grad_norm": 0.8204758889353534,
      "learning_rate": 2.455622447298298e-06,
      "loss": 0.3995,
      "step": 10853
    },
    {
      "epoch": 0.6798728448613351,
      "grad_norm": 0.9170243760988218,
      "learning_rate": 2.454749264601106e-06,
      "loss": 0.357,
      "step": 10854
    },
    {
      "epoch": 0.6799354828606774,
      "grad_norm": 0.9444422886923038,
      "learning_rate": 2.453876186666907e-06,
      "loss": 0.4238,
      "step": 10855
    },
    {
      "epoch": 0.6799981208600198,
      "grad_norm": 0.8520737912150894,
      "learning_rate": 2.453003213531636e-06,
      "loss": 0.3777,
      "step": 10856
    },
    {
      "epoch": 0.680060758859362,
      "grad_norm": 0.8762693730262922,
      "learning_rate": 2.4521303452312257e-06,
      "loss": 0.3775,
      "step": 10857
    },
    {
      "epoch": 0.6801233968587044,
      "grad_norm": 0.8363229217664236,
      "learning_rate": 2.451257581801603e-06,
      "loss": 0.3772,
      "step": 10858
    },
    {
      "epoch": 0.6801860348580466,
      "grad_norm": 0.8039009748714578,
      "learning_rate": 2.450384923278692e-06,
      "loss": 0.3766,
      "step": 10859
    },
    {
      "epoch": 0.6802486728573889,
      "grad_norm": 0.8401419288430545,
      "learning_rate": 2.4495123696984086e-06,
      "loss": 0.3933,
      "step": 10860
    },
    {
      "epoch": 0.6803113108567312,
      "grad_norm": 0.8166950716137398,
      "learning_rate": 2.4486399210966715e-06,
      "loss": 0.3843,
      "step": 10861
    },
    {
      "epoch": 0.6803739488560735,
      "grad_norm": 0.8597883653797992,
      "learning_rate": 2.4477675775093867e-06,
      "loss": 0.3241,
      "step": 10862
    },
    {
      "epoch": 0.6804365868554159,
      "grad_norm": 0.8681471218556395,
      "learning_rate": 2.446895338972464e-06,
      "loss": 0.3839,
      "step": 10863
    },
    {
      "epoch": 0.6804992248547581,
      "grad_norm": 0.8999951136156364,
      "learning_rate": 2.446023205521803e-06,
      "loss": 0.4037,
      "step": 10864
    },
    {
      "epoch": 0.6805618628541005,
      "grad_norm": 0.9325571580101071,
      "learning_rate": 2.445151177193301e-06,
      "loss": 0.4165,
      "step": 10865
    },
    {
      "epoch": 0.6806245008534427,
      "grad_norm": 0.9313463148239653,
      "learning_rate": 2.4442792540228512e-06,
      "loss": 0.4011,
      "step": 10866
    },
    {
      "epoch": 0.6806871388527851,
      "grad_norm": 0.6767057115057568,
      "learning_rate": 2.4434074360463412e-06,
      "loss": 0.4308,
      "step": 10867
    },
    {
      "epoch": 0.6807497768521273,
      "grad_norm": 0.8505091648460905,
      "learning_rate": 2.4425357232996557e-06,
      "loss": 0.412,
      "step": 10868
    },
    {
      "epoch": 0.6808124148514696,
      "grad_norm": 0.8658982624755879,
      "learning_rate": 2.441664115818672e-06,
      "loss": 0.4431,
      "step": 10869
    },
    {
      "epoch": 0.680875052850812,
      "grad_norm": 0.8766193338567385,
      "learning_rate": 2.440792613639269e-06,
      "loss": 0.3886,
      "step": 10870
    },
    {
      "epoch": 0.6809376908501542,
      "grad_norm": 0.8580090178387442,
      "learning_rate": 2.4399212167973185e-06,
      "loss": 0.3779,
      "step": 10871
    },
    {
      "epoch": 0.6810003288494966,
      "grad_norm": 0.816444161871824,
      "learning_rate": 2.4390499253286865e-06,
      "loss": 0.3798,
      "step": 10872
    },
    {
      "epoch": 0.6810629668488388,
      "grad_norm": 0.8367286941629094,
      "learning_rate": 2.438178739269234e-06,
      "loss": 0.3916,
      "step": 10873
    },
    {
      "epoch": 0.6811256048481812,
      "grad_norm": 0.8125316815359366,
      "learning_rate": 2.4373076586548204e-06,
      "loss": 0.4,
      "step": 10874
    },
    {
      "epoch": 0.6811882428475234,
      "grad_norm": 0.8667271008051443,
      "learning_rate": 2.4364366835213e-06,
      "loss": 0.4148,
      "step": 10875
    },
    {
      "epoch": 0.6812508808468658,
      "grad_norm": 0.8355740616027033,
      "learning_rate": 2.4355658139045206e-06,
      "loss": 0.3988,
      "step": 10876
    },
    {
      "epoch": 0.681313518846208,
      "grad_norm": 0.7958512095325985,
      "learning_rate": 2.4346950498403265e-06,
      "loss": 0.3424,
      "step": 10877
    },
    {
      "epoch": 0.6813761568455503,
      "grad_norm": 0.8313151440050057,
      "learning_rate": 2.4338243913645625e-06,
      "loss": 0.3698,
      "step": 10878
    },
    {
      "epoch": 0.6814387948448927,
      "grad_norm": 0.7684982647577663,
      "learning_rate": 2.432953838513061e-06,
      "loss": 0.3593,
      "step": 10879
    },
    {
      "epoch": 0.6815014328442349,
      "grad_norm": 0.6002941270374162,
      "learning_rate": 2.4320833913216573e-06,
      "loss": 0.4559,
      "step": 10880
    },
    {
      "epoch": 0.6815640708435773,
      "grad_norm": 0.7974719038087044,
      "learning_rate": 2.431213049826179e-06,
      "loss": 0.3677,
      "step": 10881
    },
    {
      "epoch": 0.6816267088429195,
      "grad_norm": 0.9365596903228881,
      "learning_rate": 2.430342814062447e-06,
      "loss": 0.3788,
      "step": 10882
    },
    {
      "epoch": 0.6816893468422619,
      "grad_norm": 0.8273257522461945,
      "learning_rate": 2.4294726840662822e-06,
      "loss": 0.3672,
      "step": 10883
    },
    {
      "epoch": 0.6817519848416042,
      "grad_norm": 0.8371836031723286,
      "learning_rate": 2.4286026598734993e-06,
      "loss": 0.3362,
      "step": 10884
    },
    {
      "epoch": 0.6818146228409464,
      "grad_norm": 0.8621486838958403,
      "learning_rate": 2.4277327415199077e-06,
      "loss": 0.4048,
      "step": 10885
    },
    {
      "epoch": 0.6818772608402888,
      "grad_norm": 0.7815086966656302,
      "learning_rate": 2.426862929041312e-06,
      "loss": 0.3651,
      "step": 10886
    },
    {
      "epoch": 0.681939898839631,
      "grad_norm": 0.7600070047822651,
      "learning_rate": 2.4259932224735176e-06,
      "loss": 0.3426,
      "step": 10887
    },
    {
      "epoch": 0.6820025368389734,
      "grad_norm": 0.8306613494673519,
      "learning_rate": 2.4251236218523188e-06,
      "loss": 0.4069,
      "step": 10888
    },
    {
      "epoch": 0.6820651748383156,
      "grad_norm": 0.9173665061290873,
      "learning_rate": 2.4242541272135107e-06,
      "loss": 0.4075,
      "step": 10889
    },
    {
      "epoch": 0.682127812837658,
      "grad_norm": 0.8868279840430403,
      "learning_rate": 2.423384738592881e-06,
      "loss": 0.4137,
      "step": 10890
    },
    {
      "epoch": 0.6821904508370003,
      "grad_norm": 0.8557437152259709,
      "learning_rate": 2.422515456026214e-06,
      "loss": 0.3806,
      "step": 10891
    },
    {
      "epoch": 0.6822530888363426,
      "grad_norm": 0.8085531915848756,
      "learning_rate": 2.421646279549289e-06,
      "loss": 0.3447,
      "step": 10892
    },
    {
      "epoch": 0.6823157268356849,
      "grad_norm": 0.6341929085572854,
      "learning_rate": 2.4207772091978817e-06,
      "loss": 0.4308,
      "step": 10893
    },
    {
      "epoch": 0.6823783648350271,
      "grad_norm": 0.8913641723112788,
      "learning_rate": 2.419908245007763e-06,
      "loss": 0.4185,
      "step": 10894
    },
    {
      "epoch": 0.6824410028343695,
      "grad_norm": 0.8625116138686819,
      "learning_rate": 2.4190393870146984e-06,
      "loss": 0.4129,
      "step": 10895
    },
    {
      "epoch": 0.6825036408337117,
      "grad_norm": 0.8250874293604501,
      "learning_rate": 2.418170635254451e-06,
      "loss": 0.3525,
      "step": 10896
    },
    {
      "epoch": 0.6825662788330541,
      "grad_norm": 0.8049518206516156,
      "learning_rate": 2.417301989762782e-06,
      "loss": 0.3797,
      "step": 10897
    },
    {
      "epoch": 0.6826289168323963,
      "grad_norm": 0.8632026887241918,
      "learning_rate": 2.4164334505754423e-06,
      "loss": 0.4043,
      "step": 10898
    },
    {
      "epoch": 0.6826915548317387,
      "grad_norm": 0.774107419518758,
      "learning_rate": 2.415565017728181e-06,
      "loss": 0.3365,
      "step": 10899
    },
    {
      "epoch": 0.682754192831081,
      "grad_norm": 0.8467056744801119,
      "learning_rate": 2.4146966912567436e-06,
      "loss": 0.3726,
      "step": 10900
    },
    {
      "epoch": 0.6828168308304233,
      "grad_norm": 0.7879059693643423,
      "learning_rate": 2.41382847119687e-06,
      "loss": 0.333,
      "step": 10901
    },
    {
      "epoch": 0.6828794688297656,
      "grad_norm": 0.8682313165177207,
      "learning_rate": 2.412960357584297e-06,
      "loss": 0.4118,
      "step": 10902
    },
    {
      "epoch": 0.6829421068291078,
      "grad_norm": 0.9095396544027348,
      "learning_rate": 2.4120923504547535e-06,
      "loss": 0.3919,
      "step": 10903
    },
    {
      "epoch": 0.6830047448284502,
      "grad_norm": 0.9143676354276622,
      "learning_rate": 2.4112244498439713e-06,
      "loss": 0.3917,
      "step": 10904
    },
    {
      "epoch": 0.6830673828277924,
      "grad_norm": 0.8689597703100855,
      "learning_rate": 2.410356655787671e-06,
      "loss": 0.4182,
      "step": 10905
    },
    {
      "epoch": 0.6831300208271348,
      "grad_norm": 0.8317110626421638,
      "learning_rate": 2.40948896832157e-06,
      "loss": 0.3846,
      "step": 10906
    },
    {
      "epoch": 0.6831926588264771,
      "grad_norm": 0.8326171040777027,
      "learning_rate": 2.4086213874813852e-06,
      "loss": 0.3718,
      "step": 10907
    },
    {
      "epoch": 0.6832552968258194,
      "grad_norm": 0.7911248554026503,
      "learning_rate": 2.4077539133028253e-06,
      "loss": 0.3634,
      "step": 10908
    },
    {
      "epoch": 0.6833179348251617,
      "grad_norm": 0.8586768667837953,
      "learning_rate": 2.4068865458215957e-06,
      "loss": 0.3853,
      "step": 10909
    },
    {
      "epoch": 0.683380572824504,
      "grad_norm": 0.8381952700508787,
      "learning_rate": 2.406019285073397e-06,
      "loss": 0.3816,
      "step": 10910
    },
    {
      "epoch": 0.6834432108238463,
      "grad_norm": 0.8591760273454863,
      "learning_rate": 2.4051521310939258e-06,
      "loss": 0.422,
      "step": 10911
    },
    {
      "epoch": 0.6835058488231885,
      "grad_norm": 0.914113362371902,
      "learning_rate": 2.404285083918873e-06,
      "loss": 0.3728,
      "step": 10912
    },
    {
      "epoch": 0.6835684868225309,
      "grad_norm": 0.8430349673151784,
      "learning_rate": 2.4034181435839293e-06,
      "loss": 0.3967,
      "step": 10913
    },
    {
      "epoch": 0.6836311248218732,
      "grad_norm": 0.8445450479863573,
      "learning_rate": 2.402551310124777e-06,
      "loss": 0.3842,
      "step": 10914
    },
    {
      "epoch": 0.6836937628212155,
      "grad_norm": 0.8606460894259514,
      "learning_rate": 2.4016845835770936e-06,
      "loss": 0.4054,
      "step": 10915
    },
    {
      "epoch": 0.6837564008205578,
      "grad_norm": 0.7938523104545677,
      "learning_rate": 2.400817963976556e-06,
      "loss": 0.377,
      "step": 10916
    },
    {
      "epoch": 0.6838190388199001,
      "grad_norm": 0.7120114299543655,
      "learning_rate": 2.399951451358834e-06,
      "loss": 0.4556,
      "step": 10917
    },
    {
      "epoch": 0.6838816768192424,
      "grad_norm": 0.896683127261559,
      "learning_rate": 2.3990850457595928e-06,
      "loss": 0.4213,
      "step": 10918
    },
    {
      "epoch": 0.6839443148185846,
      "grad_norm": 0.8510580954177077,
      "learning_rate": 2.3982187472144934e-06,
      "loss": 0.3519,
      "step": 10919
    },
    {
      "epoch": 0.684006952817927,
      "grad_norm": 0.8283578806545085,
      "learning_rate": 2.3973525557591936e-06,
      "loss": 0.3601,
      "step": 10920
    },
    {
      "epoch": 0.6840695908172693,
      "grad_norm": 0.8622372594835915,
      "learning_rate": 2.396486471429344e-06,
      "loss": 0.4037,
      "step": 10921
    },
    {
      "epoch": 0.6841322288166116,
      "grad_norm": 0.7887223009671219,
      "learning_rate": 2.395620494260596e-06,
      "loss": 0.3989,
      "step": 10922
    },
    {
      "epoch": 0.6841948668159539,
      "grad_norm": 0.8325368044031262,
      "learning_rate": 2.3947546242885922e-06,
      "loss": 0.3825,
      "step": 10923
    },
    {
      "epoch": 0.6842575048152962,
      "grad_norm": 0.8393098259654831,
      "learning_rate": 2.39388886154897e-06,
      "loss": 0.3671,
      "step": 10924
    },
    {
      "epoch": 0.6843201428146385,
      "grad_norm": 0.8772255459606986,
      "learning_rate": 2.3930232060773666e-06,
      "loss": 0.3894,
      "step": 10925
    },
    {
      "epoch": 0.6843827808139809,
      "grad_norm": 0.8558076556975458,
      "learning_rate": 2.392157657909413e-06,
      "loss": 0.3696,
      "step": 10926
    },
    {
      "epoch": 0.6844454188133231,
      "grad_norm": 0.8634261790727319,
      "learning_rate": 2.391292217080734e-06,
      "loss": 0.4069,
      "step": 10927
    },
    {
      "epoch": 0.6845080568126654,
      "grad_norm": 0.9354416949339597,
      "learning_rate": 2.3904268836269516e-06,
      "loss": 0.451,
      "step": 10928
    },
    {
      "epoch": 0.6845706948120077,
      "grad_norm": 0.808528148819946,
      "learning_rate": 2.389561657583681e-06,
      "loss": 0.3801,
      "step": 10929
    },
    {
      "epoch": 0.68463333281135,
      "grad_norm": 0.8335656507563161,
      "learning_rate": 2.388696538986539e-06,
      "loss": 0.3481,
      "step": 10930
    },
    {
      "epoch": 0.6846959708106923,
      "grad_norm": 0.8079714972906235,
      "learning_rate": 2.3878315278711322e-06,
      "loss": 0.3683,
      "step": 10931
    },
    {
      "epoch": 0.6847586088100346,
      "grad_norm": 0.8412707046101806,
      "learning_rate": 2.386966624273064e-06,
      "loss": 0.3971,
      "step": 10932
    },
    {
      "epoch": 0.684821246809377,
      "grad_norm": 0.8279513615258339,
      "learning_rate": 2.3861018282279334e-06,
      "loss": 0.399,
      "step": 10933
    },
    {
      "epoch": 0.6848838848087192,
      "grad_norm": 0.8298896284222986,
      "learning_rate": 2.3852371397713385e-06,
      "loss": 0.3649,
      "step": 10934
    },
    {
      "epoch": 0.6849465228080616,
      "grad_norm": 0.7910795588771463,
      "learning_rate": 2.3843725589388684e-06,
      "loss": 0.3896,
      "step": 10935
    },
    {
      "epoch": 0.6850091608074038,
      "grad_norm": 0.8612012385707444,
      "learning_rate": 2.3835080857661092e-06,
      "loss": 0.3914,
      "step": 10936
    },
    {
      "epoch": 0.6850717988067461,
      "grad_norm": 0.8026309954774878,
      "learning_rate": 2.382643720288642e-06,
      "loss": 0.372,
      "step": 10937
    },
    {
      "epoch": 0.6851344368060884,
      "grad_norm": 0.6101630988754337,
      "learning_rate": 2.3817794625420444e-06,
      "loss": 0.4564,
      "step": 10938
    },
    {
      "epoch": 0.6851970748054307,
      "grad_norm": 0.9180737557682271,
      "learning_rate": 2.380915312561891e-06,
      "loss": 0.4198,
      "step": 10939
    },
    {
      "epoch": 0.685259712804773,
      "grad_norm": 0.8774285364439603,
      "learning_rate": 2.3800512703837503e-06,
      "loss": 0.387,
      "step": 10940
    },
    {
      "epoch": 0.6853223508041153,
      "grad_norm": 0.7724177104163577,
      "learning_rate": 2.379187336043185e-06,
      "loss": 0.3548,
      "step": 10941
    },
    {
      "epoch": 0.6853849888034577,
      "grad_norm": 0.8375384802208372,
      "learning_rate": 2.3783235095757535e-06,
      "loss": 0.3764,
      "step": 10942
    },
    {
      "epoch": 0.6854476268027999,
      "grad_norm": 0.8504020523499036,
      "learning_rate": 2.377459791017015e-06,
      "loss": 0.3993,
      "step": 10943
    },
    {
      "epoch": 0.6855102648021422,
      "grad_norm": 0.7857221592774704,
      "learning_rate": 2.3765961804025185e-06,
      "loss": 0.3838,
      "step": 10944
    },
    {
      "epoch": 0.6855729028014845,
      "grad_norm": 0.8660723339848301,
      "learning_rate": 2.3757326777678098e-06,
      "loss": 0.3541,
      "step": 10945
    },
    {
      "epoch": 0.6856355408008268,
      "grad_norm": 0.8647533781869036,
      "learning_rate": 2.374869283148431e-06,
      "loss": 0.4109,
      "step": 10946
    },
    {
      "epoch": 0.6856981788001691,
      "grad_norm": 0.7807281295705972,
      "learning_rate": 2.3740059965799183e-06,
      "loss": 0.3137,
      "step": 10947
    },
    {
      "epoch": 0.6857608167995114,
      "grad_norm": 0.811227086410381,
      "learning_rate": 2.3731428180978076e-06,
      "loss": 0.3673,
      "step": 10948
    },
    {
      "epoch": 0.6858234547988538,
      "grad_norm": 0.8899824950503454,
      "learning_rate": 2.372279747737627e-06,
      "loss": 0.4076,
      "step": 10949
    },
    {
      "epoch": 0.685886092798196,
      "grad_norm": 0.8417453556460308,
      "learning_rate": 2.3714167855348984e-06,
      "loss": 0.3584,
      "step": 10950
    },
    {
      "epoch": 0.6859487307975384,
      "grad_norm": 0.8717860425678141,
      "learning_rate": 2.370553931525144e-06,
      "loss": 0.3894,
      "step": 10951
    },
    {
      "epoch": 0.6860113687968806,
      "grad_norm": 0.8478711619332481,
      "learning_rate": 2.3696911857438754e-06,
      "loss": 0.4047,
      "step": 10952
    },
    {
      "epoch": 0.6860740067962229,
      "grad_norm": 0.8226928544133286,
      "learning_rate": 2.368828548226608e-06,
      "loss": 0.3761,
      "step": 10953
    },
    {
      "epoch": 0.6861366447955652,
      "grad_norm": 0.928060669235318,
      "learning_rate": 2.3679660190088457e-06,
      "loss": 0.4022,
      "step": 10954
    },
    {
      "epoch": 0.6861992827949075,
      "grad_norm": 0.8418024213284435,
      "learning_rate": 2.367103598126088e-06,
      "loss": 0.3941,
      "step": 10955
    },
    {
      "epoch": 0.6862619207942499,
      "grad_norm": 0.8216380770439792,
      "learning_rate": 2.3662412856138376e-06,
      "loss": 0.3883,
      "step": 10956
    },
    {
      "epoch": 0.6863245587935921,
      "grad_norm": 0.8174896497644691,
      "learning_rate": 2.3653790815075846e-06,
      "loss": 0.4285,
      "step": 10957
    },
    {
      "epoch": 0.6863871967929345,
      "grad_norm": 0.8420694108565292,
      "learning_rate": 2.3645169858428173e-06,
      "loss": 0.401,
      "step": 10958
    },
    {
      "epoch": 0.6864498347922767,
      "grad_norm": 0.8666402907726944,
      "learning_rate": 2.3636549986550207e-06,
      "loss": 0.4389,
      "step": 10959
    },
    {
      "epoch": 0.6865124727916191,
      "grad_norm": 0.849199725006827,
      "learning_rate": 2.3627931199796734e-06,
      "loss": 0.4061,
      "step": 10960
    },
    {
      "epoch": 0.6865751107909613,
      "grad_norm": 0.8334490094286281,
      "learning_rate": 2.3619313498522484e-06,
      "loss": 0.3714,
      "step": 10961
    },
    {
      "epoch": 0.6866377487903036,
      "grad_norm": 0.7915763227286001,
      "learning_rate": 2.361069688308221e-06,
      "loss": 0.3812,
      "step": 10962
    },
    {
      "epoch": 0.686700386789646,
      "grad_norm": 0.8501176237201326,
      "learning_rate": 2.3602081353830552e-06,
      "loss": 0.374,
      "step": 10963
    },
    {
      "epoch": 0.6867630247889882,
      "grad_norm": 0.8603522638984965,
      "learning_rate": 2.3593466911122103e-06,
      "loss": 0.3915,
      "step": 10964
    },
    {
      "epoch": 0.6868256627883306,
      "grad_norm": 0.8129731779454826,
      "learning_rate": 2.3584853555311476e-06,
      "loss": 0.3665,
      "step": 10965
    },
    {
      "epoch": 0.6868883007876728,
      "grad_norm": 0.8320161629832601,
      "learning_rate": 2.357624128675318e-06,
      "loss": 0.3971,
      "step": 10966
    },
    {
      "epoch": 0.6869509387870152,
      "grad_norm": 0.8851242171575676,
      "learning_rate": 2.3567630105801696e-06,
      "loss": 0.4332,
      "step": 10967
    },
    {
      "epoch": 0.6870135767863574,
      "grad_norm": 0.8505100259692021,
      "learning_rate": 2.355902001281147e-06,
      "loss": 0.3993,
      "step": 10968
    },
    {
      "epoch": 0.6870762147856997,
      "grad_norm": 0.8055570838165919,
      "learning_rate": 2.355041100813688e-06,
      "loss": 0.3959,
      "step": 10969
    },
    {
      "epoch": 0.687138852785042,
      "grad_norm": 0.8024751614450225,
      "learning_rate": 2.3541803092132264e-06,
      "loss": 0.3298,
      "step": 10970
    },
    {
      "epoch": 0.6872014907843843,
      "grad_norm": 0.8342499946131037,
      "learning_rate": 2.353319626515197e-06,
      "loss": 0.3897,
      "step": 10971
    },
    {
      "epoch": 0.6872641287837267,
      "grad_norm": 0.7958401047160547,
      "learning_rate": 2.3524590527550223e-06,
      "loss": 0.3554,
      "step": 10972
    },
    {
      "epoch": 0.6873267667830689,
      "grad_norm": 0.8038969430516738,
      "learning_rate": 2.3515985879681224e-06,
      "loss": 0.3568,
      "step": 10973
    },
    {
      "epoch": 0.6873894047824113,
      "grad_norm": 0.8956107656186719,
      "learning_rate": 2.3507382321899187e-06,
      "loss": 0.4017,
      "step": 10974
    },
    {
      "epoch": 0.6874520427817535,
      "grad_norm": 0.774010598997273,
      "learning_rate": 2.349877985455821e-06,
      "loss": 0.3351,
      "step": 10975
    },
    {
      "epoch": 0.6875146807810959,
      "grad_norm": 0.8190531513263575,
      "learning_rate": 2.3490178478012363e-06,
      "loss": 0.4154,
      "step": 10976
    },
    {
      "epoch": 0.6875773187804382,
      "grad_norm": 0.8509588734177967,
      "learning_rate": 2.34815781926157e-06,
      "loss": 0.3836,
      "step": 10977
    },
    {
      "epoch": 0.6876399567797804,
      "grad_norm": 0.8476745445909953,
      "learning_rate": 2.34729789987222e-06,
      "loss": 0.4086,
      "step": 10978
    },
    {
      "epoch": 0.6877025947791228,
      "grad_norm": 0.8458482013757248,
      "learning_rate": 2.346438089668579e-06,
      "loss": 0.3837,
      "step": 10979
    },
    {
      "epoch": 0.687765232778465,
      "grad_norm": 0.8265940964054539,
      "learning_rate": 2.34557838868604e-06,
      "loss": 0.3597,
      "step": 10980
    },
    {
      "epoch": 0.6878278707778074,
      "grad_norm": 0.8816775204226751,
      "learning_rate": 2.3447187969599862e-06,
      "loss": 0.4082,
      "step": 10981
    },
    {
      "epoch": 0.6878905087771496,
      "grad_norm": 0.8223507024075196,
      "learning_rate": 2.3438593145258005e-06,
      "loss": 0.3631,
      "step": 10982
    },
    {
      "epoch": 0.687953146776492,
      "grad_norm": 0.8192186027573496,
      "learning_rate": 2.3429999414188594e-06,
      "loss": 0.3311,
      "step": 10983
    },
    {
      "epoch": 0.6880157847758342,
      "grad_norm": 0.8150717055075186,
      "learning_rate": 2.342140677674533e-06,
      "loss": 0.3803,
      "step": 10984
    },
    {
      "epoch": 0.6880784227751766,
      "grad_norm": 0.9725387241528289,
      "learning_rate": 2.34128152332819e-06,
      "loss": 0.4313,
      "step": 10985
    },
    {
      "epoch": 0.6881410607745189,
      "grad_norm": 0.8301772843999408,
      "learning_rate": 2.340422478415193e-06,
      "loss": 0.379,
      "step": 10986
    },
    {
      "epoch": 0.6882036987738611,
      "grad_norm": 0.6156358190841842,
      "learning_rate": 2.3395635429709e-06,
      "loss": 0.4595,
      "step": 10987
    },
    {
      "epoch": 0.6882663367732035,
      "grad_norm": 0.8383874526276811,
      "learning_rate": 2.338704717030664e-06,
      "loss": 0.4004,
      "step": 10988
    },
    {
      "epoch": 0.6883289747725457,
      "grad_norm": 0.9010148438619026,
      "learning_rate": 2.3378460006298375e-06,
      "loss": 0.4106,
      "step": 10989
    },
    {
      "epoch": 0.6883916127718881,
      "grad_norm": 0.6110054497416215,
      "learning_rate": 2.3369873938037624e-06,
      "loss": 0.4519,
      "step": 10990
    },
    {
      "epoch": 0.6884542507712303,
      "grad_norm": 0.8629039206918864,
      "learning_rate": 2.3361288965877813e-06,
      "loss": 0.3956,
      "step": 10991
    },
    {
      "epoch": 0.6885168887705727,
      "grad_norm": 0.8914890204061178,
      "learning_rate": 2.33527050901723e-06,
      "loss": 0.4118,
      "step": 10992
    },
    {
      "epoch": 0.688579526769915,
      "grad_norm": 0.800393275389066,
      "learning_rate": 2.334412231127439e-06,
      "loss": 0.356,
      "step": 10993
    },
    {
      "epoch": 0.6886421647692572,
      "grad_norm": 0.8729648500568838,
      "learning_rate": 2.333554062953736e-06,
      "loss": 0.3859,
      "step": 10994
    },
    {
      "epoch": 0.6887048027685996,
      "grad_norm": 0.8417406573407437,
      "learning_rate": 2.332696004531441e-06,
      "loss": 0.3683,
      "step": 10995
    },
    {
      "epoch": 0.6887674407679418,
      "grad_norm": 0.8726851083531093,
      "learning_rate": 2.331838055895875e-06,
      "loss": 0.4224,
      "step": 10996
    },
    {
      "epoch": 0.6888300787672842,
      "grad_norm": 0.8767331420843226,
      "learning_rate": 2.3309802170823476e-06,
      "loss": 0.3527,
      "step": 10997
    },
    {
      "epoch": 0.6888927167666264,
      "grad_norm": 0.8533154573157923,
      "learning_rate": 2.3301224881261713e-06,
      "loss": 0.3897,
      "step": 10998
    },
    {
      "epoch": 0.6889553547659688,
      "grad_norm": 0.8223270596963177,
      "learning_rate": 2.3292648690626472e-06,
      "loss": 0.3459,
      "step": 10999
    },
    {
      "epoch": 0.6890179927653111,
      "grad_norm": 0.8936607717957921,
      "learning_rate": 2.328407359927079e-06,
      "loss": 0.3892,
      "step": 11000
    },
    {
      "epoch": 0.6890806307646534,
      "grad_norm": 0.8429882327417167,
      "learning_rate": 2.32754996075476e-06,
      "loss": 0.4042,
      "step": 11001
    },
    {
      "epoch": 0.6891432687639957,
      "grad_norm": 0.8446514132020222,
      "learning_rate": 2.32669267158098e-06,
      "loss": 0.3594,
      "step": 11002
    },
    {
      "epoch": 0.6892059067633379,
      "grad_norm": 0.9426767383002366,
      "learning_rate": 2.3258354924410257e-06,
      "loss": 0.4199,
      "step": 11003
    },
    {
      "epoch": 0.6892685447626803,
      "grad_norm": 0.7735679629361464,
      "learning_rate": 2.3249784233701796e-06,
      "loss": 0.3738,
      "step": 11004
    },
    {
      "epoch": 0.6893311827620225,
      "grad_norm": 0.8735852107786174,
      "learning_rate": 2.3241214644037173e-06,
      "loss": 0.436,
      "step": 11005
    },
    {
      "epoch": 0.6893938207613649,
      "grad_norm": 0.88917471861967,
      "learning_rate": 2.323264615576911e-06,
      "loss": 0.435,
      "step": 11006
    },
    {
      "epoch": 0.6894564587607072,
      "grad_norm": 0.8517480995280866,
      "learning_rate": 2.3224078769250296e-06,
      "loss": 0.4401,
      "step": 11007
    },
    {
      "epoch": 0.6895190967600495,
      "grad_norm": 0.8381541068640533,
      "learning_rate": 2.321551248483339e-06,
      "loss": 0.3503,
      "step": 11008
    },
    {
      "epoch": 0.6895817347593918,
      "grad_norm": 0.925990884132742,
      "learning_rate": 2.3206947302870965e-06,
      "loss": 0.4003,
      "step": 11009
    },
    {
      "epoch": 0.6896443727587341,
      "grad_norm": 0.8530700804170771,
      "learning_rate": 2.3198383223715553e-06,
      "loss": 0.3943,
      "step": 11010
    },
    {
      "epoch": 0.6897070107580764,
      "grad_norm": 0.8814052144308594,
      "learning_rate": 2.318982024771967e-06,
      "loss": 0.3912,
      "step": 11011
    },
    {
      "epoch": 0.6897696487574186,
      "grad_norm": 0.8679435657897104,
      "learning_rate": 2.318125837523576e-06,
      "loss": 0.4015,
      "step": 11012
    },
    {
      "epoch": 0.689832286756761,
      "grad_norm": 0.8492903142160629,
      "learning_rate": 2.3172697606616228e-06,
      "loss": 0.3767,
      "step": 11013
    },
    {
      "epoch": 0.6898949247561033,
      "grad_norm": 0.7979644734470372,
      "learning_rate": 2.3164137942213427e-06,
      "loss": 0.3904,
      "step": 11014
    },
    {
      "epoch": 0.6899575627554456,
      "grad_norm": 0.8429615322809003,
      "learning_rate": 2.315557938237971e-06,
      "loss": 0.3938,
      "step": 11015
    },
    {
      "epoch": 0.6900202007547879,
      "grad_norm": 0.6060194548710566,
      "learning_rate": 2.314702192746731e-06,
      "loss": 0.4664,
      "step": 11016
    },
    {
      "epoch": 0.6900828387541302,
      "grad_norm": 0.851808386838099,
      "learning_rate": 2.3138465577828486e-06,
      "loss": 0.4007,
      "step": 11017
    },
    {
      "epoch": 0.6901454767534725,
      "grad_norm": 0.8477194898503372,
      "learning_rate": 2.312991033381541e-06,
      "loss": 0.389,
      "step": 11018
    },
    {
      "epoch": 0.6902081147528148,
      "grad_norm": 0.8912085941625645,
      "learning_rate": 2.3121356195780204e-06,
      "loss": 0.3646,
      "step": 11019
    },
    {
      "epoch": 0.6902707527521571,
      "grad_norm": 0.8142923816457505,
      "learning_rate": 2.311280316407497e-06,
      "loss": 0.3789,
      "step": 11020
    },
    {
      "epoch": 0.6903333907514994,
      "grad_norm": 0.8982098243136003,
      "learning_rate": 2.310425123905175e-06,
      "loss": 0.3745,
      "step": 11021
    },
    {
      "epoch": 0.6903960287508417,
      "grad_norm": 0.7774887884279847,
      "learning_rate": 2.3095700421062538e-06,
      "loss": 0.3711,
      "step": 11022
    },
    {
      "epoch": 0.690458666750184,
      "grad_norm": 0.8818645289955976,
      "learning_rate": 2.3087150710459276e-06,
      "loss": 0.4066,
      "step": 11023
    },
    {
      "epoch": 0.6905213047495263,
      "grad_norm": 0.8511022407719221,
      "learning_rate": 2.30786021075939e-06,
      "loss": 0.3955,
      "step": 11024
    },
    {
      "epoch": 0.6905839427488686,
      "grad_norm": 0.8130228348438207,
      "learning_rate": 2.3070054612818243e-06,
      "loss": 0.3677,
      "step": 11025
    },
    {
      "epoch": 0.690646580748211,
      "grad_norm": 0.9260498545587085,
      "learning_rate": 2.3061508226484153e-06,
      "loss": 0.4208,
      "step": 11026
    },
    {
      "epoch": 0.6907092187475532,
      "grad_norm": 0.812124126946776,
      "learning_rate": 2.3052962948943386e-06,
      "loss": 0.3836,
      "step": 11027
    },
    {
      "epoch": 0.6907718567468955,
      "grad_norm": 0.894822754396556,
      "learning_rate": 2.3044418780547666e-06,
      "loss": 0.4302,
      "step": 11028
    },
    {
      "epoch": 0.6908344947462378,
      "grad_norm": 0.8032161282032316,
      "learning_rate": 2.303587572164867e-06,
      "loss": 0.4267,
      "step": 11029
    },
    {
      "epoch": 0.6908971327455801,
      "grad_norm": 0.857876865746119,
      "learning_rate": 2.302733377259804e-06,
      "loss": 0.4003,
      "step": 11030
    },
    {
      "epoch": 0.6909597707449224,
      "grad_norm": 0.8543415645964715,
      "learning_rate": 2.3018792933747347e-06,
      "loss": 0.3858,
      "step": 11031
    },
    {
      "epoch": 0.6910224087442647,
      "grad_norm": 0.8101380989735327,
      "learning_rate": 2.3010253205448137e-06,
      "loss": 0.3628,
      "step": 11032
    },
    {
      "epoch": 0.691085046743607,
      "grad_norm": 0.8352317141866115,
      "learning_rate": 2.3001714588051925e-06,
      "loss": 0.3858,
      "step": 11033
    },
    {
      "epoch": 0.6911476847429493,
      "grad_norm": 0.8363833405746359,
      "learning_rate": 2.2993177081910157e-06,
      "loss": 0.4216,
      "step": 11034
    },
    {
      "epoch": 0.6912103227422917,
      "grad_norm": 0.819079981442048,
      "learning_rate": 2.2984640687374216e-06,
      "loss": 0.3467,
      "step": 11035
    },
    {
      "epoch": 0.6912729607416339,
      "grad_norm": 0.5998972794846542,
      "learning_rate": 2.297610540479549e-06,
      "loss": 0.4552,
      "step": 11036
    },
    {
      "epoch": 0.6913355987409762,
      "grad_norm": 0.7858368334826377,
      "learning_rate": 2.296757123452529e-06,
      "loss": 0.3809,
      "step": 11037
    },
    {
      "epoch": 0.6913982367403185,
      "grad_norm": 0.8604325766641525,
      "learning_rate": 2.2959038176914865e-06,
      "loss": 0.3898,
      "step": 11038
    },
    {
      "epoch": 0.6914608747396608,
      "grad_norm": 0.90145592691019,
      "learning_rate": 2.2950506232315458e-06,
      "loss": 0.4427,
      "step": 11039
    },
    {
      "epoch": 0.6915235127390031,
      "grad_norm": 0.859019164337857,
      "learning_rate": 2.294197540107821e-06,
      "loss": 0.3859,
      "step": 11040
    },
    {
      "epoch": 0.6915861507383454,
      "grad_norm": 0.8419277015705174,
      "learning_rate": 2.2933445683554294e-06,
      "loss": 0.3995,
      "step": 11041
    },
    {
      "epoch": 0.6916487887376878,
      "grad_norm": 0.804974293616934,
      "learning_rate": 2.2924917080094784e-06,
      "loss": 0.3897,
      "step": 11042
    },
    {
      "epoch": 0.69171142673703,
      "grad_norm": 0.8527175606308023,
      "learning_rate": 2.291638959105071e-06,
      "loss": 0.4153,
      "step": 11043
    },
    {
      "epoch": 0.6917740647363724,
      "grad_norm": 0.9061678499877946,
      "learning_rate": 2.2907863216773045e-06,
      "loss": 0.3926,
      "step": 11044
    },
    {
      "epoch": 0.6918367027357146,
      "grad_norm": 0.7830608952519899,
      "learning_rate": 2.2899337957612782e-06,
      "loss": 0.3556,
      "step": 11045
    },
    {
      "epoch": 0.6918993407350569,
      "grad_norm": 0.7679258470330422,
      "learning_rate": 2.2890813813920796e-06,
      "loss": 0.3812,
      "step": 11046
    },
    {
      "epoch": 0.6919619787343992,
      "grad_norm": 0.7925758199099068,
      "learning_rate": 2.288229078604795e-06,
      "loss": 0.3758,
      "step": 11047
    },
    {
      "epoch": 0.6920246167337415,
      "grad_norm": 0.8318171390008019,
      "learning_rate": 2.287376887434504e-06,
      "loss": 0.3881,
      "step": 11048
    },
    {
      "epoch": 0.6920872547330839,
      "grad_norm": 0.7540736660645286,
      "learning_rate": 2.286524807916283e-06,
      "loss": 0.3522,
      "step": 11049
    },
    {
      "epoch": 0.6921498927324261,
      "grad_norm": 0.6634093559587779,
      "learning_rate": 2.2856728400852053e-06,
      "loss": 0.437,
      "step": 11050
    },
    {
      "epoch": 0.6922125307317685,
      "grad_norm": 0.9125153322129462,
      "learning_rate": 2.284820983976338e-06,
      "loss": 0.4053,
      "step": 11051
    },
    {
      "epoch": 0.6922751687311107,
      "grad_norm": 0.7753164175174186,
      "learning_rate": 2.2839692396247427e-06,
      "loss": 0.3729,
      "step": 11052
    },
    {
      "epoch": 0.692337806730453,
      "grad_norm": 0.8611568591334047,
      "learning_rate": 2.283117607065476e-06,
      "loss": 0.3988,
      "step": 11053
    },
    {
      "epoch": 0.6924004447297953,
      "grad_norm": 0.8652800211463726,
      "learning_rate": 2.2822660863335943e-06,
      "loss": 0.4029,
      "step": 11054
    },
    {
      "epoch": 0.6924630827291376,
      "grad_norm": 0.8348509024824486,
      "learning_rate": 2.2814146774641453e-06,
      "loss": 0.3906,
      "step": 11055
    },
    {
      "epoch": 0.69252572072848,
      "grad_norm": 0.792262031423803,
      "learning_rate": 2.2805633804921727e-06,
      "loss": 0.3771,
      "step": 11056
    },
    {
      "epoch": 0.6925883587278222,
      "grad_norm": 0.8589512090713871,
      "learning_rate": 2.2797121954527156e-06,
      "loss": 0.3897,
      "step": 11057
    },
    {
      "epoch": 0.6926509967271646,
      "grad_norm": 0.8151181633707429,
      "learning_rate": 2.2788611223808083e-06,
      "loss": 0.4015,
      "step": 11058
    },
    {
      "epoch": 0.6927136347265068,
      "grad_norm": 0.8776101196256576,
      "learning_rate": 2.278010161311484e-06,
      "loss": 0.3539,
      "step": 11059
    },
    {
      "epoch": 0.6927762727258492,
      "grad_norm": 0.6065259850323337,
      "learning_rate": 2.2771593122797663e-06,
      "loss": 0.4625,
      "step": 11060
    },
    {
      "epoch": 0.6928389107251914,
      "grad_norm": 0.833600474815494,
      "learning_rate": 2.276308575320677e-06,
      "loss": 0.3763,
      "step": 11061
    },
    {
      "epoch": 0.6929015487245337,
      "grad_norm": 0.8160510825987164,
      "learning_rate": 2.27545795046923e-06,
      "loss": 0.4262,
      "step": 11062
    },
    {
      "epoch": 0.692964186723876,
      "grad_norm": 0.9002922757731833,
      "learning_rate": 2.274607437760442e-06,
      "loss": 0.4145,
      "step": 11063
    },
    {
      "epoch": 0.6930268247232183,
      "grad_norm": 0.7706870165440218,
      "learning_rate": 2.2737570372293177e-06,
      "loss": 0.3658,
      "step": 11064
    },
    {
      "epoch": 0.6930894627225607,
      "grad_norm": 0.9013489237331179,
      "learning_rate": 2.2729067489108596e-06,
      "loss": 0.3622,
      "step": 11065
    },
    {
      "epoch": 0.6931521007219029,
      "grad_norm": 0.8834019424976904,
      "learning_rate": 2.272056572840064e-06,
      "loss": 0.3786,
      "step": 11066
    },
    {
      "epoch": 0.6932147387212453,
      "grad_norm": 0.9186279934897519,
      "learning_rate": 2.2712065090519275e-06,
      "loss": 0.4289,
      "step": 11067
    },
    {
      "epoch": 0.6932773767205875,
      "grad_norm": 0.9644963221100933,
      "learning_rate": 2.2703565575814383e-06,
      "loss": 0.4121,
      "step": 11068
    },
    {
      "epoch": 0.6933400147199299,
      "grad_norm": 0.7588355034234243,
      "learning_rate": 2.2695067184635795e-06,
      "loss": 0.3575,
      "step": 11069
    },
    {
      "epoch": 0.6934026527192722,
      "grad_norm": 0.8361110526628049,
      "learning_rate": 2.2686569917333314e-06,
      "loss": 0.3663,
      "step": 11070
    },
    {
      "epoch": 0.6934652907186144,
      "grad_norm": 0.8320210462243596,
      "learning_rate": 2.267807377425666e-06,
      "loss": 0.417,
      "step": 11071
    },
    {
      "epoch": 0.6935279287179568,
      "grad_norm": 0.9148693281250789,
      "learning_rate": 2.266957875575558e-06,
      "loss": 0.4158,
      "step": 11072
    },
    {
      "epoch": 0.693590566717299,
      "grad_norm": 0.9174825537449839,
      "learning_rate": 2.266108486217971e-06,
      "loss": 0.4219,
      "step": 11073
    },
    {
      "epoch": 0.6936532047166414,
      "grad_norm": 0.7939171635957937,
      "learning_rate": 2.265259209387867e-06,
      "loss": 0.3907,
      "step": 11074
    },
    {
      "epoch": 0.6937158427159836,
      "grad_norm": 0.8213685582606708,
      "learning_rate": 2.264410045120199e-06,
      "loss": 0.3764,
      "step": 11075
    },
    {
      "epoch": 0.693778480715326,
      "grad_norm": 0.908822918817571,
      "learning_rate": 2.2635609934499232e-06,
      "loss": 0.3714,
      "step": 11076
    },
    {
      "epoch": 0.6938411187146682,
      "grad_norm": 0.92815322473849,
      "learning_rate": 2.262712054411985e-06,
      "loss": 0.4091,
      "step": 11077
    },
    {
      "epoch": 0.6939037567140105,
      "grad_norm": 0.8368122329735562,
      "learning_rate": 2.2618632280413266e-06,
      "loss": 0.3596,
      "step": 11078
    },
    {
      "epoch": 0.6939663947133529,
      "grad_norm": 0.9058652258141016,
      "learning_rate": 2.2610145143728855e-06,
      "loss": 0.4044,
      "step": 11079
    },
    {
      "epoch": 0.6940290327126951,
      "grad_norm": 0.9487699816600539,
      "learning_rate": 2.260165913441596e-06,
      "loss": 0.4182,
      "step": 11080
    },
    {
      "epoch": 0.6940916707120375,
      "grad_norm": 0.861721258363285,
      "learning_rate": 2.259317425282384e-06,
      "loss": 0.4088,
      "step": 11081
    },
    {
      "epoch": 0.6941543087113797,
      "grad_norm": 0.8836993220327286,
      "learning_rate": 2.2584690499301776e-06,
      "loss": 0.3882,
      "step": 11082
    },
    {
      "epoch": 0.6942169467107221,
      "grad_norm": 0.909573300983196,
      "learning_rate": 2.2576207874198936e-06,
      "loss": 0.4009,
      "step": 11083
    },
    {
      "epoch": 0.6942795847100643,
      "grad_norm": 0.8134664122188378,
      "learning_rate": 2.256772637786445e-06,
      "loss": 0.3791,
      "step": 11084
    },
    {
      "epoch": 0.6943422227094067,
      "grad_norm": 0.8440145318268542,
      "learning_rate": 2.2559246010647456e-06,
      "loss": 0.3817,
      "step": 11085
    },
    {
      "epoch": 0.694404860708749,
      "grad_norm": 0.6223246201308471,
      "learning_rate": 2.2550766772896995e-06,
      "loss": 0.4602,
      "step": 11086
    },
    {
      "epoch": 0.6944674987080912,
      "grad_norm": 0.9187502503094481,
      "learning_rate": 2.2542288664962066e-06,
      "loss": 0.3929,
      "step": 11087
    },
    {
      "epoch": 0.6945301367074336,
      "grad_norm": 0.8200537953447298,
      "learning_rate": 2.253381168719163e-06,
      "loss": 0.355,
      "step": 11088
    },
    {
      "epoch": 0.6945927747067758,
      "grad_norm": 0.8688936068976395,
      "learning_rate": 2.2525335839934604e-06,
      "loss": 0.3763,
      "step": 11089
    },
    {
      "epoch": 0.6946554127061182,
      "grad_norm": 0.883752136435789,
      "learning_rate": 2.2516861123539835e-06,
      "loss": 0.4357,
      "step": 11090
    },
    {
      "epoch": 0.6947180507054604,
      "grad_norm": 0.8867125185363357,
      "learning_rate": 2.2508387538356187e-06,
      "loss": 0.4251,
      "step": 11091
    },
    {
      "epoch": 0.6947806887048028,
      "grad_norm": 0.8928961519492388,
      "learning_rate": 2.2499915084732387e-06,
      "loss": 0.3751,
      "step": 11092
    },
    {
      "epoch": 0.6948433267041451,
      "grad_norm": 0.8144039005184792,
      "learning_rate": 2.2491443763017206e-06,
      "loss": 0.4015,
      "step": 11093
    },
    {
      "epoch": 0.6949059647034874,
      "grad_norm": 0.8305183618599087,
      "learning_rate": 2.2482973573559306e-06,
      "loss": 0.3728,
      "step": 11094
    },
    {
      "epoch": 0.6949686027028297,
      "grad_norm": 0.865530246221567,
      "learning_rate": 2.247450451670732e-06,
      "loss": 0.4042,
      "step": 11095
    },
    {
      "epoch": 0.6950312407021719,
      "grad_norm": 0.9035846887605126,
      "learning_rate": 2.2466036592809837e-06,
      "loss": 0.3577,
      "step": 11096
    },
    {
      "epoch": 0.6950938787015143,
      "grad_norm": 0.8464593323925345,
      "learning_rate": 2.24575698022154e-06,
      "loss": 0.3906,
      "step": 11097
    },
    {
      "epoch": 0.6951565167008565,
      "grad_norm": 0.5810091436138028,
      "learning_rate": 2.24491041452725e-06,
      "loss": 0.4217,
      "step": 11098
    },
    {
      "epoch": 0.6952191547001989,
      "grad_norm": 0.8449568975684872,
      "learning_rate": 2.2440639622329567e-06,
      "loss": 0.4211,
      "step": 11099
    },
    {
      "epoch": 0.6952817926995412,
      "grad_norm": 0.8600480645938265,
      "learning_rate": 2.2432176233735047e-06,
      "loss": 0.385,
      "step": 11100
    },
    {
      "epoch": 0.6953444306988835,
      "grad_norm": 0.9238241101846102,
      "learning_rate": 2.242371397983725e-06,
      "loss": 0.4347,
      "step": 11101
    },
    {
      "epoch": 0.6954070686982258,
      "grad_norm": 0.8251583520860104,
      "learning_rate": 2.241525286098452e-06,
      "loss": 0.3916,
      "step": 11102
    },
    {
      "epoch": 0.695469706697568,
      "grad_norm": 0.842975627129561,
      "learning_rate": 2.2406792877525103e-06,
      "loss": 0.3787,
      "step": 11103
    },
    {
      "epoch": 0.6955323446969104,
      "grad_norm": 0.9294759689711689,
      "learning_rate": 2.239833402980721e-06,
      "loss": 0.3757,
      "step": 11104
    },
    {
      "epoch": 0.6955949826962526,
      "grad_norm": 0.7792135803663062,
      "learning_rate": 2.2389876318179016e-06,
      "loss": 0.3882,
      "step": 11105
    },
    {
      "epoch": 0.695657620695595,
      "grad_norm": 0.8445777575739184,
      "learning_rate": 2.2381419742988635e-06,
      "loss": 0.384,
      "step": 11106
    },
    {
      "epoch": 0.6957202586949373,
      "grad_norm": 0.8254393257610524,
      "learning_rate": 2.2372964304584144e-06,
      "loss": 0.3743,
      "step": 11107
    },
    {
      "epoch": 0.6957828966942796,
      "grad_norm": 0.8893469262528247,
      "learning_rate": 2.236451000331355e-06,
      "loss": 0.4013,
      "step": 11108
    },
    {
      "epoch": 0.6958455346936219,
      "grad_norm": 0.8715675062306895,
      "learning_rate": 2.2356056839524875e-06,
      "loss": 0.3943,
      "step": 11109
    },
    {
      "epoch": 0.6959081726929642,
      "grad_norm": 0.8439063707850568,
      "learning_rate": 2.2347604813566016e-06,
      "loss": 0.353,
      "step": 11110
    },
    {
      "epoch": 0.6959708106923065,
      "grad_norm": 0.8687538393112991,
      "learning_rate": 2.2339153925784886e-06,
      "loss": 0.3516,
      "step": 11111
    },
    {
      "epoch": 0.6960334486916487,
      "grad_norm": 0.8801015427422296,
      "learning_rate": 2.2330704176529315e-06,
      "loss": 0.414,
      "step": 11112
    },
    {
      "epoch": 0.6960960866909911,
      "grad_norm": 0.8308464596105504,
      "learning_rate": 2.23222555661471e-06,
      "loss": 0.3978,
      "step": 11113
    },
    {
      "epoch": 0.6961587246903334,
      "grad_norm": 0.8034085056125703,
      "learning_rate": 2.2313808094985984e-06,
      "loss": 0.3727,
      "step": 11114
    },
    {
      "epoch": 0.6962213626896757,
      "grad_norm": 0.8153428435611542,
      "learning_rate": 2.2305361763393663e-06,
      "loss": 0.3669,
      "step": 11115
    },
    {
      "epoch": 0.696284000689018,
      "grad_norm": 0.8181322162575217,
      "learning_rate": 2.2296916571717793e-06,
      "loss": 0.3917,
      "step": 11116
    },
    {
      "epoch": 0.6963466386883603,
      "grad_norm": 0.858095404501356,
      "learning_rate": 2.228847252030596e-06,
      "loss": 0.3749,
      "step": 11117
    },
    {
      "epoch": 0.6964092766877026,
      "grad_norm": 0.8791929987590008,
      "learning_rate": 2.2280029609505745e-06,
      "loss": 0.4076,
      "step": 11118
    },
    {
      "epoch": 0.696471914687045,
      "grad_norm": 0.9075904257712961,
      "learning_rate": 2.2271587839664673e-06,
      "loss": 0.4036,
      "step": 11119
    },
    {
      "epoch": 0.6965345526863872,
      "grad_norm": 0.8187936460067378,
      "learning_rate": 2.226314721113019e-06,
      "loss": 0.3592,
      "step": 11120
    },
    {
      "epoch": 0.6965971906857295,
      "grad_norm": 0.8792383838391231,
      "learning_rate": 2.2254707724249712e-06,
      "loss": 0.4071,
      "step": 11121
    },
    {
      "epoch": 0.6966598286850718,
      "grad_norm": 0.8047387872623397,
      "learning_rate": 2.224626937937062e-06,
      "loss": 0.3689,
      "step": 11122
    },
    {
      "epoch": 0.6967224666844141,
      "grad_norm": 0.8256943903189772,
      "learning_rate": 2.223783217684023e-06,
      "loss": 0.3546,
      "step": 11123
    },
    {
      "epoch": 0.6967851046837564,
      "grad_norm": 0.901988200276119,
      "learning_rate": 2.222939611700582e-06,
      "loss": 0.3548,
      "step": 11124
    },
    {
      "epoch": 0.6968477426830987,
      "grad_norm": 0.8050447959780923,
      "learning_rate": 2.22209612002146e-06,
      "loss": 0.3329,
      "step": 11125
    },
    {
      "epoch": 0.696910380682441,
      "grad_norm": 0.7930698930359728,
      "learning_rate": 2.2212527426813797e-06,
      "loss": 0.3534,
      "step": 11126
    },
    {
      "epoch": 0.6969730186817833,
      "grad_norm": 0.8811597019844613,
      "learning_rate": 2.2204094797150505e-06,
      "loss": 0.3859,
      "step": 11127
    },
    {
      "epoch": 0.6970356566811257,
      "grad_norm": 0.8547682453597955,
      "learning_rate": 2.219566331157184e-06,
      "loss": 0.3883,
      "step": 11128
    },
    {
      "epoch": 0.6970982946804679,
      "grad_norm": 0.6725420534934455,
      "learning_rate": 2.2187232970424844e-06,
      "loss": 0.4402,
      "step": 11129
    },
    {
      "epoch": 0.6971609326798102,
      "grad_norm": 0.847417616601743,
      "learning_rate": 2.2178803774056494e-06,
      "loss": 0.3834,
      "step": 11130
    },
    {
      "epoch": 0.6972235706791525,
      "grad_norm": 0.8636948957654793,
      "learning_rate": 2.2170375722813746e-06,
      "loss": 0.3993,
      "step": 11131
    },
    {
      "epoch": 0.6972862086784948,
      "grad_norm": 0.9037856650557018,
      "learning_rate": 2.21619488170435e-06,
      "loss": 0.4242,
      "step": 11132
    },
    {
      "epoch": 0.6973488466778371,
      "grad_norm": 0.8520629643583332,
      "learning_rate": 2.2153523057092606e-06,
      "loss": 0.3988,
      "step": 11133
    },
    {
      "epoch": 0.6974114846771794,
      "grad_norm": 0.9114433672872067,
      "learning_rate": 2.2145098443307857e-06,
      "loss": 0.4634,
      "step": 11134
    },
    {
      "epoch": 0.6974741226765218,
      "grad_norm": 0.822998579049062,
      "learning_rate": 2.213667497603604e-06,
      "loss": 0.3823,
      "step": 11135
    },
    {
      "epoch": 0.697536760675864,
      "grad_norm": 0.9116973352588298,
      "learning_rate": 2.212825265562383e-06,
      "loss": 0.4035,
      "step": 11136
    },
    {
      "epoch": 0.6975993986752063,
      "grad_norm": 0.7771858065282452,
      "learning_rate": 2.211983148241794e-06,
      "loss": 0.3398,
      "step": 11137
    },
    {
      "epoch": 0.6976620366745486,
      "grad_norm": 0.9001861844547183,
      "learning_rate": 2.2111411456764964e-06,
      "loss": 0.383,
      "step": 11138
    },
    {
      "epoch": 0.6977246746738909,
      "grad_norm": 0.8480099451261408,
      "learning_rate": 2.210299257901146e-06,
      "loss": 0.3889,
      "step": 11139
    },
    {
      "epoch": 0.6977873126732332,
      "grad_norm": 0.956235380605185,
      "learning_rate": 2.209457484950397e-06,
      "loss": 0.4411,
      "step": 11140
    },
    {
      "epoch": 0.6978499506725755,
      "grad_norm": 0.8811955790220611,
      "learning_rate": 2.2086158268588947e-06,
      "loss": 0.4039,
      "step": 11141
    },
    {
      "epoch": 0.6979125886719179,
      "grad_norm": 0.5727675363632482,
      "learning_rate": 2.2077742836612835e-06,
      "loss": 0.4317,
      "step": 11142
    },
    {
      "epoch": 0.6979752266712601,
      "grad_norm": 0.8840698988263368,
      "learning_rate": 2.2069328553922e-06,
      "loss": 0.4045,
      "step": 11143
    },
    {
      "epoch": 0.6980378646706025,
      "grad_norm": 0.8088781891423894,
      "learning_rate": 2.2060915420862796e-06,
      "loss": 0.3835,
      "step": 11144
    },
    {
      "epoch": 0.6981005026699447,
      "grad_norm": 0.9217106903448247,
      "learning_rate": 2.2052503437781483e-06,
      "loss": 0.4321,
      "step": 11145
    },
    {
      "epoch": 0.698163140669287,
      "grad_norm": 0.8638596223400936,
      "learning_rate": 2.204409260502433e-06,
      "loss": 0.4495,
      "step": 11146
    },
    {
      "epoch": 0.6982257786686293,
      "grad_norm": 0.8093881776450633,
      "learning_rate": 2.203568292293752e-06,
      "loss": 0.3906,
      "step": 11147
    },
    {
      "epoch": 0.6982884166679716,
      "grad_norm": 0.833685381456447,
      "learning_rate": 2.202727439186718e-06,
      "loss": 0.3961,
      "step": 11148
    },
    {
      "epoch": 0.698351054667314,
      "grad_norm": 0.8250124321404212,
      "learning_rate": 2.201886701215943e-06,
      "loss": 0.3815,
      "step": 11149
    },
    {
      "epoch": 0.6984136926666562,
      "grad_norm": 0.859654153814677,
      "learning_rate": 2.2010460784160302e-06,
      "loss": 0.3461,
      "step": 11150
    },
    {
      "epoch": 0.6984763306659986,
      "grad_norm": 0.8083360462785651,
      "learning_rate": 2.200205570821579e-06,
      "loss": 0.335,
      "step": 11151
    },
    {
      "epoch": 0.6985389686653408,
      "grad_norm": 0.8725611586942609,
      "learning_rate": 2.1993651784671875e-06,
      "loss": 0.3608,
      "step": 11152
    },
    {
      "epoch": 0.6986016066646832,
      "grad_norm": 0.8343801538982706,
      "learning_rate": 2.1985249013874454e-06,
      "loss": 0.3894,
      "step": 11153
    },
    {
      "epoch": 0.6986642446640254,
      "grad_norm": 0.8225614342690245,
      "learning_rate": 2.1976847396169364e-06,
      "loss": 0.3749,
      "step": 11154
    },
    {
      "epoch": 0.6987268826633677,
      "grad_norm": 0.6187341911037807,
      "learning_rate": 2.1968446931902454e-06,
      "loss": 0.4697,
      "step": 11155
    },
    {
      "epoch": 0.69878952066271,
      "grad_norm": 0.8329521408015766,
      "learning_rate": 2.1960047621419474e-06,
      "loss": 0.3629,
      "step": 11156
    },
    {
      "epoch": 0.6988521586620523,
      "grad_norm": 0.939347847527218,
      "learning_rate": 2.1951649465066132e-06,
      "loss": 0.397,
      "step": 11157
    },
    {
      "epoch": 0.6989147966613947,
      "grad_norm": 0.80940969220344,
      "learning_rate": 2.1943252463188107e-06,
      "loss": 0.3579,
      "step": 11158
    },
    {
      "epoch": 0.6989774346607369,
      "grad_norm": 0.8311841972772418,
      "learning_rate": 2.193485661613102e-06,
      "loss": 0.402,
      "step": 11159
    },
    {
      "epoch": 0.6990400726600793,
      "grad_norm": 0.774714114757351,
      "learning_rate": 2.1926461924240427e-06,
      "loss": 0.3717,
      "step": 11160
    },
    {
      "epoch": 0.6991027106594215,
      "grad_norm": 0.8878541695685883,
      "learning_rate": 2.1918068387861892e-06,
      "loss": 0.399,
      "step": 11161
    },
    {
      "epoch": 0.6991653486587638,
      "grad_norm": 0.6410778620399384,
      "learning_rate": 2.1909676007340867e-06,
      "loss": 0.4738,
      "step": 11162
    },
    {
      "epoch": 0.6992279866581061,
      "grad_norm": 0.9049263490981012,
      "learning_rate": 2.1901284783022774e-06,
      "loss": 0.397,
      "step": 11163
    },
    {
      "epoch": 0.6992906246574484,
      "grad_norm": 0.8123411361748973,
      "learning_rate": 2.189289471525304e-06,
      "loss": 0.3901,
      "step": 11164
    },
    {
      "epoch": 0.6993532626567908,
      "grad_norm": 0.7838673322024821,
      "learning_rate": 2.1884505804376973e-06,
      "loss": 0.3519,
      "step": 11165
    },
    {
      "epoch": 0.699415900656133,
      "grad_norm": 0.8470410686658938,
      "learning_rate": 2.1876118050739858e-06,
      "loss": 0.3877,
      "step": 11166
    },
    {
      "epoch": 0.6994785386554754,
      "grad_norm": 0.8409569172273217,
      "learning_rate": 2.1867731454686954e-06,
      "loss": 0.3374,
      "step": 11167
    },
    {
      "epoch": 0.6995411766548176,
      "grad_norm": 0.803089655896206,
      "learning_rate": 2.1859346016563444e-06,
      "loss": 0.3836,
      "step": 11168
    },
    {
      "epoch": 0.69960381465416,
      "grad_norm": 0.8840835496775504,
      "learning_rate": 2.1850961736714448e-06,
      "loss": 0.3802,
      "step": 11169
    },
    {
      "epoch": 0.6996664526535022,
      "grad_norm": 0.8546493893130939,
      "learning_rate": 2.184257861548512e-06,
      "loss": 0.3817,
      "step": 11170
    },
    {
      "epoch": 0.6997290906528445,
      "grad_norm": 0.89599367753093,
      "learning_rate": 2.183419665322048e-06,
      "loss": 0.4064,
      "step": 11171
    },
    {
      "epoch": 0.6997917286521869,
      "grad_norm": 0.8701017313816476,
      "learning_rate": 2.182581585026553e-06,
      "loss": 0.3874,
      "step": 11172
    },
    {
      "epoch": 0.6998543666515291,
      "grad_norm": 0.8153342474025429,
      "learning_rate": 2.1817436206965216e-06,
      "loss": 0.3957,
      "step": 11173
    },
    {
      "epoch": 0.6999170046508715,
      "grad_norm": 0.8591981056374599,
      "learning_rate": 2.180905772366447e-06,
      "loss": 0.3554,
      "step": 11174
    },
    {
      "epoch": 0.6999796426502137,
      "grad_norm": 0.85063750619404,
      "learning_rate": 2.1800680400708147e-06,
      "loss": 0.3833,
      "step": 11175
    },
    {
      "epoch": 0.7000422806495561,
      "grad_norm": 0.9222948608478528,
      "learning_rate": 2.179230423844105e-06,
      "loss": 0.421,
      "step": 11176
    },
    {
      "epoch": 0.7001049186488983,
      "grad_norm": 0.8670301749539694,
      "learning_rate": 2.178392923720793e-06,
      "loss": 0.3785,
      "step": 11177
    },
    {
      "epoch": 0.7001675566482407,
      "grad_norm": 0.8044199887161716,
      "learning_rate": 2.1775555397353532e-06,
      "loss": 0.3824,
      "step": 11178
    },
    {
      "epoch": 0.700230194647583,
      "grad_norm": 0.8404863520643955,
      "learning_rate": 2.1767182719222517e-06,
      "loss": 0.3991,
      "step": 11179
    },
    {
      "epoch": 0.7002928326469252,
      "grad_norm": 0.8770356573643806,
      "learning_rate": 2.1758811203159504e-06,
      "loss": 0.4103,
      "step": 11180
    },
    {
      "epoch": 0.7003554706462676,
      "grad_norm": 0.8473830295732937,
      "learning_rate": 2.1750440849509056e-06,
      "loss": 0.3591,
      "step": 11181
    },
    {
      "epoch": 0.7004181086456098,
      "grad_norm": 0.8631144813484531,
      "learning_rate": 2.1742071658615692e-06,
      "loss": 0.3675,
      "step": 11182
    },
    {
      "epoch": 0.7004807466449522,
      "grad_norm": 0.8281868337373687,
      "learning_rate": 2.1733703630823926e-06,
      "loss": 0.3627,
      "step": 11183
    },
    {
      "epoch": 0.7005433846442944,
      "grad_norm": 0.9484344548532814,
      "learning_rate": 2.172533676647816e-06,
      "loss": 0.4361,
      "step": 11184
    },
    {
      "epoch": 0.7006060226436368,
      "grad_norm": 0.8496374505025228,
      "learning_rate": 2.1716971065922782e-06,
      "loss": 0.3921,
      "step": 11185
    },
    {
      "epoch": 0.7006686606429791,
      "grad_norm": 0.8921913451739357,
      "learning_rate": 2.1708606529502107e-06,
      "loss": 0.4093,
      "step": 11186
    },
    {
      "epoch": 0.7007312986423213,
      "grad_norm": 0.9233885952290137,
      "learning_rate": 2.1700243157560457e-06,
      "loss": 0.3939,
      "step": 11187
    },
    {
      "epoch": 0.7007939366416637,
      "grad_norm": 0.8609761144948589,
      "learning_rate": 2.169188095044205e-06,
      "loss": 0.3923,
      "step": 11188
    },
    {
      "epoch": 0.7008565746410059,
      "grad_norm": 0.818447972466462,
      "learning_rate": 2.1683519908491073e-06,
      "loss": 0.3757,
      "step": 11189
    },
    {
      "epoch": 0.7009192126403483,
      "grad_norm": 0.8822481617017308,
      "learning_rate": 2.167516003205168e-06,
      "loss": 0.4035,
      "step": 11190
    },
    {
      "epoch": 0.7009818506396905,
      "grad_norm": 0.8213100783877053,
      "learning_rate": 2.1666801321467932e-06,
      "loss": 0.387,
      "step": 11191
    },
    {
      "epoch": 0.7010444886390329,
      "grad_norm": 0.8184654509006938,
      "learning_rate": 2.1658443777083927e-06,
      "loss": 0.3718,
      "step": 11192
    },
    {
      "epoch": 0.7011071266383752,
      "grad_norm": 0.8930086363743978,
      "learning_rate": 2.165008739924363e-06,
      "loss": 0.3919,
      "step": 11193
    },
    {
      "epoch": 0.7011697646377175,
      "grad_norm": 0.8635783899207917,
      "learning_rate": 2.1641732188291004e-06,
      "loss": 0.3948,
      "step": 11194
    },
    {
      "epoch": 0.7012324026370598,
      "grad_norm": 0.8106898207138522,
      "learning_rate": 2.1633378144569923e-06,
      "loss": 0.4079,
      "step": 11195
    },
    {
      "epoch": 0.701295040636402,
      "grad_norm": 0.8793529848488367,
      "learning_rate": 2.1625025268424277e-06,
      "loss": 0.3836,
      "step": 11196
    },
    {
      "epoch": 0.7013576786357444,
      "grad_norm": 0.8696912067047893,
      "learning_rate": 2.1616673560197864e-06,
      "loss": 0.4127,
      "step": 11197
    },
    {
      "epoch": 0.7014203166350866,
      "grad_norm": 0.8949446513191467,
      "learning_rate": 2.1608323020234427e-06,
      "loss": 0.3952,
      "step": 11198
    },
    {
      "epoch": 0.701482954634429,
      "grad_norm": 0.831657416998631,
      "learning_rate": 2.159997364887769e-06,
      "loss": 0.4092,
      "step": 11199
    },
    {
      "epoch": 0.7015455926337713,
      "grad_norm": 0.8935253136991541,
      "learning_rate": 2.1591625446471294e-06,
      "loss": 0.4145,
      "step": 11200
    },
    {
      "epoch": 0.7016082306331136,
      "grad_norm": 0.8099616270182957,
      "learning_rate": 2.158327841335888e-06,
      "loss": 0.365,
      "step": 11201
    },
    {
      "epoch": 0.7016708686324559,
      "grad_norm": 0.9162688615657089,
      "learning_rate": 2.1574932549884e-06,
      "loss": 0.4085,
      "step": 11202
    },
    {
      "epoch": 0.7017335066317982,
      "grad_norm": 0.8608245281320759,
      "learning_rate": 2.156658785639016e-06,
      "loss": 0.373,
      "step": 11203
    },
    {
      "epoch": 0.7017961446311405,
      "grad_norm": 0.7394710799712974,
      "learning_rate": 2.1558244333220857e-06,
      "loss": 0.347,
      "step": 11204
    },
    {
      "epoch": 0.7018587826304827,
      "grad_norm": 0.8313269137566828,
      "learning_rate": 2.1549901980719494e-06,
      "loss": 0.3675,
      "step": 11205
    },
    {
      "epoch": 0.7019214206298251,
      "grad_norm": 0.8852399973488952,
      "learning_rate": 2.154156079922945e-06,
      "loss": 0.3725,
      "step": 11206
    },
    {
      "epoch": 0.7019840586291674,
      "grad_norm": 0.8312953980771192,
      "learning_rate": 2.1533220789094047e-06,
      "loss": 0.3986,
      "step": 11207
    },
    {
      "epoch": 0.7020466966285097,
      "grad_norm": 0.8896929888291809,
      "learning_rate": 2.152488195065655e-06,
      "loss": 0.3991,
      "step": 11208
    },
    {
      "epoch": 0.702109334627852,
      "grad_norm": 0.8810839807260443,
      "learning_rate": 2.151654428426019e-06,
      "loss": 0.4038,
      "step": 11209
    },
    {
      "epoch": 0.7021719726271943,
      "grad_norm": 0.8000464434997959,
      "learning_rate": 2.1508207790248163e-06,
      "loss": 0.3806,
      "step": 11210
    },
    {
      "epoch": 0.7022346106265366,
      "grad_norm": 0.9286961807453432,
      "learning_rate": 2.14998724689636e-06,
      "loss": 0.3765,
      "step": 11211
    },
    {
      "epoch": 0.702297248625879,
      "grad_norm": 0.8935934289366393,
      "learning_rate": 2.1491538320749556e-06,
      "loss": 0.4075,
      "step": 11212
    },
    {
      "epoch": 0.7023598866252212,
      "grad_norm": 0.8323863729076624,
      "learning_rate": 2.1483205345949098e-06,
      "loss": 0.3672,
      "step": 11213
    },
    {
      "epoch": 0.7024225246245634,
      "grad_norm": 0.8987267476027552,
      "learning_rate": 2.1474873544905204e-06,
      "loss": 0.4035,
      "step": 11214
    },
    {
      "epoch": 0.7024851626239058,
      "grad_norm": 0.8412790862419353,
      "learning_rate": 2.1466542917960814e-06,
      "loss": 0.4016,
      "step": 11215
    },
    {
      "epoch": 0.7025478006232481,
      "grad_norm": 0.7925281650928231,
      "learning_rate": 2.145821346545881e-06,
      "loss": 0.3728,
      "step": 11216
    },
    {
      "epoch": 0.7026104386225904,
      "grad_norm": 0.9412926753591689,
      "learning_rate": 2.1449885187742036e-06,
      "loss": 0.4062,
      "step": 11217
    },
    {
      "epoch": 0.7026730766219327,
      "grad_norm": 0.8473197783858984,
      "learning_rate": 2.1441558085153293e-06,
      "loss": 0.399,
      "step": 11218
    },
    {
      "epoch": 0.702735714621275,
      "grad_norm": 0.6222976167666503,
      "learning_rate": 2.1433232158035292e-06,
      "loss": 0.4445,
      "step": 11219
    },
    {
      "epoch": 0.7027983526206173,
      "grad_norm": 0.8428932608996511,
      "learning_rate": 2.1424907406730766e-06,
      "loss": 0.4204,
      "step": 11220
    },
    {
      "epoch": 0.7028609906199595,
      "grad_norm": 0.7776519587966538,
      "learning_rate": 2.1416583831582373e-06,
      "loss": 0.3556,
      "step": 11221
    },
    {
      "epoch": 0.7029236286193019,
      "grad_norm": 0.873568700436516,
      "learning_rate": 2.1408261432932686e-06,
      "loss": 0.4141,
      "step": 11222
    },
    {
      "epoch": 0.7029862666186442,
      "grad_norm": 0.8079589504651278,
      "learning_rate": 2.1399940211124276e-06,
      "loss": 0.3858,
      "step": 11223
    },
    {
      "epoch": 0.7030489046179865,
      "grad_norm": 0.8670379112671022,
      "learning_rate": 2.1391620166499627e-06,
      "loss": 0.4005,
      "step": 11224
    },
    {
      "epoch": 0.7031115426173288,
      "grad_norm": 0.8385485972032239,
      "learning_rate": 2.1383301299401204e-06,
      "loss": 0.4008,
      "step": 11225
    },
    {
      "epoch": 0.7031741806166711,
      "grad_norm": 0.8902450999216666,
      "learning_rate": 2.137498361017141e-06,
      "loss": 0.3999,
      "step": 11226
    },
    {
      "epoch": 0.7032368186160134,
      "grad_norm": 0.8539551405282296,
      "learning_rate": 2.136666709915261e-06,
      "loss": 0.3686,
      "step": 11227
    },
    {
      "epoch": 0.7032994566153558,
      "grad_norm": 0.7851100325044841,
      "learning_rate": 2.1358351766687084e-06,
      "loss": 0.402,
      "step": 11228
    },
    {
      "epoch": 0.703362094614698,
      "grad_norm": 0.8646289579108732,
      "learning_rate": 2.135003761311712e-06,
      "loss": 0.428,
      "step": 11229
    },
    {
      "epoch": 0.7034247326140403,
      "grad_norm": 0.8462134382596244,
      "learning_rate": 2.134172463878494e-06,
      "loss": 0.4021,
      "step": 11230
    },
    {
      "epoch": 0.7034873706133826,
      "grad_norm": 0.8952869651396211,
      "learning_rate": 2.1333412844032692e-06,
      "loss": 0.4218,
      "step": 11231
    },
    {
      "epoch": 0.7035500086127249,
      "grad_norm": 0.811593014489584,
      "learning_rate": 2.132510222920249e-06,
      "loss": 0.3943,
      "step": 11232
    },
    {
      "epoch": 0.7036126466120672,
      "grad_norm": 0.8287298186797898,
      "learning_rate": 2.1316792794636406e-06,
      "loss": 0.379,
      "step": 11233
    },
    {
      "epoch": 0.7036752846114095,
      "grad_norm": 0.8254141503379633,
      "learning_rate": 2.1308484540676455e-06,
      "loss": 0.4013,
      "step": 11234
    },
    {
      "epoch": 0.7037379226107519,
      "grad_norm": 0.7690112518963881,
      "learning_rate": 2.13001774676646e-06,
      "loss": 0.3907,
      "step": 11235
    },
    {
      "epoch": 0.7038005606100941,
      "grad_norm": 0.8794036180323845,
      "learning_rate": 2.1291871575942755e-06,
      "loss": 0.3946,
      "step": 11236
    },
    {
      "epoch": 0.7038631986094365,
      "grad_norm": 0.8393048207696658,
      "learning_rate": 2.1283566865852824e-06,
      "loss": 0.4069,
      "step": 11237
    },
    {
      "epoch": 0.7039258366087787,
      "grad_norm": 0.8169385849612377,
      "learning_rate": 2.1275263337736592e-06,
      "loss": 0.3927,
      "step": 11238
    },
    {
      "epoch": 0.703988474608121,
      "grad_norm": 0.8642617016825032,
      "learning_rate": 2.1266960991935862e-06,
      "loss": 0.4252,
      "step": 11239
    },
    {
      "epoch": 0.7040511126074633,
      "grad_norm": 0.838071293385041,
      "learning_rate": 2.125865982879236e-06,
      "loss": 0.4017,
      "step": 11240
    },
    {
      "epoch": 0.7041137506068056,
      "grad_norm": 0.8269968551898805,
      "learning_rate": 2.1250359848647754e-06,
      "loss": 0.3417,
      "step": 11241
    },
    {
      "epoch": 0.704176388606148,
      "grad_norm": 0.8317579101876242,
      "learning_rate": 2.1242061051843672e-06,
      "loss": 0.3704,
      "step": 11242
    },
    {
      "epoch": 0.7042390266054902,
      "grad_norm": 0.760305095795124,
      "learning_rate": 2.123376343872169e-06,
      "loss": 0.358,
      "step": 11243
    },
    {
      "epoch": 0.7043016646048326,
      "grad_norm": 0.8235965818330504,
      "learning_rate": 2.1225467009623353e-06,
      "loss": 0.3809,
      "step": 11244
    },
    {
      "epoch": 0.7043643026041748,
      "grad_norm": 0.8629297377911029,
      "learning_rate": 2.1217171764890103e-06,
      "loss": 0.4079,
      "step": 11245
    },
    {
      "epoch": 0.7044269406035171,
      "grad_norm": 0.8535950423220191,
      "learning_rate": 2.120887770486342e-06,
      "loss": 0.3943,
      "step": 11246
    },
    {
      "epoch": 0.7044895786028594,
      "grad_norm": 0.8761162598416384,
      "learning_rate": 2.1200584829884686e-06,
      "loss": 0.4018,
      "step": 11247
    },
    {
      "epoch": 0.7045522166022017,
      "grad_norm": 0.7635020225870693,
      "learning_rate": 2.1192293140295223e-06,
      "loss": 0.3558,
      "step": 11248
    },
    {
      "epoch": 0.704614854601544,
      "grad_norm": 0.7748434398786649,
      "learning_rate": 2.1184002636436325e-06,
      "loss": 0.361,
      "step": 11249
    },
    {
      "epoch": 0.7046774926008863,
      "grad_norm": 0.823498602955324,
      "learning_rate": 2.117571331864923e-06,
      "loss": 0.3498,
      "step": 11250
    },
    {
      "epoch": 0.7047401306002287,
      "grad_norm": 0.9157089307054879,
      "learning_rate": 2.1167425187275114e-06,
      "loss": 0.4195,
      "step": 11251
    },
    {
      "epoch": 0.7048027685995709,
      "grad_norm": 0.9015335273754236,
      "learning_rate": 2.115913824265514e-06,
      "loss": 0.4525,
      "step": 11252
    },
    {
      "epoch": 0.7048654065989133,
      "grad_norm": 0.810999382413001,
      "learning_rate": 2.1150852485130387e-06,
      "loss": 0.376,
      "step": 11253
    },
    {
      "epoch": 0.7049280445982555,
      "grad_norm": 0.8401319782246577,
      "learning_rate": 2.1142567915041878e-06,
      "loss": 0.3632,
      "step": 11254
    },
    {
      "epoch": 0.7049906825975978,
      "grad_norm": 0.917356807707226,
      "learning_rate": 2.1134284532730632e-06,
      "loss": 0.4002,
      "step": 11255
    },
    {
      "epoch": 0.7050533205969401,
      "grad_norm": 0.8226490302614652,
      "learning_rate": 2.112600233853761e-06,
      "loss": 0.3716,
      "step": 11256
    },
    {
      "epoch": 0.7051159585962824,
      "grad_norm": 0.8381860382213414,
      "learning_rate": 2.111772133280369e-06,
      "loss": 0.4121,
      "step": 11257
    },
    {
      "epoch": 0.7051785965956248,
      "grad_norm": 0.8375051062638361,
      "learning_rate": 2.1109441515869724e-06,
      "loss": 0.3768,
      "step": 11258
    },
    {
      "epoch": 0.705241234594967,
      "grad_norm": 0.8542009709310564,
      "learning_rate": 2.1101162888076503e-06,
      "loss": 0.3804,
      "step": 11259
    },
    {
      "epoch": 0.7053038725943094,
      "grad_norm": 0.8162771575260983,
      "learning_rate": 2.109288544976479e-06,
      "loss": 0.3665,
      "step": 11260
    },
    {
      "epoch": 0.7053665105936516,
      "grad_norm": 0.8735578153584213,
      "learning_rate": 2.108460920127527e-06,
      "loss": 0.3914,
      "step": 11261
    },
    {
      "epoch": 0.705429148592994,
      "grad_norm": 0.8477081492659556,
      "learning_rate": 2.1076334142948583e-06,
      "loss": 0.4099,
      "step": 11262
    },
    {
      "epoch": 0.7054917865923362,
      "grad_norm": 0.8834535446067193,
      "learning_rate": 2.1068060275125374e-06,
      "loss": 0.4032,
      "step": 11263
    },
    {
      "epoch": 0.7055544245916785,
      "grad_norm": 0.9277528069070972,
      "learning_rate": 2.105978759814617e-06,
      "loss": 0.3743,
      "step": 11264
    },
    {
      "epoch": 0.7056170625910209,
      "grad_norm": 0.8408884237594416,
      "learning_rate": 2.1051516112351466e-06,
      "loss": 0.3635,
      "step": 11265
    },
    {
      "epoch": 0.7056797005903631,
      "grad_norm": 0.832182659299251,
      "learning_rate": 2.104324581808175e-06,
      "loss": 0.3931,
      "step": 11266
    },
    {
      "epoch": 0.7057423385897055,
      "grad_norm": 0.6659339859365426,
      "learning_rate": 2.1034976715677408e-06,
      "loss": 0.4703,
      "step": 11267
    },
    {
      "epoch": 0.7058049765890477,
      "grad_norm": 0.9508029657170708,
      "learning_rate": 2.10267088054788e-06,
      "loss": 0.4154,
      "step": 11268
    },
    {
      "epoch": 0.7058676145883901,
      "grad_norm": 0.629202858008937,
      "learning_rate": 2.1018442087826235e-06,
      "loss": 0.4389,
      "step": 11269
    },
    {
      "epoch": 0.7059302525877323,
      "grad_norm": 0.8657515241915757,
      "learning_rate": 2.1010176563059974e-06,
      "loss": 0.3983,
      "step": 11270
    },
    {
      "epoch": 0.7059928905870746,
      "grad_norm": 0.8150318785237417,
      "learning_rate": 2.100191223152021e-06,
      "loss": 0.3801,
      "step": 11271
    },
    {
      "epoch": 0.706055528586417,
      "grad_norm": 0.927466784910936,
      "learning_rate": 2.0993649093547135e-06,
      "loss": 0.411,
      "step": 11272
    },
    {
      "epoch": 0.7061181665857592,
      "grad_norm": 0.826757752441679,
      "learning_rate": 2.098538714948085e-06,
      "loss": 0.385,
      "step": 11273
    },
    {
      "epoch": 0.7061808045851016,
      "grad_norm": 0.8984980652432659,
      "learning_rate": 2.0977126399661402e-06,
      "loss": 0.4227,
      "step": 11274
    },
    {
      "epoch": 0.7062434425844438,
      "grad_norm": 0.8753287546489347,
      "learning_rate": 2.0968866844428824e-06,
      "loss": 0.4168,
      "step": 11275
    },
    {
      "epoch": 0.7063060805837862,
      "grad_norm": 0.8564117152261849,
      "learning_rate": 2.0960608484123085e-06,
      "loss": 0.3818,
      "step": 11276
    },
    {
      "epoch": 0.7063687185831284,
      "grad_norm": 0.800491040270582,
      "learning_rate": 2.0952351319084093e-06,
      "loss": 0.3667,
      "step": 11277
    },
    {
      "epoch": 0.7064313565824708,
      "grad_norm": 0.8822058056649303,
      "learning_rate": 2.0944095349651706e-06,
      "loss": 0.3564,
      "step": 11278
    },
    {
      "epoch": 0.7064939945818131,
      "grad_norm": 0.8482883489583126,
      "learning_rate": 2.093584057616575e-06,
      "loss": 0.4427,
      "step": 11279
    },
    {
      "epoch": 0.7065566325811553,
      "grad_norm": 0.8287782160699172,
      "learning_rate": 2.092758699896597e-06,
      "loss": 0.3519,
      "step": 11280
    },
    {
      "epoch": 0.7066192705804977,
      "grad_norm": 0.9099358993584626,
      "learning_rate": 2.0919334618392123e-06,
      "loss": 0.3837,
      "step": 11281
    },
    {
      "epoch": 0.7066819085798399,
      "grad_norm": 0.8681566353188288,
      "learning_rate": 2.0911083434783865e-06,
      "loss": 0.4007,
      "step": 11282
    },
    {
      "epoch": 0.7067445465791823,
      "grad_norm": 0.8747676991661928,
      "learning_rate": 2.0902833448480793e-06,
      "loss": 0.3919,
      "step": 11283
    },
    {
      "epoch": 0.7068071845785245,
      "grad_norm": 0.8123615827535597,
      "learning_rate": 2.089458465982251e-06,
      "loss": 0.387,
      "step": 11284
    },
    {
      "epoch": 0.7068698225778669,
      "grad_norm": 0.7981125702843854,
      "learning_rate": 2.0886337069148526e-06,
      "loss": 0.3676,
      "step": 11285
    },
    {
      "epoch": 0.7069324605772092,
      "grad_norm": 0.5974331812997216,
      "learning_rate": 2.0878090676798313e-06,
      "loss": 0.4355,
      "step": 11286
    },
    {
      "epoch": 0.7069950985765515,
      "grad_norm": 0.8248930981138515,
      "learning_rate": 2.0869845483111295e-06,
      "loss": 0.3897,
      "step": 11287
    },
    {
      "epoch": 0.7070577365758938,
      "grad_norm": 0.8290001791127641,
      "learning_rate": 2.0861601488426825e-06,
      "loss": 0.3512,
      "step": 11288
    },
    {
      "epoch": 0.707120374575236,
      "grad_norm": 0.8317763731974456,
      "learning_rate": 2.0853358693084265e-06,
      "loss": 0.3703,
      "step": 11289
    },
    {
      "epoch": 0.7071830125745784,
      "grad_norm": 0.8587232976767644,
      "learning_rate": 2.084511709742287e-06,
      "loss": 0.4458,
      "step": 11290
    },
    {
      "epoch": 0.7072456505739206,
      "grad_norm": 0.906587670416025,
      "learning_rate": 2.0836876701781865e-06,
      "loss": 0.4131,
      "step": 11291
    },
    {
      "epoch": 0.707308288573263,
      "grad_norm": 0.9633039264171344,
      "learning_rate": 2.082863750650041e-06,
      "loss": 0.4425,
      "step": 11292
    },
    {
      "epoch": 0.7073709265726053,
      "grad_norm": 0.8608104947810815,
      "learning_rate": 2.0820399511917673e-06,
      "loss": 0.3917,
      "step": 11293
    },
    {
      "epoch": 0.7074335645719476,
      "grad_norm": 0.8251759297794218,
      "learning_rate": 2.0812162718372704e-06,
      "loss": 0.3675,
      "step": 11294
    },
    {
      "epoch": 0.7074962025712899,
      "grad_norm": 0.8792079644724808,
      "learning_rate": 2.080392712620453e-06,
      "loss": 0.3814,
      "step": 11295
    },
    {
      "epoch": 0.7075588405706321,
      "grad_norm": 0.8320557754339496,
      "learning_rate": 2.0795692735752143e-06,
      "loss": 0.3831,
      "step": 11296
    },
    {
      "epoch": 0.7076214785699745,
      "grad_norm": 0.7892240751802807,
      "learning_rate": 2.0787459547354446e-06,
      "loss": 0.3995,
      "step": 11297
    },
    {
      "epoch": 0.7076841165693167,
      "grad_norm": 0.9179898674890663,
      "learning_rate": 2.0779227561350344e-06,
      "loss": 0.3806,
      "step": 11298
    },
    {
      "epoch": 0.7077467545686591,
      "grad_norm": 0.8354564537322483,
      "learning_rate": 2.0770996778078665e-06,
      "loss": 0.3995,
      "step": 11299
    },
    {
      "epoch": 0.7078093925680013,
      "grad_norm": 0.9768568658244057,
      "learning_rate": 2.076276719787819e-06,
      "loss": 0.4233,
      "step": 11300
    },
    {
      "epoch": 0.7078720305673437,
      "grad_norm": 0.8091428292595455,
      "learning_rate": 2.0754538821087633e-06,
      "loss": 0.3513,
      "step": 11301
    },
    {
      "epoch": 0.707934668566686,
      "grad_norm": 0.8449412183105257,
      "learning_rate": 2.0746311648045674e-06,
      "loss": 0.3684,
      "step": 11302
    },
    {
      "epoch": 0.7079973065660283,
      "grad_norm": 0.8752520223710633,
      "learning_rate": 2.0738085679090974e-06,
      "loss": 0.4214,
      "step": 11303
    },
    {
      "epoch": 0.7080599445653706,
      "grad_norm": 0.8771268298872761,
      "learning_rate": 2.07298609145621e-06,
      "loss": 0.3996,
      "step": 11304
    },
    {
      "epoch": 0.7081225825647128,
      "grad_norm": 0.8592512577846504,
      "learning_rate": 2.072163735479757e-06,
      "loss": 0.4109,
      "step": 11305
    },
    {
      "epoch": 0.7081852205640552,
      "grad_norm": 0.9435681264652284,
      "learning_rate": 2.071341500013589e-06,
      "loss": 0.3995,
      "step": 11306
    },
    {
      "epoch": 0.7082478585633974,
      "grad_norm": 0.6034359619406897,
      "learning_rate": 2.070519385091549e-06,
      "loss": 0.4593,
      "step": 11307
    },
    {
      "epoch": 0.7083104965627398,
      "grad_norm": 0.8922939816694632,
      "learning_rate": 2.0696973907474743e-06,
      "loss": 0.3929,
      "step": 11308
    },
    {
      "epoch": 0.7083731345620821,
      "grad_norm": 0.9057636373010804,
      "learning_rate": 2.0688755170152e-06,
      "loss": 0.3751,
      "step": 11309
    },
    {
      "epoch": 0.7084357725614244,
      "grad_norm": 0.8923618797197569,
      "learning_rate": 2.068053763928552e-06,
      "loss": 0.441,
      "step": 11310
    },
    {
      "epoch": 0.7084984105607667,
      "grad_norm": 0.8394060352425476,
      "learning_rate": 2.0672321315213544e-06,
      "loss": 0.3909,
      "step": 11311
    },
    {
      "epoch": 0.708561048560109,
      "grad_norm": 0.8512927368063065,
      "learning_rate": 2.0664106198274283e-06,
      "loss": 0.3943,
      "step": 11312
    },
    {
      "epoch": 0.7086236865594513,
      "grad_norm": 0.7749541522426676,
      "learning_rate": 2.0655892288805857e-06,
      "loss": 0.4007,
      "step": 11313
    },
    {
      "epoch": 0.7086863245587935,
      "grad_norm": 0.8750049299237885,
      "learning_rate": 2.064767958714633e-06,
      "loss": 0.4168,
      "step": 11314
    },
    {
      "epoch": 0.7087489625581359,
      "grad_norm": 0.8400890860751221,
      "learning_rate": 2.0639468093633773e-06,
      "loss": 0.3651,
      "step": 11315
    },
    {
      "epoch": 0.7088116005574782,
      "grad_norm": 0.9512566844166928,
      "learning_rate": 2.063125780860617e-06,
      "loss": 0.3952,
      "step": 11316
    },
    {
      "epoch": 0.7088742385568205,
      "grad_norm": 0.6040070126476632,
      "learning_rate": 2.062304873240144e-06,
      "loss": 0.4397,
      "step": 11317
    },
    {
      "epoch": 0.7089368765561628,
      "grad_norm": 0.8342551302918562,
      "learning_rate": 2.0614840865357477e-06,
      "loss": 0.363,
      "step": 11318
    },
    {
      "epoch": 0.7089995145555051,
      "grad_norm": 0.8375650139294174,
      "learning_rate": 2.0606634207812116e-06,
      "loss": 0.3801,
      "step": 11319
    },
    {
      "epoch": 0.7090621525548474,
      "grad_norm": 0.8476474716080339,
      "learning_rate": 2.0598428760103133e-06,
      "loss": 0.4039,
      "step": 11320
    },
    {
      "epoch": 0.7091247905541898,
      "grad_norm": 0.7899398183168702,
      "learning_rate": 2.0590224522568295e-06,
      "loss": 0.3855,
      "step": 11321
    },
    {
      "epoch": 0.709187428553532,
      "grad_norm": 0.8360327664318987,
      "learning_rate": 2.058202149554527e-06,
      "loss": 0.3678,
      "step": 11322
    },
    {
      "epoch": 0.7092500665528743,
      "grad_norm": 0.787754803191229,
      "learning_rate": 2.0573819679371694e-06,
      "loss": 0.4266,
      "step": 11323
    },
    {
      "epoch": 0.7093127045522166,
      "grad_norm": 0.8728229197010352,
      "learning_rate": 2.0565619074385173e-06,
      "loss": 0.3813,
      "step": 11324
    },
    {
      "epoch": 0.7093753425515589,
      "grad_norm": 0.6287358556562918,
      "learning_rate": 2.0557419680923235e-06,
      "loss": 0.4412,
      "step": 11325
    },
    {
      "epoch": 0.7094379805509012,
      "grad_norm": 0.8350539565391961,
      "learning_rate": 2.054922149932336e-06,
      "loss": 0.3867,
      "step": 11326
    },
    {
      "epoch": 0.7095006185502435,
      "grad_norm": 0.6570261675446816,
      "learning_rate": 2.0541024529923007e-06,
      "loss": 0.463,
      "step": 11327
    },
    {
      "epoch": 0.7095632565495859,
      "grad_norm": 0.8484024575458463,
      "learning_rate": 2.053282877305955e-06,
      "loss": 0.4015,
      "step": 11328
    },
    {
      "epoch": 0.7096258945489281,
      "grad_norm": 0.8897962591783489,
      "learning_rate": 2.0524634229070307e-06,
      "loss": 0.3769,
      "step": 11329
    },
    {
      "epoch": 0.7096885325482704,
      "grad_norm": 0.8194808930631572,
      "learning_rate": 2.0516440898292606e-06,
      "loss": 0.3897,
      "step": 11330
    },
    {
      "epoch": 0.7097511705476127,
      "grad_norm": 0.7826308043930225,
      "learning_rate": 2.0508248781063655e-06,
      "loss": 0.3421,
      "step": 11331
    },
    {
      "epoch": 0.709813808546955,
      "grad_norm": 0.8916661398190401,
      "learning_rate": 2.050005787772068e-06,
      "loss": 0.3988,
      "step": 11332
    },
    {
      "epoch": 0.7098764465462973,
      "grad_norm": 0.8515957927041149,
      "learning_rate": 2.0491868188600794e-06,
      "loss": 0.367,
      "step": 11333
    },
    {
      "epoch": 0.7099390845456396,
      "grad_norm": 0.7851068540985393,
      "learning_rate": 2.0483679714041094e-06,
      "loss": 0.3484,
      "step": 11334
    },
    {
      "epoch": 0.710001722544982,
      "grad_norm": 0.7909834063779754,
      "learning_rate": 2.047549245437861e-06,
      "loss": 0.3612,
      "step": 11335
    },
    {
      "epoch": 0.7100643605443242,
      "grad_norm": 0.8789540841812137,
      "learning_rate": 2.046730640995034e-06,
      "loss": 0.3972,
      "step": 11336
    },
    {
      "epoch": 0.7101269985436666,
      "grad_norm": 0.6227259752655652,
      "learning_rate": 2.0459121581093223e-06,
      "loss": 0.4487,
      "step": 11337
    },
    {
      "epoch": 0.7101896365430088,
      "grad_norm": 0.8713899782621398,
      "learning_rate": 2.045093796814412e-06,
      "loss": 0.3908,
      "step": 11338
    },
    {
      "epoch": 0.7102522745423511,
      "grad_norm": 0.781541273947386,
      "learning_rate": 2.0442755571439916e-06,
      "loss": 0.3683,
      "step": 11339
    },
    {
      "epoch": 0.7103149125416934,
      "grad_norm": 0.8520209263787529,
      "learning_rate": 2.0434574391317364e-06,
      "loss": 0.4031,
      "step": 11340
    },
    {
      "epoch": 0.7103775505410357,
      "grad_norm": 0.8675088389058679,
      "learning_rate": 2.0426394428113227e-06,
      "loss": 0.4152,
      "step": 11341
    },
    {
      "epoch": 0.710440188540378,
      "grad_norm": 0.83877999559695,
      "learning_rate": 2.0418215682164194e-06,
      "loss": 0.3874,
      "step": 11342
    },
    {
      "epoch": 0.7105028265397203,
      "grad_norm": 0.8512010594708105,
      "learning_rate": 2.0410038153806888e-06,
      "loss": 0.3298,
      "step": 11343
    },
    {
      "epoch": 0.7105654645390627,
      "grad_norm": 0.827841088941387,
      "learning_rate": 2.0401861843377906e-06,
      "loss": 0.3797,
      "step": 11344
    },
    {
      "epoch": 0.7106281025384049,
      "grad_norm": 0.8887046351619825,
      "learning_rate": 2.039368675121378e-06,
      "loss": 0.4151,
      "step": 11345
    },
    {
      "epoch": 0.7106907405377473,
      "grad_norm": 0.9003479645316882,
      "learning_rate": 2.038551287765101e-06,
      "loss": 0.4065,
      "step": 11346
    },
    {
      "epoch": 0.7107533785370895,
      "grad_norm": 0.9008732284672085,
      "learning_rate": 2.0377340223026004e-06,
      "loss": 0.4062,
      "step": 11347
    },
    {
      "epoch": 0.7108160165364318,
      "grad_norm": 0.9045268248122715,
      "learning_rate": 2.0369168787675194e-06,
      "loss": 0.4122,
      "step": 11348
    },
    {
      "epoch": 0.7108786545357741,
      "grad_norm": 0.7772298118145087,
      "learning_rate": 2.0360998571934874e-06,
      "loss": 0.3766,
      "step": 11349
    },
    {
      "epoch": 0.7109412925351164,
      "grad_norm": 0.8383547994366461,
      "learning_rate": 2.0352829576141376e-06,
      "loss": 0.4197,
      "step": 11350
    },
    {
      "epoch": 0.7110039305344588,
      "grad_norm": 0.7770414809411255,
      "learning_rate": 2.034466180063091e-06,
      "loss": 0.3519,
      "step": 11351
    },
    {
      "epoch": 0.711066568533801,
      "grad_norm": 0.8140043784323261,
      "learning_rate": 2.0336495245739673e-06,
      "loss": 0.367,
      "step": 11352
    },
    {
      "epoch": 0.7111292065331434,
      "grad_norm": 0.7841087092913083,
      "learning_rate": 2.0328329911803797e-06,
      "loss": 0.4061,
      "step": 11353
    },
    {
      "epoch": 0.7111918445324856,
      "grad_norm": 0.8250293436352535,
      "learning_rate": 2.032016579915937e-06,
      "loss": 0.3673,
      "step": 11354
    },
    {
      "epoch": 0.7112544825318279,
      "grad_norm": 0.8036584764656387,
      "learning_rate": 2.0312002908142427e-06,
      "loss": 0.3386,
      "step": 11355
    },
    {
      "epoch": 0.7113171205311702,
      "grad_norm": 0.8456852955446321,
      "learning_rate": 2.0303841239088932e-06,
      "loss": 0.382,
      "step": 11356
    },
    {
      "epoch": 0.7113797585305125,
      "grad_norm": 0.6354467503602952,
      "learning_rate": 2.029568079233485e-06,
      "loss": 0.4599,
      "step": 11357
    },
    {
      "epoch": 0.7114423965298549,
      "grad_norm": 0.9101790154773329,
      "learning_rate": 2.028752156821607e-06,
      "loss": 0.4166,
      "step": 11358
    },
    {
      "epoch": 0.7115050345291971,
      "grad_norm": 0.9223044230165164,
      "learning_rate": 2.027936356706843e-06,
      "loss": 0.3724,
      "step": 11359
    },
    {
      "epoch": 0.7115676725285395,
      "grad_norm": 0.8721641952942556,
      "learning_rate": 2.0271206789227694e-06,
      "loss": 0.3624,
      "step": 11360
    },
    {
      "epoch": 0.7116303105278817,
      "grad_norm": 0.5965626401331241,
      "learning_rate": 2.02630512350296e-06,
      "loss": 0.4317,
      "step": 11361
    },
    {
      "epoch": 0.7116929485272241,
      "grad_norm": 0.7988319007120864,
      "learning_rate": 2.0254896904809847e-06,
      "loss": 0.3738,
      "step": 11362
    },
    {
      "epoch": 0.7117555865265663,
      "grad_norm": 0.8430856365890025,
      "learning_rate": 2.024674379890405e-06,
      "loss": 0.3673,
      "step": 11363
    },
    {
      "epoch": 0.7118182245259086,
      "grad_norm": 0.8436923341482927,
      "learning_rate": 2.023859191764779e-06,
      "loss": 0.3967,
      "step": 11364
    },
    {
      "epoch": 0.711880862525251,
      "grad_norm": 0.8234641259849307,
      "learning_rate": 2.023044126137662e-06,
      "loss": 0.3661,
      "step": 11365
    },
    {
      "epoch": 0.7119435005245932,
      "grad_norm": 0.8852014322107893,
      "learning_rate": 2.0222291830426e-06,
      "loss": 0.425,
      "step": 11366
    },
    {
      "epoch": 0.7120061385239356,
      "grad_norm": 0.8087581439919429,
      "learning_rate": 2.0214143625131394e-06,
      "loss": 0.4208,
      "step": 11367
    },
    {
      "epoch": 0.7120687765232778,
      "grad_norm": 0.8678508478563739,
      "learning_rate": 2.0205996645828165e-06,
      "loss": 0.3919,
      "step": 11368
    },
    {
      "epoch": 0.7121314145226202,
      "grad_norm": 0.8660205749195927,
      "learning_rate": 2.019785089285164e-06,
      "loss": 0.3756,
      "step": 11369
    },
    {
      "epoch": 0.7121940525219624,
      "grad_norm": 0.8442077528527961,
      "learning_rate": 2.0189706366537102e-06,
      "loss": 0.3964,
      "step": 11370
    },
    {
      "epoch": 0.7122566905213048,
      "grad_norm": 0.8317136206628734,
      "learning_rate": 2.018156306721979e-06,
      "loss": 0.4079,
      "step": 11371
    },
    {
      "epoch": 0.712319328520647,
      "grad_norm": 0.8995756035479973,
      "learning_rate": 2.0173420995234876e-06,
      "loss": 0.4067,
      "step": 11372
    },
    {
      "epoch": 0.7123819665199893,
      "grad_norm": 0.7335157945612467,
      "learning_rate": 2.016528015091747e-06,
      "loss": 0.3488,
      "step": 11373
    },
    {
      "epoch": 0.7124446045193317,
      "grad_norm": 0.8305618581127322,
      "learning_rate": 2.015714053460269e-06,
      "loss": 0.4028,
      "step": 11374
    },
    {
      "epoch": 0.7125072425186739,
      "grad_norm": 0.7800804426767388,
      "learning_rate": 2.0149002146625522e-06,
      "loss": 0.3635,
      "step": 11375
    },
    {
      "epoch": 0.7125698805180163,
      "grad_norm": 0.8309016373081658,
      "learning_rate": 2.014086498732099e-06,
      "loss": 0.3591,
      "step": 11376
    },
    {
      "epoch": 0.7126325185173585,
      "grad_norm": 0.8137263244591433,
      "learning_rate": 2.0132729057023997e-06,
      "loss": 0.3401,
      "step": 11377
    },
    {
      "epoch": 0.7126951565167009,
      "grad_norm": 0.8205254756842203,
      "learning_rate": 2.0124594356069416e-06,
      "loss": 0.3524,
      "step": 11378
    },
    {
      "epoch": 0.7127577945160432,
      "grad_norm": 0.7388824624703175,
      "learning_rate": 2.011646088479208e-06,
      "loss": 0.3729,
      "step": 11379
    },
    {
      "epoch": 0.7128204325153854,
      "grad_norm": 0.822748578540339,
      "learning_rate": 2.010832864352676e-06,
      "loss": 0.3921,
      "step": 11380
    },
    {
      "epoch": 0.7128830705147278,
      "grad_norm": 0.7847289048850306,
      "learning_rate": 2.0100197632608176e-06,
      "loss": 0.3761,
      "step": 11381
    },
    {
      "epoch": 0.71294570851407,
      "grad_norm": 0.8634929260395874,
      "learning_rate": 2.0092067852370995e-06,
      "loss": 0.3863,
      "step": 11382
    },
    {
      "epoch": 0.7130083465134124,
      "grad_norm": 0.812064566718495,
      "learning_rate": 2.008393930314987e-06,
      "loss": 0.3596,
      "step": 11383
    },
    {
      "epoch": 0.7130709845127546,
      "grad_norm": 0.9095251323220483,
      "learning_rate": 2.007581198527933e-06,
      "loss": 0.4052,
      "step": 11384
    },
    {
      "epoch": 0.713133622512097,
      "grad_norm": 0.823246514877354,
      "learning_rate": 2.0067685899093948e-06,
      "loss": 0.3938,
      "step": 11385
    },
    {
      "epoch": 0.7131962605114393,
      "grad_norm": 0.8867347672489106,
      "learning_rate": 2.0059561044928173e-06,
      "loss": 0.3798,
      "step": 11386
    },
    {
      "epoch": 0.7132588985107816,
      "grad_norm": 0.8724141528240843,
      "learning_rate": 2.0051437423116414e-06,
      "loss": 0.4056,
      "step": 11387
    },
    {
      "epoch": 0.7133215365101239,
      "grad_norm": 0.8425146138258806,
      "learning_rate": 2.0043315033993054e-06,
      "loss": 0.4113,
      "step": 11388
    },
    {
      "epoch": 0.7133841745094661,
      "grad_norm": 0.8691726137890807,
      "learning_rate": 2.0035193877892412e-06,
      "loss": 0.3561,
      "step": 11389
    },
    {
      "epoch": 0.7134468125088085,
      "grad_norm": 0.8632094493869509,
      "learning_rate": 2.002707395514873e-06,
      "loss": 0.374,
      "step": 11390
    },
    {
      "epoch": 0.7135094505081507,
      "grad_norm": 0.8376471053107977,
      "learning_rate": 2.0018955266096265e-06,
      "loss": 0.4039,
      "step": 11391
    },
    {
      "epoch": 0.7135720885074931,
      "grad_norm": 0.8229377533375254,
      "learning_rate": 2.001083781106916e-06,
      "loss": 0.3846,
      "step": 11392
    },
    {
      "epoch": 0.7136347265068353,
      "grad_norm": 0.8680614914972045,
      "learning_rate": 2.0002721590401545e-06,
      "loss": 0.3674,
      "step": 11393
    },
    {
      "epoch": 0.7136973645061777,
      "grad_norm": 0.9037313997391035,
      "learning_rate": 1.999460660442746e-06,
      "loss": 0.3719,
      "step": 11394
    },
    {
      "epoch": 0.71376000250552,
      "grad_norm": 0.8713324661661659,
      "learning_rate": 1.9986492853480947e-06,
      "loss": 0.4276,
      "step": 11395
    },
    {
      "epoch": 0.7138226405048623,
      "grad_norm": 0.8295871062037521,
      "learning_rate": 1.9978380337895965e-06,
      "loss": 0.3697,
      "step": 11396
    },
    {
      "epoch": 0.7138852785042046,
      "grad_norm": 0.8485417158905524,
      "learning_rate": 1.997026905800642e-06,
      "loss": 0.376,
      "step": 11397
    },
    {
      "epoch": 0.7139479165035468,
      "grad_norm": 0.8374409748955307,
      "learning_rate": 1.996215901414617e-06,
      "loss": 0.392,
      "step": 11398
    },
    {
      "epoch": 0.7140105545028892,
      "grad_norm": 0.7814842169424883,
      "learning_rate": 1.995405020664902e-06,
      "loss": 0.3762,
      "step": 11399
    },
    {
      "epoch": 0.7140731925022314,
      "grad_norm": 0.8209365291783581,
      "learning_rate": 1.9945942635848745e-06,
      "loss": 0.3959,
      "step": 11400
    },
    {
      "epoch": 0.7141358305015738,
      "grad_norm": 0.8368123582821918,
      "learning_rate": 1.993783630207906e-06,
      "loss": 0.3988,
      "step": 11401
    },
    {
      "epoch": 0.7141984685009161,
      "grad_norm": 0.7889326206644715,
      "learning_rate": 1.9929731205673614e-06,
      "loss": 0.3625,
      "step": 11402
    },
    {
      "epoch": 0.7142611065002584,
      "grad_norm": 0.9299807194711881,
      "learning_rate": 1.992162734696599e-06,
      "loss": 0.3822,
      "step": 11403
    },
    {
      "epoch": 0.7143237444996007,
      "grad_norm": 0.8662717575470088,
      "learning_rate": 1.9913524726289784e-06,
      "loss": 0.418,
      "step": 11404
    },
    {
      "epoch": 0.7143863824989429,
      "grad_norm": 0.8750184038211837,
      "learning_rate": 1.9905423343978487e-06,
      "loss": 0.4125,
      "step": 11405
    },
    {
      "epoch": 0.7144490204982853,
      "grad_norm": 0.9206728017322948,
      "learning_rate": 1.9897323200365555e-06,
      "loss": 0.3663,
      "step": 11406
    },
    {
      "epoch": 0.7145116584976275,
      "grad_norm": 0.8196138491961363,
      "learning_rate": 1.988922429578438e-06,
      "loss": 0.3909,
      "step": 11407
    },
    {
      "epoch": 0.7145742964969699,
      "grad_norm": 0.8564507902538055,
      "learning_rate": 1.9881126630568308e-06,
      "loss": 0.4106,
      "step": 11408
    },
    {
      "epoch": 0.7146369344963122,
      "grad_norm": 0.8213200429538576,
      "learning_rate": 1.987303020505067e-06,
      "loss": 0.3552,
      "step": 11409
    },
    {
      "epoch": 0.7146995724956545,
      "grad_norm": 0.8367285331229816,
      "learning_rate": 1.98649350195647e-06,
      "loss": 0.3488,
      "step": 11410
    },
    {
      "epoch": 0.7147622104949968,
      "grad_norm": 0.8200130256714342,
      "learning_rate": 1.9856841074443606e-06,
      "loss": 0.389,
      "step": 11411
    },
    {
      "epoch": 0.7148248484943391,
      "grad_norm": 0.8755859507822509,
      "learning_rate": 1.9848748370020506e-06,
      "loss": 0.3828,
      "step": 11412
    },
    {
      "epoch": 0.7148874864936814,
      "grad_norm": 0.7962207735404855,
      "learning_rate": 1.984065690662853e-06,
      "loss": 0.3644,
      "step": 11413
    },
    {
      "epoch": 0.7149501244930236,
      "grad_norm": 0.7629187156271467,
      "learning_rate": 1.9832566684600724e-06,
      "loss": 0.3745,
      "step": 11414
    },
    {
      "epoch": 0.715012762492366,
      "grad_norm": 0.8852944599910698,
      "learning_rate": 1.982447770427007e-06,
      "loss": 0.3774,
      "step": 11415
    },
    {
      "epoch": 0.7150754004917083,
      "grad_norm": 0.8248279312018882,
      "learning_rate": 1.9816389965969492e-06,
      "loss": 0.3802,
      "step": 11416
    },
    {
      "epoch": 0.7151380384910506,
      "grad_norm": 0.8246223368708766,
      "learning_rate": 1.9808303470031927e-06,
      "loss": 0.3413,
      "step": 11417
    },
    {
      "epoch": 0.7152006764903929,
      "grad_norm": 0.9145686447287099,
      "learning_rate": 1.980021821679019e-06,
      "loss": 0.3986,
      "step": 11418
    },
    {
      "epoch": 0.7152633144897352,
      "grad_norm": 0.8394278168814452,
      "learning_rate": 1.979213420657708e-06,
      "loss": 0.4014,
      "step": 11419
    },
    {
      "epoch": 0.7153259524890775,
      "grad_norm": 0.8449433278255679,
      "learning_rate": 1.9784051439725333e-06,
      "loss": 0.3813,
      "step": 11420
    },
    {
      "epoch": 0.7153885904884199,
      "grad_norm": 0.8236953899148257,
      "learning_rate": 1.9775969916567615e-06,
      "loss": 0.3596,
      "step": 11421
    },
    {
      "epoch": 0.7154512284877621,
      "grad_norm": 0.6523469352730332,
      "learning_rate": 1.976788963743661e-06,
      "loss": 0.4379,
      "step": 11422
    },
    {
      "epoch": 0.7155138664871044,
      "grad_norm": 0.8159985820017046,
      "learning_rate": 1.9759810602664877e-06,
      "loss": 0.3979,
      "step": 11423
    },
    {
      "epoch": 0.7155765044864467,
      "grad_norm": 0.8560013183099432,
      "learning_rate": 1.975173281258495e-06,
      "loss": 0.3773,
      "step": 11424
    },
    {
      "epoch": 0.715639142485789,
      "grad_norm": 0.900587848603846,
      "learning_rate": 1.97436562675293e-06,
      "loss": 0.4281,
      "step": 11425
    },
    {
      "epoch": 0.7157017804851313,
      "grad_norm": 0.8095659436209743,
      "learning_rate": 1.9735580967830393e-06,
      "loss": 0.3962,
      "step": 11426
    },
    {
      "epoch": 0.7157644184844736,
      "grad_norm": 0.8537050637648194,
      "learning_rate": 1.972750691382059e-06,
      "loss": 0.3555,
      "step": 11427
    },
    {
      "epoch": 0.715827056483816,
      "grad_norm": 0.7923542491763237,
      "learning_rate": 1.9719434105832214e-06,
      "loss": 0.3616,
      "step": 11428
    },
    {
      "epoch": 0.7158896944831582,
      "grad_norm": 1.0147843767205131,
      "learning_rate": 1.9711362544197553e-06,
      "loss": 0.3745,
      "step": 11429
    },
    {
      "epoch": 0.7159523324825006,
      "grad_norm": 0.9077943217600472,
      "learning_rate": 1.970329222924882e-06,
      "loss": 0.3937,
      "step": 11430
    },
    {
      "epoch": 0.7160149704818428,
      "grad_norm": 0.5763982071061108,
      "learning_rate": 1.9695223161318216e-06,
      "loss": 0.4339,
      "step": 11431
    },
    {
      "epoch": 0.7160776084811851,
      "grad_norm": 0.8852242307557473,
      "learning_rate": 1.968715534073785e-06,
      "loss": 0.3789,
      "step": 11432
    },
    {
      "epoch": 0.7161402464805274,
      "grad_norm": 0.8868032595008274,
      "learning_rate": 1.96790887678398e-06,
      "loss": 0.442,
      "step": 11433
    },
    {
      "epoch": 0.7162028844798697,
      "grad_norm": 0.8124366714623941,
      "learning_rate": 1.9671023442956067e-06,
      "loss": 0.3826,
      "step": 11434
    },
    {
      "epoch": 0.716265522479212,
      "grad_norm": 0.8534493452699528,
      "learning_rate": 1.9662959366418653e-06,
      "loss": 0.3972,
      "step": 11435
    },
    {
      "epoch": 0.7163281604785543,
      "grad_norm": 0.9146893109046131,
      "learning_rate": 1.9654896538559464e-06,
      "loss": 0.3881,
      "step": 11436
    },
    {
      "epoch": 0.7163907984778967,
      "grad_norm": 0.84844505246249,
      "learning_rate": 1.9646834959710364e-06,
      "loss": 0.3905,
      "step": 11437
    },
    {
      "epoch": 0.7164534364772389,
      "grad_norm": 0.8222295922995758,
      "learning_rate": 1.9638774630203174e-06,
      "loss": 0.337,
      "step": 11438
    },
    {
      "epoch": 0.7165160744765812,
      "grad_norm": 0.8170327419989263,
      "learning_rate": 1.963071555036965e-06,
      "loss": 0.3874,
      "step": 11439
    },
    {
      "epoch": 0.7165787124759235,
      "grad_norm": 0.861401181641719,
      "learning_rate": 1.9622657720541497e-06,
      "loss": 0.3845,
      "step": 11440
    },
    {
      "epoch": 0.7166413504752658,
      "grad_norm": 0.8605790329785992,
      "learning_rate": 1.961460114105041e-06,
      "loss": 0.4122,
      "step": 11441
    },
    {
      "epoch": 0.7167039884746081,
      "grad_norm": 0.9170504895538198,
      "learning_rate": 1.9606545812227955e-06,
      "loss": 0.3703,
      "step": 11442
    },
    {
      "epoch": 0.7167666264739504,
      "grad_norm": 0.8519931871582739,
      "learning_rate": 1.959849173440574e-06,
      "loss": 0.4035,
      "step": 11443
    },
    {
      "epoch": 0.7168292644732928,
      "grad_norm": 0.6076619340938435,
      "learning_rate": 1.9590438907915247e-06,
      "loss": 0.4395,
      "step": 11444
    },
    {
      "epoch": 0.716891902472635,
      "grad_norm": 0.8197019251777017,
      "learning_rate": 1.958238733308793e-06,
      "loss": 0.4225,
      "step": 11445
    },
    {
      "epoch": 0.7169545404719774,
      "grad_norm": 0.908701289584711,
      "learning_rate": 1.95743370102552e-06,
      "loss": 0.4163,
      "step": 11446
    },
    {
      "epoch": 0.7170171784713196,
      "grad_norm": 0.8431378748069619,
      "learning_rate": 1.9566287939748407e-06,
      "loss": 0.4083,
      "step": 11447
    },
    {
      "epoch": 0.7170798164706619,
      "grad_norm": 0.8431573361470804,
      "learning_rate": 1.9558240121898847e-06,
      "loss": 0.3955,
      "step": 11448
    },
    {
      "epoch": 0.7171424544700042,
      "grad_norm": 0.8383731369686466,
      "learning_rate": 1.9550193557037755e-06,
      "loss": 0.3577,
      "step": 11449
    },
    {
      "epoch": 0.7172050924693465,
      "grad_norm": 0.8046568060336687,
      "learning_rate": 1.9542148245496366e-06,
      "loss": 0.3868,
      "step": 11450
    },
    {
      "epoch": 0.7172677304686889,
      "grad_norm": 0.8404686434204288,
      "learning_rate": 1.9534104187605794e-06,
      "loss": 0.37,
      "step": 11451
    },
    {
      "epoch": 0.7173303684680311,
      "grad_norm": 0.596772441126336,
      "learning_rate": 1.952606138369717e-06,
      "loss": 0.4513,
      "step": 11452
    },
    {
      "epoch": 0.7173930064673735,
      "grad_norm": 1.0116702280408443,
      "learning_rate": 1.951801983410151e-06,
      "loss": 0.4137,
      "step": 11453
    },
    {
      "epoch": 0.7174556444667157,
      "grad_norm": 0.8973462809331181,
      "learning_rate": 1.950997953914981e-06,
      "loss": 0.4213,
      "step": 11454
    },
    {
      "epoch": 0.7175182824660581,
      "grad_norm": 0.8021907077943159,
      "learning_rate": 1.950194049917301e-06,
      "loss": 0.3615,
      "step": 11455
    },
    {
      "epoch": 0.7175809204654003,
      "grad_norm": 0.8391018606124357,
      "learning_rate": 1.9493902714502004e-06,
      "loss": 0.3767,
      "step": 11456
    },
    {
      "epoch": 0.7176435584647426,
      "grad_norm": 0.8146112610516942,
      "learning_rate": 1.948586618546763e-06,
      "loss": 0.4143,
      "step": 11457
    },
    {
      "epoch": 0.717706196464085,
      "grad_norm": 0.8784896385290132,
      "learning_rate": 1.9477830912400644e-06,
      "loss": 0.3983,
      "step": 11458
    },
    {
      "epoch": 0.7177688344634272,
      "grad_norm": 0.8102700709944469,
      "learning_rate": 1.9469796895631817e-06,
      "loss": 0.3194,
      "step": 11459
    },
    {
      "epoch": 0.7178314724627696,
      "grad_norm": 0.8835337676021063,
      "learning_rate": 1.94617641354918e-06,
      "loss": 0.3678,
      "step": 11460
    },
    {
      "epoch": 0.7178941104621118,
      "grad_norm": 0.8090161346502635,
      "learning_rate": 1.945373263231126e-06,
      "loss": 0.4185,
      "step": 11461
    },
    {
      "epoch": 0.7179567484614542,
      "grad_norm": 0.8634577151328899,
      "learning_rate": 1.944570238642075e-06,
      "loss": 0.4264,
      "step": 11462
    },
    {
      "epoch": 0.7180193864607964,
      "grad_norm": 0.8577963480573995,
      "learning_rate": 1.943767339815081e-06,
      "loss": 0.3773,
      "step": 11463
    },
    {
      "epoch": 0.7180820244601387,
      "grad_norm": 0.8436878411799618,
      "learning_rate": 1.94296456678319e-06,
      "loss": 0.3603,
      "step": 11464
    },
    {
      "epoch": 0.718144662459481,
      "grad_norm": 0.8088335814503315,
      "learning_rate": 1.9421619195794445e-06,
      "loss": 0.3896,
      "step": 11465
    },
    {
      "epoch": 0.7182073004588233,
      "grad_norm": 0.8095388734171065,
      "learning_rate": 1.941359398236882e-06,
      "loss": 0.386,
      "step": 11466
    },
    {
      "epoch": 0.7182699384581657,
      "grad_norm": 0.8592725972993882,
      "learning_rate": 1.940557002788533e-06,
      "loss": 0.3905,
      "step": 11467
    },
    {
      "epoch": 0.7183325764575079,
      "grad_norm": 0.8590766273484365,
      "learning_rate": 1.9397547332674253e-06,
      "loss": 0.391,
      "step": 11468
    },
    {
      "epoch": 0.7183952144568503,
      "grad_norm": 0.8832752572608855,
      "learning_rate": 1.938952589706583e-06,
      "loss": 0.3942,
      "step": 11469
    },
    {
      "epoch": 0.7184578524561925,
      "grad_norm": 0.8504510646667558,
      "learning_rate": 1.93815057213902e-06,
      "loss": 0.4085,
      "step": 11470
    },
    {
      "epoch": 0.7185204904555349,
      "grad_norm": 0.9308144827937689,
      "learning_rate": 1.937348680597748e-06,
      "loss": 0.4083,
      "step": 11471
    },
    {
      "epoch": 0.7185831284548772,
      "grad_norm": 0.589372422936036,
      "learning_rate": 1.9365469151157725e-06,
      "loss": 0.4437,
      "step": 11472
    },
    {
      "epoch": 0.7186457664542194,
      "grad_norm": 0.9424502972401706,
      "learning_rate": 1.935745275726094e-06,
      "loss": 0.408,
      "step": 11473
    },
    {
      "epoch": 0.7187084044535618,
      "grad_norm": 0.8502213609161442,
      "learning_rate": 1.9349437624617096e-06,
      "loss": 0.3946,
      "step": 11474
    },
    {
      "epoch": 0.718771042452904,
      "grad_norm": 0.9062266627931345,
      "learning_rate": 1.9341423753556072e-06,
      "loss": 0.3879,
      "step": 11475
    },
    {
      "epoch": 0.7188336804522464,
      "grad_norm": 0.9530254882306926,
      "learning_rate": 1.933341114440775e-06,
      "loss": 0.4006,
      "step": 11476
    },
    {
      "epoch": 0.7188963184515886,
      "grad_norm": 0.8448833072250396,
      "learning_rate": 1.93253997975019e-06,
      "loss": 0.424,
      "step": 11477
    },
    {
      "epoch": 0.718958956450931,
      "grad_norm": 0.8429112494751154,
      "learning_rate": 1.9317389713168306e-06,
      "loss": 0.3696,
      "step": 11478
    },
    {
      "epoch": 0.7190215944502732,
      "grad_norm": 0.7979377655864324,
      "learning_rate": 1.9309380891736638e-06,
      "loss": 0.3678,
      "step": 11479
    },
    {
      "epoch": 0.7190842324496156,
      "grad_norm": 0.8162333755302649,
      "learning_rate": 1.930137333353656e-06,
      "loss": 0.3422,
      "step": 11480
    },
    {
      "epoch": 0.7191468704489579,
      "grad_norm": 0.9425178070384164,
      "learning_rate": 1.9293367038897655e-06,
      "loss": 0.3782,
      "step": 11481
    },
    {
      "epoch": 0.7192095084483001,
      "grad_norm": 0.863308250257877,
      "learning_rate": 1.9285362008149455e-06,
      "loss": 0.4245,
      "step": 11482
    },
    {
      "epoch": 0.7192721464476425,
      "grad_norm": 0.8205419196714939,
      "learning_rate": 1.927735824162146e-06,
      "loss": 0.3574,
      "step": 11483
    },
    {
      "epoch": 0.7193347844469847,
      "grad_norm": 0.8853414769712296,
      "learning_rate": 1.926935573964308e-06,
      "loss": 0.4252,
      "step": 11484
    },
    {
      "epoch": 0.7193974224463271,
      "grad_norm": 0.6490218833634469,
      "learning_rate": 1.926135450254374e-06,
      "loss": 0.4325,
      "step": 11485
    },
    {
      "epoch": 0.7194600604456693,
      "grad_norm": 0.6593078903709182,
      "learning_rate": 1.925335453065274e-06,
      "loss": 0.43,
      "step": 11486
    },
    {
      "epoch": 0.7195226984450117,
      "grad_norm": 0.8160351024335553,
      "learning_rate": 1.9245355824299387e-06,
      "loss": 0.3645,
      "step": 11487
    },
    {
      "epoch": 0.719585336444354,
      "grad_norm": 0.9285099867632124,
      "learning_rate": 1.9237358383812903e-06,
      "loss": 0.3808,
      "step": 11488
    },
    {
      "epoch": 0.7196479744436962,
      "grad_norm": 0.8724501587133326,
      "learning_rate": 1.922936220952246e-06,
      "loss": 0.3628,
      "step": 11489
    },
    {
      "epoch": 0.7197106124430386,
      "grad_norm": 0.8724750128592061,
      "learning_rate": 1.9221367301757164e-06,
      "loss": 0.3826,
      "step": 11490
    },
    {
      "epoch": 0.7197732504423808,
      "grad_norm": 0.7917550571191599,
      "learning_rate": 1.9213373660846118e-06,
      "loss": 0.3899,
      "step": 11491
    },
    {
      "epoch": 0.7198358884417232,
      "grad_norm": 0.806249387084973,
      "learning_rate": 1.920538128711832e-06,
      "loss": 0.3948,
      "step": 11492
    },
    {
      "epoch": 0.7198985264410654,
      "grad_norm": 0.8356046478226828,
      "learning_rate": 1.9197390180902733e-06,
      "loss": 0.3837,
      "step": 11493
    },
    {
      "epoch": 0.7199611644404078,
      "grad_norm": 0.8013325687012477,
      "learning_rate": 1.9189400342528293e-06,
      "loss": 0.3649,
      "step": 11494
    },
    {
      "epoch": 0.7200238024397501,
      "grad_norm": 0.9317568483775337,
      "learning_rate": 1.918141177232384e-06,
      "loss": 0.4132,
      "step": 11495
    },
    {
      "epoch": 0.7200864404390924,
      "grad_norm": 0.8370349590181806,
      "learning_rate": 1.9173424470618213e-06,
      "loss": 0.3629,
      "step": 11496
    },
    {
      "epoch": 0.7201490784384347,
      "grad_norm": 0.8959812163300014,
      "learning_rate": 1.916543843774016e-06,
      "loss": 0.4067,
      "step": 11497
    },
    {
      "epoch": 0.7202117164377769,
      "grad_norm": 0.8306540050517824,
      "learning_rate": 1.915745367401838e-06,
      "loss": 0.3741,
      "step": 11498
    },
    {
      "epoch": 0.7202743544371193,
      "grad_norm": 0.8428474373467443,
      "learning_rate": 1.914947017978153e-06,
      "loss": 0.4153,
      "step": 11499
    },
    {
      "epoch": 0.7203369924364615,
      "grad_norm": 0.9127233676850921,
      "learning_rate": 1.914148795535822e-06,
      "loss": 0.361,
      "step": 11500
    },
    {
      "epoch": 0.7203996304358039,
      "grad_norm": 0.7955006020790325,
      "learning_rate": 1.9133507001076973e-06,
      "loss": 0.3705,
      "step": 11501
    },
    {
      "epoch": 0.7204622684351462,
      "grad_norm": 0.8815940496227525,
      "learning_rate": 1.912552731726632e-06,
      "loss": 0.425,
      "step": 11502
    },
    {
      "epoch": 0.7205249064344885,
      "grad_norm": 0.8061935557140069,
      "learning_rate": 1.9117548904254703e-06,
      "loss": 0.389,
      "step": 11503
    },
    {
      "epoch": 0.7205875444338308,
      "grad_norm": 0.8602831832247971,
      "learning_rate": 1.910957176237048e-06,
      "loss": 0.3989,
      "step": 11504
    },
    {
      "epoch": 0.7206501824331731,
      "grad_norm": 0.7978693577204943,
      "learning_rate": 1.9101595891942037e-06,
      "loss": 0.3998,
      "step": 11505
    },
    {
      "epoch": 0.7207128204325154,
      "grad_norm": 0.8154479049372412,
      "learning_rate": 1.9093621293297644e-06,
      "loss": 0.3667,
      "step": 11506
    },
    {
      "epoch": 0.7207754584318576,
      "grad_norm": 0.8288792296973001,
      "learning_rate": 1.9085647966765536e-06,
      "loss": 0.3767,
      "step": 11507
    },
    {
      "epoch": 0.7208380964312,
      "grad_norm": 0.8344054573646028,
      "learning_rate": 1.9077675912673893e-06,
      "loss": 0.4003,
      "step": 11508
    },
    {
      "epoch": 0.7209007344305423,
      "grad_norm": 0.8587497778040827,
      "learning_rate": 1.9069705131350846e-06,
      "loss": 0.3923,
      "step": 11509
    },
    {
      "epoch": 0.7209633724298846,
      "grad_norm": 0.6492900691994328,
      "learning_rate": 1.9061735623124462e-06,
      "loss": 0.4298,
      "step": 11510
    },
    {
      "epoch": 0.7210260104292269,
      "grad_norm": 0.8029524420103372,
      "learning_rate": 1.9053767388322796e-06,
      "loss": 0.3753,
      "step": 11511
    },
    {
      "epoch": 0.7210886484285692,
      "grad_norm": 0.9056190449502559,
      "learning_rate": 1.9045800427273815e-06,
      "loss": 0.4087,
      "step": 11512
    },
    {
      "epoch": 0.7211512864279115,
      "grad_norm": 0.8479757386959013,
      "learning_rate": 1.9037834740305411e-06,
      "loss": 0.398,
      "step": 11513
    },
    {
      "epoch": 0.7212139244272537,
      "grad_norm": 0.8949829315981095,
      "learning_rate": 1.9029870327745498e-06,
      "loss": 0.4261,
      "step": 11514
    },
    {
      "epoch": 0.7212765624265961,
      "grad_norm": 0.8534673105140321,
      "learning_rate": 1.9021907189921868e-06,
      "loss": 0.4219,
      "step": 11515
    },
    {
      "epoch": 0.7213392004259384,
      "grad_norm": 0.6495649735203132,
      "learning_rate": 1.9013945327162286e-06,
      "loss": 0.4545,
      "step": 11516
    },
    {
      "epoch": 0.7214018384252807,
      "grad_norm": 0.8795296657513703,
      "learning_rate": 1.900598473979447e-06,
      "loss": 0.4116,
      "step": 11517
    },
    {
      "epoch": 0.721464476424623,
      "grad_norm": 0.8140953137558902,
      "learning_rate": 1.8998025428146072e-06,
      "loss": 0.3678,
      "step": 11518
    },
    {
      "epoch": 0.7215271144239653,
      "grad_norm": 0.8696590713529976,
      "learning_rate": 1.8990067392544687e-06,
      "loss": 0.3655,
      "step": 11519
    },
    {
      "epoch": 0.7215897524233076,
      "grad_norm": 0.7650920670233144,
      "learning_rate": 1.8982110633317897e-06,
      "loss": 0.3755,
      "step": 11520
    },
    {
      "epoch": 0.72165239042265,
      "grad_norm": 0.8454765355302898,
      "learning_rate": 1.8974155150793188e-06,
      "loss": 0.3793,
      "step": 11521
    },
    {
      "epoch": 0.7217150284219922,
      "grad_norm": 0.8167713898088031,
      "learning_rate": 1.8966200945298003e-06,
      "loss": 0.359,
      "step": 11522
    },
    {
      "epoch": 0.7217776664213345,
      "grad_norm": 0.8284046954385099,
      "learning_rate": 1.8958248017159758e-06,
      "loss": 0.3626,
      "step": 11523
    },
    {
      "epoch": 0.7218403044206768,
      "grad_norm": 0.8175740832081264,
      "learning_rate": 1.8950296366705788e-06,
      "loss": 0.369,
      "step": 11524
    },
    {
      "epoch": 0.7219029424200191,
      "grad_norm": 0.8212561777149703,
      "learning_rate": 1.8942345994263384e-06,
      "loss": 0.3879,
      "step": 11525
    },
    {
      "epoch": 0.7219655804193614,
      "grad_norm": 0.8710844246068417,
      "learning_rate": 1.8934396900159786e-06,
      "loss": 0.3913,
      "step": 11526
    },
    {
      "epoch": 0.7220282184187037,
      "grad_norm": 0.8864178978123639,
      "learning_rate": 1.8926449084722153e-06,
      "loss": 0.3996,
      "step": 11527
    },
    {
      "epoch": 0.722090856418046,
      "grad_norm": 0.9041440653140244,
      "learning_rate": 1.8918502548277668e-06,
      "loss": 0.4181,
      "step": 11528
    },
    {
      "epoch": 0.7221534944173883,
      "grad_norm": 0.8131174006351102,
      "learning_rate": 1.8910557291153382e-06,
      "loss": 0.3741,
      "step": 11529
    },
    {
      "epoch": 0.7222161324167307,
      "grad_norm": 0.8108864519603487,
      "learning_rate": 1.8902613313676328e-06,
      "loss": 0.3477,
      "step": 11530
    },
    {
      "epoch": 0.7222787704160729,
      "grad_norm": 0.9743825205737506,
      "learning_rate": 1.8894670616173482e-06,
      "loss": 0.4233,
      "step": 11531
    },
    {
      "epoch": 0.7223414084154152,
      "grad_norm": 0.8985623285665918,
      "learning_rate": 1.8886729198971753e-06,
      "loss": 0.423,
      "step": 11532
    },
    {
      "epoch": 0.7224040464147575,
      "grad_norm": 0.905385921157256,
      "learning_rate": 1.887878906239804e-06,
      "loss": 0.4001,
      "step": 11533
    },
    {
      "epoch": 0.7224666844140998,
      "grad_norm": 0.8303658870123407,
      "learning_rate": 1.8870850206779145e-06,
      "loss": 0.3546,
      "step": 11534
    },
    {
      "epoch": 0.7225293224134421,
      "grad_norm": 0.8128278433037668,
      "learning_rate": 1.8862912632441837e-06,
      "loss": 0.3618,
      "step": 11535
    },
    {
      "epoch": 0.7225919604127844,
      "grad_norm": 0.7811717230957717,
      "learning_rate": 1.8854976339712806e-06,
      "loss": 0.3545,
      "step": 11536
    },
    {
      "epoch": 0.7226545984121268,
      "grad_norm": 0.8584645224820568,
      "learning_rate": 1.8847041328918748e-06,
      "loss": 0.4241,
      "step": 11537
    },
    {
      "epoch": 0.722717236411469,
      "grad_norm": 0.8489179918770334,
      "learning_rate": 1.8839107600386246e-06,
      "loss": 0.4229,
      "step": 11538
    },
    {
      "epoch": 0.7227798744108114,
      "grad_norm": 0.8232241524940306,
      "learning_rate": 1.8831175154441861e-06,
      "loss": 0.3883,
      "step": 11539
    },
    {
      "epoch": 0.7228425124101536,
      "grad_norm": 0.8922387852080396,
      "learning_rate": 1.8823243991412093e-06,
      "loss": 0.3975,
      "step": 11540
    },
    {
      "epoch": 0.7229051504094959,
      "grad_norm": 0.9374083386952053,
      "learning_rate": 1.8815314111623373e-06,
      "loss": 0.4343,
      "step": 11541
    },
    {
      "epoch": 0.7229677884088382,
      "grad_norm": 0.9274120169760124,
      "learning_rate": 1.8807385515402127e-06,
      "loss": 0.3969,
      "step": 11542
    },
    {
      "epoch": 0.7230304264081805,
      "grad_norm": 0.8293895193265317,
      "learning_rate": 1.8799458203074683e-06,
      "loss": 0.3671,
      "step": 11543
    },
    {
      "epoch": 0.7230930644075229,
      "grad_norm": 0.7934177586230541,
      "learning_rate": 1.8791532174967331e-06,
      "loss": 0.3831,
      "step": 11544
    },
    {
      "epoch": 0.7231557024068651,
      "grad_norm": 0.8419728600745029,
      "learning_rate": 1.8783607431406292e-06,
      "loss": 0.3811,
      "step": 11545
    },
    {
      "epoch": 0.7232183404062075,
      "grad_norm": 0.8359841043513674,
      "learning_rate": 1.8775683972717778e-06,
      "loss": 0.3513,
      "step": 11546
    },
    {
      "epoch": 0.7232809784055497,
      "grad_norm": 0.8194521060647252,
      "learning_rate": 1.8767761799227908e-06,
      "loss": 0.3905,
      "step": 11547
    },
    {
      "epoch": 0.723343616404892,
      "grad_norm": 0.8269666816331928,
      "learning_rate": 1.8759840911262762e-06,
      "loss": 0.3442,
      "step": 11548
    },
    {
      "epoch": 0.7234062544042343,
      "grad_norm": 0.9076790919495233,
      "learning_rate": 1.8751921309148363e-06,
      "loss": 0.3774,
      "step": 11549
    },
    {
      "epoch": 0.7234688924035766,
      "grad_norm": 0.8373402307061402,
      "learning_rate": 1.8744002993210663e-06,
      "loss": 0.3903,
      "step": 11550
    },
    {
      "epoch": 0.723531530402919,
      "grad_norm": 0.8589226078663483,
      "learning_rate": 1.873608596377562e-06,
      "loss": 0.4061,
      "step": 11551
    },
    {
      "epoch": 0.7235941684022612,
      "grad_norm": 0.8034849068718678,
      "learning_rate": 1.872817022116908e-06,
      "loss": 0.3894,
      "step": 11552
    },
    {
      "epoch": 0.7236568064016036,
      "grad_norm": 0.8295339811230732,
      "learning_rate": 1.8720255765716844e-06,
      "loss": 0.4074,
      "step": 11553
    },
    {
      "epoch": 0.7237194444009458,
      "grad_norm": 0.8830668590890735,
      "learning_rate": 1.8712342597744698e-06,
      "loss": 0.4187,
      "step": 11554
    },
    {
      "epoch": 0.7237820824002882,
      "grad_norm": 0.848756153091303,
      "learning_rate": 1.8704430717578342e-06,
      "loss": 0.3979,
      "step": 11555
    },
    {
      "epoch": 0.7238447203996304,
      "grad_norm": 0.8428489661620451,
      "learning_rate": 1.8696520125543427e-06,
      "loss": 0.3812,
      "step": 11556
    },
    {
      "epoch": 0.7239073583989727,
      "grad_norm": 0.9393338339662239,
      "learning_rate": 1.8688610821965553e-06,
      "loss": 0.411,
      "step": 11557
    },
    {
      "epoch": 0.723969996398315,
      "grad_norm": 0.8278678721629781,
      "learning_rate": 1.868070280717027e-06,
      "loss": 0.4,
      "step": 11558
    },
    {
      "epoch": 0.7240326343976573,
      "grad_norm": 0.6080548987287275,
      "learning_rate": 1.8672796081483051e-06,
      "loss": 0.4278,
      "step": 11559
    },
    {
      "epoch": 0.7240952723969997,
      "grad_norm": 0.8815366846492728,
      "learning_rate": 1.8664890645229384e-06,
      "loss": 0.3753,
      "step": 11560
    },
    {
      "epoch": 0.7241579103963419,
      "grad_norm": 0.8387375665737419,
      "learning_rate": 1.865698649873463e-06,
      "loss": 0.3885,
      "step": 11561
    },
    {
      "epoch": 0.7242205483956843,
      "grad_norm": 0.6370851530377826,
      "learning_rate": 1.8649083642324112e-06,
      "loss": 0.4237,
      "step": 11562
    },
    {
      "epoch": 0.7242831863950265,
      "grad_norm": 0.7789604858638645,
      "learning_rate": 1.864118207632315e-06,
      "loss": 0.3813,
      "step": 11563
    },
    {
      "epoch": 0.7243458243943689,
      "grad_norm": 0.8890011062275455,
      "learning_rate": 1.863328180105695e-06,
      "loss": 0.4317,
      "step": 11564
    },
    {
      "epoch": 0.7244084623937112,
      "grad_norm": 0.8438711548313749,
      "learning_rate": 1.8625382816850695e-06,
      "loss": 0.4141,
      "step": 11565
    },
    {
      "epoch": 0.7244711003930534,
      "grad_norm": 0.9479455441592307,
      "learning_rate": 1.861748512402951e-06,
      "loss": 0.4077,
      "step": 11566
    },
    {
      "epoch": 0.7245337383923958,
      "grad_norm": 0.8690370615909231,
      "learning_rate": 1.860958872291846e-06,
      "loss": 0.3911,
      "step": 11567
    },
    {
      "epoch": 0.724596376391738,
      "grad_norm": 0.8345970317508662,
      "learning_rate": 1.8601693613842547e-06,
      "loss": 0.4141,
      "step": 11568
    },
    {
      "epoch": 0.7246590143910804,
      "grad_norm": 0.8627231177886144,
      "learning_rate": 1.8593799797126766e-06,
      "loss": 0.3857,
      "step": 11569
    },
    {
      "epoch": 0.7247216523904226,
      "grad_norm": 0.8632526757463818,
      "learning_rate": 1.8585907273096022e-06,
      "loss": 0.4051,
      "step": 11570
    },
    {
      "epoch": 0.724784290389765,
      "grad_norm": 0.8485354343448515,
      "learning_rate": 1.8578016042075143e-06,
      "loss": 0.3856,
      "step": 11571
    },
    {
      "epoch": 0.7248469283891072,
      "grad_norm": 0.7685088392546271,
      "learning_rate": 1.8570126104388975e-06,
      "loss": 0.3797,
      "step": 11572
    },
    {
      "epoch": 0.7249095663884495,
      "grad_norm": 0.8581548290269924,
      "learning_rate": 1.8562237460362247e-06,
      "loss": 0.3972,
      "step": 11573
    },
    {
      "epoch": 0.7249722043877919,
      "grad_norm": 0.5841658173580331,
      "learning_rate": 1.8554350110319663e-06,
      "loss": 0.4397,
      "step": 11574
    },
    {
      "epoch": 0.7250348423871341,
      "grad_norm": 0.8421278587504324,
      "learning_rate": 1.8546464054585867e-06,
      "loss": 0.3831,
      "step": 11575
    },
    {
      "epoch": 0.7250974803864765,
      "grad_norm": 0.8093501374484341,
      "learning_rate": 1.8538579293485442e-06,
      "loss": 0.3598,
      "step": 11576
    },
    {
      "epoch": 0.7251601183858187,
      "grad_norm": 0.8056207762268858,
      "learning_rate": 1.8530695827342938e-06,
      "loss": 0.3551,
      "step": 11577
    },
    {
      "epoch": 0.7252227563851611,
      "grad_norm": 0.843324021396696,
      "learning_rate": 1.8522813656482818e-06,
      "loss": 0.399,
      "step": 11578
    },
    {
      "epoch": 0.7252853943845033,
      "grad_norm": 0.8261802030754893,
      "learning_rate": 1.8514932781229527e-06,
      "loss": 0.4048,
      "step": 11579
    },
    {
      "epoch": 0.7253480323838457,
      "grad_norm": 0.8197070457835945,
      "learning_rate": 1.8507053201907465e-06,
      "loss": 0.4424,
      "step": 11580
    },
    {
      "epoch": 0.725410670383188,
      "grad_norm": 0.6078657814880475,
      "learning_rate": 1.849917491884094e-06,
      "loss": 0.4416,
      "step": 11581
    },
    {
      "epoch": 0.7254733083825302,
      "grad_norm": 0.8401840294130575,
      "learning_rate": 1.8491297932354224e-06,
      "loss": 0.3572,
      "step": 11582
    },
    {
      "epoch": 0.7255359463818726,
      "grad_norm": 0.7936366815474412,
      "learning_rate": 1.8483422242771532e-06,
      "loss": 0.3742,
      "step": 11583
    },
    {
      "epoch": 0.7255985843812148,
      "grad_norm": 0.8576444537795179,
      "learning_rate": 1.8475547850417031e-06,
      "loss": 0.3695,
      "step": 11584
    },
    {
      "epoch": 0.7256612223805572,
      "grad_norm": 0.7858856078729853,
      "learning_rate": 1.8467674755614828e-06,
      "loss": 0.3208,
      "step": 11585
    },
    {
      "epoch": 0.7257238603798994,
      "grad_norm": 0.7628718822495226,
      "learning_rate": 1.8459802958688965e-06,
      "loss": 0.3942,
      "step": 11586
    },
    {
      "epoch": 0.7257864983792418,
      "grad_norm": 0.7983090309126889,
      "learning_rate": 1.845193245996349e-06,
      "loss": 0.3827,
      "step": 11587
    },
    {
      "epoch": 0.7258491363785841,
      "grad_norm": 0.9295360564916245,
      "learning_rate": 1.8444063259762314e-06,
      "loss": 0.4243,
      "step": 11588
    },
    {
      "epoch": 0.7259117743779264,
      "grad_norm": 0.845553789589911,
      "learning_rate": 1.8436195358409365e-06,
      "loss": 0.3897,
      "step": 11589
    },
    {
      "epoch": 0.7259744123772687,
      "grad_norm": 0.8862593768099601,
      "learning_rate": 1.8428328756228476e-06,
      "loss": 0.4607,
      "step": 11590
    },
    {
      "epoch": 0.7260370503766109,
      "grad_norm": 0.8371952714754041,
      "learning_rate": 1.842046345354343e-06,
      "loss": 0.3678,
      "step": 11591
    },
    {
      "epoch": 0.7260996883759533,
      "grad_norm": 0.7839763293349497,
      "learning_rate": 1.8412599450677975e-06,
      "loss": 0.4089,
      "step": 11592
    },
    {
      "epoch": 0.7261623263752955,
      "grad_norm": 0.5747511765715834,
      "learning_rate": 1.8404736747955787e-06,
      "loss": 0.4357,
      "step": 11593
    },
    {
      "epoch": 0.7262249643746379,
      "grad_norm": 0.8383959003780519,
      "learning_rate": 1.8396875345700498e-06,
      "loss": 0.3611,
      "step": 11594
    },
    {
      "epoch": 0.7262876023739802,
      "grad_norm": 0.7998382017531057,
      "learning_rate": 1.8389015244235674e-06,
      "loss": 0.3675,
      "step": 11595
    },
    {
      "epoch": 0.7263502403733225,
      "grad_norm": 0.8724650624187651,
      "learning_rate": 1.8381156443884862e-06,
      "loss": 0.4297,
      "step": 11596
    },
    {
      "epoch": 0.7264128783726648,
      "grad_norm": 0.9063330310198633,
      "learning_rate": 1.83732989449715e-06,
      "loss": 0.4164,
      "step": 11597
    },
    {
      "epoch": 0.726475516372007,
      "grad_norm": 0.8358069090342459,
      "learning_rate": 1.8365442747819046e-06,
      "loss": 0.3358,
      "step": 11598
    },
    {
      "epoch": 0.7265381543713494,
      "grad_norm": 0.9686100252110155,
      "learning_rate": 1.835758785275084e-06,
      "loss": 0.3972,
      "step": 11599
    },
    {
      "epoch": 0.7266007923706916,
      "grad_norm": 0.830922491620201,
      "learning_rate": 1.8349734260090192e-06,
      "loss": 0.3829,
      "step": 11600
    },
    {
      "epoch": 0.726663430370034,
      "grad_norm": 0.9001166877413397,
      "learning_rate": 1.8341881970160352e-06,
      "loss": 0.4118,
      "step": 11601
    },
    {
      "epoch": 0.7267260683693763,
      "grad_norm": 0.8855850651959551,
      "learning_rate": 1.8334030983284528e-06,
      "loss": 0.3489,
      "step": 11602
    },
    {
      "epoch": 0.7267887063687186,
      "grad_norm": 0.8893476084999701,
      "learning_rate": 1.8326181299785867e-06,
      "loss": 0.4143,
      "step": 11603
    },
    {
      "epoch": 0.7268513443680609,
      "grad_norm": 0.865938207760662,
      "learning_rate": 1.831833291998744e-06,
      "loss": 0.3619,
      "step": 11604
    },
    {
      "epoch": 0.7269139823674032,
      "grad_norm": 0.7965119324998581,
      "learning_rate": 1.8310485844212316e-06,
      "loss": 0.3907,
      "step": 11605
    },
    {
      "epoch": 0.7269766203667455,
      "grad_norm": 0.8119465902810905,
      "learning_rate": 1.8302640072783495e-06,
      "loss": 0.3522,
      "step": 11606
    },
    {
      "epoch": 0.7270392583660877,
      "grad_norm": 0.8916018163427769,
      "learning_rate": 1.829479560602389e-06,
      "loss": 0.4237,
      "step": 11607
    },
    {
      "epoch": 0.7271018963654301,
      "grad_norm": 0.5721348776962013,
      "learning_rate": 1.8286952444256384e-06,
      "loss": 0.4489,
      "step": 11608
    },
    {
      "epoch": 0.7271645343647724,
      "grad_norm": 0.8306188041985345,
      "learning_rate": 1.82791105878038e-06,
      "loss": 0.3899,
      "step": 11609
    },
    {
      "epoch": 0.7272271723641147,
      "grad_norm": 0.8882668730236837,
      "learning_rate": 1.8271270036988914e-06,
      "loss": 0.4008,
      "step": 11610
    },
    {
      "epoch": 0.727289810363457,
      "grad_norm": 0.8315356810320012,
      "learning_rate": 1.826343079213444e-06,
      "loss": 0.3857,
      "step": 11611
    },
    {
      "epoch": 0.7273524483627993,
      "grad_norm": 0.8754419425709976,
      "learning_rate": 1.8255592853563027e-06,
      "loss": 0.4046,
      "step": 11612
    },
    {
      "epoch": 0.7274150863621416,
      "grad_norm": 0.8369478450911685,
      "learning_rate": 1.8247756221597324e-06,
      "loss": 0.3567,
      "step": 11613
    },
    {
      "epoch": 0.727477724361484,
      "grad_norm": 0.8579502340172138,
      "learning_rate": 1.8239920896559866e-06,
      "loss": 0.4033,
      "step": 11614
    },
    {
      "epoch": 0.7275403623608262,
      "grad_norm": 0.8252353075110065,
      "learning_rate": 1.8232086878773142e-06,
      "loss": 0.3791,
      "step": 11615
    },
    {
      "epoch": 0.7276030003601685,
      "grad_norm": 0.8184518183336933,
      "learning_rate": 1.8224254168559636e-06,
      "loss": 0.4099,
      "step": 11616
    },
    {
      "epoch": 0.7276656383595108,
      "grad_norm": 0.8062776743201024,
      "learning_rate": 1.8216422766241725e-06,
      "loss": 0.4155,
      "step": 11617
    },
    {
      "epoch": 0.7277282763588531,
      "grad_norm": 0.8261265063956978,
      "learning_rate": 1.8208592672141756e-06,
      "loss": 0.3932,
      "step": 11618
    },
    {
      "epoch": 0.7277909143581954,
      "grad_norm": 0.8529836244555212,
      "learning_rate": 1.8200763886582012e-06,
      "loss": 0.3886,
      "step": 11619
    },
    {
      "epoch": 0.7278535523575377,
      "grad_norm": 0.7769068681847423,
      "learning_rate": 1.8192936409884725e-06,
      "loss": 0.3947,
      "step": 11620
    },
    {
      "epoch": 0.72791619035688,
      "grad_norm": 0.9150644665231674,
      "learning_rate": 1.8185110242372067e-06,
      "loss": 0.4327,
      "step": 11621
    },
    {
      "epoch": 0.7279788283562223,
      "grad_norm": 0.860247537465197,
      "learning_rate": 1.817728538436619e-06,
      "loss": 0.4297,
      "step": 11622
    },
    {
      "epoch": 0.7280414663555645,
      "grad_norm": 0.8694836392803998,
      "learning_rate": 1.8169461836189156e-06,
      "loss": 0.4128,
      "step": 11623
    },
    {
      "epoch": 0.7281041043549069,
      "grad_norm": 0.824258303173113,
      "learning_rate": 1.8161639598162961e-06,
      "loss": 0.4051,
      "step": 11624
    },
    {
      "epoch": 0.7281667423542492,
      "grad_norm": 0.801891482666869,
      "learning_rate": 1.8153818670609607e-06,
      "loss": 0.3575,
      "step": 11625
    },
    {
      "epoch": 0.7282293803535915,
      "grad_norm": 0.7740201755855333,
      "learning_rate": 1.8145999053850994e-06,
      "loss": 0.3384,
      "step": 11626
    },
    {
      "epoch": 0.7282920183529338,
      "grad_norm": 0.8238406314551925,
      "learning_rate": 1.813818074820896e-06,
      "loss": 0.3634,
      "step": 11627
    },
    {
      "epoch": 0.7283546563522761,
      "grad_norm": 0.8343582893013708,
      "learning_rate": 1.813036375400533e-06,
      "loss": 0.3585,
      "step": 11628
    },
    {
      "epoch": 0.7284172943516184,
      "grad_norm": 0.8749194598595453,
      "learning_rate": 1.812254807156184e-06,
      "loss": 0.3879,
      "step": 11629
    },
    {
      "epoch": 0.7284799323509608,
      "grad_norm": 0.8015800514399987,
      "learning_rate": 1.811473370120017e-06,
      "loss": 0.3739,
      "step": 11630
    },
    {
      "epoch": 0.728542570350303,
      "grad_norm": 0.8745388941274268,
      "learning_rate": 1.8106920643241998e-06,
      "loss": 0.4347,
      "step": 11631
    },
    {
      "epoch": 0.7286052083496453,
      "grad_norm": 0.9125083015359058,
      "learning_rate": 1.8099108898008894e-06,
      "loss": 0.4091,
      "step": 11632
    },
    {
      "epoch": 0.7286678463489876,
      "grad_norm": 0.8008233700778663,
      "learning_rate": 1.8091298465822365e-06,
      "loss": 0.3677,
      "step": 11633
    },
    {
      "epoch": 0.7287304843483299,
      "grad_norm": 0.8632263774546691,
      "learning_rate": 1.8083489347003941e-06,
      "loss": 0.3648,
      "step": 11634
    },
    {
      "epoch": 0.7287931223476722,
      "grad_norm": 0.8308987906476623,
      "learning_rate": 1.8075681541875013e-06,
      "loss": 0.3614,
      "step": 11635
    },
    {
      "epoch": 0.7288557603470145,
      "grad_norm": 0.8246014926186072,
      "learning_rate": 1.8067875050756956e-06,
      "loss": 0.3567,
      "step": 11636
    },
    {
      "epoch": 0.7289183983463569,
      "grad_norm": 0.9157858696674444,
      "learning_rate": 1.8060069873971091e-06,
      "loss": 0.4405,
      "step": 11637
    },
    {
      "epoch": 0.7289810363456991,
      "grad_norm": 0.8305103596732755,
      "learning_rate": 1.805226601183866e-06,
      "loss": 0.3684,
      "step": 11638
    },
    {
      "epoch": 0.7290436743450415,
      "grad_norm": 0.9221113468744381,
      "learning_rate": 1.8044463464680912e-06,
      "loss": 0.4423,
      "step": 11639
    },
    {
      "epoch": 0.7291063123443837,
      "grad_norm": 0.8687393640498731,
      "learning_rate": 1.8036662232818974e-06,
      "loss": 0.412,
      "step": 11640
    },
    {
      "epoch": 0.729168950343726,
      "grad_norm": 0.9221301952145349,
      "learning_rate": 1.8028862316573952e-06,
      "loss": 0.4102,
      "step": 11641
    },
    {
      "epoch": 0.7292315883430683,
      "grad_norm": 0.9233899532290715,
      "learning_rate": 1.8021063716266873e-06,
      "loss": 0.3935,
      "step": 11642
    },
    {
      "epoch": 0.7292942263424106,
      "grad_norm": 0.7840301522684845,
      "learning_rate": 1.8013266432218767e-06,
      "loss": 0.3858,
      "step": 11643
    },
    {
      "epoch": 0.729356864341753,
      "grad_norm": 0.6162146893584046,
      "learning_rate": 1.8005470464750552e-06,
      "loss": 0.4719,
      "step": 11644
    },
    {
      "epoch": 0.7294195023410952,
      "grad_norm": 0.8144806723881514,
      "learning_rate": 1.7997675814183108e-06,
      "loss": 0.3637,
      "step": 11645
    },
    {
      "epoch": 0.7294821403404376,
      "grad_norm": 0.8297578793050774,
      "learning_rate": 1.7989882480837267e-06,
      "loss": 0.3715,
      "step": 11646
    },
    {
      "epoch": 0.7295447783397798,
      "grad_norm": 0.8643221580632899,
      "learning_rate": 1.7982090465033792e-06,
      "loss": 0.3802,
      "step": 11647
    },
    {
      "epoch": 0.7296074163391222,
      "grad_norm": 0.8818049569771758,
      "learning_rate": 1.7974299767093433e-06,
      "loss": 0.3704,
      "step": 11648
    },
    {
      "epoch": 0.7296700543384644,
      "grad_norm": 0.8670598389880269,
      "learning_rate": 1.7966510387336834e-06,
      "loss": 0.4282,
      "step": 11649
    },
    {
      "epoch": 0.7297326923378067,
      "grad_norm": 0.8575107428799013,
      "learning_rate": 1.7958722326084616e-06,
      "loss": 0.3736,
      "step": 11650
    },
    {
      "epoch": 0.729795330337149,
      "grad_norm": 0.6390048142315615,
      "learning_rate": 1.795093558365732e-06,
      "loss": 0.4449,
      "step": 11651
    },
    {
      "epoch": 0.7298579683364913,
      "grad_norm": 0.9325939267719561,
      "learning_rate": 1.7943150160375477e-06,
      "loss": 0.4427,
      "step": 11652
    },
    {
      "epoch": 0.7299206063358337,
      "grad_norm": 0.920323926280183,
      "learning_rate": 1.7935366056559529e-06,
      "loss": 0.436,
      "step": 11653
    },
    {
      "epoch": 0.7299832443351759,
      "grad_norm": 0.8637694142047729,
      "learning_rate": 1.7927583272529858e-06,
      "loss": 0.3875,
      "step": 11654
    },
    {
      "epoch": 0.7300458823345183,
      "grad_norm": 0.9011027717760662,
      "learning_rate": 1.7919801808606819e-06,
      "loss": 0.3925,
      "step": 11655
    },
    {
      "epoch": 0.7301085203338605,
      "grad_norm": 0.8403223677900985,
      "learning_rate": 1.7912021665110674e-06,
      "loss": 0.4113,
      "step": 11656
    },
    {
      "epoch": 0.7301711583332028,
      "grad_norm": 0.7933267675159567,
      "learning_rate": 1.7904242842361685e-06,
      "loss": 0.3936,
      "step": 11657
    },
    {
      "epoch": 0.7302337963325451,
      "grad_norm": 0.8399792037835626,
      "learning_rate": 1.789646534068002e-06,
      "loss": 0.3928,
      "step": 11658
    },
    {
      "epoch": 0.7302964343318874,
      "grad_norm": 0.7805100728523086,
      "learning_rate": 1.7888689160385803e-06,
      "loss": 0.3655,
      "step": 11659
    },
    {
      "epoch": 0.7303590723312298,
      "grad_norm": 0.8274531309441931,
      "learning_rate": 1.7880914301799101e-06,
      "loss": 0.3584,
      "step": 11660
    },
    {
      "epoch": 0.730421710330572,
      "grad_norm": 0.9111621442329982,
      "learning_rate": 1.7873140765239905e-06,
      "loss": 0.3719,
      "step": 11661
    },
    {
      "epoch": 0.7304843483299144,
      "grad_norm": 0.8241040192802938,
      "learning_rate": 1.7865368551028223e-06,
      "loss": 0.3858,
      "step": 11662
    },
    {
      "epoch": 0.7305469863292566,
      "grad_norm": 0.8744389515366764,
      "learning_rate": 1.7857597659483927e-06,
      "loss": 0.3905,
      "step": 11663
    },
    {
      "epoch": 0.730609624328599,
      "grad_norm": 1.0228310489429735,
      "learning_rate": 1.7849828090926869e-06,
      "loss": 0.4211,
      "step": 11664
    },
    {
      "epoch": 0.7306722623279412,
      "grad_norm": 0.8206668391736842,
      "learning_rate": 1.7842059845676862e-06,
      "loss": 0.3764,
      "step": 11665
    },
    {
      "epoch": 0.7307349003272835,
      "grad_norm": 0.8293066981381717,
      "learning_rate": 1.7834292924053649e-06,
      "loss": 0.3916,
      "step": 11666
    },
    {
      "epoch": 0.7307975383266259,
      "grad_norm": 0.888041260474085,
      "learning_rate": 1.7826527326376908e-06,
      "loss": 0.396,
      "step": 11667
    },
    {
      "epoch": 0.7308601763259681,
      "grad_norm": 0.829244337900466,
      "learning_rate": 1.781876305296627e-06,
      "loss": 0.3707,
      "step": 11668
    },
    {
      "epoch": 0.7309228143253105,
      "grad_norm": 0.871302605157666,
      "learning_rate": 1.7811000104141318e-06,
      "loss": 0.386,
      "step": 11669
    },
    {
      "epoch": 0.7309854523246527,
      "grad_norm": 0.8750785471978336,
      "learning_rate": 1.7803238480221563e-06,
      "loss": 0.423,
      "step": 11670
    },
    {
      "epoch": 0.7310480903239951,
      "grad_norm": 0.9078770229447739,
      "learning_rate": 1.7795478181526505e-06,
      "loss": 0.3917,
      "step": 11671
    },
    {
      "epoch": 0.7311107283233373,
      "grad_norm": 0.811434402506633,
      "learning_rate": 1.7787719208375536e-06,
      "loss": 0.3747,
      "step": 11672
    },
    {
      "epoch": 0.7311733663226797,
      "grad_norm": 0.8396826879701155,
      "learning_rate": 1.7779961561088016e-06,
      "loss": 0.3836,
      "step": 11673
    },
    {
      "epoch": 0.731236004322022,
      "grad_norm": 0.823390689218042,
      "learning_rate": 1.777220523998327e-06,
      "loss": 0.3679,
      "step": 11674
    },
    {
      "epoch": 0.7312986423213642,
      "grad_norm": 0.8041324841678839,
      "learning_rate": 1.7764450245380539e-06,
      "loss": 0.4101,
      "step": 11675
    },
    {
      "epoch": 0.7313612803207066,
      "grad_norm": 0.8745814735725665,
      "learning_rate": 1.7756696577599014e-06,
      "loss": 0.4075,
      "step": 11676
    },
    {
      "epoch": 0.7314239183200488,
      "grad_norm": 0.919457117719914,
      "learning_rate": 1.774894423695785e-06,
      "loss": 0.3903,
      "step": 11677
    },
    {
      "epoch": 0.7314865563193912,
      "grad_norm": 0.8462444494655559,
      "learning_rate": 1.7741193223776127e-06,
      "loss": 0.3581,
      "step": 11678
    },
    {
      "epoch": 0.7315491943187334,
      "grad_norm": 0.9294772902472179,
      "learning_rate": 1.7733443538372858e-06,
      "loss": 0.419,
      "step": 11679
    },
    {
      "epoch": 0.7316118323180758,
      "grad_norm": 0.9390269390222808,
      "learning_rate": 1.772569518106706e-06,
      "loss": 0.389,
      "step": 11680
    },
    {
      "epoch": 0.7316744703174181,
      "grad_norm": 0.9223503354936649,
      "learning_rate": 1.7717948152177638e-06,
      "loss": 0.3679,
      "step": 11681
    },
    {
      "epoch": 0.7317371083167603,
      "grad_norm": 0.8653334652221994,
      "learning_rate": 1.7710202452023444e-06,
      "loss": 0.4118,
      "step": 11682
    },
    {
      "epoch": 0.7317997463161027,
      "grad_norm": 0.8663799730849773,
      "learning_rate": 1.7702458080923329e-06,
      "loss": 0.3897,
      "step": 11683
    },
    {
      "epoch": 0.7318623843154449,
      "grad_norm": 0.8475195702811681,
      "learning_rate": 1.7694715039196032e-06,
      "loss": 0.3889,
      "step": 11684
    },
    {
      "epoch": 0.7319250223147873,
      "grad_norm": 0.831466129459279,
      "learning_rate": 1.7686973327160261e-06,
      "loss": 0.3715,
      "step": 11685
    },
    {
      "epoch": 0.7319876603141295,
      "grad_norm": 0.9294735639590042,
      "learning_rate": 1.7679232945134668e-06,
      "loss": 0.419,
      "step": 11686
    },
    {
      "epoch": 0.7320502983134719,
      "grad_norm": 0.8787576199332603,
      "learning_rate": 1.7671493893437847e-06,
      "loss": 0.4062,
      "step": 11687
    },
    {
      "epoch": 0.7321129363128142,
      "grad_norm": 0.8452148666844048,
      "learning_rate": 1.7663756172388318e-06,
      "loss": 0.3434,
      "step": 11688
    },
    {
      "epoch": 0.7321755743121565,
      "grad_norm": 0.893393618877138,
      "learning_rate": 1.7656019782304602e-06,
      "loss": 0.3829,
      "step": 11689
    },
    {
      "epoch": 0.7322382123114988,
      "grad_norm": 0.8121257464320836,
      "learning_rate": 1.7648284723505098e-06,
      "loss": 0.3565,
      "step": 11690
    },
    {
      "epoch": 0.732300850310841,
      "grad_norm": 0.9249338303580062,
      "learning_rate": 1.764055099630822e-06,
      "loss": 0.3991,
      "step": 11691
    },
    {
      "epoch": 0.7323634883101834,
      "grad_norm": 0.8784266832630947,
      "learning_rate": 1.763281860103227e-06,
      "loss": 0.4187,
      "step": 11692
    },
    {
      "epoch": 0.7324261263095256,
      "grad_norm": 0.9071341372320457,
      "learning_rate": 1.7625087537995512e-06,
      "loss": 0.3964,
      "step": 11693
    },
    {
      "epoch": 0.732488764308868,
      "grad_norm": 0.8672218406664063,
      "learning_rate": 1.7617357807516162e-06,
      "loss": 0.3874,
      "step": 11694
    },
    {
      "epoch": 0.7325514023082103,
      "grad_norm": 0.9452802315216748,
      "learning_rate": 1.7609629409912371e-06,
      "loss": 0.3973,
      "step": 11695
    },
    {
      "epoch": 0.7326140403075526,
      "grad_norm": 0.5619930271544324,
      "learning_rate": 1.7601902345502248e-06,
      "loss": 0.4324,
      "step": 11696
    },
    {
      "epoch": 0.7326766783068949,
      "grad_norm": 0.8126362080806508,
      "learning_rate": 1.759417661460382e-06,
      "loss": 0.402,
      "step": 11697
    },
    {
      "epoch": 0.7327393163062372,
      "grad_norm": 0.8145201055443283,
      "learning_rate": 1.7586452217535116e-06,
      "loss": 0.3369,
      "step": 11698
    },
    {
      "epoch": 0.7328019543055795,
      "grad_norm": 0.9348413149317552,
      "learning_rate": 1.7578729154614032e-06,
      "loss": 0.3976,
      "step": 11699
    },
    {
      "epoch": 0.7328645923049217,
      "grad_norm": 0.8484552414071665,
      "learning_rate": 1.75710074261585e-06,
      "loss": 0.3583,
      "step": 11700
    },
    {
      "epoch": 0.7329272303042641,
      "grad_norm": 0.8172037039201933,
      "learning_rate": 1.7563287032486316e-06,
      "loss": 0.3953,
      "step": 11701
    },
    {
      "epoch": 0.7329898683036064,
      "grad_norm": 0.7902367837461853,
      "learning_rate": 1.7555567973915256e-06,
      "loss": 0.3927,
      "step": 11702
    },
    {
      "epoch": 0.7330525063029487,
      "grad_norm": 0.8505798816971754,
      "learning_rate": 1.7547850250763042e-06,
      "loss": 0.4144,
      "step": 11703
    },
    {
      "epoch": 0.733115144302291,
      "grad_norm": 0.8211427218286801,
      "learning_rate": 1.7540133863347342e-06,
      "loss": 0.3711,
      "step": 11704
    },
    {
      "epoch": 0.7331777823016333,
      "grad_norm": 0.8149989005455122,
      "learning_rate": 1.7532418811985746e-06,
      "loss": 0.3633,
      "step": 11705
    },
    {
      "epoch": 0.7332404203009756,
      "grad_norm": 0.8748539525669868,
      "learning_rate": 1.7524705096995808e-06,
      "loss": 0.4132,
      "step": 11706
    },
    {
      "epoch": 0.7333030583003178,
      "grad_norm": 0.8527397492362347,
      "learning_rate": 1.7516992718695036e-06,
      "loss": 0.3899,
      "step": 11707
    },
    {
      "epoch": 0.7333656962996602,
      "grad_norm": 0.8298736553534817,
      "learning_rate": 1.7509281677400885e-06,
      "loss": 0.3599,
      "step": 11708
    },
    {
      "epoch": 0.7334283342990024,
      "grad_norm": 0.8757712957992435,
      "learning_rate": 1.7501571973430736e-06,
      "loss": 0.355,
      "step": 11709
    },
    {
      "epoch": 0.7334909722983448,
      "grad_norm": 0.914713962961841,
      "learning_rate": 1.7493863607101914e-06,
      "loss": 0.4165,
      "step": 11710
    },
    {
      "epoch": 0.7335536102976871,
      "grad_norm": 0.8327027744282908,
      "learning_rate": 1.74861565787317e-06,
      "loss": 0.355,
      "step": 11711
    },
    {
      "epoch": 0.7336162482970294,
      "grad_norm": 0.8612865120281747,
      "learning_rate": 1.7478450888637317e-06,
      "loss": 0.3833,
      "step": 11712
    },
    {
      "epoch": 0.7336788862963717,
      "grad_norm": 0.8781669697156335,
      "learning_rate": 1.7470746537135929e-06,
      "loss": 0.3785,
      "step": 11713
    },
    {
      "epoch": 0.733741524295714,
      "grad_norm": 0.7952540971571515,
      "learning_rate": 1.7463043524544654e-06,
      "loss": 0.379,
      "step": 11714
    },
    {
      "epoch": 0.7338041622950563,
      "grad_norm": 0.8687650902623548,
      "learning_rate": 1.7455341851180523e-06,
      "loss": 0.3914,
      "step": 11715
    },
    {
      "epoch": 0.7338668002943985,
      "grad_norm": 0.8657720616492501,
      "learning_rate": 1.744764151736057e-06,
      "loss": 0.3876,
      "step": 11716
    },
    {
      "epoch": 0.7339294382937409,
      "grad_norm": 0.8794410807791561,
      "learning_rate": 1.7439942523401748e-06,
      "loss": 0.3979,
      "step": 11717
    },
    {
      "epoch": 0.7339920762930832,
      "grad_norm": 0.8889105831696783,
      "learning_rate": 1.7432244869620935e-06,
      "loss": 0.4168,
      "step": 11718
    },
    {
      "epoch": 0.7340547142924255,
      "grad_norm": 0.8863785040222056,
      "learning_rate": 1.7424548556334969e-06,
      "loss": 0.4003,
      "step": 11719
    },
    {
      "epoch": 0.7341173522917678,
      "grad_norm": 0.9021765394421237,
      "learning_rate": 1.7416853583860632e-06,
      "loss": 0.4143,
      "step": 11720
    },
    {
      "epoch": 0.7341799902911101,
      "grad_norm": 0.9119264958827329,
      "learning_rate": 1.7409159952514648e-06,
      "loss": 0.3842,
      "step": 11721
    },
    {
      "epoch": 0.7342426282904524,
      "grad_norm": 0.7892037216886343,
      "learning_rate": 1.7401467662613686e-06,
      "loss": 0.3779,
      "step": 11722
    },
    {
      "epoch": 0.7343052662897948,
      "grad_norm": 0.862071203126577,
      "learning_rate": 1.7393776714474352e-06,
      "loss": 0.383,
      "step": 11723
    },
    {
      "epoch": 0.734367904289137,
      "grad_norm": 0.9399202076051036,
      "learning_rate": 1.7386087108413242e-06,
      "loss": 0.4448,
      "step": 11724
    },
    {
      "epoch": 0.7344305422884793,
      "grad_norm": 0.8434181528658055,
      "learning_rate": 1.7378398844746812e-06,
      "loss": 0.3832,
      "step": 11725
    },
    {
      "epoch": 0.7344931802878216,
      "grad_norm": 0.7929734706936268,
      "learning_rate": 1.7370711923791567e-06,
      "loss": 0.3612,
      "step": 11726
    },
    {
      "epoch": 0.7345558182871639,
      "grad_norm": 0.8515851927591271,
      "learning_rate": 1.736302634586387e-06,
      "loss": 0.4149,
      "step": 11727
    },
    {
      "epoch": 0.7346184562865062,
      "grad_norm": 0.8552023633405632,
      "learning_rate": 1.735534211128007e-06,
      "loss": 0.3951,
      "step": 11728
    },
    {
      "epoch": 0.7346810942858485,
      "grad_norm": 0.8845055041587205,
      "learning_rate": 1.7347659220356444e-06,
      "loss": 0.3827,
      "step": 11729
    },
    {
      "epoch": 0.7347437322851909,
      "grad_norm": 0.9040422478711307,
      "learning_rate": 1.733997767340923e-06,
      "loss": 0.3768,
      "step": 11730
    },
    {
      "epoch": 0.7348063702845331,
      "grad_norm": 0.8294674605161346,
      "learning_rate": 1.7332297470754594e-06,
      "loss": 0.3924,
      "step": 11731
    },
    {
      "epoch": 0.7348690082838754,
      "grad_norm": 0.7717650179707483,
      "learning_rate": 1.7324618612708637e-06,
      "loss": 0.3525,
      "step": 11732
    },
    {
      "epoch": 0.7349316462832177,
      "grad_norm": 0.8419905180335207,
      "learning_rate": 1.7316941099587464e-06,
      "loss": 0.4261,
      "step": 11733
    },
    {
      "epoch": 0.73499428428256,
      "grad_norm": 0.8930532496794235,
      "learning_rate": 1.7309264931707048e-06,
      "loss": 0.4012,
      "step": 11734
    },
    {
      "epoch": 0.7350569222819023,
      "grad_norm": 0.8205045317301592,
      "learning_rate": 1.7301590109383365e-06,
      "loss": 0.3614,
      "step": 11735
    },
    {
      "epoch": 0.7351195602812446,
      "grad_norm": 0.8053875916176979,
      "learning_rate": 1.7293916632932306e-06,
      "loss": 0.3846,
      "step": 11736
    },
    {
      "epoch": 0.735182198280587,
      "grad_norm": 0.8769834213196138,
      "learning_rate": 1.7286244502669708e-06,
      "loss": 0.3757,
      "step": 11737
    },
    {
      "epoch": 0.7352448362799292,
      "grad_norm": 0.8931177216703691,
      "learning_rate": 1.7278573718911363e-06,
      "loss": 0.3762,
      "step": 11738
    },
    {
      "epoch": 0.7353074742792716,
      "grad_norm": 0.8098157623368545,
      "learning_rate": 1.727090428197299e-06,
      "loss": 0.3616,
      "step": 11739
    },
    {
      "epoch": 0.7353701122786138,
      "grad_norm": 0.8799611655749997,
      "learning_rate": 1.7263236192170273e-06,
      "loss": 0.3674,
      "step": 11740
    },
    {
      "epoch": 0.7354327502779561,
      "grad_norm": 0.8684804663788455,
      "learning_rate": 1.7255569449818816e-06,
      "loss": 0.388,
      "step": 11741
    },
    {
      "epoch": 0.7354953882772984,
      "grad_norm": 0.8451072723357946,
      "learning_rate": 1.7247904055234215e-06,
      "loss": 0.4045,
      "step": 11742
    },
    {
      "epoch": 0.7355580262766407,
      "grad_norm": 0.8416221343490853,
      "learning_rate": 1.724024000873194e-06,
      "loss": 0.376,
      "step": 11743
    },
    {
      "epoch": 0.735620664275983,
      "grad_norm": 0.864396114461225,
      "learning_rate": 1.723257731062749e-06,
      "loss": 0.3522,
      "step": 11744
    },
    {
      "epoch": 0.7356833022753253,
      "grad_norm": 0.8248973014900216,
      "learning_rate": 1.7224915961236239e-06,
      "loss": 0.3896,
      "step": 11745
    },
    {
      "epoch": 0.7357459402746677,
      "grad_norm": 0.8027073212810744,
      "learning_rate": 1.721725596087353e-06,
      "loss": 0.3356,
      "step": 11746
    },
    {
      "epoch": 0.7358085782740099,
      "grad_norm": 0.8485643816620712,
      "learning_rate": 1.7209597309854648e-06,
      "loss": 0.381,
      "step": 11747
    },
    {
      "epoch": 0.7358712162733523,
      "grad_norm": 0.8679015425080536,
      "learning_rate": 1.7201940008494827e-06,
      "loss": 0.4223,
      "step": 11748
    },
    {
      "epoch": 0.7359338542726945,
      "grad_norm": 0.8777241291406083,
      "learning_rate": 1.7194284057109228e-06,
      "loss": 0.3906,
      "step": 11749
    },
    {
      "epoch": 0.7359964922720368,
      "grad_norm": 0.8761097818535972,
      "learning_rate": 1.7186629456013003e-06,
      "loss": 0.405,
      "step": 11750
    },
    {
      "epoch": 0.7360591302713791,
      "grad_norm": 0.7729760752499669,
      "learning_rate": 1.7178976205521197e-06,
      "loss": 0.3635,
      "step": 11751
    },
    {
      "epoch": 0.7361217682707214,
      "grad_norm": 0.8133615243324831,
      "learning_rate": 1.7171324305948828e-06,
      "loss": 0.4,
      "step": 11752
    },
    {
      "epoch": 0.7361844062700638,
      "grad_norm": 0.8693737951340367,
      "learning_rate": 1.716367375761082e-06,
      "loss": 0.3793,
      "step": 11753
    },
    {
      "epoch": 0.736247044269406,
      "grad_norm": 0.8585713206228857,
      "learning_rate": 1.7156024560822115e-06,
      "loss": 0.4285,
      "step": 11754
    },
    {
      "epoch": 0.7363096822687484,
      "grad_norm": 0.8965390245695601,
      "learning_rate": 1.7148376715897535e-06,
      "loss": 0.3526,
      "step": 11755
    },
    {
      "epoch": 0.7363723202680906,
      "grad_norm": 0.8513703516343105,
      "learning_rate": 1.7140730223151869e-06,
      "loss": 0.4167,
      "step": 11756
    },
    {
      "epoch": 0.736434958267433,
      "grad_norm": 0.8179547627816905,
      "learning_rate": 1.7133085082899842e-06,
      "loss": 0.3795,
      "step": 11757
    },
    {
      "epoch": 0.7364975962667752,
      "grad_norm": 0.7592482360781884,
      "learning_rate": 1.7125441295456118e-06,
      "loss": 0.3203,
      "step": 11758
    },
    {
      "epoch": 0.7365602342661175,
      "grad_norm": 0.9076140923092615,
      "learning_rate": 1.7117798861135354e-06,
      "loss": 0.4222,
      "step": 11759
    },
    {
      "epoch": 0.7366228722654599,
      "grad_norm": 0.6155504468468321,
      "learning_rate": 1.7110157780252084e-06,
      "loss": 0.4357,
      "step": 11760
    },
    {
      "epoch": 0.7366855102648021,
      "grad_norm": 0.8572861612901348,
      "learning_rate": 1.710251805312083e-06,
      "loss": 0.3768,
      "step": 11761
    },
    {
      "epoch": 0.7367481482641445,
      "grad_norm": 0.8371013477167776,
      "learning_rate": 1.709487968005602e-06,
      "loss": 0.3348,
      "step": 11762
    },
    {
      "epoch": 0.7368107862634867,
      "grad_norm": 0.8300476022897114,
      "learning_rate": 1.708724266137209e-06,
      "loss": 0.3636,
      "step": 11763
    },
    {
      "epoch": 0.7368734242628291,
      "grad_norm": 0.8960396183080886,
      "learning_rate": 1.7079606997383358e-06,
      "loss": 0.4093,
      "step": 11764
    },
    {
      "epoch": 0.7369360622621713,
      "grad_norm": 0.8462346275125134,
      "learning_rate": 1.7071972688404115e-06,
      "loss": 0.3882,
      "step": 11765
    },
    {
      "epoch": 0.7369987002615136,
      "grad_norm": 0.9558666923004508,
      "learning_rate": 1.706433973474858e-06,
      "loss": 0.3874,
      "step": 11766
    },
    {
      "epoch": 0.737061338260856,
      "grad_norm": 0.839487041089926,
      "learning_rate": 1.7056708136730925e-06,
      "loss": 0.3792,
      "step": 11767
    },
    {
      "epoch": 0.7371239762601982,
      "grad_norm": 0.8857246607012824,
      "learning_rate": 1.7049077894665294e-06,
      "loss": 0.3798,
      "step": 11768
    },
    {
      "epoch": 0.7371866142595406,
      "grad_norm": 0.878869143740285,
      "learning_rate": 1.7041449008865728e-06,
      "loss": 0.3947,
      "step": 11769
    },
    {
      "epoch": 0.7372492522588828,
      "grad_norm": 0.861859092469052,
      "learning_rate": 1.7033821479646246e-06,
      "loss": 0.3906,
      "step": 11770
    },
    {
      "epoch": 0.7373118902582252,
      "grad_norm": 0.6113748382375664,
      "learning_rate": 1.702619530732077e-06,
      "loss": 0.4827,
      "step": 11771
    },
    {
      "epoch": 0.7373745282575674,
      "grad_norm": 0.8714957763553534,
      "learning_rate": 1.701857049220323e-06,
      "loss": 0.406,
      "step": 11772
    },
    {
      "epoch": 0.7374371662569098,
      "grad_norm": 0.8626818847770524,
      "learning_rate": 1.7010947034607451e-06,
      "loss": 0.3943,
      "step": 11773
    },
    {
      "epoch": 0.7374998042562521,
      "grad_norm": 0.8453622054621688,
      "learning_rate": 1.7003324934847215e-06,
      "loss": 0.3584,
      "step": 11774
    },
    {
      "epoch": 0.7375624422555943,
      "grad_norm": 0.8012928159323418,
      "learning_rate": 1.6995704193236235e-06,
      "loss": 0.3672,
      "step": 11775
    },
    {
      "epoch": 0.7376250802549367,
      "grad_norm": 0.9595840006514634,
      "learning_rate": 1.6988084810088213e-06,
      "loss": 0.4161,
      "step": 11776
    },
    {
      "epoch": 0.7376877182542789,
      "grad_norm": 0.8783492453751516,
      "learning_rate": 1.6980466785716742e-06,
      "loss": 0.4029,
      "step": 11777
    },
    {
      "epoch": 0.7377503562536213,
      "grad_norm": 0.8947081034486641,
      "learning_rate": 1.697285012043539e-06,
      "loss": 0.4392,
      "step": 11778
    },
    {
      "epoch": 0.7378129942529635,
      "grad_norm": 0.8930420610346381,
      "learning_rate": 1.6965234814557663e-06,
      "loss": 0.3881,
      "step": 11779
    },
    {
      "epoch": 0.7378756322523059,
      "grad_norm": 0.6061776176051027,
      "learning_rate": 1.6957620868396979e-06,
      "loss": 0.4517,
      "step": 11780
    },
    {
      "epoch": 0.7379382702516482,
      "grad_norm": 0.8263485725757382,
      "learning_rate": 1.6950008282266771e-06,
      "loss": 0.4095,
      "step": 11781
    },
    {
      "epoch": 0.7380009082509905,
      "grad_norm": 0.9496994825950164,
      "learning_rate": 1.6942397056480358e-06,
      "loss": 0.4474,
      "step": 11782
    },
    {
      "epoch": 0.7380635462503328,
      "grad_norm": 0.8340295000729387,
      "learning_rate": 1.6934787191351015e-06,
      "loss": 0.4192,
      "step": 11783
    },
    {
      "epoch": 0.738126184249675,
      "grad_norm": 0.8436066664129119,
      "learning_rate": 1.6927178687191953e-06,
      "loss": 0.3877,
      "step": 11784
    },
    {
      "epoch": 0.7381888222490174,
      "grad_norm": 0.9104914306471628,
      "learning_rate": 1.6919571544316377e-06,
      "loss": 0.3488,
      "step": 11785
    },
    {
      "epoch": 0.7382514602483596,
      "grad_norm": 0.8107973448712952,
      "learning_rate": 1.691196576303737e-06,
      "loss": 0.3607,
      "step": 11786
    },
    {
      "epoch": 0.738314098247702,
      "grad_norm": 0.8346553759293002,
      "learning_rate": 1.6904361343667997e-06,
      "loss": 0.3568,
      "step": 11787
    },
    {
      "epoch": 0.7383767362470443,
      "grad_norm": 0.8060257869551308,
      "learning_rate": 1.6896758286521248e-06,
      "loss": 0.3559,
      "step": 11788
    },
    {
      "epoch": 0.7384393742463866,
      "grad_norm": 0.8409859766560969,
      "learning_rate": 1.6889156591910061e-06,
      "loss": 0.3864,
      "step": 11789
    },
    {
      "epoch": 0.7385020122457289,
      "grad_norm": 0.8098985940701998,
      "learning_rate": 1.6881556260147347e-06,
      "loss": 0.393,
      "step": 11790
    },
    {
      "epoch": 0.7385646502450711,
      "grad_norm": 0.8109268615100332,
      "learning_rate": 1.6873957291545928e-06,
      "loss": 0.3764,
      "step": 11791
    },
    {
      "epoch": 0.7386272882444135,
      "grad_norm": 0.922538746099421,
      "learning_rate": 1.686635968641856e-06,
      "loss": 0.3706,
      "step": 11792
    },
    {
      "epoch": 0.7386899262437557,
      "grad_norm": 0.8985525576337892,
      "learning_rate": 1.6858763445077992e-06,
      "loss": 0.3717,
      "step": 11793
    },
    {
      "epoch": 0.7387525642430981,
      "grad_norm": 0.6460328563791166,
      "learning_rate": 1.6851168567836867e-06,
      "loss": 0.4593,
      "step": 11794
    },
    {
      "epoch": 0.7388152022424403,
      "grad_norm": 0.6006884188404923,
      "learning_rate": 1.6843575055007805e-06,
      "loss": 0.4418,
      "step": 11795
    },
    {
      "epoch": 0.7388778402417827,
      "grad_norm": 0.8755079575441096,
      "learning_rate": 1.6835982906903342e-06,
      "loss": 0.3847,
      "step": 11796
    },
    {
      "epoch": 0.738940478241125,
      "grad_norm": 0.8345966245105763,
      "learning_rate": 1.6828392123835979e-06,
      "loss": 0.3744,
      "step": 11797
    },
    {
      "epoch": 0.7390031162404673,
      "grad_norm": 0.8366498269220212,
      "learning_rate": 1.6820802706118155e-06,
      "loss": 0.3578,
      "step": 11798
    },
    {
      "epoch": 0.7390657542398096,
      "grad_norm": 0.8996955095513518,
      "learning_rate": 1.6813214654062232e-06,
      "loss": 0.403,
      "step": 11799
    },
    {
      "epoch": 0.7391283922391518,
      "grad_norm": 0.8587470154256355,
      "learning_rate": 1.680562796798057e-06,
      "loss": 0.3874,
      "step": 11800
    },
    {
      "epoch": 0.7391910302384942,
      "grad_norm": 0.7959109564537614,
      "learning_rate": 1.6798042648185409e-06,
      "loss": 0.3553,
      "step": 11801
    },
    {
      "epoch": 0.7392536682378364,
      "grad_norm": 0.8223113408855621,
      "learning_rate": 1.6790458694988988e-06,
      "loss": 0.3682,
      "step": 11802
    },
    {
      "epoch": 0.7393163062371788,
      "grad_norm": 0.9509787651666294,
      "learning_rate": 1.678287610870345e-06,
      "loss": 0.3717,
      "step": 11803
    },
    {
      "epoch": 0.7393789442365211,
      "grad_norm": 0.8721756672262173,
      "learning_rate": 1.6775294889640898e-06,
      "loss": 0.3774,
      "step": 11804
    },
    {
      "epoch": 0.7394415822358634,
      "grad_norm": 0.6454657074380385,
      "learning_rate": 1.6767715038113375e-06,
      "loss": 0.4562,
      "step": 11805
    },
    {
      "epoch": 0.7395042202352057,
      "grad_norm": 0.9285399210081571,
      "learning_rate": 1.676013655443287e-06,
      "loss": 0.4156,
      "step": 11806
    },
    {
      "epoch": 0.739566858234548,
      "grad_norm": 0.8608798220708764,
      "learning_rate": 1.6752559438911314e-06,
      "loss": 0.4018,
      "step": 11807
    },
    {
      "epoch": 0.7396294962338903,
      "grad_norm": 0.8744188962894086,
      "learning_rate": 1.6744983691860566e-06,
      "loss": 0.3891,
      "step": 11808
    },
    {
      "epoch": 0.7396921342332325,
      "grad_norm": 0.8903510524343295,
      "learning_rate": 1.673740931359248e-06,
      "loss": 0.3774,
      "step": 11809
    },
    {
      "epoch": 0.7397547722325749,
      "grad_norm": 0.8616632494368693,
      "learning_rate": 1.6729836304418778e-06,
      "loss": 0.3945,
      "step": 11810
    },
    {
      "epoch": 0.7398174102319172,
      "grad_norm": 0.8544518635148393,
      "learning_rate": 1.6722264664651212e-06,
      "loss": 0.3868,
      "step": 11811
    },
    {
      "epoch": 0.7398800482312595,
      "grad_norm": 0.8055752112590663,
      "learning_rate": 1.6714694394601405e-06,
      "loss": 0.3968,
      "step": 11812
    },
    {
      "epoch": 0.7399426862306018,
      "grad_norm": 0.9153565674718221,
      "learning_rate": 1.6707125494580956e-06,
      "loss": 0.4123,
      "step": 11813
    },
    {
      "epoch": 0.7400053242299441,
      "grad_norm": 0.8157185119202053,
      "learning_rate": 1.6699557964901403e-06,
      "loss": 0.3729,
      "step": 11814
    },
    {
      "epoch": 0.7400679622292864,
      "grad_norm": 0.7549464589845437,
      "learning_rate": 1.6691991805874225e-06,
      "loss": 0.3565,
      "step": 11815
    },
    {
      "epoch": 0.7401306002286286,
      "grad_norm": 0.8948399717919115,
      "learning_rate": 1.6684427017810844e-06,
      "loss": 0.3961,
      "step": 11816
    },
    {
      "epoch": 0.740193238227971,
      "grad_norm": 0.9155217304056809,
      "learning_rate": 1.6676863601022614e-06,
      "loss": 0.381,
      "step": 11817
    },
    {
      "epoch": 0.7402558762273133,
      "grad_norm": 0.7978925473027714,
      "learning_rate": 1.6669301555820866e-06,
      "loss": 0.3731,
      "step": 11818
    },
    {
      "epoch": 0.7403185142266556,
      "grad_norm": 0.8663129324655787,
      "learning_rate": 1.666174088251687e-06,
      "loss": 0.4009,
      "step": 11819
    },
    {
      "epoch": 0.7403811522259979,
      "grad_norm": 0.8342167436771286,
      "learning_rate": 1.6654181581421802e-06,
      "loss": 0.3659,
      "step": 11820
    },
    {
      "epoch": 0.7404437902253402,
      "grad_norm": 0.855629195411428,
      "learning_rate": 1.664662365284681e-06,
      "loss": 0.3685,
      "step": 11821
    },
    {
      "epoch": 0.7405064282246825,
      "grad_norm": 0.8438324843729923,
      "learning_rate": 1.6639067097102985e-06,
      "loss": 0.3585,
      "step": 11822
    },
    {
      "epoch": 0.7405690662240249,
      "grad_norm": 0.8661566496273404,
      "learning_rate": 1.6631511914501342e-06,
      "loss": 0.3885,
      "step": 11823
    },
    {
      "epoch": 0.7406317042233671,
      "grad_norm": 0.9050592415635598,
      "learning_rate": 1.6623958105352867e-06,
      "loss": 0.4229,
      "step": 11824
    },
    {
      "epoch": 0.7406943422227094,
      "grad_norm": 0.9313964609033556,
      "learning_rate": 1.6616405669968471e-06,
      "loss": 0.4261,
      "step": 11825
    },
    {
      "epoch": 0.7407569802220517,
      "grad_norm": 0.8503859002606946,
      "learning_rate": 1.6608854608658998e-06,
      "loss": 0.3707,
      "step": 11826
    },
    {
      "epoch": 0.740819618221394,
      "grad_norm": 0.8860057752050012,
      "learning_rate": 1.6601304921735266e-06,
      "loss": 0.3811,
      "step": 11827
    },
    {
      "epoch": 0.7408822562207363,
      "grad_norm": 0.8174177332173598,
      "learning_rate": 1.6593756609508038e-06,
      "loss": 0.3674,
      "step": 11828
    },
    {
      "epoch": 0.7409448942200786,
      "grad_norm": 0.8589202215499073,
      "learning_rate": 1.6586209672287989e-06,
      "loss": 0.3749,
      "step": 11829
    },
    {
      "epoch": 0.741007532219421,
      "grad_norm": 0.8154020151273754,
      "learning_rate": 1.657866411038575e-06,
      "loss": 0.3727,
      "step": 11830
    },
    {
      "epoch": 0.7410701702187632,
      "grad_norm": 0.8586097472686849,
      "learning_rate": 1.6571119924111905e-06,
      "loss": 0.3991,
      "step": 11831
    },
    {
      "epoch": 0.7411328082181056,
      "grad_norm": 0.8713803970626045,
      "learning_rate": 1.6563577113776964e-06,
      "loss": 0.3837,
      "step": 11832
    },
    {
      "epoch": 0.7411954462174478,
      "grad_norm": 0.7491324190424864,
      "learning_rate": 1.6556035679691395e-06,
      "loss": 0.3376,
      "step": 11833
    },
    {
      "epoch": 0.7412580842167901,
      "grad_norm": 0.8281375843416549,
      "learning_rate": 1.6548495622165595e-06,
      "loss": 0.371,
      "step": 11834
    },
    {
      "epoch": 0.7413207222161324,
      "grad_norm": 0.8380128621161567,
      "learning_rate": 1.6540956941509934e-06,
      "loss": 0.3613,
      "step": 11835
    },
    {
      "epoch": 0.7413833602154747,
      "grad_norm": 0.8303978969493737,
      "learning_rate": 1.6533419638034681e-06,
      "loss": 0.3521,
      "step": 11836
    },
    {
      "epoch": 0.741445998214817,
      "grad_norm": 0.7934103807202587,
      "learning_rate": 1.65258837120501e-06,
      "loss": 0.3624,
      "step": 11837
    },
    {
      "epoch": 0.7415086362141593,
      "grad_norm": 0.8772461756918352,
      "learning_rate": 1.6518349163866365e-06,
      "loss": 0.4329,
      "step": 11838
    },
    {
      "epoch": 0.7415712742135017,
      "grad_norm": 0.9185585895800769,
      "learning_rate": 1.6510815993793588e-06,
      "loss": 0.3873,
      "step": 11839
    },
    {
      "epoch": 0.7416339122128439,
      "grad_norm": 0.8328112545495217,
      "learning_rate": 1.6503284202141844e-06,
      "loss": 0.3823,
      "step": 11840
    },
    {
      "epoch": 0.7416965502121862,
      "grad_norm": 0.8999956214448852,
      "learning_rate": 1.6495753789221136e-06,
      "loss": 0.4153,
      "step": 11841
    },
    {
      "epoch": 0.7417591882115285,
      "grad_norm": 0.8707698009148278,
      "learning_rate": 1.6488224755341425e-06,
      "loss": 0.363,
      "step": 11842
    },
    {
      "epoch": 0.7418218262108708,
      "grad_norm": 0.9513175400368599,
      "learning_rate": 1.6480697100812582e-06,
      "loss": 0.3953,
      "step": 11843
    },
    {
      "epoch": 0.7418844642102131,
      "grad_norm": 0.8552955380710765,
      "learning_rate": 1.6473170825944484e-06,
      "loss": 0.3861,
      "step": 11844
    },
    {
      "epoch": 0.7419471022095554,
      "grad_norm": 0.8579190117080489,
      "learning_rate": 1.6465645931046886e-06,
      "loss": 0.3903,
      "step": 11845
    },
    {
      "epoch": 0.7420097402088978,
      "grad_norm": 0.8807389857162559,
      "learning_rate": 1.645812241642954e-06,
      "loss": 0.3709,
      "step": 11846
    },
    {
      "epoch": 0.74207237820824,
      "grad_norm": 0.8447124427572337,
      "learning_rate": 1.64506002824021e-06,
      "loss": 0.377,
      "step": 11847
    },
    {
      "epoch": 0.7421350162075824,
      "grad_norm": 0.9016128035041909,
      "learning_rate": 1.644307952927418e-06,
      "loss": 0.386,
      "step": 11848
    },
    {
      "epoch": 0.7421976542069246,
      "grad_norm": 0.8335771504814392,
      "learning_rate": 1.643556015735534e-06,
      "loss": 0.3803,
      "step": 11849
    },
    {
      "epoch": 0.7422602922062669,
      "grad_norm": 0.8371833752708828,
      "learning_rate": 1.6428042166955066e-06,
      "loss": 0.3749,
      "step": 11850
    },
    {
      "epoch": 0.7423229302056092,
      "grad_norm": 0.8757755887933715,
      "learning_rate": 1.642052555838281e-06,
      "loss": 0.4017,
      "step": 11851
    },
    {
      "epoch": 0.7423855682049515,
      "grad_norm": 0.9147134564510182,
      "learning_rate": 1.6413010331947942e-06,
      "loss": 0.4031,
      "step": 11852
    },
    {
      "epoch": 0.7424482062042939,
      "grad_norm": 0.7778454165207621,
      "learning_rate": 1.6405496487959815e-06,
      "loss": 0.3456,
      "step": 11853
    },
    {
      "epoch": 0.7425108442036361,
      "grad_norm": 0.8729887265002357,
      "learning_rate": 1.6397984026727677e-06,
      "loss": 0.3767,
      "step": 11854
    },
    {
      "epoch": 0.7425734822029785,
      "grad_norm": 0.9001793216399397,
      "learning_rate": 1.639047294856076e-06,
      "loss": 0.3411,
      "step": 11855
    },
    {
      "epoch": 0.7426361202023207,
      "grad_norm": 1.1026505822681958,
      "learning_rate": 1.6382963253768225e-06,
      "loss": 0.4014,
      "step": 11856
    },
    {
      "epoch": 0.7426987582016631,
      "grad_norm": 0.8825129663154513,
      "learning_rate": 1.6375454942659153e-06,
      "loss": 0.3706,
      "step": 11857
    },
    {
      "epoch": 0.7427613962010053,
      "grad_norm": 0.8626373132304608,
      "learning_rate": 1.63679480155426e-06,
      "loss": 0.3954,
      "step": 11858
    },
    {
      "epoch": 0.7428240342003476,
      "grad_norm": 0.7761317522481476,
      "learning_rate": 1.636044247272755e-06,
      "loss": 0.3847,
      "step": 11859
    },
    {
      "epoch": 0.74288667219969,
      "grad_norm": 0.8452930870960725,
      "learning_rate": 1.635293831452292e-06,
      "loss": 0.3898,
      "step": 11860
    },
    {
      "epoch": 0.7429493101990322,
      "grad_norm": 0.814954677412457,
      "learning_rate": 1.6345435541237604e-06,
      "loss": 0.3867,
      "step": 11861
    },
    {
      "epoch": 0.7430119481983746,
      "grad_norm": 0.90526297557367,
      "learning_rate": 1.6337934153180406e-06,
      "loss": 0.3872,
      "step": 11862
    },
    {
      "epoch": 0.7430745861977168,
      "grad_norm": 0.7852354276068977,
      "learning_rate": 1.633043415066008e-06,
      "loss": 0.3724,
      "step": 11863
    },
    {
      "epoch": 0.7431372241970592,
      "grad_norm": 0.836996312279598,
      "learning_rate": 1.632293553398534e-06,
      "loss": 0.3573,
      "step": 11864
    },
    {
      "epoch": 0.7431998621964014,
      "grad_norm": 0.8256371179420641,
      "learning_rate": 1.6315438303464826e-06,
      "loss": 0.3904,
      "step": 11865
    },
    {
      "epoch": 0.7432625001957438,
      "grad_norm": 0.7737962636039803,
      "learning_rate": 1.6307942459407128e-06,
      "loss": 0.3381,
      "step": 11866
    },
    {
      "epoch": 0.743325138195086,
      "grad_norm": 0.8796370808828068,
      "learning_rate": 1.6300448002120767e-06,
      "loss": 0.3679,
      "step": 11867
    },
    {
      "epoch": 0.7433877761944283,
      "grad_norm": 0.8589009828177969,
      "learning_rate": 1.6292954931914223e-06,
      "loss": 0.4023,
      "step": 11868
    },
    {
      "epoch": 0.7434504141937707,
      "grad_norm": 0.8413680639096403,
      "learning_rate": 1.6285463249095896e-06,
      "loss": 0.383,
      "step": 11869
    },
    {
      "epoch": 0.7435130521931129,
      "grad_norm": 0.8081772302151818,
      "learning_rate": 1.6277972953974174e-06,
      "loss": 0.3675,
      "step": 11870
    },
    {
      "epoch": 0.7435756901924553,
      "grad_norm": 0.9259100592922929,
      "learning_rate": 1.6270484046857342e-06,
      "loss": 0.4415,
      "step": 11871
    },
    {
      "epoch": 0.7436383281917975,
      "grad_norm": 0.9046595136047892,
      "learning_rate": 1.6262996528053638e-06,
      "loss": 0.4188,
      "step": 11872
    },
    {
      "epoch": 0.7437009661911399,
      "grad_norm": 0.5979113494552318,
      "learning_rate": 1.6255510397871267e-06,
      "loss": 0.4481,
      "step": 11873
    },
    {
      "epoch": 0.7437636041904822,
      "grad_norm": 0.9042197393744067,
      "learning_rate": 1.6248025656618355e-06,
      "loss": 0.3963,
      "step": 11874
    },
    {
      "epoch": 0.7438262421898244,
      "grad_norm": 0.93483188662867,
      "learning_rate": 1.6240542304602968e-06,
      "loss": 0.4175,
      "step": 11875
    },
    {
      "epoch": 0.7438888801891668,
      "grad_norm": 0.9142103451192279,
      "learning_rate": 1.6233060342133132e-06,
      "loss": 0.4276,
      "step": 11876
    },
    {
      "epoch": 0.743951518188509,
      "grad_norm": 0.822849293872721,
      "learning_rate": 1.6225579769516781e-06,
      "loss": 0.4122,
      "step": 11877
    },
    {
      "epoch": 0.7440141561878514,
      "grad_norm": 0.8698794743099043,
      "learning_rate": 1.6218100587061853e-06,
      "loss": 0.4258,
      "step": 11878
    },
    {
      "epoch": 0.7440767941871936,
      "grad_norm": 0.7817149532748595,
      "learning_rate": 1.621062279507617e-06,
      "loss": 0.3791,
      "step": 11879
    },
    {
      "epoch": 0.744139432186536,
      "grad_norm": 0.909483219890683,
      "learning_rate": 1.620314639386753e-06,
      "loss": 0.4084,
      "step": 11880
    },
    {
      "epoch": 0.7442020701858783,
      "grad_norm": 0.8520694354542131,
      "learning_rate": 1.6195671383743656e-06,
      "loss": 0.3717,
      "step": 11881
    },
    {
      "epoch": 0.7442647081852206,
      "grad_norm": 0.9485319452211051,
      "learning_rate": 1.6188197765012203e-06,
      "loss": 0.4232,
      "step": 11882
    },
    {
      "epoch": 0.7443273461845629,
      "grad_norm": 0.8513189199463589,
      "learning_rate": 1.6180725537980824e-06,
      "loss": 0.3637,
      "step": 11883
    },
    {
      "epoch": 0.7443899841839051,
      "grad_norm": 0.8098699914662619,
      "learning_rate": 1.6173254702957058e-06,
      "loss": 0.3562,
      "step": 11884
    },
    {
      "epoch": 0.7444526221832475,
      "grad_norm": 0.8464637476623581,
      "learning_rate": 1.6165785260248407e-06,
      "loss": 0.3475,
      "step": 11885
    },
    {
      "epoch": 0.7445152601825897,
      "grad_norm": 0.7377754149579766,
      "learning_rate": 1.61583172101623e-06,
      "loss": 0.3769,
      "step": 11886
    },
    {
      "epoch": 0.7445778981819321,
      "grad_norm": 0.8619488771945731,
      "learning_rate": 1.6150850553006153e-06,
      "loss": 0.3593,
      "step": 11887
    },
    {
      "epoch": 0.7446405361812743,
      "grad_norm": 0.8793871537236594,
      "learning_rate": 1.6143385289087282e-06,
      "loss": 0.4452,
      "step": 11888
    },
    {
      "epoch": 0.7447031741806167,
      "grad_norm": 0.8818230387213096,
      "learning_rate": 1.6135921418712959e-06,
      "loss": 0.3896,
      "step": 11889
    },
    {
      "epoch": 0.744765812179959,
      "grad_norm": 0.823266602215866,
      "learning_rate": 1.612845894219039e-06,
      "loss": 0.383,
      "step": 11890
    },
    {
      "epoch": 0.7448284501793013,
      "grad_norm": 0.8508000692902633,
      "learning_rate": 1.612099785982672e-06,
      "loss": 0.3995,
      "step": 11891
    },
    {
      "epoch": 0.7448910881786436,
      "grad_norm": 0.6491359922407461,
      "learning_rate": 1.6113538171929088e-06,
      "loss": 0.4546,
      "step": 11892
    },
    {
      "epoch": 0.7449537261779858,
      "grad_norm": 0.8483934890691623,
      "learning_rate": 1.6106079878804515e-06,
      "loss": 0.3943,
      "step": 11893
    },
    {
      "epoch": 0.7450163641773282,
      "grad_norm": 0.8020489722952857,
      "learning_rate": 1.6098622980759982e-06,
      "loss": 0.3664,
      "step": 11894
    },
    {
      "epoch": 0.7450790021766704,
      "grad_norm": 0.9182462462652727,
      "learning_rate": 1.6091167478102404e-06,
      "loss": 0.4054,
      "step": 11895
    },
    {
      "epoch": 0.7451416401760128,
      "grad_norm": 0.7998969976275621,
      "learning_rate": 1.6083713371138681e-06,
      "loss": 0.3892,
      "step": 11896
    },
    {
      "epoch": 0.7452042781753551,
      "grad_norm": 0.6339988269446581,
      "learning_rate": 1.6076260660175613e-06,
      "loss": 0.4755,
      "step": 11897
    },
    {
      "epoch": 0.7452669161746974,
      "grad_norm": 0.8167324000223108,
      "learning_rate": 1.6068809345519954e-06,
      "loss": 0.3739,
      "step": 11898
    },
    {
      "epoch": 0.7453295541740397,
      "grad_norm": 0.8758623272457485,
      "learning_rate": 1.6061359427478396e-06,
      "loss": 0.4217,
      "step": 11899
    },
    {
      "epoch": 0.7453921921733819,
      "grad_norm": 0.875641350547336,
      "learning_rate": 1.6053910906357567e-06,
      "loss": 0.3998,
      "step": 11900
    },
    {
      "epoch": 0.7454548301727243,
      "grad_norm": 0.8262546852758041,
      "learning_rate": 1.6046463782464088e-06,
      "loss": 0.4046,
      "step": 11901
    },
    {
      "epoch": 0.7455174681720665,
      "grad_norm": 0.8552060657183658,
      "learning_rate": 1.6039018056104455e-06,
      "loss": 0.3753,
      "step": 11902
    },
    {
      "epoch": 0.7455801061714089,
      "grad_norm": 0.7877645230873452,
      "learning_rate": 1.603157372758513e-06,
      "loss": 0.3363,
      "step": 11903
    },
    {
      "epoch": 0.7456427441707512,
      "grad_norm": 0.8149858404974144,
      "learning_rate": 1.6024130797212551e-06,
      "loss": 0.3798,
      "step": 11904
    },
    {
      "epoch": 0.7457053821700935,
      "grad_norm": 0.8373782972441624,
      "learning_rate": 1.6016689265293056e-06,
      "loss": 0.396,
      "step": 11905
    },
    {
      "epoch": 0.7457680201694358,
      "grad_norm": 0.6492516250448825,
      "learning_rate": 1.6009249132132937e-06,
      "loss": 0.4578,
      "step": 11906
    },
    {
      "epoch": 0.7458306581687781,
      "grad_norm": 0.885191727095637,
      "learning_rate": 1.600181039803843e-06,
      "loss": 0.3384,
      "step": 11907
    },
    {
      "epoch": 0.7458932961681204,
      "grad_norm": 0.9180938211960823,
      "learning_rate": 1.5994373063315722e-06,
      "loss": 0.4148,
      "step": 11908
    },
    {
      "epoch": 0.7459559341674626,
      "grad_norm": 0.8250442724729949,
      "learning_rate": 1.5986937128270914e-06,
      "loss": 0.4194,
      "step": 11909
    },
    {
      "epoch": 0.746018572166805,
      "grad_norm": 0.9185782233348584,
      "learning_rate": 1.59795025932101e-06,
      "loss": 0.3725,
      "step": 11910
    },
    {
      "epoch": 0.7460812101661473,
      "grad_norm": 0.7821795665146143,
      "learning_rate": 1.5972069458439278e-06,
      "loss": 0.4073,
      "step": 11911
    },
    {
      "epoch": 0.7461438481654896,
      "grad_norm": 0.8606406127854005,
      "learning_rate": 1.596463772426437e-06,
      "loss": 0.3521,
      "step": 11912
    },
    {
      "epoch": 0.7462064861648319,
      "grad_norm": 0.8180239298324905,
      "learning_rate": 1.5957207390991308e-06,
      "loss": 0.3759,
      "step": 11913
    },
    {
      "epoch": 0.7462691241641742,
      "grad_norm": 0.9280067382055457,
      "learning_rate": 1.5949778458925912e-06,
      "loss": 0.4307,
      "step": 11914
    },
    {
      "epoch": 0.7463317621635165,
      "grad_norm": 0.8232469428191261,
      "learning_rate": 1.5942350928373956e-06,
      "loss": 0.3895,
      "step": 11915
    },
    {
      "epoch": 0.7463944001628589,
      "grad_norm": 0.8494001127210588,
      "learning_rate": 1.5934924799641156e-06,
      "loss": 0.3764,
      "step": 11916
    },
    {
      "epoch": 0.7464570381622011,
      "grad_norm": 0.8233464942588791,
      "learning_rate": 1.592750007303317e-06,
      "loss": 0.3748,
      "step": 11917
    },
    {
      "epoch": 0.7465196761615434,
      "grad_norm": 0.8493570296353078,
      "learning_rate": 1.5920076748855595e-06,
      "loss": 0.3878,
      "step": 11918
    },
    {
      "epoch": 0.7465823141608857,
      "grad_norm": 0.8961262182960409,
      "learning_rate": 1.5912654827413997e-06,
      "loss": 0.3835,
      "step": 11919
    },
    {
      "epoch": 0.746644952160228,
      "grad_norm": 0.8342343667216483,
      "learning_rate": 1.5905234309013857e-06,
      "loss": 0.3704,
      "step": 11920
    },
    {
      "epoch": 0.7467075901595703,
      "grad_norm": 0.836408151192671,
      "learning_rate": 1.589781519396058e-06,
      "loss": 0.3854,
      "step": 11921
    },
    {
      "epoch": 0.7467702281589126,
      "grad_norm": 0.8076876836010475,
      "learning_rate": 1.5890397482559582e-06,
      "loss": 0.389,
      "step": 11922
    },
    {
      "epoch": 0.746832866158255,
      "grad_norm": 0.8688345555762208,
      "learning_rate": 1.5882981175116153e-06,
      "loss": 0.4059,
      "step": 11923
    },
    {
      "epoch": 0.7468955041575972,
      "grad_norm": 0.8962333221819213,
      "learning_rate": 1.587556627193555e-06,
      "loss": 0.4167,
      "step": 11924
    },
    {
      "epoch": 0.7469581421569395,
      "grad_norm": 0.6718107623112778,
      "learning_rate": 1.586815277332297e-06,
      "loss": 0.4679,
      "step": 11925
    },
    {
      "epoch": 0.7470207801562818,
      "grad_norm": 0.841996753974758,
      "learning_rate": 1.5860740679583564e-06,
      "loss": 0.3707,
      "step": 11926
    },
    {
      "epoch": 0.7470834181556241,
      "grad_norm": 0.9142455019380605,
      "learning_rate": 1.5853329991022403e-06,
      "loss": 0.4218,
      "step": 11927
    },
    {
      "epoch": 0.7471460561549664,
      "grad_norm": 0.7105677436666755,
      "learning_rate": 1.58459207079445e-06,
      "loss": 0.2911,
      "step": 11928
    },
    {
      "epoch": 0.7472086941543087,
      "grad_norm": 0.8507402184743743,
      "learning_rate": 1.5838512830654845e-06,
      "loss": 0.3796,
      "step": 11929
    },
    {
      "epoch": 0.747271332153651,
      "grad_norm": 0.7867281045043849,
      "learning_rate": 1.5831106359458354e-06,
      "loss": 0.3608,
      "step": 11930
    },
    {
      "epoch": 0.7473339701529933,
      "grad_norm": 0.8799705391134328,
      "learning_rate": 1.5823701294659872e-06,
      "loss": 0.4116,
      "step": 11931
    },
    {
      "epoch": 0.7473966081523357,
      "grad_norm": 0.8962421779456973,
      "learning_rate": 1.5816297636564187e-06,
      "loss": 0.417,
      "step": 11932
    },
    {
      "epoch": 0.7474592461516779,
      "grad_norm": 0.8698199846131511,
      "learning_rate": 1.5808895385476031e-06,
      "loss": 0.3847,
      "step": 11933
    },
    {
      "epoch": 0.7475218841510202,
      "grad_norm": 0.8530753045810614,
      "learning_rate": 1.580149454170009e-06,
      "loss": 0.3755,
      "step": 11934
    },
    {
      "epoch": 0.7475845221503625,
      "grad_norm": 0.8146859065457619,
      "learning_rate": 1.5794095105540985e-06,
      "loss": 0.3579,
      "step": 11935
    },
    {
      "epoch": 0.7476471601497048,
      "grad_norm": 0.588894589288442,
      "learning_rate": 1.578669707730327e-06,
      "loss": 0.439,
      "step": 11936
    },
    {
      "epoch": 0.7477097981490471,
      "grad_norm": 0.9079302217927938,
      "learning_rate": 1.5779300457291442e-06,
      "loss": 0.3863,
      "step": 11937
    },
    {
      "epoch": 0.7477724361483894,
      "grad_norm": 0.9000014853025456,
      "learning_rate": 1.5771905245809954e-06,
      "loss": 0.3816,
      "step": 11938
    },
    {
      "epoch": 0.7478350741477318,
      "grad_norm": 0.8333665745594241,
      "learning_rate": 1.5764511443163216e-06,
      "loss": 0.3835,
      "step": 11939
    },
    {
      "epoch": 0.747897712147074,
      "grad_norm": 0.9100275271301139,
      "learning_rate": 1.5757119049655544e-06,
      "loss": 0.4076,
      "step": 11940
    },
    {
      "epoch": 0.7479603501464164,
      "grad_norm": 0.8404770674447912,
      "learning_rate": 1.5749728065591207e-06,
      "loss": 0.3906,
      "step": 11941
    },
    {
      "epoch": 0.7480229881457586,
      "grad_norm": 0.8571678464949887,
      "learning_rate": 1.5742338491274418e-06,
      "loss": 0.394,
      "step": 11942
    },
    {
      "epoch": 0.7480856261451009,
      "grad_norm": 0.9104724858576014,
      "learning_rate": 1.5734950327009336e-06,
      "loss": 0.401,
      "step": 11943
    },
    {
      "epoch": 0.7481482641444432,
      "grad_norm": 0.8407428282688308,
      "learning_rate": 1.5727563573100058e-06,
      "loss": 0.3635,
      "step": 11944
    },
    {
      "epoch": 0.7482109021437855,
      "grad_norm": 0.8015801455113569,
      "learning_rate": 1.5720178229850608e-06,
      "loss": 0.3499,
      "step": 11945
    },
    {
      "epoch": 0.7482735401431279,
      "grad_norm": 0.5999472346039169,
      "learning_rate": 1.5712794297565003e-06,
      "loss": 0.463,
      "step": 11946
    },
    {
      "epoch": 0.7483361781424701,
      "grad_norm": 0.8488274326514242,
      "learning_rate": 1.570541177654713e-06,
      "loss": 0.3951,
      "step": 11947
    },
    {
      "epoch": 0.7483988161418125,
      "grad_norm": 0.7824717886174579,
      "learning_rate": 1.5698030667100894e-06,
      "loss": 0.3876,
      "step": 11948
    },
    {
      "epoch": 0.7484614541411547,
      "grad_norm": 0.8538961805603468,
      "learning_rate": 1.569065096953008e-06,
      "loss": 0.4132,
      "step": 11949
    },
    {
      "epoch": 0.748524092140497,
      "grad_norm": 0.8857909659072213,
      "learning_rate": 1.5683272684138439e-06,
      "loss": 0.3733,
      "step": 11950
    },
    {
      "epoch": 0.7485867301398393,
      "grad_norm": 0.8812493843687469,
      "learning_rate": 1.5675895811229664e-06,
      "loss": 0.3726,
      "step": 11951
    },
    {
      "epoch": 0.7486493681391816,
      "grad_norm": 0.8303935681873068,
      "learning_rate": 1.5668520351107391e-06,
      "loss": 0.409,
      "step": 11952
    },
    {
      "epoch": 0.748712006138524,
      "grad_norm": 0.8604884561967024,
      "learning_rate": 1.5661146304075193e-06,
      "loss": 0.4333,
      "step": 11953
    },
    {
      "epoch": 0.7487746441378662,
      "grad_norm": 0.9401877892710939,
      "learning_rate": 1.5653773670436567e-06,
      "loss": 0.3862,
      "step": 11954
    },
    {
      "epoch": 0.7488372821372086,
      "grad_norm": 0.8575874688932112,
      "learning_rate": 1.5646402450494991e-06,
      "loss": 0.3843,
      "step": 11955
    },
    {
      "epoch": 0.7488999201365508,
      "grad_norm": 0.8390913791869685,
      "learning_rate": 1.5639032644553886e-06,
      "loss": 0.3885,
      "step": 11956
    },
    {
      "epoch": 0.7489625581358932,
      "grad_norm": 0.8021998301199078,
      "learning_rate": 1.5631664252916574e-06,
      "loss": 0.3875,
      "step": 11957
    },
    {
      "epoch": 0.7490251961352354,
      "grad_norm": 0.6643164170151098,
      "learning_rate": 1.5624297275886341e-06,
      "loss": 0.4293,
      "step": 11958
    },
    {
      "epoch": 0.7490878341345777,
      "grad_norm": 0.870196420178917,
      "learning_rate": 1.561693171376641e-06,
      "loss": 0.3709,
      "step": 11959
    },
    {
      "epoch": 0.74915047213392,
      "grad_norm": 0.8479236465329602,
      "learning_rate": 1.560956756685995e-06,
      "loss": 0.4012,
      "step": 11960
    },
    {
      "epoch": 0.7492131101332623,
      "grad_norm": 0.885587748282231,
      "learning_rate": 1.5602204835470075e-06,
      "loss": 0.3893,
      "step": 11961
    },
    {
      "epoch": 0.7492757481326047,
      "grad_norm": 0.9195862575620303,
      "learning_rate": 1.5594843519899822e-06,
      "loss": 0.3898,
      "step": 11962
    },
    {
      "epoch": 0.7493383861319469,
      "grad_norm": 0.9261706819854788,
      "learning_rate": 1.5587483620452204e-06,
      "loss": 0.3719,
      "step": 11963
    },
    {
      "epoch": 0.7494010241312893,
      "grad_norm": 0.7765288124994855,
      "learning_rate": 1.5580125137430136e-06,
      "loss": 0.3668,
      "step": 11964
    },
    {
      "epoch": 0.7494636621306315,
      "grad_norm": 1.0009474452278564,
      "learning_rate": 1.5572768071136519e-06,
      "loss": 0.3951,
      "step": 11965
    },
    {
      "epoch": 0.7495263001299739,
      "grad_norm": 0.8603722519775233,
      "learning_rate": 1.556541242187416e-06,
      "loss": 0.361,
      "step": 11966
    },
    {
      "epoch": 0.7495889381293162,
      "grad_norm": 0.8268644822439459,
      "learning_rate": 1.5558058189945818e-06,
      "loss": 0.358,
      "step": 11967
    },
    {
      "epoch": 0.7496515761286584,
      "grad_norm": 0.8817672739278148,
      "learning_rate": 1.5550705375654196e-06,
      "loss": 0.4261,
      "step": 11968
    },
    {
      "epoch": 0.7497142141280008,
      "grad_norm": 0.7689901967077553,
      "learning_rate": 1.5543353979301928e-06,
      "loss": 0.369,
      "step": 11969
    },
    {
      "epoch": 0.749776852127343,
      "grad_norm": 0.8233937523660382,
      "learning_rate": 1.5536004001191612e-06,
      "loss": 0.3942,
      "step": 11970
    },
    {
      "epoch": 0.7498394901266854,
      "grad_norm": 0.8858165351981097,
      "learning_rate": 1.5528655441625746e-06,
      "loss": 0.3675,
      "step": 11971
    },
    {
      "epoch": 0.7499021281260276,
      "grad_norm": 0.6066396927793752,
      "learning_rate": 1.552130830090684e-06,
      "loss": 0.4611,
      "step": 11972
    },
    {
      "epoch": 0.74996476612537,
      "grad_norm": 0.8704888308346053,
      "learning_rate": 1.5513962579337282e-06,
      "loss": 0.3717,
      "step": 11973
    },
    {
      "epoch": 0.7500274041247122,
      "grad_norm": 0.7992132261822809,
      "learning_rate": 1.550661827721941e-06,
      "loss": 0.3777,
      "step": 11974
    },
    {
      "epoch": 0.7500900421240546,
      "grad_norm": 0.8263031706815099,
      "learning_rate": 1.549927539485555e-06,
      "loss": 0.3695,
      "step": 11975
    },
    {
      "epoch": 0.7501526801233969,
      "grad_norm": 0.8586928902613998,
      "learning_rate": 1.549193393254792e-06,
      "loss": 0.3522,
      "step": 11976
    },
    {
      "epoch": 0.7502153181227391,
      "grad_norm": 0.8548493109313451,
      "learning_rate": 1.5484593890598688e-06,
      "loss": 0.3783,
      "step": 11977
    },
    {
      "epoch": 0.7502779561220815,
      "grad_norm": 0.8523678329575161,
      "learning_rate": 1.5477255269309981e-06,
      "loss": 0.406,
      "step": 11978
    },
    {
      "epoch": 0.7503405941214237,
      "grad_norm": 0.808569482304882,
      "learning_rate": 1.5469918068983857e-06,
      "loss": 0.3322,
      "step": 11979
    },
    {
      "epoch": 0.7504032321207661,
      "grad_norm": 0.8491672799350333,
      "learning_rate": 1.5462582289922296e-06,
      "loss": 0.4096,
      "step": 11980
    },
    {
      "epoch": 0.7504658701201083,
      "grad_norm": 0.7971833584303613,
      "learning_rate": 1.5455247932427276e-06,
      "loss": 0.3509,
      "step": 11981
    },
    {
      "epoch": 0.7505285081194507,
      "grad_norm": 0.8367430127230451,
      "learning_rate": 1.5447914996800656e-06,
      "loss": 0.337,
      "step": 11982
    },
    {
      "epoch": 0.750591146118793,
      "grad_norm": 0.8563949224477432,
      "learning_rate": 1.5440583483344262e-06,
      "loss": 0.4055,
      "step": 11983
    },
    {
      "epoch": 0.7506537841181352,
      "grad_norm": 0.8491785452346478,
      "learning_rate": 1.5433253392359876e-06,
      "loss": 0.401,
      "step": 11984
    },
    {
      "epoch": 0.7507164221174776,
      "grad_norm": 0.8144619223160948,
      "learning_rate": 1.54259247241492e-06,
      "loss": 0.3905,
      "step": 11985
    },
    {
      "epoch": 0.7507790601168198,
      "grad_norm": 0.8641185109118955,
      "learning_rate": 1.5418597479013875e-06,
      "loss": 0.3715,
      "step": 11986
    },
    {
      "epoch": 0.7508416981161622,
      "grad_norm": 0.8968295887984475,
      "learning_rate": 1.54112716572555e-06,
      "loss": 0.4065,
      "step": 11987
    },
    {
      "epoch": 0.7509043361155044,
      "grad_norm": 0.5407236994527924,
      "learning_rate": 1.5403947259175583e-06,
      "loss": 0.4461,
      "step": 11988
    },
    {
      "epoch": 0.7509669741148468,
      "grad_norm": 0.8263792241123655,
      "learning_rate": 1.5396624285075639e-06,
      "loss": 0.3766,
      "step": 11989
    },
    {
      "epoch": 0.7510296121141891,
      "grad_norm": 0.8298491755745055,
      "learning_rate": 1.5389302735257061e-06,
      "loss": 0.3258,
      "step": 11990
    },
    {
      "epoch": 0.7510922501135314,
      "grad_norm": 0.8854934861887589,
      "learning_rate": 1.5381982610021206e-06,
      "loss": 0.3935,
      "step": 11991
    },
    {
      "epoch": 0.7511548881128737,
      "grad_norm": 0.889846768425515,
      "learning_rate": 1.5374663909669357e-06,
      "loss": 0.3971,
      "step": 11992
    },
    {
      "epoch": 0.7512175261122159,
      "grad_norm": 0.9406427409206626,
      "learning_rate": 1.5367346634502784e-06,
      "loss": 0.4021,
      "step": 11993
    },
    {
      "epoch": 0.7512801641115583,
      "grad_norm": 0.898920003606628,
      "learning_rate": 1.536003078482265e-06,
      "loss": 0.3792,
      "step": 11994
    },
    {
      "epoch": 0.7513428021109005,
      "grad_norm": 0.8818439404352337,
      "learning_rate": 1.5352716360930082e-06,
      "loss": 0.3882,
      "step": 11995
    },
    {
      "epoch": 0.7514054401102429,
      "grad_norm": 0.9551594206118696,
      "learning_rate": 1.5345403363126137e-06,
      "loss": 0.4314,
      "step": 11996
    },
    {
      "epoch": 0.7514680781095852,
      "grad_norm": 0.8986252417932213,
      "learning_rate": 1.5338091791711807e-06,
      "loss": 0.3826,
      "step": 11997
    },
    {
      "epoch": 0.7515307161089275,
      "grad_norm": 0.8354050909664956,
      "learning_rate": 1.5330781646988068e-06,
      "loss": 0.3674,
      "step": 11998
    },
    {
      "epoch": 0.7515933541082698,
      "grad_norm": 0.8437892890199851,
      "learning_rate": 1.532347292925579e-06,
      "loss": 0.3948,
      "step": 11999
    },
    {
      "epoch": 0.7516559921076121,
      "grad_norm": 0.887651998178469,
      "learning_rate": 1.53161656388158e-06,
      "loss": 0.4104,
      "step": 12000
    },
    {
      "epoch": 0.7517186301069544,
      "grad_norm": 0.8937004304914231,
      "learning_rate": 1.530885977596886e-06,
      "loss": 0.4185,
      "step": 12001
    },
    {
      "epoch": 0.7517812681062966,
      "grad_norm": 0.8356557978887268,
      "learning_rate": 1.53015553410157e-06,
      "loss": 0.3716,
      "step": 12002
    },
    {
      "epoch": 0.751843906105639,
      "grad_norm": 0.8719938508386742,
      "learning_rate": 1.529425233425697e-06,
      "loss": 0.3894,
      "step": 12003
    },
    {
      "epoch": 0.7519065441049813,
      "grad_norm": 0.8478615647617207,
      "learning_rate": 1.5286950755993253e-06,
      "loss": 0.3931,
      "step": 12004
    },
    {
      "epoch": 0.7519691821043236,
      "grad_norm": 0.9102791357400324,
      "learning_rate": 1.5279650606525081e-06,
      "loss": 0.4166,
      "step": 12005
    },
    {
      "epoch": 0.7520318201036659,
      "grad_norm": 0.882587472441255,
      "learning_rate": 1.5272351886152925e-06,
      "loss": 0.3677,
      "step": 12006
    },
    {
      "epoch": 0.7520944581030082,
      "grad_norm": 0.7897609497778899,
      "learning_rate": 1.526505459517722e-06,
      "loss": 0.3459,
      "step": 12007
    },
    {
      "epoch": 0.7521570961023505,
      "grad_norm": 0.9021480915944998,
      "learning_rate": 1.525775873389832e-06,
      "loss": 0.3621,
      "step": 12008
    },
    {
      "epoch": 0.7522197341016927,
      "grad_norm": 0.8866114496363339,
      "learning_rate": 1.525046430261652e-06,
      "loss": 0.3603,
      "step": 12009
    },
    {
      "epoch": 0.7522823721010351,
      "grad_norm": 0.6963022737378745,
      "learning_rate": 1.5243171301632042e-06,
      "loss": 0.4492,
      "step": 12010
    },
    {
      "epoch": 0.7523450101003774,
      "grad_norm": 0.8872131396940486,
      "learning_rate": 1.52358797312451e-06,
      "loss": 0.3938,
      "step": 12011
    },
    {
      "epoch": 0.7524076480997197,
      "grad_norm": 0.8283124681072626,
      "learning_rate": 1.52285895917558e-06,
      "loss": 0.38,
      "step": 12012
    },
    {
      "epoch": 0.752470286099062,
      "grad_norm": 0.8320077962929506,
      "learning_rate": 1.5221300883464207e-06,
      "loss": 0.3673,
      "step": 12013
    },
    {
      "epoch": 0.7525329240984043,
      "grad_norm": 0.8728298781511284,
      "learning_rate": 1.5214013606670308e-06,
      "loss": 0.4094,
      "step": 12014
    },
    {
      "epoch": 0.7525955620977466,
      "grad_norm": 0.8845317647705367,
      "learning_rate": 1.5206727761674085e-06,
      "loss": 0.3966,
      "step": 12015
    },
    {
      "epoch": 0.752658200097089,
      "grad_norm": 0.9663469692414564,
      "learning_rate": 1.51994433487754e-06,
      "loss": 0.3993,
      "step": 12016
    },
    {
      "epoch": 0.7527208380964312,
      "grad_norm": 0.9490561453142573,
      "learning_rate": 1.5192160368274088e-06,
      "loss": 0.3868,
      "step": 12017
    },
    {
      "epoch": 0.7527834760957735,
      "grad_norm": 0.8081491161280109,
      "learning_rate": 1.5184878820469917e-06,
      "loss": 0.3618,
      "step": 12018
    },
    {
      "epoch": 0.7528461140951158,
      "grad_norm": 0.8049908500896578,
      "learning_rate": 1.5177598705662593e-06,
      "loss": 0.3657,
      "step": 12019
    },
    {
      "epoch": 0.7529087520944581,
      "grad_norm": 0.7696115867496006,
      "learning_rate": 1.5170320024151758e-06,
      "loss": 0.2997,
      "step": 12020
    },
    {
      "epoch": 0.7529713900938004,
      "grad_norm": 0.8513884436413107,
      "learning_rate": 1.5163042776237031e-06,
      "loss": 0.3654,
      "step": 12021
    },
    {
      "epoch": 0.7530340280931427,
      "grad_norm": 0.8606073857279313,
      "learning_rate": 1.5155766962217927e-06,
      "loss": 0.4103,
      "step": 12022
    },
    {
      "epoch": 0.753096666092485,
      "grad_norm": 0.8340292149849713,
      "learning_rate": 1.5148492582393904e-06,
      "loss": 0.3663,
      "step": 12023
    },
    {
      "epoch": 0.7531593040918273,
      "grad_norm": 0.8334613612007321,
      "learning_rate": 1.5141219637064413e-06,
      "loss": 0.3486,
      "step": 12024
    },
    {
      "epoch": 0.7532219420911697,
      "grad_norm": 0.6307241710339819,
      "learning_rate": 1.5133948126528797e-06,
      "loss": 0.4648,
      "step": 12025
    },
    {
      "epoch": 0.7532845800905119,
      "grad_norm": 0.9541416966637157,
      "learning_rate": 1.512667805108634e-06,
      "loss": 0.4088,
      "step": 12026
    },
    {
      "epoch": 0.7533472180898542,
      "grad_norm": 0.8423412597779133,
      "learning_rate": 1.511940941103629e-06,
      "loss": 0.3912,
      "step": 12027
    },
    {
      "epoch": 0.7534098560891965,
      "grad_norm": 0.8747448159343094,
      "learning_rate": 1.5112142206677827e-06,
      "loss": 0.3969,
      "step": 12028
    },
    {
      "epoch": 0.7534724940885388,
      "grad_norm": 0.7693822848165849,
      "learning_rate": 1.5104876438310045e-06,
      "loss": 0.3603,
      "step": 12029
    },
    {
      "epoch": 0.7535351320878811,
      "grad_norm": 0.9333795522570462,
      "learning_rate": 1.509761210623204e-06,
      "loss": 0.4269,
      "step": 12030
    },
    {
      "epoch": 0.7535977700872234,
      "grad_norm": 0.8405048802783457,
      "learning_rate": 1.5090349210742806e-06,
      "loss": 0.3669,
      "step": 12031
    },
    {
      "epoch": 0.7536604080865658,
      "grad_norm": 0.8810852948344641,
      "learning_rate": 1.5083087752141257e-06,
      "loss": 0.3846,
      "step": 12032
    },
    {
      "epoch": 0.753723046085908,
      "grad_norm": 0.8631990981701155,
      "learning_rate": 1.5075827730726312e-06,
      "loss": 0.3743,
      "step": 12033
    },
    {
      "epoch": 0.7537856840852503,
      "grad_norm": 0.8939501503273081,
      "learning_rate": 1.5068569146796786e-06,
      "loss": 0.4198,
      "step": 12034
    },
    {
      "epoch": 0.7538483220845926,
      "grad_norm": 0.9062960752381407,
      "learning_rate": 1.506131200065143e-06,
      "loss": 0.4364,
      "step": 12035
    },
    {
      "epoch": 0.7539109600839349,
      "grad_norm": 0.8302509525518216,
      "learning_rate": 1.5054056292588965e-06,
      "loss": 0.3824,
      "step": 12036
    },
    {
      "epoch": 0.7539735980832772,
      "grad_norm": 0.6444690179366316,
      "learning_rate": 1.504680202290802e-06,
      "loss": 0.4401,
      "step": 12037
    },
    {
      "epoch": 0.7540362360826195,
      "grad_norm": 0.8288550252239916,
      "learning_rate": 1.5039549191907183e-06,
      "loss": 0.4137,
      "step": 12038
    },
    {
      "epoch": 0.7540988740819619,
      "grad_norm": 0.887074236445502,
      "learning_rate": 1.5032297799885004e-06,
      "loss": 0.4127,
      "step": 12039
    },
    {
      "epoch": 0.7541615120813041,
      "grad_norm": 0.8741853461113877,
      "learning_rate": 1.5025047847139918e-06,
      "loss": 0.3864,
      "step": 12040
    },
    {
      "epoch": 0.7542241500806465,
      "grad_norm": 1.0003584149816809,
      "learning_rate": 1.5017799333970373e-06,
      "loss": 0.3809,
      "step": 12041
    },
    {
      "epoch": 0.7542867880799887,
      "grad_norm": 0.9075992258664127,
      "learning_rate": 1.5010552260674705e-06,
      "loss": 0.3754,
      "step": 12042
    },
    {
      "epoch": 0.754349426079331,
      "grad_norm": 0.822777704233257,
      "learning_rate": 1.5003306627551196e-06,
      "loss": 0.3891,
      "step": 12043
    },
    {
      "epoch": 0.7544120640786733,
      "grad_norm": 0.7804710526194045,
      "learning_rate": 1.499606243489808e-06,
      "loss": 0.423,
      "step": 12044
    },
    {
      "epoch": 0.7544747020780156,
      "grad_norm": 0.7914756448036254,
      "learning_rate": 1.4988819683013533e-06,
      "loss": 0.3731,
      "step": 12045
    },
    {
      "epoch": 0.754537340077358,
      "grad_norm": 0.8919338372953055,
      "learning_rate": 1.4981578372195666e-06,
      "loss": 0.3563,
      "step": 12046
    },
    {
      "epoch": 0.7545999780767002,
      "grad_norm": 0.8859717989464533,
      "learning_rate": 1.497433850274252e-06,
      "loss": 0.4231,
      "step": 12047
    },
    {
      "epoch": 0.7546626160760426,
      "grad_norm": 0.8315980553782404,
      "learning_rate": 1.4967100074952112e-06,
      "loss": 0.3744,
      "step": 12048
    },
    {
      "epoch": 0.7547252540753848,
      "grad_norm": 0.8654507408627964,
      "learning_rate": 1.4959863089122357e-06,
      "loss": 0.4119,
      "step": 12049
    },
    {
      "epoch": 0.7547878920747272,
      "grad_norm": 0.8754364771840297,
      "learning_rate": 1.4952627545551156e-06,
      "loss": 0.4102,
      "step": 12050
    },
    {
      "epoch": 0.7548505300740694,
      "grad_norm": 0.852395707280758,
      "learning_rate": 1.494539344453631e-06,
      "loss": 0.385,
      "step": 12051
    },
    {
      "epoch": 0.7549131680734117,
      "grad_norm": 0.8622502522919606,
      "learning_rate": 1.4938160786375571e-06,
      "loss": 0.4335,
      "step": 12052
    },
    {
      "epoch": 0.754975806072754,
      "grad_norm": 0.8363698520867168,
      "learning_rate": 1.4930929571366649e-06,
      "loss": 0.3592,
      "step": 12053
    },
    {
      "epoch": 0.7550384440720963,
      "grad_norm": 0.8255128263141399,
      "learning_rate": 1.4923699799807173e-06,
      "loss": 0.3555,
      "step": 12054
    },
    {
      "epoch": 0.7551010820714387,
      "grad_norm": 0.8847000275755614,
      "learning_rate": 1.491647147199472e-06,
      "loss": 0.3839,
      "step": 12055
    },
    {
      "epoch": 0.7551637200707809,
      "grad_norm": 0.8782690476810847,
      "learning_rate": 1.49092445882268e-06,
      "loss": 0.3851,
      "step": 12056
    },
    {
      "epoch": 0.7552263580701233,
      "grad_norm": 0.8114538366080732,
      "learning_rate": 1.4902019148800906e-06,
      "loss": 0.3434,
      "step": 12057
    },
    {
      "epoch": 0.7552889960694655,
      "grad_norm": 0.8908265666607006,
      "learning_rate": 1.4894795154014396e-06,
      "loss": 0.4367,
      "step": 12058
    },
    {
      "epoch": 0.7553516340688078,
      "grad_norm": 0.9022545634382629,
      "learning_rate": 1.4887572604164651e-06,
      "loss": 0.3218,
      "step": 12059
    },
    {
      "epoch": 0.7554142720681501,
      "grad_norm": 0.8082514697085837,
      "learning_rate": 1.488035149954894e-06,
      "loss": 0.3735,
      "step": 12060
    },
    {
      "epoch": 0.7554769100674924,
      "grad_norm": 0.8190592068141946,
      "learning_rate": 1.4873131840464472e-06,
      "loss": 0.3449,
      "step": 12061
    },
    {
      "epoch": 0.7555395480668348,
      "grad_norm": 0.8412084901721537,
      "learning_rate": 1.4865913627208423e-06,
      "loss": 0.4023,
      "step": 12062
    },
    {
      "epoch": 0.755602186066177,
      "grad_norm": 0.8508167281017351,
      "learning_rate": 1.4858696860077887e-06,
      "loss": 0.3708,
      "step": 12063
    },
    {
      "epoch": 0.7556648240655194,
      "grad_norm": 0.8758496322682837,
      "learning_rate": 1.485148153936991e-06,
      "loss": 0.3916,
      "step": 12064
    },
    {
      "epoch": 0.7557274620648616,
      "grad_norm": 0.8789108499442083,
      "learning_rate": 1.4844267665381461e-06,
      "loss": 0.4069,
      "step": 12065
    },
    {
      "epoch": 0.755790100064204,
      "grad_norm": 0.889033140297008,
      "learning_rate": 1.4837055238409482e-06,
      "loss": 0.3924,
      "step": 12066
    },
    {
      "epoch": 0.7558527380635462,
      "grad_norm": 0.869399837784918,
      "learning_rate": 1.4829844258750854e-06,
      "loss": 0.4203,
      "step": 12067
    },
    {
      "epoch": 0.7559153760628885,
      "grad_norm": 0.8178829653922268,
      "learning_rate": 1.4822634726702362e-06,
      "loss": 0.3932,
      "step": 12068
    },
    {
      "epoch": 0.7559780140622309,
      "grad_norm": 0.8326968119522478,
      "learning_rate": 1.4815426642560753e-06,
      "loss": 0.3462,
      "step": 12069
    },
    {
      "epoch": 0.7560406520615731,
      "grad_norm": 0.868141065309054,
      "learning_rate": 1.4808220006622714e-06,
      "loss": 0.4209,
      "step": 12070
    },
    {
      "epoch": 0.7561032900609155,
      "grad_norm": 0.5779643001989403,
      "learning_rate": 1.4801014819184872e-06,
      "loss": 0.4322,
      "step": 12071
    },
    {
      "epoch": 0.7561659280602577,
      "grad_norm": 0.8493150641320725,
      "learning_rate": 1.4793811080543791e-06,
      "loss": 0.3707,
      "step": 12072
    },
    {
      "epoch": 0.7562285660596001,
      "grad_norm": 0.9353432798726812,
      "learning_rate": 1.4786608790995965e-06,
      "loss": 0.412,
      "step": 12073
    },
    {
      "epoch": 0.7562912040589423,
      "grad_norm": 0.8845801508952047,
      "learning_rate": 1.4779407950837876e-06,
      "loss": 0.4216,
      "step": 12074
    },
    {
      "epoch": 0.7563538420582847,
      "grad_norm": 0.8135034036090062,
      "learning_rate": 1.4772208560365875e-06,
      "loss": 0.3649,
      "step": 12075
    },
    {
      "epoch": 0.756416480057627,
      "grad_norm": 0.8224854715111544,
      "learning_rate": 1.4765010619876325e-06,
      "loss": 0.4017,
      "step": 12076
    },
    {
      "epoch": 0.7564791180569692,
      "grad_norm": 0.8370650229538394,
      "learning_rate": 1.4757814129665483e-06,
      "loss": 0.3722,
      "step": 12077
    },
    {
      "epoch": 0.7565417560563116,
      "grad_norm": 0.8777461323114917,
      "learning_rate": 1.4750619090029545e-06,
      "loss": 0.4136,
      "step": 12078
    },
    {
      "epoch": 0.7566043940556538,
      "grad_norm": 0.8055311314149846,
      "learning_rate": 1.4743425501264669e-06,
      "loss": 0.3439,
      "step": 12079
    },
    {
      "epoch": 0.7566670320549962,
      "grad_norm": 0.8627102798447374,
      "learning_rate": 1.4736233363666947e-06,
      "loss": 0.3963,
      "step": 12080
    },
    {
      "epoch": 0.7567296700543384,
      "grad_norm": 0.9237897223725997,
      "learning_rate": 1.4729042677532402e-06,
      "loss": 0.3651,
      "step": 12081
    },
    {
      "epoch": 0.7567923080536808,
      "grad_norm": 0.8751048493186416,
      "learning_rate": 1.4721853443156992e-06,
      "loss": 0.3874,
      "step": 12082
    },
    {
      "epoch": 0.7568549460530231,
      "grad_norm": 0.8078939659599695,
      "learning_rate": 1.4714665660836657e-06,
      "loss": 0.3817,
      "step": 12083
    },
    {
      "epoch": 0.7569175840523654,
      "grad_norm": 0.8859802167491553,
      "learning_rate": 1.4707479330867214e-06,
      "loss": 0.3764,
      "step": 12084
    },
    {
      "epoch": 0.7569802220517077,
      "grad_norm": 0.8357608618559447,
      "learning_rate": 1.4700294453544495e-06,
      "loss": 0.3817,
      "step": 12085
    },
    {
      "epoch": 0.7570428600510499,
      "grad_norm": 0.9007892305382708,
      "learning_rate": 1.4693111029164203e-06,
      "loss": 0.3989,
      "step": 12086
    },
    {
      "epoch": 0.7571054980503923,
      "grad_norm": 0.8613858842284355,
      "learning_rate": 1.468592905802202e-06,
      "loss": 0.405,
      "step": 12087
    },
    {
      "epoch": 0.7571681360497345,
      "grad_norm": 0.5787630543574862,
      "learning_rate": 1.4678748540413545e-06,
      "loss": 0.4498,
      "step": 12088
    },
    {
      "epoch": 0.7572307740490769,
      "grad_norm": 0.8614577077308977,
      "learning_rate": 1.467156947663434e-06,
      "loss": 0.3962,
      "step": 12089
    },
    {
      "epoch": 0.7572934120484192,
      "grad_norm": 0.8149783363975587,
      "learning_rate": 1.4664391866979887e-06,
      "loss": 0.3926,
      "step": 12090
    },
    {
      "epoch": 0.7573560500477615,
      "grad_norm": 0.837917429529752,
      "learning_rate": 1.4657215711745614e-06,
      "loss": 0.3995,
      "step": 12091
    },
    {
      "epoch": 0.7574186880471038,
      "grad_norm": 0.8489410439696408,
      "learning_rate": 1.4650041011226918e-06,
      "loss": 0.3635,
      "step": 12092
    },
    {
      "epoch": 0.757481326046446,
      "grad_norm": 0.7804052952517737,
      "learning_rate": 1.4642867765719077e-06,
      "loss": 0.3516,
      "step": 12093
    },
    {
      "epoch": 0.7575439640457884,
      "grad_norm": 0.6308275215487699,
      "learning_rate": 1.4635695975517378e-06,
      "loss": 0.477,
      "step": 12094
    },
    {
      "epoch": 0.7576066020451306,
      "grad_norm": 0.8448207262879098,
      "learning_rate": 1.4628525640916997e-06,
      "loss": 0.4399,
      "step": 12095
    },
    {
      "epoch": 0.757669240044473,
      "grad_norm": 0.8022785374886535,
      "learning_rate": 1.4621356762213062e-06,
      "loss": 0.3923,
      "step": 12096
    },
    {
      "epoch": 0.7577318780438153,
      "grad_norm": 0.5982452442792109,
      "learning_rate": 1.4614189339700652e-06,
      "loss": 0.4482,
      "step": 12097
    },
    {
      "epoch": 0.7577945160431576,
      "grad_norm": 0.9614340097255135,
      "learning_rate": 1.4607023373674778e-06,
      "loss": 0.4606,
      "step": 12098
    },
    {
      "epoch": 0.7578571540424999,
      "grad_norm": 0.8908489262647957,
      "learning_rate": 1.4599858864430377e-06,
      "loss": 0.4232,
      "step": 12099
    },
    {
      "epoch": 0.7579197920418422,
      "grad_norm": 0.87538951886884,
      "learning_rate": 1.459269581226237e-06,
      "loss": 0.4207,
      "step": 12100
    },
    {
      "epoch": 0.7579824300411845,
      "grad_norm": 0.8294292724135188,
      "learning_rate": 1.4585534217465575e-06,
      "loss": 0.3906,
      "step": 12101
    },
    {
      "epoch": 0.7580450680405267,
      "grad_norm": 0.8972783137529105,
      "learning_rate": 1.4578374080334745e-06,
      "loss": 0.4211,
      "step": 12102
    },
    {
      "epoch": 0.7581077060398691,
      "grad_norm": 0.8161628837569789,
      "learning_rate": 1.4571215401164634e-06,
      "loss": 0.3617,
      "step": 12103
    },
    {
      "epoch": 0.7581703440392114,
      "grad_norm": 0.8296136213492064,
      "learning_rate": 1.456405818024987e-06,
      "loss": 0.3888,
      "step": 12104
    },
    {
      "epoch": 0.7582329820385537,
      "grad_norm": 0.8266694684585637,
      "learning_rate": 1.4556902417885048e-06,
      "loss": 0.375,
      "step": 12105
    },
    {
      "epoch": 0.758295620037896,
      "grad_norm": 0.90250392254311,
      "learning_rate": 1.4549748114364704e-06,
      "loss": 0.386,
      "step": 12106
    },
    {
      "epoch": 0.7583582580372383,
      "grad_norm": 0.8414386087688799,
      "learning_rate": 1.4542595269983301e-06,
      "loss": 0.3414,
      "step": 12107
    },
    {
      "epoch": 0.7584208960365806,
      "grad_norm": 0.8698788675521039,
      "learning_rate": 1.4535443885035249e-06,
      "loss": 0.4012,
      "step": 12108
    },
    {
      "epoch": 0.758483534035923,
      "grad_norm": 0.9039688369997464,
      "learning_rate": 1.4528293959814921e-06,
      "loss": 0.4205,
      "step": 12109
    },
    {
      "epoch": 0.7585461720352652,
      "grad_norm": 0.6096599535270937,
      "learning_rate": 1.45211454946166e-06,
      "loss": 0.47,
      "step": 12110
    },
    {
      "epoch": 0.7586088100346075,
      "grad_norm": 0.800281967472866,
      "learning_rate": 1.4513998489734516e-06,
      "loss": 0.3973,
      "step": 12111
    },
    {
      "epoch": 0.7586714480339498,
      "grad_norm": 0.797479587430063,
      "learning_rate": 1.4506852945462824e-06,
      "loss": 0.3845,
      "step": 12112
    },
    {
      "epoch": 0.7587340860332921,
      "grad_norm": 0.9216217989610656,
      "learning_rate": 1.449970886209567e-06,
      "loss": 0.444,
      "step": 12113
    },
    {
      "epoch": 0.7587967240326344,
      "grad_norm": 0.8284214014644163,
      "learning_rate": 1.4492566239927081e-06,
      "loss": 0.3688,
      "step": 12114
    },
    {
      "epoch": 0.7588593620319767,
      "grad_norm": 0.8263567545125491,
      "learning_rate": 1.4485425079251064e-06,
      "loss": 0.3415,
      "step": 12115
    },
    {
      "epoch": 0.758922000031319,
      "grad_norm": 0.8561944383866715,
      "learning_rate": 1.447828538036154e-06,
      "loss": 0.3535,
      "step": 12116
    },
    {
      "epoch": 0.7589846380306613,
      "grad_norm": 0.8692455544846025,
      "learning_rate": 1.4471147143552366e-06,
      "loss": 0.3821,
      "step": 12117
    },
    {
      "epoch": 0.7590472760300035,
      "grad_norm": 0.8284878131614669,
      "learning_rate": 1.4464010369117386e-06,
      "loss": 0.3747,
      "step": 12118
    },
    {
      "epoch": 0.7591099140293459,
      "grad_norm": 0.8726638697485299,
      "learning_rate": 1.4456875057350339e-06,
      "loss": 0.3881,
      "step": 12119
    },
    {
      "epoch": 0.7591725520286882,
      "grad_norm": 0.8197134021885127,
      "learning_rate": 1.444974120854491e-06,
      "loss": 0.4079,
      "step": 12120
    },
    {
      "epoch": 0.7592351900280305,
      "grad_norm": 0.8764897981348678,
      "learning_rate": 1.4442608822994713e-06,
      "loss": 0.3801,
      "step": 12121
    },
    {
      "epoch": 0.7592978280273728,
      "grad_norm": 0.8514554315397287,
      "learning_rate": 1.443547790099335e-06,
      "loss": 0.3572,
      "step": 12122
    },
    {
      "epoch": 0.7593604660267151,
      "grad_norm": 0.8839789439295451,
      "learning_rate": 1.4428348442834322e-06,
      "loss": 0.3685,
      "step": 12123
    },
    {
      "epoch": 0.7594231040260574,
      "grad_norm": 0.8545258875998649,
      "learning_rate": 1.4421220448811074e-06,
      "loss": 0.3935,
      "step": 12124
    },
    {
      "epoch": 0.7594857420253998,
      "grad_norm": 0.848692878557991,
      "learning_rate": 1.4414093919216982e-06,
      "loss": 0.4061,
      "step": 12125
    },
    {
      "epoch": 0.759548380024742,
      "grad_norm": 0.8887646655648441,
      "learning_rate": 1.4406968854345405e-06,
      "loss": 0.3791,
      "step": 12126
    },
    {
      "epoch": 0.7596110180240843,
      "grad_norm": 0.8666986232780239,
      "learning_rate": 1.439984525448959e-06,
      "loss": 0.3744,
      "step": 12127
    },
    {
      "epoch": 0.7596736560234266,
      "grad_norm": 0.8458068465182685,
      "learning_rate": 1.4392723119942758e-06,
      "loss": 0.3759,
      "step": 12128
    },
    {
      "epoch": 0.7597362940227689,
      "grad_norm": 0.8959274460596769,
      "learning_rate": 1.4385602450998048e-06,
      "loss": 0.4413,
      "step": 12129
    },
    {
      "epoch": 0.7597989320221112,
      "grad_norm": 0.9182813027365249,
      "learning_rate": 1.4378483247948538e-06,
      "loss": 0.3805,
      "step": 12130
    },
    {
      "epoch": 0.7598615700214535,
      "grad_norm": 0.8185410664869485,
      "learning_rate": 1.4371365511087286e-06,
      "loss": 0.3612,
      "step": 12131
    },
    {
      "epoch": 0.7599242080207959,
      "grad_norm": 0.8273158217254062,
      "learning_rate": 1.436424924070724e-06,
      "loss": 0.4093,
      "step": 12132
    },
    {
      "epoch": 0.7599868460201381,
      "grad_norm": 0.8802481823697501,
      "learning_rate": 1.435713443710131e-06,
      "loss": 0.4003,
      "step": 12133
    },
    {
      "epoch": 0.7600494840194805,
      "grad_norm": 0.8612892029211697,
      "learning_rate": 1.4350021100562329e-06,
      "loss": 0.3752,
      "step": 12134
    },
    {
      "epoch": 0.7601121220188227,
      "grad_norm": 0.8521674629684327,
      "learning_rate": 1.4342909231383111e-06,
      "loss": 0.4209,
      "step": 12135
    },
    {
      "epoch": 0.760174760018165,
      "grad_norm": 0.8365471735770125,
      "learning_rate": 1.4335798829856368e-06,
      "loss": 0.3497,
      "step": 12136
    },
    {
      "epoch": 0.7602373980175073,
      "grad_norm": 0.8947058259618716,
      "learning_rate": 1.4328689896274762e-06,
      "loss": 0.416,
      "step": 12137
    },
    {
      "epoch": 0.7603000360168496,
      "grad_norm": 0.9100640442852441,
      "learning_rate": 1.4321582430930903e-06,
      "loss": 0.38,
      "step": 12138
    },
    {
      "epoch": 0.760362674016192,
      "grad_norm": 0.8346595032643489,
      "learning_rate": 1.4314476434117314e-06,
      "loss": 0.3886,
      "step": 12139
    },
    {
      "epoch": 0.7604253120155342,
      "grad_norm": 0.8900171016250094,
      "learning_rate": 1.4307371906126517e-06,
      "loss": 0.4049,
      "step": 12140
    },
    {
      "epoch": 0.7604879500148766,
      "grad_norm": 0.8748385984273385,
      "learning_rate": 1.430026884725092e-06,
      "loss": 0.3778,
      "step": 12141
    },
    {
      "epoch": 0.7605505880142188,
      "grad_norm": 0.9193700281573849,
      "learning_rate": 1.429316725778288e-06,
      "loss": 0.4085,
      "step": 12142
    },
    {
      "epoch": 0.7606132260135611,
      "grad_norm": 0.8517764403034263,
      "learning_rate": 1.4286067138014686e-06,
      "loss": 0.3902,
      "step": 12143
    },
    {
      "epoch": 0.7606758640129034,
      "grad_norm": 0.9105667386684505,
      "learning_rate": 1.4278968488238614e-06,
      "loss": 0.36,
      "step": 12144
    },
    {
      "epoch": 0.7607385020122457,
      "grad_norm": 0.8605863534803899,
      "learning_rate": 1.4271871308746833e-06,
      "loss": 0.4158,
      "step": 12145
    },
    {
      "epoch": 0.760801140011588,
      "grad_norm": 0.581404724954555,
      "learning_rate": 1.4264775599831459e-06,
      "loss": 0.4295,
      "step": 12146
    },
    {
      "epoch": 0.7608637780109303,
      "grad_norm": 0.8392758306452512,
      "learning_rate": 1.4257681361784552e-06,
      "loss": 0.3513,
      "step": 12147
    },
    {
      "epoch": 0.7609264160102727,
      "grad_norm": 0.8411815384199427,
      "learning_rate": 1.4250588594898102e-06,
      "loss": 0.3654,
      "step": 12148
    },
    {
      "epoch": 0.7609890540096149,
      "grad_norm": 0.8299617079723282,
      "learning_rate": 1.424349729946407e-06,
      "loss": 0.3789,
      "step": 12149
    },
    {
      "epoch": 0.7610516920089573,
      "grad_norm": 0.8959514571918932,
      "learning_rate": 1.4236407475774333e-06,
      "loss": 0.3829,
      "step": 12150
    },
    {
      "epoch": 0.7611143300082995,
      "grad_norm": 0.6026465301981816,
      "learning_rate": 1.4229319124120683e-06,
      "loss": 0.4287,
      "step": 12151
    },
    {
      "epoch": 0.7611769680076418,
      "grad_norm": 0.8065029805443577,
      "learning_rate": 1.4222232244794915e-06,
      "loss": 0.3316,
      "step": 12152
    },
    {
      "epoch": 0.7612396060069841,
      "grad_norm": 0.7968459535160386,
      "learning_rate": 1.4215146838088706e-06,
      "loss": 0.3224,
      "step": 12153
    },
    {
      "epoch": 0.7613022440063264,
      "grad_norm": 0.8475643951001758,
      "learning_rate": 1.4208062904293696e-06,
      "loss": 0.3802,
      "step": 12154
    },
    {
      "epoch": 0.7613648820056688,
      "grad_norm": 0.8340806372446653,
      "learning_rate": 1.4200980443701456e-06,
      "loss": 0.3836,
      "step": 12155
    },
    {
      "epoch": 0.761427520005011,
      "grad_norm": 0.9024225163701779,
      "learning_rate": 1.419389945660351e-06,
      "loss": 0.3734,
      "step": 12156
    },
    {
      "epoch": 0.7614901580043534,
      "grad_norm": 0.8612187242989288,
      "learning_rate": 1.4186819943291307e-06,
      "loss": 0.3831,
      "step": 12157
    },
    {
      "epoch": 0.7615527960036956,
      "grad_norm": 0.8436102091523723,
      "learning_rate": 1.4179741904056233e-06,
      "loss": 0.4005,
      "step": 12158
    },
    {
      "epoch": 0.761615434003038,
      "grad_norm": 0.8405025323603112,
      "learning_rate": 1.417266533918964e-06,
      "loss": 0.3593,
      "step": 12159
    },
    {
      "epoch": 0.7616780720023802,
      "grad_norm": 0.9150130272063589,
      "learning_rate": 1.4165590248982775e-06,
      "loss": 0.3976,
      "step": 12160
    },
    {
      "epoch": 0.7617407100017225,
      "grad_norm": 0.5399217039827935,
      "learning_rate": 1.415851663372688e-06,
      "loss": 0.4353,
      "step": 12161
    },
    {
      "epoch": 0.7618033480010649,
      "grad_norm": 0.8759214536211869,
      "learning_rate": 1.4151444493713102e-06,
      "loss": 0.3237,
      "step": 12162
    },
    {
      "epoch": 0.7618659860004071,
      "grad_norm": 0.806527025232558,
      "learning_rate": 1.4144373829232516e-06,
      "loss": 0.3832,
      "step": 12163
    },
    {
      "epoch": 0.7619286239997495,
      "grad_norm": 0.8935201645899035,
      "learning_rate": 1.4137304640576161e-06,
      "loss": 0.4146,
      "step": 12164
    },
    {
      "epoch": 0.7619912619990917,
      "grad_norm": 0.9051083301153748,
      "learning_rate": 1.4130236928035001e-06,
      "loss": 0.4014,
      "step": 12165
    },
    {
      "epoch": 0.7620538999984341,
      "grad_norm": 0.8365884918426034,
      "learning_rate": 1.4123170691899952e-06,
      "loss": 0.3938,
      "step": 12166
    },
    {
      "epoch": 0.7621165379977763,
      "grad_norm": 0.8612325268446295,
      "learning_rate": 1.4116105932461833e-06,
      "loss": 0.3732,
      "step": 12167
    },
    {
      "epoch": 0.7621791759971187,
      "grad_norm": 0.8560047304272803,
      "learning_rate": 1.4109042650011473e-06,
      "loss": 0.3833,
      "step": 12168
    },
    {
      "epoch": 0.762241813996461,
      "grad_norm": 0.9225428810652551,
      "learning_rate": 1.4101980844839564e-06,
      "loss": 0.3805,
      "step": 12169
    },
    {
      "epoch": 0.7623044519958032,
      "grad_norm": 0.8367052780872859,
      "learning_rate": 1.4094920517236805e-06,
      "loss": 0.3817,
      "step": 12170
    },
    {
      "epoch": 0.7623670899951456,
      "grad_norm": 0.942516333000648,
      "learning_rate": 1.408786166749378e-06,
      "loss": 0.4151,
      "step": 12171
    },
    {
      "epoch": 0.7624297279944878,
      "grad_norm": 0.7940558823169905,
      "learning_rate": 1.408080429590103e-06,
      "loss": 0.3862,
      "step": 12172
    },
    {
      "epoch": 0.7624923659938302,
      "grad_norm": 0.8858264866321723,
      "learning_rate": 1.4073748402749044e-06,
      "loss": 0.3825,
      "step": 12173
    },
    {
      "epoch": 0.7625550039931724,
      "grad_norm": 0.8182237648717382,
      "learning_rate": 1.4066693988328246e-06,
      "loss": 0.3641,
      "step": 12174
    },
    {
      "epoch": 0.7626176419925148,
      "grad_norm": 0.8833151484153386,
      "learning_rate": 1.405964105292899e-06,
      "loss": 0.4084,
      "step": 12175
    },
    {
      "epoch": 0.7626802799918571,
      "grad_norm": 0.7922951216724363,
      "learning_rate": 1.4052589596841566e-06,
      "loss": 0.3888,
      "step": 12176
    },
    {
      "epoch": 0.7627429179911993,
      "grad_norm": 0.8973038108809323,
      "learning_rate": 1.4045539620356225e-06,
      "loss": 0.3763,
      "step": 12177
    },
    {
      "epoch": 0.7628055559905417,
      "grad_norm": 0.6552503798120737,
      "learning_rate": 1.4038491123763158e-06,
      "loss": 0.466,
      "step": 12178
    },
    {
      "epoch": 0.7628681939898839,
      "grad_norm": 0.8997362292877513,
      "learning_rate": 1.4031444107352476e-06,
      "loss": 0.3925,
      "step": 12179
    },
    {
      "epoch": 0.7629308319892263,
      "grad_norm": 0.8322086031144545,
      "learning_rate": 1.4024398571414227e-06,
      "loss": 0.3595,
      "step": 12180
    },
    {
      "epoch": 0.7629934699885685,
      "grad_norm": 0.7974591151758381,
      "learning_rate": 1.4017354516238413e-06,
      "loss": 0.3475,
      "step": 12181
    },
    {
      "epoch": 0.7630561079879109,
      "grad_norm": 0.8321760075293948,
      "learning_rate": 1.4010311942114968e-06,
      "loss": 0.3787,
      "step": 12182
    },
    {
      "epoch": 0.7631187459872532,
      "grad_norm": 0.8438390755992784,
      "learning_rate": 1.4003270849333756e-06,
      "loss": 0.3841,
      "step": 12183
    },
    {
      "epoch": 0.7631813839865955,
      "grad_norm": 0.8823872514427169,
      "learning_rate": 1.3996231238184587e-06,
      "loss": 0.404,
      "step": 12184
    },
    {
      "epoch": 0.7632440219859378,
      "grad_norm": 0.9496868439398807,
      "learning_rate": 1.3989193108957238e-06,
      "loss": 0.4235,
      "step": 12185
    },
    {
      "epoch": 0.76330665998528,
      "grad_norm": 0.859109847342525,
      "learning_rate": 1.398215646194137e-06,
      "loss": 0.3743,
      "step": 12186
    },
    {
      "epoch": 0.7633692979846224,
      "grad_norm": 0.9607859801888406,
      "learning_rate": 1.3975121297426641e-06,
      "loss": 0.4061,
      "step": 12187
    },
    {
      "epoch": 0.7634319359839646,
      "grad_norm": 0.8304989218471871,
      "learning_rate": 1.3968087615702607e-06,
      "loss": 0.3572,
      "step": 12188
    },
    {
      "epoch": 0.763494573983307,
      "grad_norm": 0.804747260847467,
      "learning_rate": 1.3961055417058772e-06,
      "loss": 0.3372,
      "step": 12189
    },
    {
      "epoch": 0.7635572119826493,
      "grad_norm": 0.8799194405001639,
      "learning_rate": 1.3954024701784591e-06,
      "loss": 0.3996,
      "step": 12190
    },
    {
      "epoch": 0.7636198499819916,
      "grad_norm": 0.8661435233794036,
      "learning_rate": 1.394699547016944e-06,
      "loss": 0.4019,
      "step": 12191
    },
    {
      "epoch": 0.7636824879813339,
      "grad_norm": 0.8703947555515046,
      "learning_rate": 1.3939967722502646e-06,
      "loss": 0.4061,
      "step": 12192
    },
    {
      "epoch": 0.7637451259806762,
      "grad_norm": 0.8344715705171691,
      "learning_rate": 1.3932941459073456e-06,
      "loss": 0.3951,
      "step": 12193
    },
    {
      "epoch": 0.7638077639800185,
      "grad_norm": 0.8667857435777439,
      "learning_rate": 1.3925916680171108e-06,
      "loss": 0.3646,
      "step": 12194
    },
    {
      "epoch": 0.7638704019793607,
      "grad_norm": 0.9135956982314181,
      "learning_rate": 1.3918893386084715e-06,
      "loss": 0.3663,
      "step": 12195
    },
    {
      "epoch": 0.7639330399787031,
      "grad_norm": 0.5942318700825758,
      "learning_rate": 1.3911871577103375e-06,
      "loss": 0.4524,
      "step": 12196
    },
    {
      "epoch": 0.7639956779780454,
      "grad_norm": 0.8329658567539894,
      "learning_rate": 1.3904851253516106e-06,
      "loss": 0.4239,
      "step": 12197
    },
    {
      "epoch": 0.7640583159773877,
      "grad_norm": 1.1739444533116838,
      "learning_rate": 1.389783241561185e-06,
      "loss": 0.4172,
      "step": 12198
    },
    {
      "epoch": 0.76412095397673,
      "grad_norm": 0.9153387588657722,
      "learning_rate": 1.389081506367952e-06,
      "loss": 0.3718,
      "step": 12199
    },
    {
      "epoch": 0.7641835919760723,
      "grad_norm": 0.8249830806502529,
      "learning_rate": 1.388379919800794e-06,
      "loss": 0.3421,
      "step": 12200
    },
    {
      "epoch": 0.7642462299754146,
      "grad_norm": 0.8288664151670275,
      "learning_rate": 1.387678481888589e-06,
      "loss": 0.342,
      "step": 12201
    },
    {
      "epoch": 0.7643088679747568,
      "grad_norm": 0.875040188020738,
      "learning_rate": 1.3869771926602066e-06,
      "loss": 0.3958,
      "step": 12202
    },
    {
      "epoch": 0.7643715059740992,
      "grad_norm": 0.8289110277979953,
      "learning_rate": 1.3862760521445152e-06,
      "loss": 0.3686,
      "step": 12203
    },
    {
      "epoch": 0.7644341439734414,
      "grad_norm": 0.8285509174668271,
      "learning_rate": 1.385575060370371e-06,
      "loss": 0.3814,
      "step": 12204
    },
    {
      "epoch": 0.7644967819727838,
      "grad_norm": 0.8130016442810921,
      "learning_rate": 1.3848742173666291e-06,
      "loss": 0.3788,
      "step": 12205
    },
    {
      "epoch": 0.7645594199721261,
      "grad_norm": 0.8072134104851697,
      "learning_rate": 1.3841735231621357e-06,
      "loss": 0.3895,
      "step": 12206
    },
    {
      "epoch": 0.7646220579714684,
      "grad_norm": 0.9122776678779502,
      "learning_rate": 1.3834729777857314e-06,
      "loss": 0.3982,
      "step": 12207
    },
    {
      "epoch": 0.7646846959708107,
      "grad_norm": 0.8591400599056112,
      "learning_rate": 1.38277258126625e-06,
      "loss": 0.3508,
      "step": 12208
    },
    {
      "epoch": 0.764747333970153,
      "grad_norm": 0.8757206343523144,
      "learning_rate": 1.382072333632521e-06,
      "loss": 0.3682,
      "step": 12209
    },
    {
      "epoch": 0.7648099719694953,
      "grad_norm": 0.9000216742400504,
      "learning_rate": 1.3813722349133646e-06,
      "loss": 0.3923,
      "step": 12210
    },
    {
      "epoch": 0.7648726099688375,
      "grad_norm": 0.8862747282644081,
      "learning_rate": 1.3806722851376004e-06,
      "loss": 0.4425,
      "step": 12211
    },
    {
      "epoch": 0.7649352479681799,
      "grad_norm": 0.8459340531240942,
      "learning_rate": 1.3799724843340368e-06,
      "loss": 0.3822,
      "step": 12212
    },
    {
      "epoch": 0.7649978859675222,
      "grad_norm": 0.8543327393280203,
      "learning_rate": 1.379272832531476e-06,
      "loss": 0.4108,
      "step": 12213
    },
    {
      "epoch": 0.7650605239668645,
      "grad_norm": 0.8764674788870692,
      "learning_rate": 1.378573329758719e-06,
      "loss": 0.4206,
      "step": 12214
    },
    {
      "epoch": 0.7651231619662068,
      "grad_norm": 0.79571163161263,
      "learning_rate": 1.3778739760445552e-06,
      "loss": 0.3399,
      "step": 12215
    },
    {
      "epoch": 0.7651857999655491,
      "grad_norm": 0.8924452190904014,
      "learning_rate": 1.3771747714177719e-06,
      "loss": 0.3663,
      "step": 12216
    },
    {
      "epoch": 0.7652484379648914,
      "grad_norm": 0.8883699019742137,
      "learning_rate": 1.3764757159071468e-06,
      "loss": 0.4024,
      "step": 12217
    },
    {
      "epoch": 0.7653110759642338,
      "grad_norm": 0.8378904128958053,
      "learning_rate": 1.3757768095414542e-06,
      "loss": 0.4029,
      "step": 12218
    },
    {
      "epoch": 0.765373713963576,
      "grad_norm": 0.8380432347798005,
      "learning_rate": 1.3750780523494589e-06,
      "loss": 0.3751,
      "step": 12219
    },
    {
      "epoch": 0.7654363519629183,
      "grad_norm": 0.5871769837485464,
      "learning_rate": 1.3743794443599256e-06,
      "loss": 0.474,
      "step": 12220
    },
    {
      "epoch": 0.7654989899622606,
      "grad_norm": 0.8752632841571062,
      "learning_rate": 1.373680985601607e-06,
      "loss": 0.3971,
      "step": 12221
    },
    {
      "epoch": 0.7655616279616029,
      "grad_norm": 0.8945133089204861,
      "learning_rate": 1.3729826761032506e-06,
      "loss": 0.392,
      "step": 12222
    },
    {
      "epoch": 0.7656242659609452,
      "grad_norm": 0.9052446885149752,
      "learning_rate": 1.3722845158936022e-06,
      "loss": 0.3942,
      "step": 12223
    },
    {
      "epoch": 0.7656869039602875,
      "grad_norm": 1.0010257552679802,
      "learning_rate": 1.3715865050013965e-06,
      "loss": 0.4183,
      "step": 12224
    },
    {
      "epoch": 0.7657495419596299,
      "grad_norm": 0.8651119353342405,
      "learning_rate": 1.3708886434553637e-06,
      "loss": 0.3872,
      "step": 12225
    },
    {
      "epoch": 0.7658121799589721,
      "grad_norm": 0.8474481649727038,
      "learning_rate": 1.3701909312842277e-06,
      "loss": 0.3476,
      "step": 12226
    },
    {
      "epoch": 0.7658748179583144,
      "grad_norm": 0.8956496979022419,
      "learning_rate": 1.3694933685167073e-06,
      "loss": 0.4008,
      "step": 12227
    },
    {
      "epoch": 0.7659374559576567,
      "grad_norm": 0.8120681337206862,
      "learning_rate": 1.3687959551815117e-06,
      "loss": 0.3751,
      "step": 12228
    },
    {
      "epoch": 0.766000093956999,
      "grad_norm": 0.8261852463441699,
      "learning_rate": 1.3680986913073496e-06,
      "loss": 0.361,
      "step": 12229
    },
    {
      "epoch": 0.7660627319563413,
      "grad_norm": 0.9130933963427353,
      "learning_rate": 1.3674015769229204e-06,
      "loss": 0.3882,
      "step": 12230
    },
    {
      "epoch": 0.7661253699556836,
      "grad_norm": 0.8191546884633607,
      "learning_rate": 1.3667046120569145e-06,
      "loss": 0.4183,
      "step": 12231
    },
    {
      "epoch": 0.766188007955026,
      "grad_norm": 0.618787594078647,
      "learning_rate": 1.3660077967380226e-06,
      "loss": 0.4425,
      "step": 12232
    },
    {
      "epoch": 0.7662506459543682,
      "grad_norm": 0.8657862546819474,
      "learning_rate": 1.3653111309949247e-06,
      "loss": 0.3563,
      "step": 12233
    },
    {
      "epoch": 0.7663132839537106,
      "grad_norm": 0.850080431081994,
      "learning_rate": 1.364614614856295e-06,
      "loss": 0.3904,
      "step": 12234
    },
    {
      "epoch": 0.7663759219530528,
      "grad_norm": 0.8449672500295595,
      "learning_rate": 1.3639182483508024e-06,
      "loss": 0.3583,
      "step": 12235
    },
    {
      "epoch": 0.7664385599523951,
      "grad_norm": 0.8022654823269632,
      "learning_rate": 1.3632220315071082e-06,
      "loss": 0.3486,
      "step": 12236
    },
    {
      "epoch": 0.7665011979517374,
      "grad_norm": 0.8739947890307009,
      "learning_rate": 1.3625259643538718e-06,
      "loss": 0.4033,
      "step": 12237
    },
    {
      "epoch": 0.7665638359510797,
      "grad_norm": 0.8599039321160153,
      "learning_rate": 1.361830046919742e-06,
      "loss": 0.4449,
      "step": 12238
    },
    {
      "epoch": 0.766626473950422,
      "grad_norm": 0.6286663936904477,
      "learning_rate": 1.3611342792333626e-06,
      "loss": 0.4488,
      "step": 12239
    },
    {
      "epoch": 0.7666891119497643,
      "grad_norm": 0.8598966412011767,
      "learning_rate": 1.3604386613233716e-06,
      "loss": 0.3792,
      "step": 12240
    },
    {
      "epoch": 0.7667517499491067,
      "grad_norm": 0.6222692885101674,
      "learning_rate": 1.3597431932183991e-06,
      "loss": 0.4435,
      "step": 12241
    },
    {
      "epoch": 0.7668143879484489,
      "grad_norm": 0.9060732377064754,
      "learning_rate": 1.3590478749470742e-06,
      "loss": 0.404,
      "step": 12242
    },
    {
      "epoch": 0.7668770259477913,
      "grad_norm": 0.8669299732562432,
      "learning_rate": 1.3583527065380147e-06,
      "loss": 0.3622,
      "step": 12243
    },
    {
      "epoch": 0.7669396639471335,
      "grad_norm": 0.7910703676855856,
      "learning_rate": 1.3576576880198332e-06,
      "loss": 0.3617,
      "step": 12244
    },
    {
      "epoch": 0.7670023019464758,
      "grad_norm": 0.7716745320012269,
      "learning_rate": 1.3569628194211365e-06,
      "loss": 0.404,
      "step": 12245
    },
    {
      "epoch": 0.7670649399458181,
      "grad_norm": 0.8754808038602627,
      "learning_rate": 1.3562681007705276e-06,
      "loss": 0.3856,
      "step": 12246
    },
    {
      "epoch": 0.7671275779451604,
      "grad_norm": 0.923508622234522,
      "learning_rate": 1.3555735320965996e-06,
      "loss": 0.3957,
      "step": 12247
    },
    {
      "epoch": 0.7671902159445028,
      "grad_norm": 0.8581669218832356,
      "learning_rate": 1.3548791134279415e-06,
      "loss": 0.3782,
      "step": 12248
    },
    {
      "epoch": 0.767252853943845,
      "grad_norm": 0.9003232766851769,
      "learning_rate": 1.354184844793136e-06,
      "loss": 0.4163,
      "step": 12249
    },
    {
      "epoch": 0.7673154919431874,
      "grad_norm": 0.8725274676133806,
      "learning_rate": 1.3534907262207574e-06,
      "loss": 0.3907,
      "step": 12250
    },
    {
      "epoch": 0.7673781299425296,
      "grad_norm": 0.8997347758293992,
      "learning_rate": 1.352796757739379e-06,
      "loss": 0.4311,
      "step": 12251
    },
    {
      "epoch": 0.7674407679418719,
      "grad_norm": 0.822014412230924,
      "learning_rate": 1.3521029393775625e-06,
      "loss": 0.3961,
      "step": 12252
    },
    {
      "epoch": 0.7675034059412142,
      "grad_norm": 0.8272594107552561,
      "learning_rate": 1.3514092711638665e-06,
      "loss": 0.3716,
      "step": 12253
    },
    {
      "epoch": 0.7675660439405565,
      "grad_norm": 0.8906456033923704,
      "learning_rate": 1.3507157531268406e-06,
      "loss": 0.3791,
      "step": 12254
    },
    {
      "epoch": 0.7676286819398989,
      "grad_norm": 0.9387038573497993,
      "learning_rate": 1.3500223852950334e-06,
      "loss": 0.3866,
      "step": 12255
    },
    {
      "epoch": 0.7676913199392411,
      "grad_norm": 0.7884871208395204,
      "learning_rate": 1.3493291676969817e-06,
      "loss": 0.348,
      "step": 12256
    },
    {
      "epoch": 0.7677539579385835,
      "grad_norm": 0.9070487413923388,
      "learning_rate": 1.348636100361219e-06,
      "loss": 0.3967,
      "step": 12257
    },
    {
      "epoch": 0.7678165959379257,
      "grad_norm": 0.8103851404919518,
      "learning_rate": 1.3479431833162727e-06,
      "loss": 0.4112,
      "step": 12258
    },
    {
      "epoch": 0.7678792339372681,
      "grad_norm": 0.6163145405401721,
      "learning_rate": 1.3472504165906614e-06,
      "loss": 0.4359,
      "step": 12259
    },
    {
      "epoch": 0.7679418719366103,
      "grad_norm": 0.8677157888759892,
      "learning_rate": 1.3465578002129021e-06,
      "loss": 0.3793,
      "step": 12260
    },
    {
      "epoch": 0.7680045099359526,
      "grad_norm": 0.9681653462652177,
      "learning_rate": 1.3458653342115013e-06,
      "loss": 0.3968,
      "step": 12261
    },
    {
      "epoch": 0.768067147935295,
      "grad_norm": 0.841812904657243,
      "learning_rate": 1.3451730186149608e-06,
      "loss": 0.3964,
      "step": 12262
    },
    {
      "epoch": 0.7681297859346372,
      "grad_norm": 0.8733077293418391,
      "learning_rate": 1.344480853451779e-06,
      "loss": 0.3822,
      "step": 12263
    },
    {
      "epoch": 0.7681924239339796,
      "grad_norm": 0.8266265659234332,
      "learning_rate": 1.3437888387504427e-06,
      "loss": 0.3912,
      "step": 12264
    },
    {
      "epoch": 0.7682550619333218,
      "grad_norm": 0.8766775646978666,
      "learning_rate": 1.3430969745394373e-06,
      "loss": 0.4276,
      "step": 12265
    },
    {
      "epoch": 0.7683176999326642,
      "grad_norm": 0.832345945638818,
      "learning_rate": 1.3424052608472389e-06,
      "loss": 0.3911,
      "step": 12266
    },
    {
      "epoch": 0.7683803379320064,
      "grad_norm": 0.9029772326378117,
      "learning_rate": 1.3417136977023187e-06,
      "loss": 0.3985,
      "step": 12267
    },
    {
      "epoch": 0.7684429759313488,
      "grad_norm": 0.7963639359669342,
      "learning_rate": 1.3410222851331407e-06,
      "loss": 0.3755,
      "step": 12268
    },
    {
      "epoch": 0.7685056139306911,
      "grad_norm": 0.6263038791494209,
      "learning_rate": 1.3403310231681654e-06,
      "loss": 0.4595,
      "step": 12269
    },
    {
      "epoch": 0.7685682519300333,
      "grad_norm": 0.8256420051082473,
      "learning_rate": 1.339639911835845e-06,
      "loss": 0.3829,
      "step": 12270
    },
    {
      "epoch": 0.7686308899293757,
      "grad_norm": 0.852379488506237,
      "learning_rate": 1.3389489511646236e-06,
      "loss": 0.3347,
      "step": 12271
    },
    {
      "epoch": 0.7686935279287179,
      "grad_norm": 0.8585707141728599,
      "learning_rate": 1.3382581411829443e-06,
      "loss": 0.401,
      "step": 12272
    },
    {
      "epoch": 0.7687561659280603,
      "grad_norm": 0.9324896700705212,
      "learning_rate": 1.3375674819192398e-06,
      "loss": 0.385,
      "step": 12273
    },
    {
      "epoch": 0.7688188039274025,
      "grad_norm": 0.8501849052077861,
      "learning_rate": 1.3368769734019371e-06,
      "loss": 0.3334,
      "step": 12274
    },
    {
      "epoch": 0.7688814419267449,
      "grad_norm": 0.8096883810030475,
      "learning_rate": 1.3361866156594582e-06,
      "loss": 0.3975,
      "step": 12275
    },
    {
      "epoch": 0.7689440799260872,
      "grad_norm": 0.846333219715112,
      "learning_rate": 1.3354964087202188e-06,
      "loss": 0.3691,
      "step": 12276
    },
    {
      "epoch": 0.7690067179254295,
      "grad_norm": 0.8612027673259276,
      "learning_rate": 1.3348063526126253e-06,
      "loss": 0.3966,
      "step": 12277
    },
    {
      "epoch": 0.7690693559247718,
      "grad_norm": 0.8796977199554558,
      "learning_rate": 1.3341164473650842e-06,
      "loss": 0.3898,
      "step": 12278
    },
    {
      "epoch": 0.769131993924114,
      "grad_norm": 0.8721480628550387,
      "learning_rate": 1.3334266930059908e-06,
      "loss": 0.416,
      "step": 12279
    },
    {
      "epoch": 0.7691946319234564,
      "grad_norm": 0.8348490366687206,
      "learning_rate": 1.3327370895637332e-06,
      "loss": 0.3577,
      "step": 12280
    },
    {
      "epoch": 0.7692572699227986,
      "grad_norm": 0.8757652763913041,
      "learning_rate": 1.3320476370666996e-06,
      "loss": 0.3714,
      "step": 12281
    },
    {
      "epoch": 0.769319907922141,
      "grad_norm": 0.8513967452411835,
      "learning_rate": 1.3313583355432658e-06,
      "loss": 0.4232,
      "step": 12282
    },
    {
      "epoch": 0.7693825459214833,
      "grad_norm": 0.8239432119636488,
      "learning_rate": 1.330669185021804e-06,
      "loss": 0.3591,
      "step": 12283
    },
    {
      "epoch": 0.7694451839208256,
      "grad_norm": 0.8915006557559466,
      "learning_rate": 1.3299801855306799e-06,
      "loss": 0.3757,
      "step": 12284
    },
    {
      "epoch": 0.7695078219201679,
      "grad_norm": 0.9178594081581041,
      "learning_rate": 1.3292913370982524e-06,
      "loss": 0.4218,
      "step": 12285
    },
    {
      "epoch": 0.7695704599195101,
      "grad_norm": 0.928447367181959,
      "learning_rate": 1.3286026397528745e-06,
      "loss": 0.3583,
      "step": 12286
    },
    {
      "epoch": 0.7696330979188525,
      "grad_norm": 0.7783729181841863,
      "learning_rate": 1.3279140935228919e-06,
      "loss": 0.3603,
      "step": 12287
    },
    {
      "epoch": 0.7696957359181947,
      "grad_norm": 0.8401157053524564,
      "learning_rate": 1.3272256984366472e-06,
      "loss": 0.3871,
      "step": 12288
    },
    {
      "epoch": 0.7697583739175371,
      "grad_norm": 0.9408392144149074,
      "learning_rate": 1.3265374545224758e-06,
      "loss": 0.4105,
      "step": 12289
    },
    {
      "epoch": 0.7698210119168793,
      "grad_norm": 0.7957814527438397,
      "learning_rate": 1.3258493618087044e-06,
      "loss": 0.3846,
      "step": 12290
    },
    {
      "epoch": 0.7698836499162217,
      "grad_norm": 0.8623089413119133,
      "learning_rate": 1.3251614203236552e-06,
      "loss": 0.3857,
      "step": 12291
    },
    {
      "epoch": 0.769946287915564,
      "grad_norm": 0.8903758053659436,
      "learning_rate": 1.324473630095644e-06,
      "loss": 0.3831,
      "step": 12292
    },
    {
      "epoch": 0.7700089259149063,
      "grad_norm": 0.8506720586775667,
      "learning_rate": 1.3237859911529805e-06,
      "loss": 0.3791,
      "step": 12293
    },
    {
      "epoch": 0.7700715639142486,
      "grad_norm": 0.837274298096631,
      "learning_rate": 1.3230985035239679e-06,
      "loss": 0.3633,
      "step": 12294
    },
    {
      "epoch": 0.7701342019135908,
      "grad_norm": 0.87591629870011,
      "learning_rate": 1.3224111672369022e-06,
      "loss": 0.3593,
      "step": 12295
    },
    {
      "epoch": 0.7701968399129332,
      "grad_norm": 0.5899016869889776,
      "learning_rate": 1.3217239823200768e-06,
      "loss": 0.4531,
      "step": 12296
    },
    {
      "epoch": 0.7702594779122754,
      "grad_norm": 0.8636842496720666,
      "learning_rate": 1.3210369488017731e-06,
      "loss": 0.3889,
      "step": 12297
    },
    {
      "epoch": 0.7703221159116178,
      "grad_norm": 0.9140629392528077,
      "learning_rate": 1.3203500667102731e-06,
      "loss": 0.3943,
      "step": 12298
    },
    {
      "epoch": 0.7703847539109601,
      "grad_norm": 0.8680987128716712,
      "learning_rate": 1.3196633360738475e-06,
      "loss": 0.399,
      "step": 12299
    },
    {
      "epoch": 0.7704473919103024,
      "grad_norm": 0.8754438066877356,
      "learning_rate": 1.3189767569207622e-06,
      "loss": 0.3639,
      "step": 12300
    },
    {
      "epoch": 0.7705100299096447,
      "grad_norm": 0.9101517533593068,
      "learning_rate": 1.3182903292792764e-06,
      "loss": 0.4255,
      "step": 12301
    },
    {
      "epoch": 0.770572667908987,
      "grad_norm": 0.864852217140999,
      "learning_rate": 1.3176040531776447e-06,
      "loss": 0.3706,
      "step": 12302
    },
    {
      "epoch": 0.7706353059083293,
      "grad_norm": 1.001305764890246,
      "learning_rate": 1.3169179286441131e-06,
      "loss": 0.3807,
      "step": 12303
    },
    {
      "epoch": 0.7706979439076715,
      "grad_norm": 0.8725198174029146,
      "learning_rate": 1.316231955706922e-06,
      "loss": 0.3951,
      "step": 12304
    },
    {
      "epoch": 0.7707605819070139,
      "grad_norm": 0.832433413801481,
      "learning_rate": 1.3155461343943065e-06,
      "loss": 0.3573,
      "step": 12305
    },
    {
      "epoch": 0.7708232199063562,
      "grad_norm": 0.8701916708162051,
      "learning_rate": 1.3148604647344987e-06,
      "loss": 0.4212,
      "step": 12306
    },
    {
      "epoch": 0.7708858579056985,
      "grad_norm": 0.9087456117006217,
      "learning_rate": 1.3141749467557175e-06,
      "loss": 0.3535,
      "step": 12307
    },
    {
      "epoch": 0.7709484959050408,
      "grad_norm": 0.9161933143618876,
      "learning_rate": 1.3134895804861797e-06,
      "loss": 0.4136,
      "step": 12308
    },
    {
      "epoch": 0.7710111339043831,
      "grad_norm": 0.848254235899191,
      "learning_rate": 1.3128043659540945e-06,
      "loss": 0.3583,
      "step": 12309
    },
    {
      "epoch": 0.7710737719037254,
      "grad_norm": 0.9801474998198765,
      "learning_rate": 1.3121193031876666e-06,
      "loss": 0.4052,
      "step": 12310
    },
    {
      "epoch": 0.7711364099030676,
      "grad_norm": 0.8843363150764962,
      "learning_rate": 1.3114343922150929e-06,
      "loss": 0.4011,
      "step": 12311
    },
    {
      "epoch": 0.77119904790241,
      "grad_norm": 0.7566068904697433,
      "learning_rate": 1.3107496330645636e-06,
      "loss": 0.3538,
      "step": 12312
    },
    {
      "epoch": 0.7712616859017523,
      "grad_norm": 0.8735747913101004,
      "learning_rate": 1.3100650257642633e-06,
      "loss": 0.4076,
      "step": 12313
    },
    {
      "epoch": 0.7713243239010946,
      "grad_norm": 0.8381689522462522,
      "learning_rate": 1.3093805703423707e-06,
      "loss": 0.3641,
      "step": 12314
    },
    {
      "epoch": 0.7713869619004369,
      "grad_norm": 0.8875881338787945,
      "learning_rate": 1.3086962668270608e-06,
      "loss": 0.391,
      "step": 12315
    },
    {
      "epoch": 0.7714495998997792,
      "grad_norm": 0.8689477692038108,
      "learning_rate": 1.3080121152464976e-06,
      "loss": 0.3983,
      "step": 12316
    },
    {
      "epoch": 0.7715122378991215,
      "grad_norm": 0.805034511545749,
      "learning_rate": 1.3073281156288408e-06,
      "loss": 0.3687,
      "step": 12317
    },
    {
      "epoch": 0.7715748758984639,
      "grad_norm": 0.8496713211669047,
      "learning_rate": 1.3066442680022446e-06,
      "loss": 0.3606,
      "step": 12318
    },
    {
      "epoch": 0.7716375138978061,
      "grad_norm": 0.8219621642146782,
      "learning_rate": 1.3059605723948554e-06,
      "loss": 0.4005,
      "step": 12319
    },
    {
      "epoch": 0.7717001518971484,
      "grad_norm": 0.8160582202967481,
      "learning_rate": 1.3052770288348155e-06,
      "loss": 0.4053,
      "step": 12320
    },
    {
      "epoch": 0.7717627898964907,
      "grad_norm": 0.8838637168407966,
      "learning_rate": 1.3045936373502567e-06,
      "loss": 0.3667,
      "step": 12321
    },
    {
      "epoch": 0.771825427895833,
      "grad_norm": 0.8842355824779993,
      "learning_rate": 1.3039103979693107e-06,
      "loss": 0.3854,
      "step": 12322
    },
    {
      "epoch": 0.7718880658951753,
      "grad_norm": 0.7892414844447273,
      "learning_rate": 1.3032273107200983e-06,
      "loss": 0.3733,
      "step": 12323
    },
    {
      "epoch": 0.7719507038945176,
      "grad_norm": 0.760181252062533,
      "learning_rate": 1.3025443756307365e-06,
      "loss": 0.3492,
      "step": 12324
    },
    {
      "epoch": 0.77201334189386,
      "grad_norm": 0.8329584399273702,
      "learning_rate": 1.3018615927293354e-06,
      "loss": 0.3807,
      "step": 12325
    },
    {
      "epoch": 0.7720759798932022,
      "grad_norm": 0.842647905698077,
      "learning_rate": 1.3011789620439969e-06,
      "loss": 0.3462,
      "step": 12326
    },
    {
      "epoch": 0.7721386178925446,
      "grad_norm": 0.9079725623156318,
      "learning_rate": 1.3004964836028189e-06,
      "loss": 0.4182,
      "step": 12327
    },
    {
      "epoch": 0.7722012558918868,
      "grad_norm": 0.9048109028644575,
      "learning_rate": 1.2998141574338925e-06,
      "loss": 0.3853,
      "step": 12328
    },
    {
      "epoch": 0.7722638938912291,
      "grad_norm": 0.8214420143148367,
      "learning_rate": 1.2991319835653022e-06,
      "loss": 0.3993,
      "step": 12329
    },
    {
      "epoch": 0.7723265318905714,
      "grad_norm": 0.6246362478516573,
      "learning_rate": 1.2984499620251246e-06,
      "loss": 0.4571,
      "step": 12330
    },
    {
      "epoch": 0.7723891698899137,
      "grad_norm": 0.8192057052510013,
      "learning_rate": 1.2977680928414355e-06,
      "loss": 0.352,
      "step": 12331
    },
    {
      "epoch": 0.772451807889256,
      "grad_norm": 0.9625354291327105,
      "learning_rate": 1.2970863760422987e-06,
      "loss": 0.4036,
      "step": 12332
    },
    {
      "epoch": 0.7725144458885983,
      "grad_norm": 0.9337835973231406,
      "learning_rate": 1.296404811655772e-06,
      "loss": 0.4218,
      "step": 12333
    },
    {
      "epoch": 0.7725770838879407,
      "grad_norm": 0.88297485657945,
      "learning_rate": 1.2957233997099127e-06,
      "loss": 0.3974,
      "step": 12334
    },
    {
      "epoch": 0.7726397218872829,
      "grad_norm": 0.91373687410806,
      "learning_rate": 1.2950421402327656e-06,
      "loss": 0.3633,
      "step": 12335
    },
    {
      "epoch": 0.7727023598866252,
      "grad_norm": 0.8160823646348158,
      "learning_rate": 1.294361033252372e-06,
      "loss": 0.3998,
      "step": 12336
    },
    {
      "epoch": 0.7727649978859675,
      "grad_norm": 0.8190134024867194,
      "learning_rate": 1.2936800787967657e-06,
      "loss": 0.3736,
      "step": 12337
    },
    {
      "epoch": 0.7728276358853098,
      "grad_norm": 0.8278311124249005,
      "learning_rate": 1.2929992768939754e-06,
      "loss": 0.3874,
      "step": 12338
    },
    {
      "epoch": 0.7728902738846521,
      "grad_norm": 0.8257325599841278,
      "learning_rate": 1.2923186275720213e-06,
      "loss": 0.3699,
      "step": 12339
    },
    {
      "epoch": 0.7729529118839944,
      "grad_norm": 0.883432868743919,
      "learning_rate": 1.2916381308589226e-06,
      "loss": 0.3919,
      "step": 12340
    },
    {
      "epoch": 0.7730155498833368,
      "grad_norm": 0.5302492083898949,
      "learning_rate": 1.2909577867826862e-06,
      "loss": 0.4561,
      "step": 12341
    },
    {
      "epoch": 0.773078187882679,
      "grad_norm": 0.9380692862493498,
      "learning_rate": 1.2902775953713141e-06,
      "loss": 0.3854,
      "step": 12342
    },
    {
      "epoch": 0.7731408258820214,
      "grad_norm": 0.8466118783088102,
      "learning_rate": 1.2895975566528067e-06,
      "loss": 0.3928,
      "step": 12343
    },
    {
      "epoch": 0.7732034638813636,
      "grad_norm": 0.936118839948978,
      "learning_rate": 1.2889176706551527e-06,
      "loss": 0.3655,
      "step": 12344
    },
    {
      "epoch": 0.7732661018807059,
      "grad_norm": 0.8726356793394074,
      "learning_rate": 1.2882379374063358e-06,
      "loss": 0.4271,
      "step": 12345
    },
    {
      "epoch": 0.7733287398800482,
      "grad_norm": 0.6011967985217059,
      "learning_rate": 1.2875583569343348e-06,
      "loss": 0.4423,
      "step": 12346
    },
    {
      "epoch": 0.7733913778793905,
      "grad_norm": 0.8632061578173503,
      "learning_rate": 1.2868789292671186e-06,
      "loss": 0.3684,
      "step": 12347
    },
    {
      "epoch": 0.7734540158787329,
      "grad_norm": 0.8372987043481332,
      "learning_rate": 1.286199654432657e-06,
      "loss": 0.3761,
      "step": 12348
    },
    {
      "epoch": 0.7735166538780751,
      "grad_norm": 0.8443332059364841,
      "learning_rate": 1.2855205324589065e-06,
      "loss": 0.388,
      "step": 12349
    },
    {
      "epoch": 0.7735792918774175,
      "grad_norm": 0.7884892637451854,
      "learning_rate": 1.284841563373821e-06,
      "loss": 0.3554,
      "step": 12350
    },
    {
      "epoch": 0.7736419298767597,
      "grad_norm": 0.8161245481337203,
      "learning_rate": 1.284162747205344e-06,
      "loss": 0.391,
      "step": 12351
    },
    {
      "epoch": 0.7737045678761021,
      "grad_norm": 0.8653773025894754,
      "learning_rate": 1.28348408398142e-06,
      "loss": 0.3647,
      "step": 12352
    },
    {
      "epoch": 0.7737672058754443,
      "grad_norm": 0.9334148401456415,
      "learning_rate": 1.282805573729981e-06,
      "loss": 0.4278,
      "step": 12353
    },
    {
      "epoch": 0.7738298438747866,
      "grad_norm": 0.9388038489254796,
      "learning_rate": 1.2821272164789544e-06,
      "loss": 0.407,
      "step": 12354
    },
    {
      "epoch": 0.773892481874129,
      "grad_norm": 0.8202349285183792,
      "learning_rate": 1.2814490122562618e-06,
      "loss": 0.3634,
      "step": 12355
    },
    {
      "epoch": 0.7739551198734712,
      "grad_norm": 0.813603308603816,
      "learning_rate": 1.280770961089816e-06,
      "loss": 0.3852,
      "step": 12356
    },
    {
      "epoch": 0.7740177578728136,
      "grad_norm": 0.8696831403845436,
      "learning_rate": 1.2800930630075299e-06,
      "loss": 0.385,
      "step": 12357
    },
    {
      "epoch": 0.7740803958721558,
      "grad_norm": 0.9441015566193206,
      "learning_rate": 1.2794153180373032e-06,
      "loss": 0.4151,
      "step": 12358
    },
    {
      "epoch": 0.7741430338714982,
      "grad_norm": 0.8103937954557542,
      "learning_rate": 1.2787377262070328e-06,
      "loss": 0.3493,
      "step": 12359
    },
    {
      "epoch": 0.7742056718708404,
      "grad_norm": 0.6638519928599349,
      "learning_rate": 1.278060287544607e-06,
      "loss": 0.4831,
      "step": 12360
    },
    {
      "epoch": 0.7742683098701827,
      "grad_norm": 0.8879343144643036,
      "learning_rate": 1.2773830020779116e-06,
      "loss": 0.3763,
      "step": 12361
    },
    {
      "epoch": 0.774330947869525,
      "grad_norm": 0.8464242547894821,
      "learning_rate": 1.2767058698348234e-06,
      "loss": 0.3961,
      "step": 12362
    },
    {
      "epoch": 0.7743935858688673,
      "grad_norm": 0.8761276392227342,
      "learning_rate": 1.2760288908432123e-06,
      "loss": 0.3978,
      "step": 12363
    },
    {
      "epoch": 0.7744562238682097,
      "grad_norm": 0.8401883586864802,
      "learning_rate": 1.275352065130942e-06,
      "loss": 0.3803,
      "step": 12364
    },
    {
      "epoch": 0.7745188618675519,
      "grad_norm": 0.7933253600489356,
      "learning_rate": 1.274675392725873e-06,
      "loss": 0.3436,
      "step": 12365
    },
    {
      "epoch": 0.7745814998668943,
      "grad_norm": 0.6859216186287934,
      "learning_rate": 1.2739988736558568e-06,
      "loss": 0.4552,
      "step": 12366
    },
    {
      "epoch": 0.7746441378662365,
      "grad_norm": 0.8819591333049047,
      "learning_rate": 1.2733225079487387e-06,
      "loss": 0.4201,
      "step": 12367
    },
    {
      "epoch": 0.7747067758655789,
      "grad_norm": 0.8434452877895916,
      "learning_rate": 1.2726462956323575e-06,
      "loss": 0.3752,
      "step": 12368
    },
    {
      "epoch": 0.7747694138649212,
      "grad_norm": 0.8660507402347892,
      "learning_rate": 1.271970236734545e-06,
      "loss": 0.4358,
      "step": 12369
    },
    {
      "epoch": 0.7748320518642634,
      "grad_norm": 0.8849578083307087,
      "learning_rate": 1.271294331283131e-06,
      "loss": 0.37,
      "step": 12370
    },
    {
      "epoch": 0.7748946898636058,
      "grad_norm": 0.8741280966858911,
      "learning_rate": 1.270618579305935e-06,
      "loss": 0.401,
      "step": 12371
    },
    {
      "epoch": 0.774957327862948,
      "grad_norm": 0.8543666373500239,
      "learning_rate": 1.26994298083077e-06,
      "loss": 0.3993,
      "step": 12372
    },
    {
      "epoch": 0.7750199658622904,
      "grad_norm": 0.8854738056838369,
      "learning_rate": 1.2692675358854429e-06,
      "loss": 0.414,
      "step": 12373
    },
    {
      "epoch": 0.7750826038616326,
      "grad_norm": 0.9524606046843588,
      "learning_rate": 1.2685922444977577e-06,
      "loss": 0.411,
      "step": 12374
    },
    {
      "epoch": 0.775145241860975,
      "grad_norm": 0.9005337605623157,
      "learning_rate": 1.2679171066955087e-06,
      "loss": 0.4087,
      "step": 12375
    },
    {
      "epoch": 0.7752078798603173,
      "grad_norm": 0.8109933223322892,
      "learning_rate": 1.267242122506484e-06,
      "loss": 0.3442,
      "step": 12376
    },
    {
      "epoch": 0.7752705178596596,
      "grad_norm": 0.8745178836179539,
      "learning_rate": 1.2665672919584666e-06,
      "loss": 0.4255,
      "step": 12377
    },
    {
      "epoch": 0.7753331558590019,
      "grad_norm": 0.8869850763163398,
      "learning_rate": 1.2658926150792321e-06,
      "loss": 0.3737,
      "step": 12378
    },
    {
      "epoch": 0.7753957938583441,
      "grad_norm": 0.7613141643678695,
      "learning_rate": 1.2652180918965495e-06,
      "loss": 0.3462,
      "step": 12379
    },
    {
      "epoch": 0.7754584318576865,
      "grad_norm": 0.6272199561690032,
      "learning_rate": 1.2645437224381847e-06,
      "loss": 0.4403,
      "step": 12380
    },
    {
      "epoch": 0.7755210698570287,
      "grad_norm": 0.8515664859748265,
      "learning_rate": 1.2638695067318934e-06,
      "loss": 0.3928,
      "step": 12381
    },
    {
      "epoch": 0.7755837078563711,
      "grad_norm": 0.869694533946577,
      "learning_rate": 1.2631954448054257e-06,
      "loss": 0.3871,
      "step": 12382
    },
    {
      "epoch": 0.7756463458557133,
      "grad_norm": 0.8268472115828417,
      "learning_rate": 1.2625215366865284e-06,
      "loss": 0.3648,
      "step": 12383
    },
    {
      "epoch": 0.7757089838550557,
      "grad_norm": 0.8512667264589964,
      "learning_rate": 1.2618477824029378e-06,
      "loss": 0.3827,
      "step": 12384
    },
    {
      "epoch": 0.775771621854398,
      "grad_norm": 0.9044054261520558,
      "learning_rate": 1.261174181982387e-06,
      "loss": 0.4135,
      "step": 12385
    },
    {
      "epoch": 0.7758342598537403,
      "grad_norm": 0.7980711235028576,
      "learning_rate": 1.2605007354526005e-06,
      "loss": 0.387,
      "step": 12386
    },
    {
      "epoch": 0.7758968978530826,
      "grad_norm": 0.8118561406199543,
      "learning_rate": 1.2598274428412981e-06,
      "loss": 0.3578,
      "step": 12387
    },
    {
      "epoch": 0.7759595358524248,
      "grad_norm": 0.8638026019064967,
      "learning_rate": 1.259154304176191e-06,
      "loss": 0.405,
      "step": 12388
    },
    {
      "epoch": 0.7760221738517672,
      "grad_norm": 0.8023203714566688,
      "learning_rate": 1.2584813194849882e-06,
      "loss": 0.3172,
      "step": 12389
    },
    {
      "epoch": 0.7760848118511094,
      "grad_norm": 0.8719130229989496,
      "learning_rate": 1.2578084887953874e-06,
      "loss": 0.356,
      "step": 12390
    },
    {
      "epoch": 0.7761474498504518,
      "grad_norm": 0.859367762281227,
      "learning_rate": 1.2571358121350851e-06,
      "loss": 0.3633,
      "step": 12391
    },
    {
      "epoch": 0.7762100878497941,
      "grad_norm": 0.8214370841659522,
      "learning_rate": 1.2564632895317674e-06,
      "loss": 0.3865,
      "step": 12392
    },
    {
      "epoch": 0.7762727258491364,
      "grad_norm": 0.8133400859355885,
      "learning_rate": 1.2557909210131158e-06,
      "loss": 0.403,
      "step": 12393
    },
    {
      "epoch": 0.7763353638484787,
      "grad_norm": 0.8392181928546578,
      "learning_rate": 1.2551187066068048e-06,
      "loss": 0.362,
      "step": 12394
    },
    {
      "epoch": 0.7763980018478209,
      "grad_norm": 0.8447082933478081,
      "learning_rate": 1.254446646340503e-06,
      "loss": 0.4402,
      "step": 12395
    },
    {
      "epoch": 0.7764606398471633,
      "grad_norm": 0.8895647690088395,
      "learning_rate": 1.253774740241872e-06,
      "loss": 0.3888,
      "step": 12396
    },
    {
      "epoch": 0.7765232778465055,
      "grad_norm": 0.864060825967737,
      "learning_rate": 1.253102988338567e-06,
      "loss": 0.396,
      "step": 12397
    },
    {
      "epoch": 0.7765859158458479,
      "grad_norm": 0.8901097485877715,
      "learning_rate": 1.2524313906582397e-06,
      "loss": 0.3797,
      "step": 12398
    },
    {
      "epoch": 0.7766485538451902,
      "grad_norm": 0.8396445035340923,
      "learning_rate": 1.2517599472285309e-06,
      "loss": 0.375,
      "step": 12399
    },
    {
      "epoch": 0.7767111918445325,
      "grad_norm": 0.9169467201390308,
      "learning_rate": 1.2510886580770793e-06,
      "loss": 0.4272,
      "step": 12400
    },
    {
      "epoch": 0.7767738298438748,
      "grad_norm": 0.8380507412385916,
      "learning_rate": 1.2504175232315147e-06,
      "loss": 0.4052,
      "step": 12401
    },
    {
      "epoch": 0.7768364678432171,
      "grad_norm": 0.7714042704407105,
      "learning_rate": 1.2497465427194604e-06,
      "loss": 0.3751,
      "step": 12402
    },
    {
      "epoch": 0.7768991058425594,
      "grad_norm": 0.8346391511360877,
      "learning_rate": 1.2490757165685351e-06,
      "loss": 0.3652,
      "step": 12403
    },
    {
      "epoch": 0.7769617438419016,
      "grad_norm": 0.8648889488417811,
      "learning_rate": 1.248405044806349e-06,
      "loss": 0.3788,
      "step": 12404
    },
    {
      "epoch": 0.777024381841244,
      "grad_norm": 0.8859567727468758,
      "learning_rate": 1.247734527460508e-06,
      "loss": 0.4079,
      "step": 12405
    },
    {
      "epoch": 0.7770870198405863,
      "grad_norm": 0.7905909282709578,
      "learning_rate": 1.2470641645586084e-06,
      "loss": 0.3639,
      "step": 12406
    },
    {
      "epoch": 0.7771496578399286,
      "grad_norm": 0.8504930693465917,
      "learning_rate": 1.2463939561282462e-06,
      "loss": 0.3982,
      "step": 12407
    },
    {
      "epoch": 0.7772122958392709,
      "grad_norm": 0.8632289838401725,
      "learning_rate": 1.245723902197004e-06,
      "loss": 0.3791,
      "step": 12408
    },
    {
      "epoch": 0.7772749338386132,
      "grad_norm": 0.8621762644423718,
      "learning_rate": 1.245054002792464e-06,
      "loss": 0.3916,
      "step": 12409
    },
    {
      "epoch": 0.7773375718379555,
      "grad_norm": 0.8551682890961082,
      "learning_rate": 1.2443842579421982e-06,
      "loss": 0.3734,
      "step": 12410
    },
    {
      "epoch": 0.7774002098372979,
      "grad_norm": 0.9116861798887316,
      "learning_rate": 1.2437146676737738e-06,
      "loss": 0.4102,
      "step": 12411
    },
    {
      "epoch": 0.7774628478366401,
      "grad_norm": 0.9668186691818756,
      "learning_rate": 1.2430452320147506e-06,
      "loss": 0.3806,
      "step": 12412
    },
    {
      "epoch": 0.7775254858359824,
      "grad_norm": 0.5886965180339016,
      "learning_rate": 1.242375950992683e-06,
      "loss": 0.4268,
      "step": 12413
    },
    {
      "epoch": 0.7775881238353247,
      "grad_norm": 0.8589425557745741,
      "learning_rate": 1.2417068246351183e-06,
      "loss": 0.3914,
      "step": 12414
    },
    {
      "epoch": 0.777650761834667,
      "grad_norm": 0.8808942094376502,
      "learning_rate": 1.2410378529695977e-06,
      "loss": 0.4046,
      "step": 12415
    },
    {
      "epoch": 0.7777133998340093,
      "grad_norm": 0.8732217344504519,
      "learning_rate": 1.2403690360236559e-06,
      "loss": 0.4025,
      "step": 12416
    },
    {
      "epoch": 0.7777760378333516,
      "grad_norm": 0.8320248099709006,
      "learning_rate": 1.2397003738248243e-06,
      "loss": 0.4015,
      "step": 12417
    },
    {
      "epoch": 0.777838675832694,
      "grad_norm": 0.8829633354406548,
      "learning_rate": 1.239031866400623e-06,
      "loss": 0.4172,
      "step": 12418
    },
    {
      "epoch": 0.7779013138320362,
      "grad_norm": 0.8521015985364613,
      "learning_rate": 1.2383635137785683e-06,
      "loss": 0.376,
      "step": 12419
    },
    {
      "epoch": 0.7779639518313785,
      "grad_norm": 1.0258487545038506,
      "learning_rate": 1.2376953159861698e-06,
      "loss": 0.3582,
      "step": 12420
    },
    {
      "epoch": 0.7780265898307208,
      "grad_norm": 0.8721869437818351,
      "learning_rate": 1.2370272730509303e-06,
      "loss": 0.4317,
      "step": 12421
    },
    {
      "epoch": 0.7780892278300631,
      "grad_norm": 0.8742034589601159,
      "learning_rate": 1.2363593850003463e-06,
      "loss": 0.4017,
      "step": 12422
    },
    {
      "epoch": 0.7781518658294054,
      "grad_norm": 0.838142439696835,
      "learning_rate": 1.235691651861909e-06,
      "loss": 0.383,
      "step": 12423
    },
    {
      "epoch": 0.7782145038287477,
      "grad_norm": 0.8735763494739729,
      "learning_rate": 1.2350240736631004e-06,
      "loss": 0.3851,
      "step": 12424
    },
    {
      "epoch": 0.77827714182809,
      "grad_norm": 0.8220990049237745,
      "learning_rate": 1.2343566504313997e-06,
      "loss": 0.3647,
      "step": 12425
    },
    {
      "epoch": 0.7783397798274323,
      "grad_norm": 0.8552996038970379,
      "learning_rate": 1.2336893821942797e-06,
      "loss": 0.3561,
      "step": 12426
    },
    {
      "epoch": 0.7784024178267747,
      "grad_norm": 0.8783479788257794,
      "learning_rate": 1.2330222689792037e-06,
      "loss": 0.4179,
      "step": 12427
    },
    {
      "epoch": 0.7784650558261169,
      "grad_norm": 0.6509871397321537,
      "learning_rate": 1.2323553108136306e-06,
      "loss": 0.4343,
      "step": 12428
    },
    {
      "epoch": 0.7785276938254592,
      "grad_norm": 0.9433165155813781,
      "learning_rate": 1.2316885077250118e-06,
      "loss": 0.4542,
      "step": 12429
    },
    {
      "epoch": 0.7785903318248015,
      "grad_norm": 0.88755951854897,
      "learning_rate": 1.2310218597407936e-06,
      "loss": 0.4183,
      "step": 12430
    },
    {
      "epoch": 0.7786529698241438,
      "grad_norm": 0.8509520942625268,
      "learning_rate": 1.2303553668884156e-06,
      "loss": 0.4038,
      "step": 12431
    },
    {
      "epoch": 0.7787156078234861,
      "grad_norm": 0.8033388484924359,
      "learning_rate": 1.2296890291953084e-06,
      "loss": 0.3926,
      "step": 12432
    },
    {
      "epoch": 0.7787782458228284,
      "grad_norm": 0.8638064312592364,
      "learning_rate": 1.2290228466889021e-06,
      "loss": 0.3816,
      "step": 12433
    },
    {
      "epoch": 0.7788408838221708,
      "grad_norm": 0.8424050368209411,
      "learning_rate": 1.2283568193966144e-06,
      "loss": 0.3801,
      "step": 12434
    },
    {
      "epoch": 0.778903521821513,
      "grad_norm": 0.8666862643615055,
      "learning_rate": 1.227690947345861e-06,
      "loss": 0.3857,
      "step": 12435
    },
    {
      "epoch": 0.7789661598208554,
      "grad_norm": 0.8686736923936453,
      "learning_rate": 1.2270252305640485e-06,
      "loss": 0.4062,
      "step": 12436
    },
    {
      "epoch": 0.7790287978201976,
      "grad_norm": 0.8320222240225894,
      "learning_rate": 1.226359669078578e-06,
      "loss": 0.4106,
      "step": 12437
    },
    {
      "epoch": 0.7790914358195399,
      "grad_norm": 0.8165734461065778,
      "learning_rate": 1.2256942629168434e-06,
      "loss": 0.3627,
      "step": 12438
    },
    {
      "epoch": 0.7791540738188822,
      "grad_norm": 0.8550602469740148,
      "learning_rate": 1.2250290121062342e-06,
      "loss": 0.3549,
      "step": 12439
    },
    {
      "epoch": 0.7792167118182245,
      "grad_norm": 0.8752959533651217,
      "learning_rate": 1.2243639166741311e-06,
      "loss": 0.3931,
      "step": 12440
    },
    {
      "epoch": 0.7792793498175669,
      "grad_norm": 0.8553201404728615,
      "learning_rate": 1.2236989766479085e-06,
      "loss": 0.4212,
      "step": 12441
    },
    {
      "epoch": 0.7793419878169091,
      "grad_norm": 0.8068529722051702,
      "learning_rate": 1.2230341920549383e-06,
      "loss": 0.3362,
      "step": 12442
    },
    {
      "epoch": 0.7794046258162515,
      "grad_norm": 0.8213480335348555,
      "learning_rate": 1.2223695629225807e-06,
      "loss": 0.3575,
      "step": 12443
    },
    {
      "epoch": 0.7794672638155937,
      "grad_norm": 0.6427167741126153,
      "learning_rate": 1.221705089278194e-06,
      "loss": 0.4595,
      "step": 12444
    },
    {
      "epoch": 0.779529901814936,
      "grad_norm": 0.8536325412943101,
      "learning_rate": 1.2210407711491278e-06,
      "loss": 0.3648,
      "step": 12445
    },
    {
      "epoch": 0.7795925398142783,
      "grad_norm": 0.8688561616237563,
      "learning_rate": 1.2203766085627245e-06,
      "loss": 0.4071,
      "step": 12446
    },
    {
      "epoch": 0.7796551778136206,
      "grad_norm": 0.7919945309040384,
      "learning_rate": 1.2197126015463212e-06,
      "loss": 0.3483,
      "step": 12447
    },
    {
      "epoch": 0.779717815812963,
      "grad_norm": 0.8889753382056303,
      "learning_rate": 1.2190487501272497e-06,
      "loss": 0.3962,
      "step": 12448
    },
    {
      "epoch": 0.7797804538123052,
      "grad_norm": 0.8686685973920277,
      "learning_rate": 1.2183850543328313e-06,
      "loss": 0.3653,
      "step": 12449
    },
    {
      "epoch": 0.7798430918116476,
      "grad_norm": 0.9291756812324542,
      "learning_rate": 1.2177215141903875e-06,
      "loss": 0.4342,
      "step": 12450
    },
    {
      "epoch": 0.7799057298109898,
      "grad_norm": 0.8472692897593083,
      "learning_rate": 1.2170581297272282e-06,
      "loss": 0.3879,
      "step": 12451
    },
    {
      "epoch": 0.7799683678103322,
      "grad_norm": 0.9220927086234658,
      "learning_rate": 1.2163949009706572e-06,
      "loss": 0.4131,
      "step": 12452
    },
    {
      "epoch": 0.7800310058096744,
      "grad_norm": 0.7646970546406672,
      "learning_rate": 1.2157318279479758e-06,
      "loss": 0.3551,
      "step": 12453
    },
    {
      "epoch": 0.7800936438090167,
      "grad_norm": 0.8575350989105152,
      "learning_rate": 1.2150689106864744e-06,
      "loss": 0.3873,
      "step": 12454
    },
    {
      "epoch": 0.780156281808359,
      "grad_norm": 0.8089179409442001,
      "learning_rate": 1.2144061492134395e-06,
      "loss": 0.4001,
      "step": 12455
    },
    {
      "epoch": 0.7802189198077013,
      "grad_norm": 0.8945672937818295,
      "learning_rate": 1.2137435435561502e-06,
      "loss": 0.3955,
      "step": 12456
    },
    {
      "epoch": 0.7802815578070437,
      "grad_norm": 0.8832264716450391,
      "learning_rate": 1.213081093741879e-06,
      "loss": 0.3741,
      "step": 12457
    },
    {
      "epoch": 0.7803441958063859,
      "grad_norm": 0.8712700433687108,
      "learning_rate": 1.2124187997978914e-06,
      "loss": 0.3928,
      "step": 12458
    },
    {
      "epoch": 0.7804068338057283,
      "grad_norm": 0.8635938283574096,
      "learning_rate": 1.211756661751451e-06,
      "loss": 0.38,
      "step": 12459
    },
    {
      "epoch": 0.7804694718050705,
      "grad_norm": 0.8597971984703143,
      "learning_rate": 1.211094679629809e-06,
      "loss": 0.3861,
      "step": 12460
    },
    {
      "epoch": 0.7805321098044129,
      "grad_norm": 0.8709805743478284,
      "learning_rate": 1.2104328534602133e-06,
      "loss": 0.3987,
      "step": 12461
    },
    {
      "epoch": 0.7805947478037552,
      "grad_norm": 0.8923179153279734,
      "learning_rate": 1.2097711832699039e-06,
      "loss": 0.4191,
      "step": 12462
    },
    {
      "epoch": 0.7806573858030974,
      "grad_norm": 0.8841407174795513,
      "learning_rate": 1.209109669086117e-06,
      "loss": 0.4055,
      "step": 12463
    },
    {
      "epoch": 0.7807200238024398,
      "grad_norm": 0.649097438478201,
      "learning_rate": 1.2084483109360795e-06,
      "loss": 0.4617,
      "step": 12464
    },
    {
      "epoch": 0.780782661801782,
      "grad_norm": 0.8203171901401434,
      "learning_rate": 1.2077871088470138e-06,
      "loss": 0.3729,
      "step": 12465
    },
    {
      "epoch": 0.7808452998011244,
      "grad_norm": 0.8754471972946111,
      "learning_rate": 1.2071260628461346e-06,
      "loss": 0.3971,
      "step": 12466
    },
    {
      "epoch": 0.7809079378004666,
      "grad_norm": 0.6035617523868719,
      "learning_rate": 1.2064651729606485e-06,
      "loss": 0.4664,
      "step": 12467
    },
    {
      "epoch": 0.780970575799809,
      "grad_norm": 0.8819394083496126,
      "learning_rate": 1.205804439217762e-06,
      "loss": 0.4063,
      "step": 12468
    },
    {
      "epoch": 0.7810332137991512,
      "grad_norm": 0.8239023986824715,
      "learning_rate": 1.2051438616446692e-06,
      "loss": 0.3851,
      "step": 12469
    },
    {
      "epoch": 0.7810958517984935,
      "grad_norm": 0.8408721745393352,
      "learning_rate": 1.204483440268559e-06,
      "loss": 0.3728,
      "step": 12470
    },
    {
      "epoch": 0.7811584897978359,
      "grad_norm": 0.819197871647985,
      "learning_rate": 1.2038231751166135e-06,
      "loss": 0.3958,
      "step": 12471
    },
    {
      "epoch": 0.7812211277971781,
      "grad_norm": 0.8699531864584115,
      "learning_rate": 1.203163066216012e-06,
      "loss": 0.4039,
      "step": 12472
    },
    {
      "epoch": 0.7812837657965205,
      "grad_norm": 0.8477735105218411,
      "learning_rate": 1.2025031135939241e-06,
      "loss": 0.3853,
      "step": 12473
    },
    {
      "epoch": 0.7813464037958627,
      "grad_norm": 0.8137540656951348,
      "learning_rate": 1.2018433172775123e-06,
      "loss": 0.3847,
      "step": 12474
    },
    {
      "epoch": 0.7814090417952051,
      "grad_norm": 0.8340649380492705,
      "learning_rate": 1.2011836772939334e-06,
      "loss": 0.3709,
      "step": 12475
    },
    {
      "epoch": 0.7814716797945473,
      "grad_norm": 0.8766719568733403,
      "learning_rate": 1.2005241936703404e-06,
      "loss": 0.3875,
      "step": 12476
    },
    {
      "epoch": 0.7815343177938897,
      "grad_norm": 0.950701853655318,
      "learning_rate": 1.1998648664338773e-06,
      "loss": 0.4164,
      "step": 12477
    },
    {
      "epoch": 0.781596955793232,
      "grad_norm": 0.834688152608824,
      "learning_rate": 1.1992056956116816e-06,
      "loss": 0.3837,
      "step": 12478
    },
    {
      "epoch": 0.7816595937925742,
      "grad_norm": 0.8387132246295391,
      "learning_rate": 1.1985466812308843e-06,
      "loss": 0.3852,
      "step": 12479
    },
    {
      "epoch": 0.7817222317919166,
      "grad_norm": 0.836600491456288,
      "learning_rate": 1.1978878233186104e-06,
      "loss": 0.3844,
      "step": 12480
    },
    {
      "epoch": 0.7817848697912588,
      "grad_norm": 0.9224073533991326,
      "learning_rate": 1.1972291219019804e-06,
      "loss": 0.3952,
      "step": 12481
    },
    {
      "epoch": 0.7818475077906012,
      "grad_norm": 0.8800461935576587,
      "learning_rate": 1.196570577008106e-06,
      "loss": 0.4028,
      "step": 12482
    },
    {
      "epoch": 0.7819101457899434,
      "grad_norm": 0.8824934662934697,
      "learning_rate": 1.195912188664093e-06,
      "loss": 0.4268,
      "step": 12483
    },
    {
      "epoch": 0.7819727837892858,
      "grad_norm": 0.8439125834754762,
      "learning_rate": 1.1952539568970378e-06,
      "loss": 0.4106,
      "step": 12484
    },
    {
      "epoch": 0.7820354217886281,
      "grad_norm": 0.821755625928481,
      "learning_rate": 1.194595881734038e-06,
      "loss": 0.3832,
      "step": 12485
    },
    {
      "epoch": 0.7820980597879704,
      "grad_norm": 0.8623035330065902,
      "learning_rate": 1.193937963202178e-06,
      "loss": 0.3385,
      "step": 12486
    },
    {
      "epoch": 0.7821606977873127,
      "grad_norm": 0.8529232638827211,
      "learning_rate": 1.1932802013285378e-06,
      "loss": 0.4026,
      "step": 12487
    },
    {
      "epoch": 0.7822233357866549,
      "grad_norm": 0.8541063082519852,
      "learning_rate": 1.192622596140191e-06,
      "loss": 0.394,
      "step": 12488
    },
    {
      "epoch": 0.7822859737859973,
      "grad_norm": 0.8172320237653747,
      "learning_rate": 1.1919651476642035e-06,
      "loss": 0.3528,
      "step": 12489
    },
    {
      "epoch": 0.7823486117853395,
      "grad_norm": 0.8793798024649238,
      "learning_rate": 1.1913078559276393e-06,
      "loss": 0.3789,
      "step": 12490
    },
    {
      "epoch": 0.7824112497846819,
      "grad_norm": 0.891468978010751,
      "learning_rate": 1.1906507209575496e-06,
      "loss": 0.3956,
      "step": 12491
    },
    {
      "epoch": 0.7824738877840242,
      "grad_norm": 0.8910957456063036,
      "learning_rate": 1.1899937427809843e-06,
      "loss": 0.3871,
      "step": 12492
    },
    {
      "epoch": 0.7825365257833665,
      "grad_norm": 0.6155558995388807,
      "learning_rate": 1.1893369214249816e-06,
      "loss": 0.4307,
      "step": 12493
    },
    {
      "epoch": 0.7825991637827088,
      "grad_norm": 0.8593005927652628,
      "learning_rate": 1.18868025691658e-06,
      "loss": 0.3757,
      "step": 12494
    },
    {
      "epoch": 0.7826618017820511,
      "grad_norm": 0.9084884415743852,
      "learning_rate": 1.1880237492828067e-06,
      "loss": 0.4122,
      "step": 12495
    },
    {
      "epoch": 0.7827244397813934,
      "grad_norm": 0.7915471775014099,
      "learning_rate": 1.1873673985506834e-06,
      "loss": 0.3788,
      "step": 12496
    },
    {
      "epoch": 0.7827870777807356,
      "grad_norm": 0.8134793068386456,
      "learning_rate": 1.1867112047472258e-06,
      "loss": 0.3744,
      "step": 12497
    },
    {
      "epoch": 0.782849715780078,
      "grad_norm": 0.9382485776280854,
      "learning_rate": 1.1860551678994408e-06,
      "loss": 0.3813,
      "step": 12498
    },
    {
      "epoch": 0.7829123537794203,
      "grad_norm": 0.85964889984531,
      "learning_rate": 1.1853992880343346e-06,
      "loss": 0.4223,
      "step": 12499
    },
    {
      "epoch": 0.7829749917787626,
      "grad_norm": 0.8607324581181559,
      "learning_rate": 1.1847435651789018e-06,
      "loss": 0.3756,
      "step": 12500
    },
    {
      "epoch": 0.7830376297781049,
      "grad_norm": 0.8621726120912049,
      "learning_rate": 1.1840879993601307e-06,
      "loss": 0.3838,
      "step": 12501
    },
    {
      "epoch": 0.7831002677774472,
      "grad_norm": 0.8507022633390143,
      "learning_rate": 1.183432590605007e-06,
      "loss": 0.3419,
      "step": 12502
    },
    {
      "epoch": 0.7831629057767895,
      "grad_norm": 0.8391498001211138,
      "learning_rate": 1.1827773389405067e-06,
      "loss": 0.3771,
      "step": 12503
    },
    {
      "epoch": 0.7832255437761317,
      "grad_norm": 0.8534570559758352,
      "learning_rate": 1.1821222443935992e-06,
      "loss": 0.3479,
      "step": 12504
    },
    {
      "epoch": 0.7832881817754741,
      "grad_norm": 0.836627575682535,
      "learning_rate": 1.181467306991249e-06,
      "loss": 0.3563,
      "step": 12505
    },
    {
      "epoch": 0.7833508197748164,
      "grad_norm": 0.850158505494499,
      "learning_rate": 1.1808125267604131e-06,
      "loss": 0.3563,
      "step": 12506
    },
    {
      "epoch": 0.7834134577741587,
      "grad_norm": 0.9102770451913557,
      "learning_rate": 1.1801579037280424e-06,
      "loss": 0.3952,
      "step": 12507
    },
    {
      "epoch": 0.783476095773501,
      "grad_norm": 0.9587042696770496,
      "learning_rate": 1.1795034379210801e-06,
      "loss": 0.4219,
      "step": 12508
    },
    {
      "epoch": 0.7835387337728433,
      "grad_norm": 0.8754830031479091,
      "learning_rate": 1.1788491293664668e-06,
      "loss": 0.3696,
      "step": 12509
    },
    {
      "epoch": 0.7836013717721856,
      "grad_norm": 0.8697379608864397,
      "learning_rate": 1.1781949780911312e-06,
      "loss": 0.3955,
      "step": 12510
    },
    {
      "epoch": 0.783664009771528,
      "grad_norm": 0.8332060867778333,
      "learning_rate": 1.1775409841220009e-06,
      "loss": 0.3801,
      "step": 12511
    },
    {
      "epoch": 0.7837266477708702,
      "grad_norm": 0.8471572585858689,
      "learning_rate": 1.1768871474859934e-06,
      "loss": 0.4051,
      "step": 12512
    },
    {
      "epoch": 0.7837892857702125,
      "grad_norm": 0.9334339178257178,
      "learning_rate": 1.1762334682100203e-06,
      "loss": 0.4019,
      "step": 12513
    },
    {
      "epoch": 0.7838519237695548,
      "grad_norm": 0.8653171281415826,
      "learning_rate": 1.1755799463209877e-06,
      "loss": 0.3583,
      "step": 12514
    },
    {
      "epoch": 0.7839145617688971,
      "grad_norm": 0.7601389332828199,
      "learning_rate": 1.1749265818457939e-06,
      "loss": 0.3852,
      "step": 12515
    },
    {
      "epoch": 0.7839771997682394,
      "grad_norm": 0.789613967091774,
      "learning_rate": 1.174273374811332e-06,
      "loss": 0.3471,
      "step": 12516
    },
    {
      "epoch": 0.7840398377675817,
      "grad_norm": 0.8938031222906299,
      "learning_rate": 1.1736203252444872e-06,
      "loss": 0.4242,
      "step": 12517
    },
    {
      "epoch": 0.784102475766924,
      "grad_norm": 0.8723272644781694,
      "learning_rate": 1.1729674331721408e-06,
      "loss": 0.3922,
      "step": 12518
    },
    {
      "epoch": 0.7841651137662663,
      "grad_norm": 0.8427473589109223,
      "learning_rate": 1.1723146986211641e-06,
      "loss": 0.3832,
      "step": 12519
    },
    {
      "epoch": 0.7842277517656087,
      "grad_norm": 0.8477525904731877,
      "learning_rate": 1.171662121618426e-06,
      "loss": 0.3801,
      "step": 12520
    },
    {
      "epoch": 0.7842903897649509,
      "grad_norm": 0.8007579109190751,
      "learning_rate": 1.1710097021907863e-06,
      "loss": 0.326,
      "step": 12521
    },
    {
      "epoch": 0.7843530277642932,
      "grad_norm": 0.8141666616828654,
      "learning_rate": 1.1703574403650969e-06,
      "loss": 0.3609,
      "step": 12522
    },
    {
      "epoch": 0.7844156657636355,
      "grad_norm": 0.8110314859729555,
      "learning_rate": 1.169705336168207e-06,
      "loss": 0.3434,
      "step": 12523
    },
    {
      "epoch": 0.7844783037629778,
      "grad_norm": 0.6035203315559744,
      "learning_rate": 1.1690533896269557e-06,
      "loss": 0.4563,
      "step": 12524
    },
    {
      "epoch": 0.7845409417623201,
      "grad_norm": 0.8764179827479062,
      "learning_rate": 1.1684016007681782e-06,
      "loss": 0.3841,
      "step": 12525
    },
    {
      "epoch": 0.7846035797616624,
      "grad_norm": 0.8540981002897962,
      "learning_rate": 1.1677499696187006e-06,
      "loss": 0.3854,
      "step": 12526
    },
    {
      "epoch": 0.7846662177610048,
      "grad_norm": 0.8093702740982959,
      "learning_rate": 1.1670984962053456e-06,
      "loss": 0.3298,
      "step": 12527
    },
    {
      "epoch": 0.784728855760347,
      "grad_norm": 0.8763508667046342,
      "learning_rate": 1.1664471805549294e-06,
      "loss": 0.3659,
      "step": 12528
    },
    {
      "epoch": 0.7847914937596893,
      "grad_norm": 0.9142118776960689,
      "learning_rate": 1.1657960226942588e-06,
      "loss": 0.4067,
      "step": 12529
    },
    {
      "epoch": 0.7848541317590316,
      "grad_norm": 0.9178206139921472,
      "learning_rate": 1.1651450226501354e-06,
      "loss": 0.4157,
      "step": 12530
    },
    {
      "epoch": 0.7849167697583739,
      "grad_norm": 0.8511087880073451,
      "learning_rate": 1.1644941804493549e-06,
      "loss": 0.4155,
      "step": 12531
    },
    {
      "epoch": 0.7849794077577162,
      "grad_norm": 0.8791294288468114,
      "learning_rate": 1.163843496118705e-06,
      "loss": 0.4157,
      "step": 12532
    },
    {
      "epoch": 0.7850420457570585,
      "grad_norm": 0.7652534448291759,
      "learning_rate": 1.1631929696849697e-06,
      "loss": 0.3476,
      "step": 12533
    },
    {
      "epoch": 0.7851046837564009,
      "grad_norm": 0.8484990863994181,
      "learning_rate": 1.1625426011749224e-06,
      "loss": 0.394,
      "step": 12534
    },
    {
      "epoch": 0.7851673217557431,
      "grad_norm": 0.8330295214398087,
      "learning_rate": 1.161892390615335e-06,
      "loss": 0.4016,
      "step": 12535
    },
    {
      "epoch": 0.7852299597550855,
      "grad_norm": 0.9068729175260264,
      "learning_rate": 1.1612423380329678e-06,
      "loss": 0.4325,
      "step": 12536
    },
    {
      "epoch": 0.7852925977544277,
      "grad_norm": 0.908136726302521,
      "learning_rate": 1.16059244345458e-06,
      "loss": 0.3567,
      "step": 12537
    },
    {
      "epoch": 0.78535523575377,
      "grad_norm": 0.8699261599807185,
      "learning_rate": 1.1599427069069203e-06,
      "loss": 0.4127,
      "step": 12538
    },
    {
      "epoch": 0.7854178737531123,
      "grad_norm": 0.9623097158692404,
      "learning_rate": 1.1592931284167308e-06,
      "loss": 0.3789,
      "step": 12539
    },
    {
      "epoch": 0.7854805117524546,
      "grad_norm": 0.8626034928005188,
      "learning_rate": 1.1586437080107492e-06,
      "loss": 0.3802,
      "step": 12540
    },
    {
      "epoch": 0.785543149751797,
      "grad_norm": 0.8915173803094317,
      "learning_rate": 1.157994445715706e-06,
      "loss": 0.4068,
      "step": 12541
    },
    {
      "epoch": 0.7856057877511392,
      "grad_norm": 0.9344316638622888,
      "learning_rate": 1.157345341558324e-06,
      "loss": 0.3734,
      "step": 12542
    },
    {
      "epoch": 0.7856684257504816,
      "grad_norm": 0.7999672358641646,
      "learning_rate": 1.1566963955653198e-06,
      "loss": 0.3819,
      "step": 12543
    },
    {
      "epoch": 0.7857310637498238,
      "grad_norm": 0.8675701705065052,
      "learning_rate": 1.156047607763407e-06,
      "loss": 0.3668,
      "step": 12544
    },
    {
      "epoch": 0.7857937017491662,
      "grad_norm": 0.9980498018666933,
      "learning_rate": 1.1553989781792863e-06,
      "loss": 0.4194,
      "step": 12545
    },
    {
      "epoch": 0.7858563397485084,
      "grad_norm": 0.8503627040517849,
      "learning_rate": 1.1547505068396586e-06,
      "loss": 0.3792,
      "step": 12546
    },
    {
      "epoch": 0.7859189777478507,
      "grad_norm": 0.8941108668815011,
      "learning_rate": 1.1541021937712139e-06,
      "loss": 0.378,
      "step": 12547
    },
    {
      "epoch": 0.785981615747193,
      "grad_norm": 0.8288627666395484,
      "learning_rate": 1.153454039000637e-06,
      "loss": 0.3863,
      "step": 12548
    },
    {
      "epoch": 0.7860442537465353,
      "grad_norm": 0.8930496665814402,
      "learning_rate": 1.1528060425546055e-06,
      "loss": 0.3555,
      "step": 12549
    },
    {
      "epoch": 0.7861068917458777,
      "grad_norm": 0.9586383250394457,
      "learning_rate": 1.1521582044597913e-06,
      "loss": 0.3987,
      "step": 12550
    },
    {
      "epoch": 0.7861695297452199,
      "grad_norm": 0.9251726740107891,
      "learning_rate": 1.15151052474286e-06,
      "loss": 0.3975,
      "step": 12551
    },
    {
      "epoch": 0.7862321677445623,
      "grad_norm": 0.7760825067012649,
      "learning_rate": 1.1508630034304679e-06,
      "loss": 0.3253,
      "step": 12552
    },
    {
      "epoch": 0.7862948057439045,
      "grad_norm": 0.9326463251964879,
      "learning_rate": 1.150215640549271e-06,
      "loss": 0.4276,
      "step": 12553
    },
    {
      "epoch": 0.7863574437432468,
      "grad_norm": 0.8467911816758141,
      "learning_rate": 1.1495684361259106e-06,
      "loss": 0.3598,
      "step": 12554
    },
    {
      "epoch": 0.7864200817425891,
      "grad_norm": 0.8850768907093821,
      "learning_rate": 1.1489213901870306e-06,
      "loss": 0.3702,
      "step": 12555
    },
    {
      "epoch": 0.7864827197419314,
      "grad_norm": 0.8394291562691212,
      "learning_rate": 1.1482745027592601e-06,
      "loss": 0.3661,
      "step": 12556
    },
    {
      "epoch": 0.7865453577412738,
      "grad_norm": 0.8974553824856579,
      "learning_rate": 1.147627773869226e-06,
      "loss": 0.4004,
      "step": 12557
    },
    {
      "epoch": 0.786607995740616,
      "grad_norm": 0.9747020881095688,
      "learning_rate": 1.146981203543548e-06,
      "loss": 0.3862,
      "step": 12558
    },
    {
      "epoch": 0.7866706337399584,
      "grad_norm": 0.7948360944864055,
      "learning_rate": 1.1463347918088392e-06,
      "loss": 0.3311,
      "step": 12559
    },
    {
      "epoch": 0.7867332717393006,
      "grad_norm": 0.8013871823986165,
      "learning_rate": 1.1456885386917039e-06,
      "loss": 0.342,
      "step": 12560
    },
    {
      "epoch": 0.786795909738643,
      "grad_norm": 0.7904569508281576,
      "learning_rate": 1.1450424442187446e-06,
      "loss": 0.3585,
      "step": 12561
    },
    {
      "epoch": 0.7868585477379852,
      "grad_norm": 0.9295286746258085,
      "learning_rate": 1.1443965084165548e-06,
      "loss": 0.3997,
      "step": 12562
    },
    {
      "epoch": 0.7869211857373275,
      "grad_norm": 0.7859921129280641,
      "learning_rate": 1.143750731311718e-06,
      "loss": 0.3437,
      "step": 12563
    },
    {
      "epoch": 0.7869838237366699,
      "grad_norm": 0.9220150079920033,
      "learning_rate": 1.1431051129308191e-06,
      "loss": 0.4354,
      "step": 12564
    },
    {
      "epoch": 0.7870464617360121,
      "grad_norm": 0.8265061932800637,
      "learning_rate": 1.142459653300429e-06,
      "loss": 0.385,
      "step": 12565
    },
    {
      "epoch": 0.7871090997353545,
      "grad_norm": 0.8259562613414685,
      "learning_rate": 1.1418143524471159e-06,
      "loss": 0.395,
      "step": 12566
    },
    {
      "epoch": 0.7871717377346967,
      "grad_norm": 0.8922418850977815,
      "learning_rate": 1.1411692103974398e-06,
      "loss": 0.4167,
      "step": 12567
    },
    {
      "epoch": 0.7872343757340391,
      "grad_norm": 0.9220441100046687,
      "learning_rate": 1.140524227177956e-06,
      "loss": 0.4177,
      "step": 12568
    },
    {
      "epoch": 0.7872970137333813,
      "grad_norm": 0.7980269045755313,
      "learning_rate": 1.1398794028152094e-06,
      "loss": 0.3976,
      "step": 12569
    },
    {
      "epoch": 0.7873596517327237,
      "grad_norm": 1.0131259592108715,
      "learning_rate": 1.1392347373357443e-06,
      "loss": 0.4288,
      "step": 12570
    },
    {
      "epoch": 0.787422289732066,
      "grad_norm": 0.8067432752867904,
      "learning_rate": 1.138590230766094e-06,
      "loss": 0.358,
      "step": 12571
    },
    {
      "epoch": 0.7874849277314082,
      "grad_norm": 0.839617901168239,
      "learning_rate": 1.1379458831327856e-06,
      "loss": 0.3895,
      "step": 12572
    },
    {
      "epoch": 0.7875475657307506,
      "grad_norm": 0.8219542803675538,
      "learning_rate": 1.1373016944623427e-06,
      "loss": 0.3365,
      "step": 12573
    },
    {
      "epoch": 0.7876102037300928,
      "grad_norm": 0.8806636853603678,
      "learning_rate": 1.1366576647812788e-06,
      "loss": 0.3668,
      "step": 12574
    },
    {
      "epoch": 0.7876728417294352,
      "grad_norm": 0.8226454022683106,
      "learning_rate": 1.1360137941161027e-06,
      "loss": 0.3973,
      "step": 12575
    },
    {
      "epoch": 0.7877354797287774,
      "grad_norm": 0.959395050021133,
      "learning_rate": 1.1353700824933157e-06,
      "loss": 0.4064,
      "step": 12576
    },
    {
      "epoch": 0.7877981177281198,
      "grad_norm": 0.9166673214754757,
      "learning_rate": 1.1347265299394139e-06,
      "loss": 0.4014,
      "step": 12577
    },
    {
      "epoch": 0.7878607557274621,
      "grad_norm": 0.9729915466882186,
      "learning_rate": 1.1340831364808836e-06,
      "loss": 0.4254,
      "step": 12578
    },
    {
      "epoch": 0.7879233937268043,
      "grad_norm": 0.8089382619341868,
      "learning_rate": 1.133439902144211e-06,
      "loss": 0.3267,
      "step": 12579
    },
    {
      "epoch": 0.7879860317261467,
      "grad_norm": 0.8313768137378706,
      "learning_rate": 1.1327968269558692e-06,
      "loss": 0.3885,
      "step": 12580
    },
    {
      "epoch": 0.7880486697254889,
      "grad_norm": 0.9242032108179415,
      "learning_rate": 1.1321539109423263e-06,
      "loss": 0.4278,
      "step": 12581
    },
    {
      "epoch": 0.7881113077248313,
      "grad_norm": 0.8736085919046747,
      "learning_rate": 1.1315111541300483e-06,
      "loss": 0.4018,
      "step": 12582
    },
    {
      "epoch": 0.7881739457241735,
      "grad_norm": 0.879216381610442,
      "learning_rate": 1.1308685565454885e-06,
      "loss": 0.3933,
      "step": 12583
    },
    {
      "epoch": 0.7882365837235159,
      "grad_norm": 0.8693243379463957,
      "learning_rate": 1.1302261182150975e-06,
      "loss": 0.4112,
      "step": 12584
    },
    {
      "epoch": 0.7882992217228582,
      "grad_norm": 0.9245479163924568,
      "learning_rate": 1.129583839165318e-06,
      "loss": 0.4047,
      "step": 12585
    },
    {
      "epoch": 0.7883618597222005,
      "grad_norm": 0.9313087788020649,
      "learning_rate": 1.1289417194225842e-06,
      "loss": 0.3981,
      "step": 12586
    },
    {
      "epoch": 0.7884244977215428,
      "grad_norm": 0.8578031651168362,
      "learning_rate": 1.1282997590133293e-06,
      "loss": 0.3761,
      "step": 12587
    },
    {
      "epoch": 0.788487135720885,
      "grad_norm": 0.8770181245907679,
      "learning_rate": 1.1276579579639752e-06,
      "loss": 0.3792,
      "step": 12588
    },
    {
      "epoch": 0.7885497737202274,
      "grad_norm": 0.8640261512468043,
      "learning_rate": 1.127016316300938e-06,
      "loss": 0.3571,
      "step": 12589
    },
    {
      "epoch": 0.7886124117195696,
      "grad_norm": 0.8450787518959207,
      "learning_rate": 1.1263748340506264e-06,
      "loss": 0.3984,
      "step": 12590
    },
    {
      "epoch": 0.788675049718912,
      "grad_norm": 0.8863293540293675,
      "learning_rate": 1.1257335112394473e-06,
      "loss": 0.3868,
      "step": 12591
    },
    {
      "epoch": 0.7887376877182543,
      "grad_norm": 0.8399792939544477,
      "learning_rate": 1.1250923478937959e-06,
      "loss": 0.4424,
      "step": 12592
    },
    {
      "epoch": 0.7888003257175966,
      "grad_norm": 0.8192364635506901,
      "learning_rate": 1.1244513440400629e-06,
      "loss": 0.3765,
      "step": 12593
    },
    {
      "epoch": 0.7888629637169389,
      "grad_norm": 0.6182353984142571,
      "learning_rate": 1.1238104997046317e-06,
      "loss": 0.4126,
      "step": 12594
    },
    {
      "epoch": 0.7889256017162812,
      "grad_norm": 0.8809183708342481,
      "learning_rate": 1.1231698149138788e-06,
      "loss": 0.3773,
      "step": 12595
    },
    {
      "epoch": 0.7889882397156235,
      "grad_norm": 0.8520117725858163,
      "learning_rate": 1.1225292896941764e-06,
      "loss": 0.3609,
      "step": 12596
    },
    {
      "epoch": 0.7890508777149657,
      "grad_norm": 0.9269253258679584,
      "learning_rate": 1.1218889240718888e-06,
      "loss": 0.4004,
      "step": 12597
    },
    {
      "epoch": 0.7891135157143081,
      "grad_norm": 0.8451853310401818,
      "learning_rate": 1.1212487180733728e-06,
      "loss": 0.343,
      "step": 12598
    },
    {
      "epoch": 0.7891761537136504,
      "grad_norm": 0.8698372599410954,
      "learning_rate": 1.1206086717249799e-06,
      "loss": 0.4042,
      "step": 12599
    },
    {
      "epoch": 0.7892387917129927,
      "grad_norm": 0.8735373740461891,
      "learning_rate": 1.1199687850530522e-06,
      "loss": 0.3984,
      "step": 12600
    },
    {
      "epoch": 0.789301429712335,
      "grad_norm": 0.8398494756134819,
      "learning_rate": 1.1193290580839306e-06,
      "loss": 0.4091,
      "step": 12601
    },
    {
      "epoch": 0.7893640677116773,
      "grad_norm": 0.597634898509433,
      "learning_rate": 1.1186894908439455e-06,
      "loss": 0.4678,
      "step": 12602
    },
    {
      "epoch": 0.7894267057110196,
      "grad_norm": 0.840905532596803,
      "learning_rate": 1.1180500833594215e-06,
      "loss": 0.3912,
      "step": 12603
    },
    {
      "epoch": 0.789489343710362,
      "grad_norm": 0.8710162699311828,
      "learning_rate": 1.1174108356566753e-06,
      "loss": 0.3717,
      "step": 12604
    },
    {
      "epoch": 0.7895519817097042,
      "grad_norm": 0.8385308820675974,
      "learning_rate": 1.116771747762021e-06,
      "loss": 0.3737,
      "step": 12605
    },
    {
      "epoch": 0.7896146197090465,
      "grad_norm": 0.8777767513121948,
      "learning_rate": 1.1161328197017623e-06,
      "loss": 0.4109,
      "step": 12606
    },
    {
      "epoch": 0.7896772577083888,
      "grad_norm": 0.8927841442415565,
      "learning_rate": 1.1154940515021979e-06,
      "loss": 0.3905,
      "step": 12607
    },
    {
      "epoch": 0.7897398957077311,
      "grad_norm": 0.8942886696750962,
      "learning_rate": 1.114855443189619e-06,
      "loss": 0.3459,
      "step": 12608
    },
    {
      "epoch": 0.7898025337070734,
      "grad_norm": 0.8052167519777793,
      "learning_rate": 1.1142169947903098e-06,
      "loss": 0.3408,
      "step": 12609
    },
    {
      "epoch": 0.7898651717064157,
      "grad_norm": 0.859302406979004,
      "learning_rate": 1.113578706330552e-06,
      "loss": 0.3741,
      "step": 12610
    },
    {
      "epoch": 0.789927809705758,
      "grad_norm": 0.8917987913435566,
      "learning_rate": 1.1129405778366158e-06,
      "loss": 0.4072,
      "step": 12611
    },
    {
      "epoch": 0.7899904477051003,
      "grad_norm": 0.8463703405600349,
      "learning_rate": 1.1123026093347656e-06,
      "loss": 0.3638,
      "step": 12612
    },
    {
      "epoch": 0.7900530857044425,
      "grad_norm": 0.8317197832203322,
      "learning_rate": 1.1116648008512627e-06,
      "loss": 0.394,
      "step": 12613
    },
    {
      "epoch": 0.7901157237037849,
      "grad_norm": 0.5599908033273108,
      "learning_rate": 1.1110271524123584e-06,
      "loss": 0.4118,
      "step": 12614
    },
    {
      "epoch": 0.7901783617031272,
      "grad_norm": 0.6207584858572737,
      "learning_rate": 1.1103896640442985e-06,
      "loss": 0.449,
      "step": 12615
    },
    {
      "epoch": 0.7902409997024695,
      "grad_norm": 0.822705127396335,
      "learning_rate": 1.109752335773322e-06,
      "loss": 0.3815,
      "step": 12616
    },
    {
      "epoch": 0.7903036377018118,
      "grad_norm": 0.9334963578160161,
      "learning_rate": 1.1091151676256613e-06,
      "loss": 0.3956,
      "step": 12617
    },
    {
      "epoch": 0.7903662757011541,
      "grad_norm": 0.8770112302657027,
      "learning_rate": 1.1084781596275412e-06,
      "loss": 0.3676,
      "step": 12618
    },
    {
      "epoch": 0.7904289137004964,
      "grad_norm": 0.8490243165927922,
      "learning_rate": 1.1078413118051834e-06,
      "loss": 0.3391,
      "step": 12619
    },
    {
      "epoch": 0.7904915516998388,
      "grad_norm": 0.9090975941819758,
      "learning_rate": 1.1072046241848e-06,
      "loss": 0.4107,
      "step": 12620
    },
    {
      "epoch": 0.790554189699181,
      "grad_norm": 0.9059335009976701,
      "learning_rate": 1.1065680967925952e-06,
      "loss": 0.4311,
      "step": 12621
    },
    {
      "epoch": 0.7906168276985233,
      "grad_norm": 0.8679959157044181,
      "learning_rate": 1.1059317296547716e-06,
      "loss": 0.3827,
      "step": 12622
    },
    {
      "epoch": 0.7906794656978656,
      "grad_norm": 0.790533073789331,
      "learning_rate": 1.105295522797521e-06,
      "loss": 0.3434,
      "step": 12623
    },
    {
      "epoch": 0.7907421036972079,
      "grad_norm": 0.7876080744686105,
      "learning_rate": 1.1046594762470293e-06,
      "loss": 0.4153,
      "step": 12624
    },
    {
      "epoch": 0.7908047416965502,
      "grad_norm": 0.5997415484500698,
      "learning_rate": 1.1040235900294765e-06,
      "loss": 0.452,
      "step": 12625
    },
    {
      "epoch": 0.7908673796958925,
      "grad_norm": 0.7956269416036669,
      "learning_rate": 1.103387864171036e-06,
      "loss": 0.3866,
      "step": 12626
    },
    {
      "epoch": 0.7909300176952349,
      "grad_norm": 0.8538456122174175,
      "learning_rate": 1.1027522986978734e-06,
      "loss": 0.3825,
      "step": 12627
    },
    {
      "epoch": 0.7909926556945771,
      "grad_norm": 0.867656428991646,
      "learning_rate": 1.1021168936361503e-06,
      "loss": 0.3735,
      "step": 12628
    },
    {
      "epoch": 0.7910552936939195,
      "grad_norm": 0.8558457468742733,
      "learning_rate": 1.1014816490120194e-06,
      "loss": 0.3853,
      "step": 12629
    },
    {
      "epoch": 0.7911179316932617,
      "grad_norm": 0.8921446580853288,
      "learning_rate": 1.1008465648516265e-06,
      "loss": 0.3763,
      "step": 12630
    },
    {
      "epoch": 0.791180569692604,
      "grad_norm": 0.825027631348653,
      "learning_rate": 1.1002116411811137e-06,
      "loss": 0.3547,
      "step": 12631
    },
    {
      "epoch": 0.7912432076919463,
      "grad_norm": 0.8900634685150479,
      "learning_rate": 1.0995768780266136e-06,
      "loss": 0.4053,
      "step": 12632
    },
    {
      "epoch": 0.7913058456912886,
      "grad_norm": 0.8096631384442109,
      "learning_rate": 1.0989422754142532e-06,
      "loss": 0.3611,
      "step": 12633
    },
    {
      "epoch": 0.791368483690631,
      "grad_norm": 0.8979519719819447,
      "learning_rate": 1.0983078333701524e-06,
      "loss": 0.4074,
      "step": 12634
    },
    {
      "epoch": 0.7914311216899732,
      "grad_norm": 0.8752208134860319,
      "learning_rate": 1.0976735519204258e-06,
      "loss": 0.4095,
      "step": 12635
    },
    {
      "epoch": 0.7914937596893156,
      "grad_norm": 0.8851557733847465,
      "learning_rate": 1.0970394310911787e-06,
      "loss": 0.3781,
      "step": 12636
    },
    {
      "epoch": 0.7915563976886578,
      "grad_norm": 0.8385253708011943,
      "learning_rate": 1.0964054709085141e-06,
      "loss": 0.4098,
      "step": 12637
    },
    {
      "epoch": 0.7916190356880001,
      "grad_norm": 0.8410445452215256,
      "learning_rate": 1.095771671398524e-06,
      "loss": 0.3716,
      "step": 12638
    },
    {
      "epoch": 0.7916816736873424,
      "grad_norm": 0.8717988631173932,
      "learning_rate": 1.095138032587298e-06,
      "loss": 0.3856,
      "step": 12639
    },
    {
      "epoch": 0.7917443116866847,
      "grad_norm": 0.5863511275229438,
      "learning_rate": 1.0945045545009148e-06,
      "loss": 0.4428,
      "step": 12640
    },
    {
      "epoch": 0.791806949686027,
      "grad_norm": 0.8118897900505496,
      "learning_rate": 1.0938712371654496e-06,
      "loss": 0.3813,
      "step": 12641
    },
    {
      "epoch": 0.7918695876853693,
      "grad_norm": 0.8636006428571997,
      "learning_rate": 1.0932380806069686e-06,
      "loss": 0.3738,
      "step": 12642
    },
    {
      "epoch": 0.7919322256847117,
      "grad_norm": 0.7750494134827396,
      "learning_rate": 1.092605084851534e-06,
      "loss": 0.3581,
      "step": 12643
    },
    {
      "epoch": 0.7919948636840539,
      "grad_norm": 0.785496638956157,
      "learning_rate": 1.091972249925199e-06,
      "loss": 0.3479,
      "step": 12644
    },
    {
      "epoch": 0.7920575016833963,
      "grad_norm": 0.8270713459774287,
      "learning_rate": 1.09133957585401e-06,
      "loss": 0.3946,
      "step": 12645
    },
    {
      "epoch": 0.7921201396827385,
      "grad_norm": 0.9321326471054769,
      "learning_rate": 1.0907070626640116e-06,
      "loss": 0.4163,
      "step": 12646
    },
    {
      "epoch": 0.7921827776820808,
      "grad_norm": 0.8786108051564724,
      "learning_rate": 1.0900747103812342e-06,
      "loss": 0.3923,
      "step": 12647
    },
    {
      "epoch": 0.7922454156814231,
      "grad_norm": 0.806008713832629,
      "learning_rate": 1.0894425190317088e-06,
      "loss": 0.3673,
      "step": 12648
    },
    {
      "epoch": 0.7923080536807654,
      "grad_norm": 0.8976113807181694,
      "learning_rate": 1.0888104886414553e-06,
      "loss": 0.4349,
      "step": 12649
    },
    {
      "epoch": 0.7923706916801078,
      "grad_norm": 0.9270367645834846,
      "learning_rate": 1.0881786192364879e-06,
      "loss": 0.4159,
      "step": 12650
    },
    {
      "epoch": 0.79243332967945,
      "grad_norm": 0.8997108289121477,
      "learning_rate": 1.087546910842815e-06,
      "loss": 0.4275,
      "step": 12651
    },
    {
      "epoch": 0.7924959676787924,
      "grad_norm": 0.5780532063275952,
      "learning_rate": 1.0869153634864371e-06,
      "loss": 0.4639,
      "step": 12652
    },
    {
      "epoch": 0.7925586056781346,
      "grad_norm": 0.8575323407626309,
      "learning_rate": 1.0862839771933491e-06,
      "loss": 0.3554,
      "step": 12653
    },
    {
      "epoch": 0.792621243677477,
      "grad_norm": 0.8311403360871389,
      "learning_rate": 1.0856527519895376e-06,
      "loss": 0.3785,
      "step": 12654
    },
    {
      "epoch": 0.7926838816768192,
      "grad_norm": 0.8189344389762331,
      "learning_rate": 1.0850216879009872e-06,
      "loss": 0.3703,
      "step": 12655
    },
    {
      "epoch": 0.7927465196761615,
      "grad_norm": 0.9087566323591416,
      "learning_rate": 1.0843907849536689e-06,
      "loss": 0.4116,
      "step": 12656
    },
    {
      "epoch": 0.7928091576755039,
      "grad_norm": 0.6130651917232283,
      "learning_rate": 1.083760043173554e-06,
      "loss": 0.4546,
      "step": 12657
    },
    {
      "epoch": 0.7928717956748461,
      "grad_norm": 0.8775239804346403,
      "learning_rate": 1.083129462586603e-06,
      "loss": 0.3936,
      "step": 12658
    },
    {
      "epoch": 0.7929344336741885,
      "grad_norm": 0.8763540023287031,
      "learning_rate": 1.0824990432187704e-06,
      "loss": 0.3861,
      "step": 12659
    },
    {
      "epoch": 0.7929970716735307,
      "grad_norm": 0.9093882743295284,
      "learning_rate": 1.0818687850960035e-06,
      "loss": 0.4083,
      "step": 12660
    },
    {
      "epoch": 0.7930597096728731,
      "grad_norm": 0.836526001878492,
      "learning_rate": 1.0812386882442454e-06,
      "loss": 0.3766,
      "step": 12661
    },
    {
      "epoch": 0.7931223476722153,
      "grad_norm": 0.911648770826269,
      "learning_rate": 1.0806087526894305e-06,
      "loss": 0.3861,
      "step": 12662
    },
    {
      "epoch": 0.7931849856715576,
      "grad_norm": 0.8992772451276052,
      "learning_rate": 1.0799789784574848e-06,
      "loss": 0.4083,
      "step": 12663
    },
    {
      "epoch": 0.7932476236709,
      "grad_norm": 1.0113996450159366,
      "learning_rate": 1.079349365574332e-06,
      "loss": 0.3847,
      "step": 12664
    },
    {
      "epoch": 0.7933102616702422,
      "grad_norm": 0.8475065005162706,
      "learning_rate": 1.0787199140658883e-06,
      "loss": 0.3937,
      "step": 12665
    },
    {
      "epoch": 0.7933728996695846,
      "grad_norm": 0.8205202923840063,
      "learning_rate": 1.0780906239580613e-06,
      "loss": 0.384,
      "step": 12666
    },
    {
      "epoch": 0.7934355376689268,
      "grad_norm": 0.8926234557100142,
      "learning_rate": 1.0774614952767515e-06,
      "loss": 0.4043,
      "step": 12667
    },
    {
      "epoch": 0.7934981756682692,
      "grad_norm": 0.8552732714881277,
      "learning_rate": 1.0768325280478553e-06,
      "loss": 0.381,
      "step": 12668
    },
    {
      "epoch": 0.7935608136676114,
      "grad_norm": 0.7981590099004809,
      "learning_rate": 1.07620372229726e-06,
      "loss": 0.3509,
      "step": 12669
    },
    {
      "epoch": 0.7936234516669538,
      "grad_norm": 0.8442353233515729,
      "learning_rate": 1.0755750780508477e-06,
      "loss": 0.3527,
      "step": 12670
    },
    {
      "epoch": 0.7936860896662961,
      "grad_norm": 0.9265095986092947,
      "learning_rate": 1.0749465953344923e-06,
      "loss": 0.3712,
      "step": 12671
    },
    {
      "epoch": 0.7937487276656383,
      "grad_norm": 0.8188055747797444,
      "learning_rate": 1.0743182741740649e-06,
      "loss": 0.4183,
      "step": 12672
    },
    {
      "epoch": 0.7938113656649807,
      "grad_norm": 0.9183149116240987,
      "learning_rate": 1.073690114595425e-06,
      "loss": 0.4055,
      "step": 12673
    },
    {
      "epoch": 0.7938740036643229,
      "grad_norm": 0.9342348303657986,
      "learning_rate": 1.0730621166244299e-06,
      "loss": 0.4466,
      "step": 12674
    },
    {
      "epoch": 0.7939366416636653,
      "grad_norm": 0.9022680807451213,
      "learning_rate": 1.0724342802869264e-06,
      "loss": 0.4047,
      "step": 12675
    },
    {
      "epoch": 0.7939992796630075,
      "grad_norm": 0.7807491409273818,
      "learning_rate": 1.071806605608758e-06,
      "loss": 0.3186,
      "step": 12676
    },
    {
      "epoch": 0.7940619176623499,
      "grad_norm": 0.8274318545264474,
      "learning_rate": 1.0711790926157578e-06,
      "loss": 0.3814,
      "step": 12677
    },
    {
      "epoch": 0.7941245556616922,
      "grad_norm": 0.9652753259474526,
      "learning_rate": 1.0705517413337557e-06,
      "loss": 0.3958,
      "step": 12678
    },
    {
      "epoch": 0.7941871936610345,
      "grad_norm": 0.8408497737362595,
      "learning_rate": 1.0699245517885736e-06,
      "loss": 0.3942,
      "step": 12679
    },
    {
      "epoch": 0.7942498316603768,
      "grad_norm": 0.8293830315126343,
      "learning_rate": 1.069297524006025e-06,
      "loss": 0.3778,
      "step": 12680
    },
    {
      "epoch": 0.794312469659719,
      "grad_norm": 0.8788901701448908,
      "learning_rate": 1.0686706580119204e-06,
      "loss": 0.426,
      "step": 12681
    },
    {
      "epoch": 0.7943751076590614,
      "grad_norm": 0.8435570552448152,
      "learning_rate": 1.0680439538320609e-06,
      "loss": 0.3606,
      "step": 12682
    },
    {
      "epoch": 0.7944377456584036,
      "grad_norm": 0.8403595460808565,
      "learning_rate": 1.0674174114922431e-06,
      "loss": 0.3648,
      "step": 12683
    },
    {
      "epoch": 0.794500383657746,
      "grad_norm": 0.855074282066039,
      "learning_rate": 1.0667910310182545e-06,
      "loss": 0.3904,
      "step": 12684
    },
    {
      "epoch": 0.7945630216570883,
      "grad_norm": 0.8392856796740557,
      "learning_rate": 1.066164812435877e-06,
      "loss": 0.3768,
      "step": 12685
    },
    {
      "epoch": 0.7946256596564306,
      "grad_norm": 0.8575444780456306,
      "learning_rate": 1.0655387557708856e-06,
      "loss": 0.409,
      "step": 12686
    },
    {
      "epoch": 0.7946882976557729,
      "grad_norm": 0.8846081940541537,
      "learning_rate": 1.06491286104905e-06,
      "loss": 0.3603,
      "step": 12687
    },
    {
      "epoch": 0.7947509356551151,
      "grad_norm": 0.8530025768969287,
      "learning_rate": 1.064287128296131e-06,
      "loss": 0.3767,
      "step": 12688
    },
    {
      "epoch": 0.7948135736544575,
      "grad_norm": 0.7644961199816421,
      "learning_rate": 1.0636615575378834e-06,
      "loss": 0.3359,
      "step": 12689
    },
    {
      "epoch": 0.7948762116537997,
      "grad_norm": 0.8357518956747534,
      "learning_rate": 1.0630361488000574e-06,
      "loss": 0.4086,
      "step": 12690
    },
    {
      "epoch": 0.7949388496531421,
      "grad_norm": 0.8397837417332052,
      "learning_rate": 1.062410902108395e-06,
      "loss": 0.3876,
      "step": 12691
    },
    {
      "epoch": 0.7950014876524844,
      "grad_norm": 0.8433067173441143,
      "learning_rate": 1.0617858174886286e-06,
      "loss": 0.3859,
      "step": 12692
    },
    {
      "epoch": 0.7950641256518267,
      "grad_norm": 1.3014160315739554,
      "learning_rate": 1.061160894966491e-06,
      "loss": 0.379,
      "step": 12693
    },
    {
      "epoch": 0.795126763651169,
      "grad_norm": 0.9073540256244387,
      "learning_rate": 1.0605361345677019e-06,
      "loss": 0.42,
      "step": 12694
    },
    {
      "epoch": 0.7951894016505113,
      "grad_norm": 0.8865999213143042,
      "learning_rate": 1.0599115363179763e-06,
      "loss": 0.3702,
      "step": 12695
    },
    {
      "epoch": 0.7952520396498536,
      "grad_norm": 0.8889367291465299,
      "learning_rate": 1.0592871002430232e-06,
      "loss": 0.4249,
      "step": 12696
    },
    {
      "epoch": 0.7953146776491958,
      "grad_norm": 0.908479775739523,
      "learning_rate": 1.0586628263685434e-06,
      "loss": 0.4182,
      "step": 12697
    },
    {
      "epoch": 0.7953773156485382,
      "grad_norm": 0.9078028969389731,
      "learning_rate": 1.0580387147202341e-06,
      "loss": 0.3626,
      "step": 12698
    },
    {
      "epoch": 0.7954399536478804,
      "grad_norm": 0.8241063587829836,
      "learning_rate": 1.0574147653237838e-06,
      "loss": 0.3557,
      "step": 12699
    },
    {
      "epoch": 0.7955025916472228,
      "grad_norm": 0.8994043923209735,
      "learning_rate": 1.056790978204873e-06,
      "loss": 0.3577,
      "step": 12700
    },
    {
      "epoch": 0.7955652296465651,
      "grad_norm": 0.8892350287015249,
      "learning_rate": 1.0561673533891758e-06,
      "loss": 0.373,
      "step": 12701
    },
    {
      "epoch": 0.7956278676459074,
      "grad_norm": 0.8268516833893215,
      "learning_rate": 1.0555438909023642e-06,
      "loss": 0.3805,
      "step": 12702
    },
    {
      "epoch": 0.7956905056452497,
      "grad_norm": 0.889779483812979,
      "learning_rate": 1.0549205907700982e-06,
      "loss": 0.4234,
      "step": 12703
    },
    {
      "epoch": 0.795753143644592,
      "grad_norm": 0.9099650677779556,
      "learning_rate": 1.0542974530180327e-06,
      "loss": 0.3768,
      "step": 12704
    },
    {
      "epoch": 0.7958157816439343,
      "grad_norm": 0.9334918800545099,
      "learning_rate": 1.0536744776718166e-06,
      "loss": 0.4278,
      "step": 12705
    },
    {
      "epoch": 0.7958784196432765,
      "grad_norm": 0.8521032388026354,
      "learning_rate": 1.0530516647570898e-06,
      "loss": 0.3968,
      "step": 12706
    },
    {
      "epoch": 0.7959410576426189,
      "grad_norm": 0.8068091740613124,
      "learning_rate": 1.0524290142994904e-06,
      "loss": 0.3907,
      "step": 12707
    },
    {
      "epoch": 0.7960036956419612,
      "grad_norm": 0.8942302235430846,
      "learning_rate": 1.0518065263246457e-06,
      "loss": 0.369,
      "step": 12708
    },
    {
      "epoch": 0.7960663336413035,
      "grad_norm": 0.8851710822145202,
      "learning_rate": 1.0511842008581774e-06,
      "loss": 0.3923,
      "step": 12709
    },
    {
      "epoch": 0.7961289716406458,
      "grad_norm": 0.8275365678752492,
      "learning_rate": 1.0505620379256986e-06,
      "loss": 0.3608,
      "step": 12710
    },
    {
      "epoch": 0.7961916096399881,
      "grad_norm": 0.8853637199950685,
      "learning_rate": 1.0499400375528207e-06,
      "loss": 0.3485,
      "step": 12711
    },
    {
      "epoch": 0.7962542476393304,
      "grad_norm": 0.8697200931378051,
      "learning_rate": 1.0493181997651442e-06,
      "loss": 0.4293,
      "step": 12712
    },
    {
      "epoch": 0.7963168856386728,
      "grad_norm": 0.8649209403877667,
      "learning_rate": 1.048696524588264e-06,
      "loss": 0.3971,
      "step": 12713
    },
    {
      "epoch": 0.796379523638015,
      "grad_norm": 0.8757302029157517,
      "learning_rate": 1.0480750120477677e-06,
      "loss": 0.4115,
      "step": 12714
    },
    {
      "epoch": 0.7964421616373573,
      "grad_norm": 0.8769369671085957,
      "learning_rate": 1.0474536621692365e-06,
      "loss": 0.3802,
      "step": 12715
    },
    {
      "epoch": 0.7965047996366996,
      "grad_norm": 0.8409766682000331,
      "learning_rate": 1.0468324749782476e-06,
      "loss": 0.3472,
      "step": 12716
    },
    {
      "epoch": 0.7965674376360419,
      "grad_norm": 0.8421783974634877,
      "learning_rate": 1.0462114505003678e-06,
      "loss": 0.3628,
      "step": 12717
    },
    {
      "epoch": 0.7966300756353842,
      "grad_norm": 0.910293249464771,
      "learning_rate": 1.0455905887611584e-06,
      "loss": 0.4533,
      "step": 12718
    },
    {
      "epoch": 0.7966927136347265,
      "grad_norm": 0.8654381407813934,
      "learning_rate": 1.0449698897861733e-06,
      "loss": 0.39,
      "step": 12719
    },
    {
      "epoch": 0.7967553516340689,
      "grad_norm": 0.8529069523681924,
      "learning_rate": 1.0443493536009626e-06,
      "loss": 0.3851,
      "step": 12720
    },
    {
      "epoch": 0.7968179896334111,
      "grad_norm": 0.9004478516463202,
      "learning_rate": 1.043728980231067e-06,
      "loss": 0.3921,
      "step": 12721
    },
    {
      "epoch": 0.7968806276327534,
      "grad_norm": 0.8334458110882214,
      "learning_rate": 1.0431087697020204e-06,
      "loss": 0.3558,
      "step": 12722
    },
    {
      "epoch": 0.7969432656320957,
      "grad_norm": 0.8197447330094375,
      "learning_rate": 1.0424887220393503e-06,
      "loss": 0.3839,
      "step": 12723
    },
    {
      "epoch": 0.797005903631438,
      "grad_norm": 0.9066605462386369,
      "learning_rate": 1.0418688372685805e-06,
      "loss": 0.4233,
      "step": 12724
    },
    {
      "epoch": 0.7970685416307803,
      "grad_norm": 0.8836439588711488,
      "learning_rate": 1.0412491154152232e-06,
      "loss": 0.3832,
      "step": 12725
    },
    {
      "epoch": 0.7971311796301226,
      "grad_norm": 0.8688585782928075,
      "learning_rate": 1.0406295565047881e-06,
      "loss": 0.3823,
      "step": 12726
    },
    {
      "epoch": 0.797193817629465,
      "grad_norm": 0.9412892232969037,
      "learning_rate": 1.0400101605627744e-06,
      "loss": 0.4504,
      "step": 12727
    },
    {
      "epoch": 0.7972564556288072,
      "grad_norm": 0.9523238928349955,
      "learning_rate": 1.039390927614677e-06,
      "loss": 0.3978,
      "step": 12728
    },
    {
      "epoch": 0.7973190936281496,
      "grad_norm": 0.6234207997305287,
      "learning_rate": 1.038771857685985e-06,
      "loss": 0.4517,
      "step": 12729
    },
    {
      "epoch": 0.7973817316274918,
      "grad_norm": 0.8550890230783986,
      "learning_rate": 1.0381529508021786e-06,
      "loss": 0.3949,
      "step": 12730
    },
    {
      "epoch": 0.7974443696268341,
      "grad_norm": 1.0150004999949873,
      "learning_rate": 1.0375342069887323e-06,
      "loss": 0.3905,
      "step": 12731
    },
    {
      "epoch": 0.7975070076261764,
      "grad_norm": 0.8228910849121629,
      "learning_rate": 1.0369156262711123e-06,
      "loss": 0.3705,
      "step": 12732
    },
    {
      "epoch": 0.7975696456255187,
      "grad_norm": 0.826588708835123,
      "learning_rate": 1.0362972086747814e-06,
      "loss": 0.3893,
      "step": 12733
    },
    {
      "epoch": 0.797632283624861,
      "grad_norm": 0.5837220473426723,
      "learning_rate": 1.0356789542251939e-06,
      "loss": 0.4488,
      "step": 12734
    },
    {
      "epoch": 0.7976949216242033,
      "grad_norm": 0.8319897749135794,
      "learning_rate": 1.035060862947796e-06,
      "loss": 0.369,
      "step": 12735
    },
    {
      "epoch": 0.7977575596235457,
      "grad_norm": 0.9629464093938124,
      "learning_rate": 1.0344429348680285e-06,
      "loss": 0.4063,
      "step": 12736
    },
    {
      "epoch": 0.7978201976228879,
      "grad_norm": 0.8243629955768544,
      "learning_rate": 1.0338251700113267e-06,
      "loss": 0.3619,
      "step": 12737
    },
    {
      "epoch": 0.7978828356222303,
      "grad_norm": 0.8902547639837998,
      "learning_rate": 1.033207568403115e-06,
      "loss": 0.3954,
      "step": 12738
    },
    {
      "epoch": 0.7979454736215725,
      "grad_norm": 0.8090071329703602,
      "learning_rate": 1.0325901300688179e-06,
      "loss": 0.3676,
      "step": 12739
    },
    {
      "epoch": 0.7980081116209148,
      "grad_norm": 0.8107354720379732,
      "learning_rate": 1.031972855033847e-06,
      "loss": 0.3591,
      "step": 12740
    },
    {
      "epoch": 0.7980707496202571,
      "grad_norm": 0.9124359651876515,
      "learning_rate": 1.0313557433236088e-06,
      "loss": 0.4036,
      "step": 12741
    },
    {
      "epoch": 0.7981333876195994,
      "grad_norm": 0.6406916041776353,
      "learning_rate": 1.0307387949635062e-06,
      "loss": 0.4403,
      "step": 12742
    },
    {
      "epoch": 0.7981960256189418,
      "grad_norm": 0.895879415937103,
      "learning_rate": 1.0301220099789316e-06,
      "loss": 0.4038,
      "step": 12743
    },
    {
      "epoch": 0.798258663618284,
      "grad_norm": 0.8252980061952763,
      "learning_rate": 1.0295053883952716e-06,
      "loss": 0.377,
      "step": 12744
    },
    {
      "epoch": 0.7983213016176264,
      "grad_norm": 0.8373430441609334,
      "learning_rate": 1.0288889302379073e-06,
      "loss": 0.3597,
      "step": 12745
    },
    {
      "epoch": 0.7983839396169686,
      "grad_norm": 0.8576944298803667,
      "learning_rate": 1.0282726355322115e-06,
      "loss": 0.3589,
      "step": 12746
    },
    {
      "epoch": 0.7984465776163109,
      "grad_norm": 0.8251431411103242,
      "learning_rate": 1.0276565043035496e-06,
      "loss": 0.4242,
      "step": 12747
    },
    {
      "epoch": 0.7985092156156532,
      "grad_norm": 0.8098705020101363,
      "learning_rate": 1.0270405365772845e-06,
      "loss": 0.3744,
      "step": 12748
    },
    {
      "epoch": 0.7985718536149955,
      "grad_norm": 0.9200200111702185,
      "learning_rate": 1.0264247323787675e-06,
      "loss": 0.3793,
      "step": 12749
    },
    {
      "epoch": 0.7986344916143379,
      "grad_norm": 0.9160126380653592,
      "learning_rate": 1.0258090917333468e-06,
      "loss": 0.3865,
      "step": 12750
    },
    {
      "epoch": 0.7986971296136801,
      "grad_norm": 0.6238652305668035,
      "learning_rate": 1.0251936146663622e-06,
      "loss": 0.4377,
      "step": 12751
    },
    {
      "epoch": 0.7987597676130225,
      "grad_norm": 0.8806983672380236,
      "learning_rate": 1.0245783012031457e-06,
      "loss": 0.4016,
      "step": 12752
    },
    {
      "epoch": 0.7988224056123647,
      "grad_norm": 0.8146353044799716,
      "learning_rate": 1.0239631513690246e-06,
      "loss": 0.3962,
      "step": 12753
    },
    {
      "epoch": 0.7988850436117071,
      "grad_norm": 0.7989375459199375,
      "learning_rate": 1.023348165189318e-06,
      "loss": 0.3659,
      "step": 12754
    },
    {
      "epoch": 0.7989476816110493,
      "grad_norm": 0.8768730158784354,
      "learning_rate": 1.0227333426893387e-06,
      "loss": 0.3442,
      "step": 12755
    },
    {
      "epoch": 0.7990103196103916,
      "grad_norm": 0.8969705227901348,
      "learning_rate": 1.0221186838943919e-06,
      "loss": 0.3934,
      "step": 12756
    },
    {
      "epoch": 0.799072957609734,
      "grad_norm": 0.8749941063121978,
      "learning_rate": 1.0215041888297806e-06,
      "loss": 0.3713,
      "step": 12757
    },
    {
      "epoch": 0.7991355956090762,
      "grad_norm": 0.8323831011066374,
      "learning_rate": 1.0208898575207931e-06,
      "loss": 0.3809,
      "step": 12758
    },
    {
      "epoch": 0.7991982336084186,
      "grad_norm": 0.8166441867958962,
      "learning_rate": 1.020275689992719e-06,
      "loss": 0.3766,
      "step": 12759
    },
    {
      "epoch": 0.7992608716077608,
      "grad_norm": 0.6322463681543669,
      "learning_rate": 1.0196616862708363e-06,
      "loss": 0.4563,
      "step": 12760
    },
    {
      "epoch": 0.7993235096071032,
      "grad_norm": 0.811145547024655,
      "learning_rate": 1.0190478463804176e-06,
      "loss": 0.3927,
      "step": 12761
    },
    {
      "epoch": 0.7993861476064454,
      "grad_norm": 0.8812190291149108,
      "learning_rate": 1.0184341703467282e-06,
      "loss": 0.3801,
      "step": 12762
    },
    {
      "epoch": 0.7994487856057878,
      "grad_norm": 0.8104931363870541,
      "learning_rate": 1.0178206581950278e-06,
      "loss": 0.3936,
      "step": 12763
    },
    {
      "epoch": 0.7995114236051301,
      "grad_norm": 0.8817648541815524,
      "learning_rate": 1.017207309950568e-06,
      "loss": 0.3886,
      "step": 12764
    },
    {
      "epoch": 0.7995740616044723,
      "grad_norm": 0.8932591190663326,
      "learning_rate": 1.016594125638593e-06,
      "loss": 0.3588,
      "step": 12765
    },
    {
      "epoch": 0.7996366996038147,
      "grad_norm": 0.8181280125721213,
      "learning_rate": 1.015981105284345e-06,
      "loss": 0.3468,
      "step": 12766
    },
    {
      "epoch": 0.7996993376031569,
      "grad_norm": 0.8023332211238079,
      "learning_rate": 1.0153682489130522e-06,
      "loss": 0.4009,
      "step": 12767
    },
    {
      "epoch": 0.7997619756024993,
      "grad_norm": 0.6291875284984174,
      "learning_rate": 1.0147555565499435e-06,
      "loss": 0.4359,
      "step": 12768
    },
    {
      "epoch": 0.7998246136018415,
      "grad_norm": 0.8699751380931837,
      "learning_rate": 1.0141430282202357e-06,
      "loss": 0.3818,
      "step": 12769
    },
    {
      "epoch": 0.7998872516011839,
      "grad_norm": 0.9104581479778557,
      "learning_rate": 1.0135306639491405e-06,
      "loss": 0.4103,
      "step": 12770
    },
    {
      "epoch": 0.7999498896005262,
      "grad_norm": 0.908940886991088,
      "learning_rate": 1.0129184637618634e-06,
      "loss": 0.3894,
      "step": 12771
    },
    {
      "epoch": 0.8000125275998684,
      "grad_norm": 0.8809352419310713,
      "learning_rate": 1.0123064276836025e-06,
      "loss": 0.3683,
      "step": 12772
    },
    {
      "epoch": 0.8000751655992108,
      "grad_norm": 0.9143245140270528,
      "learning_rate": 1.0116945557395485e-06,
      "loss": 0.3597,
      "step": 12773
    },
    {
      "epoch": 0.800137803598553,
      "grad_norm": 0.8416454777915793,
      "learning_rate": 1.0110828479548862e-06,
      "loss": 0.3908,
      "step": 12774
    },
    {
      "epoch": 0.8002004415978954,
      "grad_norm": 0.9293757608770757,
      "learning_rate": 1.0104713043547941e-06,
      "loss": 0.3687,
      "step": 12775
    },
    {
      "epoch": 0.8002630795972376,
      "grad_norm": 0.8790950498208995,
      "learning_rate": 1.0098599249644448e-06,
      "loss": 0.3608,
      "step": 12776
    },
    {
      "epoch": 0.80032571759658,
      "grad_norm": 0.8408118383859268,
      "learning_rate": 1.009248709809002e-06,
      "loss": 0.3836,
      "step": 12777
    },
    {
      "epoch": 0.8003883555959223,
      "grad_norm": 0.6410163225015052,
      "learning_rate": 1.0086376589136226e-06,
      "loss": 0.4579,
      "step": 12778
    },
    {
      "epoch": 0.8004509935952646,
      "grad_norm": 0.909596841085858,
      "learning_rate": 1.008026772303458e-06,
      "loss": 0.3982,
      "step": 12779
    },
    {
      "epoch": 0.8005136315946069,
      "grad_norm": 0.8559137533650333,
      "learning_rate": 1.0074160500036523e-06,
      "loss": 0.358,
      "step": 12780
    },
    {
      "epoch": 0.8005762695939491,
      "grad_norm": 0.8395044377500965,
      "learning_rate": 1.0068054920393433e-06,
      "loss": 0.4067,
      "step": 12781
    },
    {
      "epoch": 0.8006389075932915,
      "grad_norm": 0.9019519942526985,
      "learning_rate": 1.00619509843566e-06,
      "loss": 0.3923,
      "step": 12782
    },
    {
      "epoch": 0.8007015455926337,
      "grad_norm": 0.8395612349543735,
      "learning_rate": 1.0055848692177294e-06,
      "loss": 0.3493,
      "step": 12783
    },
    {
      "epoch": 0.8007641835919761,
      "grad_norm": 0.8704261566038534,
      "learning_rate": 1.004974804410665e-06,
      "loss": 0.3797,
      "step": 12784
    },
    {
      "epoch": 0.8008268215913183,
      "grad_norm": 0.8719058061036661,
      "learning_rate": 1.0043649040395808e-06,
      "loss": 0.3813,
      "step": 12785
    },
    {
      "epoch": 0.8008894595906607,
      "grad_norm": 0.6286815766624659,
      "learning_rate": 1.0037551681295787e-06,
      "loss": 0.4481,
      "step": 12786
    },
    {
      "epoch": 0.800952097590003,
      "grad_norm": 0.8372900398816635,
      "learning_rate": 1.0031455967057557e-06,
      "loss": 0.396,
      "step": 12787
    },
    {
      "epoch": 0.8010147355893453,
      "grad_norm": 0.9182042044386384,
      "learning_rate": 1.0025361897932023e-06,
      "loss": 0.3938,
      "step": 12788
    },
    {
      "epoch": 0.8010773735886876,
      "grad_norm": 0.8912344247676331,
      "learning_rate": 1.0019269474170008e-06,
      "loss": 0.4026,
      "step": 12789
    },
    {
      "epoch": 0.8011400115880298,
      "grad_norm": 0.847432420884411,
      "learning_rate": 1.001317869602228e-06,
      "loss": 0.3856,
      "step": 12790
    },
    {
      "epoch": 0.8012026495873722,
      "grad_norm": 0.8267842610950057,
      "learning_rate": 1.0007089563739524e-06,
      "loss": 0.3698,
      "step": 12791
    },
    {
      "epoch": 0.8012652875867144,
      "grad_norm": 0.8176579818231531,
      "learning_rate": 1.0001002077572402e-06,
      "loss": 0.3635,
      "step": 12792
    },
    {
      "epoch": 0.8013279255860568,
      "grad_norm": 0.8482996674314407,
      "learning_rate": 9.994916237771446e-07,
      "loss": 0.3558,
      "step": 12793
    },
    {
      "epoch": 0.8013905635853991,
      "grad_norm": 0.8743124484314951,
      "learning_rate": 9.988832044587176e-07,
      "loss": 0.3818,
      "step": 12794
    },
    {
      "epoch": 0.8014532015847414,
      "grad_norm": 0.8426504385442005,
      "learning_rate": 9.982749498270005e-07,
      "loss": 0.3847,
      "step": 12795
    },
    {
      "epoch": 0.8015158395840837,
      "grad_norm": 0.8920079601237093,
      "learning_rate": 9.976668599070293e-07,
      "loss": 0.3569,
      "step": 12796
    },
    {
      "epoch": 0.8015784775834259,
      "grad_norm": 0.8586877706643378,
      "learning_rate": 9.97058934723833e-07,
      "loss": 0.3603,
      "step": 12797
    },
    {
      "epoch": 0.8016411155827683,
      "grad_norm": 0.9193613497723079,
      "learning_rate": 9.964511743024336e-07,
      "loss": 0.3619,
      "step": 12798
    },
    {
      "epoch": 0.8017037535821105,
      "grad_norm": 0.9078536990740554,
      "learning_rate": 9.958435786678471e-07,
      "loss": 0.366,
      "step": 12799
    },
    {
      "epoch": 0.8017663915814529,
      "grad_norm": 0.8976592768899251,
      "learning_rate": 9.952361478450811e-07,
      "loss": 0.3962,
      "step": 12800
    },
    {
      "epoch": 0.8018290295807952,
      "grad_norm": 0.8568266450960941,
      "learning_rate": 9.946288818591394e-07,
      "loss": 0.4357,
      "step": 12801
    },
    {
      "epoch": 0.8018916675801375,
      "grad_norm": 0.8254684837321501,
      "learning_rate": 9.940217807350155e-07,
      "loss": 0.3878,
      "step": 12802
    },
    {
      "epoch": 0.8019543055794798,
      "grad_norm": 0.5923909814402181,
      "learning_rate": 9.934148444976998e-07,
      "loss": 0.4342,
      "step": 12803
    },
    {
      "epoch": 0.8020169435788221,
      "grad_norm": 0.9101111274386255,
      "learning_rate": 9.928080731721728e-07,
      "loss": 0.3868,
      "step": 12804
    },
    {
      "epoch": 0.8020795815781644,
      "grad_norm": 0.8481401923001857,
      "learning_rate": 9.92201466783409e-07,
      "loss": 0.3615,
      "step": 12805
    },
    {
      "epoch": 0.8021422195775066,
      "grad_norm": 0.8958889485629838,
      "learning_rate": 9.915950253563773e-07,
      "loss": 0.3824,
      "step": 12806
    },
    {
      "epoch": 0.802204857576849,
      "grad_norm": 0.8425238404196935,
      "learning_rate": 9.909887489160375e-07,
      "loss": 0.379,
      "step": 12807
    },
    {
      "epoch": 0.8022674955761913,
      "grad_norm": 0.8204576315647109,
      "learning_rate": 9.903826374873443e-07,
      "loss": 0.3672,
      "step": 12808
    },
    {
      "epoch": 0.8023301335755336,
      "grad_norm": 0.8101129123952396,
      "learning_rate": 9.897766910952466e-07,
      "loss": 0.3634,
      "step": 12809
    },
    {
      "epoch": 0.8023927715748759,
      "grad_norm": 0.8561169179575965,
      "learning_rate": 9.89170909764685e-07,
      "loss": 0.4145,
      "step": 12810
    },
    {
      "epoch": 0.8024554095742182,
      "grad_norm": 0.7388103706316789,
      "learning_rate": 9.885652935205925e-07,
      "loss": 0.3176,
      "step": 12811
    },
    {
      "epoch": 0.8025180475735605,
      "grad_norm": 0.8581419659063821,
      "learning_rate": 9.879598423878978e-07,
      "loss": 0.385,
      "step": 12812
    },
    {
      "epoch": 0.8025806855729029,
      "grad_norm": 0.8500103532059387,
      "learning_rate": 9.873545563915204e-07,
      "loss": 0.3754,
      "step": 12813
    },
    {
      "epoch": 0.8026433235722451,
      "grad_norm": 0.8451410512583917,
      "learning_rate": 9.867494355563745e-07,
      "loss": 0.3799,
      "step": 12814
    },
    {
      "epoch": 0.8027059615715874,
      "grad_norm": 0.8626968242993068,
      "learning_rate": 9.861444799073672e-07,
      "loss": 0.3727,
      "step": 12815
    },
    {
      "epoch": 0.8027685995709297,
      "grad_norm": 0.9346952236277822,
      "learning_rate": 9.855396894693975e-07,
      "loss": 0.4043,
      "step": 12816
    },
    {
      "epoch": 0.802831237570272,
      "grad_norm": 0.868642189612378,
      "learning_rate": 9.849350642673577e-07,
      "loss": 0.3923,
      "step": 12817
    },
    {
      "epoch": 0.8028938755696143,
      "grad_norm": 0.877353201987724,
      "learning_rate": 9.84330604326138e-07,
      "loss": 0.3839,
      "step": 12818
    },
    {
      "epoch": 0.8029565135689566,
      "grad_norm": 0.9509274488424367,
      "learning_rate": 9.837263096706157e-07,
      "loss": 0.3625,
      "step": 12819
    },
    {
      "epoch": 0.803019151568299,
      "grad_norm": 0.8258503939802673,
      "learning_rate": 9.831221803256636e-07,
      "loss": 0.396,
      "step": 12820
    },
    {
      "epoch": 0.8030817895676412,
      "grad_norm": 0.6489747557357645,
      "learning_rate": 9.825182163161473e-07,
      "loss": 0.4576,
      "step": 12821
    },
    {
      "epoch": 0.8031444275669836,
      "grad_norm": 0.7801972984679978,
      "learning_rate": 9.81914417666928e-07,
      "loss": 0.3541,
      "step": 12822
    },
    {
      "epoch": 0.8032070655663258,
      "grad_norm": 0.8041037534290553,
      "learning_rate": 9.813107844028569e-07,
      "loss": 0.3836,
      "step": 12823
    },
    {
      "epoch": 0.8032697035656681,
      "grad_norm": 0.885235459032196,
      "learning_rate": 9.807073165487802e-07,
      "loss": 0.3753,
      "step": 12824
    },
    {
      "epoch": 0.8033323415650104,
      "grad_norm": 0.9288207007627259,
      "learning_rate": 9.80104014129536e-07,
      "loss": 0.4009,
      "step": 12825
    },
    {
      "epoch": 0.8033949795643527,
      "grad_norm": 0.766164780110576,
      "learning_rate": 9.795008771699555e-07,
      "loss": 0.3117,
      "step": 12826
    },
    {
      "epoch": 0.803457617563695,
      "grad_norm": 0.8328871247186147,
      "learning_rate": 9.788979056948666e-07,
      "loss": 0.3902,
      "step": 12827
    },
    {
      "epoch": 0.8035202555630373,
      "grad_norm": 0.8582101539724573,
      "learning_rate": 9.782950997290863e-07,
      "loss": 0.3644,
      "step": 12828
    },
    {
      "epoch": 0.8035828935623797,
      "grad_norm": 0.7953328053355502,
      "learning_rate": 9.776924592974257e-07,
      "loss": 0.3642,
      "step": 12829
    },
    {
      "epoch": 0.8036455315617219,
      "grad_norm": 0.9271910691767893,
      "learning_rate": 9.77089984424689e-07,
      "loss": 0.3699,
      "step": 12830
    },
    {
      "epoch": 0.8037081695610642,
      "grad_norm": 0.9021628687259613,
      "learning_rate": 9.764876751356766e-07,
      "loss": 0.3678,
      "step": 12831
    },
    {
      "epoch": 0.8037708075604065,
      "grad_norm": 0.8958910202965139,
      "learning_rate": 9.758855314551785e-07,
      "loss": 0.4111,
      "step": 12832
    },
    {
      "epoch": 0.8038334455597488,
      "grad_norm": 0.8241273085709235,
      "learning_rate": 9.752835534079785e-07,
      "loss": 0.4143,
      "step": 12833
    },
    {
      "epoch": 0.8038960835590911,
      "grad_norm": 0.7910616384280587,
      "learning_rate": 9.74681741018853e-07,
      "loss": 0.37,
      "step": 12834
    },
    {
      "epoch": 0.8039587215584334,
      "grad_norm": 0.8947476498174336,
      "learning_rate": 9.74080094312576e-07,
      "loss": 0.3776,
      "step": 12835
    },
    {
      "epoch": 0.8040213595577758,
      "grad_norm": 0.7723311771095349,
      "learning_rate": 9.73478613313909e-07,
      "loss": 0.3466,
      "step": 12836
    },
    {
      "epoch": 0.804083997557118,
      "grad_norm": 0.9356287081422243,
      "learning_rate": 9.728772980476103e-07,
      "loss": 0.4369,
      "step": 12837
    },
    {
      "epoch": 0.8041466355564604,
      "grad_norm": 0.9140105026026196,
      "learning_rate": 9.72276148538429e-07,
      "loss": 0.4121,
      "step": 12838
    },
    {
      "epoch": 0.8042092735558026,
      "grad_norm": 0.9010778975101709,
      "learning_rate": 9.716751648111078e-07,
      "loss": 0.3646,
      "step": 12839
    },
    {
      "epoch": 0.8042719115551449,
      "grad_norm": 0.8803071973654701,
      "learning_rate": 9.71074346890386e-07,
      "loss": 0.3668,
      "step": 12840
    },
    {
      "epoch": 0.8043345495544872,
      "grad_norm": 0.8969215026378871,
      "learning_rate": 9.704736948009913e-07,
      "loss": 0.3686,
      "step": 12841
    },
    {
      "epoch": 0.8043971875538295,
      "grad_norm": 0.7906330755980667,
      "learning_rate": 9.698732085676476e-07,
      "loss": 0.3678,
      "step": 12842
    },
    {
      "epoch": 0.8044598255531719,
      "grad_norm": 0.8116089248108201,
      "learning_rate": 9.692728882150698e-07,
      "loss": 0.3939,
      "step": 12843
    },
    {
      "epoch": 0.8045224635525141,
      "grad_norm": 0.8726857511713607,
      "learning_rate": 9.68672733767969e-07,
      "loss": 0.3633,
      "step": 12844
    },
    {
      "epoch": 0.8045851015518565,
      "grad_norm": 0.9251163401962641,
      "learning_rate": 9.680727452510464e-07,
      "loss": 0.4069,
      "step": 12845
    },
    {
      "epoch": 0.8046477395511987,
      "grad_norm": 0.840708246196672,
      "learning_rate": 9.674729226889985e-07,
      "loss": 0.3599,
      "step": 12846
    },
    {
      "epoch": 0.8047103775505411,
      "grad_norm": 0.9021700234049816,
      "learning_rate": 9.668732661065134e-07,
      "loss": 0.3682,
      "step": 12847
    },
    {
      "epoch": 0.8047730155498833,
      "grad_norm": 0.9323583892331443,
      "learning_rate": 9.66273775528272e-07,
      "loss": 0.3983,
      "step": 12848
    },
    {
      "epoch": 0.8048356535492256,
      "grad_norm": 0.7956312713271197,
      "learning_rate": 9.656744509789518e-07,
      "loss": 0.3895,
      "step": 12849
    },
    {
      "epoch": 0.804898291548568,
      "grad_norm": 0.6344044705913452,
      "learning_rate": 9.650752924832203e-07,
      "loss": 0.4498,
      "step": 12850
    },
    {
      "epoch": 0.8049609295479102,
      "grad_norm": 0.8960084881751633,
      "learning_rate": 9.64476300065738e-07,
      "loss": 0.3533,
      "step": 12851
    },
    {
      "epoch": 0.8050235675472526,
      "grad_norm": 0.8836789791494617,
      "learning_rate": 9.638774737511597e-07,
      "loss": 0.4006,
      "step": 12852
    },
    {
      "epoch": 0.8050862055465948,
      "grad_norm": 0.8192522723139921,
      "learning_rate": 9.632788135641346e-07,
      "loss": 0.3582,
      "step": 12853
    },
    {
      "epoch": 0.8051488435459372,
      "grad_norm": 0.9475612424945725,
      "learning_rate": 9.626803195293028e-07,
      "loss": 0.4064,
      "step": 12854
    },
    {
      "epoch": 0.8052114815452794,
      "grad_norm": 0.8375316649526489,
      "learning_rate": 9.620819916712987e-07,
      "loss": 0.3641,
      "step": 12855
    },
    {
      "epoch": 0.8052741195446217,
      "grad_norm": 0.851724752770884,
      "learning_rate": 9.614838300147488e-07,
      "loss": 0.3974,
      "step": 12856
    },
    {
      "epoch": 0.805336757543964,
      "grad_norm": 0.8277049049547257,
      "learning_rate": 9.60885834584273e-07,
      "loss": 0.3584,
      "step": 12857
    },
    {
      "epoch": 0.8053993955433063,
      "grad_norm": 0.9033420649611713,
      "learning_rate": 9.602880054044872e-07,
      "loss": 0.4009,
      "step": 12858
    },
    {
      "epoch": 0.8054620335426487,
      "grad_norm": 0.8347618528186865,
      "learning_rate": 9.59690342499997e-07,
      "loss": 0.3633,
      "step": 12859
    },
    {
      "epoch": 0.8055246715419909,
      "grad_norm": 0.9037703890364572,
      "learning_rate": 9.590928458954013e-07,
      "loss": 0.4135,
      "step": 12860
    },
    {
      "epoch": 0.8055873095413333,
      "grad_norm": 0.8381574829503854,
      "learning_rate": 9.584955156152948e-07,
      "loss": 0.3848,
      "step": 12861
    },
    {
      "epoch": 0.8056499475406755,
      "grad_norm": 0.872442185984471,
      "learning_rate": 9.578983516842633e-07,
      "loss": 0.381,
      "step": 12862
    },
    {
      "epoch": 0.8057125855400179,
      "grad_norm": 0.8474001027096487,
      "learning_rate": 9.573013541268855e-07,
      "loss": 0.3903,
      "step": 12863
    },
    {
      "epoch": 0.8057752235393602,
      "grad_norm": 0.8556060928824081,
      "learning_rate": 9.567045229677347e-07,
      "loss": 0.4079,
      "step": 12864
    },
    {
      "epoch": 0.8058378615387024,
      "grad_norm": 0.6576701826615069,
      "learning_rate": 9.561078582313765e-07,
      "loss": 0.4538,
      "step": 12865
    },
    {
      "epoch": 0.8059004995380448,
      "grad_norm": 0.9137245862722655,
      "learning_rate": 9.555113599423689e-07,
      "loss": 0.3733,
      "step": 12866
    },
    {
      "epoch": 0.805963137537387,
      "grad_norm": 0.829615576579356,
      "learning_rate": 9.549150281252633e-07,
      "loss": 0.3565,
      "step": 12867
    },
    {
      "epoch": 0.8060257755367294,
      "grad_norm": 0.8193802721249345,
      "learning_rate": 9.543188628046073e-07,
      "loss": 0.3341,
      "step": 12868
    },
    {
      "epoch": 0.8060884135360716,
      "grad_norm": 0.7727124387381722,
      "learning_rate": 9.53722864004936e-07,
      "loss": 0.4068,
      "step": 12869
    },
    {
      "epoch": 0.806151051535414,
      "grad_norm": 0.6452517628776667,
      "learning_rate": 9.531270317507845e-07,
      "loss": 0.4671,
      "step": 12870
    },
    {
      "epoch": 0.8062136895347563,
      "grad_norm": 0.8926866156266241,
      "learning_rate": 9.525313660666757e-07,
      "loss": 0.4045,
      "step": 12871
    },
    {
      "epoch": 0.8062763275340986,
      "grad_norm": 0.9438340031045891,
      "learning_rate": 9.519358669771268e-07,
      "loss": 0.3971,
      "step": 12872
    },
    {
      "epoch": 0.8063389655334409,
      "grad_norm": 0.8906661114353768,
      "learning_rate": 9.51340534506649e-07,
      "loss": 0.3891,
      "step": 12873
    },
    {
      "epoch": 0.8064016035327831,
      "grad_norm": 0.8096986407542536,
      "learning_rate": 9.507453686797464e-07,
      "loss": 0.3761,
      "step": 12874
    },
    {
      "epoch": 0.8064642415321255,
      "grad_norm": 0.874515299229083,
      "learning_rate": 9.501503695209158e-07,
      "loss": 0.3636,
      "step": 12875
    },
    {
      "epoch": 0.8065268795314677,
      "grad_norm": 0.7568328912506019,
      "learning_rate": 9.495555370546461e-07,
      "loss": 0.3415,
      "step": 12876
    },
    {
      "epoch": 0.8065895175308101,
      "grad_norm": 0.834327710105622,
      "learning_rate": 9.489608713054227e-07,
      "loss": 0.4108,
      "step": 12877
    },
    {
      "epoch": 0.8066521555301523,
      "grad_norm": 0.8775326005412718,
      "learning_rate": 9.483663722977232e-07,
      "loss": 0.4007,
      "step": 12878
    },
    {
      "epoch": 0.8067147935294947,
      "grad_norm": 0.8906382257824385,
      "learning_rate": 9.477720400560153e-07,
      "loss": 0.4052,
      "step": 12879
    },
    {
      "epoch": 0.806777431528837,
      "grad_norm": 0.8465224075721086,
      "learning_rate": 9.471778746047628e-07,
      "loss": 0.376,
      "step": 12880
    },
    {
      "epoch": 0.8068400695281792,
      "grad_norm": 0.8092226201435163,
      "learning_rate": 9.465838759684215e-07,
      "loss": 0.3659,
      "step": 12881
    },
    {
      "epoch": 0.8069027075275216,
      "grad_norm": 0.8002777380178409,
      "learning_rate": 9.459900441714398e-07,
      "loss": 0.3442,
      "step": 12882
    },
    {
      "epoch": 0.8069653455268638,
      "grad_norm": 0.8891825091529926,
      "learning_rate": 9.453963792382603e-07,
      "loss": 0.376,
      "step": 12883
    },
    {
      "epoch": 0.8070279835262062,
      "grad_norm": 0.89111220981614,
      "learning_rate": 9.44802881193318e-07,
      "loss": 0.3719,
      "step": 12884
    },
    {
      "epoch": 0.8070906215255484,
      "grad_norm": 0.8527838213238146,
      "learning_rate": 9.44209550061041e-07,
      "loss": 0.3452,
      "step": 12885
    },
    {
      "epoch": 0.8071532595248908,
      "grad_norm": 0.922870883811091,
      "learning_rate": 9.43616385865852e-07,
      "loss": 0.4218,
      "step": 12886
    },
    {
      "epoch": 0.8072158975242331,
      "grad_norm": 0.9554497405571056,
      "learning_rate": 9.43023388632166e-07,
      "loss": 0.3997,
      "step": 12887
    },
    {
      "epoch": 0.8072785355235754,
      "grad_norm": 0.8614465500820658,
      "learning_rate": 9.424305583843912e-07,
      "loss": 0.37,
      "step": 12888
    },
    {
      "epoch": 0.8073411735229177,
      "grad_norm": 0.9262887788508539,
      "learning_rate": 9.41837895146927e-07,
      "loss": 0.3782,
      "step": 12889
    },
    {
      "epoch": 0.8074038115222599,
      "grad_norm": 0.8383933223440881,
      "learning_rate": 9.412453989441689e-07,
      "loss": 0.3951,
      "step": 12890
    },
    {
      "epoch": 0.8074664495216023,
      "grad_norm": 0.916035220684202,
      "learning_rate": 9.406530698005028e-07,
      "loss": 0.3847,
      "step": 12891
    },
    {
      "epoch": 0.8075290875209445,
      "grad_norm": 0.8902875474015475,
      "learning_rate": 9.4006090774031e-07,
      "loss": 0.376,
      "step": 12892
    },
    {
      "epoch": 0.8075917255202869,
      "grad_norm": 0.8892764718828912,
      "learning_rate": 9.39468912787963e-07,
      "loss": 0.4314,
      "step": 12893
    },
    {
      "epoch": 0.8076543635196292,
      "grad_norm": 0.7988160252765756,
      "learning_rate": 9.388770849678302e-07,
      "loss": 0.389,
      "step": 12894
    },
    {
      "epoch": 0.8077170015189715,
      "grad_norm": 0.8026958293499501,
      "learning_rate": 9.382854243042688e-07,
      "loss": 0.3973,
      "step": 12895
    },
    {
      "epoch": 0.8077796395183138,
      "grad_norm": 0.8536486867263463,
      "learning_rate": 9.376939308216348e-07,
      "loss": 0.3293,
      "step": 12896
    },
    {
      "epoch": 0.8078422775176561,
      "grad_norm": 0.8518938308264704,
      "learning_rate": 9.371026045442727e-07,
      "loss": 0.3473,
      "step": 12897
    },
    {
      "epoch": 0.8079049155169984,
      "grad_norm": 0.9211880634595958,
      "learning_rate": 9.365114454965213e-07,
      "loss": 0.3813,
      "step": 12898
    },
    {
      "epoch": 0.8079675535163406,
      "grad_norm": 0.9461818438113997,
      "learning_rate": 9.359204537027133e-07,
      "loss": 0.3769,
      "step": 12899
    },
    {
      "epoch": 0.808030191515683,
      "grad_norm": 0.9037243501855758,
      "learning_rate": 9.353296291871733e-07,
      "loss": 0.3737,
      "step": 12900
    },
    {
      "epoch": 0.8080928295150253,
      "grad_norm": 0.8791719248194189,
      "learning_rate": 9.347389719742206e-07,
      "loss": 0.3582,
      "step": 12901
    },
    {
      "epoch": 0.8081554675143676,
      "grad_norm": 0.8938671335228917,
      "learning_rate": 9.341484820881652e-07,
      "loss": 0.3567,
      "step": 12902
    },
    {
      "epoch": 0.8082181055137099,
      "grad_norm": 0.8668627672938216,
      "learning_rate": 9.335581595533122e-07,
      "loss": 0.4259,
      "step": 12903
    },
    {
      "epoch": 0.8082807435130522,
      "grad_norm": 0.6094740470665155,
      "learning_rate": 9.329680043939621e-07,
      "loss": 0.4634,
      "step": 12904
    },
    {
      "epoch": 0.8083433815123945,
      "grad_norm": 0.87270982220204,
      "learning_rate": 9.323780166344038e-07,
      "loss": 0.403,
      "step": 12905
    },
    {
      "epoch": 0.8084060195117367,
      "grad_norm": 0.9110552291862154,
      "learning_rate": 9.317881962989212e-07,
      "loss": 0.399,
      "step": 12906
    },
    {
      "epoch": 0.8084686575110791,
      "grad_norm": 0.5910205924107854,
      "learning_rate": 9.31198543411791e-07,
      "loss": 0.4424,
      "step": 12907
    },
    {
      "epoch": 0.8085312955104214,
      "grad_norm": 0.84937104917728,
      "learning_rate": 9.306090579972848e-07,
      "loss": 0.3981,
      "step": 12908
    },
    {
      "epoch": 0.8085939335097637,
      "grad_norm": 0.867629231789529,
      "learning_rate": 9.300197400796645e-07,
      "loss": 0.4398,
      "step": 12909
    },
    {
      "epoch": 0.808656571509106,
      "grad_norm": 0.8610022399213138,
      "learning_rate": 9.294305896831873e-07,
      "loss": 0.3693,
      "step": 12910
    },
    {
      "epoch": 0.8087192095084483,
      "grad_norm": 0.8330345483105566,
      "learning_rate": 9.28841606832101e-07,
      "loss": 0.357,
      "step": 12911
    },
    {
      "epoch": 0.8087818475077906,
      "grad_norm": 0.8822588718529194,
      "learning_rate": 9.282527915506517e-07,
      "loss": 0.4087,
      "step": 12912
    },
    {
      "epoch": 0.808844485507133,
      "grad_norm": 0.8497959777936306,
      "learning_rate": 9.276641438630718e-07,
      "loss": 0.3475,
      "step": 12913
    },
    {
      "epoch": 0.8089071235064752,
      "grad_norm": 0.8601408326731518,
      "learning_rate": 9.270756637935924e-07,
      "loss": 0.3781,
      "step": 12914
    },
    {
      "epoch": 0.8089697615058175,
      "grad_norm": 0.8623987352718419,
      "learning_rate": 9.264873513664352e-07,
      "loss": 0.4525,
      "step": 12915
    },
    {
      "epoch": 0.8090323995051598,
      "grad_norm": 0.8342381896379482,
      "learning_rate": 9.258992066058142e-07,
      "loss": 0.3545,
      "step": 12916
    },
    {
      "epoch": 0.8090950375045021,
      "grad_norm": 0.865487672690156,
      "learning_rate": 9.253112295359384e-07,
      "loss": 0.3402,
      "step": 12917
    },
    {
      "epoch": 0.8091576755038444,
      "grad_norm": 0.8122800894346076,
      "learning_rate": 9.24723420181009e-07,
      "loss": 0.3786,
      "step": 12918
    },
    {
      "epoch": 0.8092203135031867,
      "grad_norm": 0.8905809933037251,
      "learning_rate": 9.241357785652183e-07,
      "loss": 0.4172,
      "step": 12919
    },
    {
      "epoch": 0.809282951502529,
      "grad_norm": 0.9252974092540868,
      "learning_rate": 9.23548304712757e-07,
      "loss": 0.3939,
      "step": 12920
    },
    {
      "epoch": 0.8093455895018713,
      "grad_norm": 0.8348872033638368,
      "learning_rate": 9.229609986478034e-07,
      "loss": 0.3482,
      "step": 12921
    },
    {
      "epoch": 0.8094082275012137,
      "grad_norm": 0.9076264191581424,
      "learning_rate": 9.223738603945314e-07,
      "loss": 0.4021,
      "step": 12922
    },
    {
      "epoch": 0.8094708655005559,
      "grad_norm": 0.8823123898733107,
      "learning_rate": 9.217868899771087e-07,
      "loss": 0.381,
      "step": 12923
    },
    {
      "epoch": 0.8095335034998982,
      "grad_norm": 0.8731444485690933,
      "learning_rate": 9.212000874196953e-07,
      "loss": 0.3836,
      "step": 12924
    },
    {
      "epoch": 0.8095961414992405,
      "grad_norm": 0.881320208747182,
      "learning_rate": 9.206134527464427e-07,
      "loss": 0.3884,
      "step": 12925
    },
    {
      "epoch": 0.8096587794985828,
      "grad_norm": 0.9413843668376268,
      "learning_rate": 9.200269859814981e-07,
      "loss": 0.4466,
      "step": 12926
    },
    {
      "epoch": 0.8097214174979251,
      "grad_norm": 0.8787637867923428,
      "learning_rate": 9.194406871489997e-07,
      "loss": 0.3891,
      "step": 12927
    },
    {
      "epoch": 0.8097840554972674,
      "grad_norm": 0.9129766533807006,
      "learning_rate": 9.188545562730789e-07,
      "loss": 0.3945,
      "step": 12928
    },
    {
      "epoch": 0.8098466934966098,
      "grad_norm": 0.9357137017248737,
      "learning_rate": 9.182685933778635e-07,
      "loss": 0.3515,
      "step": 12929
    },
    {
      "epoch": 0.809909331495952,
      "grad_norm": 0.9283030100770991,
      "learning_rate": 9.176827984874698e-07,
      "loss": 0.4165,
      "step": 12930
    },
    {
      "epoch": 0.8099719694952944,
      "grad_norm": 0.8362630069055331,
      "learning_rate": 9.170971716260091e-07,
      "loss": 0.4054,
      "step": 12931
    },
    {
      "epoch": 0.8100346074946366,
      "grad_norm": 0.8086539452858257,
      "learning_rate": 9.165117128175877e-07,
      "loss": 0.3566,
      "step": 12932
    },
    {
      "epoch": 0.8100972454939789,
      "grad_norm": 0.9214849044427641,
      "learning_rate": 9.159264220863023e-07,
      "loss": 0.3995,
      "step": 12933
    },
    {
      "epoch": 0.8101598834933212,
      "grad_norm": 0.8563095295681787,
      "learning_rate": 9.153412994562433e-07,
      "loss": 0.385,
      "step": 12934
    },
    {
      "epoch": 0.8102225214926635,
      "grad_norm": 0.8389797218914681,
      "learning_rate": 9.147563449514946e-07,
      "loss": 0.4074,
      "step": 12935
    },
    {
      "epoch": 0.8102851594920059,
      "grad_norm": 0.8159072161650869,
      "learning_rate": 9.14171558596133e-07,
      "loss": 0.3512,
      "step": 12936
    },
    {
      "epoch": 0.8103477974913481,
      "grad_norm": 0.8788292487972149,
      "learning_rate": 9.135869404142267e-07,
      "loss": 0.3844,
      "step": 12937
    },
    {
      "epoch": 0.8104104354906905,
      "grad_norm": 0.8659126735444529,
      "learning_rate": 9.130024904298423e-07,
      "loss": 0.3464,
      "step": 12938
    },
    {
      "epoch": 0.8104730734900327,
      "grad_norm": 0.8127916034008158,
      "learning_rate": 9.124182086670336e-07,
      "loss": 0.4037,
      "step": 12939
    },
    {
      "epoch": 0.810535711489375,
      "grad_norm": 0.9155656850947594,
      "learning_rate": 9.118340951498489e-07,
      "loss": 0.4058,
      "step": 12940
    },
    {
      "epoch": 0.8105983494887173,
      "grad_norm": 0.9446557459044882,
      "learning_rate": 9.112501499023335e-07,
      "loss": 0.4159,
      "step": 12941
    },
    {
      "epoch": 0.8106609874880596,
      "grad_norm": 0.8383463908690378,
      "learning_rate": 9.106663729485199e-07,
      "loss": 0.3326,
      "step": 12942
    },
    {
      "epoch": 0.810723625487402,
      "grad_norm": 0.8888246165632249,
      "learning_rate": 9.100827643124383e-07,
      "loss": 0.3649,
      "step": 12943
    },
    {
      "epoch": 0.8107862634867442,
      "grad_norm": 0.8869985112919054,
      "learning_rate": 9.094993240181088e-07,
      "loss": 0.3787,
      "step": 12944
    },
    {
      "epoch": 0.8108489014860866,
      "grad_norm": 0.8513674208279276,
      "learning_rate": 9.089160520895446e-07,
      "loss": 0.4063,
      "step": 12945
    },
    {
      "epoch": 0.8109115394854288,
      "grad_norm": 0.8673126024726577,
      "learning_rate": 9.08332948550757e-07,
      "loss": 0.4177,
      "step": 12946
    },
    {
      "epoch": 0.8109741774847712,
      "grad_norm": 0.9457709577323047,
      "learning_rate": 9.077500134257444e-07,
      "loss": 0.4186,
      "step": 12947
    },
    {
      "epoch": 0.8110368154841134,
      "grad_norm": 0.8885496926350526,
      "learning_rate": 9.071672467385012e-07,
      "loss": 0.4063,
      "step": 12948
    },
    {
      "epoch": 0.8110994534834557,
      "grad_norm": 0.814911139634187,
      "learning_rate": 9.06584648513012e-07,
      "loss": 0.4152,
      "step": 12949
    },
    {
      "epoch": 0.811162091482798,
      "grad_norm": 0.6007478104764822,
      "learning_rate": 9.060022187732598e-07,
      "loss": 0.4667,
      "step": 12950
    },
    {
      "epoch": 0.8112247294821403,
      "grad_norm": 0.9445928213602939,
      "learning_rate": 9.054199575432165e-07,
      "loss": 0.4106,
      "step": 12951
    },
    {
      "epoch": 0.8112873674814827,
      "grad_norm": 0.6297271641409337,
      "learning_rate": 9.048378648468476e-07,
      "loss": 0.4381,
      "step": 12952
    },
    {
      "epoch": 0.8113500054808249,
      "grad_norm": 0.6489716645772066,
      "learning_rate": 9.042559407081119e-07,
      "loss": 0.438,
      "step": 12953
    },
    {
      "epoch": 0.8114126434801673,
      "grad_norm": 0.8613575070506942,
      "learning_rate": 9.036741851509612e-07,
      "loss": 0.3945,
      "step": 12954
    },
    {
      "epoch": 0.8114752814795095,
      "grad_norm": 0.8487914637902619,
      "learning_rate": 9.030925981993421e-07,
      "loss": 0.3535,
      "step": 12955
    },
    {
      "epoch": 0.8115379194788519,
      "grad_norm": 0.9402340016248002,
      "learning_rate": 9.025111798771923e-07,
      "loss": 0.4175,
      "step": 12956
    },
    {
      "epoch": 0.8116005574781942,
      "grad_norm": 0.9387759525475106,
      "learning_rate": 9.019299302084428e-07,
      "loss": 0.4107,
      "step": 12957
    },
    {
      "epoch": 0.8116631954775364,
      "grad_norm": 0.9358561092019984,
      "learning_rate": 9.01348849217018e-07,
      "loss": 0.3889,
      "step": 12958
    },
    {
      "epoch": 0.8117258334768788,
      "grad_norm": 0.8533177934823342,
      "learning_rate": 9.007679369268341e-07,
      "loss": 0.3992,
      "step": 12959
    },
    {
      "epoch": 0.811788471476221,
      "grad_norm": 0.9164299933438841,
      "learning_rate": 9.001871933618045e-07,
      "loss": 0.4268,
      "step": 12960
    },
    {
      "epoch": 0.8118511094755634,
      "grad_norm": 0.8370472417372697,
      "learning_rate": 8.996066185458302e-07,
      "loss": 0.3809,
      "step": 12961
    },
    {
      "epoch": 0.8119137474749056,
      "grad_norm": 0.9142579451477072,
      "learning_rate": 8.990262125028076e-07,
      "loss": 0.4319,
      "step": 12962
    },
    {
      "epoch": 0.811976385474248,
      "grad_norm": 0.8618813783112794,
      "learning_rate": 8.984459752566288e-07,
      "loss": 0.3616,
      "step": 12963
    },
    {
      "epoch": 0.8120390234735902,
      "grad_norm": 0.886918562005584,
      "learning_rate": 8.978659068311746e-07,
      "loss": 0.3573,
      "step": 12964
    },
    {
      "epoch": 0.8121016614729325,
      "grad_norm": 0.8048287911430033,
      "learning_rate": 8.972860072503209e-07,
      "loss": 0.4161,
      "step": 12965
    },
    {
      "epoch": 0.8121642994722749,
      "grad_norm": 0.7943374825429685,
      "learning_rate": 8.96706276537937e-07,
      "loss": 0.3594,
      "step": 12966
    },
    {
      "epoch": 0.8122269374716171,
      "grad_norm": 0.8176823820249098,
      "learning_rate": 8.961267147178843e-07,
      "loss": 0.3721,
      "step": 12967
    },
    {
      "epoch": 0.8122895754709595,
      "grad_norm": 0.8131346708168399,
      "learning_rate": 8.955473218140165e-07,
      "loss": 0.4034,
      "step": 12968
    },
    {
      "epoch": 0.8123522134703017,
      "grad_norm": 0.871678366545842,
      "learning_rate": 8.94968097850184e-07,
      "loss": 0.4058,
      "step": 12969
    },
    {
      "epoch": 0.8124148514696441,
      "grad_norm": 0.9046927122286955,
      "learning_rate": 8.943890428502266e-07,
      "loss": 0.3839,
      "step": 12970
    },
    {
      "epoch": 0.8124774894689863,
      "grad_norm": 0.8482382798036103,
      "learning_rate": 8.93810156837977e-07,
      "loss": 0.4179,
      "step": 12971
    },
    {
      "epoch": 0.8125401274683287,
      "grad_norm": 0.8620510895178052,
      "learning_rate": 8.932314398372643e-07,
      "loss": 0.4045,
      "step": 12972
    },
    {
      "epoch": 0.812602765467671,
      "grad_norm": 0.8804157947360456,
      "learning_rate": 8.926528918719086e-07,
      "loss": 0.3987,
      "step": 12973
    },
    {
      "epoch": 0.8126654034670132,
      "grad_norm": 0.8524085990577561,
      "learning_rate": 8.920745129657215e-07,
      "loss": 0.3791,
      "step": 12974
    },
    {
      "epoch": 0.8127280414663556,
      "grad_norm": 0.8369324178194784,
      "learning_rate": 8.9149630314251e-07,
      "loss": 0.398,
      "step": 12975
    },
    {
      "epoch": 0.8127906794656978,
      "grad_norm": 0.8256010156346785,
      "learning_rate": 8.909182624260731e-07,
      "loss": 0.3688,
      "step": 12976
    },
    {
      "epoch": 0.8128533174650402,
      "grad_norm": 0.8690968421085195,
      "learning_rate": 8.903403908402025e-07,
      "loss": 0.3594,
      "step": 12977
    },
    {
      "epoch": 0.8129159554643824,
      "grad_norm": 0.9349021316275747,
      "learning_rate": 8.897626884086851e-07,
      "loss": 0.4341,
      "step": 12978
    },
    {
      "epoch": 0.8129785934637248,
      "grad_norm": 0.8687867001559869,
      "learning_rate": 8.891851551552988e-07,
      "loss": 0.3937,
      "step": 12979
    },
    {
      "epoch": 0.8130412314630671,
      "grad_norm": 0.8369727738710956,
      "learning_rate": 8.886077911038127e-07,
      "loss": 0.3829,
      "step": 12980
    },
    {
      "epoch": 0.8131038694624094,
      "grad_norm": 0.8685859503886456,
      "learning_rate": 8.880305962779944e-07,
      "loss": 0.4086,
      "step": 12981
    },
    {
      "epoch": 0.8131665074617517,
      "grad_norm": 0.9022687103660793,
      "learning_rate": 8.874535707016003e-07,
      "loss": 0.3903,
      "step": 12982
    },
    {
      "epoch": 0.8132291454610939,
      "grad_norm": 0.8366285791546489,
      "learning_rate": 8.868767143983803e-07,
      "loss": 0.3996,
      "step": 12983
    },
    {
      "epoch": 0.8132917834604363,
      "grad_norm": 0.8462312130928076,
      "learning_rate": 8.86300027392078e-07,
      "loss": 0.366,
      "step": 12984
    },
    {
      "epoch": 0.8133544214597785,
      "grad_norm": 0.9055323710076557,
      "learning_rate": 8.857235097064304e-07,
      "loss": 0.3837,
      "step": 12985
    },
    {
      "epoch": 0.8134170594591209,
      "grad_norm": 0.7939101884482986,
      "learning_rate": 8.851471613651652e-07,
      "loss": 0.3625,
      "step": 12986
    },
    {
      "epoch": 0.8134796974584632,
      "grad_norm": 0.8378535101090849,
      "learning_rate": 8.845709823920079e-07,
      "loss": 0.3532,
      "step": 12987
    },
    {
      "epoch": 0.8135423354578055,
      "grad_norm": 0.829625878647554,
      "learning_rate": 8.839949728106717e-07,
      "loss": 0.3998,
      "step": 12988
    },
    {
      "epoch": 0.8136049734571478,
      "grad_norm": 0.9242495314117991,
      "learning_rate": 8.834191326448677e-07,
      "loss": 0.4236,
      "step": 12989
    },
    {
      "epoch": 0.81366761145649,
      "grad_norm": 0.8896964905095741,
      "learning_rate": 8.828434619182963e-07,
      "loss": 0.4195,
      "step": 12990
    },
    {
      "epoch": 0.8137302494558324,
      "grad_norm": 0.9071299038983545,
      "learning_rate": 8.822679606546524e-07,
      "loss": 0.3869,
      "step": 12991
    },
    {
      "epoch": 0.8137928874551746,
      "grad_norm": 0.8534689194666545,
      "learning_rate": 8.816926288776228e-07,
      "loss": 0.3862,
      "step": 12992
    },
    {
      "epoch": 0.813855525454517,
      "grad_norm": 0.8744524604632617,
      "learning_rate": 8.811174666108895e-07,
      "loss": 0.3654,
      "step": 12993
    },
    {
      "epoch": 0.8139181634538593,
      "grad_norm": 0.8132148686338835,
      "learning_rate": 8.805424738781259e-07,
      "loss": 0.3802,
      "step": 12994
    },
    {
      "epoch": 0.8139808014532016,
      "grad_norm": 0.8280293985357411,
      "learning_rate": 8.799676507029975e-07,
      "loss": 0.3927,
      "step": 12995
    },
    {
      "epoch": 0.8140434394525439,
      "grad_norm": 0.8653953009115111,
      "learning_rate": 8.793929971091664e-07,
      "loss": 0.3805,
      "step": 12996
    },
    {
      "epoch": 0.8141060774518862,
      "grad_norm": 0.8836067704471523,
      "learning_rate": 8.788185131202826e-07,
      "loss": 0.421,
      "step": 12997
    },
    {
      "epoch": 0.8141687154512285,
      "grad_norm": 0.8302525894708621,
      "learning_rate": 8.782441987599955e-07,
      "loss": 0.3583,
      "step": 12998
    },
    {
      "epoch": 0.8142313534505707,
      "grad_norm": 0.8051489303066962,
      "learning_rate": 8.776700540519423e-07,
      "loss": 0.3701,
      "step": 12999
    },
    {
      "epoch": 0.8142939914499131,
      "grad_norm": 0.8529031124168767,
      "learning_rate": 8.770960790197547e-07,
      "loss": 0.3468,
      "step": 13000
    },
    {
      "epoch": 0.8143566294492554,
      "grad_norm": 0.8812199467817505,
      "learning_rate": 8.765222736870576e-07,
      "loss": 0.3723,
      "step": 13001
    },
    {
      "epoch": 0.8144192674485977,
      "grad_norm": 0.8707640236921274,
      "learning_rate": 8.75948638077469e-07,
      "loss": 0.399,
      "step": 13002
    },
    {
      "epoch": 0.81448190544794,
      "grad_norm": 0.8002282972107165,
      "learning_rate": 8.753751722146003e-07,
      "loss": 0.3572,
      "step": 13003
    },
    {
      "epoch": 0.8145445434472823,
      "grad_norm": 0.8681858876936007,
      "learning_rate": 8.748018761220533e-07,
      "loss": 0.4231,
      "step": 13004
    },
    {
      "epoch": 0.8146071814466246,
      "grad_norm": 0.859180422076744,
      "learning_rate": 8.742287498234281e-07,
      "loss": 0.3782,
      "step": 13005
    },
    {
      "epoch": 0.814669819445967,
      "grad_norm": 0.8075166290307283,
      "learning_rate": 8.736557933423123e-07,
      "loss": 0.3386,
      "step": 13006
    },
    {
      "epoch": 0.8147324574453092,
      "grad_norm": 0.8339716696746414,
      "learning_rate": 8.73083006702291e-07,
      "loss": 0.386,
      "step": 13007
    },
    {
      "epoch": 0.8147950954446515,
      "grad_norm": 0.8781486209708311,
      "learning_rate": 8.725103899269394e-07,
      "loss": 0.409,
      "step": 13008
    },
    {
      "epoch": 0.8148577334439938,
      "grad_norm": 0.9607109205638528,
      "learning_rate": 8.719379430398262e-07,
      "loss": 0.4147,
      "step": 13009
    },
    {
      "epoch": 0.8149203714433361,
      "grad_norm": 0.909489882973935,
      "learning_rate": 8.713656660645136e-07,
      "loss": 0.3803,
      "step": 13010
    },
    {
      "epoch": 0.8149830094426784,
      "grad_norm": 0.9204919274251563,
      "learning_rate": 8.707935590245565e-07,
      "loss": 0.3785,
      "step": 13011
    },
    {
      "epoch": 0.8150456474420207,
      "grad_norm": 0.8719444439198919,
      "learning_rate": 8.702216219435022e-07,
      "loss": 0.3936,
      "step": 13012
    },
    {
      "epoch": 0.815108285441363,
      "grad_norm": 0.8484879036203571,
      "learning_rate": 8.696498548448923e-07,
      "loss": 0.3808,
      "step": 13013
    },
    {
      "epoch": 0.8151709234407053,
      "grad_norm": 0.9584494769010727,
      "learning_rate": 8.6907825775226e-07,
      "loss": 0.3831,
      "step": 13014
    },
    {
      "epoch": 0.8152335614400475,
      "grad_norm": 0.9174475969180309,
      "learning_rate": 8.685068306891353e-07,
      "loss": 0.3691,
      "step": 13015
    },
    {
      "epoch": 0.8152961994393899,
      "grad_norm": 0.8509452532348487,
      "learning_rate": 8.679355736790357e-07,
      "loss": 0.3896,
      "step": 13016
    },
    {
      "epoch": 0.8153588374387322,
      "grad_norm": 0.8253547032205383,
      "learning_rate": 8.673644867454756e-07,
      "loss": 0.3927,
      "step": 13017
    },
    {
      "epoch": 0.8154214754380745,
      "grad_norm": 0.8475849366085396,
      "learning_rate": 8.667935699119595e-07,
      "loss": 0.3966,
      "step": 13018
    },
    {
      "epoch": 0.8154841134374168,
      "grad_norm": 0.8593024455172785,
      "learning_rate": 8.662228232019876e-07,
      "loss": 0.3948,
      "step": 13019
    },
    {
      "epoch": 0.8155467514367591,
      "grad_norm": 0.8142144015353295,
      "learning_rate": 8.656522466390515e-07,
      "loss": 0.323,
      "step": 13020
    },
    {
      "epoch": 0.8156093894361014,
      "grad_norm": 0.8953642284635266,
      "learning_rate": 8.650818402466354e-07,
      "loss": 0.3836,
      "step": 13021
    },
    {
      "epoch": 0.8156720274354438,
      "grad_norm": 0.9342555388192858,
      "learning_rate": 8.645116040482187e-07,
      "loss": 0.4308,
      "step": 13022
    },
    {
      "epoch": 0.815734665434786,
      "grad_norm": 0.8769833725961786,
      "learning_rate": 8.639415380672717e-07,
      "loss": 0.3887,
      "step": 13023
    },
    {
      "epoch": 0.8157973034341283,
      "grad_norm": 0.8669357214936964,
      "learning_rate": 8.633716423272592e-07,
      "loss": 0.3842,
      "step": 13024
    },
    {
      "epoch": 0.8158599414334706,
      "grad_norm": 0.8726161005343261,
      "learning_rate": 8.628019168516377e-07,
      "loss": 0.4271,
      "step": 13025
    },
    {
      "epoch": 0.8159225794328129,
      "grad_norm": 0.8447672495214898,
      "learning_rate": 8.622323616638573e-07,
      "loss": 0.4032,
      "step": 13026
    },
    {
      "epoch": 0.8159852174321552,
      "grad_norm": 0.8844973686622366,
      "learning_rate": 8.616629767873614e-07,
      "loss": 0.4185,
      "step": 13027
    },
    {
      "epoch": 0.8160478554314975,
      "grad_norm": 0.8473088285872525,
      "learning_rate": 8.610937622455851e-07,
      "loss": 0.3693,
      "step": 13028
    },
    {
      "epoch": 0.8161104934308399,
      "grad_norm": 0.8278814662939127,
      "learning_rate": 8.605247180619574e-07,
      "loss": 0.3814,
      "step": 13029
    },
    {
      "epoch": 0.8161731314301821,
      "grad_norm": 0.8116829504077905,
      "learning_rate": 8.599558442598998e-07,
      "loss": 0.3234,
      "step": 13030
    },
    {
      "epoch": 0.8162357694295245,
      "grad_norm": 0.9129802938675701,
      "learning_rate": 8.593871408628291e-07,
      "loss": 0.3843,
      "step": 13031
    },
    {
      "epoch": 0.8162984074288667,
      "grad_norm": 0.8544176450289587,
      "learning_rate": 8.588186078941513e-07,
      "loss": 0.3745,
      "step": 13032
    },
    {
      "epoch": 0.816361045428209,
      "grad_norm": 0.8138029251783616,
      "learning_rate": 8.582502453772696e-07,
      "loss": 0.3919,
      "step": 13033
    },
    {
      "epoch": 0.8164236834275513,
      "grad_norm": 0.8781899454611052,
      "learning_rate": 8.576820533355762e-07,
      "loss": 0.4149,
      "step": 13034
    },
    {
      "epoch": 0.8164863214268936,
      "grad_norm": 0.8725069465021005,
      "learning_rate": 8.571140317924581e-07,
      "loss": 0.4371,
      "step": 13035
    },
    {
      "epoch": 0.816548959426236,
      "grad_norm": 0.6895969097562928,
      "learning_rate": 8.56546180771296e-07,
      "loss": 0.4647,
      "step": 13036
    },
    {
      "epoch": 0.8166115974255782,
      "grad_norm": 0.8791878132397207,
      "learning_rate": 8.559785002954618e-07,
      "loss": 0.3939,
      "step": 13037
    },
    {
      "epoch": 0.8166742354249206,
      "grad_norm": 0.8221224220044754,
      "learning_rate": 8.554109903883218e-07,
      "loss": 0.3808,
      "step": 13038
    },
    {
      "epoch": 0.8167368734242628,
      "grad_norm": 0.7752315434431156,
      "learning_rate": 8.548436510732333e-07,
      "loss": 0.3698,
      "step": 13039
    },
    {
      "epoch": 0.8167995114236052,
      "grad_norm": 0.8551750474260927,
      "learning_rate": 8.542764823735506e-07,
      "loss": 0.3797,
      "step": 13040
    },
    {
      "epoch": 0.8168621494229474,
      "grad_norm": 0.8630305677354914,
      "learning_rate": 8.537094843126164e-07,
      "loss": 0.3602,
      "step": 13041
    },
    {
      "epoch": 0.8169247874222897,
      "grad_norm": 0.6825962876188603,
      "learning_rate": 8.531426569137708e-07,
      "loss": 0.4673,
      "step": 13042
    },
    {
      "epoch": 0.816987425421632,
      "grad_norm": 0.6471080253358855,
      "learning_rate": 8.525760002003431e-07,
      "loss": 0.4578,
      "step": 13043
    },
    {
      "epoch": 0.8170500634209743,
      "grad_norm": 0.8626598204923709,
      "learning_rate": 8.520095141956569e-07,
      "loss": 0.3862,
      "step": 13044
    },
    {
      "epoch": 0.8171127014203167,
      "grad_norm": 0.8359818157556665,
      "learning_rate": 8.514431989230288e-07,
      "loss": 0.3765,
      "step": 13045
    },
    {
      "epoch": 0.8171753394196589,
      "grad_norm": 0.8718828496370461,
      "learning_rate": 8.508770544057688e-07,
      "loss": 0.3867,
      "step": 13046
    },
    {
      "epoch": 0.8172379774190013,
      "grad_norm": 0.8368725586075016,
      "learning_rate": 8.503110806671783e-07,
      "loss": 0.3543,
      "step": 13047
    },
    {
      "epoch": 0.8173006154183435,
      "grad_norm": 0.8607228984242866,
      "learning_rate": 8.497452777305548e-07,
      "loss": 0.3772,
      "step": 13048
    },
    {
      "epoch": 0.8173632534176858,
      "grad_norm": 0.8702121397624598,
      "learning_rate": 8.491796456191864e-07,
      "loss": 0.3864,
      "step": 13049
    },
    {
      "epoch": 0.8174258914170281,
      "grad_norm": 0.9741379598000708,
      "learning_rate": 8.486141843563539e-07,
      "loss": 0.4123,
      "step": 13050
    },
    {
      "epoch": 0.8174885294163704,
      "grad_norm": 0.8736546144219929,
      "learning_rate": 8.480488939653315e-07,
      "loss": 0.4249,
      "step": 13051
    },
    {
      "epoch": 0.8175511674157128,
      "grad_norm": 0.8685691778423474,
      "learning_rate": 8.474837744693876e-07,
      "loss": 0.3806,
      "step": 13052
    },
    {
      "epoch": 0.817613805415055,
      "grad_norm": 0.8481046022557316,
      "learning_rate": 8.469188258917826e-07,
      "loss": 0.3839,
      "step": 13053
    },
    {
      "epoch": 0.8176764434143974,
      "grad_norm": 0.8330477268709156,
      "learning_rate": 8.463540482557697e-07,
      "loss": 0.4029,
      "step": 13054
    },
    {
      "epoch": 0.8177390814137396,
      "grad_norm": 0.941296090713639,
      "learning_rate": 8.457894415845952e-07,
      "loss": 0.3828,
      "step": 13055
    },
    {
      "epoch": 0.817801719413082,
      "grad_norm": 0.8092580384735099,
      "learning_rate": 8.452250059014971e-07,
      "loss": 0.3809,
      "step": 13056
    },
    {
      "epoch": 0.8178643574124242,
      "grad_norm": 0.7748784072113423,
      "learning_rate": 8.446607412297098e-07,
      "loss": 0.359,
      "step": 13057
    },
    {
      "epoch": 0.8179269954117665,
      "grad_norm": 0.9387474317607185,
      "learning_rate": 8.440966475924584e-07,
      "loss": 0.4039,
      "step": 13058
    },
    {
      "epoch": 0.8179896334111089,
      "grad_norm": 0.8927948494553825,
      "learning_rate": 8.435327250129594e-07,
      "loss": 0.3853,
      "step": 13059
    },
    {
      "epoch": 0.8180522714104511,
      "grad_norm": 0.8469859926860324,
      "learning_rate": 8.429689735144242e-07,
      "loss": 0.3666,
      "step": 13060
    },
    {
      "epoch": 0.8181149094097935,
      "grad_norm": 0.8534501882548563,
      "learning_rate": 8.424053931200587e-07,
      "loss": 0.366,
      "step": 13061
    },
    {
      "epoch": 0.8181775474091357,
      "grad_norm": 0.8785465136345493,
      "learning_rate": 8.418419838530594e-07,
      "loss": 0.3961,
      "step": 13062
    },
    {
      "epoch": 0.8182401854084781,
      "grad_norm": 0.9285946598123678,
      "learning_rate": 8.412787457366156e-07,
      "loss": 0.3965,
      "step": 13063
    },
    {
      "epoch": 0.8183028234078203,
      "grad_norm": 0.8513372287710865,
      "learning_rate": 8.407156787939103e-07,
      "loss": 0.3964,
      "step": 13064
    },
    {
      "epoch": 0.8183654614071627,
      "grad_norm": 0.836086568922767,
      "learning_rate": 8.401527830481182e-07,
      "loss": 0.3809,
      "step": 13065
    },
    {
      "epoch": 0.818428099406505,
      "grad_norm": 0.860784805732971,
      "learning_rate": 8.395900585224115e-07,
      "loss": 0.366,
      "step": 13066
    },
    {
      "epoch": 0.8184907374058472,
      "grad_norm": 0.7761201825960998,
      "learning_rate": 8.390275052399499e-07,
      "loss": 0.3445,
      "step": 13067
    },
    {
      "epoch": 0.8185533754051896,
      "grad_norm": 0.8631881607369682,
      "learning_rate": 8.384651232238883e-07,
      "loss": 0.3524,
      "step": 13068
    },
    {
      "epoch": 0.8186160134045318,
      "grad_norm": 0.7585279087601859,
      "learning_rate": 8.379029124973737e-07,
      "loss": 0.3286,
      "step": 13069
    },
    {
      "epoch": 0.8186786514038742,
      "grad_norm": 0.8246997580057457,
      "learning_rate": 8.373408730835486e-07,
      "loss": 0.3853,
      "step": 13070
    },
    {
      "epoch": 0.8187412894032164,
      "grad_norm": 0.8616695512410256,
      "learning_rate": 8.367790050055463e-07,
      "loss": 0.3525,
      "step": 13071
    },
    {
      "epoch": 0.8188039274025588,
      "grad_norm": 0.892024220731985,
      "learning_rate": 8.362173082864927e-07,
      "loss": 0.4017,
      "step": 13072
    },
    {
      "epoch": 0.8188665654019011,
      "grad_norm": 0.8762184839644747,
      "learning_rate": 8.35655782949506e-07,
      "loss": 0.4041,
      "step": 13073
    },
    {
      "epoch": 0.8189292034012433,
      "grad_norm": 0.9387349867677839,
      "learning_rate": 8.350944290177021e-07,
      "loss": 0.3842,
      "step": 13074
    },
    {
      "epoch": 0.8189918414005857,
      "grad_norm": 0.8105985634419185,
      "learning_rate": 8.345332465141836e-07,
      "loss": 0.3087,
      "step": 13075
    },
    {
      "epoch": 0.8190544793999279,
      "grad_norm": 0.9043141586795492,
      "learning_rate": 8.339722354620505e-07,
      "loss": 0.415,
      "step": 13076
    },
    {
      "epoch": 0.8191171173992703,
      "grad_norm": 0.8352443954471811,
      "learning_rate": 8.334113958843936e-07,
      "loss": 0.3592,
      "step": 13077
    },
    {
      "epoch": 0.8191797553986125,
      "grad_norm": 0.8501396926386581,
      "learning_rate": 8.328507278042953e-07,
      "loss": 0.4051,
      "step": 13078
    },
    {
      "epoch": 0.8192423933979549,
      "grad_norm": 0.8517194053105127,
      "learning_rate": 8.322902312448361e-07,
      "loss": 0.3831,
      "step": 13079
    },
    {
      "epoch": 0.8193050313972972,
      "grad_norm": 0.8775932993479559,
      "learning_rate": 8.317299062290841e-07,
      "loss": 0.3924,
      "step": 13080
    },
    {
      "epoch": 0.8193676693966395,
      "grad_norm": 0.8745244536393091,
      "learning_rate": 8.311697527801032e-07,
      "loss": 0.3639,
      "step": 13081
    },
    {
      "epoch": 0.8194303073959818,
      "grad_norm": 0.9448847126435325,
      "learning_rate": 8.306097709209476e-07,
      "loss": 0.391,
      "step": 13082
    },
    {
      "epoch": 0.819492945395324,
      "grad_norm": 0.7801958126407247,
      "learning_rate": 8.300499606746692e-07,
      "loss": 0.3777,
      "step": 13083
    },
    {
      "epoch": 0.8195555833946664,
      "grad_norm": 0.8732064577806804,
      "learning_rate": 8.29490322064308e-07,
      "loss": 0.3843,
      "step": 13084
    },
    {
      "epoch": 0.8196182213940086,
      "grad_norm": 0.8835495990390112,
      "learning_rate": 8.289308551128994e-07,
      "loss": 0.3735,
      "step": 13085
    },
    {
      "epoch": 0.819680859393351,
      "grad_norm": 0.8324790307457738,
      "learning_rate": 8.283715598434711e-07,
      "loss": 0.3854,
      "step": 13086
    },
    {
      "epoch": 0.8197434973926933,
      "grad_norm": 0.8619570721005607,
      "learning_rate": 8.278124362790435e-07,
      "loss": 0.3665,
      "step": 13087
    },
    {
      "epoch": 0.8198061353920356,
      "grad_norm": 0.8572878669101907,
      "learning_rate": 8.272534844426294e-07,
      "loss": 0.3908,
      "step": 13088
    },
    {
      "epoch": 0.8198687733913779,
      "grad_norm": 0.9267951352257306,
      "learning_rate": 8.266947043572376e-07,
      "loss": 0.3754,
      "step": 13089
    },
    {
      "epoch": 0.8199314113907202,
      "grad_norm": 0.8450548034860031,
      "learning_rate": 8.261360960458664e-07,
      "loss": 0.3927,
      "step": 13090
    },
    {
      "epoch": 0.8199940493900625,
      "grad_norm": 0.9032227655414511,
      "learning_rate": 8.255776595315069e-07,
      "loss": 0.3834,
      "step": 13091
    },
    {
      "epoch": 0.8200566873894047,
      "grad_norm": 0.8050974352946285,
      "learning_rate": 8.250193948371466e-07,
      "loss": 0.3865,
      "step": 13092
    },
    {
      "epoch": 0.8201193253887471,
      "grad_norm": 0.8002521095460114,
      "learning_rate": 8.244613019857633e-07,
      "loss": 0.3524,
      "step": 13093
    },
    {
      "epoch": 0.8201819633880894,
      "grad_norm": 0.8998709247600494,
      "learning_rate": 8.239033810003277e-07,
      "loss": 0.4168,
      "step": 13094
    },
    {
      "epoch": 0.8202446013874317,
      "grad_norm": 0.8991488241024856,
      "learning_rate": 8.233456319038041e-07,
      "loss": 0.4156,
      "step": 13095
    },
    {
      "epoch": 0.820307239386774,
      "grad_norm": 0.87922315830797,
      "learning_rate": 8.227880547191497e-07,
      "loss": 0.3858,
      "step": 13096
    },
    {
      "epoch": 0.8203698773861163,
      "grad_norm": 0.8319372529344741,
      "learning_rate": 8.222306494693133e-07,
      "loss": 0.3724,
      "step": 13097
    },
    {
      "epoch": 0.8204325153854586,
      "grad_norm": 0.9950605073822612,
      "learning_rate": 8.216734161772393e-07,
      "loss": 0.3935,
      "step": 13098
    },
    {
      "epoch": 0.8204951533848008,
      "grad_norm": 0.8920368067795043,
      "learning_rate": 8.211163548658624e-07,
      "loss": 0.4141,
      "step": 13099
    },
    {
      "epoch": 0.8205577913841432,
      "grad_norm": 0.882866160259302,
      "learning_rate": 8.205594655581128e-07,
      "loss": 0.3976,
      "step": 13100
    },
    {
      "epoch": 0.8206204293834855,
      "grad_norm": 0.8221515622031429,
      "learning_rate": 8.200027482769113e-07,
      "loss": 0.4062,
      "step": 13101
    },
    {
      "epoch": 0.8206830673828278,
      "grad_norm": 0.7777902244721443,
      "learning_rate": 8.194462030451727e-07,
      "loss": 0.3491,
      "step": 13102
    },
    {
      "epoch": 0.8207457053821701,
      "grad_norm": 0.8405730944075652,
      "learning_rate": 8.188898298858045e-07,
      "loss": 0.3734,
      "step": 13103
    },
    {
      "epoch": 0.8208083433815124,
      "grad_norm": 0.9258188396999897,
      "learning_rate": 8.18333628821707e-07,
      "loss": 0.4358,
      "step": 13104
    },
    {
      "epoch": 0.8208709813808547,
      "grad_norm": 0.6899050317564238,
      "learning_rate": 8.177775998757731e-07,
      "loss": 0.4108,
      "step": 13105
    },
    {
      "epoch": 0.820933619380197,
      "grad_norm": 0.8965673109588665,
      "learning_rate": 8.172217430708884e-07,
      "loss": 0.3873,
      "step": 13106
    },
    {
      "epoch": 0.8209962573795393,
      "grad_norm": 0.8769454127754752,
      "learning_rate": 8.166660584299341e-07,
      "loss": 0.3739,
      "step": 13107
    },
    {
      "epoch": 0.8210588953788815,
      "grad_norm": 0.9268388789878439,
      "learning_rate": 8.161105459757806e-07,
      "loss": 0.423,
      "step": 13108
    },
    {
      "epoch": 0.8211215333782239,
      "grad_norm": 0.9237364096771132,
      "learning_rate": 8.155552057312944e-07,
      "loss": 0.4114,
      "step": 13109
    },
    {
      "epoch": 0.8211841713775662,
      "grad_norm": 0.9103525515739535,
      "learning_rate": 8.150000377193329e-07,
      "loss": 0.3539,
      "step": 13110
    },
    {
      "epoch": 0.8212468093769085,
      "grad_norm": 0.6825789826274761,
      "learning_rate": 8.144450419627464e-07,
      "loss": 0.465,
      "step": 13111
    },
    {
      "epoch": 0.8213094473762508,
      "grad_norm": 0.8914062900169679,
      "learning_rate": 8.138902184843789e-07,
      "loss": 0.4488,
      "step": 13112
    },
    {
      "epoch": 0.8213720853755931,
      "grad_norm": 0.8431198425919294,
      "learning_rate": 8.13335567307067e-07,
      "loss": 0.3739,
      "step": 13113
    },
    {
      "epoch": 0.8214347233749354,
      "grad_norm": 0.8997368229848637,
      "learning_rate": 8.127810884536402e-07,
      "loss": 0.3804,
      "step": 13114
    },
    {
      "epoch": 0.8214973613742778,
      "grad_norm": 0.8950402072620695,
      "learning_rate": 8.122267819469198e-07,
      "loss": 0.3675,
      "step": 13115
    },
    {
      "epoch": 0.82155999937362,
      "grad_norm": 0.8378369095912671,
      "learning_rate": 8.116726478097237e-07,
      "loss": 0.3977,
      "step": 13116
    },
    {
      "epoch": 0.8216226373729623,
      "grad_norm": 0.8746848961521297,
      "learning_rate": 8.111186860648573e-07,
      "loss": 0.3914,
      "step": 13117
    },
    {
      "epoch": 0.8216852753723046,
      "grad_norm": 0.932568195564416,
      "learning_rate": 8.105648967351243e-07,
      "loss": 0.3883,
      "step": 13118
    },
    {
      "epoch": 0.8217479133716469,
      "grad_norm": 0.8716784104243331,
      "learning_rate": 8.10011279843318e-07,
      "loss": 0.4063,
      "step": 13119
    },
    {
      "epoch": 0.8218105513709892,
      "grad_norm": 0.6597891452564006,
      "learning_rate": 8.094578354122251e-07,
      "loss": 0.445,
      "step": 13120
    },
    {
      "epoch": 0.8218731893703315,
      "grad_norm": 0.8469447502588512,
      "learning_rate": 8.089045634646259e-07,
      "loss": 0.3891,
      "step": 13121
    },
    {
      "epoch": 0.8219358273696739,
      "grad_norm": 0.8577807419785699,
      "learning_rate": 8.083514640232925e-07,
      "loss": 0.3977,
      "step": 13122
    },
    {
      "epoch": 0.8219984653690161,
      "grad_norm": 0.851359595290133,
      "learning_rate": 8.077985371109909e-07,
      "loss": 0.3818,
      "step": 13123
    },
    {
      "epoch": 0.8220611033683585,
      "grad_norm": 0.8369662160427659,
      "learning_rate": 8.072457827504782e-07,
      "loss": 0.3634,
      "step": 13124
    },
    {
      "epoch": 0.8221237413677007,
      "grad_norm": 0.9516127528790349,
      "learning_rate": 8.066932009645074e-07,
      "loss": 0.3625,
      "step": 13125
    },
    {
      "epoch": 0.822186379367043,
      "grad_norm": 0.9058016238958901,
      "learning_rate": 8.06140791775824e-07,
      "loss": 0.4261,
      "step": 13126
    },
    {
      "epoch": 0.8222490173663853,
      "grad_norm": 0.9485916218880598,
      "learning_rate": 8.055885552071641e-07,
      "loss": 0.3872,
      "step": 13127
    },
    {
      "epoch": 0.8223116553657276,
      "grad_norm": 0.8431792083167041,
      "learning_rate": 8.050364912812575e-07,
      "loss": 0.3974,
      "step": 13128
    },
    {
      "epoch": 0.82237429336507,
      "grad_norm": 0.8604738623304615,
      "learning_rate": 8.044846000208278e-07,
      "loss": 0.391,
      "step": 13129
    },
    {
      "epoch": 0.8224369313644122,
      "grad_norm": 0.9313073623202568,
      "learning_rate": 8.039328814485909e-07,
      "loss": 0.3943,
      "step": 13130
    },
    {
      "epoch": 0.8224995693637546,
      "grad_norm": 0.8681803076568452,
      "learning_rate": 8.03381335587255e-07,
      "loss": 0.3647,
      "step": 13131
    },
    {
      "epoch": 0.8225622073630968,
      "grad_norm": 0.88235744307142,
      "learning_rate": 8.028299624595209e-07,
      "loss": 0.3793,
      "step": 13132
    },
    {
      "epoch": 0.8226248453624391,
      "grad_norm": 0.8381350321110403,
      "learning_rate": 8.022787620880862e-07,
      "loss": 0.3845,
      "step": 13133
    },
    {
      "epoch": 0.8226874833617814,
      "grad_norm": 0.9528040594647559,
      "learning_rate": 8.017277344956353e-07,
      "loss": 0.414,
      "step": 13134
    },
    {
      "epoch": 0.8227501213611237,
      "grad_norm": 0.8032097423339682,
      "learning_rate": 8.011768797048514e-07,
      "loss": 0.383,
      "step": 13135
    },
    {
      "epoch": 0.822812759360466,
      "grad_norm": 0.936236713476028,
      "learning_rate": 8.006261977384066e-07,
      "loss": 0.4298,
      "step": 13136
    },
    {
      "epoch": 0.8228753973598083,
      "grad_norm": 0.8158751899364992,
      "learning_rate": 8.00075688618967e-07,
      "loss": 0.393,
      "step": 13137
    },
    {
      "epoch": 0.8229380353591507,
      "grad_norm": 0.8498957959911019,
      "learning_rate": 7.995253523691915e-07,
      "loss": 0.4173,
      "step": 13138
    },
    {
      "epoch": 0.8230006733584929,
      "grad_norm": 0.7990877695888677,
      "learning_rate": 7.98975189011732e-07,
      "loss": 0.373,
      "step": 13139
    },
    {
      "epoch": 0.8230633113578353,
      "grad_norm": 0.839319123263176,
      "learning_rate": 7.984251985692338e-07,
      "loss": 0.3811,
      "step": 13140
    },
    {
      "epoch": 0.8231259493571775,
      "grad_norm": 0.8388042676561289,
      "learning_rate": 7.978753810643324e-07,
      "loss": 0.3581,
      "step": 13141
    },
    {
      "epoch": 0.8231885873565198,
      "grad_norm": 0.8178405690427871,
      "learning_rate": 7.973257365196624e-07,
      "loss": 0.3684,
      "step": 13142
    },
    {
      "epoch": 0.8232512253558621,
      "grad_norm": 0.6008436620959752,
      "learning_rate": 7.967762649578431e-07,
      "loss": 0.4606,
      "step": 13143
    },
    {
      "epoch": 0.8233138633552044,
      "grad_norm": 0.883367704503284,
      "learning_rate": 7.962269664014943e-07,
      "loss": 0.4216,
      "step": 13144
    },
    {
      "epoch": 0.8233765013545468,
      "grad_norm": 0.8433278827473459,
      "learning_rate": 7.956778408732241e-07,
      "loss": 0.4281,
      "step": 13145
    },
    {
      "epoch": 0.823439139353889,
      "grad_norm": 0.8185004593870768,
      "learning_rate": 7.951288883956343e-07,
      "loss": 0.3251,
      "step": 13146
    },
    {
      "epoch": 0.8235017773532314,
      "grad_norm": 0.8687165649548859,
      "learning_rate": 7.945801089913196e-07,
      "loss": 0.3591,
      "step": 13147
    },
    {
      "epoch": 0.8235644153525736,
      "grad_norm": 0.8471989818323972,
      "learning_rate": 7.940315026828688e-07,
      "loss": 0.3592,
      "step": 13148
    },
    {
      "epoch": 0.823627053351916,
      "grad_norm": 0.8148501051170058,
      "learning_rate": 7.934830694928614e-07,
      "loss": 0.3972,
      "step": 13149
    },
    {
      "epoch": 0.8236896913512582,
      "grad_norm": 0.8448270608968893,
      "learning_rate": 7.929348094438705e-07,
      "loss": 0.3823,
      "step": 13150
    },
    {
      "epoch": 0.8237523293506005,
      "grad_norm": 0.826289132195028,
      "learning_rate": 7.923867225584653e-07,
      "loss": 0.3383,
      "step": 13151
    },
    {
      "epoch": 0.8238149673499429,
      "grad_norm": 0.8880542565405274,
      "learning_rate": 7.918388088592022e-07,
      "loss": 0.3745,
      "step": 13152
    },
    {
      "epoch": 0.8238776053492851,
      "grad_norm": 0.8328302664288244,
      "learning_rate": 7.912910683686359e-07,
      "loss": 0.3226,
      "step": 13153
    },
    {
      "epoch": 0.8239402433486275,
      "grad_norm": 0.8620386355582756,
      "learning_rate": 7.907435011093101e-07,
      "loss": 0.3726,
      "step": 13154
    },
    {
      "epoch": 0.8240028813479697,
      "grad_norm": 0.8519944767881104,
      "learning_rate": 7.901961071037634e-07,
      "loss": 0.3545,
      "step": 13155
    },
    {
      "epoch": 0.8240655193473121,
      "grad_norm": 0.8619833448771369,
      "learning_rate": 7.896488863745261e-07,
      "loss": 0.3849,
      "step": 13156
    },
    {
      "epoch": 0.8241281573466543,
      "grad_norm": 0.8145696141604697,
      "learning_rate": 7.891018389441224e-07,
      "loss": 0.3738,
      "step": 13157
    },
    {
      "epoch": 0.8241907953459966,
      "grad_norm": 0.664891925344058,
      "learning_rate": 7.885549648350671e-07,
      "loss": 0.4309,
      "step": 13158
    },
    {
      "epoch": 0.824253433345339,
      "grad_norm": 0.8553198487108292,
      "learning_rate": 7.880082640698727e-07,
      "loss": 0.3791,
      "step": 13159
    },
    {
      "epoch": 0.8243160713446812,
      "grad_norm": 0.8585940006994185,
      "learning_rate": 7.874617366710396e-07,
      "loss": 0.3697,
      "step": 13160
    },
    {
      "epoch": 0.8243787093440236,
      "grad_norm": 0.8180329040630634,
      "learning_rate": 7.869153826610626e-07,
      "loss": 0.3803,
      "step": 13161
    },
    {
      "epoch": 0.8244413473433658,
      "grad_norm": 0.8445235036235914,
      "learning_rate": 7.863692020624314e-07,
      "loss": 0.3658,
      "step": 13162
    },
    {
      "epoch": 0.8245039853427082,
      "grad_norm": 0.8704978176682914,
      "learning_rate": 7.858231948976253e-07,
      "loss": 0.4284,
      "step": 13163
    },
    {
      "epoch": 0.8245666233420504,
      "grad_norm": 0.8976888620528521,
      "learning_rate": 7.852773611891195e-07,
      "loss": 0.3972,
      "step": 13164
    },
    {
      "epoch": 0.8246292613413928,
      "grad_norm": 0.8420933143750846,
      "learning_rate": 7.847317009593797e-07,
      "loss": 0.3698,
      "step": 13165
    },
    {
      "epoch": 0.8246918993407351,
      "grad_norm": 0.8323338141721551,
      "learning_rate": 7.841862142308649e-07,
      "loss": 0.4169,
      "step": 13166
    },
    {
      "epoch": 0.8247545373400773,
      "grad_norm": 0.8511665463092252,
      "learning_rate": 7.83640901026027e-07,
      "loss": 0.3641,
      "step": 13167
    },
    {
      "epoch": 0.8248171753394197,
      "grad_norm": 0.7992765163325808,
      "learning_rate": 7.83095761367314e-07,
      "loss": 0.3398,
      "step": 13168
    },
    {
      "epoch": 0.8248798133387619,
      "grad_norm": 0.8382096250779467,
      "learning_rate": 7.825507952771616e-07,
      "loss": 0.3594,
      "step": 13169
    },
    {
      "epoch": 0.8249424513381043,
      "grad_norm": 0.8472874633854303,
      "learning_rate": 7.820060027780002e-07,
      "loss": 0.3778,
      "step": 13170
    },
    {
      "epoch": 0.8250050893374465,
      "grad_norm": 0.7999275202145562,
      "learning_rate": 7.814613838922558e-07,
      "loss": 0.356,
      "step": 13171
    },
    {
      "epoch": 0.8250677273367889,
      "grad_norm": 0.806795477241581,
      "learning_rate": 7.809169386423438e-07,
      "loss": 0.3668,
      "step": 13172
    },
    {
      "epoch": 0.8251303653361312,
      "grad_norm": 0.6367946291426326,
      "learning_rate": 7.80372667050674e-07,
      "loss": 0.4516,
      "step": 13173
    },
    {
      "epoch": 0.8251930033354735,
      "grad_norm": 0.862820268196397,
      "learning_rate": 7.798285691396479e-07,
      "loss": 0.4078,
      "step": 13174
    },
    {
      "epoch": 0.8252556413348158,
      "grad_norm": 0.7982510496828896,
      "learning_rate": 7.792846449316616e-07,
      "loss": 0.3377,
      "step": 13175
    },
    {
      "epoch": 0.825318279334158,
      "grad_norm": 0.7768398048558595,
      "learning_rate": 7.787408944491015e-07,
      "loss": 0.3432,
      "step": 13176
    },
    {
      "epoch": 0.8253809173335004,
      "grad_norm": 0.8389682867343532,
      "learning_rate": 7.781973177143509e-07,
      "loss": 0.3698,
      "step": 13177
    },
    {
      "epoch": 0.8254435553328426,
      "grad_norm": 0.8759669279566097,
      "learning_rate": 7.776539147497819e-07,
      "loss": 0.3979,
      "step": 13178
    },
    {
      "epoch": 0.825506193332185,
      "grad_norm": 0.8842873634214855,
      "learning_rate": 7.771106855777622e-07,
      "loss": 0.4037,
      "step": 13179
    },
    {
      "epoch": 0.8255688313315273,
      "grad_norm": 0.9415770871728483,
      "learning_rate": 7.765676302206487e-07,
      "loss": 0.4212,
      "step": 13180
    },
    {
      "epoch": 0.8256314693308696,
      "grad_norm": 0.879031288427798,
      "learning_rate": 7.760247487007966e-07,
      "loss": 0.415,
      "step": 13181
    },
    {
      "epoch": 0.8256941073302119,
      "grad_norm": 0.8206321756542123,
      "learning_rate": 7.7548204104055e-07,
      "loss": 0.3881,
      "step": 13182
    },
    {
      "epoch": 0.8257567453295541,
      "grad_norm": 0.6218567897908684,
      "learning_rate": 7.749395072622473e-07,
      "loss": 0.4541,
      "step": 13183
    },
    {
      "epoch": 0.8258193833288965,
      "grad_norm": 0.8274335813678547,
      "learning_rate": 7.743971473882167e-07,
      "loss": 0.3938,
      "step": 13184
    },
    {
      "epoch": 0.8258820213282387,
      "grad_norm": 0.9147732804462686,
      "learning_rate": 7.738549614407854e-07,
      "loss": 0.4008,
      "step": 13185
    },
    {
      "epoch": 0.8259446593275811,
      "grad_norm": 0.8068885048117015,
      "learning_rate": 7.733129494422681e-07,
      "loss": 0.3613,
      "step": 13186
    },
    {
      "epoch": 0.8260072973269234,
      "grad_norm": 0.815596144631582,
      "learning_rate": 7.727711114149744e-07,
      "loss": 0.3989,
      "step": 13187
    },
    {
      "epoch": 0.8260699353262657,
      "grad_norm": 0.8379221322810751,
      "learning_rate": 7.722294473812064e-07,
      "loss": 0.3647,
      "step": 13188
    },
    {
      "epoch": 0.826132573325608,
      "grad_norm": 0.876074519774937,
      "learning_rate": 7.716879573632574e-07,
      "loss": 0.4358,
      "step": 13189
    },
    {
      "epoch": 0.8261952113249503,
      "grad_norm": 0.9485913007185944,
      "learning_rate": 7.71146641383419e-07,
      "loss": 0.4451,
      "step": 13190
    },
    {
      "epoch": 0.8262578493242926,
      "grad_norm": 0.8454628887359736,
      "learning_rate": 7.706054994639689e-07,
      "loss": 0.388,
      "step": 13191
    },
    {
      "epoch": 0.8263204873236348,
      "grad_norm": 0.844603812185292,
      "learning_rate": 7.700645316271821e-07,
      "loss": 0.3844,
      "step": 13192
    },
    {
      "epoch": 0.8263831253229772,
      "grad_norm": 0.8460861193420489,
      "learning_rate": 7.695237378953224e-07,
      "loss": 0.3804,
      "step": 13193
    },
    {
      "epoch": 0.8264457633223194,
      "grad_norm": 0.824921602534395,
      "learning_rate": 7.689831182906526e-07,
      "loss": 0.3823,
      "step": 13194
    },
    {
      "epoch": 0.8265084013216618,
      "grad_norm": 0.8048289666268725,
      "learning_rate": 7.684426728354227e-07,
      "loss": 0.3499,
      "step": 13195
    },
    {
      "epoch": 0.8265710393210041,
      "grad_norm": 0.8443033858114539,
      "learning_rate": 7.679024015518782e-07,
      "loss": 0.3998,
      "step": 13196
    },
    {
      "epoch": 0.8266336773203464,
      "grad_norm": 0.8847341969936166,
      "learning_rate": 7.673623044622557e-07,
      "loss": 0.3707,
      "step": 13197
    },
    {
      "epoch": 0.8266963153196887,
      "grad_norm": 0.8355106234958525,
      "learning_rate": 7.668223815887859e-07,
      "loss": 0.361,
      "step": 13198
    },
    {
      "epoch": 0.826758953319031,
      "grad_norm": 0.8177233358807602,
      "learning_rate": 7.662826329536932e-07,
      "loss": 0.3692,
      "step": 13199
    },
    {
      "epoch": 0.8268215913183733,
      "grad_norm": 0.9051109084391217,
      "learning_rate": 7.657430585791937e-07,
      "loss": 0.3644,
      "step": 13200
    },
    {
      "epoch": 0.8268842293177155,
      "grad_norm": 0.8601182256911316,
      "learning_rate": 7.652036584874955e-07,
      "loss": 0.4066,
      "step": 13201
    },
    {
      "epoch": 0.8269468673170579,
      "grad_norm": 0.8027772901708257,
      "learning_rate": 7.646644327007996e-07,
      "loss": 0.397,
      "step": 13202
    },
    {
      "epoch": 0.8270095053164002,
      "grad_norm": 0.880512156222641,
      "learning_rate": 7.641253812413024e-07,
      "loss": 0.3912,
      "step": 13203
    },
    {
      "epoch": 0.8270721433157425,
      "grad_norm": 0.8805756972507103,
      "learning_rate": 7.635865041311913e-07,
      "loss": 0.3687,
      "step": 13204
    },
    {
      "epoch": 0.8271347813150848,
      "grad_norm": 0.8696713327644341,
      "learning_rate": 7.630478013926463e-07,
      "loss": 0.3628,
      "step": 13205
    },
    {
      "epoch": 0.8271974193144271,
      "grad_norm": 0.9128632509441876,
      "learning_rate": 7.625092730478395e-07,
      "loss": 0.4157,
      "step": 13206
    },
    {
      "epoch": 0.8272600573137694,
      "grad_norm": 0.8438632333421012,
      "learning_rate": 7.619709191189367e-07,
      "loss": 0.342,
      "step": 13207
    },
    {
      "epoch": 0.8273226953131116,
      "grad_norm": 0.8126838871171056,
      "learning_rate": 7.614327396280985e-07,
      "loss": 0.3701,
      "step": 13208
    },
    {
      "epoch": 0.827385333312454,
      "grad_norm": 0.8173501560580816,
      "learning_rate": 7.60894734597476e-07,
      "loss": 0.3857,
      "step": 13209
    },
    {
      "epoch": 0.8274479713117963,
      "grad_norm": 0.8953442793857092,
      "learning_rate": 7.603569040492115e-07,
      "loss": 0.358,
      "step": 13210
    },
    {
      "epoch": 0.8275106093111386,
      "grad_norm": 0.8140999543379897,
      "learning_rate": 7.598192480054445e-07,
      "loss": 0.3774,
      "step": 13211
    },
    {
      "epoch": 0.8275732473104809,
      "grad_norm": 0.8039510727530088,
      "learning_rate": 7.592817664883051e-07,
      "loss": 0.3815,
      "step": 13212
    },
    {
      "epoch": 0.8276358853098232,
      "grad_norm": 0.8412355606585095,
      "learning_rate": 7.587444595199145e-07,
      "loss": 0.3918,
      "step": 13213
    },
    {
      "epoch": 0.8276985233091655,
      "grad_norm": 0.884781098714875,
      "learning_rate": 7.582073271223894e-07,
      "loss": 0.4101,
      "step": 13214
    },
    {
      "epoch": 0.8277611613085079,
      "grad_norm": 0.88734828936073,
      "learning_rate": 7.576703693178384e-07,
      "loss": 0.3607,
      "step": 13215
    },
    {
      "epoch": 0.8278237993078501,
      "grad_norm": 0.8136848849869747,
      "learning_rate": 7.571335861283607e-07,
      "loss": 0.3504,
      "step": 13216
    },
    {
      "epoch": 0.8278864373071924,
      "grad_norm": 0.7790699769368101,
      "learning_rate": 7.56596977576054e-07,
      "loss": 0.3707,
      "step": 13217
    },
    {
      "epoch": 0.8279490753065347,
      "grad_norm": 0.8850769364457022,
      "learning_rate": 7.560605436830027e-07,
      "loss": 0.3915,
      "step": 13218
    },
    {
      "epoch": 0.828011713305877,
      "grad_norm": 0.9162774612411548,
      "learning_rate": 7.555242844712863e-07,
      "loss": 0.3966,
      "step": 13219
    },
    {
      "epoch": 0.8280743513052193,
      "grad_norm": 0.8596282775454123,
      "learning_rate": 7.549881999629793e-07,
      "loss": 0.3805,
      "step": 13220
    },
    {
      "epoch": 0.8281369893045616,
      "grad_norm": 0.8592437041604798,
      "learning_rate": 7.544522901801455e-07,
      "loss": 0.3498,
      "step": 13221
    },
    {
      "epoch": 0.828199627303904,
      "grad_norm": 0.8531870542934844,
      "learning_rate": 7.539165551448435e-07,
      "loss": 0.3734,
      "step": 13222
    },
    {
      "epoch": 0.8282622653032462,
      "grad_norm": 0.8725951813155489,
      "learning_rate": 7.533809948791248e-07,
      "loss": 0.3569,
      "step": 13223
    },
    {
      "epoch": 0.8283249033025886,
      "grad_norm": 0.6126662214299605,
      "learning_rate": 7.528456094050324e-07,
      "loss": 0.4486,
      "step": 13224
    },
    {
      "epoch": 0.8283875413019308,
      "grad_norm": 0.8901972345030736,
      "learning_rate": 7.523103987446023e-07,
      "loss": 0.4077,
      "step": 13225
    },
    {
      "epoch": 0.8284501793012731,
      "grad_norm": 0.9408893576676104,
      "learning_rate": 7.517753629198643e-07,
      "loss": 0.4018,
      "step": 13226
    },
    {
      "epoch": 0.8285128173006154,
      "grad_norm": 0.8334120872237485,
      "learning_rate": 7.512405019528418e-07,
      "loss": 0.3622,
      "step": 13227
    },
    {
      "epoch": 0.8285754552999577,
      "grad_norm": 0.8682237923862406,
      "learning_rate": 7.507058158655473e-07,
      "loss": 0.3907,
      "step": 13228
    },
    {
      "epoch": 0.8286380932993,
      "grad_norm": 0.8426318036294099,
      "learning_rate": 7.501713046799913e-07,
      "loss": 0.3065,
      "step": 13229
    },
    {
      "epoch": 0.8287007312986423,
      "grad_norm": 0.8506202745508297,
      "learning_rate": 7.496369684181736e-07,
      "loss": 0.3879,
      "step": 13230
    },
    {
      "epoch": 0.8287633692979847,
      "grad_norm": 0.8553513039751964,
      "learning_rate": 7.491028071020872e-07,
      "loss": 0.3658,
      "step": 13231
    },
    {
      "epoch": 0.8288260072973269,
      "grad_norm": 0.8626253651035126,
      "learning_rate": 7.485688207537179e-07,
      "loss": 0.3735,
      "step": 13232
    },
    {
      "epoch": 0.8288886452966693,
      "grad_norm": 0.8919008659363387,
      "learning_rate": 7.480350093950456e-07,
      "loss": 0.3302,
      "step": 13233
    },
    {
      "epoch": 0.8289512832960115,
      "grad_norm": 0.9819615370382709,
      "learning_rate": 7.475013730480407e-07,
      "loss": 0.3907,
      "step": 13234
    },
    {
      "epoch": 0.8290139212953538,
      "grad_norm": 0.8750407524621948,
      "learning_rate": 7.469679117346684e-07,
      "loss": 0.3614,
      "step": 13235
    },
    {
      "epoch": 0.8290765592946961,
      "grad_norm": 0.7750421422978954,
      "learning_rate": 7.464346254768856e-07,
      "loss": 0.3275,
      "step": 13236
    },
    {
      "epoch": 0.8291391972940384,
      "grad_norm": 0.6495754404939887,
      "learning_rate": 7.459015142966442e-07,
      "loss": 0.473,
      "step": 13237
    },
    {
      "epoch": 0.8292018352933808,
      "grad_norm": 0.8337157855834809,
      "learning_rate": 7.453685782158865e-07,
      "loss": 0.3733,
      "step": 13238
    },
    {
      "epoch": 0.829264473292723,
      "grad_norm": 0.8330997724867111,
      "learning_rate": 7.448358172565473e-07,
      "loss": 0.3625,
      "step": 13239
    },
    {
      "epoch": 0.8293271112920654,
      "grad_norm": 0.8379852396904802,
      "learning_rate": 7.443032314405563e-07,
      "loss": 0.372,
      "step": 13240
    },
    {
      "epoch": 0.8293897492914076,
      "grad_norm": 0.8536318717395994,
      "learning_rate": 7.437708207898337e-07,
      "loss": 0.3869,
      "step": 13241
    },
    {
      "epoch": 0.8294523872907499,
      "grad_norm": 0.8637310357674601,
      "learning_rate": 7.432385853262941e-07,
      "loss": 0.3994,
      "step": 13242
    },
    {
      "epoch": 0.8295150252900922,
      "grad_norm": 0.8108281762152755,
      "learning_rate": 7.427065250718435e-07,
      "loss": 0.3514,
      "step": 13243
    },
    {
      "epoch": 0.8295776632894345,
      "grad_norm": 0.9363124412411521,
      "learning_rate": 7.421746400483837e-07,
      "loss": 0.3881,
      "step": 13244
    },
    {
      "epoch": 0.8296403012887769,
      "grad_norm": 0.8440469545562534,
      "learning_rate": 7.41642930277805e-07,
      "loss": 0.3781,
      "step": 13245
    },
    {
      "epoch": 0.8297029392881191,
      "grad_norm": 0.9446064759510692,
      "learning_rate": 7.411113957819943e-07,
      "loss": 0.4009,
      "step": 13246
    },
    {
      "epoch": 0.8297655772874615,
      "grad_norm": 0.8352025110765492,
      "learning_rate": 7.405800365828291e-07,
      "loss": 0.3861,
      "step": 13247
    },
    {
      "epoch": 0.8298282152868037,
      "grad_norm": 0.8394929643441329,
      "learning_rate": 7.400488527021804e-07,
      "loss": 0.352,
      "step": 13248
    },
    {
      "epoch": 0.8298908532861461,
      "grad_norm": 0.7967884701903237,
      "learning_rate": 7.395178441619116e-07,
      "loss": 0.3219,
      "step": 13249
    },
    {
      "epoch": 0.8299534912854883,
      "grad_norm": 0.7816225001627685,
      "learning_rate": 7.389870109838793e-07,
      "loss": 0.351,
      "step": 13250
    },
    {
      "epoch": 0.8300161292848306,
      "grad_norm": 0.9122976982523106,
      "learning_rate": 7.384563531899325e-07,
      "loss": 0.3891,
      "step": 13251
    },
    {
      "epoch": 0.830078767284173,
      "grad_norm": 0.8417115899125148,
      "learning_rate": 7.379258708019116e-07,
      "loss": 0.3872,
      "step": 13252
    },
    {
      "epoch": 0.8301414052835152,
      "grad_norm": 0.905047744583743,
      "learning_rate": 7.373955638416546e-07,
      "loss": 0.4108,
      "step": 13253
    },
    {
      "epoch": 0.8302040432828576,
      "grad_norm": 0.8938652907997369,
      "learning_rate": 7.368654323309859e-07,
      "loss": 0.4417,
      "step": 13254
    },
    {
      "epoch": 0.8302666812821998,
      "grad_norm": 0.8405740177766212,
      "learning_rate": 7.363354762917285e-07,
      "loss": 0.3719,
      "step": 13255
    },
    {
      "epoch": 0.8303293192815422,
      "grad_norm": 0.8952436531434531,
      "learning_rate": 7.358056957456944e-07,
      "loss": 0.4146,
      "step": 13256
    },
    {
      "epoch": 0.8303919572808844,
      "grad_norm": 0.8142069948382233,
      "learning_rate": 7.352760907146889e-07,
      "loss": 0.3994,
      "step": 13257
    },
    {
      "epoch": 0.8304545952802268,
      "grad_norm": 0.7926373023722658,
      "learning_rate": 7.347466612205112e-07,
      "loss": 0.3706,
      "step": 13258
    },
    {
      "epoch": 0.8305172332795691,
      "grad_norm": 0.8824087193882049,
      "learning_rate": 7.342174072849528e-07,
      "loss": 0.4041,
      "step": 13259
    },
    {
      "epoch": 0.8305798712789113,
      "grad_norm": 0.8552266974933367,
      "learning_rate": 7.33688328929798e-07,
      "loss": 0.3984,
      "step": 13260
    },
    {
      "epoch": 0.8306425092782537,
      "grad_norm": 0.8412234809241439,
      "learning_rate": 7.331594261768215e-07,
      "loss": 0.3934,
      "step": 13261
    },
    {
      "epoch": 0.8307051472775959,
      "grad_norm": 0.6217991515962579,
      "learning_rate": 7.326306990477955e-07,
      "loss": 0.4524,
      "step": 13262
    },
    {
      "epoch": 0.8307677852769383,
      "grad_norm": 0.8310482555993631,
      "learning_rate": 7.321021475644835e-07,
      "loss": 0.4104,
      "step": 13263
    },
    {
      "epoch": 0.8308304232762805,
      "grad_norm": 0.9625646662162506,
      "learning_rate": 7.315737717486388e-07,
      "loss": 0.4423,
      "step": 13264
    },
    {
      "epoch": 0.8308930612756229,
      "grad_norm": 0.914483735237155,
      "learning_rate": 7.3104557162201e-07,
      "loss": 0.3729,
      "step": 13265
    },
    {
      "epoch": 0.8309556992749652,
      "grad_norm": 0.7991791642826039,
      "learning_rate": 7.305175472063375e-07,
      "loss": 0.3343,
      "step": 13266
    },
    {
      "epoch": 0.8310183372743074,
      "grad_norm": 0.9726334787712739,
      "learning_rate": 7.299896985233556e-07,
      "loss": 0.401,
      "step": 13267
    },
    {
      "epoch": 0.8310809752736498,
      "grad_norm": 0.903143532941678,
      "learning_rate": 7.294620255947899e-07,
      "loss": 0.4112,
      "step": 13268
    },
    {
      "epoch": 0.831143613272992,
      "grad_norm": 0.9163218777526667,
      "learning_rate": 7.289345284423594e-07,
      "loss": 0.3997,
      "step": 13269
    },
    {
      "epoch": 0.8312062512723344,
      "grad_norm": 0.6304682545907689,
      "learning_rate": 7.284072070877768e-07,
      "loss": 0.4392,
      "step": 13270
    },
    {
      "epoch": 0.8312688892716766,
      "grad_norm": 0.8201485857204753,
      "learning_rate": 7.278800615527471e-07,
      "loss": 0.3664,
      "step": 13271
    },
    {
      "epoch": 0.831331527271019,
      "grad_norm": 0.8522246839030108,
      "learning_rate": 7.273530918589655e-07,
      "loss": 0.3604,
      "step": 13272
    },
    {
      "epoch": 0.8313941652703613,
      "grad_norm": 0.914467420047328,
      "learning_rate": 7.268262980281249e-07,
      "loss": 0.4191,
      "step": 13273
    },
    {
      "epoch": 0.8314568032697036,
      "grad_norm": 0.773818004349468,
      "learning_rate": 7.262996800819066e-07,
      "loss": 0.354,
      "step": 13274
    },
    {
      "epoch": 0.8315194412690459,
      "grad_norm": 0.8456549105164357,
      "learning_rate": 7.257732380419874e-07,
      "loss": 0.3646,
      "step": 13275
    },
    {
      "epoch": 0.8315820792683881,
      "grad_norm": 0.927235587245696,
      "learning_rate": 7.252469719300347e-07,
      "loss": 0.4051,
      "step": 13276
    },
    {
      "epoch": 0.8316447172677305,
      "grad_norm": 0.6006713549325716,
      "learning_rate": 7.247208817677099e-07,
      "loss": 0.4524,
      "step": 13277
    },
    {
      "epoch": 0.8317073552670727,
      "grad_norm": 0.8979575404213022,
      "learning_rate": 7.241949675766663e-07,
      "loss": 0.3573,
      "step": 13278
    },
    {
      "epoch": 0.8317699932664151,
      "grad_norm": 0.8207427759705662,
      "learning_rate": 7.236692293785524e-07,
      "loss": 0.378,
      "step": 13279
    },
    {
      "epoch": 0.8318326312657573,
      "grad_norm": 0.6328326725708041,
      "learning_rate": 7.23143667195007e-07,
      "loss": 0.4186,
      "step": 13280
    },
    {
      "epoch": 0.8318952692650997,
      "grad_norm": 0.8331961214040625,
      "learning_rate": 7.226182810476606e-07,
      "loss": 0.3903,
      "step": 13281
    },
    {
      "epoch": 0.831957907264442,
      "grad_norm": 0.8411052066265421,
      "learning_rate": 7.220930709581409e-07,
      "loss": 0.4178,
      "step": 13282
    },
    {
      "epoch": 0.8320205452637843,
      "grad_norm": 0.9357320480441172,
      "learning_rate": 7.215680369480649e-07,
      "loss": 0.3939,
      "step": 13283
    },
    {
      "epoch": 0.8320831832631266,
      "grad_norm": 0.8776458164979177,
      "learning_rate": 7.210431790390421e-07,
      "loss": 0.4113,
      "step": 13284
    },
    {
      "epoch": 0.8321458212624688,
      "grad_norm": 0.8435053486163226,
      "learning_rate": 7.205184972526763e-07,
      "loss": 0.365,
      "step": 13285
    },
    {
      "epoch": 0.8322084592618112,
      "grad_norm": 0.8669467002681015,
      "learning_rate": 7.199939916105636e-07,
      "loss": 0.3709,
      "step": 13286
    },
    {
      "epoch": 0.8322710972611534,
      "grad_norm": 0.8798944715901688,
      "learning_rate": 7.194696621342912e-07,
      "loss": 0.3737,
      "step": 13287
    },
    {
      "epoch": 0.8323337352604958,
      "grad_norm": 0.8348874326692152,
      "learning_rate": 7.189455088454433e-07,
      "loss": 0.3481,
      "step": 13288
    },
    {
      "epoch": 0.8323963732598381,
      "grad_norm": 0.8959455594270543,
      "learning_rate": 7.184215317655929e-07,
      "loss": 0.3947,
      "step": 13289
    },
    {
      "epoch": 0.8324590112591804,
      "grad_norm": 0.844639525896671,
      "learning_rate": 7.17897730916306e-07,
      "loss": 0.3305,
      "step": 13290
    },
    {
      "epoch": 0.8325216492585227,
      "grad_norm": 0.8329633970386181,
      "learning_rate": 7.173741063191442e-07,
      "loss": 0.3945,
      "step": 13291
    },
    {
      "epoch": 0.8325842872578649,
      "grad_norm": 0.9173274835862989,
      "learning_rate": 7.168506579956591e-07,
      "loss": 0.3914,
      "step": 13292
    },
    {
      "epoch": 0.8326469252572073,
      "grad_norm": 0.6069555127876489,
      "learning_rate": 7.163273859673964e-07,
      "loss": 0.4205,
      "step": 13293
    },
    {
      "epoch": 0.8327095632565495,
      "grad_norm": 0.8192407557565851,
      "learning_rate": 7.158042902558932e-07,
      "loss": 0.3756,
      "step": 13294
    },
    {
      "epoch": 0.8327722012558919,
      "grad_norm": 0.9339642252931983,
      "learning_rate": 7.152813708826789e-07,
      "loss": 0.4306,
      "step": 13295
    },
    {
      "epoch": 0.8328348392552342,
      "grad_norm": 0.872604078145448,
      "learning_rate": 7.147586278692803e-07,
      "loss": 0.3974,
      "step": 13296
    },
    {
      "epoch": 0.8328974772545765,
      "grad_norm": 0.591653727060034,
      "learning_rate": 7.142360612372123e-07,
      "loss": 0.43,
      "step": 13297
    },
    {
      "epoch": 0.8329601152539188,
      "grad_norm": 0.865684680978511,
      "learning_rate": 7.137136710079828e-07,
      "loss": 0.3634,
      "step": 13298
    },
    {
      "epoch": 0.8330227532532611,
      "grad_norm": 0.9203161282543497,
      "learning_rate": 7.131914572030934e-07,
      "loss": 0.3656,
      "step": 13299
    },
    {
      "epoch": 0.8330853912526034,
      "grad_norm": 0.7955463189648971,
      "learning_rate": 7.126694198440404e-07,
      "loss": 0.3649,
      "step": 13300
    },
    {
      "epoch": 0.8331480292519456,
      "grad_norm": 0.807346399625569,
      "learning_rate": 7.121475589523102e-07,
      "loss": 0.3531,
      "step": 13301
    },
    {
      "epoch": 0.833210667251288,
      "grad_norm": 0.9652389817031489,
      "learning_rate": 7.116258745493815e-07,
      "loss": 0.4578,
      "step": 13302
    },
    {
      "epoch": 0.8332733052506303,
      "grad_norm": 0.8057869898046381,
      "learning_rate": 7.111043666567286e-07,
      "loss": 0.3392,
      "step": 13303
    },
    {
      "epoch": 0.8333359432499726,
      "grad_norm": 0.8550233218378096,
      "learning_rate": 7.105830352958143e-07,
      "loss": 0.3665,
      "step": 13304
    },
    {
      "epoch": 0.8333985812493149,
      "grad_norm": 0.8632052569976382,
      "learning_rate": 7.100618804880999e-07,
      "loss": 0.4304,
      "step": 13305
    },
    {
      "epoch": 0.8334612192486572,
      "grad_norm": 0.8992261008642877,
      "learning_rate": 7.095409022550348e-07,
      "loss": 0.363,
      "step": 13306
    },
    {
      "epoch": 0.8335238572479995,
      "grad_norm": 0.8434924019831872,
      "learning_rate": 7.090201006180624e-07,
      "loss": 0.3468,
      "step": 13307
    },
    {
      "epoch": 0.8335864952473419,
      "grad_norm": 0.9127263616731589,
      "learning_rate": 7.08499475598618e-07,
      "loss": 0.3747,
      "step": 13308
    },
    {
      "epoch": 0.8336491332466841,
      "grad_norm": 0.868388675116896,
      "learning_rate": 7.079790272181325e-07,
      "loss": 0.4136,
      "step": 13309
    },
    {
      "epoch": 0.8337117712460264,
      "grad_norm": 0.8749421293618557,
      "learning_rate": 7.074587554980272e-07,
      "loss": 0.3767,
      "step": 13310
    },
    {
      "epoch": 0.8337744092453687,
      "grad_norm": 0.8451485412895492,
      "learning_rate": 7.069386604597161e-07,
      "loss": 0.4145,
      "step": 13311
    },
    {
      "epoch": 0.833837047244711,
      "grad_norm": 0.6754659171889088,
      "learning_rate": 7.064187421246066e-07,
      "loss": 0.4503,
      "step": 13312
    },
    {
      "epoch": 0.8338996852440533,
      "grad_norm": 0.6341610233523386,
      "learning_rate": 7.058990005140975e-07,
      "loss": 0.4581,
      "step": 13313
    },
    {
      "epoch": 0.8339623232433956,
      "grad_norm": 0.8518130438805624,
      "learning_rate": 7.053794356495841e-07,
      "loss": 0.3636,
      "step": 13314
    },
    {
      "epoch": 0.834024961242738,
      "grad_norm": 0.8655682656311624,
      "learning_rate": 7.048600475524497e-07,
      "loss": 0.3661,
      "step": 13315
    },
    {
      "epoch": 0.8340875992420802,
      "grad_norm": 0.8127274876041458,
      "learning_rate": 7.04340836244073e-07,
      "loss": 0.3683,
      "step": 13316
    },
    {
      "epoch": 0.8341502372414225,
      "grad_norm": 0.8674867213136642,
      "learning_rate": 7.038218017458248e-07,
      "loss": 0.3917,
      "step": 13317
    },
    {
      "epoch": 0.8342128752407648,
      "grad_norm": 0.80390267647694,
      "learning_rate": 7.033029440790679e-07,
      "loss": 0.3289,
      "step": 13318
    },
    {
      "epoch": 0.8342755132401071,
      "grad_norm": 0.8042772938268499,
      "learning_rate": 7.027842632651604e-07,
      "loss": 0.3439,
      "step": 13319
    },
    {
      "epoch": 0.8343381512394494,
      "grad_norm": 0.8392862258908023,
      "learning_rate": 7.022657593254501e-07,
      "loss": 0.3565,
      "step": 13320
    },
    {
      "epoch": 0.8344007892387917,
      "grad_norm": 0.6239198753538151,
      "learning_rate": 7.017474322812773e-07,
      "loss": 0.4431,
      "step": 13321
    },
    {
      "epoch": 0.834463427238134,
      "grad_norm": 0.7867323776931664,
      "learning_rate": 7.012292821539796e-07,
      "loss": 0.4038,
      "step": 13322
    },
    {
      "epoch": 0.8345260652374763,
      "grad_norm": 0.8356350368533002,
      "learning_rate": 7.007113089648826e-07,
      "loss": 0.3801,
      "step": 13323
    },
    {
      "epoch": 0.8345887032368187,
      "grad_norm": 0.8936255588857506,
      "learning_rate": 7.001935127353054e-07,
      "loss": 0.4081,
      "step": 13324
    },
    {
      "epoch": 0.8346513412361609,
      "grad_norm": 0.8141195350490259,
      "learning_rate": 6.996758934865621e-07,
      "loss": 0.3573,
      "step": 13325
    },
    {
      "epoch": 0.8347139792355032,
      "grad_norm": 0.8856845166252009,
      "learning_rate": 6.991584512399569e-07,
      "loss": 0.3885,
      "step": 13326
    },
    {
      "epoch": 0.8347766172348455,
      "grad_norm": 0.8504351873696487,
      "learning_rate": 6.986411860167869e-07,
      "loss": 0.3691,
      "step": 13327
    },
    {
      "epoch": 0.8348392552341878,
      "grad_norm": 0.8461589051060646,
      "learning_rate": 6.981240978383452e-07,
      "loss": 0.4166,
      "step": 13328
    },
    {
      "epoch": 0.8349018932335301,
      "grad_norm": 0.7925263045307597,
      "learning_rate": 6.976071867259144e-07,
      "loss": 0.3683,
      "step": 13329
    },
    {
      "epoch": 0.8349645312328724,
      "grad_norm": 0.8206527307366399,
      "learning_rate": 6.970904527007688e-07,
      "loss": 0.3626,
      "step": 13330
    },
    {
      "epoch": 0.8350271692322148,
      "grad_norm": 0.8625524410959958,
      "learning_rate": 6.9657389578418e-07,
      "loss": 0.3493,
      "step": 13331
    },
    {
      "epoch": 0.835089807231557,
      "grad_norm": 0.9317167244829467,
      "learning_rate": 6.960575159974087e-07,
      "loss": 0.4004,
      "step": 13332
    },
    {
      "epoch": 0.8351524452308994,
      "grad_norm": 0.7839222482375139,
      "learning_rate": 6.955413133617089e-07,
      "loss": 0.3862,
      "step": 13333
    },
    {
      "epoch": 0.8352150832302416,
      "grad_norm": 0.86047183790989,
      "learning_rate": 6.950252878983271e-07,
      "loss": 0.4036,
      "step": 13334
    },
    {
      "epoch": 0.8352777212295839,
      "grad_norm": 0.8289127945788617,
      "learning_rate": 6.945094396285041e-07,
      "loss": 0.3636,
      "step": 13335
    },
    {
      "epoch": 0.8353403592289262,
      "grad_norm": 0.851164857339236,
      "learning_rate": 6.939937685734704e-07,
      "loss": 0.3858,
      "step": 13336
    },
    {
      "epoch": 0.8354029972282685,
      "grad_norm": 0.8398507166420847,
      "learning_rate": 6.934782747544533e-07,
      "loss": 0.3717,
      "step": 13337
    },
    {
      "epoch": 0.8354656352276109,
      "grad_norm": 0.8575418762566672,
      "learning_rate": 6.9296295819267e-07,
      "loss": 0.3874,
      "step": 13338
    },
    {
      "epoch": 0.8355282732269531,
      "grad_norm": 0.8423344282108245,
      "learning_rate": 6.924478189093292e-07,
      "loss": 0.3522,
      "step": 13339
    },
    {
      "epoch": 0.8355909112262955,
      "grad_norm": 0.8393884090182125,
      "learning_rate": 6.919328569256372e-07,
      "loss": 0.3993,
      "step": 13340
    },
    {
      "epoch": 0.8356535492256377,
      "grad_norm": 0.7989117771673035,
      "learning_rate": 6.914180722627883e-07,
      "loss": 0.3444,
      "step": 13341
    },
    {
      "epoch": 0.8357161872249801,
      "grad_norm": 0.8131763607681222,
      "learning_rate": 6.909034649419716e-07,
      "loss": 0.3927,
      "step": 13342
    },
    {
      "epoch": 0.8357788252243223,
      "grad_norm": 0.8826774137728727,
      "learning_rate": 6.903890349843678e-07,
      "loss": 0.3636,
      "step": 13343
    },
    {
      "epoch": 0.8358414632236646,
      "grad_norm": 0.8711432128694586,
      "learning_rate": 6.898747824111512e-07,
      "loss": 0.382,
      "step": 13344
    },
    {
      "epoch": 0.835904101223007,
      "grad_norm": 0.8251002896982228,
      "learning_rate": 6.893607072434877e-07,
      "loss": 0.391,
      "step": 13345
    },
    {
      "epoch": 0.8359667392223492,
      "grad_norm": 0.8544252666802568,
      "learning_rate": 6.88846809502538e-07,
      "loss": 0.4016,
      "step": 13346
    },
    {
      "epoch": 0.8360293772216916,
      "grad_norm": 0.8737318130745472,
      "learning_rate": 6.883330892094537e-07,
      "loss": 0.4057,
      "step": 13347
    },
    {
      "epoch": 0.8360920152210338,
      "grad_norm": 0.8569779576611987,
      "learning_rate": 6.878195463853798e-07,
      "loss": 0.3886,
      "step": 13348
    },
    {
      "epoch": 0.8361546532203762,
      "grad_norm": 0.8805777031562242,
      "learning_rate": 6.873061810514548e-07,
      "loss": 0.3561,
      "step": 13349
    },
    {
      "epoch": 0.8362172912197184,
      "grad_norm": 0.8514561512600024,
      "learning_rate": 6.86792993228807e-07,
      "loss": 0.3563,
      "step": 13350
    },
    {
      "epoch": 0.8362799292190607,
      "grad_norm": 0.8930537479545364,
      "learning_rate": 6.862799829385603e-07,
      "loss": 0.3568,
      "step": 13351
    },
    {
      "epoch": 0.836342567218403,
      "grad_norm": 0.8069125638326715,
      "learning_rate": 6.857671502018304e-07,
      "loss": 0.3796,
      "step": 13352
    },
    {
      "epoch": 0.8364052052177453,
      "grad_norm": 0.895469951336712,
      "learning_rate": 6.852544950397256e-07,
      "loss": 0.4167,
      "step": 13353
    },
    {
      "epoch": 0.8364678432170877,
      "grad_norm": 0.6273540279974923,
      "learning_rate": 6.847420174733449e-07,
      "loss": 0.4365,
      "step": 13354
    },
    {
      "epoch": 0.8365304812164299,
      "grad_norm": 0.8266919184750026,
      "learning_rate": 6.842297175237855e-07,
      "loss": 0.4082,
      "step": 13355
    },
    {
      "epoch": 0.8365931192157723,
      "grad_norm": 0.878479599568586,
      "learning_rate": 6.837175952121305e-07,
      "loss": 0.3985,
      "step": 13356
    },
    {
      "epoch": 0.8366557572151145,
      "grad_norm": 0.7659752196379502,
      "learning_rate": 6.832056505594614e-07,
      "loss": 0.3777,
      "step": 13357
    },
    {
      "epoch": 0.8367183952144569,
      "grad_norm": 0.8847051606578954,
      "learning_rate": 6.826938835868491e-07,
      "loss": 0.3531,
      "step": 13358
    },
    {
      "epoch": 0.8367810332137992,
      "grad_norm": 0.899375404251128,
      "learning_rate": 6.821822943153577e-07,
      "loss": 0.3549,
      "step": 13359
    },
    {
      "epoch": 0.8368436712131414,
      "grad_norm": 0.9197454371111351,
      "learning_rate": 6.816708827660451e-07,
      "loss": 0.3784,
      "step": 13360
    },
    {
      "epoch": 0.8369063092124838,
      "grad_norm": 0.8878260031557852,
      "learning_rate": 6.811596489599598e-07,
      "loss": 0.3994,
      "step": 13361
    },
    {
      "epoch": 0.836968947211826,
      "grad_norm": 0.8366915630196498,
      "learning_rate": 6.80648592918145e-07,
      "loss": 0.4154,
      "step": 13362
    },
    {
      "epoch": 0.8370315852111684,
      "grad_norm": 0.8058486626034892,
      "learning_rate": 6.80137714661635e-07,
      "loss": 0.3531,
      "step": 13363
    },
    {
      "epoch": 0.8370942232105106,
      "grad_norm": 0.9815596964757765,
      "learning_rate": 6.796270142114586e-07,
      "loss": 0.3769,
      "step": 13364
    },
    {
      "epoch": 0.837156861209853,
      "grad_norm": 0.8187409920325248,
      "learning_rate": 6.791164915886373e-07,
      "loss": 0.3604,
      "step": 13365
    },
    {
      "epoch": 0.8372194992091953,
      "grad_norm": 0.8463553123287934,
      "learning_rate": 6.78606146814183e-07,
      "loss": 0.3646,
      "step": 13366
    },
    {
      "epoch": 0.8372821372085376,
      "grad_norm": 0.8715499072204559,
      "learning_rate": 6.780959799091013e-07,
      "loss": 0.4205,
      "step": 13367
    },
    {
      "epoch": 0.8373447752078799,
      "grad_norm": 0.9205973691453876,
      "learning_rate": 6.775859908943921e-07,
      "loss": 0.3814,
      "step": 13368
    },
    {
      "epoch": 0.8374074132072221,
      "grad_norm": 0.8038204871238435,
      "learning_rate": 6.770761797910453e-07,
      "loss": 0.3541,
      "step": 13369
    },
    {
      "epoch": 0.8374700512065645,
      "grad_norm": 0.7747659623789664,
      "learning_rate": 6.765665466200455e-07,
      "loss": 0.3848,
      "step": 13370
    },
    {
      "epoch": 0.8375326892059067,
      "grad_norm": 0.9392436184187727,
      "learning_rate": 6.76057091402369e-07,
      "loss": 0.4206,
      "step": 13371
    },
    {
      "epoch": 0.8375953272052491,
      "grad_norm": 0.8537503472646294,
      "learning_rate": 6.755478141589844e-07,
      "loss": 0.3416,
      "step": 13372
    },
    {
      "epoch": 0.8376579652045913,
      "grad_norm": 0.8829636562927211,
      "learning_rate": 6.75038714910854e-07,
      "loss": 0.3886,
      "step": 13373
    },
    {
      "epoch": 0.8377206032039337,
      "grad_norm": 0.9067802038629943,
      "learning_rate": 6.745297936789341e-07,
      "loss": 0.4277,
      "step": 13374
    },
    {
      "epoch": 0.837783241203276,
      "grad_norm": 0.8696998221476943,
      "learning_rate": 6.74021050484171e-07,
      "loss": 0.3824,
      "step": 13375
    },
    {
      "epoch": 0.8378458792026182,
      "grad_norm": 0.8228442136922991,
      "learning_rate": 6.735124853475039e-07,
      "loss": 0.3922,
      "step": 13376
    },
    {
      "epoch": 0.8379085172019606,
      "grad_norm": 0.8390721414624841,
      "learning_rate": 6.730040982898661e-07,
      "loss": 0.3835,
      "step": 13377
    },
    {
      "epoch": 0.8379711552013028,
      "grad_norm": 0.9071869937076835,
      "learning_rate": 6.724958893321832e-07,
      "loss": 0.3877,
      "step": 13378
    },
    {
      "epoch": 0.8380337932006452,
      "grad_norm": 0.802235818599097,
      "learning_rate": 6.719878584953721e-07,
      "loss": 0.3805,
      "step": 13379
    },
    {
      "epoch": 0.8380964311999874,
      "grad_norm": 0.8601543573921945,
      "learning_rate": 6.714800058003435e-07,
      "loss": 0.3706,
      "step": 13380
    },
    {
      "epoch": 0.8381590691993298,
      "grad_norm": 0.8557390001949448,
      "learning_rate": 6.709723312680017e-07,
      "loss": 0.3817,
      "step": 13381
    },
    {
      "epoch": 0.8382217071986721,
      "grad_norm": 1.3013802424528618,
      "learning_rate": 6.704648349192417e-07,
      "loss": 0.4012,
      "step": 13382
    },
    {
      "epoch": 0.8382843451980144,
      "grad_norm": 0.9724764526116069,
      "learning_rate": 6.699575167749534e-07,
      "loss": 0.4009,
      "step": 13383
    },
    {
      "epoch": 0.8383469831973567,
      "grad_norm": 0.897661419638479,
      "learning_rate": 6.694503768560178e-07,
      "loss": 0.3995,
      "step": 13384
    },
    {
      "epoch": 0.8384096211966989,
      "grad_norm": 0.8123614443760716,
      "learning_rate": 6.689434151833085e-07,
      "loss": 0.3469,
      "step": 13385
    },
    {
      "epoch": 0.8384722591960413,
      "grad_norm": 0.9148886154399811,
      "learning_rate": 6.684366317776919e-07,
      "loss": 0.3343,
      "step": 13386
    },
    {
      "epoch": 0.8385348971953835,
      "grad_norm": 0.8420655858934348,
      "learning_rate": 6.679300266600275e-07,
      "loss": 0.3673,
      "step": 13387
    },
    {
      "epoch": 0.8385975351947259,
      "grad_norm": 0.5674007869930938,
      "learning_rate": 6.674235998511669e-07,
      "loss": 0.4312,
      "step": 13388
    },
    {
      "epoch": 0.8386601731940682,
      "grad_norm": 0.881880779663093,
      "learning_rate": 6.669173513719546e-07,
      "loss": 0.3972,
      "step": 13389
    },
    {
      "epoch": 0.8387228111934105,
      "grad_norm": 0.8210278858282445,
      "learning_rate": 6.664112812432294e-07,
      "loss": 0.3499,
      "step": 13390
    },
    {
      "epoch": 0.8387854491927528,
      "grad_norm": 0.9353567103964641,
      "learning_rate": 6.659053894858186e-07,
      "loss": 0.4032,
      "step": 13391
    },
    {
      "epoch": 0.8388480871920951,
      "grad_norm": 0.8565213950948777,
      "learning_rate": 6.653996761205484e-07,
      "loss": 0.3624,
      "step": 13392
    },
    {
      "epoch": 0.8389107251914374,
      "grad_norm": 0.8533902968869246,
      "learning_rate": 6.648941411682313e-07,
      "loss": 0.373,
      "step": 13393
    },
    {
      "epoch": 0.8389733631907796,
      "grad_norm": 0.8083031424577041,
      "learning_rate": 6.643887846496766e-07,
      "loss": 0.377,
      "step": 13394
    },
    {
      "epoch": 0.839036001190122,
      "grad_norm": 0.9077269372145351,
      "learning_rate": 6.638836065856835e-07,
      "loss": 0.4189,
      "step": 13395
    },
    {
      "epoch": 0.8390986391894643,
      "grad_norm": 0.8445598275685782,
      "learning_rate": 6.633786069970466e-07,
      "loss": 0.3672,
      "step": 13396
    },
    {
      "epoch": 0.8391612771888066,
      "grad_norm": 0.807201928917584,
      "learning_rate": 6.628737859045508e-07,
      "loss": 0.3186,
      "step": 13397
    },
    {
      "epoch": 0.8392239151881489,
      "grad_norm": 0.9158004377601894,
      "learning_rate": 6.623691433289741e-07,
      "loss": 0.3953,
      "step": 13398
    },
    {
      "epoch": 0.8392865531874912,
      "grad_norm": 0.8074793227881015,
      "learning_rate": 6.618646792910893e-07,
      "loss": 0.3516,
      "step": 13399
    },
    {
      "epoch": 0.8393491911868335,
      "grad_norm": 0.8232857796053843,
      "learning_rate": 6.613603938116603e-07,
      "loss": 0.3348,
      "step": 13400
    },
    {
      "epoch": 0.8394118291861757,
      "grad_norm": 0.7769459229979659,
      "learning_rate": 6.608562869114409e-07,
      "loss": 0.3289,
      "step": 13401
    },
    {
      "epoch": 0.8394744671855181,
      "grad_norm": 0.8789448488179713,
      "learning_rate": 6.603523586111832e-07,
      "loss": 0.3835,
      "step": 13402
    },
    {
      "epoch": 0.8395371051848604,
      "grad_norm": 0.9137880062940446,
      "learning_rate": 6.598486089316286e-07,
      "loss": 0.4289,
      "step": 13403
    },
    {
      "epoch": 0.8395997431842027,
      "grad_norm": 0.8586776094276706,
      "learning_rate": 6.593450378935102e-07,
      "loss": 0.3829,
      "step": 13404
    },
    {
      "epoch": 0.839662381183545,
      "grad_norm": 0.9168507827650292,
      "learning_rate": 6.588416455175561e-07,
      "loss": 0.4322,
      "step": 13405
    },
    {
      "epoch": 0.8397250191828873,
      "grad_norm": 0.9235478425460929,
      "learning_rate": 6.58338431824484e-07,
      "loss": 0.3999,
      "step": 13406
    },
    {
      "epoch": 0.8397876571822296,
      "grad_norm": 0.8866489192385818,
      "learning_rate": 6.578353968350093e-07,
      "loss": 0.375,
      "step": 13407
    },
    {
      "epoch": 0.839850295181572,
      "grad_norm": 0.9025011987049032,
      "learning_rate": 6.573325405698355e-07,
      "loss": 0.4061,
      "step": 13408
    },
    {
      "epoch": 0.8399129331809142,
      "grad_norm": 0.8578449131160866,
      "learning_rate": 6.568298630496606e-07,
      "loss": 0.3903,
      "step": 13409
    },
    {
      "epoch": 0.8399755711802565,
      "grad_norm": 0.8558058118092244,
      "learning_rate": 6.563273642951734e-07,
      "loss": 0.4243,
      "step": 13410
    },
    {
      "epoch": 0.8400382091795988,
      "grad_norm": 0.8469086665475503,
      "learning_rate": 6.558250443270597e-07,
      "loss": 0.3771,
      "step": 13411
    },
    {
      "epoch": 0.8401008471789411,
      "grad_norm": 0.8139064667010322,
      "learning_rate": 6.55322903165993e-07,
      "loss": 0.3471,
      "step": 13412
    },
    {
      "epoch": 0.8401634851782834,
      "grad_norm": 0.8588018630531655,
      "learning_rate": 6.548209408326417e-07,
      "loss": 0.38,
      "step": 13413
    },
    {
      "epoch": 0.8402261231776257,
      "grad_norm": 0.8164579791450353,
      "learning_rate": 6.543191573476676e-07,
      "loss": 0.3641,
      "step": 13414
    },
    {
      "epoch": 0.840288761176968,
      "grad_norm": 0.9115285290132784,
      "learning_rate": 6.538175527317219e-07,
      "loss": 0.4188,
      "step": 13415
    },
    {
      "epoch": 0.8403513991763103,
      "grad_norm": 0.8381640578260668,
      "learning_rate": 6.533161270054545e-07,
      "loss": 0.3479,
      "step": 13416
    },
    {
      "epoch": 0.8404140371756527,
      "grad_norm": 0.8157501508143342,
      "learning_rate": 6.528148801895013e-07,
      "loss": 0.374,
      "step": 13417
    },
    {
      "epoch": 0.8404766751749949,
      "grad_norm": 0.8903054984898472,
      "learning_rate": 6.523138123044948e-07,
      "loss": 0.4018,
      "step": 13418
    },
    {
      "epoch": 0.8405393131743372,
      "grad_norm": 0.9120818609896209,
      "learning_rate": 6.518129233710574e-07,
      "loss": 0.3827,
      "step": 13419
    },
    {
      "epoch": 0.8406019511736795,
      "grad_norm": 0.9506775313746499,
      "learning_rate": 6.513122134098083e-07,
      "loss": 0.3955,
      "step": 13420
    },
    {
      "epoch": 0.8406645891730218,
      "grad_norm": 0.8240356821510335,
      "learning_rate": 6.508116824413557e-07,
      "loss": 0.3882,
      "step": 13421
    },
    {
      "epoch": 0.8407272271723641,
      "grad_norm": 0.8718175395974685,
      "learning_rate": 6.503113304863013e-07,
      "loss": 0.3824,
      "step": 13422
    },
    {
      "epoch": 0.8407898651717064,
      "grad_norm": 0.8535920528435978,
      "learning_rate": 6.498111575652399e-07,
      "loss": 0.4049,
      "step": 13423
    },
    {
      "epoch": 0.8408525031710488,
      "grad_norm": 0.896789091598635,
      "learning_rate": 6.493111636987582e-07,
      "loss": 0.417,
      "step": 13424
    },
    {
      "epoch": 0.840915141170391,
      "grad_norm": 0.8621167606620789,
      "learning_rate": 6.488113489074371e-07,
      "loss": 0.3959,
      "step": 13425
    },
    {
      "epoch": 0.8409777791697333,
      "grad_norm": 0.8187669954588288,
      "learning_rate": 6.483117132118483e-07,
      "loss": 0.3521,
      "step": 13426
    },
    {
      "epoch": 0.8410404171690756,
      "grad_norm": 0.8672206077009958,
      "learning_rate": 6.478122566325573e-07,
      "loss": 0.3686,
      "step": 13427
    },
    {
      "epoch": 0.8411030551684179,
      "grad_norm": 0.8840812520743236,
      "learning_rate": 6.473129791901206e-07,
      "loss": 0.3653,
      "step": 13428
    },
    {
      "epoch": 0.8411656931677602,
      "grad_norm": 0.8463201133268394,
      "learning_rate": 6.468138809050911e-07,
      "loss": 0.3781,
      "step": 13429
    },
    {
      "epoch": 0.8412283311671025,
      "grad_norm": 0.8781225366273275,
      "learning_rate": 6.463149617980097e-07,
      "loss": 0.407,
      "step": 13430
    },
    {
      "epoch": 0.8412909691664449,
      "grad_norm": 0.8767631358310709,
      "learning_rate": 6.458162218894127e-07,
      "loss": 0.3633,
      "step": 13431
    },
    {
      "epoch": 0.8413536071657871,
      "grad_norm": 0.8354146338969092,
      "learning_rate": 6.453176611998268e-07,
      "loss": 0.3687,
      "step": 13432
    },
    {
      "epoch": 0.8414162451651295,
      "grad_norm": 0.7933776242378255,
      "learning_rate": 6.44819279749776e-07,
      "loss": 0.3253,
      "step": 13433
    },
    {
      "epoch": 0.8414788831644717,
      "grad_norm": 0.8527197988945147,
      "learning_rate": 6.443210775597714e-07,
      "loss": 0.4011,
      "step": 13434
    },
    {
      "epoch": 0.841541521163814,
      "grad_norm": 0.845544656269711,
      "learning_rate": 6.438230546503199e-07,
      "loss": 0.3514,
      "step": 13435
    },
    {
      "epoch": 0.8416041591631563,
      "grad_norm": 0.8413525835787151,
      "learning_rate": 6.433252110419202e-07,
      "loss": 0.3791,
      "step": 13436
    },
    {
      "epoch": 0.8416667971624986,
      "grad_norm": 0.9283534437890152,
      "learning_rate": 6.428275467550621e-07,
      "loss": 0.3927,
      "step": 13437
    },
    {
      "epoch": 0.841729435161841,
      "grad_norm": 0.8272663693749889,
      "learning_rate": 6.423300618102319e-07,
      "loss": 0.3458,
      "step": 13438
    },
    {
      "epoch": 0.8417920731611832,
      "grad_norm": 0.8572203065978905,
      "learning_rate": 6.418327562279053e-07,
      "loss": 0.4214,
      "step": 13439
    },
    {
      "epoch": 0.8418547111605256,
      "grad_norm": 0.771873061687504,
      "learning_rate": 6.413356300285517e-07,
      "loss": 0.3614,
      "step": 13440
    },
    {
      "epoch": 0.8419173491598678,
      "grad_norm": 0.8924971802216533,
      "learning_rate": 6.408386832326314e-07,
      "loss": 0.3683,
      "step": 13441
    },
    {
      "epoch": 0.8419799871592102,
      "grad_norm": 0.8282906609756984,
      "learning_rate": 6.403419158606005e-07,
      "loss": 0.3476,
      "step": 13442
    },
    {
      "epoch": 0.8420426251585524,
      "grad_norm": 0.837090680146753,
      "learning_rate": 6.398453279329059e-07,
      "loss": 0.3762,
      "step": 13443
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.8876421501867011,
      "learning_rate": 6.39348919469987e-07,
      "loss": 0.4179,
      "step": 13444
    },
    {
      "epoch": 0.842167901157237,
      "grad_norm": 0.8725016672444784,
      "learning_rate": 6.388526904922759e-07,
      "loss": 0.4044,
      "step": 13445
    },
    {
      "epoch": 0.8422305391565793,
      "grad_norm": 0.7643037820323705,
      "learning_rate": 6.383566410201974e-07,
      "loss": 0.3304,
      "step": 13446
    },
    {
      "epoch": 0.8422931771559217,
      "grad_norm": 0.9063343403928082,
      "learning_rate": 6.37860771074168e-07,
      "loss": 0.4036,
      "step": 13447
    },
    {
      "epoch": 0.8423558151552639,
      "grad_norm": 0.9465322961690545,
      "learning_rate": 6.373650806746001e-07,
      "loss": 0.424,
      "step": 13448
    },
    {
      "epoch": 0.8424184531546063,
      "grad_norm": 0.8689567351144044,
      "learning_rate": 6.368695698418941e-07,
      "loss": 0.355,
      "step": 13449
    },
    {
      "epoch": 0.8424810911539485,
      "grad_norm": 0.9224533735441196,
      "learning_rate": 6.363742385964478e-07,
      "loss": 0.3743,
      "step": 13450
    },
    {
      "epoch": 0.8425437291532909,
      "grad_norm": 0.8341410815871985,
      "learning_rate": 6.358790869586473e-07,
      "loss": 0.4216,
      "step": 13451
    },
    {
      "epoch": 0.8426063671526332,
      "grad_norm": 0.9264091148551347,
      "learning_rate": 6.353841149488738e-07,
      "loss": 0.448,
      "step": 13452
    },
    {
      "epoch": 0.8426690051519754,
      "grad_norm": 0.991757671416981,
      "learning_rate": 6.348893225875002e-07,
      "loss": 0.4152,
      "step": 13453
    },
    {
      "epoch": 0.8427316431513178,
      "grad_norm": 0.8288662897362408,
      "learning_rate": 6.343947098948927e-07,
      "loss": 0.3619,
      "step": 13454
    },
    {
      "epoch": 0.84279428115066,
      "grad_norm": 0.8641143323595434,
      "learning_rate": 6.339002768914087e-07,
      "loss": 0.3668,
      "step": 13455
    },
    {
      "epoch": 0.8428569191500024,
      "grad_norm": 0.8582869699296476,
      "learning_rate": 6.334060235973988e-07,
      "loss": 0.3793,
      "step": 13456
    },
    {
      "epoch": 0.8429195571493446,
      "grad_norm": 0.8875056162899714,
      "learning_rate": 6.329119500332087e-07,
      "loss": 0.3984,
      "step": 13457
    },
    {
      "epoch": 0.842982195148687,
      "grad_norm": 0.873106232542641,
      "learning_rate": 6.324180562191722e-07,
      "loss": 0.4055,
      "step": 13458
    },
    {
      "epoch": 0.8430448331480292,
      "grad_norm": 0.8797998897041956,
      "learning_rate": 6.319243421756199e-07,
      "loss": 0.4049,
      "step": 13459
    },
    {
      "epoch": 0.8431074711473715,
      "grad_norm": 0.9227071860616315,
      "learning_rate": 6.314308079228726e-07,
      "loss": 0.3934,
      "step": 13460
    },
    {
      "epoch": 0.8431701091467139,
      "grad_norm": 0.8270894127392496,
      "learning_rate": 6.30937453481244e-07,
      "loss": 0.3922,
      "step": 13461
    },
    {
      "epoch": 0.8432327471460561,
      "grad_norm": 0.8874719331180978,
      "learning_rate": 6.304442788710413e-07,
      "loss": 0.4234,
      "step": 13462
    },
    {
      "epoch": 0.8432953851453985,
      "grad_norm": 0.8853742511156081,
      "learning_rate": 6.299512841125626e-07,
      "loss": 0.3808,
      "step": 13463
    },
    {
      "epoch": 0.8433580231447407,
      "grad_norm": 0.8716409365955213,
      "learning_rate": 6.294584692261002e-07,
      "loss": 0.3864,
      "step": 13464
    },
    {
      "epoch": 0.8434206611440831,
      "grad_norm": 0.8473789342758893,
      "learning_rate": 6.289658342319372e-07,
      "loss": 0.3693,
      "step": 13465
    },
    {
      "epoch": 0.8434832991434253,
      "grad_norm": 0.8559662899002163,
      "learning_rate": 6.284733791503533e-07,
      "loss": 0.4064,
      "step": 13466
    },
    {
      "epoch": 0.8435459371427677,
      "grad_norm": 0.8445037959365513,
      "learning_rate": 6.279811040016148e-07,
      "loss": 0.3886,
      "step": 13467
    },
    {
      "epoch": 0.84360857514211,
      "grad_norm": 0.5724963556141492,
      "learning_rate": 6.274890088059871e-07,
      "loss": 0.4415,
      "step": 13468
    },
    {
      "epoch": 0.8436712131414522,
      "grad_norm": 0.8805284562472635,
      "learning_rate": 6.269970935837233e-07,
      "loss": 0.3948,
      "step": 13469
    },
    {
      "epoch": 0.8437338511407946,
      "grad_norm": 0.8727643168080884,
      "learning_rate": 6.265053583550706e-07,
      "loss": 0.3823,
      "step": 13470
    },
    {
      "epoch": 0.8437964891401368,
      "grad_norm": 0.9106523884515172,
      "learning_rate": 6.260138031402696e-07,
      "loss": 0.4241,
      "step": 13471
    },
    {
      "epoch": 0.8438591271394792,
      "grad_norm": 0.8593460410587551,
      "learning_rate": 6.255224279595518e-07,
      "loss": 0.4101,
      "step": 13472
    },
    {
      "epoch": 0.8439217651388214,
      "grad_norm": 0.8908902398765288,
      "learning_rate": 6.250312328331431e-07,
      "loss": 0.3689,
      "step": 13473
    },
    {
      "epoch": 0.8439844031381638,
      "grad_norm": 0.8447879167943739,
      "learning_rate": 6.245402177812599e-07,
      "loss": 0.4092,
      "step": 13474
    },
    {
      "epoch": 0.8440470411375061,
      "grad_norm": 0.8603790975812267,
      "learning_rate": 6.240493828241128e-07,
      "loss": 0.361,
      "step": 13475
    },
    {
      "epoch": 0.8441096791368484,
      "grad_norm": 0.8971279279944685,
      "learning_rate": 6.235587279819072e-07,
      "loss": 0.4156,
      "step": 13476
    },
    {
      "epoch": 0.8441723171361907,
      "grad_norm": 0.885161796455134,
      "learning_rate": 6.230682532748361e-07,
      "loss": 0.3985,
      "step": 13477
    },
    {
      "epoch": 0.8442349551355329,
      "grad_norm": 0.8587595859844966,
      "learning_rate": 6.225779587230885e-07,
      "loss": 0.398,
      "step": 13478
    },
    {
      "epoch": 0.8442975931348753,
      "grad_norm": 0.9276407495182495,
      "learning_rate": 6.220878443468442e-07,
      "loss": 0.4076,
      "step": 13479
    },
    {
      "epoch": 0.8443602311342175,
      "grad_norm": 0.8868490353229708,
      "learning_rate": 6.215979101662773e-07,
      "loss": 0.3896,
      "step": 13480
    },
    {
      "epoch": 0.8444228691335599,
      "grad_norm": 0.8815731076610963,
      "learning_rate": 6.211081562015526e-07,
      "loss": 0.42,
      "step": 13481
    },
    {
      "epoch": 0.8444855071329022,
      "grad_norm": 0.6592336872579033,
      "learning_rate": 6.206185824728295e-07,
      "loss": 0.4725,
      "step": 13482
    },
    {
      "epoch": 0.8445481451322445,
      "grad_norm": 0.875774182378357,
      "learning_rate": 6.20129189000257e-07,
      "loss": 0.3872,
      "step": 13483
    },
    {
      "epoch": 0.8446107831315868,
      "grad_norm": 0.8376421434096163,
      "learning_rate": 6.196399758039795e-07,
      "loss": 0.3566,
      "step": 13484
    },
    {
      "epoch": 0.844673421130929,
      "grad_norm": 0.8991582210984852,
      "learning_rate": 6.191509429041353e-07,
      "loss": 0.3764,
      "step": 13485
    },
    {
      "epoch": 0.8447360591302714,
      "grad_norm": 0.8467525462288455,
      "learning_rate": 6.186620903208512e-07,
      "loss": 0.3748,
      "step": 13486
    },
    {
      "epoch": 0.8447986971296136,
      "grad_norm": 0.8292662954653534,
      "learning_rate": 6.181734180742488e-07,
      "loss": 0.3453,
      "step": 13487
    },
    {
      "epoch": 0.844861335128956,
      "grad_norm": 0.8289206052954443,
      "learning_rate": 6.176849261844419e-07,
      "loss": 0.3504,
      "step": 13488
    },
    {
      "epoch": 0.8449239731282983,
      "grad_norm": 0.9179156856456723,
      "learning_rate": 6.171966146715369e-07,
      "loss": 0.3635,
      "step": 13489
    },
    {
      "epoch": 0.8449866111276406,
      "grad_norm": 0.6079404975364304,
      "learning_rate": 6.167084835556325e-07,
      "loss": 0.4775,
      "step": 13490
    },
    {
      "epoch": 0.8450492491269829,
      "grad_norm": 0.853358307620942,
      "learning_rate": 6.162205328568194e-07,
      "loss": 0.4022,
      "step": 13491
    },
    {
      "epoch": 0.8451118871263252,
      "grad_norm": 0.934476966997766,
      "learning_rate": 6.157327625951843e-07,
      "loss": 0.3802,
      "step": 13492
    },
    {
      "epoch": 0.8451745251256675,
      "grad_norm": 0.8951684014997312,
      "learning_rate": 6.15245172790801e-07,
      "loss": 0.4186,
      "step": 13493
    },
    {
      "epoch": 0.8452371631250097,
      "grad_norm": 0.8292029594401948,
      "learning_rate": 6.147577634637413e-07,
      "loss": 0.3673,
      "step": 13494
    },
    {
      "epoch": 0.8452998011243521,
      "grad_norm": 0.8922187926256429,
      "learning_rate": 6.14270534634066e-07,
      "loss": 0.3805,
      "step": 13495
    },
    {
      "epoch": 0.8453624391236944,
      "grad_norm": 0.845809755884411,
      "learning_rate": 6.137834863218295e-07,
      "loss": 0.3678,
      "step": 13496
    },
    {
      "epoch": 0.8454250771230367,
      "grad_norm": 0.8794388990558779,
      "learning_rate": 6.13296618547079e-07,
      "loss": 0.3842,
      "step": 13497
    },
    {
      "epoch": 0.845487715122379,
      "grad_norm": 0.8704064973423138,
      "learning_rate": 6.128099313298541e-07,
      "loss": 0.4124,
      "step": 13498
    },
    {
      "epoch": 0.8455503531217213,
      "grad_norm": 0.9335877945336223,
      "learning_rate": 6.123234246901866e-07,
      "loss": 0.3794,
      "step": 13499
    },
    {
      "epoch": 0.8456129911210636,
      "grad_norm": 0.8471420420935851,
      "learning_rate": 6.118370986480999e-07,
      "loss": 0.385,
      "step": 13500
    },
    {
      "epoch": 0.845675629120406,
      "grad_norm": 0.8252787111283435,
      "learning_rate": 6.113509532236139e-07,
      "loss": 0.3682,
      "step": 13501
    },
    {
      "epoch": 0.8457382671197482,
      "grad_norm": 0.8555479919410461,
      "learning_rate": 6.108649884367362e-07,
      "loss": 0.3653,
      "step": 13502
    },
    {
      "epoch": 0.8458009051190905,
      "grad_norm": 0.8958184971108341,
      "learning_rate": 6.103792043074713e-07,
      "loss": 0.3953,
      "step": 13503
    },
    {
      "epoch": 0.8458635431184328,
      "grad_norm": 0.8750337001785596,
      "learning_rate": 6.098936008558131e-07,
      "loss": 0.3998,
      "step": 13504
    },
    {
      "epoch": 0.8459261811177751,
      "grad_norm": 0.9084089040861821,
      "learning_rate": 6.094081781017497e-07,
      "loss": 0.3621,
      "step": 13505
    },
    {
      "epoch": 0.8459888191171174,
      "grad_norm": 0.9309589479875249,
      "learning_rate": 6.089229360652599e-07,
      "loss": 0.3785,
      "step": 13506
    },
    {
      "epoch": 0.8460514571164597,
      "grad_norm": 0.91643434871522,
      "learning_rate": 6.084378747663172e-07,
      "loss": 0.3781,
      "step": 13507
    },
    {
      "epoch": 0.846114095115802,
      "grad_norm": 0.9074609767558326,
      "learning_rate": 6.079529942248868e-07,
      "loss": 0.3716,
      "step": 13508
    },
    {
      "epoch": 0.8461767331151443,
      "grad_norm": 0.9074489456338919,
      "learning_rate": 6.07468294460925e-07,
      "loss": 0.3565,
      "step": 13509
    },
    {
      "epoch": 0.8462393711144865,
      "grad_norm": 0.8887907413872156,
      "learning_rate": 6.069837754943853e-07,
      "loss": 0.3858,
      "step": 13510
    },
    {
      "epoch": 0.8463020091138289,
      "grad_norm": 0.8113872564801081,
      "learning_rate": 6.064994373452071e-07,
      "loss": 0.3146,
      "step": 13511
    },
    {
      "epoch": 0.8463646471131712,
      "grad_norm": 0.9196473947884839,
      "learning_rate": 6.060152800333286e-07,
      "loss": 0.421,
      "step": 13512
    },
    {
      "epoch": 0.8464272851125135,
      "grad_norm": 0.8817431776195372,
      "learning_rate": 6.055313035786769e-07,
      "loss": 0.3808,
      "step": 13513
    },
    {
      "epoch": 0.8464899231118558,
      "grad_norm": 0.874414527911658,
      "learning_rate": 6.050475080011725e-07,
      "loss": 0.4012,
      "step": 13514
    },
    {
      "epoch": 0.8465525611111981,
      "grad_norm": 0.9567828070255532,
      "learning_rate": 6.045638933207282e-07,
      "loss": 0.4257,
      "step": 13515
    },
    {
      "epoch": 0.8466151991105404,
      "grad_norm": 0.848178873758519,
      "learning_rate": 6.040804595572497e-07,
      "loss": 0.3857,
      "step": 13516
    },
    {
      "epoch": 0.8466778371098828,
      "grad_norm": 0.854881750316407,
      "learning_rate": 6.03597206730635e-07,
      "loss": 0.4152,
      "step": 13517
    },
    {
      "epoch": 0.846740475109225,
      "grad_norm": 0.8407716505888015,
      "learning_rate": 6.031141348607761e-07,
      "loss": 0.3885,
      "step": 13518
    },
    {
      "epoch": 0.8468031131085673,
      "grad_norm": 0.9734085431501454,
      "learning_rate": 6.026312439675553e-07,
      "loss": 0.4049,
      "step": 13519
    },
    {
      "epoch": 0.8468657511079096,
      "grad_norm": 0.8087929054869709,
      "learning_rate": 6.021485340708477e-07,
      "loss": 0.3502,
      "step": 13520
    },
    {
      "epoch": 0.8469283891072519,
      "grad_norm": 0.8899431139964624,
      "learning_rate": 6.016660051905238e-07,
      "loss": 0.3517,
      "step": 13521
    },
    {
      "epoch": 0.8469910271065942,
      "grad_norm": 0.8608179915210714,
      "learning_rate": 6.01183657346443e-07,
      "loss": 0.3627,
      "step": 13522
    },
    {
      "epoch": 0.8470536651059365,
      "grad_norm": 0.8722120279285496,
      "learning_rate": 6.007014905584602e-07,
      "loss": 0.3832,
      "step": 13523
    },
    {
      "epoch": 0.8471163031052789,
      "grad_norm": 0.8690799817086744,
      "learning_rate": 6.002195048464199e-07,
      "loss": 0.3894,
      "step": 13524
    },
    {
      "epoch": 0.8471789411046211,
      "grad_norm": 0.8694013394420095,
      "learning_rate": 5.997377002301618e-07,
      "loss": 0.359,
      "step": 13525
    },
    {
      "epoch": 0.8472415791039635,
      "grad_norm": 0.7993384863797472,
      "learning_rate": 5.992560767295158e-07,
      "loss": 0.3851,
      "step": 13526
    },
    {
      "epoch": 0.8473042171033057,
      "grad_norm": 0.9055435975169182,
      "learning_rate": 5.98774634364307e-07,
      "loss": 0.4048,
      "step": 13527
    },
    {
      "epoch": 0.847366855102648,
      "grad_norm": 0.7934534933565518,
      "learning_rate": 5.982933731543517e-07,
      "loss": 0.3224,
      "step": 13528
    },
    {
      "epoch": 0.8474294931019903,
      "grad_norm": 0.8243557863017559,
      "learning_rate": 5.978122931194569e-07,
      "loss": 0.335,
      "step": 13529
    },
    {
      "epoch": 0.8474921311013326,
      "grad_norm": 0.9019012941182575,
      "learning_rate": 5.973313942794256e-07,
      "loss": 0.4091,
      "step": 13530
    },
    {
      "epoch": 0.847554769100675,
      "grad_norm": 0.8794565688210341,
      "learning_rate": 5.968506766540521e-07,
      "loss": 0.3669,
      "step": 13531
    },
    {
      "epoch": 0.8476174071000172,
      "grad_norm": 0.834386362820316,
      "learning_rate": 5.963701402631217e-07,
      "loss": 0.3875,
      "step": 13532
    },
    {
      "epoch": 0.8476800450993596,
      "grad_norm": 0.8556046812149115,
      "learning_rate": 5.958897851264139e-07,
      "loss": 0.3452,
      "step": 13533
    },
    {
      "epoch": 0.8477426830987018,
      "grad_norm": 0.8369894626784755,
      "learning_rate": 5.954096112636987e-07,
      "loss": 0.371,
      "step": 13534
    },
    {
      "epoch": 0.8478053210980441,
      "grad_norm": 0.9498125880536274,
      "learning_rate": 5.949296186947429e-07,
      "loss": 0.3897,
      "step": 13535
    },
    {
      "epoch": 0.8478679590973864,
      "grad_norm": 0.9415369779676381,
      "learning_rate": 5.944498074393013e-07,
      "loss": 0.4592,
      "step": 13536
    },
    {
      "epoch": 0.8479305970967287,
      "grad_norm": 0.5753217547917915,
      "learning_rate": 5.93970177517123e-07,
      "loss": 0.4466,
      "step": 13537
    },
    {
      "epoch": 0.847993235096071,
      "grad_norm": 0.9319425221783904,
      "learning_rate": 5.934907289479508e-07,
      "loss": 0.3551,
      "step": 13538
    },
    {
      "epoch": 0.8480558730954133,
      "grad_norm": 0.8673981660026089,
      "learning_rate": 5.930114617515165e-07,
      "loss": 0.3746,
      "step": 13539
    },
    {
      "epoch": 0.8481185110947557,
      "grad_norm": 0.9226842193455315,
      "learning_rate": 5.925323759475494e-07,
      "loss": 0.4017,
      "step": 13540
    },
    {
      "epoch": 0.8481811490940979,
      "grad_norm": 0.8609729198902301,
      "learning_rate": 5.92053471555768e-07,
      "loss": 0.412,
      "step": 13541
    },
    {
      "epoch": 0.8482437870934403,
      "grad_norm": 0.8489135368834813,
      "learning_rate": 5.915747485958834e-07,
      "loss": 0.4029,
      "step": 13542
    },
    {
      "epoch": 0.8483064250927825,
      "grad_norm": 0.9414986975533244,
      "learning_rate": 5.910962070875998e-07,
      "loss": 0.3936,
      "step": 13543
    },
    {
      "epoch": 0.8483690630921248,
      "grad_norm": 0.8116632236922077,
      "learning_rate": 5.906178470506158e-07,
      "loss": 0.3706,
      "step": 13544
    },
    {
      "epoch": 0.8484317010914671,
      "grad_norm": 0.852392394787291,
      "learning_rate": 5.90139668504619e-07,
      "loss": 0.4068,
      "step": 13545
    },
    {
      "epoch": 0.8484943390908094,
      "grad_norm": 0.8839829033527947,
      "learning_rate": 5.896616714692927e-07,
      "loss": 0.3874,
      "step": 13546
    },
    {
      "epoch": 0.8485569770901518,
      "grad_norm": 0.8458435528688907,
      "learning_rate": 5.891838559643104e-07,
      "loss": 0.3726,
      "step": 13547
    },
    {
      "epoch": 0.848619615089494,
      "grad_norm": 0.8253035996881153,
      "learning_rate": 5.887062220093381e-07,
      "loss": 0.3744,
      "step": 13548
    },
    {
      "epoch": 0.8486822530888364,
      "grad_norm": 0.7954313606442143,
      "learning_rate": 5.882287696240374e-07,
      "loss": 0.3896,
      "step": 13549
    },
    {
      "epoch": 0.8487448910881786,
      "grad_norm": 0.8703428498709661,
      "learning_rate": 5.877514988280597e-07,
      "loss": 0.3654,
      "step": 13550
    },
    {
      "epoch": 0.848807529087521,
      "grad_norm": 0.8280652887647062,
      "learning_rate": 5.872744096410487e-07,
      "loss": 0.3562,
      "step": 13551
    },
    {
      "epoch": 0.8488701670868632,
      "grad_norm": 0.8915210432397387,
      "learning_rate": 5.867975020826416e-07,
      "loss": 0.3731,
      "step": 13552
    },
    {
      "epoch": 0.8489328050862055,
      "grad_norm": 0.8413047251664084,
      "learning_rate": 5.863207761724693e-07,
      "loss": 0.3697,
      "step": 13553
    },
    {
      "epoch": 0.8489954430855479,
      "grad_norm": 0.9283711805374978,
      "learning_rate": 5.858442319301532e-07,
      "loss": 0.4031,
      "step": 13554
    },
    {
      "epoch": 0.8490580810848901,
      "grad_norm": 0.9249085403960884,
      "learning_rate": 5.853678693753073e-07,
      "loss": 0.4184,
      "step": 13555
    },
    {
      "epoch": 0.8491207190842325,
      "grad_norm": 0.8595576368285577,
      "learning_rate": 5.848916885275397e-07,
      "loss": 0.3724,
      "step": 13556
    },
    {
      "epoch": 0.8491833570835747,
      "grad_norm": 0.5759684904164073,
      "learning_rate": 5.84415689406449e-07,
      "loss": 0.4298,
      "step": 13557
    },
    {
      "epoch": 0.8492459950829171,
      "grad_norm": 0.8552970149158085,
      "learning_rate": 5.839398720316286e-07,
      "loss": 0.3618,
      "step": 13558
    },
    {
      "epoch": 0.8493086330822593,
      "grad_norm": 0.8423071498484144,
      "learning_rate": 5.834642364226628e-07,
      "loss": 0.355,
      "step": 13559
    },
    {
      "epoch": 0.8493712710816017,
      "grad_norm": 0.8633931737450061,
      "learning_rate": 5.829887825991276e-07,
      "loss": 0.3853,
      "step": 13560
    },
    {
      "epoch": 0.849433909080944,
      "grad_norm": 0.841565347111892,
      "learning_rate": 5.825135105805952e-07,
      "loss": 0.3516,
      "step": 13561
    },
    {
      "epoch": 0.8494965470802862,
      "grad_norm": 0.8266688028569953,
      "learning_rate": 5.820384203866269e-07,
      "loss": 0.3538,
      "step": 13562
    },
    {
      "epoch": 0.8495591850796286,
      "grad_norm": 0.8400981257737311,
      "learning_rate": 5.815635120367769e-07,
      "loss": 0.3775,
      "step": 13563
    },
    {
      "epoch": 0.8496218230789708,
      "grad_norm": 0.8426603308713919,
      "learning_rate": 5.810887855505926e-07,
      "loss": 0.3927,
      "step": 13564
    },
    {
      "epoch": 0.8496844610783132,
      "grad_norm": 0.9276927840536991,
      "learning_rate": 5.806142409476145e-07,
      "loss": 0.4366,
      "step": 13565
    },
    {
      "epoch": 0.8497470990776554,
      "grad_norm": 0.600439422972364,
      "learning_rate": 5.801398782473733e-07,
      "loss": 0.4543,
      "step": 13566
    },
    {
      "epoch": 0.8498097370769978,
      "grad_norm": 0.9033021929518298,
      "learning_rate": 5.796656974693964e-07,
      "loss": 0.3795,
      "step": 13567
    },
    {
      "epoch": 0.8498723750763401,
      "grad_norm": 0.9701658700266792,
      "learning_rate": 5.791916986331996e-07,
      "loss": 0.4256,
      "step": 13568
    },
    {
      "epoch": 0.8499350130756823,
      "grad_norm": 0.8875873352257336,
      "learning_rate": 5.787178817582917e-07,
      "loss": 0.3761,
      "step": 13569
    },
    {
      "epoch": 0.8499976510750247,
      "grad_norm": 0.5889435819839354,
      "learning_rate": 5.78244246864178e-07,
      "loss": 0.4184,
      "step": 13570
    },
    {
      "epoch": 0.8500602890743669,
      "grad_norm": 0.8019882765885019,
      "learning_rate": 5.777707939703515e-07,
      "loss": 0.3416,
      "step": 13571
    },
    {
      "epoch": 0.8501229270737093,
      "grad_norm": 0.8750515255606793,
      "learning_rate": 5.772975230963002e-07,
      "loss": 0.399,
      "step": 13572
    },
    {
      "epoch": 0.8501855650730515,
      "grad_norm": 0.8747359650768225,
      "learning_rate": 5.768244342615031e-07,
      "loss": 0.3832,
      "step": 13573
    },
    {
      "epoch": 0.8502482030723939,
      "grad_norm": 0.8198915859990994,
      "learning_rate": 5.763515274854337e-07,
      "loss": 0.3818,
      "step": 13574
    },
    {
      "epoch": 0.8503108410717362,
      "grad_norm": 0.8175970031022336,
      "learning_rate": 5.758788027875556e-07,
      "loss": 0.4116,
      "step": 13575
    },
    {
      "epoch": 0.8503734790710785,
      "grad_norm": 0.8555839239854939,
      "learning_rate": 5.754062601873278e-07,
      "loss": 0.3941,
      "step": 13576
    },
    {
      "epoch": 0.8504361170704208,
      "grad_norm": 0.9021158468077546,
      "learning_rate": 5.749338997041997e-07,
      "loss": 0.4334,
      "step": 13577
    },
    {
      "epoch": 0.850498755069763,
      "grad_norm": 0.9003606038178533,
      "learning_rate": 5.744617213576131e-07,
      "loss": 0.4071,
      "step": 13578
    },
    {
      "epoch": 0.8505613930691054,
      "grad_norm": 0.8912903689857077,
      "learning_rate": 5.73989725167004e-07,
      "loss": 0.4158,
      "step": 13579
    },
    {
      "epoch": 0.8506240310684476,
      "grad_norm": 0.9150853251976403,
      "learning_rate": 5.735179111517997e-07,
      "loss": 0.3813,
      "step": 13580
    },
    {
      "epoch": 0.85068666906779,
      "grad_norm": 0.8949027503327376,
      "learning_rate": 5.73046279331419e-07,
      "loss": 0.3705,
      "step": 13581
    },
    {
      "epoch": 0.8507493070671323,
      "grad_norm": 0.945072085002833,
      "learning_rate": 5.725748297252759e-07,
      "loss": 0.4136,
      "step": 13582
    },
    {
      "epoch": 0.8508119450664746,
      "grad_norm": 0.8088606643403846,
      "learning_rate": 5.721035623527743e-07,
      "loss": 0.3658,
      "step": 13583
    },
    {
      "epoch": 0.8508745830658169,
      "grad_norm": 0.8851857706765173,
      "learning_rate": 5.716324772333115e-07,
      "loss": 0.3995,
      "step": 13584
    },
    {
      "epoch": 0.8509372210651592,
      "grad_norm": 0.8698503550327896,
      "learning_rate": 5.711615743862775e-07,
      "loss": 0.3717,
      "step": 13585
    },
    {
      "epoch": 0.8509998590645015,
      "grad_norm": 0.8911375802140209,
      "learning_rate": 5.706908538310546e-07,
      "loss": 0.3896,
      "step": 13586
    },
    {
      "epoch": 0.8510624970638437,
      "grad_norm": 0.9102401632813593,
      "learning_rate": 5.7022031558702e-07,
      "loss": 0.3805,
      "step": 13587
    },
    {
      "epoch": 0.8511251350631861,
      "grad_norm": 0.8603668603488097,
      "learning_rate": 5.69749959673539e-07,
      "loss": 0.3838,
      "step": 13588
    },
    {
      "epoch": 0.8511877730625284,
      "grad_norm": 0.8356716892866951,
      "learning_rate": 5.692797861099719e-07,
      "loss": 0.3531,
      "step": 13589
    },
    {
      "epoch": 0.8512504110618707,
      "grad_norm": 0.8830975682527782,
      "learning_rate": 5.688097949156712e-07,
      "loss": 0.3635,
      "step": 13590
    },
    {
      "epoch": 0.851313049061213,
      "grad_norm": 0.8381340274066198,
      "learning_rate": 5.68339986109982e-07,
      "loss": 0.3701,
      "step": 13591
    },
    {
      "epoch": 0.8513756870605553,
      "grad_norm": 0.8793360511484049,
      "learning_rate": 5.678703597122409e-07,
      "loss": 0.377,
      "step": 13592
    },
    {
      "epoch": 0.8514383250598976,
      "grad_norm": 0.8305559712599914,
      "learning_rate": 5.674009157417793e-07,
      "loss": 0.3426,
      "step": 13593
    },
    {
      "epoch": 0.8515009630592398,
      "grad_norm": 0.8717425326151015,
      "learning_rate": 5.669316542179171e-07,
      "loss": 0.4117,
      "step": 13594
    },
    {
      "epoch": 0.8515636010585822,
      "grad_norm": 0.8729282447067457,
      "learning_rate": 5.664625751599711e-07,
      "loss": 0.3988,
      "step": 13595
    },
    {
      "epoch": 0.8516262390579245,
      "grad_norm": 0.8537718395847764,
      "learning_rate": 5.659936785872494e-07,
      "loss": 0.3808,
      "step": 13596
    },
    {
      "epoch": 0.8516888770572668,
      "grad_norm": 0.8415484879623086,
      "learning_rate": 5.655249645190508e-07,
      "loss": 0.3996,
      "step": 13597
    },
    {
      "epoch": 0.8517515150566091,
      "grad_norm": 0.7871422364398799,
      "learning_rate": 5.650564329746677e-07,
      "loss": 0.3528,
      "step": 13598
    },
    {
      "epoch": 0.8518141530559514,
      "grad_norm": 0.6146225781811805,
      "learning_rate": 5.645880839733853e-07,
      "loss": 0.4566,
      "step": 13599
    },
    {
      "epoch": 0.8518767910552937,
      "grad_norm": 0.9232299879411361,
      "learning_rate": 5.641199175344803e-07,
      "loss": 0.3877,
      "step": 13600
    },
    {
      "epoch": 0.851939429054636,
      "grad_norm": 0.9274236912383736,
      "learning_rate": 5.636519336772228e-07,
      "loss": 0.3777,
      "step": 13601
    },
    {
      "epoch": 0.8520020670539783,
      "grad_norm": 0.8703513055944838,
      "learning_rate": 5.631841324208743e-07,
      "loss": 0.3619,
      "step": 13602
    },
    {
      "epoch": 0.8520647050533205,
      "grad_norm": 0.8148042335253884,
      "learning_rate": 5.627165137846919e-07,
      "loss": 0.3972,
      "step": 13603
    },
    {
      "epoch": 0.8521273430526629,
      "grad_norm": 0.6280404129256465,
      "learning_rate": 5.622490777879197e-07,
      "loss": 0.4539,
      "step": 13604
    },
    {
      "epoch": 0.8521899810520052,
      "grad_norm": 0.8152891775882993,
      "learning_rate": 5.617818244498002e-07,
      "loss": 0.3606,
      "step": 13605
    },
    {
      "epoch": 0.8522526190513475,
      "grad_norm": 0.8769198245257509,
      "learning_rate": 5.613147537895647e-07,
      "loss": 0.376,
      "step": 13606
    },
    {
      "epoch": 0.8523152570506898,
      "grad_norm": 0.8572978294714848,
      "learning_rate": 5.608478658264376e-07,
      "loss": 0.3658,
      "step": 13607
    },
    {
      "epoch": 0.8523778950500321,
      "grad_norm": 0.8424245264919895,
      "learning_rate": 5.603811605796366e-07,
      "loss": 0.3922,
      "step": 13608
    },
    {
      "epoch": 0.8524405330493744,
      "grad_norm": 0.8666905434734636,
      "learning_rate": 5.599146380683712e-07,
      "loss": 0.4107,
      "step": 13609
    },
    {
      "epoch": 0.8525031710487168,
      "grad_norm": 0.8494694457329665,
      "learning_rate": 5.594482983118432e-07,
      "loss": 0.3312,
      "step": 13610
    },
    {
      "epoch": 0.852565809048059,
      "grad_norm": 0.8620432574549403,
      "learning_rate": 5.589821413292462e-07,
      "loss": 0.3344,
      "step": 13611
    },
    {
      "epoch": 0.8526284470474013,
      "grad_norm": 0.9532780059636449,
      "learning_rate": 5.585161671397687e-07,
      "loss": 0.4264,
      "step": 13612
    },
    {
      "epoch": 0.8526910850467436,
      "grad_norm": 0.9156396831586116,
      "learning_rate": 5.58050375762591e-07,
      "loss": 0.411,
      "step": 13613
    },
    {
      "epoch": 0.8527537230460859,
      "grad_norm": 0.8627191851838208,
      "learning_rate": 5.575847672168849e-07,
      "loss": 0.3993,
      "step": 13614
    },
    {
      "epoch": 0.8528163610454282,
      "grad_norm": 0.8735428206416087,
      "learning_rate": 5.571193415218135e-07,
      "loss": 0.3711,
      "step": 13615
    },
    {
      "epoch": 0.8528789990447705,
      "grad_norm": 0.8577028981359086,
      "learning_rate": 5.566540986965352e-07,
      "loss": 0.3873,
      "step": 13616
    },
    {
      "epoch": 0.8529416370441129,
      "grad_norm": 0.8778330086158294,
      "learning_rate": 5.561890387601987e-07,
      "loss": 0.3547,
      "step": 13617
    },
    {
      "epoch": 0.8530042750434551,
      "grad_norm": 0.6174842030915503,
      "learning_rate": 5.557241617319465e-07,
      "loss": 0.4459,
      "step": 13618
    },
    {
      "epoch": 0.8530669130427974,
      "grad_norm": 0.5808859741150131,
      "learning_rate": 5.552594676309115e-07,
      "loss": 0.4377,
      "step": 13619
    },
    {
      "epoch": 0.8531295510421397,
      "grad_norm": 0.8260630269693926,
      "learning_rate": 5.547949564762229e-07,
      "loss": 0.3598,
      "step": 13620
    },
    {
      "epoch": 0.853192189041482,
      "grad_norm": 0.8694272010962936,
      "learning_rate": 5.543306282869976e-07,
      "loss": 0.3967,
      "step": 13621
    },
    {
      "epoch": 0.8532548270408243,
      "grad_norm": 0.9325510860998407,
      "learning_rate": 5.538664830823498e-07,
      "loss": 0.4221,
      "step": 13622
    },
    {
      "epoch": 0.8533174650401666,
      "grad_norm": 0.8694209470642589,
      "learning_rate": 5.534025208813831e-07,
      "loss": 0.4046,
      "step": 13623
    },
    {
      "epoch": 0.853380103039509,
      "grad_norm": 0.8591403994914437,
      "learning_rate": 5.529387417031939e-07,
      "loss": 0.3906,
      "step": 13624
    },
    {
      "epoch": 0.8534427410388512,
      "grad_norm": 0.8789346384675177,
      "learning_rate": 5.524751455668715e-07,
      "loss": 0.3453,
      "step": 13625
    },
    {
      "epoch": 0.8535053790381936,
      "grad_norm": 0.9363231597249752,
      "learning_rate": 5.520117324914975e-07,
      "loss": 0.4344,
      "step": 13626
    },
    {
      "epoch": 0.8535680170375358,
      "grad_norm": 0.8766747785843215,
      "learning_rate": 5.515485024961465e-07,
      "loss": 0.3281,
      "step": 13627
    },
    {
      "epoch": 0.8536306550368781,
      "grad_norm": 0.8467781769686614,
      "learning_rate": 5.510854555998834e-07,
      "loss": 0.4056,
      "step": 13628
    },
    {
      "epoch": 0.8536932930362204,
      "grad_norm": 0.8285473057869693,
      "learning_rate": 5.5062259182177e-07,
      "loss": 0.3487,
      "step": 13629
    },
    {
      "epoch": 0.8537559310355627,
      "grad_norm": 0.8730928200975602,
      "learning_rate": 5.501599111808565e-07,
      "loss": 0.3267,
      "step": 13630
    },
    {
      "epoch": 0.853818569034905,
      "grad_norm": 0.8111800146790324,
      "learning_rate": 5.496974136961857e-07,
      "loss": 0.3259,
      "step": 13631
    },
    {
      "epoch": 0.8538812070342473,
      "grad_norm": 0.8822865585976095,
      "learning_rate": 5.492350993867968e-07,
      "loss": 0.388,
      "step": 13632
    },
    {
      "epoch": 0.8539438450335897,
      "grad_norm": 0.9054582047294051,
      "learning_rate": 5.487729682717169e-07,
      "loss": 0.3763,
      "step": 13633
    },
    {
      "epoch": 0.8540064830329319,
      "grad_norm": 0.9192378713598537,
      "learning_rate": 5.483110203699682e-07,
      "loss": 0.3834,
      "step": 13634
    },
    {
      "epoch": 0.8540691210322743,
      "grad_norm": 0.8756359321846818,
      "learning_rate": 5.478492557005638e-07,
      "loss": 0.3941,
      "step": 13635
    },
    {
      "epoch": 0.8541317590316165,
      "grad_norm": 0.8372954299336627,
      "learning_rate": 5.473876742825107e-07,
      "loss": 0.3511,
      "step": 13636
    },
    {
      "epoch": 0.8541943970309588,
      "grad_norm": 0.8739980075163479,
      "learning_rate": 5.469262761348065e-07,
      "loss": 0.3696,
      "step": 13637
    },
    {
      "epoch": 0.8542570350303011,
      "grad_norm": 0.5970876987541094,
      "learning_rate": 5.464650612764438e-07,
      "loss": 0.4267,
      "step": 13638
    },
    {
      "epoch": 0.8543196730296434,
      "grad_norm": 0.8468703155177748,
      "learning_rate": 5.460040297264063e-07,
      "loss": 0.3641,
      "step": 13639
    },
    {
      "epoch": 0.8543823110289858,
      "grad_norm": 0.8629078836980016,
      "learning_rate": 5.455431815036683e-07,
      "loss": 0.3642,
      "step": 13640
    },
    {
      "epoch": 0.854444949028328,
      "grad_norm": 0.8979459900377185,
      "learning_rate": 5.450825166272006e-07,
      "loss": 0.3924,
      "step": 13641
    },
    {
      "epoch": 0.8545075870276704,
      "grad_norm": 0.8763008156434878,
      "learning_rate": 5.446220351159637e-07,
      "loss": 0.4052,
      "step": 13642
    },
    {
      "epoch": 0.8545702250270126,
      "grad_norm": 0.9164421200661367,
      "learning_rate": 5.441617369889108e-07,
      "loss": 0.3822,
      "step": 13643
    },
    {
      "epoch": 0.8546328630263549,
      "grad_norm": 0.8880461399585213,
      "learning_rate": 5.437016222649877e-07,
      "loss": 0.3781,
      "step": 13644
    },
    {
      "epoch": 0.8546955010256972,
      "grad_norm": 0.8569740524015446,
      "learning_rate": 5.432416909631321e-07,
      "loss": 0.4051,
      "step": 13645
    },
    {
      "epoch": 0.8547581390250395,
      "grad_norm": 0.876554432042823,
      "learning_rate": 5.427819431022768e-07,
      "loss": 0.3924,
      "step": 13646
    },
    {
      "epoch": 0.8548207770243819,
      "grad_norm": 0.889136602186312,
      "learning_rate": 5.423223787013438e-07,
      "loss": 0.3754,
      "step": 13647
    },
    {
      "epoch": 0.8548834150237241,
      "grad_norm": 0.9236379344613429,
      "learning_rate": 5.418629977792489e-07,
      "loss": 0.4308,
      "step": 13648
    },
    {
      "epoch": 0.8549460530230665,
      "grad_norm": 0.8684483459058218,
      "learning_rate": 5.414038003549e-07,
      "loss": 0.3998,
      "step": 13649
    },
    {
      "epoch": 0.8550086910224087,
      "grad_norm": 0.8004628646174152,
      "learning_rate": 5.409447864471989e-07,
      "loss": 0.3713,
      "step": 13650
    },
    {
      "epoch": 0.8550713290217511,
      "grad_norm": 0.8915323069637459,
      "learning_rate": 5.40485956075038e-07,
      "loss": 0.3844,
      "step": 13651
    },
    {
      "epoch": 0.8551339670210933,
      "grad_norm": 0.9527773922611982,
      "learning_rate": 5.400273092573028e-07,
      "loss": 0.4021,
      "step": 13652
    },
    {
      "epoch": 0.8551966050204356,
      "grad_norm": 0.9168259285886996,
      "learning_rate": 5.395688460128718e-07,
      "loss": 0.3753,
      "step": 13653
    },
    {
      "epoch": 0.855259243019778,
      "grad_norm": 0.9729970410909411,
      "learning_rate": 5.391105663606133e-07,
      "loss": 0.4137,
      "step": 13654
    },
    {
      "epoch": 0.8553218810191202,
      "grad_norm": 0.8601966258095626,
      "learning_rate": 5.386524703193934e-07,
      "loss": 0.3712,
      "step": 13655
    },
    {
      "epoch": 0.8553845190184626,
      "grad_norm": 0.8688997986502452,
      "learning_rate": 5.381945579080655e-07,
      "loss": 0.3713,
      "step": 13656
    },
    {
      "epoch": 0.8554471570178048,
      "grad_norm": 0.9262186943947636,
      "learning_rate": 5.377368291454782e-07,
      "loss": 0.3953,
      "step": 13657
    },
    {
      "epoch": 0.8555097950171472,
      "grad_norm": 0.9671198445364305,
      "learning_rate": 5.372792840504704e-07,
      "loss": 0.4671,
      "step": 13658
    },
    {
      "epoch": 0.8555724330164894,
      "grad_norm": 0.9735898669989651,
      "learning_rate": 5.368219226418758e-07,
      "loss": 0.4156,
      "step": 13659
    },
    {
      "epoch": 0.8556350710158318,
      "grad_norm": 0.9078989502213282,
      "learning_rate": 5.363647449385201e-07,
      "loss": 0.3598,
      "step": 13660
    },
    {
      "epoch": 0.8556977090151741,
      "grad_norm": 0.7806398395034392,
      "learning_rate": 5.359077509592192e-07,
      "loss": 0.353,
      "step": 13661
    },
    {
      "epoch": 0.8557603470145163,
      "grad_norm": 0.8828688419291478,
      "learning_rate": 5.354509407227842e-07,
      "loss": 0.3998,
      "step": 13662
    },
    {
      "epoch": 0.8558229850138587,
      "grad_norm": 0.8150688722798433,
      "learning_rate": 5.349943142480163e-07,
      "loss": 0.3643,
      "step": 13663
    },
    {
      "epoch": 0.8558856230132009,
      "grad_norm": 0.8206207551097752,
      "learning_rate": 5.34537871553712e-07,
      "loss": 0.393,
      "step": 13664
    },
    {
      "epoch": 0.8559482610125433,
      "grad_norm": 0.9076449868437412,
      "learning_rate": 5.340816126586579e-07,
      "loss": 0.383,
      "step": 13665
    },
    {
      "epoch": 0.8560108990118855,
      "grad_norm": 0.9407692310620932,
      "learning_rate": 5.336255375816334e-07,
      "loss": 0.3478,
      "step": 13666
    },
    {
      "epoch": 0.8560735370112279,
      "grad_norm": 0.824791442855143,
      "learning_rate": 5.33169646341411e-07,
      "loss": 0.3944,
      "step": 13667
    },
    {
      "epoch": 0.8561361750105702,
      "grad_norm": 0.6052659712595495,
      "learning_rate": 5.327139389567537e-07,
      "loss": 0.4442,
      "step": 13668
    },
    {
      "epoch": 0.8561988130099125,
      "grad_norm": 0.8013745162958942,
      "learning_rate": 5.322584154464211e-07,
      "loss": 0.39,
      "step": 13669
    },
    {
      "epoch": 0.8562614510092548,
      "grad_norm": 0.9694743385758487,
      "learning_rate": 5.318030758291609e-07,
      "loss": 0.353,
      "step": 13670
    },
    {
      "epoch": 0.856324089008597,
      "grad_norm": 0.8989776325995084,
      "learning_rate": 5.313479201237149e-07,
      "loss": 0.3594,
      "step": 13671
    },
    {
      "epoch": 0.8563867270079394,
      "grad_norm": 0.8046220201450625,
      "learning_rate": 5.30892948348819e-07,
      "loss": 0.3732,
      "step": 13672
    },
    {
      "epoch": 0.8564493650072816,
      "grad_norm": 0.9100291734290538,
      "learning_rate": 5.304381605231985e-07,
      "loss": 0.4338,
      "step": 13673
    },
    {
      "epoch": 0.856512003006624,
      "grad_norm": 0.8854894896867925,
      "learning_rate": 5.299835566655732e-07,
      "loss": 0.3961,
      "step": 13674
    },
    {
      "epoch": 0.8565746410059663,
      "grad_norm": 0.9416181672450481,
      "learning_rate": 5.295291367946542e-07,
      "loss": 0.3833,
      "step": 13675
    },
    {
      "epoch": 0.8566372790053086,
      "grad_norm": 0.8842015682348576,
      "learning_rate": 5.290749009291457e-07,
      "loss": 0.386,
      "step": 13676
    },
    {
      "epoch": 0.8566999170046509,
      "grad_norm": 0.9377192425206471,
      "learning_rate": 5.286208490877431e-07,
      "loss": 0.3893,
      "step": 13677
    },
    {
      "epoch": 0.8567625550039931,
      "grad_norm": 0.9057716056299234,
      "learning_rate": 5.281669812891371e-07,
      "loss": 0.3732,
      "step": 13678
    },
    {
      "epoch": 0.8568251930033355,
      "grad_norm": 0.795428373762524,
      "learning_rate": 5.277132975520083e-07,
      "loss": 0.3944,
      "step": 13679
    },
    {
      "epoch": 0.8568878310026777,
      "grad_norm": 0.8848640798943369,
      "learning_rate": 5.272597978950294e-07,
      "loss": 0.4023,
      "step": 13680
    },
    {
      "epoch": 0.8569504690020201,
      "grad_norm": 0.9168144832975446,
      "learning_rate": 5.268064823368685e-07,
      "loss": 0.4085,
      "step": 13681
    },
    {
      "epoch": 0.8570131070013624,
      "grad_norm": 0.8416359483221546,
      "learning_rate": 5.263533508961827e-07,
      "loss": 0.3644,
      "step": 13682
    },
    {
      "epoch": 0.8570757450007047,
      "grad_norm": 0.876962275011439,
      "learning_rate": 5.259004035916237e-07,
      "loss": 0.4226,
      "step": 13683
    },
    {
      "epoch": 0.857138383000047,
      "grad_norm": 0.8608845327254689,
      "learning_rate": 5.254476404418341e-07,
      "loss": 0.384,
      "step": 13684
    },
    {
      "epoch": 0.8572010209993893,
      "grad_norm": 0.9225313709498415,
      "learning_rate": 5.249950614654509e-07,
      "loss": 0.4209,
      "step": 13685
    },
    {
      "epoch": 0.8572636589987316,
      "grad_norm": 0.9197207801602236,
      "learning_rate": 5.245426666810999e-07,
      "loss": 0.35,
      "step": 13686
    },
    {
      "epoch": 0.8573262969980738,
      "grad_norm": 0.927452834965069,
      "learning_rate": 5.240904561074045e-07,
      "loss": 0.3838,
      "step": 13687
    },
    {
      "epoch": 0.8573889349974162,
      "grad_norm": 0.8499331035056412,
      "learning_rate": 5.236384297629765e-07,
      "loss": 0.4205,
      "step": 13688
    },
    {
      "epoch": 0.8574515729967584,
      "grad_norm": 0.9312937488891896,
      "learning_rate": 5.23186587666421e-07,
      "loss": 0.4034,
      "step": 13689
    },
    {
      "epoch": 0.8575142109961008,
      "grad_norm": 0.8973578347504347,
      "learning_rate": 5.227349298363376e-07,
      "loss": 0.4027,
      "step": 13690
    },
    {
      "epoch": 0.8575768489954431,
      "grad_norm": 0.8652605209401537,
      "learning_rate": 5.222834562913148e-07,
      "loss": 0.3747,
      "step": 13691
    },
    {
      "epoch": 0.8576394869947854,
      "grad_norm": 0.8878975542338722,
      "learning_rate": 5.218321670499365e-07,
      "loss": 0.4354,
      "step": 13692
    },
    {
      "epoch": 0.8577021249941277,
      "grad_norm": 0.885429715112323,
      "learning_rate": 5.213810621307775e-07,
      "loss": 0.4013,
      "step": 13693
    },
    {
      "epoch": 0.85776476299347,
      "grad_norm": 0.5967361298901782,
      "learning_rate": 5.209301415524054e-07,
      "loss": 0.4394,
      "step": 13694
    },
    {
      "epoch": 0.8578274009928123,
      "grad_norm": 0.8655168563700054,
      "learning_rate": 5.204794053333784e-07,
      "loss": 0.3636,
      "step": 13695
    },
    {
      "epoch": 0.8578900389921545,
      "grad_norm": 0.8573350522078227,
      "learning_rate": 5.200288534922521e-07,
      "loss": 0.3625,
      "step": 13696
    },
    {
      "epoch": 0.8579526769914969,
      "grad_norm": 0.5609471891112332,
      "learning_rate": 5.195784860475684e-07,
      "loss": 0.4355,
      "step": 13697
    },
    {
      "epoch": 0.8580153149908392,
      "grad_norm": 0.8737498468310649,
      "learning_rate": 5.191283030178668e-07,
      "loss": 0.4006,
      "step": 13698
    },
    {
      "epoch": 0.8580779529901815,
      "grad_norm": 0.8760419187210269,
      "learning_rate": 5.186783044216765e-07,
      "loss": 0.4099,
      "step": 13699
    },
    {
      "epoch": 0.8581405909895238,
      "grad_norm": 0.8482603888506393,
      "learning_rate": 5.18228490277518e-07,
      "loss": 0.3755,
      "step": 13700
    },
    {
      "epoch": 0.8582032289888661,
      "grad_norm": 0.8758611476429602,
      "learning_rate": 5.177788606039075e-07,
      "loss": 0.4156,
      "step": 13701
    },
    {
      "epoch": 0.8582658669882084,
      "grad_norm": 0.8442818931315692,
      "learning_rate": 5.173294154193509e-07,
      "loss": 0.3719,
      "step": 13702
    },
    {
      "epoch": 0.8583285049875506,
      "grad_norm": 0.8371389265607231,
      "learning_rate": 5.168801547423474e-07,
      "loss": 0.3914,
      "step": 13703
    },
    {
      "epoch": 0.858391142986893,
      "grad_norm": 0.6401703711597313,
      "learning_rate": 5.164310785913878e-07,
      "loss": 0.4729,
      "step": 13704
    },
    {
      "epoch": 0.8584537809862353,
      "grad_norm": 0.962125601695983,
      "learning_rate": 5.15982186984958e-07,
      "loss": 0.3672,
      "step": 13705
    },
    {
      "epoch": 0.8585164189855776,
      "grad_norm": 0.9330807502874855,
      "learning_rate": 5.155334799415329e-07,
      "loss": 0.412,
      "step": 13706
    },
    {
      "epoch": 0.8585790569849199,
      "grad_norm": 0.8895689937792356,
      "learning_rate": 5.150849574795835e-07,
      "loss": 0.3788,
      "step": 13707
    },
    {
      "epoch": 0.8586416949842622,
      "grad_norm": 0.8031167978894036,
      "learning_rate": 5.146366196175689e-07,
      "loss": 0.3811,
      "step": 13708
    },
    {
      "epoch": 0.8587043329836045,
      "grad_norm": 0.8429234782105429,
      "learning_rate": 5.141884663739444e-07,
      "loss": 0.3752,
      "step": 13709
    },
    {
      "epoch": 0.8587669709829469,
      "grad_norm": 0.9547653877167716,
      "learning_rate": 5.137404977671546e-07,
      "loss": 0.4172,
      "step": 13710
    },
    {
      "epoch": 0.8588296089822891,
      "grad_norm": 0.8805012436908901,
      "learning_rate": 5.13292713815639e-07,
      "loss": 0.3667,
      "step": 13711
    },
    {
      "epoch": 0.8588922469816314,
      "grad_norm": 0.8652319340230994,
      "learning_rate": 5.128451145378278e-07,
      "loss": 0.4055,
      "step": 13712
    },
    {
      "epoch": 0.8589548849809737,
      "grad_norm": 0.9168331188597493,
      "learning_rate": 5.123976999521435e-07,
      "loss": 0.3792,
      "step": 13713
    },
    {
      "epoch": 0.859017522980316,
      "grad_norm": 0.8629591358433758,
      "learning_rate": 5.119504700770039e-07,
      "loss": 0.4033,
      "step": 13714
    },
    {
      "epoch": 0.8590801609796583,
      "grad_norm": 0.8334082580813527,
      "learning_rate": 5.115034249308154e-07,
      "loss": 0.4087,
      "step": 13715
    },
    {
      "epoch": 0.8591427989790006,
      "grad_norm": 0.9675392749821111,
      "learning_rate": 5.110565645319793e-07,
      "loss": 0.4087,
      "step": 13716
    },
    {
      "epoch": 0.859205436978343,
      "grad_norm": 0.8665322634921095,
      "learning_rate": 5.106098888988886e-07,
      "loss": 0.3809,
      "step": 13717
    },
    {
      "epoch": 0.8592680749776852,
      "grad_norm": 0.8370386940611247,
      "learning_rate": 5.101633980499283e-07,
      "loss": 0.3604,
      "step": 13718
    },
    {
      "epoch": 0.8593307129770276,
      "grad_norm": 0.8146188066929537,
      "learning_rate": 5.097170920034755e-07,
      "loss": 0.3715,
      "step": 13719
    },
    {
      "epoch": 0.8593933509763698,
      "grad_norm": 0.8477896573222959,
      "learning_rate": 5.092709707779008e-07,
      "loss": 0.4042,
      "step": 13720
    },
    {
      "epoch": 0.8594559889757121,
      "grad_norm": 0.9148307549451704,
      "learning_rate": 5.088250343915663e-07,
      "loss": 0.3663,
      "step": 13721
    },
    {
      "epoch": 0.8595186269750544,
      "grad_norm": 0.8653460522491639,
      "learning_rate": 5.083792828628264e-07,
      "loss": 0.3783,
      "step": 13722
    },
    {
      "epoch": 0.8595812649743967,
      "grad_norm": 0.8758961438646781,
      "learning_rate": 5.079337162100284e-07,
      "loss": 0.3485,
      "step": 13723
    },
    {
      "epoch": 0.859643902973739,
      "grad_norm": 0.6048639263220748,
      "learning_rate": 5.074883344515136e-07,
      "loss": 0.4426,
      "step": 13724
    },
    {
      "epoch": 0.8597065409730813,
      "grad_norm": 0.891984820349102,
      "learning_rate": 5.070431376056129e-07,
      "loss": 0.3982,
      "step": 13725
    },
    {
      "epoch": 0.8597691789724237,
      "grad_norm": 0.8216029025570459,
      "learning_rate": 5.065981256906505e-07,
      "loss": 0.3517,
      "step": 13726
    },
    {
      "epoch": 0.8598318169717659,
      "grad_norm": 0.8552427114359128,
      "learning_rate": 5.061532987249434e-07,
      "loss": 0.3558,
      "step": 13727
    },
    {
      "epoch": 0.8598944549711082,
      "grad_norm": 0.9741076932024423,
      "learning_rate": 5.057086567268005e-07,
      "loss": 0.3916,
      "step": 13728
    },
    {
      "epoch": 0.8599570929704505,
      "grad_norm": 0.8417000829041485,
      "learning_rate": 5.052641997145235e-07,
      "loss": 0.3992,
      "step": 13729
    },
    {
      "epoch": 0.8600197309697928,
      "grad_norm": 0.8657350759366026,
      "learning_rate": 5.048199277064054e-07,
      "loss": 0.3922,
      "step": 13730
    },
    {
      "epoch": 0.8600823689691351,
      "grad_norm": 0.8902305114716077,
      "learning_rate": 5.043758407207344e-07,
      "loss": 0.4075,
      "step": 13731
    },
    {
      "epoch": 0.8601450069684774,
      "grad_norm": 0.8467937480082643,
      "learning_rate": 5.039319387757874e-07,
      "loss": 0.3729,
      "step": 13732
    },
    {
      "epoch": 0.8602076449678198,
      "grad_norm": 0.9774390608383293,
      "learning_rate": 5.034882218898368e-07,
      "loss": 0.4039,
      "step": 13733
    },
    {
      "epoch": 0.860270282967162,
      "grad_norm": 0.8766293980688911,
      "learning_rate": 5.030446900811459e-07,
      "loss": 0.3699,
      "step": 13734
    },
    {
      "epoch": 0.8603329209665044,
      "grad_norm": 0.8421236938517598,
      "learning_rate": 5.026013433679699e-07,
      "loss": 0.3861,
      "step": 13735
    },
    {
      "epoch": 0.8603955589658466,
      "grad_norm": 0.8306821179874758,
      "learning_rate": 5.021581817685578e-07,
      "loss": 0.3841,
      "step": 13736
    },
    {
      "epoch": 0.8604581969651889,
      "grad_norm": 0.8642238597084698,
      "learning_rate": 5.017152053011499e-07,
      "loss": 0.3447,
      "step": 13737
    },
    {
      "epoch": 0.8605208349645312,
      "grad_norm": 0.8943175983776553,
      "learning_rate": 5.012724139839786e-07,
      "loss": 0.414,
      "step": 13738
    },
    {
      "epoch": 0.8605834729638735,
      "grad_norm": 0.6156659528991284,
      "learning_rate": 5.008298078352686e-07,
      "loss": 0.4416,
      "step": 13739
    },
    {
      "epoch": 0.8606461109632159,
      "grad_norm": 0.7898717757786741,
      "learning_rate": 5.0038738687324e-07,
      "loss": 0.3505,
      "step": 13740
    },
    {
      "epoch": 0.8607087489625581,
      "grad_norm": 0.881015970486942,
      "learning_rate": 4.999451511161013e-07,
      "loss": 0.3789,
      "step": 13741
    },
    {
      "epoch": 0.8607713869619005,
      "grad_norm": 0.8408227839598164,
      "learning_rate": 4.995031005820561e-07,
      "loss": 0.3874,
      "step": 13742
    },
    {
      "epoch": 0.8608340249612427,
      "grad_norm": 0.8503344680206444,
      "learning_rate": 4.990612352892982e-07,
      "loss": 0.3678,
      "step": 13743
    },
    {
      "epoch": 0.8608966629605851,
      "grad_norm": 0.9000409448255654,
      "learning_rate": 4.986195552560158e-07,
      "loss": 0.3842,
      "step": 13744
    },
    {
      "epoch": 0.8609593009599273,
      "grad_norm": 0.8741976566138334,
      "learning_rate": 4.981780605003872e-07,
      "loss": 0.3492,
      "step": 13745
    },
    {
      "epoch": 0.8610219389592696,
      "grad_norm": 0.9111996215527114,
      "learning_rate": 4.977367510405861e-07,
      "loss": 0.3799,
      "step": 13746
    },
    {
      "epoch": 0.861084576958612,
      "grad_norm": 0.8117166103369041,
      "learning_rate": 4.972956268947754e-07,
      "loss": 0.3907,
      "step": 13747
    },
    {
      "epoch": 0.8611472149579542,
      "grad_norm": 0.9084631019085709,
      "learning_rate": 4.96854688081112e-07,
      "loss": 0.3875,
      "step": 13748
    },
    {
      "epoch": 0.8612098529572966,
      "grad_norm": 0.897190040670462,
      "learning_rate": 4.96413934617746e-07,
      "loss": 0.4214,
      "step": 13749
    },
    {
      "epoch": 0.8612724909566388,
      "grad_norm": 0.8388433841765985,
      "learning_rate": 4.959733665228173e-07,
      "loss": 0.3932,
      "step": 13750
    },
    {
      "epoch": 0.8613351289559812,
      "grad_norm": 0.8903657351273051,
      "learning_rate": 4.955329838144624e-07,
      "loss": 0.4135,
      "step": 13751
    },
    {
      "epoch": 0.8613977669553234,
      "grad_norm": 0.8654030176744897,
      "learning_rate": 4.950927865108051e-07,
      "loss": 0.35,
      "step": 13752
    },
    {
      "epoch": 0.8614604049546657,
      "grad_norm": 0.8891932726958427,
      "learning_rate": 4.946527746299656e-07,
      "loss": 0.3997,
      "step": 13753
    },
    {
      "epoch": 0.8615230429540081,
      "grad_norm": 0.7780057065769187,
      "learning_rate": 4.942129481900537e-07,
      "loss": 0.3559,
      "step": 13754
    },
    {
      "epoch": 0.8615856809533503,
      "grad_norm": 0.8746983935484882,
      "learning_rate": 4.937733072091733e-07,
      "loss": 0.4408,
      "step": 13755
    },
    {
      "epoch": 0.8616483189526927,
      "grad_norm": 0.8569465589399727,
      "learning_rate": 4.933338517054193e-07,
      "loss": 0.3951,
      "step": 13756
    },
    {
      "epoch": 0.8617109569520349,
      "grad_norm": 0.8864322420174312,
      "learning_rate": 4.928945816968811e-07,
      "loss": 0.398,
      "step": 13757
    },
    {
      "epoch": 0.8617735949513773,
      "grad_norm": 0.9001170112304511,
      "learning_rate": 4.924554972016382e-07,
      "loss": 0.3856,
      "step": 13758
    },
    {
      "epoch": 0.8618362329507195,
      "grad_norm": 0.8890179277310638,
      "learning_rate": 4.920165982377645e-07,
      "loss": 0.3828,
      "step": 13759
    },
    {
      "epoch": 0.8618988709500619,
      "grad_norm": 0.9138836567815534,
      "learning_rate": 4.915778848233227e-07,
      "loss": 0.3586,
      "step": 13760
    },
    {
      "epoch": 0.8619615089494042,
      "grad_norm": 0.8557558411493272,
      "learning_rate": 4.911393569763728e-07,
      "loss": 0.3834,
      "step": 13761
    },
    {
      "epoch": 0.8620241469487464,
      "grad_norm": 0.8242477169403968,
      "learning_rate": 4.907010147149643e-07,
      "loss": 0.3769,
      "step": 13762
    },
    {
      "epoch": 0.8620867849480888,
      "grad_norm": 0.8766060725910297,
      "learning_rate": 4.902628580571389e-07,
      "loss": 0.3498,
      "step": 13763
    },
    {
      "epoch": 0.862149422947431,
      "grad_norm": 0.8590489190228072,
      "learning_rate": 4.89824887020931e-07,
      "loss": 0.3719,
      "step": 13764
    },
    {
      "epoch": 0.8622120609467734,
      "grad_norm": 0.9710226342372779,
      "learning_rate": 4.893871016243673e-07,
      "loss": 0.3893,
      "step": 13765
    },
    {
      "epoch": 0.8622746989461156,
      "grad_norm": 0.9356705850790152,
      "learning_rate": 4.889495018854684e-07,
      "loss": 0.375,
      "step": 13766
    },
    {
      "epoch": 0.862337336945458,
      "grad_norm": 0.8351262949349904,
      "learning_rate": 4.885120878222455e-07,
      "loss": 0.3814,
      "step": 13767
    },
    {
      "epoch": 0.8623999749448003,
      "grad_norm": 0.8044872445154431,
      "learning_rate": 4.880748594527024e-07,
      "loss": 0.3308,
      "step": 13768
    },
    {
      "epoch": 0.8624626129441426,
      "grad_norm": 0.8589294315259303,
      "learning_rate": 4.876378167948348e-07,
      "loss": 0.3761,
      "step": 13769
    },
    {
      "epoch": 0.8625252509434849,
      "grad_norm": 0.8133521086896282,
      "learning_rate": 4.872009598666327e-07,
      "loss": 0.3758,
      "step": 13770
    },
    {
      "epoch": 0.8625878889428271,
      "grad_norm": 0.8298187635633326,
      "learning_rate": 4.867642886860769e-07,
      "loss": 0.3772,
      "step": 13771
    },
    {
      "epoch": 0.8626505269421695,
      "grad_norm": 0.8457973281226696,
      "learning_rate": 4.863278032711405e-07,
      "loss": 0.3571,
      "step": 13772
    },
    {
      "epoch": 0.8627131649415117,
      "grad_norm": 0.865475005926299,
      "learning_rate": 4.858915036397899e-07,
      "loss": 0.3851,
      "step": 13773
    },
    {
      "epoch": 0.8627758029408541,
      "grad_norm": 0.8383037165284962,
      "learning_rate": 4.854553898099817e-07,
      "loss": 0.3608,
      "step": 13774
    },
    {
      "epoch": 0.8628384409401963,
      "grad_norm": 0.8727173896872842,
      "learning_rate": 4.850194617996684e-07,
      "loss": 0.3779,
      "step": 13775
    },
    {
      "epoch": 0.8629010789395387,
      "grad_norm": 0.8682526222668105,
      "learning_rate": 4.845837196267922e-07,
      "loss": 0.3959,
      "step": 13776
    },
    {
      "epoch": 0.862963716938881,
      "grad_norm": 0.8367348947381307,
      "learning_rate": 4.841481633092881e-07,
      "loss": 0.3461,
      "step": 13777
    },
    {
      "epoch": 0.8630263549382233,
      "grad_norm": 0.8503502086892161,
      "learning_rate": 4.837127928650831e-07,
      "loss": 0.3956,
      "step": 13778
    },
    {
      "epoch": 0.8630889929375656,
      "grad_norm": 0.8891158897321481,
      "learning_rate": 4.832776083120983e-07,
      "loss": 0.3994,
      "step": 13779
    },
    {
      "epoch": 0.8631516309369078,
      "grad_norm": 0.8895065971027935,
      "learning_rate": 4.828426096682459e-07,
      "loss": 0.3668,
      "step": 13780
    },
    {
      "epoch": 0.8632142689362502,
      "grad_norm": 0.8071677192741932,
      "learning_rate": 4.824077969514301e-07,
      "loss": 0.3943,
      "step": 13781
    },
    {
      "epoch": 0.8632769069355924,
      "grad_norm": 0.8543507102303598,
      "learning_rate": 4.819731701795466e-07,
      "loss": 0.3573,
      "step": 13782
    },
    {
      "epoch": 0.8633395449349348,
      "grad_norm": 0.8951818612899446,
      "learning_rate": 4.815387293704871e-07,
      "loss": 0.3932,
      "step": 13783
    },
    {
      "epoch": 0.8634021829342771,
      "grad_norm": 0.9442320673352149,
      "learning_rate": 4.811044745421323e-07,
      "loss": 0.3739,
      "step": 13784
    },
    {
      "epoch": 0.8634648209336194,
      "grad_norm": 0.867871468874045,
      "learning_rate": 4.806704057123562e-07,
      "loss": 0.3431,
      "step": 13785
    },
    {
      "epoch": 0.8635274589329617,
      "grad_norm": 0.9183000673225998,
      "learning_rate": 4.802365228990247e-07,
      "loss": 0.3896,
      "step": 13786
    },
    {
      "epoch": 0.8635900969323039,
      "grad_norm": 0.8546154030113424,
      "learning_rate": 4.798028261199961e-07,
      "loss": 0.3561,
      "step": 13787
    },
    {
      "epoch": 0.8636527349316463,
      "grad_norm": 0.8769919651670247,
      "learning_rate": 4.793693153931228e-07,
      "loss": 0.3653,
      "step": 13788
    },
    {
      "epoch": 0.8637153729309885,
      "grad_norm": 0.8145536228873638,
      "learning_rate": 4.789359907362479e-07,
      "loss": 0.3743,
      "step": 13789
    },
    {
      "epoch": 0.8637780109303309,
      "grad_norm": 0.7839745253760134,
      "learning_rate": 4.785028521672063e-07,
      "loss": 0.3494,
      "step": 13790
    },
    {
      "epoch": 0.8638406489296732,
      "grad_norm": 0.8601428138122504,
      "learning_rate": 4.780698997038264e-07,
      "loss": 0.3903,
      "step": 13791
    },
    {
      "epoch": 0.8639032869290155,
      "grad_norm": 0.8653211815365994,
      "learning_rate": 4.77637133363929e-07,
      "loss": 0.3692,
      "step": 13792
    },
    {
      "epoch": 0.8639659249283578,
      "grad_norm": 0.8896887147478905,
      "learning_rate": 4.772045531653269e-07,
      "loss": 0.3545,
      "step": 13793
    },
    {
      "epoch": 0.8640285629277001,
      "grad_norm": 0.5863800216636046,
      "learning_rate": 4.767721591258245e-07,
      "loss": 0.4449,
      "step": 13794
    },
    {
      "epoch": 0.8640912009270424,
      "grad_norm": 0.9042727896399118,
      "learning_rate": 4.7633995126321996e-07,
      "loss": 0.3728,
      "step": 13795
    },
    {
      "epoch": 0.8641538389263846,
      "grad_norm": 0.7484176358255392,
      "learning_rate": 4.759079295953017e-07,
      "loss": 0.343,
      "step": 13796
    },
    {
      "epoch": 0.864216476925727,
      "grad_norm": 0.8354783191604294,
      "learning_rate": 4.754760941398534e-07,
      "loss": 0.3768,
      "step": 13797
    },
    {
      "epoch": 0.8642791149250693,
      "grad_norm": 0.8649937292027429,
      "learning_rate": 4.7504444491464907e-07,
      "loss": 0.3918,
      "step": 13798
    },
    {
      "epoch": 0.8643417529244116,
      "grad_norm": 0.8754148712436157,
      "learning_rate": 4.746129819374551e-07,
      "loss": 0.3723,
      "step": 13799
    },
    {
      "epoch": 0.8644043909237539,
      "grad_norm": 0.7646263675282471,
      "learning_rate": 4.7418170522603e-07,
      "loss": 0.3643,
      "step": 13800
    },
    {
      "epoch": 0.8644670289230962,
      "grad_norm": 0.9299180038050359,
      "learning_rate": 4.737506147981269e-07,
      "loss": 0.3915,
      "step": 13801
    },
    {
      "epoch": 0.8645296669224385,
      "grad_norm": 0.8579285173198999,
      "learning_rate": 4.733197106714887e-07,
      "loss": 0.403,
      "step": 13802
    },
    {
      "epoch": 0.8645923049217809,
      "grad_norm": 0.9095600742703492,
      "learning_rate": 4.7288899286385136e-07,
      "loss": 0.376,
      "step": 13803
    },
    {
      "epoch": 0.8646549429211231,
      "grad_norm": 0.9248934653752223,
      "learning_rate": 4.7245846139294283e-07,
      "loss": 0.3796,
      "step": 13804
    },
    {
      "epoch": 0.8647175809204654,
      "grad_norm": 0.8953194611183068,
      "learning_rate": 4.72028116276485e-07,
      "loss": 0.3854,
      "step": 13805
    },
    {
      "epoch": 0.8647802189198077,
      "grad_norm": 0.8991398728706873,
      "learning_rate": 4.715979575321894e-07,
      "loss": 0.4381,
      "step": 13806
    },
    {
      "epoch": 0.86484285691915,
      "grad_norm": 0.9120380156888438,
      "learning_rate": 4.7116798517776283e-07,
      "loss": 0.3803,
      "step": 13807
    },
    {
      "epoch": 0.8649054949184923,
      "grad_norm": 0.8855047240022172,
      "learning_rate": 4.707381992309018e-07,
      "loss": 0.3996,
      "step": 13808
    },
    {
      "epoch": 0.8649681329178346,
      "grad_norm": 0.8033432435832717,
      "learning_rate": 4.703085997092982e-07,
      "loss": 0.3337,
      "step": 13809
    },
    {
      "epoch": 0.865030770917177,
      "grad_norm": 0.8740501804905639,
      "learning_rate": 4.698791866306335e-07,
      "loss": 0.3395,
      "step": 13810
    },
    {
      "epoch": 0.8650934089165192,
      "grad_norm": 0.8142502506400007,
      "learning_rate": 4.694499600125824e-07,
      "loss": 0.3744,
      "step": 13811
    },
    {
      "epoch": 0.8651560469158615,
      "grad_norm": 0.8500607463718322,
      "learning_rate": 4.690209198728113e-07,
      "loss": 0.3868,
      "step": 13812
    },
    {
      "epoch": 0.8652186849152038,
      "grad_norm": 0.8080796702290612,
      "learning_rate": 4.6859206622898055e-07,
      "loss": 0.3911,
      "step": 13813
    },
    {
      "epoch": 0.8652813229145461,
      "grad_norm": 0.9049681994754131,
      "learning_rate": 4.681633990987416e-07,
      "loss": 0.4093,
      "step": 13814
    },
    {
      "epoch": 0.8653439609138884,
      "grad_norm": 0.8439248722408916,
      "learning_rate": 4.6773491849973707e-07,
      "loss": 0.4183,
      "step": 13815
    },
    {
      "epoch": 0.8654065989132307,
      "grad_norm": 0.9286631619802013,
      "learning_rate": 4.673066244496055e-07,
      "loss": 0.4125,
      "step": 13816
    },
    {
      "epoch": 0.865469236912573,
      "grad_norm": 0.8027528298815414,
      "learning_rate": 4.668785169659734e-07,
      "loss": 0.3412,
      "step": 13817
    },
    {
      "epoch": 0.8655318749119153,
      "grad_norm": 0.8744693270986267,
      "learning_rate": 4.664505960664639e-07,
      "loss": 0.3953,
      "step": 13818
    },
    {
      "epoch": 0.8655945129112577,
      "grad_norm": 0.8913720091561943,
      "learning_rate": 4.660228617686896e-07,
      "loss": 0.3887,
      "step": 13819
    },
    {
      "epoch": 0.8656571509105999,
      "grad_norm": 0.872721748154102,
      "learning_rate": 4.6559531409025584e-07,
      "loss": 0.4169,
      "step": 13820
    },
    {
      "epoch": 0.8657197889099422,
      "grad_norm": 0.8019018608019984,
      "learning_rate": 4.6516795304876074e-07,
      "loss": 0.3709,
      "step": 13821
    },
    {
      "epoch": 0.8657824269092845,
      "grad_norm": 0.6227030743189155,
      "learning_rate": 4.6474077866179413e-07,
      "loss": 0.4517,
      "step": 13822
    },
    {
      "epoch": 0.8658450649086268,
      "grad_norm": 0.8932789221618552,
      "learning_rate": 4.643137909469386e-07,
      "loss": 0.4017,
      "step": 13823
    },
    {
      "epoch": 0.8659077029079691,
      "grad_norm": 0.8888013966685622,
      "learning_rate": 4.6388698992176904e-07,
      "loss": 0.4196,
      "step": 13824
    },
    {
      "epoch": 0.8659703409073114,
      "grad_norm": 0.5789002103597685,
      "learning_rate": 4.634603756038536e-07,
      "loss": 0.4393,
      "step": 13825
    },
    {
      "epoch": 0.8660329789066538,
      "grad_norm": 0.8497385132082178,
      "learning_rate": 4.6303394801074987e-07,
      "loss": 0.3731,
      "step": 13826
    },
    {
      "epoch": 0.866095616905996,
      "grad_norm": 0.9445724191633232,
      "learning_rate": 4.626077071600121e-07,
      "loss": 0.436,
      "step": 13827
    },
    {
      "epoch": 0.8661582549053384,
      "grad_norm": 0.8979856956270541,
      "learning_rate": 4.6218165306918304e-07,
      "loss": 0.3693,
      "step": 13828
    },
    {
      "epoch": 0.8662208929046806,
      "grad_norm": 0.8496615766506856,
      "learning_rate": 4.617557857557997e-07,
      "loss": 0.3939,
      "step": 13829
    },
    {
      "epoch": 0.8662835309040229,
      "grad_norm": 0.8383373980050857,
      "learning_rate": 4.6133010523739033e-07,
      "loss": 0.3898,
      "step": 13830
    },
    {
      "epoch": 0.8663461689033652,
      "grad_norm": 0.911310906065585,
      "learning_rate": 4.609046115314764e-07,
      "loss": 0.4036,
      "step": 13831
    },
    {
      "epoch": 0.8664088069027075,
      "grad_norm": 0.9280652719587484,
      "learning_rate": 4.604793046555711e-07,
      "loss": 0.3895,
      "step": 13832
    },
    {
      "epoch": 0.8664714449020499,
      "grad_norm": 0.8037297671914061,
      "learning_rate": 4.6005418462717944e-07,
      "loss": 0.3606,
      "step": 13833
    },
    {
      "epoch": 0.8665340829013921,
      "grad_norm": 0.9142486345439699,
      "learning_rate": 4.596292514638001e-07,
      "loss": 0.4045,
      "step": 13834
    },
    {
      "epoch": 0.8665967209007345,
      "grad_norm": 0.845092642888046,
      "learning_rate": 4.5920450518292414e-07,
      "loss": 0.3852,
      "step": 13835
    },
    {
      "epoch": 0.8666593589000767,
      "grad_norm": 0.7999766935654136,
      "learning_rate": 4.587799458020342e-07,
      "loss": 0.3565,
      "step": 13836
    },
    {
      "epoch": 0.866721996899419,
      "grad_norm": 0.8919019898054643,
      "learning_rate": 4.5835557333860414e-07,
      "loss": 0.3682,
      "step": 13837
    },
    {
      "epoch": 0.8667846348987613,
      "grad_norm": 0.8759318615884889,
      "learning_rate": 4.579313878101016e-07,
      "loss": 0.3841,
      "step": 13838
    },
    {
      "epoch": 0.8668472728981036,
      "grad_norm": 0.8089519065227289,
      "learning_rate": 4.5750738923398595e-07,
      "loss": 0.3825,
      "step": 13839
    },
    {
      "epoch": 0.866909910897446,
      "grad_norm": 0.9463892895456552,
      "learning_rate": 4.5708357762770993e-07,
      "loss": 0.3938,
      "step": 13840
    },
    {
      "epoch": 0.8669725488967882,
      "grad_norm": 0.8656091335346049,
      "learning_rate": 4.566599530087157e-07,
      "loss": 0.3759,
      "step": 13841
    },
    {
      "epoch": 0.8670351868961306,
      "grad_norm": 0.8542633373212092,
      "learning_rate": 4.5623651539444204e-07,
      "loss": 0.4094,
      "step": 13842
    },
    {
      "epoch": 0.8670978248954728,
      "grad_norm": 0.8538734474958843,
      "learning_rate": 4.5581326480231616e-07,
      "loss": 0.3818,
      "step": 13843
    },
    {
      "epoch": 0.8671604628948152,
      "grad_norm": 0.8655348528370287,
      "learning_rate": 4.553902012497602e-07,
      "loss": 0.377,
      "step": 13844
    },
    {
      "epoch": 0.8672231008941574,
      "grad_norm": 0.8491655335605421,
      "learning_rate": 4.549673247541875e-07,
      "loss": 0.3738,
      "step": 13845
    },
    {
      "epoch": 0.8672857388934997,
      "grad_norm": 0.8221681285006079,
      "learning_rate": 4.54544635333003e-07,
      "loss": 0.3795,
      "step": 13846
    },
    {
      "epoch": 0.867348376892842,
      "grad_norm": 0.845634941347183,
      "learning_rate": 4.541221330036055e-07,
      "loss": 0.3463,
      "step": 13847
    },
    {
      "epoch": 0.8674110148921843,
      "grad_norm": 0.8379125806179762,
      "learning_rate": 4.53699817783384e-07,
      "loss": 0.3443,
      "step": 13848
    },
    {
      "epoch": 0.8674736528915267,
      "grad_norm": 0.8925230543440063,
      "learning_rate": 4.532776896897223e-07,
      "loss": 0.3836,
      "step": 13849
    },
    {
      "epoch": 0.8675362908908689,
      "grad_norm": 0.8611948146945935,
      "learning_rate": 4.5285574873999374e-07,
      "loss": 0.3686,
      "step": 13850
    },
    {
      "epoch": 0.8675989288902113,
      "grad_norm": 0.8398999987053327,
      "learning_rate": 4.524339949515677e-07,
      "loss": 0.3573,
      "step": 13851
    },
    {
      "epoch": 0.8676615668895535,
      "grad_norm": 0.8619327730379104,
      "learning_rate": 4.5201242834180146e-07,
      "loss": 0.4405,
      "step": 13852
    },
    {
      "epoch": 0.8677242048888959,
      "grad_norm": 0.8402619212590428,
      "learning_rate": 4.515910489280484e-07,
      "loss": 0.3582,
      "step": 13853
    },
    {
      "epoch": 0.8677868428882382,
      "grad_norm": 0.8972680099893688,
      "learning_rate": 4.511698567276523e-07,
      "loss": 0.4085,
      "step": 13854
    },
    {
      "epoch": 0.8678494808875804,
      "grad_norm": 0.8751854350606874,
      "learning_rate": 4.507488517579495e-07,
      "loss": 0.3866,
      "step": 13855
    },
    {
      "epoch": 0.8679121188869228,
      "grad_norm": 0.8594588059574562,
      "learning_rate": 4.5032803403626757e-07,
      "loss": 0.3626,
      "step": 13856
    },
    {
      "epoch": 0.867974756886265,
      "grad_norm": 0.9104522090164895,
      "learning_rate": 4.4990740357992893e-07,
      "loss": 0.4172,
      "step": 13857
    },
    {
      "epoch": 0.8680373948856074,
      "grad_norm": 0.8935177577542089,
      "learning_rate": 4.4948696040624576e-07,
      "loss": 0.3577,
      "step": 13858
    },
    {
      "epoch": 0.8681000328849496,
      "grad_norm": 0.909545967000386,
      "learning_rate": 4.490667045325231e-07,
      "loss": 0.3553,
      "step": 13859
    },
    {
      "epoch": 0.868162670884292,
      "grad_norm": 0.9007230130304591,
      "learning_rate": 4.4864663597606053e-07,
      "loss": 0.3598,
      "step": 13860
    },
    {
      "epoch": 0.8682253088836343,
      "grad_norm": 0.8299254824121292,
      "learning_rate": 4.482267547541463e-07,
      "loss": 0.3595,
      "step": 13861
    },
    {
      "epoch": 0.8682879468829765,
      "grad_norm": 0.8865958016141636,
      "learning_rate": 4.4780706088406455e-07,
      "loss": 0.3749,
      "step": 13862
    },
    {
      "epoch": 0.8683505848823189,
      "grad_norm": 0.858016752664445,
      "learning_rate": 4.4738755438308855e-07,
      "loss": 0.3628,
      "step": 13863
    },
    {
      "epoch": 0.8684132228816611,
      "grad_norm": 0.9210708565953851,
      "learning_rate": 4.469682352684868e-07,
      "loss": 0.3592,
      "step": 13864
    },
    {
      "epoch": 0.8684758608810035,
      "grad_norm": 0.8564218084407555,
      "learning_rate": 4.46549103557517e-07,
      "loss": 0.4614,
      "step": 13865
    },
    {
      "epoch": 0.8685384988803457,
      "grad_norm": 0.9124081251975897,
      "learning_rate": 4.461301592674311e-07,
      "loss": 0.3749,
      "step": 13866
    },
    {
      "epoch": 0.8686011368796881,
      "grad_norm": 0.9789294657262992,
      "learning_rate": 4.457114024154724e-07,
      "loss": 0.4337,
      "step": 13867
    },
    {
      "epoch": 0.8686637748790303,
      "grad_norm": 0.8111321819538473,
      "learning_rate": 4.452928330188788e-07,
      "loss": 0.3164,
      "step": 13868
    },
    {
      "epoch": 0.8687264128783727,
      "grad_norm": 0.8889581242826818,
      "learning_rate": 4.4487445109487715e-07,
      "loss": 0.3899,
      "step": 13869
    },
    {
      "epoch": 0.868789050877715,
      "grad_norm": 0.8095997911433007,
      "learning_rate": 4.4445625666068805e-07,
      "loss": 0.3255,
      "step": 13870
    },
    {
      "epoch": 0.8688516888770572,
      "grad_norm": 0.8331175746052548,
      "learning_rate": 4.4403824973352603e-07,
      "loss": 0.3491,
      "step": 13871
    },
    {
      "epoch": 0.8689143268763996,
      "grad_norm": 0.7515904347767565,
      "learning_rate": 4.4362043033059465e-07,
      "loss": 0.3566,
      "step": 13872
    },
    {
      "epoch": 0.8689769648757418,
      "grad_norm": 0.873253877375975,
      "learning_rate": 4.4320279846909234e-07,
      "loss": 0.3951,
      "step": 13873
    },
    {
      "epoch": 0.8690396028750842,
      "grad_norm": 0.8247141535269152,
      "learning_rate": 4.4278535416620914e-07,
      "loss": 0.3599,
      "step": 13874
    },
    {
      "epoch": 0.8691022408744264,
      "grad_norm": 0.8640043254086202,
      "learning_rate": 4.423680974391259e-07,
      "loss": 0.3965,
      "step": 13875
    },
    {
      "epoch": 0.8691648788737688,
      "grad_norm": 0.6846731395503498,
      "learning_rate": 4.4195102830501713e-07,
      "loss": 0.4325,
      "step": 13876
    },
    {
      "epoch": 0.8692275168731111,
      "grad_norm": 0.8089529326371696,
      "learning_rate": 4.415341467810508e-07,
      "loss": 0.3688,
      "step": 13877
    },
    {
      "epoch": 0.8692901548724534,
      "grad_norm": 0.813189553611974,
      "learning_rate": 4.4111745288438534e-07,
      "loss": 0.3679,
      "step": 13878
    },
    {
      "epoch": 0.8693527928717957,
      "grad_norm": 0.9349181066654012,
      "learning_rate": 4.4070094663217045e-07,
      "loss": 0.3805,
      "step": 13879
    },
    {
      "epoch": 0.8694154308711379,
      "grad_norm": 0.8259697898310369,
      "learning_rate": 4.402846280415518e-07,
      "loss": 0.389,
      "step": 13880
    },
    {
      "epoch": 0.8694780688704803,
      "grad_norm": 0.8420366197476745,
      "learning_rate": 4.3986849712966463e-07,
      "loss": 0.3756,
      "step": 13881
    },
    {
      "epoch": 0.8695407068698225,
      "grad_norm": 0.8740170115271229,
      "learning_rate": 4.394525539136363e-07,
      "loss": 0.3967,
      "step": 13882
    },
    {
      "epoch": 0.8696033448691649,
      "grad_norm": 0.8515608594632003,
      "learning_rate": 4.39036798410587e-07,
      "loss": 0.4007,
      "step": 13883
    },
    {
      "epoch": 0.8696659828685072,
      "grad_norm": 0.6120517274976706,
      "learning_rate": 4.3862123063762973e-07,
      "loss": 0.4328,
      "step": 13884
    },
    {
      "epoch": 0.8697286208678495,
      "grad_norm": 0.8516000315963618,
      "learning_rate": 4.382058506118686e-07,
      "loss": 0.3951,
      "step": 13885
    },
    {
      "epoch": 0.8697912588671918,
      "grad_norm": 0.8344926599356235,
      "learning_rate": 4.3779065835040213e-07,
      "loss": 0.3749,
      "step": 13886
    },
    {
      "epoch": 0.8698538968665341,
      "grad_norm": 0.8931245736541353,
      "learning_rate": 4.3737565387031886e-07,
      "loss": 0.3549,
      "step": 13887
    },
    {
      "epoch": 0.8699165348658764,
      "grad_norm": 0.8572852790489308,
      "learning_rate": 4.3696083718869965e-07,
      "loss": 0.4382,
      "step": 13888
    },
    {
      "epoch": 0.8699791728652186,
      "grad_norm": 0.8628940905655298,
      "learning_rate": 4.3654620832262075e-07,
      "loss": 0.3887,
      "step": 13889
    },
    {
      "epoch": 0.870041810864561,
      "grad_norm": 0.8198091114303546,
      "learning_rate": 4.361317672891463e-07,
      "loss": 0.3891,
      "step": 13890
    },
    {
      "epoch": 0.8701044488639033,
      "grad_norm": 0.8987995747482761,
      "learning_rate": 4.3571751410533545e-07,
      "loss": 0.3645,
      "step": 13891
    },
    {
      "epoch": 0.8701670868632456,
      "grad_norm": 0.8776714206541099,
      "learning_rate": 4.3530344878823907e-07,
      "loss": 0.3856,
      "step": 13892
    },
    {
      "epoch": 0.8702297248625879,
      "grad_norm": 0.6165117133523341,
      "learning_rate": 4.3488957135489895e-07,
      "loss": 0.4526,
      "step": 13893
    },
    {
      "epoch": 0.8702923628619302,
      "grad_norm": 0.8162826875658168,
      "learning_rate": 4.344758818223527e-07,
      "loss": 0.3955,
      "step": 13894
    },
    {
      "epoch": 0.8703550008612725,
      "grad_norm": 0.8069022734048876,
      "learning_rate": 4.3406238020762605e-07,
      "loss": 0.3914,
      "step": 13895
    },
    {
      "epoch": 0.8704176388606147,
      "grad_norm": 0.8503107281342543,
      "learning_rate": 4.336490665277393e-07,
      "loss": 0.3914,
      "step": 13896
    },
    {
      "epoch": 0.8704802768599571,
      "grad_norm": 0.789785898744827,
      "learning_rate": 4.33235940799705e-07,
      "loss": 0.3957,
      "step": 13897
    },
    {
      "epoch": 0.8705429148592994,
      "grad_norm": 0.8418693686418479,
      "learning_rate": 4.328230030405256e-07,
      "loss": 0.3318,
      "step": 13898
    },
    {
      "epoch": 0.8706055528586417,
      "grad_norm": 0.8637060931678925,
      "learning_rate": 4.324102532672003e-07,
      "loss": 0.3583,
      "step": 13899
    },
    {
      "epoch": 0.870668190857984,
      "grad_norm": 0.8458314315259697,
      "learning_rate": 4.319976914967167e-07,
      "loss": 0.3549,
      "step": 13900
    },
    {
      "epoch": 0.8707308288573263,
      "grad_norm": 0.8012902978533278,
      "learning_rate": 4.3158531774605605e-07,
      "loss": 0.3537,
      "step": 13901
    },
    {
      "epoch": 0.8707934668566686,
      "grad_norm": 0.9298577359103105,
      "learning_rate": 4.311731320321905e-07,
      "loss": 0.3846,
      "step": 13902
    },
    {
      "epoch": 0.870856104856011,
      "grad_norm": 0.7978926315588077,
      "learning_rate": 4.30761134372088e-07,
      "loss": 0.3401,
      "step": 13903
    },
    {
      "epoch": 0.8709187428553532,
      "grad_norm": 0.8898285396282537,
      "learning_rate": 4.303493247827051e-07,
      "loss": 0.4018,
      "step": 13904
    },
    {
      "epoch": 0.8709813808546955,
      "grad_norm": 0.9074306753158243,
      "learning_rate": 4.299377032809926e-07,
      "loss": 0.4164,
      "step": 13905
    },
    {
      "epoch": 0.8710440188540378,
      "grad_norm": 0.8896132463654455,
      "learning_rate": 4.2952626988389204e-07,
      "loss": 0.4352,
      "step": 13906
    },
    {
      "epoch": 0.8711066568533801,
      "grad_norm": 0.8256147879690885,
      "learning_rate": 4.2911502460833755e-07,
      "loss": 0.3801,
      "step": 13907
    },
    {
      "epoch": 0.8711692948527224,
      "grad_norm": 0.9259281708258635,
      "learning_rate": 4.2870396747125844e-07,
      "loss": 0.3746,
      "step": 13908
    },
    {
      "epoch": 0.8712319328520647,
      "grad_norm": 0.9193399712728869,
      "learning_rate": 4.282930984895722e-07,
      "loss": 0.3942,
      "step": 13909
    },
    {
      "epoch": 0.871294570851407,
      "grad_norm": 0.8896615538660791,
      "learning_rate": 4.2788241768019046e-07,
      "loss": 0.3384,
      "step": 13910
    },
    {
      "epoch": 0.8713572088507493,
      "grad_norm": 0.8328492388533009,
      "learning_rate": 4.274719250600162e-07,
      "loss": 0.3765,
      "step": 13911
    },
    {
      "epoch": 0.8714198468500917,
      "grad_norm": 0.8768002184253157,
      "learning_rate": 4.2706162064594713e-07,
      "loss": 0.4126,
      "step": 13912
    },
    {
      "epoch": 0.8714824848494339,
      "grad_norm": 0.8303829435246688,
      "learning_rate": 4.266515044548708e-07,
      "loss": 0.3718,
      "step": 13913
    },
    {
      "epoch": 0.8715451228487762,
      "grad_norm": 0.841158533688779,
      "learning_rate": 4.2624157650366706e-07,
      "loss": 0.3246,
      "step": 13914
    },
    {
      "epoch": 0.8716077608481185,
      "grad_norm": 0.8720825359106462,
      "learning_rate": 4.2583183680920905e-07,
      "loss": 0.3631,
      "step": 13915
    },
    {
      "epoch": 0.8716703988474608,
      "grad_norm": 0.8267263270454842,
      "learning_rate": 4.2542228538836104e-07,
      "loss": 0.4091,
      "step": 13916
    },
    {
      "epoch": 0.8717330368468031,
      "grad_norm": 0.8575235482032028,
      "learning_rate": 4.250129222579813e-07,
      "loss": 0.4016,
      "step": 13917
    },
    {
      "epoch": 0.8717956748461454,
      "grad_norm": 0.8792651210231471,
      "learning_rate": 4.2460374743491904e-07,
      "loss": 0.4145,
      "step": 13918
    },
    {
      "epoch": 0.8718583128454878,
      "grad_norm": 0.9213290077109294,
      "learning_rate": 4.2419476093601476e-07,
      "loss": 0.4103,
      "step": 13919
    },
    {
      "epoch": 0.87192095084483,
      "grad_norm": 0.8658992957756924,
      "learning_rate": 4.2378596277810435e-07,
      "loss": 0.3769,
      "step": 13920
    },
    {
      "epoch": 0.8719835888441723,
      "grad_norm": 0.8859925058293484,
      "learning_rate": 4.233773529780133e-07,
      "loss": 0.4031,
      "step": 13921
    },
    {
      "epoch": 0.8720462268435146,
      "grad_norm": 0.9508902777898273,
      "learning_rate": 4.229689315525598e-07,
      "loss": 0.4,
      "step": 13922
    },
    {
      "epoch": 0.8721088648428569,
      "grad_norm": 0.8334072039366847,
      "learning_rate": 4.225606985185543e-07,
      "loss": 0.3431,
      "step": 13923
    },
    {
      "epoch": 0.8721715028421992,
      "grad_norm": 0.7707005557231189,
      "learning_rate": 4.2215265389280057e-07,
      "loss": 0.3245,
      "step": 13924
    },
    {
      "epoch": 0.8722341408415415,
      "grad_norm": 0.8966814545196528,
      "learning_rate": 4.217447976920919e-07,
      "loss": 0.4051,
      "step": 13925
    },
    {
      "epoch": 0.8722967788408839,
      "grad_norm": 0.8181578126195963,
      "learning_rate": 4.213371299332181e-07,
      "loss": 0.3352,
      "step": 13926
    },
    {
      "epoch": 0.8723594168402261,
      "grad_norm": 0.8368885987231633,
      "learning_rate": 4.2092965063295797e-07,
      "loss": 0.3658,
      "step": 13927
    },
    {
      "epoch": 0.8724220548395685,
      "grad_norm": 0.9056519817043908,
      "learning_rate": 4.2052235980808266e-07,
      "loss": 0.4032,
      "step": 13928
    },
    {
      "epoch": 0.8724846928389107,
      "grad_norm": 0.8858576369605243,
      "learning_rate": 4.201152574753581e-07,
      "loss": 0.3492,
      "step": 13929
    },
    {
      "epoch": 0.872547330838253,
      "grad_norm": 0.8793644472199093,
      "learning_rate": 4.1970834365153924e-07,
      "loss": 0.3984,
      "step": 13930
    },
    {
      "epoch": 0.8726099688375953,
      "grad_norm": 0.6188823401754527,
      "learning_rate": 4.193016183533749e-07,
      "loss": 0.4522,
      "step": 13931
    },
    {
      "epoch": 0.8726726068369376,
      "grad_norm": 0.901229290419351,
      "learning_rate": 4.1889508159760617e-07,
      "loss": 0.38,
      "step": 13932
    },
    {
      "epoch": 0.87273524483628,
      "grad_norm": 0.8923963908740498,
      "learning_rate": 4.184887334009663e-07,
      "loss": 0.4148,
      "step": 13933
    },
    {
      "epoch": 0.8727978828356222,
      "grad_norm": 0.8703847536349492,
      "learning_rate": 4.180825737801808e-07,
      "loss": 0.3247,
      "step": 13934
    },
    {
      "epoch": 0.8728605208349646,
      "grad_norm": 0.9079613107406762,
      "learning_rate": 4.1767660275196576e-07,
      "loss": 0.3752,
      "step": 13935
    },
    {
      "epoch": 0.8729231588343068,
      "grad_norm": 0.5892297969567482,
      "learning_rate": 4.1727082033303333e-07,
      "loss": 0.4544,
      "step": 13936
    },
    {
      "epoch": 0.8729857968336492,
      "grad_norm": 0.8596083742034788,
      "learning_rate": 4.1686522654008355e-07,
      "loss": 0.3854,
      "step": 13937
    },
    {
      "epoch": 0.8730484348329914,
      "grad_norm": 0.8437883271207367,
      "learning_rate": 4.16459821389813e-07,
      "loss": 0.3872,
      "step": 13938
    },
    {
      "epoch": 0.8731110728323337,
      "grad_norm": 0.8736948253583842,
      "learning_rate": 4.160546048989067e-07,
      "loss": 0.3923,
      "step": 13939
    },
    {
      "epoch": 0.873173710831676,
      "grad_norm": 0.8440985654996446,
      "learning_rate": 4.156495770840435e-07,
      "loss": 0.3671,
      "step": 13940
    },
    {
      "epoch": 0.8732363488310183,
      "grad_norm": 0.5662930023977545,
      "learning_rate": 4.1524473796189457e-07,
      "loss": 0.4396,
      "step": 13941
    },
    {
      "epoch": 0.8732989868303607,
      "grad_norm": 0.9139948407817744,
      "learning_rate": 4.148400875491232e-07,
      "loss": 0.3872,
      "step": 13942
    },
    {
      "epoch": 0.8733616248297029,
      "grad_norm": 0.9214811995581579,
      "learning_rate": 4.1443562586238495e-07,
      "loss": 0.433,
      "step": 13943
    },
    {
      "epoch": 0.8734242628290453,
      "grad_norm": 0.8490096892344934,
      "learning_rate": 4.1403135291832707e-07,
      "loss": 0.4061,
      "step": 13944
    },
    {
      "epoch": 0.8734869008283875,
      "grad_norm": 0.8834315296705941,
      "learning_rate": 4.1362726873358905e-07,
      "loss": 0.3755,
      "step": 13945
    },
    {
      "epoch": 0.8735495388277298,
      "grad_norm": 0.8131168681414157,
      "learning_rate": 4.132233733248053e-07,
      "loss": 0.3788,
      "step": 13946
    },
    {
      "epoch": 0.8736121768270722,
      "grad_norm": 0.9172422705734891,
      "learning_rate": 4.128196667085987e-07,
      "loss": 0.3745,
      "step": 13947
    },
    {
      "epoch": 0.8736748148264144,
      "grad_norm": 0.8943969574150749,
      "learning_rate": 4.1241614890158657e-07,
      "loss": 0.4238,
      "step": 13948
    },
    {
      "epoch": 0.8737374528257568,
      "grad_norm": 0.8657438070541444,
      "learning_rate": 4.120128199203766e-07,
      "loss": 0.3999,
      "step": 13949
    },
    {
      "epoch": 0.873800090825099,
      "grad_norm": 0.8262011971237845,
      "learning_rate": 4.116096797815705e-07,
      "loss": 0.3501,
      "step": 13950
    },
    {
      "epoch": 0.8738627288244414,
      "grad_norm": 0.8866630062525608,
      "learning_rate": 4.112067285017618e-07,
      "loss": 0.3735,
      "step": 13951
    },
    {
      "epoch": 0.8739253668237836,
      "grad_norm": 0.8592342752404211,
      "learning_rate": 4.1080396609753493e-07,
      "loss": 0.3969,
      "step": 13952
    },
    {
      "epoch": 0.873988004823126,
      "grad_norm": 0.8168305209139017,
      "learning_rate": 4.1040139258546996e-07,
      "loss": 0.3954,
      "step": 13953
    },
    {
      "epoch": 0.8740506428224682,
      "grad_norm": 0.8626309616723855,
      "learning_rate": 4.099990079821342e-07,
      "loss": 0.4085,
      "step": 13954
    },
    {
      "epoch": 0.8741132808218105,
      "grad_norm": 0.937247266378566,
      "learning_rate": 4.0959681230409223e-07,
      "loss": 0.399,
      "step": 13955
    },
    {
      "epoch": 0.8741759188211529,
      "grad_norm": 0.9022972887479958,
      "learning_rate": 4.0919480556789736e-07,
      "loss": 0.3645,
      "step": 13956
    },
    {
      "epoch": 0.8742385568204951,
      "grad_norm": 0.8243971427612125,
      "learning_rate": 4.087929877900965e-07,
      "loss": 0.3581,
      "step": 13957
    },
    {
      "epoch": 0.8743011948198375,
      "grad_norm": 0.8953102055632833,
      "learning_rate": 4.0839135898722846e-07,
      "loss": 0.3724,
      "step": 13958
    },
    {
      "epoch": 0.8743638328191797,
      "grad_norm": 0.8953504528062478,
      "learning_rate": 4.079899191758241e-07,
      "loss": 0.3875,
      "step": 13959
    },
    {
      "epoch": 0.8744264708185221,
      "grad_norm": 0.9024800668432699,
      "learning_rate": 4.075886683724073e-07,
      "loss": 0.3656,
      "step": 13960
    },
    {
      "epoch": 0.8744891088178643,
      "grad_norm": 0.8411056267042303,
      "learning_rate": 4.071876065934921e-07,
      "loss": 0.3814,
      "step": 13961
    },
    {
      "epoch": 0.8745517468172067,
      "grad_norm": 0.86152720361257,
      "learning_rate": 4.067867338555875e-07,
      "loss": 0.3852,
      "step": 13962
    },
    {
      "epoch": 0.874614384816549,
      "grad_norm": 0.9116374427866112,
      "learning_rate": 4.0638605017519483e-07,
      "loss": 0.4089,
      "step": 13963
    },
    {
      "epoch": 0.8746770228158912,
      "grad_norm": 0.8519244711766205,
      "learning_rate": 4.059855555688047e-07,
      "loss": 0.3867,
      "step": 13964
    },
    {
      "epoch": 0.8747396608152336,
      "grad_norm": 0.8725372632363664,
      "learning_rate": 4.055852500529017e-07,
      "loss": 0.4286,
      "step": 13965
    },
    {
      "epoch": 0.8748022988145758,
      "grad_norm": 0.9299708417124968,
      "learning_rate": 4.0518513364396274e-07,
      "loss": 0.3893,
      "step": 13966
    },
    {
      "epoch": 0.8748649368139182,
      "grad_norm": 0.6203309127695815,
      "learning_rate": 4.047852063584562e-07,
      "loss": 0.4269,
      "step": 13967
    },
    {
      "epoch": 0.8749275748132604,
      "grad_norm": 0.8747718339064404,
      "learning_rate": 4.043854682128434e-07,
      "loss": 0.3758,
      "step": 13968
    },
    {
      "epoch": 0.8749902128126028,
      "grad_norm": 0.7909883702144181,
      "learning_rate": 4.0398591922357787e-07,
      "loss": 0.3591,
      "step": 13969
    },
    {
      "epoch": 0.8750528508119451,
      "grad_norm": 0.93488494835849,
      "learning_rate": 4.0358655940710425e-07,
      "loss": 0.3895,
      "step": 13970
    },
    {
      "epoch": 0.8751154888112873,
      "grad_norm": 0.8831675389634042,
      "learning_rate": 4.0318738877986096e-07,
      "loss": 0.3676,
      "step": 13971
    },
    {
      "epoch": 0.8751781268106297,
      "grad_norm": 0.7978160658273911,
      "learning_rate": 4.0278840735827886e-07,
      "loss": 0.3663,
      "step": 13972
    },
    {
      "epoch": 0.8752407648099719,
      "grad_norm": 0.8464567616086645,
      "learning_rate": 4.0238961515877916e-07,
      "loss": 0.3776,
      "step": 13973
    },
    {
      "epoch": 0.8753034028093143,
      "grad_norm": 0.8442449807009321,
      "learning_rate": 4.0199101219777603e-07,
      "loss": 0.383,
      "step": 13974
    },
    {
      "epoch": 0.8753660408086565,
      "grad_norm": 0.5858962125536575,
      "learning_rate": 4.015925984916769e-07,
      "loss": 0.4509,
      "step": 13975
    },
    {
      "epoch": 0.8754286788079989,
      "grad_norm": 0.8782590172054058,
      "learning_rate": 4.011943740568791e-07,
      "loss": 0.3896,
      "step": 13976
    },
    {
      "epoch": 0.8754913168073412,
      "grad_norm": 0.8770972773781855,
      "learning_rate": 4.007963389097752e-07,
      "loss": 0.3611,
      "step": 13977
    },
    {
      "epoch": 0.8755539548066835,
      "grad_norm": 0.663842646635357,
      "learning_rate": 4.0039849306674593e-07,
      "loss": 0.4435,
      "step": 13978
    },
    {
      "epoch": 0.8756165928060258,
      "grad_norm": 0.9707687812934754,
      "learning_rate": 4.0000083654416986e-07,
      "loss": 0.4263,
      "step": 13979
    },
    {
      "epoch": 0.875679230805368,
      "grad_norm": 0.6626638693457534,
      "learning_rate": 3.9960336935841336e-07,
      "loss": 0.4721,
      "step": 13980
    },
    {
      "epoch": 0.8757418688047104,
      "grad_norm": 0.6531425008480715,
      "learning_rate": 3.9920609152583445e-07,
      "loss": 0.4748,
      "step": 13981
    },
    {
      "epoch": 0.8758045068040526,
      "grad_norm": 0.8503861230772385,
      "learning_rate": 3.988090030627884e-07,
      "loss": 0.3708,
      "step": 13982
    },
    {
      "epoch": 0.875867144803395,
      "grad_norm": 0.8601808784570271,
      "learning_rate": 3.984121039856176e-07,
      "loss": 0.3891,
      "step": 13983
    },
    {
      "epoch": 0.8759297828027373,
      "grad_norm": 0.8105002372155308,
      "learning_rate": 3.9801539431065847e-07,
      "loss": 0.3675,
      "step": 13984
    },
    {
      "epoch": 0.8759924208020796,
      "grad_norm": 0.8796916883826137,
      "learning_rate": 3.9761887405423973e-07,
      "loss": 0.3716,
      "step": 13985
    },
    {
      "epoch": 0.8760550588014219,
      "grad_norm": 0.5920412668180547,
      "learning_rate": 3.972225432326826e-07,
      "loss": 0.4698,
      "step": 13986
    },
    {
      "epoch": 0.8761176968007642,
      "grad_norm": 0.8514595727323399,
      "learning_rate": 3.9682640186229914e-07,
      "loss": 0.3809,
      "step": 13987
    },
    {
      "epoch": 0.8761803348001065,
      "grad_norm": 0.8191068016837589,
      "learning_rate": 3.9643044995939626e-07,
      "loss": 0.3614,
      "step": 13988
    },
    {
      "epoch": 0.8762429727994487,
      "grad_norm": 0.8568856188588921,
      "learning_rate": 3.960346875402704e-07,
      "loss": 0.4032,
      "step": 13989
    },
    {
      "epoch": 0.8763056107987911,
      "grad_norm": 0.9177572504083826,
      "learning_rate": 3.956391146212102e-07,
      "loss": 0.3785,
      "step": 13990
    },
    {
      "epoch": 0.8763682487981334,
      "grad_norm": 0.8393966374061629,
      "learning_rate": 3.952437312184998e-07,
      "loss": 0.4063,
      "step": 13991
    },
    {
      "epoch": 0.8764308867974757,
      "grad_norm": 0.8957679061940997,
      "learning_rate": 3.9484853734841234e-07,
      "loss": 0.3832,
      "step": 13992
    },
    {
      "epoch": 0.876493524796818,
      "grad_norm": 0.8384232757620583,
      "learning_rate": 3.944535330272131e-07,
      "loss": 0.3372,
      "step": 13993
    },
    {
      "epoch": 0.8765561627961603,
      "grad_norm": 0.8957113675791107,
      "learning_rate": 3.9405871827116196e-07,
      "loss": 0.3593,
      "step": 13994
    },
    {
      "epoch": 0.8766188007955026,
      "grad_norm": 0.8818781757911657,
      "learning_rate": 3.9366409309650857e-07,
      "loss": 0.393,
      "step": 13995
    },
    {
      "epoch": 0.876681438794845,
      "grad_norm": 0.9274266921043032,
      "learning_rate": 3.932696575194955e-07,
      "loss": 0.3942,
      "step": 13996
    },
    {
      "epoch": 0.8767440767941872,
      "grad_norm": 0.7862644920822456,
      "learning_rate": 3.9287541155635875e-07,
      "loss": 0.3743,
      "step": 13997
    },
    {
      "epoch": 0.8768067147935295,
      "grad_norm": 0.9231606640591249,
      "learning_rate": 3.9248135522332587e-07,
      "loss": 0.3907,
      "step": 13998
    },
    {
      "epoch": 0.8768693527928718,
      "grad_norm": 0.8808602474700556,
      "learning_rate": 3.920874885366144e-07,
      "loss": 0.4056,
      "step": 13999
    },
    {
      "epoch": 0.8769319907922141,
      "grad_norm": 0.8659827858195399,
      "learning_rate": 3.9169381151243803e-07,
      "loss": 0.3927,
      "step": 14000
    },
    {
      "epoch": 0.8769946287915564,
      "grad_norm": 0.8687907357353045,
      "learning_rate": 3.9130032416699935e-07,
      "loss": 0.3865,
      "step": 14001
    },
    {
      "epoch": 0.8770572667908987,
      "grad_norm": 0.9505375962799076,
      "learning_rate": 3.909070265164955e-07,
      "loss": 0.4055,
      "step": 14002
    },
    {
      "epoch": 0.877119904790241,
      "grad_norm": 0.9151438818580124,
      "learning_rate": 3.905139185771134e-07,
      "loss": 0.362,
      "step": 14003
    },
    {
      "epoch": 0.8771825427895833,
      "grad_norm": 0.9225728521590839,
      "learning_rate": 3.9012100036503295e-07,
      "loss": 0.3852,
      "step": 14004
    },
    {
      "epoch": 0.8772451807889255,
      "grad_norm": 0.8325311895509329,
      "learning_rate": 3.89728271896429e-07,
      "loss": 0.3549,
      "step": 14005
    },
    {
      "epoch": 0.8773078187882679,
      "grad_norm": 0.8761844960970022,
      "learning_rate": 3.8933573318746475e-07,
      "loss": 0.3826,
      "step": 14006
    },
    {
      "epoch": 0.8773704567876102,
      "grad_norm": 0.8572863028429251,
      "learning_rate": 3.889433842542978e-07,
      "loss": 0.3786,
      "step": 14007
    },
    {
      "epoch": 0.8774330947869525,
      "grad_norm": 0.8942567198737035,
      "learning_rate": 3.885512251130763e-07,
      "loss": 0.3745,
      "step": 14008
    },
    {
      "epoch": 0.8774957327862948,
      "grad_norm": 0.8602726977422481,
      "learning_rate": 3.8815925577994296e-07,
      "loss": 0.3854,
      "step": 14009
    },
    {
      "epoch": 0.8775583707856371,
      "grad_norm": 0.8397052955880704,
      "learning_rate": 3.877674762710304e-07,
      "loss": 0.3603,
      "step": 14010
    },
    {
      "epoch": 0.8776210087849794,
      "grad_norm": 0.8740749811325975,
      "learning_rate": 3.873758866024652e-07,
      "loss": 0.3942,
      "step": 14011
    },
    {
      "epoch": 0.8776836467843218,
      "grad_norm": 0.9524843033579785,
      "learning_rate": 3.869844867903638e-07,
      "loss": 0.4259,
      "step": 14012
    },
    {
      "epoch": 0.877746284783664,
      "grad_norm": 0.5845218943909671,
      "learning_rate": 3.865932768508368e-07,
      "loss": 0.4271,
      "step": 14013
    },
    {
      "epoch": 0.8778089227830063,
      "grad_norm": 0.7981546253108208,
      "learning_rate": 3.862022567999879e-07,
      "loss": 0.3729,
      "step": 14014
    },
    {
      "epoch": 0.8778715607823486,
      "grad_norm": 0.8360023681106464,
      "learning_rate": 3.8581142665391036e-07,
      "loss": 0.3699,
      "step": 14015
    },
    {
      "epoch": 0.8779341987816909,
      "grad_norm": 0.7828346663051295,
      "learning_rate": 3.8542078642869075e-07,
      "loss": 0.3666,
      "step": 14016
    },
    {
      "epoch": 0.8779968367810332,
      "grad_norm": 0.8655193539966062,
      "learning_rate": 3.850303361404067e-07,
      "loss": 0.3689,
      "step": 14017
    },
    {
      "epoch": 0.8780594747803755,
      "grad_norm": 0.6518644144966935,
      "learning_rate": 3.846400758051322e-07,
      "loss": 0.4732,
      "step": 14018
    },
    {
      "epoch": 0.8781221127797179,
      "grad_norm": 0.856924094572329,
      "learning_rate": 3.8425000543892864e-07,
      "loss": 0.3882,
      "step": 14019
    },
    {
      "epoch": 0.8781847507790601,
      "grad_norm": 0.5984445183652974,
      "learning_rate": 3.8386012505785164e-07,
      "loss": 0.4558,
      "step": 14020
    },
    {
      "epoch": 0.8782473887784025,
      "grad_norm": 0.8654805500881209,
      "learning_rate": 3.8347043467794777e-07,
      "loss": 0.414,
      "step": 14021
    },
    {
      "epoch": 0.8783100267777447,
      "grad_norm": 0.9750271354594201,
      "learning_rate": 3.8308093431525863e-07,
      "loss": 0.3662,
      "step": 14022
    },
    {
      "epoch": 0.878372664777087,
      "grad_norm": 0.8297080128228211,
      "learning_rate": 3.826916239858147e-07,
      "loss": 0.3386,
      "step": 14023
    },
    {
      "epoch": 0.8784353027764293,
      "grad_norm": 0.8450030830167623,
      "learning_rate": 3.82302503705641e-07,
      "loss": 0.3613,
      "step": 14024
    },
    {
      "epoch": 0.8784979407757716,
      "grad_norm": 0.8972798699061757,
      "learning_rate": 3.819135734907531e-07,
      "loss": 0.4001,
      "step": 14025
    },
    {
      "epoch": 0.878560578775114,
      "grad_norm": 0.8900471339767992,
      "learning_rate": 3.8152483335715975e-07,
      "loss": 0.3809,
      "step": 14026
    },
    {
      "epoch": 0.8786232167744562,
      "grad_norm": 0.5899951275435747,
      "learning_rate": 3.8113628332086095e-07,
      "loss": 0.4238,
      "step": 14027
    },
    {
      "epoch": 0.8786858547737986,
      "grad_norm": 0.8358224708022404,
      "learning_rate": 3.807479233978506e-07,
      "loss": 0.4234,
      "step": 14028
    },
    {
      "epoch": 0.8787484927731408,
      "grad_norm": 0.901965451966369,
      "learning_rate": 3.8035975360411317e-07,
      "loss": 0.3912,
      "step": 14029
    },
    {
      "epoch": 0.8788111307724831,
      "grad_norm": 0.8673685878857976,
      "learning_rate": 3.799717739556247e-07,
      "loss": 0.4124,
      "step": 14030
    },
    {
      "epoch": 0.8788737687718254,
      "grad_norm": 0.8866022785116802,
      "learning_rate": 3.795839844683563e-07,
      "loss": 0.3278,
      "step": 14031
    },
    {
      "epoch": 0.8789364067711677,
      "grad_norm": 0.6467003243293497,
      "learning_rate": 3.791963851582692e-07,
      "loss": 0.4458,
      "step": 14032
    },
    {
      "epoch": 0.87899904477051,
      "grad_norm": 0.9262926256422755,
      "learning_rate": 3.788089760413161e-07,
      "loss": 0.3915,
      "step": 14033
    },
    {
      "epoch": 0.8790616827698523,
      "grad_norm": 0.9373236466417613,
      "learning_rate": 3.784217571334436e-07,
      "loss": 0.3673,
      "step": 14034
    },
    {
      "epoch": 0.8791243207691947,
      "grad_norm": 0.8316813353975687,
      "learning_rate": 3.780347284505892e-07,
      "loss": 0.3865,
      "step": 14035
    },
    {
      "epoch": 0.8791869587685369,
      "grad_norm": 0.9229254150070696,
      "learning_rate": 3.7764789000868216e-07,
      "loss": 0.4069,
      "step": 14036
    },
    {
      "epoch": 0.8792495967678793,
      "grad_norm": 0.8920173027233362,
      "learning_rate": 3.772612418236471e-07,
      "loss": 0.4138,
      "step": 14037
    },
    {
      "epoch": 0.8793122347672215,
      "grad_norm": 0.9383583481790605,
      "learning_rate": 3.7687478391139734e-07,
      "loss": 0.3878,
      "step": 14038
    },
    {
      "epoch": 0.8793748727665638,
      "grad_norm": 0.916573895500343,
      "learning_rate": 3.7648851628783844e-07,
      "loss": 0.3824,
      "step": 14039
    },
    {
      "epoch": 0.8794375107659061,
      "grad_norm": 0.8637816172082884,
      "learning_rate": 3.7610243896887164e-07,
      "loss": 0.3804,
      "step": 14040
    },
    {
      "epoch": 0.8795001487652484,
      "grad_norm": 0.9185018656814263,
      "learning_rate": 3.757165519703865e-07,
      "loss": 0.3667,
      "step": 14041
    },
    {
      "epoch": 0.8795627867645908,
      "grad_norm": 0.8643257345552875,
      "learning_rate": 3.753308553082663e-07,
      "loss": 0.3916,
      "step": 14042
    },
    {
      "epoch": 0.879625424763933,
      "grad_norm": 0.8711543083968465,
      "learning_rate": 3.7494534899838674e-07,
      "loss": 0.3676,
      "step": 14043
    },
    {
      "epoch": 0.8796880627632754,
      "grad_norm": 0.8672265883413147,
      "learning_rate": 3.7456003305661457e-07,
      "loss": 0.3756,
      "step": 14044
    },
    {
      "epoch": 0.8797507007626176,
      "grad_norm": 0.8869002116073331,
      "learning_rate": 3.7417490749880937e-07,
      "loss": 0.318,
      "step": 14045
    },
    {
      "epoch": 0.87981333876196,
      "grad_norm": 0.8499055453855936,
      "learning_rate": 3.737899723408245e-07,
      "loss": 0.3388,
      "step": 14046
    },
    {
      "epoch": 0.8798759767613022,
      "grad_norm": 0.8611924323968503,
      "learning_rate": 3.7340522759850226e-07,
      "loss": 0.3465,
      "step": 14047
    },
    {
      "epoch": 0.8799386147606445,
      "grad_norm": 0.8693711241509673,
      "learning_rate": 3.730206732876801e-07,
      "loss": 0.3687,
      "step": 14048
    },
    {
      "epoch": 0.8800012527599869,
      "grad_norm": 0.8702292196325648,
      "learning_rate": 3.7263630942418636e-07,
      "loss": 0.363,
      "step": 14049
    },
    {
      "epoch": 0.8800638907593291,
      "grad_norm": 0.8816062489964316,
      "learning_rate": 3.722521360238407e-07,
      "loss": 0.3589,
      "step": 14050
    },
    {
      "epoch": 0.8801265287586715,
      "grad_norm": 0.8841859784624069,
      "learning_rate": 3.718681531024559e-07,
      "loss": 0.391,
      "step": 14051
    },
    {
      "epoch": 0.8801891667580137,
      "grad_norm": 0.9129761228032405,
      "learning_rate": 3.7148436067583724e-07,
      "loss": 0.4,
      "step": 14052
    },
    {
      "epoch": 0.8802518047573561,
      "grad_norm": 0.8482373584734912,
      "learning_rate": 3.711007587597809e-07,
      "loss": 0.3888,
      "step": 14053
    },
    {
      "epoch": 0.8803144427566983,
      "grad_norm": 0.8743831320818731,
      "learning_rate": 3.7071734737007593e-07,
      "loss": 0.3456,
      "step": 14054
    },
    {
      "epoch": 0.8803770807560406,
      "grad_norm": 0.8982528806505776,
      "learning_rate": 3.7033412652250524e-07,
      "loss": 0.3714,
      "step": 14055
    },
    {
      "epoch": 0.880439718755383,
      "grad_norm": 0.8920676945708844,
      "learning_rate": 3.699510962328401e-07,
      "loss": 0.3867,
      "step": 14056
    },
    {
      "epoch": 0.8805023567547252,
      "grad_norm": 0.8446523518959315,
      "learning_rate": 3.6956825651684846e-07,
      "loss": 0.3922,
      "step": 14057
    },
    {
      "epoch": 0.8805649947540676,
      "grad_norm": 0.88424753801567,
      "learning_rate": 3.691856073902866e-07,
      "loss": 0.3913,
      "step": 14058
    },
    {
      "epoch": 0.8806276327534098,
      "grad_norm": 0.7617755219784613,
      "learning_rate": 3.6880314886890424e-07,
      "loss": 0.3428,
      "step": 14059
    },
    {
      "epoch": 0.8806902707527522,
      "grad_norm": 0.9219559189729135,
      "learning_rate": 3.684208809684442e-07,
      "loss": 0.4107,
      "step": 14060
    },
    {
      "epoch": 0.8807529087520944,
      "grad_norm": 0.8110942778954472,
      "learning_rate": 3.680388037046406e-07,
      "loss": 0.3415,
      "step": 14061
    },
    {
      "epoch": 0.8808155467514368,
      "grad_norm": 0.8496581343189424,
      "learning_rate": 3.6765691709321924e-07,
      "loss": 0.3417,
      "step": 14062
    },
    {
      "epoch": 0.8808781847507791,
      "grad_norm": 0.8743632628999576,
      "learning_rate": 3.672752211498981e-07,
      "loss": 0.3785,
      "step": 14063
    },
    {
      "epoch": 0.8809408227501213,
      "grad_norm": 0.853056364552332,
      "learning_rate": 3.6689371589039013e-07,
      "loss": 0.4037,
      "step": 14064
    },
    {
      "epoch": 0.8810034607494637,
      "grad_norm": 0.8873356277306939,
      "learning_rate": 3.6651240133039557e-07,
      "loss": 0.377,
      "step": 14065
    },
    {
      "epoch": 0.8810660987488059,
      "grad_norm": 0.8226704308556567,
      "learning_rate": 3.661312774856113e-07,
      "loss": 0.4159,
      "step": 14066
    },
    {
      "epoch": 0.8811287367481483,
      "grad_norm": 0.8807792092548994,
      "learning_rate": 3.657503443717236e-07,
      "loss": 0.4255,
      "step": 14067
    },
    {
      "epoch": 0.8811913747474905,
      "grad_norm": 0.7892046076307677,
      "learning_rate": 3.653696020044123e-07,
      "loss": 0.3687,
      "step": 14068
    },
    {
      "epoch": 0.8812540127468329,
      "grad_norm": 0.9489188309208688,
      "learning_rate": 3.6498905039934807e-07,
      "loss": 0.3895,
      "step": 14069
    },
    {
      "epoch": 0.8813166507461752,
      "grad_norm": 0.8624131752685421,
      "learning_rate": 3.646086895721951e-07,
      "loss": 0.385,
      "step": 14070
    },
    {
      "epoch": 0.8813792887455175,
      "grad_norm": 0.837574044055321,
      "learning_rate": 3.642285195386086e-07,
      "loss": 0.3928,
      "step": 14071
    },
    {
      "epoch": 0.8814419267448598,
      "grad_norm": 0.9171089083657346,
      "learning_rate": 3.6384854031423553e-07,
      "loss": 0.3851,
      "step": 14072
    },
    {
      "epoch": 0.881504564744202,
      "grad_norm": 0.872999612305069,
      "learning_rate": 3.6346875191471785e-07,
      "loss": 0.405,
      "step": 14073
    },
    {
      "epoch": 0.8815672027435444,
      "grad_norm": 0.8560025676106903,
      "learning_rate": 3.6308915435568746e-07,
      "loss": 0.3659,
      "step": 14074
    },
    {
      "epoch": 0.8816298407428866,
      "grad_norm": 0.8144684885418181,
      "learning_rate": 3.62709747652768e-07,
      "loss": 0.3539,
      "step": 14075
    },
    {
      "epoch": 0.881692478742229,
      "grad_norm": 0.8101483322922356,
      "learning_rate": 3.6233053182157586e-07,
      "loss": 0.3535,
      "step": 14076
    },
    {
      "epoch": 0.8817551167415713,
      "grad_norm": 0.7938247172348455,
      "learning_rate": 3.619515068777202e-07,
      "loss": 0.3712,
      "step": 14077
    },
    {
      "epoch": 0.8818177547409136,
      "grad_norm": 0.8890379648984326,
      "learning_rate": 3.615726728368013e-07,
      "loss": 0.3845,
      "step": 14078
    },
    {
      "epoch": 0.8818803927402559,
      "grad_norm": 0.8904373137281539,
      "learning_rate": 3.6119402971441175e-07,
      "loss": 0.3725,
      "step": 14079
    },
    {
      "epoch": 0.8819430307395982,
      "grad_norm": 0.8743556084524559,
      "learning_rate": 3.608155775261374e-07,
      "loss": 0.3717,
      "step": 14080
    },
    {
      "epoch": 0.8820056687389405,
      "grad_norm": 0.823137991681904,
      "learning_rate": 3.6043731628755405e-07,
      "loss": 0.3501,
      "step": 14081
    },
    {
      "epoch": 0.8820683067382827,
      "grad_norm": 0.5823822220330094,
      "learning_rate": 3.600592460142316e-07,
      "loss": 0.4493,
      "step": 14082
    },
    {
      "epoch": 0.8821309447376251,
      "grad_norm": 0.5993733911583382,
      "learning_rate": 3.5968136672173304e-07,
      "loss": 0.4598,
      "step": 14083
    },
    {
      "epoch": 0.8821935827369674,
      "grad_norm": 0.933344157969982,
      "learning_rate": 3.5930367842561044e-07,
      "loss": 0.4586,
      "step": 14084
    },
    {
      "epoch": 0.8822562207363097,
      "grad_norm": 0.9120339359347482,
      "learning_rate": 3.589261811414096e-07,
      "loss": 0.3899,
      "step": 14085
    },
    {
      "epoch": 0.882318858735652,
      "grad_norm": 0.8700332734489158,
      "learning_rate": 3.5854887488466883e-07,
      "loss": 0.3932,
      "step": 14086
    },
    {
      "epoch": 0.8823814967349943,
      "grad_norm": 0.8431711204302517,
      "learning_rate": 3.5817175967091776e-07,
      "loss": 0.3283,
      "step": 14087
    },
    {
      "epoch": 0.8824441347343366,
      "grad_norm": 0.8150247395336991,
      "learning_rate": 3.5779483551567905e-07,
      "loss": 0.3672,
      "step": 14088
    },
    {
      "epoch": 0.8825067727336788,
      "grad_norm": 0.9323135742470443,
      "learning_rate": 3.574181024344653e-07,
      "loss": 0.3989,
      "step": 14089
    },
    {
      "epoch": 0.8825694107330212,
      "grad_norm": 0.874515238365937,
      "learning_rate": 3.5704156044278517e-07,
      "loss": 0.4025,
      "step": 14090
    },
    {
      "epoch": 0.8826320487323635,
      "grad_norm": 0.8299726112452122,
      "learning_rate": 3.5666520955613517e-07,
      "loss": 0.3342,
      "step": 14091
    },
    {
      "epoch": 0.8826946867317058,
      "grad_norm": 0.8338126150329953,
      "learning_rate": 3.562890497900079e-07,
      "loss": 0.3617,
      "step": 14092
    },
    {
      "epoch": 0.8827573247310481,
      "grad_norm": 0.8397092308966333,
      "learning_rate": 3.559130811598849e-07,
      "loss": 0.334,
      "step": 14093
    },
    {
      "epoch": 0.8828199627303904,
      "grad_norm": 0.8753685149400108,
      "learning_rate": 3.5553730368124205e-07,
      "loss": 0.3712,
      "step": 14094
    },
    {
      "epoch": 0.8828826007297327,
      "grad_norm": 0.6040394088585379,
      "learning_rate": 3.5516171736954543e-07,
      "loss": 0.4189,
      "step": 14095
    },
    {
      "epoch": 0.882945238729075,
      "grad_norm": 0.5581709959139709,
      "learning_rate": 3.5478632224025477e-07,
      "loss": 0.4686,
      "step": 14096
    },
    {
      "epoch": 0.8830078767284173,
      "grad_norm": 0.8594430768100829,
      "learning_rate": 3.544111183088217e-07,
      "loss": 0.3867,
      "step": 14097
    },
    {
      "epoch": 0.8830705147277595,
      "grad_norm": 0.8503566481043175,
      "learning_rate": 3.540361055906877e-07,
      "loss": 0.3758,
      "step": 14098
    },
    {
      "epoch": 0.8831331527271019,
      "grad_norm": 0.9364975446432597,
      "learning_rate": 3.5366128410129154e-07,
      "loss": 0.3707,
      "step": 14099
    },
    {
      "epoch": 0.8831957907264442,
      "grad_norm": 0.8837198544144171,
      "learning_rate": 3.5328665385605754e-07,
      "loss": 0.3587,
      "step": 14100
    },
    {
      "epoch": 0.8832584287257865,
      "grad_norm": 0.8146900635395964,
      "learning_rate": 3.529122148704089e-07,
      "loss": 0.3584,
      "step": 14101
    },
    {
      "epoch": 0.8833210667251288,
      "grad_norm": 0.6328927858042217,
      "learning_rate": 3.5253796715975564e-07,
      "loss": 0.4678,
      "step": 14102
    },
    {
      "epoch": 0.8833837047244711,
      "grad_norm": 0.8810356297421909,
      "learning_rate": 3.5216391073950253e-07,
      "loss": 0.4014,
      "step": 14103
    },
    {
      "epoch": 0.8834463427238134,
      "grad_norm": 0.8742729587890181,
      "learning_rate": 3.517900456250456e-07,
      "loss": 0.3792,
      "step": 14104
    },
    {
      "epoch": 0.8835089807231558,
      "grad_norm": 0.880999493755953,
      "learning_rate": 3.5141637183177314e-07,
      "loss": 0.3933,
      "step": 14105
    },
    {
      "epoch": 0.883571618722498,
      "grad_norm": 0.8723673162840717,
      "learning_rate": 3.510428893750645e-07,
      "loss": 0.3953,
      "step": 14106
    },
    {
      "epoch": 0.8836342567218403,
      "grad_norm": 0.9081683204522124,
      "learning_rate": 3.5066959827029454e-07,
      "loss": 0.3773,
      "step": 14107
    },
    {
      "epoch": 0.8836968947211826,
      "grad_norm": 0.7595608421678699,
      "learning_rate": 3.5029649853282667e-07,
      "loss": 0.322,
      "step": 14108
    },
    {
      "epoch": 0.8837595327205249,
      "grad_norm": 0.8228244424157228,
      "learning_rate": 3.4992359017801737e-07,
      "loss": 0.3903,
      "step": 14109
    },
    {
      "epoch": 0.8838221707198672,
      "grad_norm": 0.9054445471221643,
      "learning_rate": 3.495508732212166e-07,
      "loss": 0.3966,
      "step": 14110
    },
    {
      "epoch": 0.8838848087192095,
      "grad_norm": 0.9392096654073443,
      "learning_rate": 3.491783476777655e-07,
      "loss": 0.4065,
      "step": 14111
    },
    {
      "epoch": 0.8839474467185519,
      "grad_norm": 0.8461453209344013,
      "learning_rate": 3.488060135629967e-07,
      "loss": 0.3303,
      "step": 14112
    },
    {
      "epoch": 0.8840100847178941,
      "grad_norm": 0.8273006003327861,
      "learning_rate": 3.4843387089223524e-07,
      "loss": 0.3678,
      "step": 14113
    },
    {
      "epoch": 0.8840727227172364,
      "grad_norm": 0.8851312032299019,
      "learning_rate": 3.480619196807994e-07,
      "loss": 0.3743,
      "step": 14114
    },
    {
      "epoch": 0.8841353607165787,
      "grad_norm": 0.8621146924212044,
      "learning_rate": 3.47690159943998e-07,
      "loss": 0.3658,
      "step": 14115
    },
    {
      "epoch": 0.884197998715921,
      "grad_norm": 0.8457316012437817,
      "learning_rate": 3.4731859169713335e-07,
      "loss": 0.3595,
      "step": 14116
    },
    {
      "epoch": 0.8842606367152633,
      "grad_norm": 0.8643759071263697,
      "learning_rate": 3.469472149554992e-07,
      "loss": 0.35,
      "step": 14117
    },
    {
      "epoch": 0.8843232747146056,
      "grad_norm": 0.5634378036707722,
      "learning_rate": 3.465760297343812e-07,
      "loss": 0.4421,
      "step": 14118
    },
    {
      "epoch": 0.884385912713948,
      "grad_norm": 0.845646075873081,
      "learning_rate": 3.462050360490571e-07,
      "loss": 0.3602,
      "step": 14119
    },
    {
      "epoch": 0.8844485507132902,
      "grad_norm": 0.9337562296802016,
      "learning_rate": 3.45834233914798e-07,
      "loss": 0.3711,
      "step": 14120
    },
    {
      "epoch": 0.8845111887126326,
      "grad_norm": 0.8477467591396314,
      "learning_rate": 3.454636233468656e-07,
      "loss": 0.3568,
      "step": 14121
    },
    {
      "epoch": 0.8845738267119748,
      "grad_norm": 0.8218167601932878,
      "learning_rate": 3.4509320436051486e-07,
      "loss": 0.3867,
      "step": 14122
    },
    {
      "epoch": 0.8846364647113171,
      "grad_norm": 0.8968271728038693,
      "learning_rate": 3.4472297697099143e-07,
      "loss": 0.3817,
      "step": 14123
    },
    {
      "epoch": 0.8846991027106594,
      "grad_norm": 0.8996502077854004,
      "learning_rate": 3.443529411935331e-07,
      "loss": 0.379,
      "step": 14124
    },
    {
      "epoch": 0.8847617407100017,
      "grad_norm": 0.866806582896012,
      "learning_rate": 3.4398309704337327e-07,
      "loss": 0.4121,
      "step": 14125
    },
    {
      "epoch": 0.884824378709344,
      "grad_norm": 0.9321864450297336,
      "learning_rate": 3.4361344453573367e-07,
      "loss": 0.4131,
      "step": 14126
    },
    {
      "epoch": 0.8848870167086863,
      "grad_norm": 0.8043033437245863,
      "learning_rate": 3.432439836858281e-07,
      "loss": 0.3566,
      "step": 14127
    },
    {
      "epoch": 0.8849496547080287,
      "grad_norm": 0.9006637224311673,
      "learning_rate": 3.42874714508864e-07,
      "loss": 0.4239,
      "step": 14128
    },
    {
      "epoch": 0.8850122927073709,
      "grad_norm": 0.9133230454471212,
      "learning_rate": 3.4250563702004246e-07,
      "loss": 0.3617,
      "step": 14129
    },
    {
      "epoch": 0.8850749307067133,
      "grad_norm": 0.8473170805492286,
      "learning_rate": 3.4213675123455246e-07,
      "loss": 0.3653,
      "step": 14130
    },
    {
      "epoch": 0.8851375687060555,
      "grad_norm": 0.8647850798279353,
      "learning_rate": 3.417680571675791e-07,
      "loss": 0.4197,
      "step": 14131
    },
    {
      "epoch": 0.8852002067053978,
      "grad_norm": 0.8217464676602,
      "learning_rate": 3.413995548342958e-07,
      "loss": 0.3714,
      "step": 14132
    },
    {
      "epoch": 0.8852628447047401,
      "grad_norm": 0.8809711647381091,
      "learning_rate": 3.4103124424987254e-07,
      "loss": 0.3536,
      "step": 14133
    },
    {
      "epoch": 0.8853254827040824,
      "grad_norm": 0.8827847853349143,
      "learning_rate": 3.406631254294679e-07,
      "loss": 0.3796,
      "step": 14134
    },
    {
      "epoch": 0.8853881207034248,
      "grad_norm": 0.7894367457211906,
      "learning_rate": 3.402951983882341e-07,
      "loss": 0.3659,
      "step": 14135
    },
    {
      "epoch": 0.885450758702767,
      "grad_norm": 0.8251006057737569,
      "learning_rate": 3.3992746314131406e-07,
      "loss": 0.3677,
      "step": 14136
    },
    {
      "epoch": 0.8855133967021094,
      "grad_norm": 0.5724255168927116,
      "learning_rate": 3.395599197038446e-07,
      "loss": 0.4286,
      "step": 14137
    },
    {
      "epoch": 0.8855760347014516,
      "grad_norm": 0.9335370595375754,
      "learning_rate": 3.3919256809095414e-07,
      "loss": 0.3886,
      "step": 14138
    },
    {
      "epoch": 0.8856386727007939,
      "grad_norm": 0.8807299720930952,
      "learning_rate": 3.3882540831776334e-07,
      "loss": 0.3919,
      "step": 14139
    },
    {
      "epoch": 0.8857013107001362,
      "grad_norm": 0.9078783482700185,
      "learning_rate": 3.384584403993829e-07,
      "loss": 0.3826,
      "step": 14140
    },
    {
      "epoch": 0.8857639486994785,
      "grad_norm": 0.8497822749481374,
      "learning_rate": 3.380916643509186e-07,
      "loss": 0.3727,
      "step": 14141
    },
    {
      "epoch": 0.8858265866988209,
      "grad_norm": 0.8025360607613096,
      "learning_rate": 3.3772508018746654e-07,
      "loss": 0.3719,
      "step": 14142
    },
    {
      "epoch": 0.8858892246981631,
      "grad_norm": 0.9759372599950686,
      "learning_rate": 3.3735868792411586e-07,
      "loss": 0.4315,
      "step": 14143
    },
    {
      "epoch": 0.8859518626975055,
      "grad_norm": 0.8452484573987421,
      "learning_rate": 3.369924875759473e-07,
      "loss": 0.357,
      "step": 14144
    },
    {
      "epoch": 0.8860145006968477,
      "grad_norm": 0.8358125422950737,
      "learning_rate": 3.3662647915803316e-07,
      "loss": 0.3745,
      "step": 14145
    },
    {
      "epoch": 0.8860771386961901,
      "grad_norm": 0.8891326805850921,
      "learning_rate": 3.362606626854381e-07,
      "loss": 0.378,
      "step": 14146
    },
    {
      "epoch": 0.8861397766955323,
      "grad_norm": 0.8649282077806922,
      "learning_rate": 3.3589503817322013e-07,
      "loss": 0.3667,
      "step": 14147
    },
    {
      "epoch": 0.8862024146948746,
      "grad_norm": 0.8805046215430001,
      "learning_rate": 3.3552960563642825e-07,
      "loss": 0.3544,
      "step": 14148
    },
    {
      "epoch": 0.886265052694217,
      "grad_norm": 0.8927620056005346,
      "learning_rate": 3.3516436509010376e-07,
      "loss": 0.3854,
      "step": 14149
    },
    {
      "epoch": 0.8863276906935592,
      "grad_norm": 0.8670087814445315,
      "learning_rate": 3.347993165492791e-07,
      "loss": 0.3733,
      "step": 14150
    },
    {
      "epoch": 0.8863903286929016,
      "grad_norm": 0.909730038480672,
      "learning_rate": 3.3443446002898117e-07,
      "loss": 0.4095,
      "step": 14151
    },
    {
      "epoch": 0.8864529666922438,
      "grad_norm": 0.8881914015687823,
      "learning_rate": 3.340697955442268e-07,
      "loss": 0.3866,
      "step": 14152
    },
    {
      "epoch": 0.8865156046915862,
      "grad_norm": 0.7965095461813618,
      "learning_rate": 3.337053231100257e-07,
      "loss": 0.3276,
      "step": 14153
    },
    {
      "epoch": 0.8865782426909284,
      "grad_norm": 0.8731046701698273,
      "learning_rate": 3.333410427413797e-07,
      "loss": 0.428,
      "step": 14154
    },
    {
      "epoch": 0.8866408806902708,
      "grad_norm": 0.9898300739279613,
      "learning_rate": 3.329769544532818e-07,
      "loss": 0.4239,
      "step": 14155
    },
    {
      "epoch": 0.8867035186896131,
      "grad_norm": 0.8678281193520022,
      "learning_rate": 3.32613058260719e-07,
      "loss": 0.4046,
      "step": 14156
    },
    {
      "epoch": 0.8867661566889553,
      "grad_norm": 0.8873112320200859,
      "learning_rate": 3.3224935417866975e-07,
      "loss": 0.373,
      "step": 14157
    },
    {
      "epoch": 0.8868287946882977,
      "grad_norm": 0.8729816944464479,
      "learning_rate": 3.318858422221022e-07,
      "loss": 0.4055,
      "step": 14158
    },
    {
      "epoch": 0.8868914326876399,
      "grad_norm": 0.929275924279114,
      "learning_rate": 3.315225224059809e-07,
      "loss": 0.3928,
      "step": 14159
    },
    {
      "epoch": 0.8869540706869823,
      "grad_norm": 0.9248730487723744,
      "learning_rate": 3.311593947452585e-07,
      "loss": 0.4144,
      "step": 14160
    },
    {
      "epoch": 0.8870167086863245,
      "grad_norm": 0.9093388623389898,
      "learning_rate": 3.3079645925488237e-07,
      "loss": 0.4028,
      "step": 14161
    },
    {
      "epoch": 0.8870793466856669,
      "grad_norm": 0.9788734470169854,
      "learning_rate": 3.304337159497911e-07,
      "loss": 0.3845,
      "step": 14162
    },
    {
      "epoch": 0.8871419846850092,
      "grad_norm": 0.8994315783820973,
      "learning_rate": 3.3007116484491397e-07,
      "loss": 0.3602,
      "step": 14163
    },
    {
      "epoch": 0.8872046226843514,
      "grad_norm": 0.8217694884211492,
      "learning_rate": 3.297088059551751e-07,
      "loss": 0.371,
      "step": 14164
    },
    {
      "epoch": 0.8872672606836938,
      "grad_norm": 0.8727464593393304,
      "learning_rate": 3.293466392954875e-07,
      "loss": 0.4185,
      "step": 14165
    },
    {
      "epoch": 0.887329898683036,
      "grad_norm": 0.9468517012075914,
      "learning_rate": 3.2898466488075984e-07,
      "loss": 0.3507,
      "step": 14166
    },
    {
      "epoch": 0.8873925366823784,
      "grad_norm": 0.8874440950162152,
      "learning_rate": 3.286228827258897e-07,
      "loss": 0.3992,
      "step": 14167
    },
    {
      "epoch": 0.8874551746817206,
      "grad_norm": 0.891262905015541,
      "learning_rate": 3.282612928457696e-07,
      "loss": 0.3682,
      "step": 14168
    },
    {
      "epoch": 0.887517812681063,
      "grad_norm": 0.8642864066813948,
      "learning_rate": 3.27899895255282e-07,
      "loss": 0.3781,
      "step": 14169
    },
    {
      "epoch": 0.8875804506804053,
      "grad_norm": 0.8184322939655596,
      "learning_rate": 3.275386899693017e-07,
      "loss": 0.3478,
      "step": 14170
    },
    {
      "epoch": 0.8876430886797476,
      "grad_norm": 0.8135944283011708,
      "learning_rate": 3.271776770026963e-07,
      "loss": 0.358,
      "step": 14171
    },
    {
      "epoch": 0.8877057266790899,
      "grad_norm": 0.9189499078004679,
      "learning_rate": 3.26816856370325e-07,
      "loss": 0.3751,
      "step": 14172
    },
    {
      "epoch": 0.8877683646784321,
      "grad_norm": 0.8749130649599405,
      "learning_rate": 3.2645622808703867e-07,
      "loss": 0.3933,
      "step": 14173
    },
    {
      "epoch": 0.8878310026777745,
      "grad_norm": 0.7569357064970345,
      "learning_rate": 3.2609579216768163e-07,
      "loss": 0.3457,
      "step": 14174
    },
    {
      "epoch": 0.8878936406771167,
      "grad_norm": 0.9429824826366419,
      "learning_rate": 3.257355486270891e-07,
      "loss": 0.4076,
      "step": 14175
    },
    {
      "epoch": 0.8879562786764591,
      "grad_norm": 0.8919815355646258,
      "learning_rate": 3.253754974800888e-07,
      "loss": 0.3822,
      "step": 14176
    },
    {
      "epoch": 0.8880189166758014,
      "grad_norm": 0.825935006127046,
      "learning_rate": 3.250156387415015e-07,
      "loss": 0.3509,
      "step": 14177
    },
    {
      "epoch": 0.8880815546751437,
      "grad_norm": 0.7744085478657369,
      "learning_rate": 3.2465597242613835e-07,
      "loss": 0.3299,
      "step": 14178
    },
    {
      "epoch": 0.888144192674486,
      "grad_norm": 0.9164533516503423,
      "learning_rate": 3.242964985488029e-07,
      "loss": 0.3768,
      "step": 14179
    },
    {
      "epoch": 0.8882068306738283,
      "grad_norm": 0.952609441245086,
      "learning_rate": 3.2393721712429106e-07,
      "loss": 0.3917,
      "step": 14180
    },
    {
      "epoch": 0.8882694686731706,
      "grad_norm": 0.9121089239973922,
      "learning_rate": 3.2357812816739163e-07,
      "loss": 0.4207,
      "step": 14181
    },
    {
      "epoch": 0.8883321066725128,
      "grad_norm": 0.5761341837018129,
      "learning_rate": 3.232192316928845e-07,
      "loss": 0.4534,
      "step": 14182
    },
    {
      "epoch": 0.8883947446718552,
      "grad_norm": 0.8910453729738204,
      "learning_rate": 3.228605277155411e-07,
      "loss": 0.3807,
      "step": 14183
    },
    {
      "epoch": 0.8884573826711974,
      "grad_norm": 0.8834185338363633,
      "learning_rate": 3.2250201625012633e-07,
      "loss": 0.4102,
      "step": 14184
    },
    {
      "epoch": 0.8885200206705398,
      "grad_norm": 0.9100211816038437,
      "learning_rate": 3.221436973113978e-07,
      "loss": 0.416,
      "step": 14185
    },
    {
      "epoch": 0.8885826586698821,
      "grad_norm": 0.8950916872352738,
      "learning_rate": 3.2178557091410265e-07,
      "loss": 0.3744,
      "step": 14186
    },
    {
      "epoch": 0.8886452966692244,
      "grad_norm": 0.8872764601220826,
      "learning_rate": 3.2142763707298183e-07,
      "loss": 0.4199,
      "step": 14187
    },
    {
      "epoch": 0.8887079346685667,
      "grad_norm": 0.8701717967527733,
      "learning_rate": 3.210698958027675e-07,
      "loss": 0.4043,
      "step": 14188
    },
    {
      "epoch": 0.888770572667909,
      "grad_norm": 0.825321906504695,
      "learning_rate": 3.20712347118185e-07,
      "loss": 0.3928,
      "step": 14189
    },
    {
      "epoch": 0.8888332106672513,
      "grad_norm": 0.8875606256155797,
      "learning_rate": 3.2035499103395097e-07,
      "loss": 0.3745,
      "step": 14190
    },
    {
      "epoch": 0.8888958486665935,
      "grad_norm": 0.8754503638669873,
      "learning_rate": 3.199978275647736e-07,
      "loss": 0.4061,
      "step": 14191
    },
    {
      "epoch": 0.8889584866659359,
      "grad_norm": 0.9496106093456748,
      "learning_rate": 3.196408567253545e-07,
      "loss": 0.4285,
      "step": 14192
    },
    {
      "epoch": 0.8890211246652782,
      "grad_norm": 0.9545679278286293,
      "learning_rate": 3.192840785303858e-07,
      "loss": 0.411,
      "step": 14193
    },
    {
      "epoch": 0.8890837626646205,
      "grad_norm": 0.8691831683280746,
      "learning_rate": 3.189274929945541e-07,
      "loss": 0.3841,
      "step": 14194
    },
    {
      "epoch": 0.8891464006639628,
      "grad_norm": 0.9208601229239332,
      "learning_rate": 3.1857110013253534e-07,
      "loss": 0.4084,
      "step": 14195
    },
    {
      "epoch": 0.8892090386633051,
      "grad_norm": 0.8446722442649119,
      "learning_rate": 3.1821489995899956e-07,
      "loss": 0.3804,
      "step": 14196
    },
    {
      "epoch": 0.8892716766626474,
      "grad_norm": 0.9171523850020002,
      "learning_rate": 3.178588924886067e-07,
      "loss": 0.4205,
      "step": 14197
    },
    {
      "epoch": 0.8893343146619896,
      "grad_norm": 0.8722283455678899,
      "learning_rate": 3.175030777360116e-07,
      "loss": 0.3822,
      "step": 14198
    },
    {
      "epoch": 0.889396952661332,
      "grad_norm": 0.9258630987631375,
      "learning_rate": 3.171474557158588e-07,
      "loss": 0.4346,
      "step": 14199
    },
    {
      "epoch": 0.8894595906606743,
      "grad_norm": 0.9061721798033777,
      "learning_rate": 3.1679202644278484e-07,
      "loss": 0.3447,
      "step": 14200
    },
    {
      "epoch": 0.8895222286600166,
      "grad_norm": 0.8432470295542694,
      "learning_rate": 3.1643678993142135e-07,
      "loss": 0.3656,
      "step": 14201
    },
    {
      "epoch": 0.8895848666593589,
      "grad_norm": 0.6018727073089741,
      "learning_rate": 3.1608174619638775e-07,
      "loss": 0.4467,
      "step": 14202
    },
    {
      "epoch": 0.8896475046587012,
      "grad_norm": 0.8722214601106364,
      "learning_rate": 3.1572689525230014e-07,
      "loss": 0.388,
      "step": 14203
    },
    {
      "epoch": 0.8897101426580435,
      "grad_norm": 0.8516150906680048,
      "learning_rate": 3.153722371137624e-07,
      "loss": 0.3683,
      "step": 14204
    },
    {
      "epoch": 0.8897727806573859,
      "grad_norm": 0.9024221976879122,
      "learning_rate": 3.150177717953734e-07,
      "loss": 0.3456,
      "step": 14205
    },
    {
      "epoch": 0.8898354186567281,
      "grad_norm": 0.873899444351979,
      "learning_rate": 3.14663499311722e-07,
      "loss": 0.3616,
      "step": 14206
    },
    {
      "epoch": 0.8898980566560704,
      "grad_norm": 0.8368910439334751,
      "learning_rate": 3.14309419677391e-07,
      "loss": 0.3562,
      "step": 14207
    },
    {
      "epoch": 0.8899606946554127,
      "grad_norm": 0.8463259891920614,
      "learning_rate": 3.139555329069538e-07,
      "loss": 0.3561,
      "step": 14208
    },
    {
      "epoch": 0.890023332654755,
      "grad_norm": 0.8395401272203865,
      "learning_rate": 3.1360183901497645e-07,
      "loss": 0.3965,
      "step": 14209
    },
    {
      "epoch": 0.8900859706540973,
      "grad_norm": 0.8921601699061367,
      "learning_rate": 3.1324833801601727e-07,
      "loss": 0.3895,
      "step": 14210
    },
    {
      "epoch": 0.8901486086534396,
      "grad_norm": 0.8588177079393438,
      "learning_rate": 3.1289502992462585e-07,
      "loss": 0.396,
      "step": 14211
    },
    {
      "epoch": 0.890211246652782,
      "grad_norm": 0.8391175561474526,
      "learning_rate": 3.12541914755346e-07,
      "loss": 0.3826,
      "step": 14212
    },
    {
      "epoch": 0.8902738846521242,
      "grad_norm": 0.8872233729317084,
      "learning_rate": 3.1218899252271064e-07,
      "loss": 0.3759,
      "step": 14213
    },
    {
      "epoch": 0.8903365226514666,
      "grad_norm": 0.8544865413618571,
      "learning_rate": 3.118362632412464e-07,
      "loss": 0.4007,
      "step": 14214
    },
    {
      "epoch": 0.8903991606508088,
      "grad_norm": 0.8111990916113396,
      "learning_rate": 3.114837269254717e-07,
      "loss": 0.3793,
      "step": 14215
    },
    {
      "epoch": 0.8904617986501511,
      "grad_norm": 0.8439602334399934,
      "learning_rate": 3.1113138358989714e-07,
      "loss": 0.4032,
      "step": 14216
    },
    {
      "epoch": 0.8905244366494934,
      "grad_norm": 0.5938635553722339,
      "learning_rate": 3.1077923324902446e-07,
      "loss": 0.4466,
      "step": 14217
    },
    {
      "epoch": 0.8905870746488357,
      "grad_norm": 0.8922524268926572,
      "learning_rate": 3.104272759173499e-07,
      "loss": 0.3702,
      "step": 14218
    },
    {
      "epoch": 0.890649712648178,
      "grad_norm": 0.8401567254487564,
      "learning_rate": 3.1007551160935847e-07,
      "loss": 0.381,
      "step": 14219
    },
    {
      "epoch": 0.8907123506475203,
      "grad_norm": 0.7985084191577794,
      "learning_rate": 3.0972394033952913e-07,
      "loss": 0.3544,
      "step": 14220
    },
    {
      "epoch": 0.8907749886468627,
      "grad_norm": 0.9206210144605022,
      "learning_rate": 3.093725621223337e-07,
      "loss": 0.3632,
      "step": 14221
    },
    {
      "epoch": 0.8908376266462049,
      "grad_norm": 0.8169331511420946,
      "learning_rate": 3.0902137697223443e-07,
      "loss": 0.3651,
      "step": 14222
    },
    {
      "epoch": 0.8909002646455472,
      "grad_norm": 0.8522609145530996,
      "learning_rate": 3.086703849036854e-07,
      "loss": 0.3688,
      "step": 14223
    },
    {
      "epoch": 0.8909629026448895,
      "grad_norm": 0.8836669000377733,
      "learning_rate": 3.083195859311344e-07,
      "loss": 0.389,
      "step": 14224
    },
    {
      "epoch": 0.8910255406442318,
      "grad_norm": 0.8371939977650827,
      "learning_rate": 3.0796898006902e-07,
      "loss": 0.3986,
      "step": 14225
    },
    {
      "epoch": 0.8910881786435741,
      "grad_norm": 0.8424563106499953,
      "learning_rate": 3.0761856733177273e-07,
      "loss": 0.3673,
      "step": 14226
    },
    {
      "epoch": 0.8911508166429164,
      "grad_norm": 0.6295762817358482,
      "learning_rate": 3.0726834773381673e-07,
      "loss": 0.4353,
      "step": 14227
    },
    {
      "epoch": 0.8912134546422588,
      "grad_norm": 0.8528321479668526,
      "learning_rate": 3.0691832128956657e-07,
      "loss": 0.3759,
      "step": 14228
    },
    {
      "epoch": 0.891276092641601,
      "grad_norm": 0.8890280879202304,
      "learning_rate": 3.0656848801342896e-07,
      "loss": 0.3764,
      "step": 14229
    },
    {
      "epoch": 0.8913387306409434,
      "grad_norm": 0.8606850915881802,
      "learning_rate": 3.062188479198036e-07,
      "loss": 0.3749,
      "step": 14230
    },
    {
      "epoch": 0.8914013686402856,
      "grad_norm": 0.8736303435311897,
      "learning_rate": 3.0586940102308225e-07,
      "loss": 0.3533,
      "step": 14231
    },
    {
      "epoch": 0.8914640066396279,
      "grad_norm": 0.6618818023827332,
      "learning_rate": 3.0552014733764735e-07,
      "loss": 0.4482,
      "step": 14232
    },
    {
      "epoch": 0.8915266446389702,
      "grad_norm": 0.8703837610088229,
      "learning_rate": 3.051710868778746e-07,
      "loss": 0.4155,
      "step": 14233
    },
    {
      "epoch": 0.8915892826383125,
      "grad_norm": 0.9094828978281655,
      "learning_rate": 3.048222196581313e-07,
      "loss": 0.3973,
      "step": 14234
    },
    {
      "epoch": 0.8916519206376549,
      "grad_norm": 0.9378379055028611,
      "learning_rate": 3.044735456927761e-07,
      "loss": 0.4013,
      "step": 14235
    },
    {
      "epoch": 0.8917145586369971,
      "grad_norm": 0.7552568791757666,
      "learning_rate": 3.041250649961619e-07,
      "loss": 0.3649,
      "step": 14236
    },
    {
      "epoch": 0.8917771966363395,
      "grad_norm": 0.9076441363414376,
      "learning_rate": 3.0377677758263224e-07,
      "loss": 0.4255,
      "step": 14237
    },
    {
      "epoch": 0.8918398346356817,
      "grad_norm": 0.6843482942907136,
      "learning_rate": 3.0342868346652065e-07,
      "loss": 0.4421,
      "step": 14238
    },
    {
      "epoch": 0.8919024726350241,
      "grad_norm": 0.8312571098340201,
      "learning_rate": 3.0308078266215733e-07,
      "loss": 0.3456,
      "step": 14239
    },
    {
      "epoch": 0.8919651106343663,
      "grad_norm": 0.9339371205812677,
      "learning_rate": 3.027330751838603e-07,
      "loss": 0.3615,
      "step": 14240
    },
    {
      "epoch": 0.8920277486337086,
      "grad_norm": 0.8891884614159249,
      "learning_rate": 3.0238556104594194e-07,
      "loss": 0.3855,
      "step": 14241
    },
    {
      "epoch": 0.892090386633051,
      "grad_norm": 0.9014248571564232,
      "learning_rate": 3.0203824026270647e-07,
      "loss": 0.4122,
      "step": 14242
    },
    {
      "epoch": 0.8921530246323932,
      "grad_norm": 0.9463904306159475,
      "learning_rate": 3.0169111284844734e-07,
      "loss": 0.4043,
      "step": 14243
    },
    {
      "epoch": 0.8922156626317356,
      "grad_norm": 0.8396936380888939,
      "learning_rate": 3.013441788174554e-07,
      "loss": 0.3465,
      "step": 14244
    },
    {
      "epoch": 0.8922783006310778,
      "grad_norm": 0.8422617743237424,
      "learning_rate": 3.0099743818400915e-07,
      "loss": 0.39,
      "step": 14245
    },
    {
      "epoch": 0.8923409386304202,
      "grad_norm": 0.9001467283134987,
      "learning_rate": 3.006508909623801e-07,
      "loss": 0.3432,
      "step": 14246
    },
    {
      "epoch": 0.8924035766297624,
      "grad_norm": 0.8606127445645676,
      "learning_rate": 3.0030453716683283e-07,
      "loss": 0.3476,
      "step": 14247
    },
    {
      "epoch": 0.8924662146291047,
      "grad_norm": 0.8422779015307474,
      "learning_rate": 2.9995837681162266e-07,
      "loss": 0.422,
      "step": 14248
    },
    {
      "epoch": 0.8925288526284471,
      "grad_norm": 0.9865600476303554,
      "learning_rate": 2.9961240991099815e-07,
      "loss": 0.4091,
      "step": 14249
    },
    {
      "epoch": 0.8925914906277893,
      "grad_norm": 0.9040843180141844,
      "learning_rate": 2.992666364791996e-07,
      "loss": 0.4058,
      "step": 14250
    },
    {
      "epoch": 0.8926541286271317,
      "grad_norm": 0.921621279436494,
      "learning_rate": 2.98921056530459e-07,
      "loss": 0.4081,
      "step": 14251
    },
    {
      "epoch": 0.8927167666264739,
      "grad_norm": 0.9154067340901727,
      "learning_rate": 2.9857567007899936e-07,
      "loss": 0.3849,
      "step": 14252
    },
    {
      "epoch": 0.8927794046258163,
      "grad_norm": 0.877227061951378,
      "learning_rate": 2.9823047713903874e-07,
      "loss": 0.3624,
      "step": 14253
    },
    {
      "epoch": 0.8928420426251585,
      "grad_norm": 0.8844977247406983,
      "learning_rate": 2.9788547772478416e-07,
      "loss": 0.3796,
      "step": 14254
    },
    {
      "epoch": 0.8929046806245009,
      "grad_norm": 0.606997204513245,
      "learning_rate": 2.975406718504359e-07,
      "loss": 0.4398,
      "step": 14255
    },
    {
      "epoch": 0.8929673186238432,
      "grad_norm": 0.8936313526669126,
      "learning_rate": 2.971960595301865e-07,
      "loss": 0.4026,
      "step": 14256
    },
    {
      "epoch": 0.8930299566231854,
      "grad_norm": 0.8537703002552546,
      "learning_rate": 2.968516407782196e-07,
      "loss": 0.3799,
      "step": 14257
    },
    {
      "epoch": 0.8930925946225278,
      "grad_norm": 0.9229462677520888,
      "learning_rate": 2.9650741560871276e-07,
      "loss": 0.3895,
      "step": 14258
    },
    {
      "epoch": 0.89315523262187,
      "grad_norm": 0.8241883197760892,
      "learning_rate": 2.961633840358341e-07,
      "loss": 0.3579,
      "step": 14259
    },
    {
      "epoch": 0.8932178706212124,
      "grad_norm": 0.8586553048254622,
      "learning_rate": 2.9581954607374285e-07,
      "loss": 0.3643,
      "step": 14260
    },
    {
      "epoch": 0.8932805086205546,
      "grad_norm": 0.9229188574053797,
      "learning_rate": 2.9547590173659214e-07,
      "loss": 0.3675,
      "step": 14261
    },
    {
      "epoch": 0.893343146619897,
      "grad_norm": 0.6074005858205944,
      "learning_rate": 2.9513245103852673e-07,
      "loss": 0.4499,
      "step": 14262
    },
    {
      "epoch": 0.8934057846192393,
      "grad_norm": 0.8694321991620504,
      "learning_rate": 2.947891939936831e-07,
      "loss": 0.3935,
      "step": 14263
    },
    {
      "epoch": 0.8934684226185816,
      "grad_norm": 0.8003269050269973,
      "learning_rate": 2.9444613061619e-07,
      "loss": 0.3572,
      "step": 14264
    },
    {
      "epoch": 0.8935310606179239,
      "grad_norm": 0.8945601830812837,
      "learning_rate": 2.941032609201672e-07,
      "loss": 0.401,
      "step": 14265
    },
    {
      "epoch": 0.8935936986172661,
      "grad_norm": 0.8558445521079702,
      "learning_rate": 2.937605849197267e-07,
      "loss": 0.3892,
      "step": 14266
    },
    {
      "epoch": 0.8936563366166085,
      "grad_norm": 0.8678800366783391,
      "learning_rate": 2.934181026289751e-07,
      "loss": 0.3751,
      "step": 14267
    },
    {
      "epoch": 0.8937189746159507,
      "grad_norm": 0.8388575190402071,
      "learning_rate": 2.9307581406200767e-07,
      "loss": 0.3648,
      "step": 14268
    },
    {
      "epoch": 0.8937816126152931,
      "grad_norm": 0.8168956216038544,
      "learning_rate": 2.927337192329127e-07,
      "loss": 0.3311,
      "step": 14269
    },
    {
      "epoch": 0.8938442506146353,
      "grad_norm": 0.9052638822604693,
      "learning_rate": 2.9239181815577276e-07,
      "loss": 0.3773,
      "step": 14270
    },
    {
      "epoch": 0.8939068886139777,
      "grad_norm": 0.8580523577522058,
      "learning_rate": 2.920501108446588e-07,
      "loss": 0.3686,
      "step": 14271
    },
    {
      "epoch": 0.89396952661332,
      "grad_norm": 0.6502320752355706,
      "learning_rate": 2.9170859731363634e-07,
      "loss": 0.4475,
      "step": 14272
    },
    {
      "epoch": 0.8940321646126622,
      "grad_norm": 0.8224231152388702,
      "learning_rate": 2.9136727757676174e-07,
      "loss": 0.3733,
      "step": 14273
    },
    {
      "epoch": 0.8940948026120046,
      "grad_norm": 0.8847875245199115,
      "learning_rate": 2.9102615164808444e-07,
      "loss": 0.4011,
      "step": 14274
    },
    {
      "epoch": 0.8941574406113468,
      "grad_norm": 0.8689336099060786,
      "learning_rate": 2.9068521954164373e-07,
      "loss": 0.3948,
      "step": 14275
    },
    {
      "epoch": 0.8942200786106892,
      "grad_norm": 0.7966955632856474,
      "learning_rate": 2.9034448127147395e-07,
      "loss": 0.3471,
      "step": 14276
    },
    {
      "epoch": 0.8942827166100314,
      "grad_norm": 0.9095458048538375,
      "learning_rate": 2.900039368515994e-07,
      "loss": 0.3838,
      "step": 14277
    },
    {
      "epoch": 0.8943453546093738,
      "grad_norm": 0.8702720264499085,
      "learning_rate": 2.8966358629603674e-07,
      "loss": 0.4069,
      "step": 14278
    },
    {
      "epoch": 0.8944079926087161,
      "grad_norm": 0.8350601903000797,
      "learning_rate": 2.893234296187958e-07,
      "loss": 0.3752,
      "step": 14279
    },
    {
      "epoch": 0.8944706306080584,
      "grad_norm": 0.8576602573859469,
      "learning_rate": 2.88983466833877e-07,
      "loss": 0.3697,
      "step": 14280
    },
    {
      "epoch": 0.8945332686074007,
      "grad_norm": 0.8572752220021613,
      "learning_rate": 2.886436979552726e-07,
      "loss": 0.3788,
      "step": 14281
    },
    {
      "epoch": 0.8945959066067429,
      "grad_norm": 0.9052531980586812,
      "learning_rate": 2.883041229969685e-07,
      "loss": 0.3847,
      "step": 14282
    },
    {
      "epoch": 0.8946585446060853,
      "grad_norm": 0.8711931443177112,
      "learning_rate": 2.879647419729409e-07,
      "loss": 0.3853,
      "step": 14283
    },
    {
      "epoch": 0.8947211826054275,
      "grad_norm": 0.583111971250103,
      "learning_rate": 2.8762555489715904e-07,
      "loss": 0.4349,
      "step": 14284
    },
    {
      "epoch": 0.8947838206047699,
      "grad_norm": 0.9131456160112209,
      "learning_rate": 2.8728656178358403e-07,
      "loss": 0.3974,
      "step": 14285
    },
    {
      "epoch": 0.8948464586041122,
      "grad_norm": 0.8617824832701602,
      "learning_rate": 2.869477626461692e-07,
      "loss": 0.3968,
      "step": 14286
    },
    {
      "epoch": 0.8949090966034545,
      "grad_norm": 0.8249753483041508,
      "learning_rate": 2.866091574988583e-07,
      "loss": 0.3764,
      "step": 14287
    },
    {
      "epoch": 0.8949717346027968,
      "grad_norm": 0.8858894821617089,
      "learning_rate": 2.8627074635559024e-07,
      "loss": 0.3709,
      "step": 14288
    },
    {
      "epoch": 0.8950343726021391,
      "grad_norm": 0.8132275951013057,
      "learning_rate": 2.859325292302928e-07,
      "loss": 0.3832,
      "step": 14289
    },
    {
      "epoch": 0.8950970106014814,
      "grad_norm": 0.8707994119636704,
      "learning_rate": 2.855945061368881e-07,
      "loss": 0.4024,
      "step": 14290
    },
    {
      "epoch": 0.8951596486008236,
      "grad_norm": 0.8784313181292672,
      "learning_rate": 2.852566770892878e-07,
      "loss": 0.3995,
      "step": 14291
    },
    {
      "epoch": 0.895222286600166,
      "grad_norm": 0.8252688927769382,
      "learning_rate": 2.849190421013981e-07,
      "loss": 0.3196,
      "step": 14292
    },
    {
      "epoch": 0.8952849245995083,
      "grad_norm": 0.9488189226492064,
      "learning_rate": 2.845816011871161e-07,
      "loss": 0.3754,
      "step": 14293
    },
    {
      "epoch": 0.8953475625988506,
      "grad_norm": 0.8386899188431887,
      "learning_rate": 2.842443543603296e-07,
      "loss": 0.3829,
      "step": 14294
    },
    {
      "epoch": 0.8954102005981929,
      "grad_norm": 0.950962895317383,
      "learning_rate": 2.8390730163492087e-07,
      "loss": 0.4105,
      "step": 14295
    },
    {
      "epoch": 0.8954728385975352,
      "grad_norm": 0.855336416668441,
      "learning_rate": 2.8357044302476323e-07,
      "loss": 0.3878,
      "step": 14296
    },
    {
      "epoch": 0.8955354765968775,
      "grad_norm": 0.8749179516755972,
      "learning_rate": 2.832337785437222e-07,
      "loss": 0.3666,
      "step": 14297
    },
    {
      "epoch": 0.8955981145962199,
      "grad_norm": 0.8372517062641676,
      "learning_rate": 2.828973082056541e-07,
      "loss": 0.3914,
      "step": 14298
    },
    {
      "epoch": 0.8956607525955621,
      "grad_norm": 1.0025817881588908,
      "learning_rate": 2.825610320244082e-07,
      "loss": 0.3831,
      "step": 14299
    },
    {
      "epoch": 0.8957233905949044,
      "grad_norm": 0.8504756876033578,
      "learning_rate": 2.8222495001382575e-07,
      "loss": 0.3566,
      "step": 14300
    },
    {
      "epoch": 0.8957860285942467,
      "grad_norm": 0.8299523301624452,
      "learning_rate": 2.8188906218774017e-07,
      "loss": 0.3711,
      "step": 14301
    },
    {
      "epoch": 0.895848666593589,
      "grad_norm": 0.6952208423337554,
      "learning_rate": 2.8155336855997583e-07,
      "loss": 0.4385,
      "step": 14302
    },
    {
      "epoch": 0.8959113045929313,
      "grad_norm": 0.8895673205881611,
      "learning_rate": 2.812178691443512e-07,
      "loss": 0.3872,
      "step": 14303
    },
    {
      "epoch": 0.8959739425922736,
      "grad_norm": 0.9086479888840019,
      "learning_rate": 2.8088256395467415e-07,
      "loss": 0.372,
      "step": 14304
    },
    {
      "epoch": 0.896036580591616,
      "grad_norm": 0.8486778281935715,
      "learning_rate": 2.805474530047475e-07,
      "loss": 0.3725,
      "step": 14305
    },
    {
      "epoch": 0.8960992185909582,
      "grad_norm": 0.8964174170791154,
      "learning_rate": 2.8021253630836407e-07,
      "loss": 0.4062,
      "step": 14306
    },
    {
      "epoch": 0.8961618565903005,
      "grad_norm": 0.8390722313980362,
      "learning_rate": 2.7987781387930845e-07,
      "loss": 0.3729,
      "step": 14307
    },
    {
      "epoch": 0.8962244945896428,
      "grad_norm": 0.8310924536567917,
      "learning_rate": 2.795432857313579e-07,
      "loss": 0.3509,
      "step": 14308
    },
    {
      "epoch": 0.8962871325889851,
      "grad_norm": 0.8687737278748768,
      "learning_rate": 2.7920895187828203e-07,
      "loss": 0.3586,
      "step": 14309
    },
    {
      "epoch": 0.8963497705883274,
      "grad_norm": 0.8729981348021092,
      "learning_rate": 2.788748123338414e-07,
      "loss": 0.3542,
      "step": 14310
    },
    {
      "epoch": 0.8964124085876697,
      "grad_norm": 0.8188075396144217,
      "learning_rate": 2.7854086711178953e-07,
      "loss": 0.3532,
      "step": 14311
    },
    {
      "epoch": 0.896475046587012,
      "grad_norm": 0.8684432717043526,
      "learning_rate": 2.782071162258726e-07,
      "loss": 0.3956,
      "step": 14312
    },
    {
      "epoch": 0.8965376845863543,
      "grad_norm": 0.9198315314740477,
      "learning_rate": 2.778735596898263e-07,
      "loss": 0.4069,
      "step": 14313
    },
    {
      "epoch": 0.8966003225856967,
      "grad_norm": 0.9048224539095053,
      "learning_rate": 2.775401975173819e-07,
      "loss": 0.3881,
      "step": 14314
    },
    {
      "epoch": 0.8966629605850389,
      "grad_norm": 0.9944439110007334,
      "learning_rate": 2.77207029722259e-07,
      "loss": 0.3919,
      "step": 14315
    },
    {
      "epoch": 0.8967255985843812,
      "grad_norm": 0.9098680227712398,
      "learning_rate": 2.7687405631817153e-07,
      "loss": 0.4097,
      "step": 14316
    },
    {
      "epoch": 0.8967882365837235,
      "grad_norm": 0.8881117467311148,
      "learning_rate": 2.765412773188247e-07,
      "loss": 0.3668,
      "step": 14317
    },
    {
      "epoch": 0.8968508745830658,
      "grad_norm": 0.7936293603409523,
      "learning_rate": 2.762086927379154e-07,
      "loss": 0.3444,
      "step": 14318
    },
    {
      "epoch": 0.8969135125824081,
      "grad_norm": 0.8916321889803542,
      "learning_rate": 2.758763025891331e-07,
      "loss": 0.3884,
      "step": 14319
    },
    {
      "epoch": 0.8969761505817504,
      "grad_norm": 0.8834657223492456,
      "learning_rate": 2.755441068861586e-07,
      "loss": 0.3995,
      "step": 14320
    },
    {
      "epoch": 0.8970387885810928,
      "grad_norm": 0.9089543795199876,
      "learning_rate": 2.7521210564266544e-07,
      "loss": 0.3709,
      "step": 14321
    },
    {
      "epoch": 0.897101426580435,
      "grad_norm": 0.8976842264133892,
      "learning_rate": 2.7488029887231927e-07,
      "loss": 0.375,
      "step": 14322
    },
    {
      "epoch": 0.8971640645797774,
      "grad_norm": 0.8442826184231725,
      "learning_rate": 2.7454868658877756e-07,
      "loss": 0.3349,
      "step": 14323
    },
    {
      "epoch": 0.8972267025791196,
      "grad_norm": 0.8773495453447628,
      "learning_rate": 2.7421726880568876e-07,
      "loss": 0.3585,
      "step": 14324
    },
    {
      "epoch": 0.8972893405784619,
      "grad_norm": 0.9413839925227435,
      "learning_rate": 2.738860455366943e-07,
      "loss": 0.4063,
      "step": 14325
    },
    {
      "epoch": 0.8973519785778042,
      "grad_norm": 0.8913655974760417,
      "learning_rate": 2.735550167954276e-07,
      "loss": 0.4182,
      "step": 14326
    },
    {
      "epoch": 0.8974146165771465,
      "grad_norm": 0.8508440386189441,
      "learning_rate": 2.7322418259551343e-07,
      "loss": 0.3927,
      "step": 14327
    },
    {
      "epoch": 0.8974772545764889,
      "grad_norm": 0.8396421102933385,
      "learning_rate": 2.728935429505686e-07,
      "loss": 0.3688,
      "step": 14328
    },
    {
      "epoch": 0.8975398925758311,
      "grad_norm": 0.8958656884171844,
      "learning_rate": 2.7256309787420387e-07,
      "loss": 0.3786,
      "step": 14329
    },
    {
      "epoch": 0.8976025305751735,
      "grad_norm": 0.9454556329839228,
      "learning_rate": 2.722328473800184e-07,
      "loss": 0.3954,
      "step": 14330
    },
    {
      "epoch": 0.8976651685745157,
      "grad_norm": 0.8256003624733897,
      "learning_rate": 2.719027914816075e-07,
      "loss": 0.3552,
      "step": 14331
    },
    {
      "epoch": 0.897727806573858,
      "grad_norm": 0.6740657054494643,
      "learning_rate": 2.7157293019255513e-07,
      "loss": 0.4349,
      "step": 14332
    },
    {
      "epoch": 0.8977904445732003,
      "grad_norm": 0.6809996305570334,
      "learning_rate": 2.7124326352643835e-07,
      "loss": 0.4571,
      "step": 14333
    },
    {
      "epoch": 0.8978530825725426,
      "grad_norm": 0.9311410540901185,
      "learning_rate": 2.7091379149682683e-07,
      "loss": 0.4457,
      "step": 14334
    },
    {
      "epoch": 0.897915720571885,
      "grad_norm": 0.9027428981278023,
      "learning_rate": 2.705845141172814e-07,
      "loss": 0.4034,
      "step": 14335
    },
    {
      "epoch": 0.8979783585712272,
      "grad_norm": 0.8671635641539329,
      "learning_rate": 2.70255431401355e-07,
      "loss": 0.4089,
      "step": 14336
    },
    {
      "epoch": 0.8980409965705696,
      "grad_norm": 0.8598228437675774,
      "learning_rate": 2.699265433625925e-07,
      "loss": 0.4106,
      "step": 14337
    },
    {
      "epoch": 0.8981036345699118,
      "grad_norm": 0.6301652939104607,
      "learning_rate": 2.6959785001453187e-07,
      "loss": 0.446,
      "step": 14338
    },
    {
      "epoch": 0.8981662725692542,
      "grad_norm": 0.846251183250362,
      "learning_rate": 2.6926935137070175e-07,
      "loss": 0.3846,
      "step": 14339
    },
    {
      "epoch": 0.8982289105685964,
      "grad_norm": 0.9028681905787441,
      "learning_rate": 2.68941047444623e-07,
      "loss": 0.3977,
      "step": 14340
    },
    {
      "epoch": 0.8982915485679387,
      "grad_norm": 0.8050973030138306,
      "learning_rate": 2.6861293824980874e-07,
      "loss": 0.3618,
      "step": 14341
    },
    {
      "epoch": 0.898354186567281,
      "grad_norm": 0.8944452941333808,
      "learning_rate": 2.682850237997647e-07,
      "loss": 0.4559,
      "step": 14342
    },
    {
      "epoch": 0.8984168245666233,
      "grad_norm": 0.8546997562662199,
      "learning_rate": 2.679573041079875e-07,
      "loss": 0.3703,
      "step": 14343
    },
    {
      "epoch": 0.8984794625659657,
      "grad_norm": 0.8273607919730627,
      "learning_rate": 2.676297791879656e-07,
      "loss": 0.3732,
      "step": 14344
    },
    {
      "epoch": 0.8985421005653079,
      "grad_norm": 0.9126584601679723,
      "learning_rate": 2.6730244905318e-07,
      "loss": 0.4214,
      "step": 14345
    },
    {
      "epoch": 0.8986047385646503,
      "grad_norm": 0.8567741548370126,
      "learning_rate": 2.6697531371710383e-07,
      "loss": 0.3639,
      "step": 14346
    },
    {
      "epoch": 0.8986673765639925,
      "grad_norm": 0.886672215601459,
      "learning_rate": 2.666483731932029e-07,
      "loss": 0.4224,
      "step": 14347
    },
    {
      "epoch": 0.8987300145633349,
      "grad_norm": 0.8478267726650823,
      "learning_rate": 2.663216274949332e-07,
      "loss": 0.3517,
      "step": 14348
    },
    {
      "epoch": 0.8987926525626772,
      "grad_norm": 0.6668028519301397,
      "learning_rate": 2.6599507663574387e-07,
      "loss": 0.4618,
      "step": 14349
    },
    {
      "epoch": 0.8988552905620194,
      "grad_norm": 0.9261982836674114,
      "learning_rate": 2.656687206290759e-07,
      "loss": 0.4314,
      "step": 14350
    },
    {
      "epoch": 0.8989179285613618,
      "grad_norm": 0.870072947906364,
      "learning_rate": 2.653425594883624e-07,
      "loss": 0.3948,
      "step": 14351
    },
    {
      "epoch": 0.898980566560704,
      "grad_norm": 0.8733515106669485,
      "learning_rate": 2.6501659322702765e-07,
      "loss": 0.3841,
      "step": 14352
    },
    {
      "epoch": 0.8990432045600464,
      "grad_norm": 0.9430715956108922,
      "learning_rate": 2.6469082185848926e-07,
      "loss": 0.3918,
      "step": 14353
    },
    {
      "epoch": 0.8991058425593886,
      "grad_norm": 0.8545059394755627,
      "learning_rate": 2.643652453961548e-07,
      "loss": 0.3895,
      "step": 14354
    },
    {
      "epoch": 0.899168480558731,
      "grad_norm": 0.897254044339704,
      "learning_rate": 2.640398638534264e-07,
      "loss": 0.3723,
      "step": 14355
    },
    {
      "epoch": 0.8992311185580733,
      "grad_norm": 0.9280590569320921,
      "learning_rate": 2.6371467724369603e-07,
      "loss": 0.3733,
      "step": 14356
    },
    {
      "epoch": 0.8992937565574155,
      "grad_norm": 0.8146484399743495,
      "learning_rate": 2.633896855803486e-07,
      "loss": 0.3423,
      "step": 14357
    },
    {
      "epoch": 0.8993563945567579,
      "grad_norm": 0.8021205909896214,
      "learning_rate": 2.6306488887676065e-07,
      "loss": 0.3234,
      "step": 14358
    },
    {
      "epoch": 0.8994190325561001,
      "grad_norm": 0.6088044986905019,
      "learning_rate": 2.627402871463014e-07,
      "loss": 0.4631,
      "step": 14359
    },
    {
      "epoch": 0.8994816705554425,
      "grad_norm": 0.7854999457070483,
      "learning_rate": 2.6241588040233136e-07,
      "loss": 0.3688,
      "step": 14360
    },
    {
      "epoch": 0.8995443085547847,
      "grad_norm": 0.8757503344093331,
      "learning_rate": 2.6209166865820257e-07,
      "loss": 0.3869,
      "step": 14361
    },
    {
      "epoch": 0.8996069465541271,
      "grad_norm": 0.8611928481024373,
      "learning_rate": 2.617676519272605e-07,
      "loss": 0.3522,
      "step": 14362
    },
    {
      "epoch": 0.8996695845534693,
      "grad_norm": 0.8197197451050644,
      "learning_rate": 2.6144383022284056e-07,
      "loss": 0.355,
      "step": 14363
    },
    {
      "epoch": 0.8997322225528117,
      "grad_norm": 0.8730077339880595,
      "learning_rate": 2.6112020355827327e-07,
      "loss": 0.3721,
      "step": 14364
    },
    {
      "epoch": 0.899794860552154,
      "grad_norm": 0.9199556058031785,
      "learning_rate": 2.6079677194687737e-07,
      "loss": 0.4029,
      "step": 14365
    },
    {
      "epoch": 0.8998574985514962,
      "grad_norm": 0.9507776839171186,
      "learning_rate": 2.6047353540196606e-07,
      "loss": 0.4324,
      "step": 14366
    },
    {
      "epoch": 0.8999201365508386,
      "grad_norm": 0.9401565204335052,
      "learning_rate": 2.601504939368432e-07,
      "loss": 0.4291,
      "step": 14367
    },
    {
      "epoch": 0.8999827745501808,
      "grad_norm": 0.9124622722494016,
      "learning_rate": 2.598276475648065e-07,
      "loss": 0.4135,
      "step": 14368
    },
    {
      "epoch": 0.9000454125495232,
      "grad_norm": 0.8798743802062009,
      "learning_rate": 2.595049962991436e-07,
      "loss": 0.3665,
      "step": 14369
    },
    {
      "epoch": 0.9001080505488654,
      "grad_norm": 0.8726020208403374,
      "learning_rate": 2.5918254015313506e-07,
      "loss": 0.3597,
      "step": 14370
    },
    {
      "epoch": 0.9001706885482078,
      "grad_norm": 0.8950567930691518,
      "learning_rate": 2.588602791400524e-07,
      "loss": 0.4136,
      "step": 14371
    },
    {
      "epoch": 0.9002333265475501,
      "grad_norm": 0.5703242044518737,
      "learning_rate": 2.585382132731606e-07,
      "loss": 0.4336,
      "step": 14372
    },
    {
      "epoch": 0.9002959645468924,
      "grad_norm": 0.8910507101092005,
      "learning_rate": 2.582163425657169e-07,
      "loss": 0.3723,
      "step": 14373
    },
    {
      "epoch": 0.9003586025462347,
      "grad_norm": 0.878492075785251,
      "learning_rate": 2.5789466703096787e-07,
      "loss": 0.4246,
      "step": 14374
    },
    {
      "epoch": 0.9004212405455769,
      "grad_norm": 0.870934467384153,
      "learning_rate": 2.5757318668215514e-07,
      "loss": 0.3796,
      "step": 14375
    },
    {
      "epoch": 0.9004838785449193,
      "grad_norm": 0.9577991353792757,
      "learning_rate": 2.5725190153250925e-07,
      "loss": 0.4078,
      "step": 14376
    },
    {
      "epoch": 0.9005465165442615,
      "grad_norm": 0.6815931758777708,
      "learning_rate": 2.5693081159525626e-07,
      "loss": 0.4406,
      "step": 14377
    },
    {
      "epoch": 0.9006091545436039,
      "grad_norm": 0.8782054545509501,
      "learning_rate": 2.566099168836117e-07,
      "loss": 0.3724,
      "step": 14378
    },
    {
      "epoch": 0.9006717925429462,
      "grad_norm": 0.8951872034579966,
      "learning_rate": 2.562892174107828e-07,
      "loss": 0.3855,
      "step": 14379
    },
    {
      "epoch": 0.9007344305422885,
      "grad_norm": 0.8450140842002598,
      "learning_rate": 2.559687131899702e-07,
      "loss": 0.3798,
      "step": 14380
    },
    {
      "epoch": 0.9007970685416308,
      "grad_norm": 0.882098244326199,
      "learning_rate": 2.55648404234366e-07,
      "loss": 0.4098,
      "step": 14381
    },
    {
      "epoch": 0.900859706540973,
      "grad_norm": 0.7882850187339416,
      "learning_rate": 2.553282905571541e-07,
      "loss": 0.3617,
      "step": 14382
    },
    {
      "epoch": 0.9009223445403154,
      "grad_norm": 0.866853563820558,
      "learning_rate": 2.5500837217151075e-07,
      "loss": 0.3808,
      "step": 14383
    },
    {
      "epoch": 0.9009849825396576,
      "grad_norm": 0.5931417882276466,
      "learning_rate": 2.5468864909060307e-07,
      "loss": 0.4332,
      "step": 14384
    },
    {
      "epoch": 0.901047620539,
      "grad_norm": 0.8585313129115665,
      "learning_rate": 2.5436912132759175e-07,
      "loss": 0.3963,
      "step": 14385
    },
    {
      "epoch": 0.9011102585383423,
      "grad_norm": 0.870269729303079,
      "learning_rate": 2.540497888956273e-07,
      "loss": 0.4012,
      "step": 14386
    },
    {
      "epoch": 0.9011728965376846,
      "grad_norm": 0.8722951040171596,
      "learning_rate": 2.5373065180785525e-07,
      "loss": 0.3298,
      "step": 14387
    },
    {
      "epoch": 0.9012355345370269,
      "grad_norm": 0.5912488748131159,
      "learning_rate": 2.534117100774108e-07,
      "loss": 0.452,
      "step": 14388
    },
    {
      "epoch": 0.9012981725363692,
      "grad_norm": 0.8082891855073645,
      "learning_rate": 2.5309296371742007e-07,
      "loss": 0.3646,
      "step": 14389
    },
    {
      "epoch": 0.9013608105357115,
      "grad_norm": 0.8368153230715094,
      "learning_rate": 2.5277441274100536e-07,
      "loss": 0.3685,
      "step": 14390
    },
    {
      "epoch": 0.9014234485350537,
      "grad_norm": 0.8260312454815939,
      "learning_rate": 2.524560571612761e-07,
      "loss": 0.3577,
      "step": 14391
    },
    {
      "epoch": 0.9014860865343961,
      "grad_norm": 0.8992186601616257,
      "learning_rate": 2.5213789699133737e-07,
      "loss": 0.3852,
      "step": 14392
    },
    {
      "epoch": 0.9015487245337384,
      "grad_norm": 0.9142421357377482,
      "learning_rate": 2.518199322442838e-07,
      "loss": 0.4073,
      "step": 14393
    },
    {
      "epoch": 0.9016113625330807,
      "grad_norm": 0.8416155840986792,
      "learning_rate": 2.515021629332032e-07,
      "loss": 0.3994,
      "step": 14394
    },
    {
      "epoch": 0.901674000532423,
      "grad_norm": 0.874397983904406,
      "learning_rate": 2.511845890711745e-07,
      "loss": 0.3862,
      "step": 14395
    },
    {
      "epoch": 0.9017366385317653,
      "grad_norm": 0.8956852623137096,
      "learning_rate": 2.508672106712701e-07,
      "loss": 0.3447,
      "step": 14396
    },
    {
      "epoch": 0.9017992765311076,
      "grad_norm": 0.8350867065223302,
      "learning_rate": 2.5055002774655226e-07,
      "loss": 0.3557,
      "step": 14397
    },
    {
      "epoch": 0.90186191453045,
      "grad_norm": 0.8831885074392142,
      "learning_rate": 2.502330403100767e-07,
      "loss": 0.3745,
      "step": 14398
    },
    {
      "epoch": 0.9019245525297922,
      "grad_norm": 0.8241025712588292,
      "learning_rate": 2.499162483748913e-07,
      "loss": 0.3666,
      "step": 14399
    },
    {
      "epoch": 0.9019871905291345,
      "grad_norm": 0.9751017462595527,
      "learning_rate": 2.4959965195403504e-07,
      "loss": 0.459,
      "step": 14400
    },
    {
      "epoch": 0.9020498285284768,
      "grad_norm": 0.8480102596897426,
      "learning_rate": 2.492832510605386e-07,
      "loss": 0.3642,
      "step": 14401
    },
    {
      "epoch": 0.9021124665278191,
      "grad_norm": 0.8345207767022091,
      "learning_rate": 2.4896704570742504e-07,
      "loss": 0.3952,
      "step": 14402
    },
    {
      "epoch": 0.9021751045271614,
      "grad_norm": 0.9889788454504901,
      "learning_rate": 2.486510359077099e-07,
      "loss": 0.3534,
      "step": 14403
    },
    {
      "epoch": 0.9022377425265037,
      "grad_norm": 0.890864050087619,
      "learning_rate": 2.483352216743995e-07,
      "loss": 0.3486,
      "step": 14404
    },
    {
      "epoch": 0.902300380525846,
      "grad_norm": 0.8603018663919936,
      "learning_rate": 2.4801960302049397e-07,
      "loss": 0.3897,
      "step": 14405
    },
    {
      "epoch": 0.9023630185251883,
      "grad_norm": 0.9734171398842055,
      "learning_rate": 2.47704179958983e-07,
      "loss": 0.4398,
      "step": 14406
    },
    {
      "epoch": 0.9024256565245307,
      "grad_norm": 0.8145832290407347,
      "learning_rate": 2.4738895250285054e-07,
      "loss": 0.3764,
      "step": 14407
    },
    {
      "epoch": 0.9024882945238729,
      "grad_norm": 0.9308113785097916,
      "learning_rate": 2.4707392066507077e-07,
      "loss": 0.4078,
      "step": 14408
    },
    {
      "epoch": 0.9025509325232152,
      "grad_norm": 0.8892975092171601,
      "learning_rate": 2.46759084458611e-07,
      "loss": 0.3873,
      "step": 14409
    },
    {
      "epoch": 0.9026135705225575,
      "grad_norm": 0.8944351696909383,
      "learning_rate": 2.464444438964292e-07,
      "loss": 0.4194,
      "step": 14410
    },
    {
      "epoch": 0.9026762085218998,
      "grad_norm": 0.9006192166296613,
      "learning_rate": 2.461299989914767e-07,
      "loss": 0.3861,
      "step": 14411
    },
    {
      "epoch": 0.9027388465212421,
      "grad_norm": 0.8287012517650527,
      "learning_rate": 2.458157497566954e-07,
      "loss": 0.3562,
      "step": 14412
    },
    {
      "epoch": 0.9028014845205844,
      "grad_norm": 0.8554093395783705,
      "learning_rate": 2.4550169620501996e-07,
      "loss": 0.4021,
      "step": 14413
    },
    {
      "epoch": 0.9028641225199268,
      "grad_norm": 0.8545475626769443,
      "learning_rate": 2.451878383493778e-07,
      "loss": 0.3919,
      "step": 14414
    },
    {
      "epoch": 0.902926760519269,
      "grad_norm": 0.8581170571147995,
      "learning_rate": 2.4487417620268625e-07,
      "loss": 0.3592,
      "step": 14415
    },
    {
      "epoch": 0.9029893985186113,
      "grad_norm": 0.7824429965350672,
      "learning_rate": 2.445607097778563e-07,
      "loss": 0.3474,
      "step": 14416
    },
    {
      "epoch": 0.9030520365179536,
      "grad_norm": 0.8966511671281959,
      "learning_rate": 2.442474390877908e-07,
      "loss": 0.3853,
      "step": 14417
    },
    {
      "epoch": 0.9031146745172959,
      "grad_norm": 0.8899360853077366,
      "learning_rate": 2.4393436414538287e-07,
      "loss": 0.3693,
      "step": 14418
    },
    {
      "epoch": 0.9031773125166382,
      "grad_norm": 0.8783688488496341,
      "learning_rate": 2.4362148496351987e-07,
      "loss": 0.3135,
      "step": 14419
    },
    {
      "epoch": 0.9032399505159805,
      "grad_norm": 0.8232678739240394,
      "learning_rate": 2.4330880155507887e-07,
      "loss": 0.3551,
      "step": 14420
    },
    {
      "epoch": 0.9033025885153229,
      "grad_norm": 0.8661415200026101,
      "learning_rate": 2.4299631393293057e-07,
      "loss": 0.3891,
      "step": 14421
    },
    {
      "epoch": 0.9033652265146651,
      "grad_norm": 0.6366136877983773,
      "learning_rate": 2.4268402210993633e-07,
      "loss": 0.4485,
      "step": 14422
    },
    {
      "epoch": 0.9034278645140075,
      "grad_norm": 0.9245819591797088,
      "learning_rate": 2.4237192609895157e-07,
      "loss": 0.4136,
      "step": 14423
    },
    {
      "epoch": 0.9034905025133497,
      "grad_norm": 0.9034428390435895,
      "learning_rate": 2.4206002591282027e-07,
      "loss": 0.4049,
      "step": 14424
    },
    {
      "epoch": 0.903553140512692,
      "grad_norm": 0.7624294905256396,
      "learning_rate": 2.4174832156438287e-07,
      "loss": 0.3509,
      "step": 14425
    },
    {
      "epoch": 0.9036157785120343,
      "grad_norm": 0.8651396353382869,
      "learning_rate": 2.4143681306646684e-07,
      "loss": 0.3531,
      "step": 14426
    },
    {
      "epoch": 0.9036784165113766,
      "grad_norm": 0.863425376649713,
      "learning_rate": 2.411255004318952e-07,
      "loss": 0.3939,
      "step": 14427
    },
    {
      "epoch": 0.903741054510719,
      "grad_norm": 0.8872063469276973,
      "learning_rate": 2.4081438367348163e-07,
      "loss": 0.4002,
      "step": 14428
    },
    {
      "epoch": 0.9038036925100612,
      "grad_norm": 0.9009244407051864,
      "learning_rate": 2.405034628040309e-07,
      "loss": 0.3766,
      "step": 14429
    },
    {
      "epoch": 0.9038663305094036,
      "grad_norm": 0.9743765265445637,
      "learning_rate": 2.401927378363411e-07,
      "loss": 0.4346,
      "step": 14430
    },
    {
      "epoch": 0.9039289685087458,
      "grad_norm": 0.8719161107373345,
      "learning_rate": 2.398822087832009e-07,
      "loss": 0.3756,
      "step": 14431
    },
    {
      "epoch": 0.9039916065080882,
      "grad_norm": 0.9690463532025732,
      "learning_rate": 2.395718756573928e-07,
      "loss": 0.3966,
      "step": 14432
    },
    {
      "epoch": 0.9040542445074304,
      "grad_norm": 0.8810532919925511,
      "learning_rate": 2.3926173847169056e-07,
      "loss": 0.3809,
      "step": 14433
    },
    {
      "epoch": 0.9041168825067727,
      "grad_norm": 0.8627233524954493,
      "learning_rate": 2.3895179723885886e-07,
      "loss": 0.4181,
      "step": 14434
    },
    {
      "epoch": 0.904179520506115,
      "grad_norm": 0.8478546718392607,
      "learning_rate": 2.3864205197165426e-07,
      "loss": 0.3682,
      "step": 14435
    },
    {
      "epoch": 0.9042421585054573,
      "grad_norm": 0.8659823669810899,
      "learning_rate": 2.3833250268282705e-07,
      "loss": 0.3619,
      "step": 14436
    },
    {
      "epoch": 0.9043047965047997,
      "grad_norm": 0.8798139203433948,
      "learning_rate": 2.3802314938511817e-07,
      "loss": 0.3961,
      "step": 14437
    },
    {
      "epoch": 0.9043674345041419,
      "grad_norm": 0.8696691348029164,
      "learning_rate": 2.3771399209125968e-07,
      "loss": 0.4133,
      "step": 14438
    },
    {
      "epoch": 0.9044300725034843,
      "grad_norm": 0.885283877231372,
      "learning_rate": 2.3740503081397693e-07,
      "loss": 0.3783,
      "step": 14439
    },
    {
      "epoch": 0.9044927105028265,
      "grad_norm": 0.9248617635464865,
      "learning_rate": 2.3709626556598752e-07,
      "loss": 0.4145,
      "step": 14440
    },
    {
      "epoch": 0.9045553485021688,
      "grad_norm": 0.9204900670019661,
      "learning_rate": 2.3678769635999909e-07,
      "loss": 0.4511,
      "step": 14441
    },
    {
      "epoch": 0.9046179865015112,
      "grad_norm": 0.835234136896742,
      "learning_rate": 2.3647932320871426e-07,
      "loss": 0.3832,
      "step": 14442
    },
    {
      "epoch": 0.9046806245008534,
      "grad_norm": 0.610582583527176,
      "learning_rate": 2.3617114612482394e-07,
      "loss": 0.4313,
      "step": 14443
    },
    {
      "epoch": 0.9047432625001958,
      "grad_norm": 0.8192334585012531,
      "learning_rate": 2.3586316512101416e-07,
      "loss": 0.3423,
      "step": 14444
    },
    {
      "epoch": 0.904805900499538,
      "grad_norm": 0.9610886973173783,
      "learning_rate": 2.3555538020996027e-07,
      "loss": 0.4279,
      "step": 14445
    },
    {
      "epoch": 0.9048685384988804,
      "grad_norm": 0.8353728850468612,
      "learning_rate": 2.3524779140433163e-07,
      "loss": 0.3574,
      "step": 14446
    },
    {
      "epoch": 0.9049311764982226,
      "grad_norm": 0.8605998246142004,
      "learning_rate": 2.3494039871678808e-07,
      "loss": 0.3904,
      "step": 14447
    },
    {
      "epoch": 0.904993814497565,
      "grad_norm": 0.8132921120647685,
      "learning_rate": 2.3463320215998176e-07,
      "loss": 0.3601,
      "step": 14448
    },
    {
      "epoch": 0.9050564524969072,
      "grad_norm": 0.8435621411626051,
      "learning_rate": 2.3432620174655752e-07,
      "loss": 0.354,
      "step": 14449
    },
    {
      "epoch": 0.9051190904962495,
      "grad_norm": 0.9197515517514555,
      "learning_rate": 2.3401939748915137e-07,
      "loss": 0.3946,
      "step": 14450
    },
    {
      "epoch": 0.9051817284955919,
      "grad_norm": 0.8891667347648184,
      "learning_rate": 2.3371278940039155e-07,
      "loss": 0.3786,
      "step": 14451
    },
    {
      "epoch": 0.9052443664949341,
      "grad_norm": 0.862105256016247,
      "learning_rate": 2.3340637749289796e-07,
      "loss": 0.3999,
      "step": 14452
    },
    {
      "epoch": 0.9053070044942765,
      "grad_norm": 0.8836864718129698,
      "learning_rate": 2.3310016177928274e-07,
      "loss": 0.3691,
      "step": 14453
    },
    {
      "epoch": 0.9053696424936187,
      "grad_norm": 0.8920959496937879,
      "learning_rate": 2.327941422721497e-07,
      "loss": 0.4249,
      "step": 14454
    },
    {
      "epoch": 0.9054322804929611,
      "grad_norm": 0.9171864767909105,
      "learning_rate": 2.3248831898409484e-07,
      "loss": 0.3998,
      "step": 14455
    },
    {
      "epoch": 0.9054949184923033,
      "grad_norm": 0.847121601618576,
      "learning_rate": 2.3218269192770537e-07,
      "loss": 0.3576,
      "step": 14456
    },
    {
      "epoch": 0.9055575564916457,
      "grad_norm": 0.8928182277097533,
      "learning_rate": 2.3187726111556063e-07,
      "loss": 0.3975,
      "step": 14457
    },
    {
      "epoch": 0.905620194490988,
      "grad_norm": 0.8739392872772436,
      "learning_rate": 2.3157202656023335e-07,
      "loss": 0.3536,
      "step": 14458
    },
    {
      "epoch": 0.9056828324903302,
      "grad_norm": 0.9051170366784979,
      "learning_rate": 2.3126698827428572e-07,
      "loss": 0.416,
      "step": 14459
    },
    {
      "epoch": 0.9057454704896726,
      "grad_norm": 0.7903708239522551,
      "learning_rate": 2.309621462702749e-07,
      "loss": 0.3588,
      "step": 14460
    },
    {
      "epoch": 0.9058081084890148,
      "grad_norm": 0.8500044732243955,
      "learning_rate": 2.3065750056074755e-07,
      "loss": 0.3877,
      "step": 14461
    },
    {
      "epoch": 0.9058707464883572,
      "grad_norm": 0.9265837096061915,
      "learning_rate": 2.3035305115824248e-07,
      "loss": 0.3815,
      "step": 14462
    },
    {
      "epoch": 0.9059333844876994,
      "grad_norm": 0.8410374295688078,
      "learning_rate": 2.3004879807529078e-07,
      "loss": 0.3746,
      "step": 14463
    },
    {
      "epoch": 0.9059960224870418,
      "grad_norm": 0.9595233541848897,
      "learning_rate": 2.2974474132441638e-07,
      "loss": 0.3703,
      "step": 14464
    },
    {
      "epoch": 0.9060586604863841,
      "grad_norm": 0.8830118589884784,
      "learning_rate": 2.294408809181331e-07,
      "loss": 0.372,
      "step": 14465
    },
    {
      "epoch": 0.9061212984857263,
      "grad_norm": 0.8289650484912829,
      "learning_rate": 2.2913721686894875e-07,
      "loss": 0.3637,
      "step": 14466
    },
    {
      "epoch": 0.9061839364850687,
      "grad_norm": 0.902207428254577,
      "learning_rate": 2.2883374918936275e-07,
      "loss": 0.4116,
      "step": 14467
    },
    {
      "epoch": 0.9062465744844109,
      "grad_norm": 0.9295550773870617,
      "learning_rate": 2.28530477891864e-07,
      "loss": 0.4029,
      "step": 14468
    },
    {
      "epoch": 0.9063092124837533,
      "grad_norm": 0.8569427925556324,
      "learning_rate": 2.2822740298893754e-07,
      "loss": 0.377,
      "step": 14469
    },
    {
      "epoch": 0.9063718504830955,
      "grad_norm": 0.886597382092008,
      "learning_rate": 2.2792452449305668e-07,
      "loss": 0.3625,
      "step": 14470
    },
    {
      "epoch": 0.9064344884824379,
      "grad_norm": 0.9244995358395378,
      "learning_rate": 2.2762184241668818e-07,
      "loss": 0.3792,
      "step": 14471
    },
    {
      "epoch": 0.9064971264817802,
      "grad_norm": 0.8506875288103419,
      "learning_rate": 2.2731935677229033e-07,
      "loss": 0.3608,
      "step": 14472
    },
    {
      "epoch": 0.9065597644811225,
      "grad_norm": 0.839552096796751,
      "learning_rate": 2.2701706757231378e-07,
      "loss": 0.373,
      "step": 14473
    },
    {
      "epoch": 0.9066224024804648,
      "grad_norm": 0.9025039781100302,
      "learning_rate": 2.267149748291997e-07,
      "loss": 0.3723,
      "step": 14474
    },
    {
      "epoch": 0.906685040479807,
      "grad_norm": 0.7967677359527542,
      "learning_rate": 2.264130785553842e-07,
      "loss": 0.3693,
      "step": 14475
    },
    {
      "epoch": 0.9067476784791494,
      "grad_norm": 0.889209257871171,
      "learning_rate": 2.261113787632918e-07,
      "loss": 0.3823,
      "step": 14476
    },
    {
      "epoch": 0.9068103164784916,
      "grad_norm": 0.8992084273983385,
      "learning_rate": 2.2580987546534207e-07,
      "loss": 0.3785,
      "step": 14477
    },
    {
      "epoch": 0.906872954477834,
      "grad_norm": 0.9180612474431262,
      "learning_rate": 2.2550856867394278e-07,
      "loss": 0.4192,
      "step": 14478
    },
    {
      "epoch": 0.9069355924771763,
      "grad_norm": 0.9169431105324396,
      "learning_rate": 2.2520745840149738e-07,
      "loss": 0.3898,
      "step": 14479
    },
    {
      "epoch": 0.9069982304765186,
      "grad_norm": 0.8271375012003271,
      "learning_rate": 2.2490654466039986e-07,
      "loss": 0.3717,
      "step": 14480
    },
    {
      "epoch": 0.9070608684758609,
      "grad_norm": 0.5951050444114396,
      "learning_rate": 2.2460582746303472e-07,
      "loss": 0.4636,
      "step": 14481
    },
    {
      "epoch": 0.9071235064752032,
      "grad_norm": 0.8218655199972478,
      "learning_rate": 2.2430530682178043e-07,
      "loss": 0.3322,
      "step": 14482
    },
    {
      "epoch": 0.9071861444745455,
      "grad_norm": 2.704555899610304,
      "learning_rate": 2.240049827490054e-07,
      "loss": 0.4291,
      "step": 14483
    },
    {
      "epoch": 0.9072487824738877,
      "grad_norm": 0.9365497208953425,
      "learning_rate": 2.2370485525707197e-07,
      "loss": 0.3821,
      "step": 14484
    },
    {
      "epoch": 0.9073114204732301,
      "grad_norm": 0.834089950420208,
      "learning_rate": 2.234049243583336e-07,
      "loss": 0.3638,
      "step": 14485
    },
    {
      "epoch": 0.9073740584725724,
      "grad_norm": 0.9065284463869558,
      "learning_rate": 2.2310519006513542e-07,
      "loss": 0.3645,
      "step": 14486
    },
    {
      "epoch": 0.9074366964719147,
      "grad_norm": 0.845364635226946,
      "learning_rate": 2.228056523898131e-07,
      "loss": 0.3629,
      "step": 14487
    },
    {
      "epoch": 0.907499334471257,
      "grad_norm": 0.8925117627911082,
      "learning_rate": 2.2250631134469792e-07,
      "loss": 0.4067,
      "step": 14488
    },
    {
      "epoch": 0.9075619724705993,
      "grad_norm": 0.9958688989535294,
      "learning_rate": 2.2220716694210942e-07,
      "loss": 0.3912,
      "step": 14489
    },
    {
      "epoch": 0.9076246104699416,
      "grad_norm": 0.8268102721594539,
      "learning_rate": 2.2190821919436056e-07,
      "loss": 0.3593,
      "step": 14490
    },
    {
      "epoch": 0.9076872484692838,
      "grad_norm": 0.911712917894971,
      "learning_rate": 2.2160946811375596e-07,
      "loss": 0.4143,
      "step": 14491
    },
    {
      "epoch": 0.9077498864686262,
      "grad_norm": 0.8334587493760908,
      "learning_rate": 2.2131091371259295e-07,
      "loss": 0.3792,
      "step": 14492
    },
    {
      "epoch": 0.9078125244679685,
      "grad_norm": 0.8448920211394677,
      "learning_rate": 2.2101255600316007e-07,
      "loss": 0.3878,
      "step": 14493
    },
    {
      "epoch": 0.9078751624673108,
      "grad_norm": 0.6182964019852157,
      "learning_rate": 2.2071439499773694e-07,
      "loss": 0.4557,
      "step": 14494
    },
    {
      "epoch": 0.9079378004666531,
      "grad_norm": 0.8754175588265944,
      "learning_rate": 2.2041643070859653e-07,
      "loss": 0.4313,
      "step": 14495
    },
    {
      "epoch": 0.9080004384659954,
      "grad_norm": 0.8996053292381104,
      "learning_rate": 2.2011866314800234e-07,
      "loss": 0.3983,
      "step": 14496
    },
    {
      "epoch": 0.9080630764653377,
      "grad_norm": 0.9542122239884184,
      "learning_rate": 2.198210923282118e-07,
      "loss": 0.4016,
      "step": 14497
    },
    {
      "epoch": 0.90812571446468,
      "grad_norm": 0.8963903683169457,
      "learning_rate": 2.1952371826147233e-07,
      "loss": 0.3935,
      "step": 14498
    },
    {
      "epoch": 0.9081883524640223,
      "grad_norm": 0.9305797885727992,
      "learning_rate": 2.1922654096002416e-07,
      "loss": 0.4286,
      "step": 14499
    },
    {
      "epoch": 0.9082509904633645,
      "grad_norm": 0.8976302851539563,
      "learning_rate": 2.1892956043609804e-07,
      "loss": 0.3982,
      "step": 14500
    },
    {
      "epoch": 0.9083136284627069,
      "grad_norm": 0.7840204598265965,
      "learning_rate": 2.186327767019192e-07,
      "loss": 0.3758,
      "step": 14501
    },
    {
      "epoch": 0.9083762664620492,
      "grad_norm": 0.7946842154361382,
      "learning_rate": 2.1833618976970284e-07,
      "loss": 0.3407,
      "step": 14502
    },
    {
      "epoch": 0.9084389044613915,
      "grad_norm": 0.8662399558515259,
      "learning_rate": 2.1803979965165646e-07,
      "loss": 0.3839,
      "step": 14503
    },
    {
      "epoch": 0.9085015424607338,
      "grad_norm": 0.8784802830767026,
      "learning_rate": 2.177436063599797e-07,
      "loss": 0.4042,
      "step": 14504
    },
    {
      "epoch": 0.9085641804600761,
      "grad_norm": 0.8631294216620459,
      "learning_rate": 2.1744760990686287e-07,
      "loss": 0.3569,
      "step": 14505
    },
    {
      "epoch": 0.9086268184594184,
      "grad_norm": 0.8819106150962902,
      "learning_rate": 2.1715181030449062e-07,
      "loss": 0.3829,
      "step": 14506
    },
    {
      "epoch": 0.9086894564587608,
      "grad_norm": 0.8902747930937875,
      "learning_rate": 2.1685620756503766e-07,
      "loss": 0.3732,
      "step": 14507
    },
    {
      "epoch": 0.908752094458103,
      "grad_norm": 0.899150800174386,
      "learning_rate": 2.1656080170067094e-07,
      "loss": 0.4179,
      "step": 14508
    },
    {
      "epoch": 0.9088147324574453,
      "grad_norm": 0.8036554442096415,
      "learning_rate": 2.1626559272354908e-07,
      "loss": 0.3514,
      "step": 14509
    },
    {
      "epoch": 0.9088773704567876,
      "grad_norm": 0.8831437374791827,
      "learning_rate": 2.1597058064582398e-07,
      "loss": 0.3767,
      "step": 14510
    },
    {
      "epoch": 0.9089400084561299,
      "grad_norm": 0.8124511399584443,
      "learning_rate": 2.156757654796371e-07,
      "loss": 0.3401,
      "step": 14511
    },
    {
      "epoch": 0.9090026464554722,
      "grad_norm": 0.6196710157677477,
      "learning_rate": 2.1538114723712422e-07,
      "loss": 0.4505,
      "step": 14512
    },
    {
      "epoch": 0.9090652844548145,
      "grad_norm": 0.9215828705768413,
      "learning_rate": 2.150867259304118e-07,
      "loss": 0.4375,
      "step": 14513
    },
    {
      "epoch": 0.9091279224541569,
      "grad_norm": 0.8101617248335953,
      "learning_rate": 2.1479250157161735e-07,
      "loss": 0.34,
      "step": 14514
    },
    {
      "epoch": 0.9091905604534991,
      "grad_norm": 0.8623013427544753,
      "learning_rate": 2.144984741728512e-07,
      "loss": 0.3252,
      "step": 14515
    },
    {
      "epoch": 0.9092531984528415,
      "grad_norm": 0.8268906408811817,
      "learning_rate": 2.1420464374621642e-07,
      "loss": 0.3766,
      "step": 14516
    },
    {
      "epoch": 0.9093158364521837,
      "grad_norm": 0.8872473569589333,
      "learning_rate": 2.1391101030380668e-07,
      "loss": 0.3596,
      "step": 14517
    },
    {
      "epoch": 0.909378474451526,
      "grad_norm": 0.9122549876090975,
      "learning_rate": 2.1361757385770843e-07,
      "loss": 0.4184,
      "step": 14518
    },
    {
      "epoch": 0.9094411124508683,
      "grad_norm": 0.870424847267297,
      "learning_rate": 2.1332433441999923e-07,
      "loss": 0.3954,
      "step": 14519
    },
    {
      "epoch": 0.9095037504502106,
      "grad_norm": 0.834374117451356,
      "learning_rate": 2.1303129200274886e-07,
      "loss": 0.3725,
      "step": 14520
    },
    {
      "epoch": 0.909566388449553,
      "grad_norm": 0.8284406014822289,
      "learning_rate": 2.1273844661801878e-07,
      "loss": 0.3879,
      "step": 14521
    },
    {
      "epoch": 0.9096290264488952,
      "grad_norm": 0.5969791788161002,
      "learning_rate": 2.1244579827786326e-07,
      "loss": 0.4292,
      "step": 14522
    },
    {
      "epoch": 0.9096916644482376,
      "grad_norm": 0.5629512855859092,
      "learning_rate": 2.121533469943271e-07,
      "loss": 0.4378,
      "step": 14523
    },
    {
      "epoch": 0.9097543024475798,
      "grad_norm": 0.822374010894613,
      "learning_rate": 2.1186109277944732e-07,
      "loss": 0.3504,
      "step": 14524
    },
    {
      "epoch": 0.9098169404469221,
      "grad_norm": 0.8930091055400947,
      "learning_rate": 2.1156903564525432e-07,
      "loss": 0.3761,
      "step": 14525
    },
    {
      "epoch": 0.9098795784462644,
      "grad_norm": 0.8333905139450267,
      "learning_rate": 2.1127717560376794e-07,
      "loss": 0.3563,
      "step": 14526
    },
    {
      "epoch": 0.9099422164456067,
      "grad_norm": 0.9192758876531361,
      "learning_rate": 2.109855126670024e-07,
      "loss": 0.43,
      "step": 14527
    },
    {
      "epoch": 0.910004854444949,
      "grad_norm": 0.9638136796342194,
      "learning_rate": 2.1069404684696148e-07,
      "loss": 0.3936,
      "step": 14528
    },
    {
      "epoch": 0.9100674924442913,
      "grad_norm": 0.7845641270774674,
      "learning_rate": 2.1040277815564336e-07,
      "loss": 0.3731,
      "step": 14529
    },
    {
      "epoch": 0.9101301304436337,
      "grad_norm": 0.7751697757840038,
      "learning_rate": 2.1011170660503565e-07,
      "loss": 0.3296,
      "step": 14530
    },
    {
      "epoch": 0.9101927684429759,
      "grad_norm": 0.8462184633834104,
      "learning_rate": 2.098208322071188e-07,
      "loss": 0.3655,
      "step": 14531
    },
    {
      "epoch": 0.9102554064423183,
      "grad_norm": 0.8193606633065464,
      "learning_rate": 2.0953015497386597e-07,
      "loss": 0.3068,
      "step": 14532
    },
    {
      "epoch": 0.9103180444416605,
      "grad_norm": 0.833623164964156,
      "learning_rate": 2.092396749172404e-07,
      "loss": 0.3507,
      "step": 14533
    },
    {
      "epoch": 0.9103806824410028,
      "grad_norm": 0.843677412879372,
      "learning_rate": 2.0894939204919916e-07,
      "loss": 0.399,
      "step": 14534
    },
    {
      "epoch": 0.9104433204403451,
      "grad_norm": 0.9164679358555833,
      "learning_rate": 2.0865930638169053e-07,
      "loss": 0.3845,
      "step": 14535
    },
    {
      "epoch": 0.9105059584396874,
      "grad_norm": 0.8918338834525462,
      "learning_rate": 2.0836941792665432e-07,
      "loss": 0.4017,
      "step": 14536
    },
    {
      "epoch": 0.9105685964390298,
      "grad_norm": 0.832323405719235,
      "learning_rate": 2.0807972669602217e-07,
      "loss": 0.3889,
      "step": 14537
    },
    {
      "epoch": 0.910631234438372,
      "grad_norm": 0.8606160096063852,
      "learning_rate": 2.077902327017184e-07,
      "loss": 0.3694,
      "step": 14538
    },
    {
      "epoch": 0.9106938724377144,
      "grad_norm": 0.8965167049908673,
      "learning_rate": 2.0750093595565735e-07,
      "loss": 0.3929,
      "step": 14539
    },
    {
      "epoch": 0.9107565104370566,
      "grad_norm": 0.6184377201607902,
      "learning_rate": 2.0721183646974785e-07,
      "loss": 0.4522,
      "step": 14540
    },
    {
      "epoch": 0.910819148436399,
      "grad_norm": 0.8609456763313632,
      "learning_rate": 2.069229342558887e-07,
      "loss": 0.3912,
      "step": 14541
    },
    {
      "epoch": 0.9108817864357412,
      "grad_norm": 0.8419065430874886,
      "learning_rate": 2.0663422932597044e-07,
      "loss": 0.3769,
      "step": 14542
    },
    {
      "epoch": 0.9109444244350835,
      "grad_norm": 0.9072836301890276,
      "learning_rate": 2.0634572169187682e-07,
      "loss": 0.4335,
      "step": 14543
    },
    {
      "epoch": 0.9110070624344259,
      "grad_norm": 0.6280693700625949,
      "learning_rate": 2.0605741136548396e-07,
      "loss": 0.4115,
      "step": 14544
    },
    {
      "epoch": 0.9110697004337681,
      "grad_norm": 0.6510622909595432,
      "learning_rate": 2.0576929835865732e-07,
      "loss": 0.4265,
      "step": 14545
    },
    {
      "epoch": 0.9111323384331105,
      "grad_norm": 0.9589818628430679,
      "learning_rate": 2.054813826832569e-07,
      "loss": 0.4292,
      "step": 14546
    },
    {
      "epoch": 0.9111949764324527,
      "grad_norm": 0.8697673958160035,
      "learning_rate": 2.0519366435113208e-07,
      "loss": 0.3421,
      "step": 14547
    },
    {
      "epoch": 0.9112576144317951,
      "grad_norm": 0.8690575246531955,
      "learning_rate": 2.0490614337412618e-07,
      "loss": 0.4229,
      "step": 14548
    },
    {
      "epoch": 0.9113202524311373,
      "grad_norm": 0.8832377581795572,
      "learning_rate": 2.046188197640736e-07,
      "loss": 0.3587,
      "step": 14549
    },
    {
      "epoch": 0.9113828904304796,
      "grad_norm": 0.8995928388056197,
      "learning_rate": 2.043316935327999e-07,
      "loss": 0.388,
      "step": 14550
    },
    {
      "epoch": 0.911445528429822,
      "grad_norm": 0.8617696179619232,
      "learning_rate": 2.0404476469212453e-07,
      "loss": 0.3613,
      "step": 14551
    },
    {
      "epoch": 0.9115081664291642,
      "grad_norm": 0.8697711551147823,
      "learning_rate": 2.0375803325385578e-07,
      "loss": 0.4149,
      "step": 14552
    },
    {
      "epoch": 0.9115708044285066,
      "grad_norm": 0.9348521317927773,
      "learning_rate": 2.034714992297976e-07,
      "loss": 0.4202,
      "step": 14553
    },
    {
      "epoch": 0.9116334424278488,
      "grad_norm": 0.9044066204784527,
      "learning_rate": 2.031851626317427e-07,
      "loss": 0.3445,
      "step": 14554
    },
    {
      "epoch": 0.9116960804271912,
      "grad_norm": 0.9336030872966831,
      "learning_rate": 2.0289902347147672e-07,
      "loss": 0.4229,
      "step": 14555
    },
    {
      "epoch": 0.9117587184265334,
      "grad_norm": 0.7917761028379916,
      "learning_rate": 2.0261308176077743e-07,
      "loss": 0.345,
      "step": 14556
    },
    {
      "epoch": 0.9118213564258758,
      "grad_norm": 0.8489674624866208,
      "learning_rate": 2.0232733751141376e-07,
      "loss": 0.3807,
      "step": 14557
    },
    {
      "epoch": 0.9118839944252181,
      "grad_norm": 1.2952041033619557,
      "learning_rate": 2.0204179073514795e-07,
      "loss": 0.4087,
      "step": 14558
    },
    {
      "epoch": 0.9119466324245603,
      "grad_norm": 0.8494492475237228,
      "learning_rate": 2.017564414437312e-07,
      "loss": 0.414,
      "step": 14559
    },
    {
      "epoch": 0.9120092704239027,
      "grad_norm": 0.8514332301504445,
      "learning_rate": 2.0147128964891126e-07,
      "loss": 0.3797,
      "step": 14560
    },
    {
      "epoch": 0.9120719084232449,
      "grad_norm": 0.8575108186935899,
      "learning_rate": 2.0118633536242272e-07,
      "loss": 0.3692,
      "step": 14561
    },
    {
      "epoch": 0.9121345464225873,
      "grad_norm": 0.8971263742967712,
      "learning_rate": 2.0090157859599614e-07,
      "loss": 0.3958,
      "step": 14562
    },
    {
      "epoch": 0.9121971844219295,
      "grad_norm": 0.8852048912540622,
      "learning_rate": 2.0061701936135102e-07,
      "loss": 0.3729,
      "step": 14563
    },
    {
      "epoch": 0.9122598224212719,
      "grad_norm": 0.8091321452144214,
      "learning_rate": 2.0033265767020026e-07,
      "loss": 0.359,
      "step": 14564
    },
    {
      "epoch": 0.9123224604206142,
      "grad_norm": 0.8192105544946708,
      "learning_rate": 2.0004849353424838e-07,
      "loss": 0.3858,
      "step": 14565
    },
    {
      "epoch": 0.9123850984199565,
      "grad_norm": 0.9208487360264388,
      "learning_rate": 1.9976452696519155e-07,
      "loss": 0.4269,
      "step": 14566
    },
    {
      "epoch": 0.9124477364192988,
      "grad_norm": 0.8454606741126122,
      "learning_rate": 1.9948075797471766e-07,
      "loss": 0.3661,
      "step": 14567
    },
    {
      "epoch": 0.912510374418641,
      "grad_norm": 0.7926150349362686,
      "learning_rate": 1.991971865745057e-07,
      "loss": 0.3617,
      "step": 14568
    },
    {
      "epoch": 0.9125730124179834,
      "grad_norm": 0.9233191477009968,
      "learning_rate": 1.9891381277622968e-07,
      "loss": 0.3615,
      "step": 14569
    },
    {
      "epoch": 0.9126356504173256,
      "grad_norm": 0.7796172021920009,
      "learning_rate": 1.986306365915519e-07,
      "loss": 0.3642,
      "step": 14570
    },
    {
      "epoch": 0.912698288416668,
      "grad_norm": 0.9302718848533903,
      "learning_rate": 1.9834765803212807e-07,
      "loss": 0.3886,
      "step": 14571
    },
    {
      "epoch": 0.9127609264160103,
      "grad_norm": 0.8437031841415149,
      "learning_rate": 1.9806487710960664e-07,
      "loss": 0.3943,
      "step": 14572
    },
    {
      "epoch": 0.9128235644153526,
      "grad_norm": 0.8238586193302636,
      "learning_rate": 1.9778229383562608e-07,
      "loss": 0.3639,
      "step": 14573
    },
    {
      "epoch": 0.9128862024146949,
      "grad_norm": 0.6012005467216828,
      "learning_rate": 1.9749990822181707e-07,
      "loss": 0.443,
      "step": 14574
    },
    {
      "epoch": 0.9129488404140371,
      "grad_norm": 0.9049530853214635,
      "learning_rate": 1.9721772027980368e-07,
      "loss": 0.4005,
      "step": 14575
    },
    {
      "epoch": 0.9130114784133795,
      "grad_norm": 0.8884528604523468,
      "learning_rate": 1.9693573002119937e-07,
      "loss": 0.3957,
      "step": 14576
    },
    {
      "epoch": 0.9130741164127217,
      "grad_norm": 0.8316383700627232,
      "learning_rate": 1.966539374576132e-07,
      "loss": 0.3998,
      "step": 14577
    },
    {
      "epoch": 0.9131367544120641,
      "grad_norm": 0.8446552122103382,
      "learning_rate": 1.9637234260064197e-07,
      "loss": 0.3993,
      "step": 14578
    },
    {
      "epoch": 0.9131993924114064,
      "grad_norm": 0.823613454066151,
      "learning_rate": 1.9609094546187647e-07,
      "loss": 0.3738,
      "step": 14579
    },
    {
      "epoch": 0.9132620304107487,
      "grad_norm": 0.8448139865984626,
      "learning_rate": 1.958097460528996e-07,
      "loss": 0.3841,
      "step": 14580
    },
    {
      "epoch": 0.913324668410091,
      "grad_norm": 0.8742507250744033,
      "learning_rate": 1.955287443852849e-07,
      "loss": 0.3809,
      "step": 14581
    },
    {
      "epoch": 0.9133873064094333,
      "grad_norm": 0.8457330393278842,
      "learning_rate": 1.952479404705998e-07,
      "loss": 0.3741,
      "step": 14582
    },
    {
      "epoch": 0.9134499444087756,
      "grad_norm": 0.9210817729944507,
      "learning_rate": 1.949673343204006e-07,
      "loss": 0.37,
      "step": 14583
    },
    {
      "epoch": 0.9135125824081178,
      "grad_norm": 0.8044618509076671,
      "learning_rate": 1.9468692594623805e-07,
      "loss": 0.3389,
      "step": 14584
    },
    {
      "epoch": 0.9135752204074602,
      "grad_norm": 0.9182852108223399,
      "learning_rate": 1.9440671535965239e-07,
      "loss": 0.4244,
      "step": 14585
    },
    {
      "epoch": 0.9136378584068024,
      "grad_norm": 0.8561050781748168,
      "learning_rate": 1.9412670257217936e-07,
      "loss": 0.3705,
      "step": 14586
    },
    {
      "epoch": 0.9137004964061448,
      "grad_norm": 0.9158719573305588,
      "learning_rate": 1.9384688759534363e-07,
      "loss": 0.4253,
      "step": 14587
    },
    {
      "epoch": 0.9137631344054871,
      "grad_norm": 0.8078956568747874,
      "learning_rate": 1.9356727044066103e-07,
      "loss": 0.3675,
      "step": 14588
    },
    {
      "epoch": 0.9138257724048294,
      "grad_norm": 0.8358340722977022,
      "learning_rate": 1.932878511196423e-07,
      "loss": 0.3763,
      "step": 14589
    },
    {
      "epoch": 0.9138884104041717,
      "grad_norm": 0.8142360198572128,
      "learning_rate": 1.9300862964378774e-07,
      "loss": 0.4056,
      "step": 14590
    },
    {
      "epoch": 0.913951048403514,
      "grad_norm": 0.6251688781410166,
      "learning_rate": 1.9272960602459035e-07,
      "loss": 0.4482,
      "step": 14591
    },
    {
      "epoch": 0.9140136864028563,
      "grad_norm": 0.9019783659831833,
      "learning_rate": 1.9245078027353426e-07,
      "loss": 0.3971,
      "step": 14592
    },
    {
      "epoch": 0.9140763244021985,
      "grad_norm": 0.9385954198260807,
      "learning_rate": 1.9217215240209697e-07,
      "loss": 0.4379,
      "step": 14593
    },
    {
      "epoch": 0.9141389624015409,
      "grad_norm": 0.799174369169606,
      "learning_rate": 1.9189372242174542e-07,
      "loss": 0.3479,
      "step": 14594
    },
    {
      "epoch": 0.9142016004008832,
      "grad_norm": 0.8477388708884852,
      "learning_rate": 1.91615490343941e-07,
      "loss": 0.4093,
      "step": 14595
    },
    {
      "epoch": 0.9142642384002255,
      "grad_norm": 1.0180491024602643,
      "learning_rate": 1.9133745618013567e-07,
      "loss": 0.4367,
      "step": 14596
    },
    {
      "epoch": 0.9143268763995678,
      "grad_norm": 0.876495455857534,
      "learning_rate": 1.9105961994177246e-07,
      "loss": 0.3915,
      "step": 14597
    },
    {
      "epoch": 0.9143895143989101,
      "grad_norm": 0.9067958793249522,
      "learning_rate": 1.9078198164028894e-07,
      "loss": 0.4106,
      "step": 14598
    },
    {
      "epoch": 0.9144521523982524,
      "grad_norm": 0.8756331128699929,
      "learning_rate": 1.9050454128711148e-07,
      "loss": 0.3874,
      "step": 14599
    },
    {
      "epoch": 0.9145147903975946,
      "grad_norm": 0.9129895395573933,
      "learning_rate": 1.9022729889365933e-07,
      "loss": 0.425,
      "step": 14600
    },
    {
      "epoch": 0.914577428396937,
      "grad_norm": 0.8349924471808255,
      "learning_rate": 1.8995025447134496e-07,
      "loss": 0.3732,
      "step": 14601
    },
    {
      "epoch": 0.9146400663962793,
      "grad_norm": 0.9229007484398715,
      "learning_rate": 1.8967340803156985e-07,
      "loss": 0.4004,
      "step": 14602
    },
    {
      "epoch": 0.9147027043956216,
      "grad_norm": 0.8457595880349401,
      "learning_rate": 1.89396759585731e-07,
      "loss": 0.3228,
      "step": 14603
    },
    {
      "epoch": 0.9147653423949639,
      "grad_norm": 0.8912367320006188,
      "learning_rate": 1.891203091452143e-07,
      "loss": 0.3613,
      "step": 14604
    },
    {
      "epoch": 0.9148279803943062,
      "grad_norm": 0.9113192261735382,
      "learning_rate": 1.8884405672139838e-07,
      "loss": 0.3984,
      "step": 14605
    },
    {
      "epoch": 0.9148906183936485,
      "grad_norm": 0.9363902083437875,
      "learning_rate": 1.885680023256542e-07,
      "loss": 0.3652,
      "step": 14606
    },
    {
      "epoch": 0.9149532563929909,
      "grad_norm": 0.8392296020760586,
      "learning_rate": 1.8829214596934376e-07,
      "loss": 0.3635,
      "step": 14607
    },
    {
      "epoch": 0.9150158943923331,
      "grad_norm": 0.8469015276185888,
      "learning_rate": 1.8801648766382187e-07,
      "loss": 0.369,
      "step": 14608
    },
    {
      "epoch": 0.9150785323916754,
      "grad_norm": 0.8800578305901238,
      "learning_rate": 1.8774102742043497e-07,
      "loss": 0.3427,
      "step": 14609
    },
    {
      "epoch": 0.9151411703910177,
      "grad_norm": 0.8955785344250305,
      "learning_rate": 1.8746576525052017e-07,
      "loss": 0.4333,
      "step": 14610
    },
    {
      "epoch": 0.91520380839036,
      "grad_norm": 0.8939193355077921,
      "learning_rate": 1.8719070116540727e-07,
      "loss": 0.3907,
      "step": 14611
    },
    {
      "epoch": 0.9152664463897023,
      "grad_norm": 0.9607461838215237,
      "learning_rate": 1.8691583517641887e-07,
      "loss": 0.4228,
      "step": 14612
    },
    {
      "epoch": 0.9153290843890446,
      "grad_norm": 0.8333820953686849,
      "learning_rate": 1.8664116729486814e-07,
      "loss": 0.3848,
      "step": 14613
    },
    {
      "epoch": 0.915391722388387,
      "grad_norm": 0.9906622003629626,
      "learning_rate": 1.863666975320605e-07,
      "loss": 0.4518,
      "step": 14614
    },
    {
      "epoch": 0.9154543603877292,
      "grad_norm": 0.8704696777314882,
      "learning_rate": 1.86092425899293e-07,
      "loss": 0.352,
      "step": 14615
    },
    {
      "epoch": 0.9155169983870716,
      "grad_norm": 0.8739631300968358,
      "learning_rate": 1.858183524078544e-07,
      "loss": 0.3658,
      "step": 14616
    },
    {
      "epoch": 0.9155796363864138,
      "grad_norm": 0.9004253320365077,
      "learning_rate": 1.8554447706902622e-07,
      "loss": 0.3906,
      "step": 14617
    },
    {
      "epoch": 0.9156422743857561,
      "grad_norm": 0.9045262262915919,
      "learning_rate": 1.8527079989408113e-07,
      "loss": 0.4091,
      "step": 14618
    },
    {
      "epoch": 0.9157049123850984,
      "grad_norm": 0.6484553383263393,
      "learning_rate": 1.849973208942829e-07,
      "loss": 0.4296,
      "step": 14619
    },
    {
      "epoch": 0.9157675503844407,
      "grad_norm": 0.8527038114534683,
      "learning_rate": 1.8472404008088918e-07,
      "loss": 0.3878,
      "step": 14620
    },
    {
      "epoch": 0.915830188383783,
      "grad_norm": 0.8903022802402267,
      "learning_rate": 1.8445095746514763e-07,
      "loss": 0.3588,
      "step": 14621
    },
    {
      "epoch": 0.9158928263831253,
      "grad_norm": 0.9076804279059656,
      "learning_rate": 1.841780730582987e-07,
      "loss": 0.3805,
      "step": 14622
    },
    {
      "epoch": 0.9159554643824677,
      "grad_norm": 0.8538237020628726,
      "learning_rate": 1.83905386871574e-07,
      "loss": 0.4052,
      "step": 14623
    },
    {
      "epoch": 0.9160181023818099,
      "grad_norm": 0.8196870752205448,
      "learning_rate": 1.8363289891619728e-07,
      "loss": 0.3601,
      "step": 14624
    },
    {
      "epoch": 0.9160807403811523,
      "grad_norm": 0.8366849545131746,
      "learning_rate": 1.8336060920338405e-07,
      "loss": 0.3707,
      "step": 14625
    },
    {
      "epoch": 0.9161433783804945,
      "grad_norm": 0.8802054485865731,
      "learning_rate": 1.8308851774434312e-07,
      "loss": 0.4103,
      "step": 14626
    },
    {
      "epoch": 0.9162060163798368,
      "grad_norm": 0.8682617357088023,
      "learning_rate": 1.8281662455027215e-07,
      "loss": 0.4059,
      "step": 14627
    },
    {
      "epoch": 0.9162686543791791,
      "grad_norm": 0.8952820263376404,
      "learning_rate": 1.825449296323628e-07,
      "loss": 0.3911,
      "step": 14628
    },
    {
      "epoch": 0.9163312923785214,
      "grad_norm": 0.8673713012727365,
      "learning_rate": 1.822734330017989e-07,
      "loss": 0.3861,
      "step": 14629
    },
    {
      "epoch": 0.9163939303778638,
      "grad_norm": 0.832598636140544,
      "learning_rate": 1.820021346697548e-07,
      "loss": 0.3702,
      "step": 14630
    },
    {
      "epoch": 0.916456568377206,
      "grad_norm": 0.8365237659581025,
      "learning_rate": 1.8173103464739717e-07,
      "loss": 0.383,
      "step": 14631
    },
    {
      "epoch": 0.9165192063765484,
      "grad_norm": 0.8410069306307457,
      "learning_rate": 1.8146013294588427e-07,
      "loss": 0.3567,
      "step": 14632
    },
    {
      "epoch": 0.9165818443758906,
      "grad_norm": 0.8709949044719849,
      "learning_rate": 1.8118942957636666e-07,
      "loss": 0.3859,
      "step": 14633
    },
    {
      "epoch": 0.9166444823752329,
      "grad_norm": 0.8317712005163819,
      "learning_rate": 1.8091892454998595e-07,
      "loss": 0.3568,
      "step": 14634
    },
    {
      "epoch": 0.9167071203745752,
      "grad_norm": 0.8748629190555344,
      "learning_rate": 1.8064861787787768e-07,
      "loss": 0.409,
      "step": 14635
    },
    {
      "epoch": 0.9167697583739175,
      "grad_norm": 0.8306565486180815,
      "learning_rate": 1.803785095711663e-07,
      "loss": 0.3652,
      "step": 14636
    },
    {
      "epoch": 0.9168323963732599,
      "grad_norm": 0.8306150441006032,
      "learning_rate": 1.801085996409696e-07,
      "loss": 0.3689,
      "step": 14637
    },
    {
      "epoch": 0.9168950343726021,
      "grad_norm": 0.9222494702524204,
      "learning_rate": 1.7983888809839811e-07,
      "loss": 0.3856,
      "step": 14638
    },
    {
      "epoch": 0.9169576723719445,
      "grad_norm": 0.980174896540621,
      "learning_rate": 1.7956937495455295e-07,
      "loss": 0.425,
      "step": 14639
    },
    {
      "epoch": 0.9170203103712867,
      "grad_norm": 0.8364452845338674,
      "learning_rate": 1.7930006022052694e-07,
      "loss": 0.3725,
      "step": 14640
    },
    {
      "epoch": 0.9170829483706291,
      "grad_norm": 0.8243818859532487,
      "learning_rate": 1.7903094390740505e-07,
      "loss": 0.3907,
      "step": 14641
    },
    {
      "epoch": 0.9171455863699713,
      "grad_norm": 0.8714631617923024,
      "learning_rate": 1.7876202602626457e-07,
      "loss": 0.4006,
      "step": 14642
    },
    {
      "epoch": 0.9172082243693136,
      "grad_norm": 0.8608068354982259,
      "learning_rate": 1.7849330658817278e-07,
      "loss": 0.3799,
      "step": 14643
    },
    {
      "epoch": 0.917270862368656,
      "grad_norm": 0.9645747703626029,
      "learning_rate": 1.7822478560419242e-07,
      "loss": 0.3748,
      "step": 14644
    },
    {
      "epoch": 0.9173335003679982,
      "grad_norm": 0.8159120042182578,
      "learning_rate": 1.7795646308537417e-07,
      "loss": 0.3336,
      "step": 14645
    },
    {
      "epoch": 0.9173961383673406,
      "grad_norm": 0.9276269374408209,
      "learning_rate": 1.776883390427636e-07,
      "loss": 0.3641,
      "step": 14646
    },
    {
      "epoch": 0.9174587763666828,
      "grad_norm": 0.8982690821382054,
      "learning_rate": 1.7742041348739636e-07,
      "loss": 0.3753,
      "step": 14647
    },
    {
      "epoch": 0.9175214143660252,
      "grad_norm": 0.794047761322713,
      "learning_rate": 1.771526864302997e-07,
      "loss": 0.3298,
      "step": 14648
    },
    {
      "epoch": 0.9175840523653674,
      "grad_norm": 1.009714043720396,
      "learning_rate": 1.7688515788249316e-07,
      "loss": 0.394,
      "step": 14649
    },
    {
      "epoch": 0.9176466903647098,
      "grad_norm": 0.8224101186173519,
      "learning_rate": 1.7661782785498959e-07,
      "loss": 0.3397,
      "step": 14650
    },
    {
      "epoch": 0.9177093283640521,
      "grad_norm": 0.8243043312325466,
      "learning_rate": 1.7635069635879077e-07,
      "loss": 0.365,
      "step": 14651
    },
    {
      "epoch": 0.9177719663633943,
      "grad_norm": 0.8930657760382512,
      "learning_rate": 1.7608376340489286e-07,
      "loss": 0.3722,
      "step": 14652
    },
    {
      "epoch": 0.9178346043627367,
      "grad_norm": 0.8786578291218512,
      "learning_rate": 1.7581702900428211e-07,
      "loss": 0.4,
      "step": 14653
    },
    {
      "epoch": 0.9178972423620789,
      "grad_norm": 0.8830951325357927,
      "learning_rate": 1.7555049316793804e-07,
      "loss": 0.4192,
      "step": 14654
    },
    {
      "epoch": 0.9179598803614213,
      "grad_norm": 0.9441034903145736,
      "learning_rate": 1.7528415590683133e-07,
      "loss": 0.387,
      "step": 14655
    },
    {
      "epoch": 0.9180225183607635,
      "grad_norm": 0.8447530105785896,
      "learning_rate": 1.7501801723192437e-07,
      "loss": 0.381,
      "step": 14656
    },
    {
      "epoch": 0.9180851563601059,
      "grad_norm": 0.8696518803051337,
      "learning_rate": 1.7475207715417163e-07,
      "loss": 0.382,
      "step": 14657
    },
    {
      "epoch": 0.9181477943594482,
      "grad_norm": 0.886163386541185,
      "learning_rate": 1.744863356845189e-07,
      "loss": 0.4008,
      "step": 14658
    },
    {
      "epoch": 0.9182104323587904,
      "grad_norm": 0.8567787738911893,
      "learning_rate": 1.7422079283390404e-07,
      "loss": 0.4104,
      "step": 14659
    },
    {
      "epoch": 0.9182730703581328,
      "grad_norm": 0.871876646413901,
      "learning_rate": 1.7395544861325718e-07,
      "loss": 0.3673,
      "step": 14660
    },
    {
      "epoch": 0.918335708357475,
      "grad_norm": 0.8675901159912227,
      "learning_rate": 1.736903030334991e-07,
      "loss": 0.3796,
      "step": 14661
    },
    {
      "epoch": 0.9183983463568174,
      "grad_norm": 0.9079797443122608,
      "learning_rate": 1.7342535610554488e-07,
      "loss": 0.3857,
      "step": 14662
    },
    {
      "epoch": 0.9184609843561596,
      "grad_norm": 0.56651834085239,
      "learning_rate": 1.7316060784029808e-07,
      "loss": 0.4667,
      "step": 14663
    },
    {
      "epoch": 0.918523622355502,
      "grad_norm": 0.8935188995572623,
      "learning_rate": 1.7289605824865718e-07,
      "loss": 0.3888,
      "step": 14664
    },
    {
      "epoch": 0.9185862603548443,
      "grad_norm": 0.8257029455718833,
      "learning_rate": 1.7263170734151013e-07,
      "loss": 0.3782,
      "step": 14665
    },
    {
      "epoch": 0.9186488983541866,
      "grad_norm": 0.7739864835045562,
      "learning_rate": 1.7236755512973824e-07,
      "loss": 0.3326,
      "step": 14666
    },
    {
      "epoch": 0.9187115363535289,
      "grad_norm": 0.8132792668192733,
      "learning_rate": 1.7210360162421391e-07,
      "loss": 0.3326,
      "step": 14667
    },
    {
      "epoch": 0.9187741743528711,
      "grad_norm": 0.8268939323045604,
      "learning_rate": 1.718398468358018e-07,
      "loss": 0.3911,
      "step": 14668
    },
    {
      "epoch": 0.9188368123522135,
      "grad_norm": 0.9318172145930547,
      "learning_rate": 1.715762907753571e-07,
      "loss": 0.3942,
      "step": 14669
    },
    {
      "epoch": 0.9188994503515557,
      "grad_norm": 0.8719330939825359,
      "learning_rate": 1.7131293345372835e-07,
      "loss": 0.3962,
      "step": 14670
    },
    {
      "epoch": 0.9189620883508981,
      "grad_norm": 0.8554845256202654,
      "learning_rate": 1.7104977488175522e-07,
      "loss": 0.3796,
      "step": 14671
    },
    {
      "epoch": 0.9190247263502404,
      "grad_norm": 0.9468012877077068,
      "learning_rate": 1.7078681507027017e-07,
      "loss": 0.3929,
      "step": 14672
    },
    {
      "epoch": 0.9190873643495827,
      "grad_norm": 0.8927901505399598,
      "learning_rate": 1.705240540300962e-07,
      "loss": 0.3962,
      "step": 14673
    },
    {
      "epoch": 0.919150002348925,
      "grad_norm": 0.7493932371054397,
      "learning_rate": 1.702614917720491e-07,
      "loss": 0.3494,
      "step": 14674
    },
    {
      "epoch": 0.9192126403482673,
      "grad_norm": 0.903848323609536,
      "learning_rate": 1.6999912830693467e-07,
      "loss": 0.3721,
      "step": 14675
    },
    {
      "epoch": 0.9192752783476096,
      "grad_norm": 0.8664793441007657,
      "learning_rate": 1.6973696364555314e-07,
      "loss": 0.3833,
      "step": 14676
    },
    {
      "epoch": 0.9193379163469518,
      "grad_norm": 0.8628408126673278,
      "learning_rate": 1.6947499779869425e-07,
      "loss": 0.3756,
      "step": 14677
    },
    {
      "epoch": 0.9194005543462942,
      "grad_norm": 0.8049012490904607,
      "learning_rate": 1.69213230777141e-07,
      "loss": 0.3688,
      "step": 14678
    },
    {
      "epoch": 0.9194631923456364,
      "grad_norm": 0.9301200680437386,
      "learning_rate": 1.6895166259166817e-07,
      "loss": 0.3987,
      "step": 14679
    },
    {
      "epoch": 0.9195258303449788,
      "grad_norm": 0.9015276109132928,
      "learning_rate": 1.6869029325304097e-07,
      "loss": 0.3581,
      "step": 14680
    },
    {
      "epoch": 0.9195884683443211,
      "grad_norm": 0.5656968627631848,
      "learning_rate": 1.6842912277201862e-07,
      "loss": 0.4422,
      "step": 14681
    },
    {
      "epoch": 0.9196511063436634,
      "grad_norm": 0.8580661346064088,
      "learning_rate": 1.6816815115935026e-07,
      "loss": 0.3847,
      "step": 14682
    },
    {
      "epoch": 0.9197137443430057,
      "grad_norm": 0.5576955191482998,
      "learning_rate": 1.6790737842577786e-07,
      "loss": 0.4151,
      "step": 14683
    },
    {
      "epoch": 0.9197763823423479,
      "grad_norm": 0.8667029746845923,
      "learning_rate": 1.6764680458203453e-07,
      "loss": 0.3602,
      "step": 14684
    },
    {
      "epoch": 0.9198390203416903,
      "grad_norm": 0.8704208810388455,
      "learning_rate": 1.6738642963884555e-07,
      "loss": 0.3723,
      "step": 14685
    },
    {
      "epoch": 0.9199016583410325,
      "grad_norm": 0.8432333727805947,
      "learning_rate": 1.671262536069279e-07,
      "loss": 0.3531,
      "step": 14686
    },
    {
      "epoch": 0.9199642963403749,
      "grad_norm": 0.8010457956969766,
      "learning_rate": 1.6686627649699028e-07,
      "loss": 0.358,
      "step": 14687
    },
    {
      "epoch": 0.9200269343397172,
      "grad_norm": 0.8191031675857771,
      "learning_rate": 1.6660649831973464e-07,
      "loss": 0.3808,
      "step": 14688
    },
    {
      "epoch": 0.9200895723390595,
      "grad_norm": 0.8763759328058821,
      "learning_rate": 1.6634691908585187e-07,
      "loss": 0.3987,
      "step": 14689
    },
    {
      "epoch": 0.9201522103384018,
      "grad_norm": 0.8410040823690403,
      "learning_rate": 1.660875388060279e-07,
      "loss": 0.3872,
      "step": 14690
    },
    {
      "epoch": 0.9202148483377441,
      "grad_norm": 0.9005790939474518,
      "learning_rate": 1.658283574909375e-07,
      "loss": 0.371,
      "step": 14691
    },
    {
      "epoch": 0.9202774863370864,
      "grad_norm": 0.8203826705390574,
      "learning_rate": 1.6556937515124938e-07,
      "loss": 0.3866,
      "step": 14692
    },
    {
      "epoch": 0.9203401243364286,
      "grad_norm": 0.8319142132305005,
      "learning_rate": 1.6531059179762333e-07,
      "loss": 0.3574,
      "step": 14693
    },
    {
      "epoch": 0.920402762335771,
      "grad_norm": 0.9424611396075832,
      "learning_rate": 1.650520074407108e-07,
      "loss": 0.4208,
      "step": 14694
    },
    {
      "epoch": 0.9204654003351133,
      "grad_norm": 0.8674179255005526,
      "learning_rate": 1.6479362209115445e-07,
      "loss": 0.3321,
      "step": 14695
    },
    {
      "epoch": 0.9205280383344556,
      "grad_norm": 0.9531663922781224,
      "learning_rate": 1.645354357595902e-07,
      "loss": 0.404,
      "step": 14696
    },
    {
      "epoch": 0.9205906763337979,
      "grad_norm": 0.8986312321566398,
      "learning_rate": 1.6427744845664506e-07,
      "loss": 0.3844,
      "step": 14697
    },
    {
      "epoch": 0.9206533143331402,
      "grad_norm": 0.9317387134377791,
      "learning_rate": 1.640196601929378e-07,
      "loss": 0.4197,
      "step": 14698
    },
    {
      "epoch": 0.9207159523324825,
      "grad_norm": 0.9229465544677935,
      "learning_rate": 1.6376207097907826e-07,
      "loss": 0.4056,
      "step": 14699
    },
    {
      "epoch": 0.9207785903318249,
      "grad_norm": 0.8951608289670081,
      "learning_rate": 1.6350468082567018e-07,
      "loss": 0.4103,
      "step": 14700
    },
    {
      "epoch": 0.9208412283311671,
      "grad_norm": 0.8778509790032191,
      "learning_rate": 1.6324748974330728e-07,
      "loss": 0.3773,
      "step": 14701
    },
    {
      "epoch": 0.9209038663305094,
      "grad_norm": 0.8701376953624392,
      "learning_rate": 1.6299049774257503e-07,
      "loss": 0.3673,
      "step": 14702
    },
    {
      "epoch": 0.9209665043298517,
      "grad_norm": 0.8703177454258434,
      "learning_rate": 1.6273370483405216e-07,
      "loss": 0.3636,
      "step": 14703
    },
    {
      "epoch": 0.921029142329194,
      "grad_norm": 0.8033007174368356,
      "learning_rate": 1.6247711102830687e-07,
      "loss": 0.3976,
      "step": 14704
    },
    {
      "epoch": 0.9210917803285363,
      "grad_norm": 0.8630477042594399,
      "learning_rate": 1.622207163359024e-07,
      "loss": 0.4073,
      "step": 14705
    },
    {
      "epoch": 0.9211544183278786,
      "grad_norm": 0.8340873092347382,
      "learning_rate": 1.6196452076739143e-07,
      "loss": 0.3926,
      "step": 14706
    },
    {
      "epoch": 0.921217056327221,
      "grad_norm": 0.6381933796970108,
      "learning_rate": 1.617085243333183e-07,
      "loss": 0.4362,
      "step": 14707
    },
    {
      "epoch": 0.9212796943265632,
      "grad_norm": 0.862160194036841,
      "learning_rate": 1.614527270442201e-07,
      "loss": 0.3503,
      "step": 14708
    },
    {
      "epoch": 0.9213423323259055,
      "grad_norm": 0.8701286557549277,
      "learning_rate": 1.6119712891062566e-07,
      "loss": 0.3676,
      "step": 14709
    },
    {
      "epoch": 0.9214049703252478,
      "grad_norm": 0.8345434737563768,
      "learning_rate": 1.6094172994305603e-07,
      "loss": 0.3976,
      "step": 14710
    },
    {
      "epoch": 0.9214676083245901,
      "grad_norm": 0.5860917157941177,
      "learning_rate": 1.6068653015202328e-07,
      "loss": 0.4412,
      "step": 14711
    },
    {
      "epoch": 0.9215302463239324,
      "grad_norm": 0.9074843540978847,
      "learning_rate": 1.6043152954803077e-07,
      "loss": 0.392,
      "step": 14712
    },
    {
      "epoch": 0.9215928843232747,
      "grad_norm": 0.87021248833068,
      "learning_rate": 1.6017672814157392e-07,
      "loss": 0.379,
      "step": 14713
    },
    {
      "epoch": 0.921655522322617,
      "grad_norm": 0.8711845489167479,
      "learning_rate": 1.5992212594314215e-07,
      "loss": 0.3888,
      "step": 14714
    },
    {
      "epoch": 0.9217181603219593,
      "grad_norm": 0.8787048095293242,
      "learning_rate": 1.596677229632143e-07,
      "loss": 0.3949,
      "step": 14715
    },
    {
      "epoch": 0.9217807983213017,
      "grad_norm": 0.8727650926833521,
      "learning_rate": 1.594135192122609e-07,
      "loss": 0.383,
      "step": 14716
    },
    {
      "epoch": 0.9218434363206439,
      "grad_norm": 0.8456146637483302,
      "learning_rate": 1.591595147007452e-07,
      "loss": 0.3814,
      "step": 14717
    },
    {
      "epoch": 0.9219060743199862,
      "grad_norm": 0.8674350305693941,
      "learning_rate": 1.5890570943912275e-07,
      "loss": 0.4286,
      "step": 14718
    },
    {
      "epoch": 0.9219687123193285,
      "grad_norm": 0.9521371063110421,
      "learning_rate": 1.586521034378402e-07,
      "loss": 0.3912,
      "step": 14719
    },
    {
      "epoch": 0.9220313503186708,
      "grad_norm": 0.8941422319783079,
      "learning_rate": 1.5839869670733532e-07,
      "loss": 0.4253,
      "step": 14720
    },
    {
      "epoch": 0.9220939883180131,
      "grad_norm": 0.8971905389329219,
      "learning_rate": 1.5814548925803863e-07,
      "loss": 0.3545,
      "step": 14721
    },
    {
      "epoch": 0.9221566263173554,
      "grad_norm": 0.8870241398058054,
      "learning_rate": 1.5789248110037237e-07,
      "loss": 0.4117,
      "step": 14722
    },
    {
      "epoch": 0.9222192643166978,
      "grad_norm": 0.8066239040174703,
      "learning_rate": 1.576396722447504e-07,
      "loss": 0.3532,
      "step": 14723
    },
    {
      "epoch": 0.92228190231604,
      "grad_norm": 0.8358823176851482,
      "learning_rate": 1.5738706270157832e-07,
      "loss": 0.3882,
      "step": 14724
    },
    {
      "epoch": 0.9223445403153824,
      "grad_norm": 0.870347429869684,
      "learning_rate": 1.5713465248125338e-07,
      "loss": 0.3976,
      "step": 14725
    },
    {
      "epoch": 0.9224071783147246,
      "grad_norm": 0.9476285041159658,
      "learning_rate": 1.56882441594165e-07,
      "loss": 0.4198,
      "step": 14726
    },
    {
      "epoch": 0.9224698163140669,
      "grad_norm": 0.9701170447208158,
      "learning_rate": 1.5663043005069433e-07,
      "loss": 0.4657,
      "step": 14727
    },
    {
      "epoch": 0.9225324543134092,
      "grad_norm": 0.9323239847495163,
      "learning_rate": 1.5637861786121422e-07,
      "loss": 0.4054,
      "step": 14728
    },
    {
      "epoch": 0.9225950923127515,
      "grad_norm": 0.8411628675788348,
      "learning_rate": 1.561270050360897e-07,
      "loss": 0.371,
      "step": 14729
    },
    {
      "epoch": 0.9226577303120939,
      "grad_norm": 0.9323844332405403,
      "learning_rate": 1.5587559158567578e-07,
      "loss": 0.4455,
      "step": 14730
    },
    {
      "epoch": 0.9227203683114361,
      "grad_norm": 0.8627126415481418,
      "learning_rate": 1.55624377520322e-07,
      "loss": 0.353,
      "step": 14731
    },
    {
      "epoch": 0.9227830063107785,
      "grad_norm": 0.8798562791391245,
      "learning_rate": 1.5537336285036842e-07,
      "loss": 0.3872,
      "step": 14732
    },
    {
      "epoch": 0.9228456443101207,
      "grad_norm": 0.8669605469895654,
      "learning_rate": 1.551225475861462e-07,
      "loss": 0.3668,
      "step": 14733
    },
    {
      "epoch": 0.9229082823094631,
      "grad_norm": 0.8469651957055837,
      "learning_rate": 1.5487193173797875e-07,
      "loss": 0.3606,
      "step": 14734
    },
    {
      "epoch": 0.9229709203088053,
      "grad_norm": 0.8818273202018854,
      "learning_rate": 1.5462151531618174e-07,
      "loss": 0.377,
      "step": 14735
    },
    {
      "epoch": 0.9230335583081476,
      "grad_norm": 0.8747548761336114,
      "learning_rate": 1.5437129833106356e-07,
      "loss": 0.3772,
      "step": 14736
    },
    {
      "epoch": 0.92309619630749,
      "grad_norm": 0.866207698472162,
      "learning_rate": 1.5412128079292154e-07,
      "loss": 0.3824,
      "step": 14737
    },
    {
      "epoch": 0.9231588343068322,
      "grad_norm": 0.6501220761793048,
      "learning_rate": 1.538714627120469e-07,
      "loss": 0.4498,
      "step": 14738
    },
    {
      "epoch": 0.9232214723061746,
      "grad_norm": 0.9843882874404228,
      "learning_rate": 1.5362184409872194e-07,
      "loss": 0.4079,
      "step": 14739
    },
    {
      "epoch": 0.9232841103055168,
      "grad_norm": 0.8867269921736035,
      "learning_rate": 1.5337242496322234e-07,
      "loss": 0.3621,
      "step": 14740
    },
    {
      "epoch": 0.9233467483048592,
      "grad_norm": 0.8937706730956598,
      "learning_rate": 1.531232053158127e-07,
      "loss": 0.3353,
      "step": 14741
    },
    {
      "epoch": 0.9234093863042014,
      "grad_norm": 0.928627934025503,
      "learning_rate": 1.5287418516675202e-07,
      "loss": 0.4199,
      "step": 14742
    },
    {
      "epoch": 0.9234720243035437,
      "grad_norm": 0.7880696250055326,
      "learning_rate": 1.5262536452628929e-07,
      "loss": 0.3789,
      "step": 14743
    },
    {
      "epoch": 0.9235346623028861,
      "grad_norm": 0.8925507061483414,
      "learning_rate": 1.5237674340466635e-07,
      "loss": 0.393,
      "step": 14744
    },
    {
      "epoch": 0.9235973003022283,
      "grad_norm": 0.8911311821883656,
      "learning_rate": 1.5212832181211613e-07,
      "loss": 0.3669,
      "step": 14745
    },
    {
      "epoch": 0.9236599383015707,
      "grad_norm": 0.8655947731574158,
      "learning_rate": 1.518800997588643e-07,
      "loss": 0.3957,
      "step": 14746
    },
    {
      "epoch": 0.9237225763009129,
      "grad_norm": 0.8655935167762343,
      "learning_rate": 1.5163207725512718e-07,
      "loss": 0.4103,
      "step": 14747
    },
    {
      "epoch": 0.9237852143002553,
      "grad_norm": 0.8202327832086467,
      "learning_rate": 1.5138425431111327e-07,
      "loss": 0.3804,
      "step": 14748
    },
    {
      "epoch": 0.9238478522995975,
      "grad_norm": 0.8797047388307302,
      "learning_rate": 1.5113663093702434e-07,
      "loss": 0.3945,
      "step": 14749
    },
    {
      "epoch": 0.9239104902989399,
      "grad_norm": 0.9433878849978943,
      "learning_rate": 1.5088920714305122e-07,
      "loss": 0.4002,
      "step": 14750
    },
    {
      "epoch": 0.9239731282982822,
      "grad_norm": 0.8268468528005631,
      "learning_rate": 1.5064198293937849e-07,
      "loss": 0.3713,
      "step": 14751
    },
    {
      "epoch": 0.9240357662976244,
      "grad_norm": 0.926740667748047,
      "learning_rate": 1.503949583361819e-07,
      "loss": 0.4132,
      "step": 14752
    },
    {
      "epoch": 0.9240984042969668,
      "grad_norm": 0.8756922386679483,
      "learning_rate": 1.501481333436289e-07,
      "loss": 0.3642,
      "step": 14753
    },
    {
      "epoch": 0.924161042296309,
      "grad_norm": 0.8992174410982819,
      "learning_rate": 1.4990150797187854e-07,
      "loss": 0.3925,
      "step": 14754
    },
    {
      "epoch": 0.9242236802956514,
      "grad_norm": 0.9211233852893941,
      "learning_rate": 1.4965508223108272e-07,
      "loss": 0.3878,
      "step": 14755
    },
    {
      "epoch": 0.9242863182949936,
      "grad_norm": 0.8192392210080767,
      "learning_rate": 1.4940885613138335e-07,
      "loss": 0.3685,
      "step": 14756
    },
    {
      "epoch": 0.924348956294336,
      "grad_norm": 0.8393876083120878,
      "learning_rate": 1.4916282968291617e-07,
      "loss": 0.4181,
      "step": 14757
    },
    {
      "epoch": 0.9244115942936783,
      "grad_norm": 0.6109444737881098,
      "learning_rate": 1.4891700289580758e-07,
      "loss": 0.449,
      "step": 14758
    },
    {
      "epoch": 0.9244742322930206,
      "grad_norm": 0.7785353875808478,
      "learning_rate": 1.4867137578017553e-07,
      "loss": 0.3343,
      "step": 14759
    },
    {
      "epoch": 0.9245368702923629,
      "grad_norm": 0.8576340463057066,
      "learning_rate": 1.4842594834612978e-07,
      "loss": 0.3675,
      "step": 14760
    },
    {
      "epoch": 0.9245995082917051,
      "grad_norm": 0.9163171630507834,
      "learning_rate": 1.4818072060377275e-07,
      "loss": 0.3946,
      "step": 14761
    },
    {
      "epoch": 0.9246621462910475,
      "grad_norm": 0.9389890631051475,
      "learning_rate": 1.4793569256319806e-07,
      "loss": 0.4194,
      "step": 14762
    },
    {
      "epoch": 0.9247247842903897,
      "grad_norm": 0.8159189939022967,
      "learning_rate": 1.4769086423448985e-07,
      "loss": 0.3418,
      "step": 14763
    },
    {
      "epoch": 0.9247874222897321,
      "grad_norm": 0.8626756667531456,
      "learning_rate": 1.4744623562772676e-07,
      "loss": 0.4092,
      "step": 14764
    },
    {
      "epoch": 0.9248500602890743,
      "grad_norm": 0.9630489414101225,
      "learning_rate": 1.4720180675297736e-07,
      "loss": 0.3723,
      "step": 14765
    },
    {
      "epoch": 0.9249126982884167,
      "grad_norm": 0.8244360493749009,
      "learning_rate": 1.469575776203025e-07,
      "loss": 0.3555,
      "step": 14766
    },
    {
      "epoch": 0.924975336287759,
      "grad_norm": 0.8536466871007101,
      "learning_rate": 1.467135482397547e-07,
      "loss": 0.3735,
      "step": 14767
    },
    {
      "epoch": 0.9250379742871012,
      "grad_norm": 0.8224869151772132,
      "learning_rate": 1.464697186213776e-07,
      "loss": 0.3885,
      "step": 14768
    },
    {
      "epoch": 0.9251006122864436,
      "grad_norm": 0.8176680232550725,
      "learning_rate": 1.4622608877520816e-07,
      "loss": 0.425,
      "step": 14769
    },
    {
      "epoch": 0.9251632502857858,
      "grad_norm": 0.868449000623306,
      "learning_rate": 1.4598265871127336e-07,
      "loss": 0.42,
      "step": 14770
    },
    {
      "epoch": 0.9252258882851282,
      "grad_norm": 0.8806399274656759,
      "learning_rate": 1.4573942843959355e-07,
      "loss": 0.3882,
      "step": 14771
    },
    {
      "epoch": 0.9252885262844704,
      "grad_norm": 0.8529272246331876,
      "learning_rate": 1.4549639797017955e-07,
      "loss": 0.3369,
      "step": 14772
    },
    {
      "epoch": 0.9253511642838128,
      "grad_norm": 0.9493763637548642,
      "learning_rate": 1.4525356731303508e-07,
      "loss": 0.383,
      "step": 14773
    },
    {
      "epoch": 0.9254138022831551,
      "grad_norm": 0.8542992384023457,
      "learning_rate": 1.450109364781549e-07,
      "loss": 0.3643,
      "step": 14774
    },
    {
      "epoch": 0.9254764402824974,
      "grad_norm": 0.8593425841409795,
      "learning_rate": 1.4476850547552546e-07,
      "loss": 0.3747,
      "step": 14775
    },
    {
      "epoch": 0.9255390782818397,
      "grad_norm": 0.8679999164442935,
      "learning_rate": 1.44526274315126e-07,
      "loss": 0.3788,
      "step": 14776
    },
    {
      "epoch": 0.9256017162811819,
      "grad_norm": 0.8616763060917516,
      "learning_rate": 1.442842430069258e-07,
      "loss": 0.3895,
      "step": 14777
    },
    {
      "epoch": 0.9256643542805243,
      "grad_norm": 0.9205589679185712,
      "learning_rate": 1.4404241156088793e-07,
      "loss": 0.3986,
      "step": 14778
    },
    {
      "epoch": 0.9257269922798665,
      "grad_norm": 0.8497547402807707,
      "learning_rate": 1.4380077998696562e-07,
      "loss": 0.3743,
      "step": 14779
    },
    {
      "epoch": 0.9257896302792089,
      "grad_norm": 0.9601213052860507,
      "learning_rate": 1.435593482951042e-07,
      "loss": 0.4144,
      "step": 14780
    },
    {
      "epoch": 0.9258522682785512,
      "grad_norm": 0.8836591760942433,
      "learning_rate": 1.433181164952413e-07,
      "loss": 0.3867,
      "step": 14781
    },
    {
      "epoch": 0.9259149062778935,
      "grad_norm": 0.8760520368926288,
      "learning_rate": 1.4307708459730619e-07,
      "loss": 0.4138,
      "step": 14782
    },
    {
      "epoch": 0.9259775442772358,
      "grad_norm": 0.8895842564908901,
      "learning_rate": 1.428362526112198e-07,
      "loss": 0.4087,
      "step": 14783
    },
    {
      "epoch": 0.9260401822765781,
      "grad_norm": 0.840200838812197,
      "learning_rate": 1.4259562054689536e-07,
      "loss": 0.3464,
      "step": 14784
    },
    {
      "epoch": 0.9261028202759204,
      "grad_norm": 0.8615470422687604,
      "learning_rate": 1.4235518841423657e-07,
      "loss": 0.3383,
      "step": 14785
    },
    {
      "epoch": 0.9261654582752626,
      "grad_norm": 0.9310291339020254,
      "learning_rate": 1.4211495622313998e-07,
      "loss": 0.3982,
      "step": 14786
    },
    {
      "epoch": 0.926228096274605,
      "grad_norm": 0.9075455857482108,
      "learning_rate": 1.4187492398349324e-07,
      "loss": 0.362,
      "step": 14787
    },
    {
      "epoch": 0.9262907342739473,
      "grad_norm": 0.8375365990183425,
      "learning_rate": 1.416350917051762e-07,
      "loss": 0.3815,
      "step": 14788
    },
    {
      "epoch": 0.9263533722732896,
      "grad_norm": 0.8950199824015383,
      "learning_rate": 1.4139545939805988e-07,
      "loss": 0.3952,
      "step": 14789
    },
    {
      "epoch": 0.9264160102726319,
      "grad_norm": 0.8726064605248296,
      "learning_rate": 1.4115602707200914e-07,
      "loss": 0.3987,
      "step": 14790
    },
    {
      "epoch": 0.9264786482719742,
      "grad_norm": 0.8463332551594024,
      "learning_rate": 1.409167947368778e-07,
      "loss": 0.3545,
      "step": 14791
    },
    {
      "epoch": 0.9265412862713165,
      "grad_norm": 0.9275502728661408,
      "learning_rate": 1.4067776240251352e-07,
      "loss": 0.4418,
      "step": 14792
    },
    {
      "epoch": 0.9266039242706587,
      "grad_norm": 0.8339378557199282,
      "learning_rate": 1.40438930078754e-07,
      "loss": 0.3697,
      "step": 14793
    },
    {
      "epoch": 0.9266665622700011,
      "grad_norm": 0.8207154579906358,
      "learning_rate": 1.4020029777543075e-07,
      "loss": 0.3304,
      "step": 14794
    },
    {
      "epoch": 0.9267292002693434,
      "grad_norm": 0.9048810287278517,
      "learning_rate": 1.399618655023649e-07,
      "loss": 0.3863,
      "step": 14795
    },
    {
      "epoch": 0.9267918382686857,
      "grad_norm": 0.8697015934525428,
      "learning_rate": 1.3972363326937078e-07,
      "loss": 0.3641,
      "step": 14796
    },
    {
      "epoch": 0.926854476268028,
      "grad_norm": 0.8352039808586214,
      "learning_rate": 1.3948560108625386e-07,
      "loss": 0.4061,
      "step": 14797
    },
    {
      "epoch": 0.9269171142673703,
      "grad_norm": 0.8626541857777776,
      "learning_rate": 1.3924776896281135e-07,
      "loss": 0.354,
      "step": 14798
    },
    {
      "epoch": 0.9269797522667126,
      "grad_norm": 0.8211911033377771,
      "learning_rate": 1.3901013690883315e-07,
      "loss": 0.3897,
      "step": 14799
    },
    {
      "epoch": 0.927042390266055,
      "grad_norm": 0.8543986577915138,
      "learning_rate": 1.387727049340998e-07,
      "loss": 0.3929,
      "step": 14800
    },
    {
      "epoch": 0.9271050282653972,
      "grad_norm": 0.9144360453186952,
      "learning_rate": 1.3853547304838454e-07,
      "loss": 0.3865,
      "step": 14801
    },
    {
      "epoch": 0.9271676662647395,
      "grad_norm": 0.8372001662173909,
      "learning_rate": 1.3829844126145186e-07,
      "loss": 0.3543,
      "step": 14802
    },
    {
      "epoch": 0.9272303042640818,
      "grad_norm": 0.9095526456520832,
      "learning_rate": 1.3806160958305727e-07,
      "loss": 0.4161,
      "step": 14803
    },
    {
      "epoch": 0.9272929422634241,
      "grad_norm": 0.8595976302790328,
      "learning_rate": 1.378249780229496e-07,
      "loss": 0.3997,
      "step": 14804
    },
    {
      "epoch": 0.9273555802627664,
      "grad_norm": 0.8866447014570322,
      "learning_rate": 1.375885465908683e-07,
      "loss": 0.3677,
      "step": 14805
    },
    {
      "epoch": 0.9274182182621087,
      "grad_norm": 0.8486575829675538,
      "learning_rate": 1.3735231529654446e-07,
      "loss": 0.3828,
      "step": 14806
    },
    {
      "epoch": 0.927480856261451,
      "grad_norm": 0.8787071780766434,
      "learning_rate": 1.3711628414970203e-07,
      "loss": 0.39,
      "step": 14807
    },
    {
      "epoch": 0.9275434942607933,
      "grad_norm": 0.9008417473472965,
      "learning_rate": 1.3688045316005595e-07,
      "loss": 0.3688,
      "step": 14808
    },
    {
      "epoch": 0.9276061322601357,
      "grad_norm": 0.8459581663878987,
      "learning_rate": 1.366448223373129e-07,
      "loss": 0.3521,
      "step": 14809
    },
    {
      "epoch": 0.9276687702594779,
      "grad_norm": 0.8539597241033332,
      "learning_rate": 1.3640939169117185e-07,
      "loss": 0.355,
      "step": 14810
    },
    {
      "epoch": 0.9277314082588202,
      "grad_norm": 0.9188623978334577,
      "learning_rate": 1.3617416123132275e-07,
      "loss": 0.3959,
      "step": 14811
    },
    {
      "epoch": 0.9277940462581625,
      "grad_norm": 0.8822292944047937,
      "learning_rate": 1.3593913096744848e-07,
      "loss": 0.3723,
      "step": 14812
    },
    {
      "epoch": 0.9278566842575048,
      "grad_norm": 0.8248233297112681,
      "learning_rate": 1.3570430090922238e-07,
      "loss": 0.388,
      "step": 14813
    },
    {
      "epoch": 0.9279193222568471,
      "grad_norm": 0.9041088586667478,
      "learning_rate": 1.3546967106630948e-07,
      "loss": 0.3848,
      "step": 14814
    },
    {
      "epoch": 0.9279819602561894,
      "grad_norm": 0.5702302141268994,
      "learning_rate": 1.352352414483682e-07,
      "loss": 0.4358,
      "step": 14815
    },
    {
      "epoch": 0.9280445982555318,
      "grad_norm": 0.9415040254149123,
      "learning_rate": 1.350010120650469e-07,
      "loss": 0.4007,
      "step": 14816
    },
    {
      "epoch": 0.928107236254874,
      "grad_norm": 0.8225867270837077,
      "learning_rate": 1.347669829259879e-07,
      "loss": 0.3606,
      "step": 14817
    },
    {
      "epoch": 0.9281698742542163,
      "grad_norm": 0.9034243581324374,
      "learning_rate": 1.3453315404082178e-07,
      "loss": 0.4105,
      "step": 14818
    },
    {
      "epoch": 0.9282325122535586,
      "grad_norm": 0.9134099059678861,
      "learning_rate": 1.342995254191748e-07,
      "loss": 0.4373,
      "step": 14819
    },
    {
      "epoch": 0.9282951502529009,
      "grad_norm": 0.8194571711691807,
      "learning_rate": 1.34066097070662e-07,
      "loss": 0.4004,
      "step": 14820
    },
    {
      "epoch": 0.9283577882522432,
      "grad_norm": 0.8895313448270207,
      "learning_rate": 1.3383286900489234e-07,
      "loss": 0.3348,
      "step": 14821
    },
    {
      "epoch": 0.9284204262515855,
      "grad_norm": 0.7787666992445146,
      "learning_rate": 1.3359984123146487e-07,
      "loss": 0.3488,
      "step": 14822
    },
    {
      "epoch": 0.9284830642509279,
      "grad_norm": 0.8870908822860342,
      "learning_rate": 1.333670137599713e-07,
      "loss": 0.341,
      "step": 14823
    },
    {
      "epoch": 0.9285457022502701,
      "grad_norm": 0.819408302851515,
      "learning_rate": 1.33134386599994e-07,
      "loss": 0.3385,
      "step": 14824
    },
    {
      "epoch": 0.9286083402496125,
      "grad_norm": 0.8065629235770662,
      "learning_rate": 1.329019597611092e-07,
      "loss": 0.3481,
      "step": 14825
    },
    {
      "epoch": 0.9286709782489547,
      "grad_norm": 0.8549365432085776,
      "learning_rate": 1.3266973325288314e-07,
      "loss": 0.3757,
      "step": 14826
    },
    {
      "epoch": 0.928733616248297,
      "grad_norm": 0.8550025460734466,
      "learning_rate": 1.324377070848748e-07,
      "loss": 0.398,
      "step": 14827
    },
    {
      "epoch": 0.9287962542476393,
      "grad_norm": 0.8495817991751914,
      "learning_rate": 1.322058812666327e-07,
      "loss": 0.3709,
      "step": 14828
    },
    {
      "epoch": 0.9288588922469816,
      "grad_norm": 0.8762292154676864,
      "learning_rate": 1.3197425580770085e-07,
      "loss": 0.3699,
      "step": 14829
    },
    {
      "epoch": 0.928921530246324,
      "grad_norm": 0.8730399124947688,
      "learning_rate": 1.3174283071761217e-07,
      "loss": 0.3476,
      "step": 14830
    },
    {
      "epoch": 0.9289841682456662,
      "grad_norm": 0.8844284552804863,
      "learning_rate": 1.3151160600589185e-07,
      "loss": 0.3772,
      "step": 14831
    },
    {
      "epoch": 0.9290468062450086,
      "grad_norm": 0.7921222323021608,
      "learning_rate": 1.312805816820578e-07,
      "loss": 0.3525,
      "step": 14832
    },
    {
      "epoch": 0.9291094442443508,
      "grad_norm": 0.8324825878217739,
      "learning_rate": 1.3104975775561802e-07,
      "loss": 0.3813,
      "step": 14833
    },
    {
      "epoch": 0.9291720822436932,
      "grad_norm": 0.8347876316994729,
      "learning_rate": 1.3081913423607484e-07,
      "loss": 0.3231,
      "step": 14834
    },
    {
      "epoch": 0.9292347202430354,
      "grad_norm": 0.8505363068282926,
      "learning_rate": 1.3058871113291905e-07,
      "loss": 0.3344,
      "step": 14835
    },
    {
      "epoch": 0.9292973582423777,
      "grad_norm": 0.8546499017360409,
      "learning_rate": 1.3035848845563637e-07,
      "loss": 0.3818,
      "step": 14836
    },
    {
      "epoch": 0.92935999624172,
      "grad_norm": 0.7734214993592076,
      "learning_rate": 1.3012846621370144e-07,
      "loss": 0.3188,
      "step": 14837
    },
    {
      "epoch": 0.9294226342410623,
      "grad_norm": 0.9162882430224971,
      "learning_rate": 1.298986444165834e-07,
      "loss": 0.3817,
      "step": 14838
    },
    {
      "epoch": 0.9294852722404047,
      "grad_norm": 0.835383660757356,
      "learning_rate": 1.2966902307374073e-07,
      "loss": 0.3987,
      "step": 14839
    },
    {
      "epoch": 0.9295479102397469,
      "grad_norm": 0.8143057255523661,
      "learning_rate": 1.2943960219462537e-07,
      "loss": 0.3958,
      "step": 14840
    },
    {
      "epoch": 0.9296105482390893,
      "grad_norm": 0.9256229615024858,
      "learning_rate": 1.2921038178867972e-07,
      "loss": 0.4176,
      "step": 14841
    },
    {
      "epoch": 0.9296731862384315,
      "grad_norm": 0.8689653060955106,
      "learning_rate": 1.289813618653385e-07,
      "loss": 0.3489,
      "step": 14842
    },
    {
      "epoch": 0.9297358242377739,
      "grad_norm": 0.8824659921072899,
      "learning_rate": 1.2875254243402913e-07,
      "loss": 0.3594,
      "step": 14843
    },
    {
      "epoch": 0.9297984622371162,
      "grad_norm": 0.8627417836952641,
      "learning_rate": 1.285239235041691e-07,
      "loss": 0.4255,
      "step": 14844
    },
    {
      "epoch": 0.9298611002364584,
      "grad_norm": 0.8493714958244634,
      "learning_rate": 1.2829550508516864e-07,
      "loss": 0.3633,
      "step": 14845
    },
    {
      "epoch": 0.9299237382358008,
      "grad_norm": 0.8857338694271654,
      "learning_rate": 1.2806728718642913e-07,
      "loss": 0.4049,
      "step": 14846
    },
    {
      "epoch": 0.929986376235143,
      "grad_norm": 0.8849269430732126,
      "learning_rate": 1.2783926981734474e-07,
      "loss": 0.3917,
      "step": 14847
    },
    {
      "epoch": 0.9300490142344854,
      "grad_norm": 0.9105546162479768,
      "learning_rate": 1.2761145298730016e-07,
      "loss": 0.4223,
      "step": 14848
    },
    {
      "epoch": 0.9301116522338276,
      "grad_norm": 0.8287636934119735,
      "learning_rate": 1.2738383670567234e-07,
      "loss": 0.3521,
      "step": 14849
    },
    {
      "epoch": 0.93017429023317,
      "grad_norm": 0.9480743438140772,
      "learning_rate": 1.271564209818299e-07,
      "loss": 0.3912,
      "step": 14850
    },
    {
      "epoch": 0.9302369282325123,
      "grad_norm": 0.8659408455713699,
      "learning_rate": 1.2692920582513425e-07,
      "loss": 0.3699,
      "step": 14851
    },
    {
      "epoch": 0.9302995662318545,
      "grad_norm": 0.8989106039705924,
      "learning_rate": 1.2670219124493678e-07,
      "loss": 0.4085,
      "step": 14852
    },
    {
      "epoch": 0.9303622042311969,
      "grad_norm": 0.936227841389022,
      "learning_rate": 1.2647537725058222e-07,
      "loss": 0.4087,
      "step": 14853
    },
    {
      "epoch": 0.9304248422305391,
      "grad_norm": 0.796468287554183,
      "learning_rate": 1.2624876385140482e-07,
      "loss": 0.3674,
      "step": 14854
    },
    {
      "epoch": 0.9304874802298815,
      "grad_norm": 0.8741601600161825,
      "learning_rate": 1.2602235105673323e-07,
      "loss": 0.3568,
      "step": 14855
    },
    {
      "epoch": 0.9305501182292237,
      "grad_norm": 0.8426381944408357,
      "learning_rate": 1.2579613887588603e-07,
      "loss": 0.3807,
      "step": 14856
    },
    {
      "epoch": 0.9306127562285661,
      "grad_norm": 0.8284514768097777,
      "learning_rate": 1.2557012731817476e-07,
      "loss": 0.3468,
      "step": 14857
    },
    {
      "epoch": 0.9306753942279083,
      "grad_norm": 0.8792954939370311,
      "learning_rate": 1.253443163929019e-07,
      "loss": 0.4166,
      "step": 14858
    },
    {
      "epoch": 0.9307380322272507,
      "grad_norm": 0.8834380715743249,
      "learning_rate": 1.2511870610936115e-07,
      "loss": 0.3745,
      "step": 14859
    },
    {
      "epoch": 0.930800670226593,
      "grad_norm": 0.866331568649274,
      "learning_rate": 1.2489329647684013e-07,
      "loss": 0.4207,
      "step": 14860
    },
    {
      "epoch": 0.9308633082259352,
      "grad_norm": 0.8578320411914555,
      "learning_rate": 1.246680875046158e-07,
      "loss": 0.4138,
      "step": 14861
    },
    {
      "epoch": 0.9309259462252776,
      "grad_norm": 0.8618282090216588,
      "learning_rate": 1.2444307920195742e-07,
      "loss": 0.355,
      "step": 14862
    },
    {
      "epoch": 0.9309885842246198,
      "grad_norm": 0.8626318707935294,
      "learning_rate": 1.2421827157812705e-07,
      "loss": 0.4052,
      "step": 14863
    },
    {
      "epoch": 0.9310512222239622,
      "grad_norm": 0.8332374341115997,
      "learning_rate": 1.2399366464237728e-07,
      "loss": 0.3664,
      "step": 14864
    },
    {
      "epoch": 0.9311138602233044,
      "grad_norm": 0.9098542887360828,
      "learning_rate": 1.237692584039535e-07,
      "loss": 0.3891,
      "step": 14865
    },
    {
      "epoch": 0.9311764982226468,
      "grad_norm": 0.8724387342321722,
      "learning_rate": 1.2354505287209272e-07,
      "loss": 0.3686,
      "step": 14866
    },
    {
      "epoch": 0.9312391362219891,
      "grad_norm": 0.9057349082443297,
      "learning_rate": 1.2332104805602153e-07,
      "loss": 0.3943,
      "step": 14867
    },
    {
      "epoch": 0.9313017742213314,
      "grad_norm": 0.8858741700690045,
      "learning_rate": 1.2309724396496192e-07,
      "loss": 0.3801,
      "step": 14868
    },
    {
      "epoch": 0.9313644122206737,
      "grad_norm": 0.9066682026826358,
      "learning_rate": 1.228736406081249e-07,
      "loss": 0.3629,
      "step": 14869
    },
    {
      "epoch": 0.9314270502200159,
      "grad_norm": 0.8868420095059182,
      "learning_rate": 1.2265023799471422e-07,
      "loss": 0.3679,
      "step": 14870
    },
    {
      "epoch": 0.9314896882193583,
      "grad_norm": 0.8481306255943579,
      "learning_rate": 1.224270361339247e-07,
      "loss": 0.3527,
      "step": 14871
    },
    {
      "epoch": 0.9315523262187005,
      "grad_norm": 0.8374382722773357,
      "learning_rate": 1.2220403503494405e-07,
      "loss": 0.3429,
      "step": 14872
    },
    {
      "epoch": 0.9316149642180429,
      "grad_norm": 0.9280709175444355,
      "learning_rate": 1.2198123470695044e-07,
      "loss": 0.3919,
      "step": 14873
    },
    {
      "epoch": 0.9316776022173852,
      "grad_norm": 0.8407332406081014,
      "learning_rate": 1.2175863515911434e-07,
      "loss": 0.3722,
      "step": 14874
    },
    {
      "epoch": 0.9317402402167275,
      "grad_norm": 0.9117235540767225,
      "learning_rate": 1.2153623640059843e-07,
      "loss": 0.4215,
      "step": 14875
    },
    {
      "epoch": 0.9318028782160698,
      "grad_norm": 0.8971909911983352,
      "learning_rate": 1.2131403844055645e-07,
      "loss": 0.3772,
      "step": 14876
    },
    {
      "epoch": 0.931865516215412,
      "grad_norm": 0.8358524098185524,
      "learning_rate": 1.2109204128813446e-07,
      "loss": 0.3972,
      "step": 14877
    },
    {
      "epoch": 0.9319281542147544,
      "grad_norm": 0.890081534261119,
      "learning_rate": 1.2087024495246958e-07,
      "loss": 0.3945,
      "step": 14878
    },
    {
      "epoch": 0.9319907922140966,
      "grad_norm": 0.9042606416102673,
      "learning_rate": 1.206486494426906e-07,
      "loss": 0.3744,
      "step": 14879
    },
    {
      "epoch": 0.932053430213439,
      "grad_norm": 0.8935887559624153,
      "learning_rate": 1.2042725476791916e-07,
      "loss": 0.4037,
      "step": 14880
    },
    {
      "epoch": 0.9321160682127813,
      "grad_norm": 0.8432360463470633,
      "learning_rate": 1.2020606093726794e-07,
      "loss": 0.3943,
      "step": 14881
    },
    {
      "epoch": 0.9321787062121236,
      "grad_norm": 0.8010867924941264,
      "learning_rate": 1.1998506795984077e-07,
      "loss": 0.3369,
      "step": 14882
    },
    {
      "epoch": 0.9322413442114659,
      "grad_norm": 0.8825084900600176,
      "learning_rate": 1.197642758447337e-07,
      "loss": 0.3662,
      "step": 14883
    },
    {
      "epoch": 0.9323039822108082,
      "grad_norm": 0.9142896129921322,
      "learning_rate": 1.1954368460103504e-07,
      "loss": 0.3837,
      "step": 14884
    },
    {
      "epoch": 0.9323666202101505,
      "grad_norm": 0.9181599635802045,
      "learning_rate": 1.1932329423782363e-07,
      "loss": 0.3906,
      "step": 14885
    },
    {
      "epoch": 0.9324292582094927,
      "grad_norm": 0.5817856240342016,
      "learning_rate": 1.191031047641722e-07,
      "loss": 0.4217,
      "step": 14886
    },
    {
      "epoch": 0.9324918962088351,
      "grad_norm": 0.8601607004318278,
      "learning_rate": 1.188831161891424e-07,
      "loss": 0.4175,
      "step": 14887
    },
    {
      "epoch": 0.9325545342081774,
      "grad_norm": 0.9355345647014428,
      "learning_rate": 1.1866332852178975e-07,
      "loss": 0.3964,
      "step": 14888
    },
    {
      "epoch": 0.9326171722075197,
      "grad_norm": 0.5847617300628062,
      "learning_rate": 1.1844374177116091e-07,
      "loss": 0.4703,
      "step": 14889
    },
    {
      "epoch": 0.932679810206862,
      "grad_norm": 0.9064154445897123,
      "learning_rate": 1.1822435594629312e-07,
      "loss": 0.4149,
      "step": 14890
    },
    {
      "epoch": 0.9327424482062043,
      "grad_norm": 0.6441901940234684,
      "learning_rate": 1.1800517105621745e-07,
      "loss": 0.4591,
      "step": 14891
    },
    {
      "epoch": 0.9328050862055466,
      "grad_norm": 0.819823209810609,
      "learning_rate": 1.1778618710995393e-07,
      "loss": 0.3965,
      "step": 14892
    },
    {
      "epoch": 0.932867724204889,
      "grad_norm": 0.5801108588221369,
      "learning_rate": 1.1756740411651757e-07,
      "loss": 0.4351,
      "step": 14893
    },
    {
      "epoch": 0.9329303622042312,
      "grad_norm": 0.8281394658359641,
      "learning_rate": 1.1734882208491393e-07,
      "loss": 0.3883,
      "step": 14894
    },
    {
      "epoch": 0.9329930002035735,
      "grad_norm": 0.8452440010724656,
      "learning_rate": 1.1713044102413861e-07,
      "loss": 0.3674,
      "step": 14895
    },
    {
      "epoch": 0.9330556382029158,
      "grad_norm": 0.8683728555191663,
      "learning_rate": 1.1691226094318053e-07,
      "loss": 0.3848,
      "step": 14896
    },
    {
      "epoch": 0.9331182762022581,
      "grad_norm": 0.8319066024520386,
      "learning_rate": 1.1669428185102027e-07,
      "loss": 0.4023,
      "step": 14897
    },
    {
      "epoch": 0.9331809142016004,
      "grad_norm": 0.8374522756863786,
      "learning_rate": 1.1647650375662956e-07,
      "loss": 0.3685,
      "step": 14898
    },
    {
      "epoch": 0.9332435522009427,
      "grad_norm": 0.776031069049152,
      "learning_rate": 1.1625892666897288e-07,
      "loss": 0.3175,
      "step": 14899
    },
    {
      "epoch": 0.933306190200285,
      "grad_norm": 0.9117908241665251,
      "learning_rate": 1.1604155059700417e-07,
      "loss": 0.3858,
      "step": 14900
    },
    {
      "epoch": 0.9333688281996273,
      "grad_norm": 0.8330558173417701,
      "learning_rate": 1.1582437554967241e-07,
      "loss": 0.3665,
      "step": 14901
    },
    {
      "epoch": 0.9334314661989696,
      "grad_norm": 0.8509185211188688,
      "learning_rate": 1.1560740153591598e-07,
      "loss": 0.3234,
      "step": 14902
    },
    {
      "epoch": 0.9334941041983119,
      "grad_norm": 0.8566830658180128,
      "learning_rate": 1.1539062856466499e-07,
      "loss": 0.3871,
      "step": 14903
    },
    {
      "epoch": 0.9335567421976542,
      "grad_norm": 1.039867810436582,
      "learning_rate": 1.1517405664484338e-07,
      "loss": 0.3954,
      "step": 14904
    },
    {
      "epoch": 0.9336193801969965,
      "grad_norm": 0.8281852721134726,
      "learning_rate": 1.1495768578536348e-07,
      "loss": 0.3264,
      "step": 14905
    },
    {
      "epoch": 0.9336820181963388,
      "grad_norm": 0.6132401579052688,
      "learning_rate": 1.1474151599513206e-07,
      "loss": 0.4536,
      "step": 14906
    },
    {
      "epoch": 0.9337446561956811,
      "grad_norm": 0.8612396303987934,
      "learning_rate": 1.1452554728304699e-07,
      "loss": 0.4071,
      "step": 14907
    },
    {
      "epoch": 0.9338072941950234,
      "grad_norm": 0.8243059038108371,
      "learning_rate": 1.1430977965799673e-07,
      "loss": 0.3722,
      "step": 14908
    },
    {
      "epoch": 0.9338699321943658,
      "grad_norm": 0.877374230428213,
      "learning_rate": 1.1409421312886249e-07,
      "loss": 0.3956,
      "step": 14909
    },
    {
      "epoch": 0.933932570193708,
      "grad_norm": 0.8745324126909703,
      "learning_rate": 1.1387884770451773e-07,
      "loss": 0.4001,
      "step": 14910
    },
    {
      "epoch": 0.9339952081930503,
      "grad_norm": 0.8311582738807403,
      "learning_rate": 1.1366368339382649e-07,
      "loss": 0.3862,
      "step": 14911
    },
    {
      "epoch": 0.9340578461923926,
      "grad_norm": 0.8382069614807007,
      "learning_rate": 1.1344872020564502e-07,
      "loss": 0.3565,
      "step": 14912
    },
    {
      "epoch": 0.9341204841917349,
      "grad_norm": 0.9540585497575519,
      "learning_rate": 1.1323395814882121e-07,
      "loss": 0.396,
      "step": 14913
    },
    {
      "epoch": 0.9341831221910772,
      "grad_norm": 0.8686014264073995,
      "learning_rate": 1.1301939723219468e-07,
      "loss": 0.37,
      "step": 14914
    },
    {
      "epoch": 0.9342457601904195,
      "grad_norm": 0.8874265054588759,
      "learning_rate": 1.128050374645967e-07,
      "loss": 0.401,
      "step": 14915
    },
    {
      "epoch": 0.9343083981897619,
      "grad_norm": 0.8727242587406843,
      "learning_rate": 1.1259087885485076e-07,
      "loss": 0.3613,
      "step": 14916
    },
    {
      "epoch": 0.9343710361891041,
      "grad_norm": 0.9265680980118758,
      "learning_rate": 1.1237692141177148e-07,
      "loss": 0.4319,
      "step": 14917
    },
    {
      "epoch": 0.9344336741884465,
      "grad_norm": 0.9075158892460304,
      "learning_rate": 1.1216316514416458e-07,
      "loss": 0.3868,
      "step": 14918
    },
    {
      "epoch": 0.9344963121877887,
      "grad_norm": 0.8696507785491144,
      "learning_rate": 1.1194961006082972e-07,
      "loss": 0.3928,
      "step": 14919
    },
    {
      "epoch": 0.934558950187131,
      "grad_norm": 0.9562524196568379,
      "learning_rate": 1.117362561705554e-07,
      "loss": 0.4034,
      "step": 14920
    },
    {
      "epoch": 0.9346215881864733,
      "grad_norm": 0.9268965423506195,
      "learning_rate": 1.1152310348212458e-07,
      "loss": 0.3991,
      "step": 14921
    },
    {
      "epoch": 0.9346842261858156,
      "grad_norm": 0.8522373128349177,
      "learning_rate": 1.1131015200431028e-07,
      "loss": 0.3787,
      "step": 14922
    },
    {
      "epoch": 0.934746864185158,
      "grad_norm": 0.8880620992786948,
      "learning_rate": 1.1109740174587768e-07,
      "loss": 0.4087,
      "step": 14923
    },
    {
      "epoch": 0.9348095021845002,
      "grad_norm": 0.8929341367082161,
      "learning_rate": 1.1088485271558313e-07,
      "loss": 0.3966,
      "step": 14924
    },
    {
      "epoch": 0.9348721401838426,
      "grad_norm": 0.9215826792863062,
      "learning_rate": 1.1067250492217574e-07,
      "loss": 0.3795,
      "step": 14925
    },
    {
      "epoch": 0.9349347781831848,
      "grad_norm": 0.8887104624430465,
      "learning_rate": 1.1046035837439517e-07,
      "loss": 0.3701,
      "step": 14926
    },
    {
      "epoch": 0.9349974161825271,
      "grad_norm": 0.858962210785154,
      "learning_rate": 1.1024841308097389e-07,
      "loss": 0.3865,
      "step": 14927
    },
    {
      "epoch": 0.9350600541818694,
      "grad_norm": 0.7515150448219543,
      "learning_rate": 1.1003666905063548e-07,
      "loss": 0.3668,
      "step": 14928
    },
    {
      "epoch": 0.9351226921812117,
      "grad_norm": 0.8845589376856708,
      "learning_rate": 1.0982512629209518e-07,
      "loss": 0.3829,
      "step": 14929
    },
    {
      "epoch": 0.935185330180554,
      "grad_norm": 0.8646364203795543,
      "learning_rate": 1.0961378481406104e-07,
      "loss": 0.3964,
      "step": 14930
    },
    {
      "epoch": 0.9352479681798963,
      "grad_norm": 0.9416062064011014,
      "learning_rate": 1.0940264462523109e-07,
      "loss": 0.4358,
      "step": 14931
    },
    {
      "epoch": 0.9353106061792387,
      "grad_norm": 0.8132364995271923,
      "learning_rate": 1.091917057342956e-07,
      "loss": 0.3729,
      "step": 14932
    },
    {
      "epoch": 0.9353732441785809,
      "grad_norm": 0.7012394066785659,
      "learning_rate": 1.0898096814993763e-07,
      "loss": 0.45,
      "step": 14933
    },
    {
      "epoch": 0.9354358821779233,
      "grad_norm": 0.8308493328312976,
      "learning_rate": 1.0877043188083081e-07,
      "loss": 0.3619,
      "step": 14934
    },
    {
      "epoch": 0.9354985201772655,
      "grad_norm": 0.8396614289530001,
      "learning_rate": 1.0856009693564041e-07,
      "loss": 0.3903,
      "step": 14935
    },
    {
      "epoch": 0.9355611581766078,
      "grad_norm": 0.9269502216604076,
      "learning_rate": 1.083499633230245e-07,
      "loss": 0.3798,
      "step": 14936
    },
    {
      "epoch": 0.9356237961759502,
      "grad_norm": 0.9039424739597433,
      "learning_rate": 1.0814003105163229e-07,
      "loss": 0.413,
      "step": 14937
    },
    {
      "epoch": 0.9356864341752924,
      "grad_norm": 0.860078920809883,
      "learning_rate": 1.0793030013010408e-07,
      "loss": 0.3807,
      "step": 14938
    },
    {
      "epoch": 0.9357490721746348,
      "grad_norm": 1.0183515065285744,
      "learning_rate": 1.0772077056707297e-07,
      "loss": 0.3767,
      "step": 14939
    },
    {
      "epoch": 0.935811710173977,
      "grad_norm": 0.9220386064898971,
      "learning_rate": 1.0751144237116317e-07,
      "loss": 0.3761,
      "step": 14940
    },
    {
      "epoch": 0.9358743481733194,
      "grad_norm": 0.9171624644242028,
      "learning_rate": 1.0730231555099002e-07,
      "loss": 0.3686,
      "step": 14941
    },
    {
      "epoch": 0.9359369861726616,
      "grad_norm": 0.6394471700600383,
      "learning_rate": 1.0709339011516218e-07,
      "loss": 0.4479,
      "step": 14942
    },
    {
      "epoch": 0.935999624172004,
      "grad_norm": 0.8323268372080883,
      "learning_rate": 1.0688466607227831e-07,
      "loss": 0.3964,
      "step": 14943
    },
    {
      "epoch": 0.9360622621713462,
      "grad_norm": 0.8683920779425349,
      "learning_rate": 1.0667614343092936e-07,
      "loss": 0.4211,
      "step": 14944
    },
    {
      "epoch": 0.9361249001706885,
      "grad_norm": 0.8443952710857786,
      "learning_rate": 1.0646782219969953e-07,
      "loss": 0.3555,
      "step": 14945
    },
    {
      "epoch": 0.9361875381700309,
      "grad_norm": 0.8542372288695079,
      "learning_rate": 1.06259702387162e-07,
      "loss": 0.3733,
      "step": 14946
    },
    {
      "epoch": 0.9362501761693731,
      "grad_norm": 0.8756341473143918,
      "learning_rate": 1.0605178400188265e-07,
      "loss": 0.4085,
      "step": 14947
    },
    {
      "epoch": 0.9363128141687155,
      "grad_norm": 0.8395323694616363,
      "learning_rate": 1.0584406705242134e-07,
      "loss": 0.3879,
      "step": 14948
    },
    {
      "epoch": 0.9363754521680577,
      "grad_norm": 0.8752522875424086,
      "learning_rate": 1.0563655154732621e-07,
      "loss": 0.4162,
      "step": 14949
    },
    {
      "epoch": 0.9364380901674001,
      "grad_norm": 0.8727472902257055,
      "learning_rate": 1.0542923749513934e-07,
      "loss": 0.3922,
      "step": 14950
    },
    {
      "epoch": 0.9365007281667423,
      "grad_norm": 0.9785931016371348,
      "learning_rate": 1.0522212490439332e-07,
      "loss": 0.3963,
      "step": 14951
    },
    {
      "epoch": 0.9365633661660847,
      "grad_norm": 0.835453200218373,
      "learning_rate": 1.0501521378361246e-07,
      "loss": 0.3578,
      "step": 14952
    },
    {
      "epoch": 0.936626004165427,
      "grad_norm": 0.8539136491733887,
      "learning_rate": 1.0480850414131494e-07,
      "loss": 0.352,
      "step": 14953
    },
    {
      "epoch": 0.9366886421647692,
      "grad_norm": 0.900775773879923,
      "learning_rate": 1.0460199598600729e-07,
      "loss": 0.3766,
      "step": 14954
    },
    {
      "epoch": 0.9367512801641116,
      "grad_norm": 0.8530717712453164,
      "learning_rate": 1.0439568932619048e-07,
      "loss": 0.3535,
      "step": 14955
    },
    {
      "epoch": 0.9368139181634538,
      "grad_norm": 0.8577573719642143,
      "learning_rate": 1.0418958417035551e-07,
      "loss": 0.3648,
      "step": 14956
    },
    {
      "epoch": 0.9368765561627962,
      "grad_norm": 0.8542352818234393,
      "learning_rate": 1.0398368052698616e-07,
      "loss": 0.3978,
      "step": 14957
    },
    {
      "epoch": 0.9369391941621384,
      "grad_norm": 0.9034953315849998,
      "learning_rate": 1.0377797840455727e-07,
      "loss": 0.3406,
      "step": 14958
    },
    {
      "epoch": 0.9370018321614808,
      "grad_norm": 0.8400733216734257,
      "learning_rate": 1.0357247781153601e-07,
      "loss": 0.3997,
      "step": 14959
    },
    {
      "epoch": 0.9370644701608231,
      "grad_norm": 0.8642730735896035,
      "learning_rate": 1.0336717875638003e-07,
      "loss": 0.352,
      "step": 14960
    },
    {
      "epoch": 0.9371271081601653,
      "grad_norm": 0.6254392045709497,
      "learning_rate": 1.0316208124753923e-07,
      "loss": 0.4494,
      "step": 14961
    },
    {
      "epoch": 0.9371897461595077,
      "grad_norm": 0.968849501168239,
      "learning_rate": 1.0295718529345689e-07,
      "loss": 0.3793,
      "step": 14962
    },
    {
      "epoch": 0.9372523841588499,
      "grad_norm": 0.8725691216494189,
      "learning_rate": 1.0275249090256567e-07,
      "loss": 0.4197,
      "step": 14963
    },
    {
      "epoch": 0.9373150221581923,
      "grad_norm": 0.9433696454355214,
      "learning_rate": 1.0254799808329053e-07,
      "loss": 0.4143,
      "step": 14964
    },
    {
      "epoch": 0.9373776601575345,
      "grad_norm": 0.9201601452343265,
      "learning_rate": 1.0234370684404915e-07,
      "loss": 0.367,
      "step": 14965
    },
    {
      "epoch": 0.9374402981568769,
      "grad_norm": 0.9122288976991932,
      "learning_rate": 1.0213961719324927e-07,
      "loss": 0.4277,
      "step": 14966
    },
    {
      "epoch": 0.9375029361562192,
      "grad_norm": 0.902421172514538,
      "learning_rate": 1.0193572913929195e-07,
      "loss": 0.456,
      "step": 14967
    },
    {
      "epoch": 0.9375655741555615,
      "grad_norm": 0.926927865999392,
      "learning_rate": 1.0173204269056936e-07,
      "loss": 0.3933,
      "step": 14968
    },
    {
      "epoch": 0.9376282121549038,
      "grad_norm": 0.7752449100387977,
      "learning_rate": 1.0152855785546534e-07,
      "loss": 0.3835,
      "step": 14969
    },
    {
      "epoch": 0.937690850154246,
      "grad_norm": 0.8409916037786301,
      "learning_rate": 1.0132527464235431e-07,
      "loss": 0.3621,
      "step": 14970
    },
    {
      "epoch": 0.9377534881535884,
      "grad_norm": 0.9387741937405402,
      "learning_rate": 1.0112219305960514e-07,
      "loss": 0.3894,
      "step": 14971
    },
    {
      "epoch": 0.9378161261529306,
      "grad_norm": 0.5727712530606373,
      "learning_rate": 1.0091931311557557e-07,
      "loss": 0.4377,
      "step": 14972
    },
    {
      "epoch": 0.937878764152273,
      "grad_norm": 0.8823421174016246,
      "learning_rate": 1.0071663481861615e-07,
      "loss": 0.3554,
      "step": 14973
    },
    {
      "epoch": 0.9379414021516153,
      "grad_norm": 0.9213356261454885,
      "learning_rate": 1.0051415817706911e-07,
      "loss": 0.3924,
      "step": 14974
    },
    {
      "epoch": 0.9380040401509576,
      "grad_norm": 0.906537426064168,
      "learning_rate": 1.0031188319926887e-07,
      "loss": 0.3895,
      "step": 14975
    },
    {
      "epoch": 0.9380666781502999,
      "grad_norm": 0.9045452568728798,
      "learning_rate": 1.0010980989354102e-07,
      "loss": 0.4245,
      "step": 14976
    },
    {
      "epoch": 0.9381293161496422,
      "grad_norm": 0.8936935632060466,
      "learning_rate": 9.990793826820333e-08,
      "loss": 0.4335,
      "step": 14977
    },
    {
      "epoch": 0.9381919541489845,
      "grad_norm": 0.8162069515934607,
      "learning_rate": 9.970626833156361e-08,
      "loss": 0.3391,
      "step": 14978
    },
    {
      "epoch": 0.9382545921483267,
      "grad_norm": 0.8866911733321454,
      "learning_rate": 9.950480009192354e-08,
      "loss": 0.3909,
      "step": 14979
    },
    {
      "epoch": 0.9383172301476691,
      "grad_norm": 0.9686447526653326,
      "learning_rate": 9.930353355757594e-08,
      "loss": 0.4082,
      "step": 14980
    },
    {
      "epoch": 0.9383798681470114,
      "grad_norm": 0.8422468624440923,
      "learning_rate": 9.910246873680418e-08,
      "loss": 0.3657,
      "step": 14981
    },
    {
      "epoch": 0.9384425061463537,
      "grad_norm": 0.9541171253903789,
      "learning_rate": 9.890160563788498e-08,
      "loss": 0.4016,
      "step": 14982
    },
    {
      "epoch": 0.938505144145696,
      "grad_norm": 0.7685938104964755,
      "learning_rate": 9.870094426908506e-08,
      "loss": 0.3361,
      "step": 14983
    },
    {
      "epoch": 0.9385677821450383,
      "grad_norm": 0.6462953196914702,
      "learning_rate": 9.850048463866336e-08,
      "loss": 0.4403,
      "step": 14984
    },
    {
      "epoch": 0.9386304201443806,
      "grad_norm": 0.843608234335462,
      "learning_rate": 9.830022675487216e-08,
      "loss": 0.3687,
      "step": 14985
    },
    {
      "epoch": 0.9386930581437228,
      "grad_norm": 0.9245198949508849,
      "learning_rate": 9.810017062595322e-08,
      "loss": 0.384,
      "step": 14986
    },
    {
      "epoch": 0.9387556961430652,
      "grad_norm": 0.8547119702805077,
      "learning_rate": 9.790031626014107e-08,
      "loss": 0.3624,
      "step": 14987
    },
    {
      "epoch": 0.9388183341424075,
      "grad_norm": 0.899765247436963,
      "learning_rate": 9.770066366566189e-08,
      "loss": 0.392,
      "step": 14988
    },
    {
      "epoch": 0.9388809721417498,
      "grad_norm": 0.8284542660554622,
      "learning_rate": 9.750121285073355e-08,
      "loss": 0.3433,
      "step": 14989
    },
    {
      "epoch": 0.9389436101410921,
      "grad_norm": 0.6139394693173027,
      "learning_rate": 9.730196382356504e-08,
      "loss": 0.4652,
      "step": 14990
    },
    {
      "epoch": 0.9390062481404344,
      "grad_norm": 0.8634082258614973,
      "learning_rate": 9.71029165923576e-08,
      "loss": 0.385,
      "step": 14991
    },
    {
      "epoch": 0.9390688861397767,
      "grad_norm": 0.9024714751119672,
      "learning_rate": 9.690407116530409e-08,
      "loss": 0.3607,
      "step": 14992
    },
    {
      "epoch": 0.939131524139119,
      "grad_norm": 0.9360065497260018,
      "learning_rate": 9.670542755058909e-08,
      "loss": 0.3454,
      "step": 14993
    },
    {
      "epoch": 0.9391941621384613,
      "grad_norm": 0.931992438878749,
      "learning_rate": 9.650698575638883e-08,
      "loss": 0.3957,
      "step": 14994
    },
    {
      "epoch": 0.9392568001378035,
      "grad_norm": 0.879882422065157,
      "learning_rate": 9.630874579087124e-08,
      "loss": 0.3859,
      "step": 14995
    },
    {
      "epoch": 0.9393194381371459,
      "grad_norm": 0.9779258276358039,
      "learning_rate": 9.611070766219533e-08,
      "loss": 0.4083,
      "step": 14996
    },
    {
      "epoch": 0.9393820761364882,
      "grad_norm": 0.867932464455154,
      "learning_rate": 9.591287137851346e-08,
      "loss": 0.3713,
      "step": 14997
    },
    {
      "epoch": 0.9394447141358305,
      "grad_norm": 0.8571455880730287,
      "learning_rate": 9.571523694796803e-08,
      "loss": 0.3811,
      "step": 14998
    },
    {
      "epoch": 0.9395073521351728,
      "grad_norm": 0.8403767786999081,
      "learning_rate": 9.551780437869362e-08,
      "loss": 0.4187,
      "step": 14999
    },
    {
      "epoch": 0.9395699901345151,
      "grad_norm": 0.9188640649905074,
      "learning_rate": 9.532057367881653e-08,
      "loss": 0.4296,
      "step": 15000
    },
    {
      "epoch": 0.9396326281338574,
      "grad_norm": 0.8440263208321244,
      "learning_rate": 9.512354485645525e-08,
      "loss": 0.3911,
      "step": 15001
    },
    {
      "epoch": 0.9396952661331998,
      "grad_norm": 0.8275315943371241,
      "learning_rate": 9.492671791971886e-08,
      "loss": 0.3646,
      "step": 15002
    },
    {
      "epoch": 0.939757904132542,
      "grad_norm": 0.8966430970268109,
      "learning_rate": 9.473009287670976e-08,
      "loss": 0.3621,
      "step": 15003
    },
    {
      "epoch": 0.9398205421318843,
      "grad_norm": 0.8629530687734365,
      "learning_rate": 9.453366973551981e-08,
      "loss": 0.405,
      "step": 15004
    },
    {
      "epoch": 0.9398831801312266,
      "grad_norm": 0.8578869630172165,
      "learning_rate": 9.433744850423476e-08,
      "loss": 0.3564,
      "step": 15005
    },
    {
      "epoch": 0.9399458181305689,
      "grad_norm": 0.873351697704251,
      "learning_rate": 9.414142919093094e-08,
      "loss": 0.3676,
      "step": 15006
    },
    {
      "epoch": 0.9400084561299112,
      "grad_norm": 0.8807952309818285,
      "learning_rate": 9.39456118036769e-08,
      "loss": 0.359,
      "step": 15007
    },
    {
      "epoch": 0.9400710941292535,
      "grad_norm": 0.9198203717044129,
      "learning_rate": 9.374999635053172e-08,
      "loss": 0.3985,
      "step": 15008
    },
    {
      "epoch": 0.9401337321285959,
      "grad_norm": 0.8284009255083131,
      "learning_rate": 9.355458283954733e-08,
      "loss": 0.3747,
      "step": 15009
    },
    {
      "epoch": 0.9401963701279381,
      "grad_norm": 0.8768815568479759,
      "learning_rate": 9.335937127876727e-08,
      "loss": 0.3992,
      "step": 15010
    },
    {
      "epoch": 0.9402590081272804,
      "grad_norm": 0.864491603673651,
      "learning_rate": 9.316436167622567e-08,
      "loss": 0.3972,
      "step": 15011
    },
    {
      "epoch": 0.9403216461266227,
      "grad_norm": 0.9208867569380882,
      "learning_rate": 9.296955403995e-08,
      "loss": 0.3785,
      "step": 15012
    },
    {
      "epoch": 0.940384284125965,
      "grad_norm": 0.9057563514088213,
      "learning_rate": 9.277494837795775e-08,
      "loss": 0.4086,
      "step": 15013
    },
    {
      "epoch": 0.9404469221253073,
      "grad_norm": 0.8918759174071023,
      "learning_rate": 9.258054469825972e-08,
      "loss": 0.4103,
      "step": 15014
    },
    {
      "epoch": 0.9405095601246496,
      "grad_norm": 0.8288735299812139,
      "learning_rate": 9.238634300885785e-08,
      "loss": 0.3955,
      "step": 15015
    },
    {
      "epoch": 0.940572198123992,
      "grad_norm": 0.877303180540676,
      "learning_rate": 9.219234331774462e-08,
      "loss": 0.3415,
      "step": 15016
    },
    {
      "epoch": 0.9406348361233342,
      "grad_norm": 0.9188588153149,
      "learning_rate": 9.199854563290588e-08,
      "loss": 0.3884,
      "step": 15017
    },
    {
      "epoch": 0.9406974741226766,
      "grad_norm": 0.9045105432006103,
      "learning_rate": 9.180494996231747e-08,
      "loss": 0.4267,
      "step": 15018
    },
    {
      "epoch": 0.9407601121220188,
      "grad_norm": 0.8147147901146932,
      "learning_rate": 9.161155631394858e-08,
      "loss": 0.3678,
      "step": 15019
    },
    {
      "epoch": 0.9408227501213611,
      "grad_norm": 0.90698343581865,
      "learning_rate": 9.141836469575893e-08,
      "loss": 0.3799,
      "step": 15020
    },
    {
      "epoch": 0.9408853881207034,
      "grad_norm": 0.8680999491769914,
      "learning_rate": 9.122537511569996e-08,
      "loss": 0.3611,
      "step": 15021
    },
    {
      "epoch": 0.9409480261200457,
      "grad_norm": 0.8494937901651554,
      "learning_rate": 9.10325875817164e-08,
      "loss": 0.3749,
      "step": 15022
    },
    {
      "epoch": 0.941010664119388,
      "grad_norm": 0.8856961129738505,
      "learning_rate": 9.084000210174248e-08,
      "loss": 0.3799,
      "step": 15023
    },
    {
      "epoch": 0.9410733021187303,
      "grad_norm": 0.9282434430789732,
      "learning_rate": 9.064761868370519e-08,
      "loss": 0.3984,
      "step": 15024
    },
    {
      "epoch": 0.9411359401180727,
      "grad_norm": 0.7780995323790996,
      "learning_rate": 9.04554373355232e-08,
      "loss": 0.3728,
      "step": 15025
    },
    {
      "epoch": 0.9411985781174149,
      "grad_norm": 0.8554683251549439,
      "learning_rate": 9.026345806510683e-08,
      "loss": 0.3839,
      "step": 15026
    },
    {
      "epoch": 0.9412612161167573,
      "grad_norm": 0.8513877869088127,
      "learning_rate": 9.007168088035756e-08,
      "loss": 0.3775,
      "step": 15027
    },
    {
      "epoch": 0.9413238541160995,
      "grad_norm": 0.8367091568398095,
      "learning_rate": 8.988010578916962e-08,
      "loss": 0.3808,
      "step": 15028
    },
    {
      "epoch": 0.9413864921154418,
      "grad_norm": 0.8777310807067898,
      "learning_rate": 8.968873279942724e-08,
      "loss": 0.3639,
      "step": 15029
    },
    {
      "epoch": 0.9414491301147841,
      "grad_norm": 0.8473700447344359,
      "learning_rate": 8.949756191900804e-08,
      "loss": 0.3372,
      "step": 15030
    },
    {
      "epoch": 0.9415117681141264,
      "grad_norm": 0.853888837729198,
      "learning_rate": 8.930659315578127e-08,
      "loss": 0.3942,
      "step": 15031
    },
    {
      "epoch": 0.9415744061134688,
      "grad_norm": 0.9311133491957395,
      "learning_rate": 8.91158265176062e-08,
      "loss": 0.3962,
      "step": 15032
    },
    {
      "epoch": 0.941637044112811,
      "grad_norm": 0.8657334304350252,
      "learning_rate": 8.892526201233543e-08,
      "loss": 0.3766,
      "step": 15033
    },
    {
      "epoch": 0.9416996821121534,
      "grad_norm": 0.8692807422768174,
      "learning_rate": 8.873489964781212e-08,
      "loss": 0.3811,
      "step": 15034
    },
    {
      "epoch": 0.9417623201114956,
      "grad_norm": 0.8106144448290615,
      "learning_rate": 8.854473943187225e-08,
      "loss": 0.3677,
      "step": 15035
    },
    {
      "epoch": 0.941824958110838,
      "grad_norm": 0.8849624764857373,
      "learning_rate": 8.835478137234233e-08,
      "loss": 0.3661,
      "step": 15036
    },
    {
      "epoch": 0.9418875961101802,
      "grad_norm": 0.8983354437111402,
      "learning_rate": 8.816502547704109e-08,
      "loss": 0.4237,
      "step": 15037
    },
    {
      "epoch": 0.9419502341095225,
      "grad_norm": 0.8570807125445875,
      "learning_rate": 8.797547175377897e-08,
      "loss": 0.3344,
      "step": 15038
    },
    {
      "epoch": 0.9420128721088649,
      "grad_norm": 0.8285971868945287,
      "learning_rate": 8.778612021035804e-08,
      "loss": 0.3631,
      "step": 15039
    },
    {
      "epoch": 0.9420755101082071,
      "grad_norm": 0.8681284486776323,
      "learning_rate": 8.759697085457264e-08,
      "loss": 0.3764,
      "step": 15040
    },
    {
      "epoch": 0.9421381481075495,
      "grad_norm": 0.8543070758281515,
      "learning_rate": 8.740802369420764e-08,
      "loss": 0.3881,
      "step": 15041
    },
    {
      "epoch": 0.9422007861068917,
      "grad_norm": 0.8950293250706944,
      "learning_rate": 8.721927873704073e-08,
      "loss": 0.3883,
      "step": 15042
    },
    {
      "epoch": 0.9422634241062341,
      "grad_norm": 0.8581271202830603,
      "learning_rate": 8.703073599083955e-08,
      "loss": 0.3944,
      "step": 15043
    },
    {
      "epoch": 0.9423260621055763,
      "grad_norm": 0.5852656299861249,
      "learning_rate": 8.684239546336515e-08,
      "loss": 0.4338,
      "step": 15044
    },
    {
      "epoch": 0.9423887001049186,
      "grad_norm": 0.932925579221156,
      "learning_rate": 8.665425716236963e-08,
      "loss": 0.4264,
      "step": 15045
    },
    {
      "epoch": 0.942451338104261,
      "grad_norm": 0.8465205949018723,
      "learning_rate": 8.646632109559683e-08,
      "loss": 0.3716,
      "step": 15046
    },
    {
      "epoch": 0.9425139761036032,
      "grad_norm": 0.8884334160768661,
      "learning_rate": 8.62785872707822e-08,
      "loss": 0.387,
      "step": 15047
    },
    {
      "epoch": 0.9425766141029456,
      "grad_norm": 0.856945213293012,
      "learning_rate": 8.60910556956529e-08,
      "loss": 0.3465,
      "step": 15048
    },
    {
      "epoch": 0.9426392521022878,
      "grad_norm": 0.9275148367817185,
      "learning_rate": 8.590372637792832e-08,
      "loss": 0.4536,
      "step": 15049
    },
    {
      "epoch": 0.9427018901016302,
      "grad_norm": 0.8574629232050859,
      "learning_rate": 8.57165993253184e-08,
      "loss": 0.3616,
      "step": 15050
    },
    {
      "epoch": 0.9427645281009724,
      "grad_norm": 0.9062461195690011,
      "learning_rate": 8.552967454552475e-08,
      "loss": 0.4133,
      "step": 15051
    },
    {
      "epoch": 0.9428271661003148,
      "grad_norm": 0.9039352241779474,
      "learning_rate": 8.534295204624232e-08,
      "loss": 0.3704,
      "step": 15052
    },
    {
      "epoch": 0.9428898040996571,
      "grad_norm": 0.8735891616919268,
      "learning_rate": 8.515643183515665e-08,
      "loss": 0.3464,
      "step": 15053
    },
    {
      "epoch": 0.9429524420989993,
      "grad_norm": 0.9924452003357368,
      "learning_rate": 8.497011391994381e-08,
      "loss": 0.4286,
      "step": 15054
    },
    {
      "epoch": 0.9430150800983417,
      "grad_norm": 0.8295186577279857,
      "learning_rate": 8.478399830827321e-08,
      "loss": 0.3759,
      "step": 15055
    },
    {
      "epoch": 0.9430777180976839,
      "grad_norm": 0.8779085132111857,
      "learning_rate": 8.459808500780597e-08,
      "loss": 0.3677,
      "step": 15056
    },
    {
      "epoch": 0.9431403560970263,
      "grad_norm": 0.8402429373829123,
      "learning_rate": 8.441237402619373e-08,
      "loss": 0.335,
      "step": 15057
    },
    {
      "epoch": 0.9432029940963685,
      "grad_norm": 0.8038403406607492,
      "learning_rate": 8.422686537108037e-08,
      "loss": 0.3641,
      "step": 15058
    },
    {
      "epoch": 0.9432656320957109,
      "grad_norm": 0.9203091974024096,
      "learning_rate": 8.404155905010203e-08,
      "loss": 0.3889,
      "step": 15059
    },
    {
      "epoch": 0.9433282700950532,
      "grad_norm": 0.7656766000779617,
      "learning_rate": 8.385645507088536e-08,
      "loss": 0.3418,
      "step": 15060
    },
    {
      "epoch": 0.9433909080943955,
      "grad_norm": 0.8313866937484016,
      "learning_rate": 8.367155344104982e-08,
      "loss": 0.3689,
      "step": 15061
    },
    {
      "epoch": 0.9434535460937378,
      "grad_norm": 0.8853392771852597,
      "learning_rate": 8.348685416820546e-08,
      "loss": 0.3811,
      "step": 15062
    },
    {
      "epoch": 0.94351618409308,
      "grad_norm": 0.8989111291103703,
      "learning_rate": 8.330235725995395e-08,
      "loss": 0.3747,
      "step": 15063
    },
    {
      "epoch": 0.9435788220924224,
      "grad_norm": 0.9115295698050226,
      "learning_rate": 8.311806272389089e-08,
      "loss": 0.3651,
      "step": 15064
    },
    {
      "epoch": 0.9436414600917646,
      "grad_norm": 0.9041824540782172,
      "learning_rate": 8.293397056760078e-08,
      "loss": 0.3873,
      "step": 15065
    },
    {
      "epoch": 0.943704098091107,
      "grad_norm": 0.9215098490840182,
      "learning_rate": 8.275008079866142e-08,
      "loss": 0.3611,
      "step": 15066
    },
    {
      "epoch": 0.9437667360904493,
      "grad_norm": 0.9029324620680333,
      "learning_rate": 8.256639342464068e-08,
      "loss": 0.3797,
      "step": 15067
    },
    {
      "epoch": 0.9438293740897916,
      "grad_norm": 0.8882971994184312,
      "learning_rate": 8.238290845310027e-08,
      "loss": 0.4231,
      "step": 15068
    },
    {
      "epoch": 0.9438920120891339,
      "grad_norm": 0.8546895311559969,
      "learning_rate": 8.219962589159192e-08,
      "loss": 0.3766,
      "step": 15069
    },
    {
      "epoch": 0.9439546500884761,
      "grad_norm": 0.9219635817465593,
      "learning_rate": 8.201654574766016e-08,
      "loss": 0.4483,
      "step": 15070
    },
    {
      "epoch": 0.9440172880878185,
      "grad_norm": 0.8736467468696671,
      "learning_rate": 8.183366802884007e-08,
      "loss": 0.3589,
      "step": 15071
    },
    {
      "epoch": 0.9440799260871607,
      "grad_norm": 0.9318194626142455,
      "learning_rate": 8.165099274265897e-08,
      "loss": 0.4156,
      "step": 15072
    },
    {
      "epoch": 0.9441425640865031,
      "grad_norm": 0.8827914191864529,
      "learning_rate": 8.146851989663585e-08,
      "loss": 0.3801,
      "step": 15073
    },
    {
      "epoch": 0.9442052020858454,
      "grad_norm": 0.8478875135479487,
      "learning_rate": 8.128624949828134e-08,
      "loss": 0.4021,
      "step": 15074
    },
    {
      "epoch": 0.9442678400851877,
      "grad_norm": 0.9463648967542433,
      "learning_rate": 8.110418155509836e-08,
      "loss": 0.3961,
      "step": 15075
    },
    {
      "epoch": 0.94433047808453,
      "grad_norm": 0.8013979175614064,
      "learning_rate": 8.092231607457924e-08,
      "loss": 0.3784,
      "step": 15076
    },
    {
      "epoch": 0.9443931160838723,
      "grad_norm": 0.8682717915605606,
      "learning_rate": 8.074065306421131e-08,
      "loss": 0.3851,
      "step": 15077
    },
    {
      "epoch": 0.9444557540832146,
      "grad_norm": 0.9173896154108943,
      "learning_rate": 8.055919253147082e-08,
      "loss": 0.3994,
      "step": 15078
    },
    {
      "epoch": 0.9445183920825568,
      "grad_norm": 0.8420364197043873,
      "learning_rate": 8.037793448382736e-08,
      "loss": 0.3749,
      "step": 15079
    },
    {
      "epoch": 0.9445810300818992,
      "grad_norm": 0.9029445915224482,
      "learning_rate": 8.019687892874106e-08,
      "loss": 0.3792,
      "step": 15080
    },
    {
      "epoch": 0.9446436680812414,
      "grad_norm": 0.8164357510595971,
      "learning_rate": 8.00160258736643e-08,
      "loss": 0.3511,
      "step": 15081
    },
    {
      "epoch": 0.9447063060805838,
      "grad_norm": 0.579050584196693,
      "learning_rate": 7.983537532604114e-08,
      "loss": 0.4457,
      "step": 15082
    },
    {
      "epoch": 0.9447689440799261,
      "grad_norm": 0.8501706104779784,
      "learning_rate": 7.965492729330725e-08,
      "loss": 0.3832,
      "step": 15083
    },
    {
      "epoch": 0.9448315820792684,
      "grad_norm": 0.8502739767099133,
      "learning_rate": 7.947468178289009e-08,
      "loss": 0.4133,
      "step": 15084
    },
    {
      "epoch": 0.9448942200786107,
      "grad_norm": 0.9306206949448025,
      "learning_rate": 7.929463880220756e-08,
      "loss": 0.442,
      "step": 15085
    },
    {
      "epoch": 0.944956858077953,
      "grad_norm": 0.6325357973729725,
      "learning_rate": 7.911479835867153e-08,
      "loss": 0.4259,
      "step": 15086
    },
    {
      "epoch": 0.9450194960772953,
      "grad_norm": 0.8753759279752198,
      "learning_rate": 7.893516045968386e-08,
      "loss": 0.381,
      "step": 15087
    },
    {
      "epoch": 0.9450821340766375,
      "grad_norm": 0.8602884644192695,
      "learning_rate": 7.875572511263863e-08,
      "loss": 0.3682,
      "step": 15088
    },
    {
      "epoch": 0.9451447720759799,
      "grad_norm": 0.8388082108833839,
      "learning_rate": 7.857649232492049e-08,
      "loss": 0.3582,
      "step": 15089
    },
    {
      "epoch": 0.9452074100753222,
      "grad_norm": 0.7862719259706356,
      "learning_rate": 7.839746210390742e-08,
      "loss": 0.3568,
      "step": 15090
    },
    {
      "epoch": 0.9452700480746645,
      "grad_norm": 0.8736361478322904,
      "learning_rate": 7.821863445696853e-08,
      "loss": 0.3962,
      "step": 15091
    },
    {
      "epoch": 0.9453326860740068,
      "grad_norm": 0.8997244324522835,
      "learning_rate": 7.804000939146461e-08,
      "loss": 0.376,
      "step": 15092
    },
    {
      "epoch": 0.9453953240733491,
      "grad_norm": 0.8825800846685155,
      "learning_rate": 7.786158691474699e-08,
      "loss": 0.3487,
      "step": 15093
    },
    {
      "epoch": 0.9454579620726914,
      "grad_norm": 0.9301132749807726,
      "learning_rate": 7.768336703416035e-08,
      "loss": 0.4218,
      "step": 15094
    },
    {
      "epoch": 0.9455206000720336,
      "grad_norm": 0.8879169455258805,
      "learning_rate": 7.750534975703939e-08,
      "loss": 0.4122,
      "step": 15095
    },
    {
      "epoch": 0.945583238071376,
      "grad_norm": 0.8799609519821208,
      "learning_rate": 7.732753509071212e-08,
      "loss": 0.3609,
      "step": 15096
    },
    {
      "epoch": 0.9456458760707183,
      "grad_norm": 0.9059213652778431,
      "learning_rate": 7.714992304249769e-08,
      "loss": 0.3892,
      "step": 15097
    },
    {
      "epoch": 0.9457085140700606,
      "grad_norm": 0.8708754889219446,
      "learning_rate": 7.697251361970526e-08,
      "loss": 0.3958,
      "step": 15098
    },
    {
      "epoch": 0.9457711520694029,
      "grad_norm": 0.8391827622005361,
      "learning_rate": 7.679530682963843e-08,
      "loss": 0.3678,
      "step": 15099
    },
    {
      "epoch": 0.9458337900687452,
      "grad_norm": 0.830268964751129,
      "learning_rate": 7.66183026795908e-08,
      "loss": 0.3677,
      "step": 15100
    },
    {
      "epoch": 0.9458964280680875,
      "grad_norm": 0.9324853484210957,
      "learning_rate": 7.644150117684767e-08,
      "loss": 0.3759,
      "step": 15101
    },
    {
      "epoch": 0.9459590660674299,
      "grad_norm": 0.8331055740304505,
      "learning_rate": 7.626490232868599e-08,
      "loss": 0.3571,
      "step": 15102
    },
    {
      "epoch": 0.9460217040667721,
      "grad_norm": 0.6008206658765382,
      "learning_rate": 7.60885061423744e-08,
      "loss": 0.4347,
      "step": 15103
    },
    {
      "epoch": 0.9460843420661144,
      "grad_norm": 0.809035032656066,
      "learning_rate": 7.591231262517428e-08,
      "loss": 0.3238,
      "step": 15104
    },
    {
      "epoch": 0.9461469800654567,
      "grad_norm": 0.8919737390462726,
      "learning_rate": 7.57363217843371e-08,
      "loss": 0.4016,
      "step": 15105
    },
    {
      "epoch": 0.946209618064799,
      "grad_norm": 0.8401098248494416,
      "learning_rate": 7.556053362710702e-08,
      "loss": 0.3801,
      "step": 15106
    },
    {
      "epoch": 0.9462722560641413,
      "grad_norm": 0.9457438234499866,
      "learning_rate": 7.53849481607194e-08,
      "loss": 0.4178,
      "step": 15107
    },
    {
      "epoch": 0.9463348940634836,
      "grad_norm": 0.9524289002847265,
      "learning_rate": 7.52095653924012e-08,
      "loss": 0.4229,
      "step": 15108
    },
    {
      "epoch": 0.946397532062826,
      "grad_norm": 0.8852801752813008,
      "learning_rate": 7.503438532937169e-08,
      "loss": 0.4169,
      "step": 15109
    },
    {
      "epoch": 0.9464601700621682,
      "grad_norm": 0.9324892131854013,
      "learning_rate": 7.485940797884118e-08,
      "loss": 0.403,
      "step": 15110
    },
    {
      "epoch": 0.9465228080615106,
      "grad_norm": 0.8880535208384499,
      "learning_rate": 7.468463334801113e-08,
      "loss": 0.3653,
      "step": 15111
    },
    {
      "epoch": 0.9465854460608528,
      "grad_norm": 0.8083086875165644,
      "learning_rate": 7.45100614440758e-08,
      "loss": 0.3421,
      "step": 15112
    },
    {
      "epoch": 0.9466480840601951,
      "grad_norm": 0.8836173237227464,
      "learning_rate": 7.433569227422056e-08,
      "loss": 0.3835,
      "step": 15113
    },
    {
      "epoch": 0.9467107220595374,
      "grad_norm": 0.7757431784710443,
      "learning_rate": 7.416152584562242e-08,
      "loss": 0.3258,
      "step": 15114
    },
    {
      "epoch": 0.9467733600588797,
      "grad_norm": 0.872938663273693,
      "learning_rate": 7.398756216545011e-08,
      "loss": 0.3888,
      "step": 15115
    },
    {
      "epoch": 0.946835998058222,
      "grad_norm": 0.6445781179228164,
      "learning_rate": 7.381380124086457e-08,
      "loss": 0.4723,
      "step": 15116
    },
    {
      "epoch": 0.9468986360575643,
      "grad_norm": 0.9466855288261795,
      "learning_rate": 7.364024307901674e-08,
      "loss": 0.4269,
      "step": 15117
    },
    {
      "epoch": 0.9469612740569067,
      "grad_norm": 0.8535603313552126,
      "learning_rate": 7.346688768705146e-08,
      "loss": 0.3858,
      "step": 15118
    },
    {
      "epoch": 0.9470239120562489,
      "grad_norm": 0.7860891134415974,
      "learning_rate": 7.329373507210302e-08,
      "loss": 0.3476,
      "step": 15119
    },
    {
      "epoch": 0.9470865500555912,
      "grad_norm": 0.8996584469426612,
      "learning_rate": 7.312078524129906e-08,
      "loss": 0.3961,
      "step": 15120
    },
    {
      "epoch": 0.9471491880549335,
      "grad_norm": 0.8680335030152334,
      "learning_rate": 7.294803820175833e-08,
      "loss": 0.3623,
      "step": 15121
    },
    {
      "epoch": 0.9472118260542758,
      "grad_norm": 0.8650025392507383,
      "learning_rate": 7.277549396059013e-08,
      "loss": 0.3759,
      "step": 15122
    },
    {
      "epoch": 0.9472744640536181,
      "grad_norm": 0.9643424482817751,
      "learning_rate": 7.260315252489769e-08,
      "loss": 0.4138,
      "step": 15123
    },
    {
      "epoch": 0.9473371020529604,
      "grad_norm": 0.8516200160862064,
      "learning_rate": 7.243101390177365e-08,
      "loss": 0.3824,
      "step": 15124
    },
    {
      "epoch": 0.9473997400523028,
      "grad_norm": 0.8415437457003867,
      "learning_rate": 7.225907809830401e-08,
      "loss": 0.3664,
      "step": 15125
    },
    {
      "epoch": 0.947462378051645,
      "grad_norm": 0.8023137413848173,
      "learning_rate": 7.208734512156534e-08,
      "loss": 0.3444,
      "step": 15126
    },
    {
      "epoch": 0.9475250160509874,
      "grad_norm": 0.8649747159617228,
      "learning_rate": 7.191581497862588e-08,
      "loss": 0.3777,
      "step": 15127
    },
    {
      "epoch": 0.9475876540503296,
      "grad_norm": 0.812909994197236,
      "learning_rate": 7.174448767654662e-08,
      "loss": 0.403,
      "step": 15128
    },
    {
      "epoch": 0.9476502920496719,
      "grad_norm": 0.8587039360717753,
      "learning_rate": 7.157336322237862e-08,
      "loss": 0.4062,
      "step": 15129
    },
    {
      "epoch": 0.9477129300490142,
      "grad_norm": 0.8866525834827717,
      "learning_rate": 7.140244162316623e-08,
      "loss": 0.4236,
      "step": 15130
    },
    {
      "epoch": 0.9477755680483565,
      "grad_norm": 0.9053852611884159,
      "learning_rate": 7.123172288594327e-08,
      "loss": 0.3577,
      "step": 15131
    },
    {
      "epoch": 0.9478382060476989,
      "grad_norm": 0.9015876408314591,
      "learning_rate": 7.1061207017738e-08,
      "loss": 0.3814,
      "step": 15132
    },
    {
      "epoch": 0.9479008440470411,
      "grad_norm": 0.9272912175763967,
      "learning_rate": 7.089089402556815e-08,
      "loss": 0.4045,
      "step": 15133
    },
    {
      "epoch": 0.9479634820463835,
      "grad_norm": 0.7901990319626707,
      "learning_rate": 7.072078391644421e-08,
      "loss": 0.3566,
      "step": 15134
    },
    {
      "epoch": 0.9480261200457257,
      "grad_norm": 0.9149550104214828,
      "learning_rate": 7.055087669736782e-08,
      "loss": 0.3951,
      "step": 15135
    },
    {
      "epoch": 0.9480887580450681,
      "grad_norm": 0.8583201376145567,
      "learning_rate": 7.038117237533227e-08,
      "loss": 0.3388,
      "step": 15136
    },
    {
      "epoch": 0.9481513960444103,
      "grad_norm": 0.805067196436258,
      "learning_rate": 7.021167095732251e-08,
      "loss": 0.305,
      "step": 15137
    },
    {
      "epoch": 0.9482140340437526,
      "grad_norm": 0.9226588394363644,
      "learning_rate": 7.00423724503152e-08,
      "loss": 0.3667,
      "step": 15138
    },
    {
      "epoch": 0.948276672043095,
      "grad_norm": 0.8098700623771306,
      "learning_rate": 6.987327686127921e-08,
      "loss": 0.39,
      "step": 15139
    },
    {
      "epoch": 0.9483393100424372,
      "grad_norm": 0.8252514292232077,
      "learning_rate": 6.970438419717396e-08,
      "loss": 0.3621,
      "step": 15140
    },
    {
      "epoch": 0.9484019480417796,
      "grad_norm": 0.9008757871531998,
      "learning_rate": 6.953569446495112e-08,
      "loss": 0.4405,
      "step": 15141
    },
    {
      "epoch": 0.9484645860411218,
      "grad_norm": 0.9368795598292678,
      "learning_rate": 6.936720767155459e-08,
      "loss": 0.4004,
      "step": 15142
    },
    {
      "epoch": 0.9485272240404642,
      "grad_norm": 0.9726035271030847,
      "learning_rate": 6.919892382391879e-08,
      "loss": 0.4238,
      "step": 15143
    },
    {
      "epoch": 0.9485898620398064,
      "grad_norm": 0.646511547724569,
      "learning_rate": 6.903084292897044e-08,
      "loss": 0.4683,
      "step": 15144
    },
    {
      "epoch": 0.9486525000391488,
      "grad_norm": 0.8478179710725083,
      "learning_rate": 6.886296499362844e-08,
      "loss": 0.3685,
      "step": 15145
    },
    {
      "epoch": 0.9487151380384911,
      "grad_norm": 0.8383188356509546,
      "learning_rate": 6.869529002480168e-08,
      "loss": 0.403,
      "step": 15146
    },
    {
      "epoch": 0.9487777760378333,
      "grad_norm": 0.8815048659386333,
      "learning_rate": 6.85278180293919e-08,
      "loss": 0.3836,
      "step": 15147
    },
    {
      "epoch": 0.9488404140371757,
      "grad_norm": 0.7926881952603503,
      "learning_rate": 6.836054901429245e-08,
      "loss": 0.3949,
      "step": 15148
    },
    {
      "epoch": 0.9489030520365179,
      "grad_norm": 0.9806884527468106,
      "learning_rate": 6.819348298638839e-08,
      "loss": 0.4027,
      "step": 15149
    },
    {
      "epoch": 0.9489656900358603,
      "grad_norm": 0.9373877024027468,
      "learning_rate": 6.802661995255533e-08,
      "loss": 0.4429,
      "step": 15150
    },
    {
      "epoch": 0.9490283280352025,
      "grad_norm": 0.850290935452023,
      "learning_rate": 6.785995991966276e-08,
      "loss": 0.3968,
      "step": 15151
    },
    {
      "epoch": 0.9490909660345449,
      "grad_norm": 0.6667995819587709,
      "learning_rate": 6.769350289456966e-08,
      "loss": 0.4353,
      "step": 15152
    },
    {
      "epoch": 0.9491536040338872,
      "grad_norm": 0.8988012124252165,
      "learning_rate": 6.75272488841272e-08,
      "loss": 0.4008,
      "step": 15153
    },
    {
      "epoch": 0.9492162420332294,
      "grad_norm": 0.7947650443158673,
      "learning_rate": 6.73611978951788e-08,
      "loss": 0.3263,
      "step": 15154
    },
    {
      "epoch": 0.9492788800325718,
      "grad_norm": 0.8374899957768258,
      "learning_rate": 6.7195349934559e-08,
      "loss": 0.3828,
      "step": 15155
    },
    {
      "epoch": 0.949341518031914,
      "grad_norm": 0.6168483389885411,
      "learning_rate": 6.7029705009094e-08,
      "loss": 0.4506,
      "step": 15156
    },
    {
      "epoch": 0.9494041560312564,
      "grad_norm": 0.8318807129083625,
      "learning_rate": 6.686426312560224e-08,
      "loss": 0.3852,
      "step": 15157
    },
    {
      "epoch": 0.9494667940305986,
      "grad_norm": 0.8393372337770555,
      "learning_rate": 6.66990242908927e-08,
      "loss": 0.3829,
      "step": 15158
    },
    {
      "epoch": 0.949529432029941,
      "grad_norm": 0.87593612123407,
      "learning_rate": 6.653398851176718e-08,
      "loss": 0.3998,
      "step": 15159
    },
    {
      "epoch": 0.9495920700292833,
      "grad_norm": 0.908124428373624,
      "learning_rate": 6.636915579501801e-08,
      "loss": 0.4054,
      "step": 15160
    },
    {
      "epoch": 0.9496547080286256,
      "grad_norm": 0.8825382751226057,
      "learning_rate": 6.620452614743089e-08,
      "loss": 0.4286,
      "step": 15161
    },
    {
      "epoch": 0.9497173460279679,
      "grad_norm": 0.8162059340917535,
      "learning_rate": 6.604009957578095e-08,
      "loss": 0.3649,
      "step": 15162
    },
    {
      "epoch": 0.9497799840273101,
      "grad_norm": 0.8709000184141983,
      "learning_rate": 6.58758760868361e-08,
      "loss": 0.405,
      "step": 15163
    },
    {
      "epoch": 0.9498426220266525,
      "grad_norm": 0.8461824038331884,
      "learning_rate": 6.571185568735594e-08,
      "loss": 0.3363,
      "step": 15164
    },
    {
      "epoch": 0.9499052600259947,
      "grad_norm": 0.8482351107950012,
      "learning_rate": 6.55480383840923e-08,
      "loss": 0.3688,
      "step": 15165
    },
    {
      "epoch": 0.9499678980253371,
      "grad_norm": 0.8783912390452302,
      "learning_rate": 6.538442418378643e-08,
      "loss": 0.3847,
      "step": 15166
    },
    {
      "epoch": 0.9500305360246794,
      "grad_norm": 0.8091029985868348,
      "learning_rate": 6.522101309317352e-08,
      "loss": 0.3946,
      "step": 15167
    },
    {
      "epoch": 0.9500931740240217,
      "grad_norm": 0.9746687304978656,
      "learning_rate": 6.505780511897985e-08,
      "loss": 0.4584,
      "step": 15168
    },
    {
      "epoch": 0.950155812023364,
      "grad_norm": 0.5546194949179687,
      "learning_rate": 6.489480026792339e-08,
      "loss": 0.4494,
      "step": 15169
    },
    {
      "epoch": 0.9502184500227063,
      "grad_norm": 0.8545235857456099,
      "learning_rate": 6.473199854671264e-08,
      "loss": 0.3924,
      "step": 15170
    },
    {
      "epoch": 0.9502810880220486,
      "grad_norm": 0.8992308773200962,
      "learning_rate": 6.456939996204892e-08,
      "loss": 0.4016,
      "step": 15171
    },
    {
      "epoch": 0.9503437260213908,
      "grad_norm": 0.8576701654237787,
      "learning_rate": 6.440700452062465e-08,
      "loss": 0.3629,
      "step": 15172
    },
    {
      "epoch": 0.9504063640207332,
      "grad_norm": 0.8036932817332635,
      "learning_rate": 6.424481222912449e-08,
      "loss": 0.3307,
      "step": 15173
    },
    {
      "epoch": 0.9504690020200754,
      "grad_norm": 0.6218048671045283,
      "learning_rate": 6.408282309422308e-08,
      "loss": 0.4356,
      "step": 15174
    },
    {
      "epoch": 0.9505316400194178,
      "grad_norm": 0.8967478035001041,
      "learning_rate": 6.392103712258957e-08,
      "loss": 0.4192,
      "step": 15175
    },
    {
      "epoch": 0.9505942780187601,
      "grad_norm": 0.891487185519194,
      "learning_rate": 6.375945432088249e-08,
      "loss": 0.3934,
      "step": 15176
    },
    {
      "epoch": 0.9506569160181024,
      "grad_norm": 0.8663686283785116,
      "learning_rate": 6.359807469575153e-08,
      "loss": 0.3916,
      "step": 15177
    },
    {
      "epoch": 0.9507195540174447,
      "grad_norm": 0.8546869862189056,
      "learning_rate": 6.343689825384081e-08,
      "loss": 0.381,
      "step": 15178
    },
    {
      "epoch": 0.9507821920167869,
      "grad_norm": 0.8529755531848552,
      "learning_rate": 6.327592500178393e-08,
      "loss": 0.3596,
      "step": 15179
    },
    {
      "epoch": 0.9508448300161293,
      "grad_norm": 0.9240758688583309,
      "learning_rate": 6.311515494620613e-08,
      "loss": 0.3971,
      "step": 15180
    },
    {
      "epoch": 0.9509074680154715,
      "grad_norm": 0.5949771221986326,
      "learning_rate": 6.295458809372434e-08,
      "loss": 0.4372,
      "step": 15181
    },
    {
      "epoch": 0.9509701060148139,
      "grad_norm": 0.9031642832076464,
      "learning_rate": 6.279422445094829e-08,
      "loss": 0.3821,
      "step": 15182
    },
    {
      "epoch": 0.9510327440141562,
      "grad_norm": 0.9778693046006339,
      "learning_rate": 6.263406402447824e-08,
      "loss": 0.3559,
      "step": 15183
    },
    {
      "epoch": 0.9510953820134985,
      "grad_norm": 0.8307377455103001,
      "learning_rate": 6.247410682090671e-08,
      "loss": 0.3819,
      "step": 15184
    },
    {
      "epoch": 0.9511580200128408,
      "grad_norm": 0.9049246454439309,
      "learning_rate": 6.231435284681786e-08,
      "loss": 0.3966,
      "step": 15185
    },
    {
      "epoch": 0.9512206580121831,
      "grad_norm": 0.955229242956645,
      "learning_rate": 6.215480210878644e-08,
      "loss": 0.3962,
      "step": 15186
    },
    {
      "epoch": 0.9512832960115254,
      "grad_norm": 0.8942132938914066,
      "learning_rate": 6.199545461337941e-08,
      "loss": 0.3474,
      "step": 15187
    },
    {
      "epoch": 0.9513459340108676,
      "grad_norm": 0.8165754647841473,
      "learning_rate": 6.183631036715598e-08,
      "loss": 0.3496,
      "step": 15188
    },
    {
      "epoch": 0.95140857201021,
      "grad_norm": 0.8966908377544361,
      "learning_rate": 6.167736937666702e-08,
      "loss": 0.3664,
      "step": 15189
    },
    {
      "epoch": 0.9514712100095523,
      "grad_norm": 0.8562943336959343,
      "learning_rate": 6.151863164845396e-08,
      "loss": 0.3535,
      "step": 15190
    },
    {
      "epoch": 0.9515338480088946,
      "grad_norm": 0.9350706316788144,
      "learning_rate": 6.136009718905044e-08,
      "loss": 0.3943,
      "step": 15191
    },
    {
      "epoch": 0.9515964860082369,
      "grad_norm": 0.6520634803629606,
      "learning_rate": 6.12017660049824e-08,
      "loss": 0.4268,
      "step": 15192
    },
    {
      "epoch": 0.9516591240075792,
      "grad_norm": 0.6130850505446637,
      "learning_rate": 6.104363810276626e-08,
      "loss": 0.443,
      "step": 15193
    },
    {
      "epoch": 0.9517217620069215,
      "grad_norm": 0.9101618799282681,
      "learning_rate": 6.08857134889107e-08,
      "loss": 0.4063,
      "step": 15194
    },
    {
      "epoch": 0.9517844000062639,
      "grad_norm": 0.8383014823639624,
      "learning_rate": 6.072799216991609e-08,
      "loss": 0.3762,
      "step": 15195
    },
    {
      "epoch": 0.9518470380056061,
      "grad_norm": 0.8827515174053461,
      "learning_rate": 6.057047415227335e-08,
      "loss": 0.4107,
      "step": 15196
    },
    {
      "epoch": 0.9519096760049484,
      "grad_norm": 0.8749623569337858,
      "learning_rate": 6.041315944246728e-08,
      "loss": 0.3797,
      "step": 15197
    },
    {
      "epoch": 0.9519723140042907,
      "grad_norm": 0.9580130370878973,
      "learning_rate": 6.025604804697215e-08,
      "loss": 0.3602,
      "step": 15198
    },
    {
      "epoch": 0.952034952003633,
      "grad_norm": 0.8929533683675394,
      "learning_rate": 6.009913997225559e-08,
      "loss": 0.3848,
      "step": 15199
    },
    {
      "epoch": 0.9520975900029753,
      "grad_norm": 0.9007927096604686,
      "learning_rate": 5.994243522477405e-08,
      "loss": 0.3766,
      "step": 15200
    },
    {
      "epoch": 0.9521602280023176,
      "grad_norm": 0.9161683784071581,
      "learning_rate": 5.978593381097964e-08,
      "loss": 0.3964,
      "step": 15201
    },
    {
      "epoch": 0.95222286600166,
      "grad_norm": 0.8142244308388425,
      "learning_rate": 5.962963573731329e-08,
      "loss": 0.3709,
      "step": 15202
    },
    {
      "epoch": 0.9522855040010022,
      "grad_norm": 0.8583021495087235,
      "learning_rate": 5.947354101020819e-08,
      "loss": 0.3877,
      "step": 15203
    },
    {
      "epoch": 0.9523481420003445,
      "grad_norm": 0.8647342606418376,
      "learning_rate": 5.9317649636088656e-08,
      "loss": 0.4133,
      "step": 15204
    },
    {
      "epoch": 0.9524107799996868,
      "grad_norm": 0.8523429609816765,
      "learning_rate": 5.916196162137178e-08,
      "loss": 0.387,
      "step": 15205
    },
    {
      "epoch": 0.9524734179990291,
      "grad_norm": 0.8895751423374791,
      "learning_rate": 5.9006476972465754e-08,
      "loss": 0.4193,
      "step": 15206
    },
    {
      "epoch": 0.9525360559983714,
      "grad_norm": 0.9050616724266982,
      "learning_rate": 5.8851195695770467e-08,
      "loss": 0.4108,
      "step": 15207
    },
    {
      "epoch": 0.9525986939977137,
      "grad_norm": 0.7964308989773079,
      "learning_rate": 5.869611779767636e-08,
      "loss": 0.3441,
      "step": 15208
    },
    {
      "epoch": 0.952661331997056,
      "grad_norm": 0.842916369948338,
      "learning_rate": 5.854124328456778e-08,
      "loss": 0.3718,
      "step": 15209
    },
    {
      "epoch": 0.9527239699963983,
      "grad_norm": 0.9530411615397253,
      "learning_rate": 5.8386572162818514e-08,
      "loss": 0.4018,
      "step": 15210
    },
    {
      "epoch": 0.9527866079957407,
      "grad_norm": 0.8497925526907734,
      "learning_rate": 5.823210443879568e-08,
      "loss": 0.3979,
      "step": 15211
    },
    {
      "epoch": 0.9528492459950829,
      "grad_norm": 0.8752777519650528,
      "learning_rate": 5.807784011885587e-08,
      "loss": 0.369,
      "step": 15212
    },
    {
      "epoch": 0.9529118839944252,
      "grad_norm": 0.8780369617534569,
      "learning_rate": 5.7923779209350104e-08,
      "loss": 0.3841,
      "step": 15213
    },
    {
      "epoch": 0.9529745219937675,
      "grad_norm": 0.6906437893288655,
      "learning_rate": 5.776992171661833e-08,
      "loss": 0.4486,
      "step": 15214
    },
    {
      "epoch": 0.9530371599931098,
      "grad_norm": 0.8271541204829124,
      "learning_rate": 5.761626764699435e-08,
      "loss": 0.3915,
      "step": 15215
    },
    {
      "epoch": 0.9530997979924521,
      "grad_norm": 0.8972640522528235,
      "learning_rate": 5.7462817006802006e-08,
      "loss": 0.3887,
      "step": 15216
    },
    {
      "epoch": 0.9531624359917944,
      "grad_norm": 0.963509829319896,
      "learning_rate": 5.730956980235736e-08,
      "loss": 0.4056,
      "step": 15217
    },
    {
      "epoch": 0.9532250739911368,
      "grad_norm": 0.9076880298166101,
      "learning_rate": 5.715652603996813e-08,
      "loss": 0.379,
      "step": 15218
    },
    {
      "epoch": 0.953287711990479,
      "grad_norm": 0.9114870763500346,
      "learning_rate": 5.700368572593429e-08,
      "loss": 0.3683,
      "step": 15219
    },
    {
      "epoch": 0.9533503499898214,
      "grad_norm": 0.859914603338617,
      "learning_rate": 5.68510488665458e-08,
      "loss": 0.393,
      "step": 15220
    },
    {
      "epoch": 0.9534129879891636,
      "grad_norm": 0.8942071799406345,
      "learning_rate": 5.669861546808542e-08,
      "loss": 0.424,
      "step": 15221
    },
    {
      "epoch": 0.9534756259885059,
      "grad_norm": 0.8561670097500427,
      "learning_rate": 5.654638553682812e-08,
      "loss": 0.3498,
      "step": 15222
    },
    {
      "epoch": 0.9535382639878482,
      "grad_norm": 0.8811259244659206,
      "learning_rate": 5.639435907903834e-08,
      "loss": 0.3708,
      "step": 15223
    },
    {
      "epoch": 0.9536009019871905,
      "grad_norm": 0.9133383112285534,
      "learning_rate": 5.6242536100974964e-08,
      "loss": 0.447,
      "step": 15224
    },
    {
      "epoch": 0.9536635399865329,
      "grad_norm": 0.8444589111367443,
      "learning_rate": 5.609091660888633e-08,
      "loss": 0.3699,
      "step": 15225
    },
    {
      "epoch": 0.9537261779858751,
      "grad_norm": 0.9566524156043652,
      "learning_rate": 5.5939500609012986e-08,
      "loss": 0.3937,
      "step": 15226
    },
    {
      "epoch": 0.9537888159852175,
      "grad_norm": 0.8387826920445289,
      "learning_rate": 5.578828810758774e-08,
      "loss": 0.3691,
      "step": 15227
    },
    {
      "epoch": 0.9538514539845597,
      "grad_norm": 0.8930262895042048,
      "learning_rate": 5.563727911083394e-08,
      "loss": 0.395,
      "step": 15228
    },
    {
      "epoch": 0.953914091983902,
      "grad_norm": 0.9009484591859717,
      "learning_rate": 5.548647362496773e-08,
      "loss": 0.3895,
      "step": 15229
    },
    {
      "epoch": 0.9539767299832443,
      "grad_norm": 0.8951663985320926,
      "learning_rate": 5.5335871656195804e-08,
      "loss": 0.3712,
      "step": 15230
    },
    {
      "epoch": 0.9540393679825866,
      "grad_norm": 0.8614378610643855,
      "learning_rate": 5.518547321071765e-08,
      "loss": 0.3706,
      "step": 15231
    },
    {
      "epoch": 0.954102005981929,
      "grad_norm": 0.9296830845718519,
      "learning_rate": 5.503527829472277e-08,
      "loss": 0.3994,
      "step": 15232
    },
    {
      "epoch": 0.9541646439812712,
      "grad_norm": 0.9262552220034107,
      "learning_rate": 5.488528691439343e-08,
      "loss": 0.3768,
      "step": 15233
    },
    {
      "epoch": 0.9542272819806136,
      "grad_norm": 0.9028274462355177,
      "learning_rate": 5.4735499075903586e-08,
      "loss": 0.4121,
      "step": 15234
    },
    {
      "epoch": 0.9542899199799558,
      "grad_norm": 0.6430835958847367,
      "learning_rate": 5.458591478541886e-08,
      "loss": 0.4671,
      "step": 15235
    },
    {
      "epoch": 0.9543525579792982,
      "grad_norm": 0.8963391449964239,
      "learning_rate": 5.4436534049095436e-08,
      "loss": 0.3785,
      "step": 15236
    },
    {
      "epoch": 0.9544151959786404,
      "grad_norm": 0.9240025824155932,
      "learning_rate": 5.4287356873082284e-08,
      "loss": 0.4159,
      "step": 15237
    },
    {
      "epoch": 0.9544778339779827,
      "grad_norm": 0.8764848645026678,
      "learning_rate": 5.4138383263520054e-08,
      "loss": 0.3603,
      "step": 15238
    },
    {
      "epoch": 0.9545404719773251,
      "grad_norm": 0.8737078942415979,
      "learning_rate": 5.398961322653939e-08,
      "loss": 0.3879,
      "step": 15239
    },
    {
      "epoch": 0.9546031099766673,
      "grad_norm": 0.8048156342816535,
      "learning_rate": 5.3841046768264846e-08,
      "loss": 0.3592,
      "step": 15240
    },
    {
      "epoch": 0.9546657479760097,
      "grad_norm": 0.9274565500265314,
      "learning_rate": 5.3692683894810414e-08,
      "loss": 0.385,
      "step": 15241
    },
    {
      "epoch": 0.9547283859753519,
      "grad_norm": 0.8794174706305752,
      "learning_rate": 5.3544524612282876e-08,
      "loss": 0.3946,
      "step": 15242
    },
    {
      "epoch": 0.9547910239746943,
      "grad_norm": 0.9116926777392583,
      "learning_rate": 5.339656892678069e-08,
      "loss": 0.3915,
      "step": 15243
    },
    {
      "epoch": 0.9548536619740365,
      "grad_norm": 0.9106171366490637,
      "learning_rate": 5.3248816844394536e-08,
      "loss": 0.4043,
      "step": 15244
    },
    {
      "epoch": 0.9549162999733789,
      "grad_norm": 0.8790682716636106,
      "learning_rate": 5.310126837120511e-08,
      "loss": 0.3992,
      "step": 15245
    },
    {
      "epoch": 0.9549789379727212,
      "grad_norm": 0.844796654705665,
      "learning_rate": 5.295392351328532e-08,
      "loss": 0.3398,
      "step": 15246
    },
    {
      "epoch": 0.9550415759720634,
      "grad_norm": 0.8854191039286493,
      "learning_rate": 5.280678227670088e-08,
      "loss": 0.3618,
      "step": 15247
    },
    {
      "epoch": 0.9551042139714058,
      "grad_norm": 1.0037023562393483,
      "learning_rate": 5.2659844667506954e-08,
      "loss": 0.3922,
      "step": 15248
    },
    {
      "epoch": 0.955166851970748,
      "grad_norm": 0.5740176928433699,
      "learning_rate": 5.2513110691752575e-08,
      "loss": 0.4411,
      "step": 15249
    },
    {
      "epoch": 0.9552294899700904,
      "grad_norm": 0.9353010861426966,
      "learning_rate": 5.236658035547681e-08,
      "loss": 0.3584,
      "step": 15250
    },
    {
      "epoch": 0.9552921279694326,
      "grad_norm": 0.9285336075176985,
      "learning_rate": 5.2220253664710394e-08,
      "loss": 0.3918,
      "step": 15251
    },
    {
      "epoch": 0.955354765968775,
      "grad_norm": 0.8925130769824358,
      "learning_rate": 5.207413062547684e-08,
      "loss": 0.3731,
      "step": 15252
    },
    {
      "epoch": 0.9554174039681173,
      "grad_norm": 0.8724212743089182,
      "learning_rate": 5.1928211243790215e-08,
      "loss": 0.3481,
      "step": 15253
    },
    {
      "epoch": 0.9554800419674596,
      "grad_norm": 0.9051831566450509,
      "learning_rate": 5.178249552565739e-08,
      "loss": 0.3834,
      "step": 15254
    },
    {
      "epoch": 0.9555426799668019,
      "grad_norm": 0.8695245780739478,
      "learning_rate": 5.163698347707524e-08,
      "loss": 0.3691,
      "step": 15255
    },
    {
      "epoch": 0.9556053179661441,
      "grad_norm": 0.8629111028847275,
      "learning_rate": 5.1491675104033415e-08,
      "loss": 0.3919,
      "step": 15256
    },
    {
      "epoch": 0.9556679559654865,
      "grad_norm": 0.8328583604156387,
      "learning_rate": 5.1346570412512674e-08,
      "loss": 0.3847,
      "step": 15257
    },
    {
      "epoch": 0.9557305939648287,
      "grad_norm": 0.9151284564858712,
      "learning_rate": 5.1201669408486034e-08,
      "loss": 0.3959,
      "step": 15258
    },
    {
      "epoch": 0.9557932319641711,
      "grad_norm": 0.9031206909069097,
      "learning_rate": 5.105697209791649e-08,
      "loss": 0.3886,
      "step": 15259
    },
    {
      "epoch": 0.9558558699635133,
      "grad_norm": 0.8401680262234839,
      "learning_rate": 5.091247848676095e-08,
      "loss": 0.4005,
      "step": 15260
    },
    {
      "epoch": 0.9559185079628557,
      "grad_norm": 0.8770636343681987,
      "learning_rate": 5.076818858096577e-08,
      "loss": 0.3872,
      "step": 15261
    },
    {
      "epoch": 0.955981145962198,
      "grad_norm": 0.8827060041355576,
      "learning_rate": 5.06241023864712e-08,
      "loss": 0.381,
      "step": 15262
    },
    {
      "epoch": 0.9560437839615402,
      "grad_norm": 0.9026497752806967,
      "learning_rate": 5.048021990920748e-08,
      "loss": 0.4168,
      "step": 15263
    },
    {
      "epoch": 0.9561064219608826,
      "grad_norm": 0.8360569473300837,
      "learning_rate": 5.0336541155096006e-08,
      "loss": 0.3757,
      "step": 15264
    },
    {
      "epoch": 0.9561690599602248,
      "grad_norm": 0.8562164362803881,
      "learning_rate": 5.019306613005148e-08,
      "loss": 0.3639,
      "step": 15265
    },
    {
      "epoch": 0.9562316979595672,
      "grad_norm": 0.9818202293436474,
      "learning_rate": 5.004979483997918e-08,
      "loss": 0.3885,
      "step": 15266
    },
    {
      "epoch": 0.9562943359589094,
      "grad_norm": 0.8978690350702947,
      "learning_rate": 4.9906727290776055e-08,
      "loss": 0.4365,
      "step": 15267
    },
    {
      "epoch": 0.9563569739582518,
      "grad_norm": 0.934801068521005,
      "learning_rate": 4.976386348833018e-08,
      "loss": 0.4082,
      "step": 15268
    },
    {
      "epoch": 0.9564196119575941,
      "grad_norm": 0.8593009793375922,
      "learning_rate": 4.9621203438522946e-08,
      "loss": 0.3855,
      "step": 15269
    },
    {
      "epoch": 0.9564822499569364,
      "grad_norm": 0.8332581823311529,
      "learning_rate": 4.947874714722578e-08,
      "loss": 0.3456,
      "step": 15270
    },
    {
      "epoch": 0.9565448879562787,
      "grad_norm": 0.909906988910217,
      "learning_rate": 4.933649462030232e-08,
      "loss": 0.3959,
      "step": 15271
    },
    {
      "epoch": 0.9566075259556209,
      "grad_norm": 0.8632977407262562,
      "learning_rate": 4.919444586360789e-08,
      "loss": 0.4241,
      "step": 15272
    },
    {
      "epoch": 0.9566701639549633,
      "grad_norm": 0.8844778222207289,
      "learning_rate": 4.905260088298835e-08,
      "loss": 0.3936,
      "step": 15273
    },
    {
      "epoch": 0.9567328019543055,
      "grad_norm": 0.8775418105656873,
      "learning_rate": 4.89109596842835e-08,
      "loss": 0.3738,
      "step": 15274
    },
    {
      "epoch": 0.9567954399536479,
      "grad_norm": 0.8977451832989181,
      "learning_rate": 4.876952227332199e-08,
      "loss": 0.3914,
      "step": 15275
    },
    {
      "epoch": 0.9568580779529902,
      "grad_norm": 0.8717042390319484,
      "learning_rate": 4.862828865592584e-08,
      "loss": 0.3881,
      "step": 15276
    },
    {
      "epoch": 0.9569207159523325,
      "grad_norm": 0.8251224606835292,
      "learning_rate": 4.848725883790817e-08,
      "loss": 0.348,
      "step": 15277
    },
    {
      "epoch": 0.9569833539516748,
      "grad_norm": 0.8366950048258435,
      "learning_rate": 4.834643282507434e-08,
      "loss": 0.36,
      "step": 15278
    },
    {
      "epoch": 0.9570459919510171,
      "grad_norm": 0.5864545573239239,
      "learning_rate": 4.8205810623220274e-08,
      "loss": 0.4456,
      "step": 15279
    },
    {
      "epoch": 0.9571086299503594,
      "grad_norm": 0.8615858268494393,
      "learning_rate": 4.8065392238134664e-08,
      "loss": 0.3822,
      "step": 15280
    },
    {
      "epoch": 0.9571712679497016,
      "grad_norm": 0.8642434648876063,
      "learning_rate": 4.792517767559623e-08,
      "loss": 0.3744,
      "step": 15281
    },
    {
      "epoch": 0.957233905949044,
      "grad_norm": 0.814450208416004,
      "learning_rate": 4.778516694137647e-08,
      "loss": 0.3418,
      "step": 15282
    },
    {
      "epoch": 0.9572965439483863,
      "grad_norm": 0.8956806910980577,
      "learning_rate": 4.764536004123854e-08,
      "loss": 0.4317,
      "step": 15283
    },
    {
      "epoch": 0.9573591819477286,
      "grad_norm": 0.9075066538680118,
      "learning_rate": 4.750575698093674e-08,
      "loss": 0.4369,
      "step": 15284
    },
    {
      "epoch": 0.9574218199470709,
      "grad_norm": 0.8645897916064694,
      "learning_rate": 4.736635776621701e-08,
      "loss": 0.3765,
      "step": 15285
    },
    {
      "epoch": 0.9574844579464132,
      "grad_norm": 0.8586174737368258,
      "learning_rate": 4.722716240281755e-08,
      "loss": 0.3235,
      "step": 15286
    },
    {
      "epoch": 0.9575470959457555,
      "grad_norm": 0.8689420298855414,
      "learning_rate": 4.708817089646767e-08,
      "loss": 0.4049,
      "step": 15287
    },
    {
      "epoch": 0.9576097339450977,
      "grad_norm": 0.8931855200576905,
      "learning_rate": 4.6949383252887224e-08,
      "loss": 0.3764,
      "step": 15288
    },
    {
      "epoch": 0.9576723719444401,
      "grad_norm": 0.8711537083543348,
      "learning_rate": 4.681079947778999e-08,
      "loss": 0.3536,
      "step": 15289
    },
    {
      "epoch": 0.9577350099437824,
      "grad_norm": 0.8794931082806541,
      "learning_rate": 4.667241957687974e-08,
      "loss": 0.3676,
      "step": 15290
    },
    {
      "epoch": 0.9577976479431247,
      "grad_norm": 0.8410188432181981,
      "learning_rate": 4.65342435558519e-08,
      "loss": 0.3997,
      "step": 15291
    },
    {
      "epoch": 0.957860285942467,
      "grad_norm": 0.8108072124762906,
      "learning_rate": 4.6396271420394156e-08,
      "loss": 0.3878,
      "step": 15292
    },
    {
      "epoch": 0.9579229239418093,
      "grad_norm": 0.8724914537789534,
      "learning_rate": 4.625850317618475e-08,
      "loss": 0.3531,
      "step": 15293
    },
    {
      "epoch": 0.9579855619411516,
      "grad_norm": 0.6350242764695084,
      "learning_rate": 4.612093882889523e-08,
      "loss": 0.4479,
      "step": 15294
    },
    {
      "epoch": 0.958048199940494,
      "grad_norm": 0.860772348907484,
      "learning_rate": 4.598357838418721e-08,
      "loss": 0.3812,
      "step": 15295
    },
    {
      "epoch": 0.9581108379398362,
      "grad_norm": 0.6239764831342037,
      "learning_rate": 4.584642184771504e-08,
      "loss": 0.4465,
      "step": 15296
    },
    {
      "epoch": 0.9581734759391785,
      "grad_norm": 0.9386718348732345,
      "learning_rate": 4.5709469225123096e-08,
      "loss": 0.3842,
      "step": 15297
    },
    {
      "epoch": 0.9582361139385208,
      "grad_norm": 0.8504973015566546,
      "learning_rate": 4.557272052204909e-08,
      "loss": 0.3999,
      "step": 15298
    },
    {
      "epoch": 0.9582987519378631,
      "grad_norm": 0.8810495086535847,
      "learning_rate": 4.543617574412185e-08,
      "loss": 0.3739,
      "step": 15299
    },
    {
      "epoch": 0.9583613899372054,
      "grad_norm": 0.8679170919544106,
      "learning_rate": 4.529983489696077e-08,
      "loss": 0.3623,
      "step": 15300
    },
    {
      "epoch": 0.9584240279365477,
      "grad_norm": 0.896068138141103,
      "learning_rate": 4.516369798617859e-08,
      "loss": 0.3995,
      "step": 15301
    },
    {
      "epoch": 0.95848666593589,
      "grad_norm": 0.8723492253812636,
      "learning_rate": 4.502776501737749e-08,
      "loss": 0.3838,
      "step": 15302
    },
    {
      "epoch": 0.9585493039352323,
      "grad_norm": 0.7747608936620805,
      "learning_rate": 4.489203599615411e-08,
      "loss": 0.3358,
      "step": 15303
    },
    {
      "epoch": 0.9586119419345747,
      "grad_norm": 0.8470925841621801,
      "learning_rate": 4.4756510928093965e-08,
      "loss": 0.3668,
      "step": 15304
    },
    {
      "epoch": 0.9586745799339169,
      "grad_norm": 0.8939239232424739,
      "learning_rate": 4.462118981877539e-08,
      "loss": 0.3711,
      "step": 15305
    },
    {
      "epoch": 0.9587372179332592,
      "grad_norm": 0.8635011635000369,
      "learning_rate": 4.448607267376836e-08,
      "loss": 0.4024,
      "step": 15306
    },
    {
      "epoch": 0.9587998559326015,
      "grad_norm": 0.8985912525370945,
      "learning_rate": 4.435115949863456e-08,
      "loss": 0.4104,
      "step": 15307
    },
    {
      "epoch": 0.9588624939319438,
      "grad_norm": 0.8817117091057654,
      "learning_rate": 4.4216450298926736e-08,
      "loss": 0.3617,
      "step": 15308
    },
    {
      "epoch": 0.9589251319312861,
      "grad_norm": 0.905955124032304,
      "learning_rate": 4.408194508018992e-08,
      "loss": 0.4174,
      "step": 15309
    },
    {
      "epoch": 0.9589877699306284,
      "grad_norm": 0.8954924473245595,
      "learning_rate": 4.394764384796024e-08,
      "loss": 0.372,
      "step": 15310
    },
    {
      "epoch": 0.9590504079299708,
      "grad_norm": 0.8634253049096153,
      "learning_rate": 4.381354660776493e-08,
      "loss": 0.392,
      "step": 15311
    },
    {
      "epoch": 0.959113045929313,
      "grad_norm": 0.855700492545292,
      "learning_rate": 4.367965336512403e-08,
      "loss": 0.3578,
      "step": 15312
    },
    {
      "epoch": 0.9591756839286553,
      "grad_norm": 0.8857635696051541,
      "learning_rate": 4.354596412554868e-08,
      "loss": 0.4047,
      "step": 15313
    },
    {
      "epoch": 0.9592383219279976,
      "grad_norm": 0.8358799558002553,
      "learning_rate": 4.3412478894541145e-08,
      "loss": 0.3565,
      "step": 15314
    },
    {
      "epoch": 0.9593009599273399,
      "grad_norm": 0.9259269322925344,
      "learning_rate": 4.327919767759592e-08,
      "loss": 0.382,
      "step": 15315
    },
    {
      "epoch": 0.9593635979266822,
      "grad_norm": 0.8827408557729712,
      "learning_rate": 4.3146120480199726e-08,
      "loss": 0.393,
      "step": 15316
    },
    {
      "epoch": 0.9594262359260245,
      "grad_norm": 0.8495653463173831,
      "learning_rate": 4.3013247307828746e-08,
      "loss": 0.3999,
      "step": 15317
    },
    {
      "epoch": 0.9594888739253669,
      "grad_norm": 0.80530272604344,
      "learning_rate": 4.288057816595248e-08,
      "loss": 0.3686,
      "step": 15318
    },
    {
      "epoch": 0.9595515119247091,
      "grad_norm": 0.8837064988689668,
      "learning_rate": 4.2748113060031574e-08,
      "loss": 0.3592,
      "step": 15319
    },
    {
      "epoch": 0.9596141499240515,
      "grad_norm": 0.8600026123983309,
      "learning_rate": 4.261585199551832e-08,
      "loss": 0.3552,
      "step": 15320
    },
    {
      "epoch": 0.9596767879233937,
      "grad_norm": 0.8654465728683391,
      "learning_rate": 4.2483794977857264e-08,
      "loss": 0.338,
      "step": 15321
    },
    {
      "epoch": 0.959739425922736,
      "grad_norm": 0.9153208149562682,
      "learning_rate": 4.2351942012483495e-08,
      "loss": 0.3354,
      "step": 15322
    },
    {
      "epoch": 0.9598020639220783,
      "grad_norm": 0.866859519564023,
      "learning_rate": 4.2220293104823227e-08,
      "loss": 0.3864,
      "step": 15323
    },
    {
      "epoch": 0.9598647019214206,
      "grad_norm": 0.8863362030696739,
      "learning_rate": 4.208884826029658e-08,
      "loss": 0.3936,
      "step": 15324
    },
    {
      "epoch": 0.959927339920763,
      "grad_norm": 0.8587414642228819,
      "learning_rate": 4.195760748431254e-08,
      "loss": 0.4096,
      "step": 15325
    },
    {
      "epoch": 0.9599899779201052,
      "grad_norm": 0.8432904721126594,
      "learning_rate": 4.182657078227403e-08,
      "loss": 0.3506,
      "step": 15326
    },
    {
      "epoch": 0.9600526159194476,
      "grad_norm": 0.8004953909229986,
      "learning_rate": 4.16957381595745e-08,
      "loss": 0.3717,
      "step": 15327
    },
    {
      "epoch": 0.9601152539187898,
      "grad_norm": 0.8909361408817863,
      "learning_rate": 4.156510962159799e-08,
      "loss": 0.3919,
      "step": 15328
    },
    {
      "epoch": 0.9601778919181322,
      "grad_norm": 0.9332616344343582,
      "learning_rate": 4.143468517372185e-08,
      "loss": 0.3865,
      "step": 15329
    },
    {
      "epoch": 0.9602405299174744,
      "grad_norm": 0.9187151296010319,
      "learning_rate": 4.1304464821315114e-08,
      "loss": 0.4433,
      "step": 15330
    },
    {
      "epoch": 0.9603031679168167,
      "grad_norm": 0.792129185651848,
      "learning_rate": 4.1174448569736846e-08,
      "loss": 0.3563,
      "step": 15331
    },
    {
      "epoch": 0.960365805916159,
      "grad_norm": 0.9254371344397172,
      "learning_rate": 4.104463642433831e-08,
      "loss": 0.393,
      "step": 15332
    },
    {
      "epoch": 0.9604284439155013,
      "grad_norm": 0.9191179374093187,
      "learning_rate": 4.091502839046302e-08,
      "loss": 0.3286,
      "step": 15333
    },
    {
      "epoch": 0.9604910819148437,
      "grad_norm": 0.6435470101395275,
      "learning_rate": 4.078562447344558e-08,
      "loss": 0.4481,
      "step": 15334
    },
    {
      "epoch": 0.9605537199141859,
      "grad_norm": 0.8587576988064328,
      "learning_rate": 4.065642467861286e-08,
      "loss": 0.374,
      "step": 15335
    },
    {
      "epoch": 0.9606163579135283,
      "grad_norm": 0.9007243091756292,
      "learning_rate": 4.05274290112817e-08,
      "loss": 0.4232,
      "step": 15336
    },
    {
      "epoch": 0.9606789959128705,
      "grad_norm": 0.9141629787079678,
      "learning_rate": 4.039863747676176e-08,
      "loss": 0.4193,
      "step": 15337
    },
    {
      "epoch": 0.9607416339122128,
      "grad_norm": 0.9261849861608009,
      "learning_rate": 4.02700500803549e-08,
      "loss": 0.4013,
      "step": 15338
    },
    {
      "epoch": 0.9608042719115552,
      "grad_norm": 0.8482849075669768,
      "learning_rate": 4.014166682735354e-08,
      "loss": 0.3231,
      "step": 15339
    },
    {
      "epoch": 0.9608669099108974,
      "grad_norm": 0.9052748192137557,
      "learning_rate": 4.0013487723041256e-08,
      "loss": 0.4023,
      "step": 15340
    },
    {
      "epoch": 0.9609295479102398,
      "grad_norm": 1.0008872664434418,
      "learning_rate": 3.988551277269492e-08,
      "loss": 0.3923,
      "step": 15341
    },
    {
      "epoch": 0.960992185909582,
      "grad_norm": 0.7785873246734197,
      "learning_rate": 3.975774198158144e-08,
      "loss": 0.3642,
      "step": 15342
    },
    {
      "epoch": 0.9610548239089244,
      "grad_norm": 0.8135601522254703,
      "learning_rate": 3.963017535495939e-08,
      "loss": 0.3213,
      "step": 15343
    },
    {
      "epoch": 0.9611174619082666,
      "grad_norm": 0.8802374030901752,
      "learning_rate": 3.9502812898080664e-08,
      "loss": 0.3712,
      "step": 15344
    },
    {
      "epoch": 0.961180099907609,
      "grad_norm": 0.8740405827297618,
      "learning_rate": 3.937565461618664e-08,
      "loss": 0.383,
      "step": 15345
    },
    {
      "epoch": 0.9612427379069513,
      "grad_norm": 0.9033242104465589,
      "learning_rate": 3.924870051451146e-08,
      "loss": 0.4379,
      "step": 15346
    },
    {
      "epoch": 0.9613053759062935,
      "grad_norm": 0.8853119623081632,
      "learning_rate": 3.912195059828039e-08,
      "loss": 0.3942,
      "step": 15347
    },
    {
      "epoch": 0.9613680139056359,
      "grad_norm": 0.8658876111158962,
      "learning_rate": 3.899540487271092e-08,
      "loss": 0.3827,
      "step": 15348
    },
    {
      "epoch": 0.9614306519049781,
      "grad_norm": 0.8212473629545832,
      "learning_rate": 3.8869063343011106e-08,
      "loss": 0.3641,
      "step": 15349
    },
    {
      "epoch": 0.9614932899043205,
      "grad_norm": 0.8714586178618858,
      "learning_rate": 3.874292601438179e-08,
      "loss": 0.3921,
      "step": 15350
    },
    {
      "epoch": 0.9615559279036627,
      "grad_norm": 0.8613215077712058,
      "learning_rate": 3.861699289201437e-08,
      "loss": 0.3401,
      "step": 15351
    },
    {
      "epoch": 0.9616185659030051,
      "grad_norm": 0.8469106780717525,
      "learning_rate": 3.849126398109193e-08,
      "loss": 0.3796,
      "step": 15352
    },
    {
      "epoch": 0.9616812039023473,
      "grad_norm": 0.896139734344734,
      "learning_rate": 3.836573928679033e-08,
      "loss": 0.4266,
      "step": 15353
    },
    {
      "epoch": 0.9617438419016897,
      "grad_norm": 0.8937707720733193,
      "learning_rate": 3.824041881427543e-08,
      "loss": 0.4073,
      "step": 15354
    },
    {
      "epoch": 0.961806479901032,
      "grad_norm": 0.6374670147624867,
      "learning_rate": 3.8115302568706435e-08,
      "loss": 0.4478,
      "step": 15355
    },
    {
      "epoch": 0.9618691179003742,
      "grad_norm": 0.6442835997468715,
      "learning_rate": 3.799039055523257e-08,
      "loss": 0.4496,
      "step": 15356
    },
    {
      "epoch": 0.9619317558997166,
      "grad_norm": 0.8915447684133789,
      "learning_rate": 3.7865682778994716e-08,
      "loss": 0.3495,
      "step": 15357
    },
    {
      "epoch": 0.9619943938990588,
      "grad_norm": 0.8814450830168428,
      "learning_rate": 3.7741179245126545e-08,
      "loss": 0.4177,
      "step": 15358
    },
    {
      "epoch": 0.9620570318984012,
      "grad_norm": 0.8764824163394052,
      "learning_rate": 3.761687995875285e-08,
      "loss": 0.3938,
      "step": 15359
    },
    {
      "epoch": 0.9621196698977434,
      "grad_norm": 0.8120958720789679,
      "learning_rate": 3.749278492498898e-08,
      "loss": 0.3787,
      "step": 15360
    },
    {
      "epoch": 0.9621823078970858,
      "grad_norm": 0.908215980122687,
      "learning_rate": 3.7368894148943625e-08,
      "loss": 0.3706,
      "step": 15361
    },
    {
      "epoch": 0.9622449458964281,
      "grad_norm": 0.9166272470814161,
      "learning_rate": 3.724520763571549e-08,
      "loss": 0.3835,
      "step": 15362
    },
    {
      "epoch": 0.9623075838957704,
      "grad_norm": 0.9350434424488853,
      "learning_rate": 3.712172539039549e-08,
      "loss": 0.3837,
      "step": 15363
    },
    {
      "epoch": 0.9623702218951127,
      "grad_norm": 0.843709482143319,
      "learning_rate": 3.699844741806624e-08,
      "loss": 0.4011,
      "step": 15364
    },
    {
      "epoch": 0.9624328598944549,
      "grad_norm": 0.7926094324984028,
      "learning_rate": 3.6875373723802565e-08,
      "loss": 0.3351,
      "step": 15365
    },
    {
      "epoch": 0.9624954978937973,
      "grad_norm": 0.8551186438211217,
      "learning_rate": 3.675250431266986e-08,
      "loss": 0.3826,
      "step": 15366
    },
    {
      "epoch": 0.9625581358931395,
      "grad_norm": 0.8944623418581544,
      "learning_rate": 3.662983918972518e-08,
      "loss": 0.3777,
      "step": 15367
    },
    {
      "epoch": 0.9626207738924819,
      "grad_norm": 0.9234970121926226,
      "learning_rate": 3.650737836001728e-08,
      "loss": 0.3666,
      "step": 15368
    },
    {
      "epoch": 0.9626834118918242,
      "grad_norm": 0.8329719825031984,
      "learning_rate": 3.638512182858711e-08,
      "loss": 0.3965,
      "step": 15369
    },
    {
      "epoch": 0.9627460498911665,
      "grad_norm": 0.8199981951948601,
      "learning_rate": 3.626306960046621e-08,
      "loss": 0.3828,
      "step": 15370
    },
    {
      "epoch": 0.9628086878905088,
      "grad_norm": 0.8299375185763909,
      "learning_rate": 3.614122168067946e-08,
      "loss": 0.3542,
      "step": 15371
    },
    {
      "epoch": 0.962871325889851,
      "grad_norm": 0.8939190319146617,
      "learning_rate": 3.601957807424061e-08,
      "loss": 0.4338,
      "step": 15372
    },
    {
      "epoch": 0.9629339638891934,
      "grad_norm": 0.6815749487331217,
      "learning_rate": 3.589813878615733e-08,
      "loss": 0.4618,
      "step": 15373
    },
    {
      "epoch": 0.9629966018885356,
      "grad_norm": 0.9020730381513189,
      "learning_rate": 3.57769038214284e-08,
      "loss": 0.3737,
      "step": 15374
    },
    {
      "epoch": 0.963059239887878,
      "grad_norm": 0.8266436130791119,
      "learning_rate": 3.565587318504371e-08,
      "loss": 0.3554,
      "step": 15375
    },
    {
      "epoch": 0.9631218778872203,
      "grad_norm": 0.6137759692116093,
      "learning_rate": 3.553504688198428e-08,
      "loss": 0.4255,
      "step": 15376
    },
    {
      "epoch": 0.9631845158865626,
      "grad_norm": 0.8334345586445071,
      "learning_rate": 3.541442491722447e-08,
      "loss": 0.3622,
      "step": 15377
    },
    {
      "epoch": 0.9632471538859049,
      "grad_norm": 0.8683782199701677,
      "learning_rate": 3.5294007295727515e-08,
      "loss": 0.3821,
      "step": 15378
    },
    {
      "epoch": 0.9633097918852472,
      "grad_norm": 0.8793176052096168,
      "learning_rate": 3.517379402245113e-08,
      "loss": 0.396,
      "step": 15379
    },
    {
      "epoch": 0.9633724298845895,
      "grad_norm": 0.8761870770501404,
      "learning_rate": 3.5053785102342476e-08,
      "loss": 0.377,
      "step": 15380
    },
    {
      "epoch": 0.9634350678839317,
      "grad_norm": 0.885357661471309,
      "learning_rate": 3.4933980540342024e-08,
      "loss": 0.3584,
      "step": 15381
    },
    {
      "epoch": 0.9634977058832741,
      "grad_norm": 0.8179891477091322,
      "learning_rate": 3.4814380341380296e-08,
      "loss": 0.3697,
      "step": 15382
    },
    {
      "epoch": 0.9635603438826164,
      "grad_norm": 0.8541799036182447,
      "learning_rate": 3.469498451038056e-08,
      "loss": 0.3793,
      "step": 15383
    },
    {
      "epoch": 0.9636229818819587,
      "grad_norm": 0.7858610693931908,
      "learning_rate": 3.457579305225667e-08,
      "loss": 0.3598,
      "step": 15384
    },
    {
      "epoch": 0.963685619881301,
      "grad_norm": 0.8251004743259767,
      "learning_rate": 3.44568059719147e-08,
      "loss": 0.3844,
      "step": 15385
    },
    {
      "epoch": 0.9637482578806433,
      "grad_norm": 0.8334686013298904,
      "learning_rate": 3.43380232742524e-08,
      "loss": 0.3824,
      "step": 15386
    },
    {
      "epoch": 0.9638108958799856,
      "grad_norm": 0.9077021886169134,
      "learning_rate": 3.421944496415863e-08,
      "loss": 0.3933,
      "step": 15387
    },
    {
      "epoch": 0.963873533879328,
      "grad_norm": 0.8033782328970996,
      "learning_rate": 3.410107104651394e-08,
      "loss": 0.3561,
      "step": 15388
    },
    {
      "epoch": 0.9639361718786702,
      "grad_norm": 0.8324352618288592,
      "learning_rate": 3.3982901526191106e-08,
      "loss": 0.3535,
      "step": 15389
    },
    {
      "epoch": 0.9639988098780125,
      "grad_norm": 0.9232741009370211,
      "learning_rate": 3.386493640805399e-08,
      "loss": 0.3858,
      "step": 15390
    },
    {
      "epoch": 0.9640614478773548,
      "grad_norm": 0.9762739272316912,
      "learning_rate": 3.3747175696957626e-08,
      "loss": 0.4084,
      "step": 15391
    },
    {
      "epoch": 0.9641240858766971,
      "grad_norm": 0.8459060839581716,
      "learning_rate": 3.362961939774923e-08,
      "loss": 0.4249,
      "step": 15392
    },
    {
      "epoch": 0.9641867238760394,
      "grad_norm": 0.8376993723899859,
      "learning_rate": 3.35122675152677e-08,
      "loss": 0.3599,
      "step": 15393
    },
    {
      "epoch": 0.9642493618753817,
      "grad_norm": 0.8758462410362228,
      "learning_rate": 3.339512005434309e-08,
      "loss": 0.4122,
      "step": 15394
    },
    {
      "epoch": 0.964311999874724,
      "grad_norm": 0.8817189381216336,
      "learning_rate": 3.3278177019797075e-08,
      "loss": 0.3655,
      "step": 15395
    },
    {
      "epoch": 0.9643746378740663,
      "grad_norm": 0.8806962003683689,
      "learning_rate": 3.3161438416443036e-08,
      "loss": 0.3946,
      "step": 15396
    },
    {
      "epoch": 0.9644372758734086,
      "grad_norm": 0.8822860955924249,
      "learning_rate": 3.304490424908602e-08,
      "loss": 0.3788,
      "step": 15397
    },
    {
      "epoch": 0.9644999138727509,
      "grad_norm": 0.8533238922418697,
      "learning_rate": 3.292857452252274e-08,
      "loss": 0.3994,
      "step": 15398
    },
    {
      "epoch": 0.9645625518720932,
      "grad_norm": 0.6066658964466997,
      "learning_rate": 3.2812449241541587e-08,
      "loss": 0.4322,
      "step": 15399
    },
    {
      "epoch": 0.9646251898714355,
      "grad_norm": 0.6146233853024793,
      "learning_rate": 3.2696528410921524e-08,
      "loss": 0.4462,
      "step": 15400
    },
    {
      "epoch": 0.9646878278707778,
      "grad_norm": 0.7844581900298092,
      "learning_rate": 3.258081203543484e-08,
      "loss": 0.4021,
      "step": 15401
    },
    {
      "epoch": 0.9647504658701201,
      "grad_norm": 0.8370720977185571,
      "learning_rate": 3.246530011984328e-08,
      "loss": 0.3853,
      "step": 15402
    },
    {
      "epoch": 0.9648131038694624,
      "grad_norm": 0.9027322019971741,
      "learning_rate": 3.234999266890249e-08,
      "loss": 0.3732,
      "step": 15403
    },
    {
      "epoch": 0.9648757418688048,
      "grad_norm": 0.8719036342776846,
      "learning_rate": 3.223488968735811e-08,
      "loss": 0.3796,
      "step": 15404
    },
    {
      "epoch": 0.964938379868147,
      "grad_norm": 0.8793114166234907,
      "learning_rate": 3.2119991179947464e-08,
      "loss": 0.3685,
      "step": 15405
    },
    {
      "epoch": 0.9650010178674893,
      "grad_norm": 0.9220126505661838,
      "learning_rate": 3.2005297151400105e-08,
      "loss": 0.3355,
      "step": 15406
    },
    {
      "epoch": 0.9650636558668316,
      "grad_norm": 0.8973396057434276,
      "learning_rate": 3.1890807606436705e-08,
      "loss": 0.4094,
      "step": 15407
    },
    {
      "epoch": 0.9651262938661739,
      "grad_norm": 0.9102606546153139,
      "learning_rate": 3.1776522549769594e-08,
      "loss": 0.3805,
      "step": 15408
    },
    {
      "epoch": 0.9651889318655162,
      "grad_norm": 0.837544816045977,
      "learning_rate": 3.166244198610336e-08,
      "loss": 0.3498,
      "step": 15409
    },
    {
      "epoch": 0.9652515698648585,
      "grad_norm": 0.9243009899231045,
      "learning_rate": 3.154856592013311e-08,
      "loss": 0.3638,
      "step": 15410
    },
    {
      "epoch": 0.9653142078642009,
      "grad_norm": 0.8681198020056655,
      "learning_rate": 3.143489435654623e-08,
      "loss": 0.4379,
      "step": 15411
    },
    {
      "epoch": 0.9653768458635431,
      "grad_norm": 0.9028621300032101,
      "learning_rate": 3.132142730002119e-08,
      "loss": 0.3192,
      "step": 15412
    },
    {
      "epoch": 0.9654394838628855,
      "grad_norm": 0.8946054255959653,
      "learning_rate": 3.1208164755228145e-08,
      "loss": 0.3542,
      "step": 15413
    },
    {
      "epoch": 0.9655021218622277,
      "grad_norm": 0.8639932920786096,
      "learning_rate": 3.109510672682947e-08,
      "loss": 0.3799,
      "step": 15414
    },
    {
      "epoch": 0.96556475986157,
      "grad_norm": 0.8902467252533232,
      "learning_rate": 3.0982253219478674e-08,
      "loss": 0.3907,
      "step": 15415
    },
    {
      "epoch": 0.9656273978609123,
      "grad_norm": 0.8245346305680328,
      "learning_rate": 3.086960423782037e-08,
      "loss": 0.3825,
      "step": 15416
    },
    {
      "epoch": 0.9656900358602546,
      "grad_norm": 0.8679844600605278,
      "learning_rate": 3.075715978649141e-08,
      "loss": 0.4119,
      "step": 15417
    },
    {
      "epoch": 0.965752673859597,
      "grad_norm": 0.856254855634409,
      "learning_rate": 3.0644919870120304e-08,
      "loss": 0.3707,
      "step": 15418
    },
    {
      "epoch": 0.9658153118589392,
      "grad_norm": 0.8074088061015086,
      "learning_rate": 3.0532884493326695e-08,
      "loss": 0.3468,
      "step": 15419
    },
    {
      "epoch": 0.9658779498582816,
      "grad_norm": 0.8357063582157184,
      "learning_rate": 3.04210536607219e-08,
      "loss": 0.3903,
      "step": 15420
    },
    {
      "epoch": 0.9659405878576238,
      "grad_norm": 0.8931454113661679,
      "learning_rate": 3.0309427376908893e-08,
      "loss": 0.3637,
      "step": 15421
    },
    {
      "epoch": 0.9660032258569661,
      "grad_norm": 0.8301947860670977,
      "learning_rate": 3.019800564648234e-08,
      "loss": 0.3343,
      "step": 15422
    },
    {
      "epoch": 0.9660658638563084,
      "grad_norm": 0.9024804483825647,
      "learning_rate": 3.0086788474028015e-08,
      "loss": 0.3896,
      "step": 15423
    },
    {
      "epoch": 0.9661285018556507,
      "grad_norm": 0.8719966482887124,
      "learning_rate": 2.997577586412448e-08,
      "loss": 0.3987,
      "step": 15424
    },
    {
      "epoch": 0.966191139854993,
      "grad_norm": 0.8538237712556744,
      "learning_rate": 2.9864967821340296e-08,
      "loss": 0.4214,
      "step": 15425
    },
    {
      "epoch": 0.9662537778543353,
      "grad_norm": 0.5993328661666445,
      "learning_rate": 2.9754364350236265e-08,
      "loss": 0.4196,
      "step": 15426
    },
    {
      "epoch": 0.9663164158536777,
      "grad_norm": 0.8610070933638883,
      "learning_rate": 2.9643965455364852e-08,
      "loss": 0.4037,
      "step": 15427
    },
    {
      "epoch": 0.9663790538530199,
      "grad_norm": 0.9352625631633119,
      "learning_rate": 2.9533771141271316e-08,
      "loss": 0.3709,
      "step": 15428
    },
    {
      "epoch": 0.9664416918523623,
      "grad_norm": 0.6080258045872599,
      "learning_rate": 2.9423781412489804e-08,
      "loss": 0.4405,
      "step": 15429
    },
    {
      "epoch": 0.9665043298517045,
      "grad_norm": 0.9477778289102553,
      "learning_rate": 2.9313996273547807e-08,
      "loss": 0.3948,
      "step": 15430
    },
    {
      "epoch": 0.9665669678510468,
      "grad_norm": 0.6065947703772894,
      "learning_rate": 2.9204415728964487e-08,
      "loss": 0.4312,
      "step": 15431
    },
    {
      "epoch": 0.9666296058503892,
      "grad_norm": 0.8943844212340013,
      "learning_rate": 2.909503978325012e-08,
      "loss": 0.3542,
      "step": 15432
    },
    {
      "epoch": 0.9666922438497314,
      "grad_norm": 0.8946677555776431,
      "learning_rate": 2.8985868440906672e-08,
      "loss": 0.4006,
      "step": 15433
    },
    {
      "epoch": 0.9667548818490738,
      "grad_norm": 0.8372665316120074,
      "learning_rate": 2.8876901706427208e-08,
      "loss": 0.3825,
      "step": 15434
    },
    {
      "epoch": 0.966817519848416,
      "grad_norm": 0.8727088110971003,
      "learning_rate": 2.8768139584297026e-08,
      "loss": 0.3773,
      "step": 15435
    },
    {
      "epoch": 0.9668801578477584,
      "grad_norm": 0.9136520664291765,
      "learning_rate": 2.865958207899311e-08,
      "loss": 0.3876,
      "step": 15436
    },
    {
      "epoch": 0.9669427958471006,
      "grad_norm": 0.882533458865436,
      "learning_rate": 2.8551229194984098e-08,
      "loss": 0.3964,
      "step": 15437
    },
    {
      "epoch": 0.967005433846443,
      "grad_norm": 0.6267922626957882,
      "learning_rate": 2.8443080936728652e-08,
      "loss": 0.4496,
      "step": 15438
    },
    {
      "epoch": 0.9670680718457852,
      "grad_norm": 0.907557982367982,
      "learning_rate": 2.833513730867876e-08,
      "loss": 0.4034,
      "step": 15439
    },
    {
      "epoch": 0.9671307098451275,
      "grad_norm": 0.8789509610846663,
      "learning_rate": 2.8227398315277544e-08,
      "loss": 0.3868,
      "step": 15440
    },
    {
      "epoch": 0.9671933478444699,
      "grad_norm": 0.8489270152978361,
      "learning_rate": 2.8119863960959225e-08,
      "loss": 0.3484,
      "step": 15441
    },
    {
      "epoch": 0.9672559858438121,
      "grad_norm": 0.8417123667107964,
      "learning_rate": 2.8012534250150824e-08,
      "loss": 0.3778,
      "step": 15442
    },
    {
      "epoch": 0.9673186238431545,
      "grad_norm": 0.8876150916811427,
      "learning_rate": 2.7905409187268806e-08,
      "loss": 0.4022,
      "step": 15443
    },
    {
      "epoch": 0.9673812618424967,
      "grad_norm": 0.9295910295455142,
      "learning_rate": 2.7798488776722422e-08,
      "loss": 0.4015,
      "step": 15444
    },
    {
      "epoch": 0.9674438998418391,
      "grad_norm": 0.9109692630299568,
      "learning_rate": 2.7691773022914257e-08,
      "loss": 0.3934,
      "step": 15445
    },
    {
      "epoch": 0.9675065378411813,
      "grad_norm": 0.8817209310005781,
      "learning_rate": 2.7585261930234696e-08,
      "loss": 0.3549,
      "step": 15446
    },
    {
      "epoch": 0.9675691758405236,
      "grad_norm": 0.8156650657128284,
      "learning_rate": 2.7478955503069116e-08,
      "loss": 0.3608,
      "step": 15447
    },
    {
      "epoch": 0.967631813839866,
      "grad_norm": 0.8998186826492154,
      "learning_rate": 2.7372853745792905e-08,
      "loss": 0.3669,
      "step": 15448
    },
    {
      "epoch": 0.9676944518392082,
      "grad_norm": 0.839900093778335,
      "learning_rate": 2.726695666277257e-08,
      "loss": 0.3754,
      "step": 15449
    },
    {
      "epoch": 0.9677570898385506,
      "grad_norm": 1.0325104531530294,
      "learning_rate": 2.7161264258367405e-08,
      "loss": 0.3513,
      "step": 15450
    },
    {
      "epoch": 0.9678197278378928,
      "grad_norm": 0.6063397094832422,
      "learning_rate": 2.7055776536927815e-08,
      "loss": 0.4611,
      "step": 15451
    },
    {
      "epoch": 0.9678823658372352,
      "grad_norm": 0.8449274323131678,
      "learning_rate": 2.6950493502795328e-08,
      "loss": 0.3744,
      "step": 15452
    },
    {
      "epoch": 0.9679450038365774,
      "grad_norm": 0.8895088164223637,
      "learning_rate": 2.68454151603037e-08,
      "loss": 0.3617,
      "step": 15453
    },
    {
      "epoch": 0.9680076418359198,
      "grad_norm": 0.8926202703977455,
      "learning_rate": 2.674054151377725e-08,
      "loss": 0.3769,
      "step": 15454
    },
    {
      "epoch": 0.9680702798352621,
      "grad_norm": 0.8708195415629615,
      "learning_rate": 2.6635872567533638e-08,
      "loss": 0.3817,
      "step": 15455
    },
    {
      "epoch": 0.9681329178346043,
      "grad_norm": 0.8807694805115854,
      "learning_rate": 2.653140832588108e-08,
      "loss": 0.387,
      "step": 15456
    },
    {
      "epoch": 0.9681955558339467,
      "grad_norm": 0.824042730121043,
      "learning_rate": 2.6427148793117808e-08,
      "loss": 0.3682,
      "step": 15457
    },
    {
      "epoch": 0.9682581938332889,
      "grad_norm": 0.8477880614793277,
      "learning_rate": 2.632309397353705e-08,
      "loss": 0.3787,
      "step": 15458
    },
    {
      "epoch": 0.9683208318326313,
      "grad_norm": 0.8147261519998235,
      "learning_rate": 2.621924387142094e-08,
      "loss": 0.3914,
      "step": 15459
    },
    {
      "epoch": 0.9683834698319735,
      "grad_norm": 0.9216044531334834,
      "learning_rate": 2.6115598491043837e-08,
      "loss": 0.3946,
      "step": 15460
    },
    {
      "epoch": 0.9684461078313159,
      "grad_norm": 0.855202580657381,
      "learning_rate": 2.601215783667177e-08,
      "loss": 0.3825,
      "step": 15461
    },
    {
      "epoch": 0.9685087458306582,
      "grad_norm": 0.8577544287478877,
      "learning_rate": 2.5908921912562445e-08,
      "loss": 0.3699,
      "step": 15462
    },
    {
      "epoch": 0.9685713838300005,
      "grad_norm": 0.8526912880306324,
      "learning_rate": 2.5805890722964687e-08,
      "loss": 0.3646,
      "step": 15463
    },
    {
      "epoch": 0.9686340218293428,
      "grad_norm": 0.8772139232915068,
      "learning_rate": 2.5703064272120103e-08,
      "loss": 0.3724,
      "step": 15464
    },
    {
      "epoch": 0.968696659828685,
      "grad_norm": 0.8280154789355384,
      "learning_rate": 2.5600442564260864e-08,
      "loss": 0.3615,
      "step": 15465
    },
    {
      "epoch": 0.9687592978280274,
      "grad_norm": 0.8215037061541222,
      "learning_rate": 2.549802560361081e-08,
      "loss": 0.3752,
      "step": 15466
    },
    {
      "epoch": 0.9688219358273696,
      "grad_norm": 0.876564630560865,
      "learning_rate": 2.5395813394384904e-08,
      "loss": 0.43,
      "step": 15467
    },
    {
      "epoch": 0.968884573826712,
      "grad_norm": 0.8266428880812274,
      "learning_rate": 2.5293805940790894e-08,
      "loss": 0.3727,
      "step": 15468
    },
    {
      "epoch": 0.9689472118260543,
      "grad_norm": 0.8992013061686593,
      "learning_rate": 2.5192003247027086e-08,
      "loss": 0.3616,
      "step": 15469
    },
    {
      "epoch": 0.9690098498253966,
      "grad_norm": 0.8709869401722116,
      "learning_rate": 2.5090405317284015e-08,
      "loss": 0.4017,
      "step": 15470
    },
    {
      "epoch": 0.9690724878247389,
      "grad_norm": 0.8740690988827476,
      "learning_rate": 2.4989012155743343e-08,
      "loss": 0.3349,
      "step": 15471
    },
    {
      "epoch": 0.9691351258240812,
      "grad_norm": 0.8315350072737148,
      "learning_rate": 2.4887823766577835e-08,
      "loss": 0.3689,
      "step": 15472
    },
    {
      "epoch": 0.9691977638234235,
      "grad_norm": 0.8878894635806208,
      "learning_rate": 2.478684015395305e-08,
      "loss": 0.3944,
      "step": 15473
    },
    {
      "epoch": 0.9692604018227657,
      "grad_norm": 0.8588358493992754,
      "learning_rate": 2.468606132202511e-08,
      "loss": 0.3661,
      "step": 15474
    },
    {
      "epoch": 0.9693230398221081,
      "grad_norm": 0.5639330338370152,
      "learning_rate": 2.4585487274942922e-08,
      "loss": 0.4507,
      "step": 15475
    },
    {
      "epoch": 0.9693856778214504,
      "grad_norm": 0.8738309592213866,
      "learning_rate": 2.448511801684539e-08,
      "loss": 0.4056,
      "step": 15476
    },
    {
      "epoch": 0.9694483158207927,
      "grad_norm": 0.8789353709100012,
      "learning_rate": 2.4384953551863655e-08,
      "loss": 0.4265,
      "step": 15477
    },
    {
      "epoch": 0.969510953820135,
      "grad_norm": 0.8530436814725615,
      "learning_rate": 2.428499388412109e-08,
      "loss": 0.3854,
      "step": 15478
    },
    {
      "epoch": 0.9695735918194773,
      "grad_norm": 0.8590354223729002,
      "learning_rate": 2.418523901773162e-08,
      "loss": 0.3817,
      "step": 15479
    },
    {
      "epoch": 0.9696362298188196,
      "grad_norm": 0.8454693754835346,
      "learning_rate": 2.408568895680141e-08,
      "loss": 0.3348,
      "step": 15480
    },
    {
      "epoch": 0.9696988678181618,
      "grad_norm": 0.8304642686624326,
      "learning_rate": 2.398634370542774e-08,
      "loss": 0.3382,
      "step": 15481
    },
    {
      "epoch": 0.9697615058175042,
      "grad_norm": 0.5667523256130544,
      "learning_rate": 2.3887203267699554e-08,
      "loss": 0.4238,
      "step": 15482
    },
    {
      "epoch": 0.9698241438168465,
      "grad_norm": 0.8777421270979004,
      "learning_rate": 2.3788267647697484e-08,
      "loss": 0.3699,
      "step": 15483
    },
    {
      "epoch": 0.9698867818161888,
      "grad_norm": 0.8437603821639976,
      "learning_rate": 2.3689536849494375e-08,
      "loss": 0.3934,
      "step": 15484
    },
    {
      "epoch": 0.9699494198155311,
      "grad_norm": 0.8913104005894753,
      "learning_rate": 2.3591010877153653e-08,
      "loss": 0.3799,
      "step": 15485
    },
    {
      "epoch": 0.9700120578148734,
      "grad_norm": 0.631730538081904,
      "learning_rate": 2.34926897347304e-08,
      "loss": 0.4716,
      "step": 15486
    },
    {
      "epoch": 0.9700746958142157,
      "grad_norm": 0.8520639787986684,
      "learning_rate": 2.3394573426271384e-08,
      "loss": 0.3679,
      "step": 15487
    },
    {
      "epoch": 0.970137333813558,
      "grad_norm": 0.613195926446623,
      "learning_rate": 2.3296661955816146e-08,
      "loss": 0.4389,
      "step": 15488
    },
    {
      "epoch": 0.9701999718129003,
      "grad_norm": 0.8862931508421793,
      "learning_rate": 2.319895532739369e-08,
      "loss": 0.3802,
      "step": 15489
    },
    {
      "epoch": 0.9702626098122425,
      "grad_norm": 0.8023470388298567,
      "learning_rate": 2.3101453545025797e-08,
      "loss": 0.3387,
      "step": 15490
    },
    {
      "epoch": 0.9703252478115849,
      "grad_norm": 0.8326551876503574,
      "learning_rate": 2.300415661272537e-08,
      "loss": 0.3863,
      "step": 15491
    },
    {
      "epoch": 0.9703878858109272,
      "grad_norm": 0.8836691961038243,
      "learning_rate": 2.290706453449809e-08,
      "loss": 0.3962,
      "step": 15492
    },
    {
      "epoch": 0.9704505238102695,
      "grad_norm": 0.8337424327400395,
      "learning_rate": 2.281017731434021e-08,
      "loss": 0.3649,
      "step": 15493
    },
    {
      "epoch": 0.9705131618096118,
      "grad_norm": 0.8810665943089141,
      "learning_rate": 2.271349495623909e-08,
      "loss": 0.3777,
      "step": 15494
    },
    {
      "epoch": 0.9705757998089541,
      "grad_norm": 0.7815184021376865,
      "learning_rate": 2.261701746417433e-08,
      "loss": 0.3025,
      "step": 15495
    },
    {
      "epoch": 0.9706384378082964,
      "grad_norm": 0.8650993625611085,
      "learning_rate": 2.2520744842116637e-08,
      "loss": 0.3546,
      "step": 15496
    },
    {
      "epoch": 0.9707010758076388,
      "grad_norm": 0.8896082669313604,
      "learning_rate": 2.242467709402951e-08,
      "loss": 0.3842,
      "step": 15497
    },
    {
      "epoch": 0.970763713806981,
      "grad_norm": 1.087475607049159,
      "learning_rate": 2.2328814223865902e-08,
      "loss": 0.3781,
      "step": 15498
    },
    {
      "epoch": 0.9708263518063233,
      "grad_norm": 0.9646850477555676,
      "learning_rate": 2.223315623557265e-08,
      "loss": 0.3847,
      "step": 15499
    },
    {
      "epoch": 0.9708889898056656,
      "grad_norm": 0.8156299032973611,
      "learning_rate": 2.2137703133086052e-08,
      "loss": 0.3314,
      "step": 15500
    },
    {
      "epoch": 0.9709516278050079,
      "grad_norm": 0.916430795838837,
      "learning_rate": 2.20424549203363e-08,
      "loss": 0.4117,
      "step": 15501
    },
    {
      "epoch": 0.9710142658043502,
      "grad_norm": 0.9417271406773436,
      "learning_rate": 2.1947411601242475e-08,
      "loss": 0.4319,
      "step": 15502
    },
    {
      "epoch": 0.9710769038036925,
      "grad_norm": 0.8927067960977508,
      "learning_rate": 2.185257317971756e-08,
      "loss": 0.4205,
      "step": 15503
    },
    {
      "epoch": 0.9711395418030349,
      "grad_norm": 0.8656732348886744,
      "learning_rate": 2.175793965966455e-08,
      "loss": 0.33,
      "step": 15504
    },
    {
      "epoch": 0.9712021798023771,
      "grad_norm": 0.8520842694662745,
      "learning_rate": 2.1663511044978658e-08,
      "loss": 0.3819,
      "step": 15505
    },
    {
      "epoch": 0.9712648178017194,
      "grad_norm": 0.8825635187542115,
      "learning_rate": 2.156928733954733e-08,
      "loss": 0.3583,
      "step": 15506
    },
    {
      "epoch": 0.9713274558010617,
      "grad_norm": 0.9205804375801561,
      "learning_rate": 2.1475268547247465e-08,
      "loss": 0.3942,
      "step": 15507
    },
    {
      "epoch": 0.971390093800404,
      "grad_norm": 0.8452031758732282,
      "learning_rate": 2.13814546719493e-08,
      "loss": 0.4157,
      "step": 15508
    },
    {
      "epoch": 0.9714527317997463,
      "grad_norm": 0.5964417940924551,
      "learning_rate": 2.12878457175153e-08,
      "loss": 0.4362,
      "step": 15509
    },
    {
      "epoch": 0.9715153697990886,
      "grad_norm": 0.8441093085017638,
      "learning_rate": 2.1194441687797383e-08,
      "loss": 0.3425,
      "step": 15510
    },
    {
      "epoch": 0.971578007798431,
      "grad_norm": 0.9187044616850477,
      "learning_rate": 2.1101242586640258e-08,
      "loss": 0.4031,
      "step": 15511
    },
    {
      "epoch": 0.9716406457977732,
      "grad_norm": 0.8136106876328638,
      "learning_rate": 2.1008248417880295e-08,
      "loss": 0.3648,
      "step": 15512
    },
    {
      "epoch": 0.9717032837971156,
      "grad_norm": 0.8921511339361118,
      "learning_rate": 2.091545918534499e-08,
      "loss": 0.3638,
      "step": 15513
    },
    {
      "epoch": 0.9717659217964578,
      "grad_norm": 0.8596286761105801,
      "learning_rate": 2.082287489285295e-08,
      "loss": 0.3615,
      "step": 15514
    },
    {
      "epoch": 0.9718285597958001,
      "grad_norm": 0.818539638186491,
      "learning_rate": 2.0730495544216134e-08,
      "loss": 0.3642,
      "step": 15515
    },
    {
      "epoch": 0.9718911977951424,
      "grad_norm": 0.8219156205183785,
      "learning_rate": 2.0638321143236496e-08,
      "loss": 0.3804,
      "step": 15516
    },
    {
      "epoch": 0.9719538357944847,
      "grad_norm": 1.0047880800569389,
      "learning_rate": 2.0546351693707113e-08,
      "loss": 0.3915,
      "step": 15517
    },
    {
      "epoch": 0.972016473793827,
      "grad_norm": 0.8904642228279224,
      "learning_rate": 2.04545871994144e-08,
      "loss": 0.4047,
      "step": 15518
    },
    {
      "epoch": 0.9720791117931693,
      "grad_norm": 0.8555023635490892,
      "learning_rate": 2.0363027664135337e-08,
      "loss": 0.3627,
      "step": 15519
    },
    {
      "epoch": 0.9721417497925117,
      "grad_norm": 0.9354070127846124,
      "learning_rate": 2.0271673091638022e-08,
      "loss": 0.393,
      "step": 15520
    },
    {
      "epoch": 0.9722043877918539,
      "grad_norm": 0.89223724558515,
      "learning_rate": 2.0180523485683336e-08,
      "loss": 0.3732,
      "step": 15521
    },
    {
      "epoch": 0.9722670257911963,
      "grad_norm": 0.8054023595279617,
      "learning_rate": 2.0089578850022717e-08,
      "loss": 0.3817,
      "step": 15522
    },
    {
      "epoch": 0.9723296637905385,
      "grad_norm": 0.8570897917997682,
      "learning_rate": 1.9998839188398733e-08,
      "loss": 0.3813,
      "step": 15523
    },
    {
      "epoch": 0.9723923017898808,
      "grad_norm": 0.908730777849681,
      "learning_rate": 1.990830450454728e-08,
      "loss": 0.3806,
      "step": 15524
    },
    {
      "epoch": 0.9724549397892231,
      "grad_norm": 0.8802477202936226,
      "learning_rate": 1.9817974802193717e-08,
      "loss": 0.4377,
      "step": 15525
    },
    {
      "epoch": 0.9725175777885654,
      "grad_norm": 0.8495482881452687,
      "learning_rate": 1.9727850085057287e-08,
      "loss": 0.3955,
      "step": 15526
    },
    {
      "epoch": 0.9725802157879078,
      "grad_norm": 0.9219715356078818,
      "learning_rate": 1.963793035684669e-08,
      "loss": 0.4025,
      "step": 15527
    },
    {
      "epoch": 0.97264285378725,
      "grad_norm": 0.9438259131276932,
      "learning_rate": 1.9548215621263412e-08,
      "loss": 0.3763,
      "step": 15528
    },
    {
      "epoch": 0.9727054917865924,
      "grad_norm": 0.9613642943391018,
      "learning_rate": 1.945870588200005e-08,
      "loss": 0.4117,
      "step": 15529
    },
    {
      "epoch": 0.9727681297859346,
      "grad_norm": 0.9390714553653292,
      "learning_rate": 1.9369401142740886e-08,
      "loss": 0.3647,
      "step": 15530
    },
    {
      "epoch": 0.9728307677852769,
      "grad_norm": 0.8669143396108595,
      "learning_rate": 1.9280301407161304e-08,
      "loss": 0.3756,
      "step": 15531
    },
    {
      "epoch": 0.9728934057846192,
      "grad_norm": 0.8341619847556833,
      "learning_rate": 1.919140667892949e-08,
      "loss": 0.3942,
      "step": 15532
    },
    {
      "epoch": 0.9729560437839615,
      "grad_norm": 0.8691716066055313,
      "learning_rate": 1.910271696170307e-08,
      "loss": 0.4,
      "step": 15533
    },
    {
      "epoch": 0.9730186817833039,
      "grad_norm": 0.8211138885859116,
      "learning_rate": 1.9014232259134125e-08,
      "loss": 0.3904,
      "step": 15534
    },
    {
      "epoch": 0.9730813197826461,
      "grad_norm": 0.8775086807461668,
      "learning_rate": 1.8925952574863627e-08,
      "loss": 0.3898,
      "step": 15535
    },
    {
      "epoch": 0.9731439577819885,
      "grad_norm": 0.9256680325498727,
      "learning_rate": 1.8837877912524784e-08,
      "loss": 0.4081,
      "step": 15536
    },
    {
      "epoch": 0.9732065957813307,
      "grad_norm": 0.8905936807019417,
      "learning_rate": 1.8750008275744136e-08,
      "loss": 0.3667,
      "step": 15537
    },
    {
      "epoch": 0.9732692337806731,
      "grad_norm": 0.9510096638958044,
      "learning_rate": 1.8662343668137685e-08,
      "loss": 0.4028,
      "step": 15538
    },
    {
      "epoch": 0.9733318717800153,
      "grad_norm": 0.8457618824338141,
      "learning_rate": 1.857488409331365e-08,
      "loss": 0.4045,
      "step": 15539
    },
    {
      "epoch": 0.9733945097793576,
      "grad_norm": 0.8799760017811097,
      "learning_rate": 1.8487629554871932e-08,
      "loss": 0.3944,
      "step": 15540
    },
    {
      "epoch": 0.9734571477787,
      "grad_norm": 0.8618069016081504,
      "learning_rate": 1.8400580056404105e-08,
      "loss": 0.4183,
      "step": 15541
    },
    {
      "epoch": 0.9735197857780422,
      "grad_norm": 0.5884803615992632,
      "learning_rate": 1.8313735601492855e-08,
      "loss": 0.4395,
      "step": 15542
    },
    {
      "epoch": 0.9735824237773846,
      "grad_norm": 0.8395287484676058,
      "learning_rate": 1.8227096193713102e-08,
      "loss": 0.373,
      "step": 15543
    },
    {
      "epoch": 0.9736450617767268,
      "grad_norm": 0.821531578984945,
      "learning_rate": 1.8140661836630325e-08,
      "loss": 0.361,
      "step": 15544
    },
    {
      "epoch": 0.9737076997760692,
      "grad_norm": 0.8409568930122362,
      "learning_rate": 1.8054432533803346e-08,
      "loss": 0.3459,
      "step": 15545
    },
    {
      "epoch": 0.9737703377754114,
      "grad_norm": 0.9055649304991141,
      "learning_rate": 1.7968408288779883e-08,
      "loss": 0.3876,
      "step": 15546
    },
    {
      "epoch": 0.9738329757747538,
      "grad_norm": 0.6152339420307192,
      "learning_rate": 1.7882589105101546e-08,
      "loss": 0.4573,
      "step": 15547
    },
    {
      "epoch": 0.9738956137740961,
      "grad_norm": 0.8202780242161521,
      "learning_rate": 1.7796974986301064e-08,
      "loss": 0.3765,
      "step": 15548
    },
    {
      "epoch": 0.9739582517734383,
      "grad_norm": 0.8751082787434763,
      "learning_rate": 1.7711565935901176e-08,
      "loss": 0.353,
      "step": 15549
    },
    {
      "epoch": 0.9740208897727807,
      "grad_norm": 0.8753208426160325,
      "learning_rate": 1.7626361957418514e-08,
      "loss": 0.3575,
      "step": 15550
    },
    {
      "epoch": 0.9740835277721229,
      "grad_norm": 0.8890187705877944,
      "learning_rate": 1.754136305435916e-08,
      "loss": 0.382,
      "step": 15551
    },
    {
      "epoch": 0.9741461657714653,
      "grad_norm": 0.844943097766584,
      "learning_rate": 1.7456569230222543e-08,
      "loss": 0.3437,
      "step": 15552
    },
    {
      "epoch": 0.9742088037708075,
      "grad_norm": 0.8924217961087381,
      "learning_rate": 1.7371980488498087e-08,
      "loss": 0.3891,
      "step": 15553
    },
    {
      "epoch": 0.9742714417701499,
      "grad_norm": 0.83003332640552,
      "learning_rate": 1.7287596832667453e-08,
      "loss": 0.346,
      "step": 15554
    },
    {
      "epoch": 0.9743340797694922,
      "grad_norm": 0.7852897493908926,
      "learning_rate": 1.720341826620453e-08,
      "loss": 0.3604,
      "step": 15555
    },
    {
      "epoch": 0.9743967177688344,
      "grad_norm": 0.851837898819784,
      "learning_rate": 1.711944479257377e-08,
      "loss": 0.3553,
      "step": 15556
    },
    {
      "epoch": 0.9744593557681768,
      "grad_norm": 0.8602878636442035,
      "learning_rate": 1.7035676415231294e-08,
      "loss": 0.3658,
      "step": 15557
    },
    {
      "epoch": 0.974521993767519,
      "grad_norm": 0.7925543549742983,
      "learning_rate": 1.6952113137625458e-08,
      "loss": 0.3961,
      "step": 15558
    },
    {
      "epoch": 0.9745846317668614,
      "grad_norm": 0.8365403938391225,
      "learning_rate": 1.6868754963195176e-08,
      "loss": 0.3571,
      "step": 15559
    },
    {
      "epoch": 0.9746472697662036,
      "grad_norm": 0.9112453449871689,
      "learning_rate": 1.6785601895372705e-08,
      "loss": 0.4191,
      "step": 15560
    },
    {
      "epoch": 0.974709907765546,
      "grad_norm": 0.8741393108464988,
      "learning_rate": 1.6702653937579194e-08,
      "loss": 0.3478,
      "step": 15561
    },
    {
      "epoch": 0.9747725457648883,
      "grad_norm": 0.6053530384974088,
      "learning_rate": 1.6619911093229135e-08,
      "loss": 0.4542,
      "step": 15562
    },
    {
      "epoch": 0.9748351837642306,
      "grad_norm": 0.8350430491967757,
      "learning_rate": 1.6537373365729246e-08,
      "loss": 0.3901,
      "step": 15563
    },
    {
      "epoch": 0.9748978217635729,
      "grad_norm": 0.894962998401825,
      "learning_rate": 1.645504075847515e-08,
      "loss": 0.4032,
      "step": 15564
    },
    {
      "epoch": 0.9749604597629151,
      "grad_norm": 0.6530313205758047,
      "learning_rate": 1.6372913274856907e-08,
      "loss": 0.4682,
      "step": 15565
    },
    {
      "epoch": 0.9750230977622575,
      "grad_norm": 0.8539551808324553,
      "learning_rate": 1.6290990918255144e-08,
      "loss": 0.3726,
      "step": 15566
    },
    {
      "epoch": 0.9750857357615997,
      "grad_norm": 0.8968540564407895,
      "learning_rate": 1.6209273692040507e-08,
      "loss": 0.3784,
      "step": 15567
    },
    {
      "epoch": 0.9751483737609421,
      "grad_norm": 0.8302701769595529,
      "learning_rate": 1.612776159957752e-08,
      "loss": 0.3507,
      "step": 15568
    },
    {
      "epoch": 0.9752110117602844,
      "grad_norm": 0.8438879591757934,
      "learning_rate": 1.604645464422072e-08,
      "loss": 0.3802,
      "step": 15569
    },
    {
      "epoch": 0.9752736497596267,
      "grad_norm": 0.6503180422545215,
      "learning_rate": 1.5965352829316883e-08,
      "loss": 0.45,
      "step": 15570
    },
    {
      "epoch": 0.975336287758969,
      "grad_norm": 0.9089955107040247,
      "learning_rate": 1.588445615820444e-08,
      "loss": 0.3797,
      "step": 15571
    },
    {
      "epoch": 0.9753989257583113,
      "grad_norm": 0.8957836808093547,
      "learning_rate": 1.580376463421296e-08,
      "loss": 0.4156,
      "step": 15572
    },
    {
      "epoch": 0.9754615637576536,
      "grad_norm": 0.5977263467429681,
      "learning_rate": 1.572327826066311e-08,
      "loss": 0.4258,
      "step": 15573
    },
    {
      "epoch": 0.9755242017569958,
      "grad_norm": 0.888267168689836,
      "learning_rate": 1.5642997040868912e-08,
      "loss": 0.3722,
      "step": 15574
    },
    {
      "epoch": 0.9755868397563382,
      "grad_norm": 0.6279163489864883,
      "learning_rate": 1.5562920978133834e-08,
      "loss": 0.4166,
      "step": 15575
    },
    {
      "epoch": 0.9756494777556804,
      "grad_norm": 0.8952849857472264,
      "learning_rate": 1.548305007575357e-08,
      "loss": 0.3687,
      "step": 15576
    },
    {
      "epoch": 0.9757121157550228,
      "grad_norm": 0.9240450920671383,
      "learning_rate": 1.54033843370166e-08,
      "loss": 0.3781,
      "step": 15577
    },
    {
      "epoch": 0.9757747537543651,
      "grad_norm": 0.8834392641669582,
      "learning_rate": 1.5323923765201975e-08,
      "loss": 0.3572,
      "step": 15578
    },
    {
      "epoch": 0.9758373917537074,
      "grad_norm": 0.8739250808814808,
      "learning_rate": 1.5244668363579296e-08,
      "loss": 0.3437,
      "step": 15579
    },
    {
      "epoch": 0.9759000297530497,
      "grad_norm": 0.8892102058901472,
      "learning_rate": 1.516561813541151e-08,
      "loss": 0.3895,
      "step": 15580
    },
    {
      "epoch": 0.975962667752392,
      "grad_norm": 0.8611910236892352,
      "learning_rate": 1.508677308395212e-08,
      "loss": 0.374,
      "step": 15581
    },
    {
      "epoch": 0.9760253057517343,
      "grad_norm": 0.9207111056590773,
      "learning_rate": 1.5008133212445763e-08,
      "loss": 0.3726,
      "step": 15582
    },
    {
      "epoch": 0.9760879437510765,
      "grad_norm": 0.7288690506337524,
      "learning_rate": 1.4929698524130397e-08,
      "loss": 0.3313,
      "step": 15583
    },
    {
      "epoch": 0.9761505817504189,
      "grad_norm": 0.8409234871754191,
      "learning_rate": 1.4851469022234e-08,
      "loss": 0.3523,
      "step": 15584
    },
    {
      "epoch": 0.9762132197497612,
      "grad_norm": 0.8900219490907414,
      "learning_rate": 1.4773444709976214e-08,
      "loss": 0.358,
      "step": 15585
    },
    {
      "epoch": 0.9762758577491035,
      "grad_norm": 0.8958946516593604,
      "learning_rate": 1.4695625590568918e-08,
      "loss": 0.3784,
      "step": 15586
    },
    {
      "epoch": 0.9763384957484458,
      "grad_norm": 0.8966450500190144,
      "learning_rate": 1.4618011667215104e-08,
      "loss": 0.4144,
      "step": 15587
    },
    {
      "epoch": 0.9764011337477881,
      "grad_norm": 0.8158735874720768,
      "learning_rate": 1.4540602943108883e-08,
      "loss": 0.3551,
      "step": 15588
    },
    {
      "epoch": 0.9764637717471304,
      "grad_norm": 0.9052695838862146,
      "learning_rate": 1.4463399421437152e-08,
      "loss": 0.3871,
      "step": 15589
    },
    {
      "epoch": 0.9765264097464726,
      "grad_norm": 0.7998960761372458,
      "learning_rate": 1.4386401105376812e-08,
      "loss": 0.3487,
      "step": 15590
    },
    {
      "epoch": 0.976589047745815,
      "grad_norm": 0.6128113198090288,
      "learning_rate": 1.4309607998098108e-08,
      "loss": 0.4312,
      "step": 15591
    },
    {
      "epoch": 0.9766516857451573,
      "grad_norm": 0.8511040762239117,
      "learning_rate": 1.423302010276073e-08,
      "loss": 0.3651,
      "step": 15592
    },
    {
      "epoch": 0.9767143237444996,
      "grad_norm": 0.8543717548599701,
      "learning_rate": 1.4156637422518271e-08,
      "loss": 0.3971,
      "step": 15593
    },
    {
      "epoch": 0.9767769617438419,
      "grad_norm": 0.8294673584483591,
      "learning_rate": 1.4080459960513771e-08,
      "loss": 0.375,
      "step": 15594
    },
    {
      "epoch": 0.9768395997431842,
      "grad_norm": 0.9320882995883175,
      "learning_rate": 1.4004487719883054e-08,
      "loss": 0.3787,
      "step": 15595
    },
    {
      "epoch": 0.9769022377425265,
      "grad_norm": 0.8635636842400572,
      "learning_rate": 1.3928720703753063e-08,
      "loss": 0.3766,
      "step": 15596
    },
    {
      "epoch": 0.9769648757418689,
      "grad_norm": 0.9241114003484158,
      "learning_rate": 1.385315891524186e-08,
      "loss": 0.4099,
      "step": 15597
    },
    {
      "epoch": 0.9770275137412111,
      "grad_norm": 0.9311301288361196,
      "learning_rate": 1.3777802357460845e-08,
      "loss": 0.3605,
      "step": 15598
    },
    {
      "epoch": 0.9770901517405534,
      "grad_norm": 0.9418603320027465,
      "learning_rate": 1.3702651033510872e-08,
      "loss": 0.4029,
      "step": 15599
    },
    {
      "epoch": 0.9771527897398957,
      "grad_norm": 0.8618782680595218,
      "learning_rate": 1.362770494648502e-08,
      "loss": 0.3757,
      "step": 15600
    },
    {
      "epoch": 0.977215427739238,
      "grad_norm": 0.8677808109132824,
      "learning_rate": 1.3552964099468046e-08,
      "loss": 0.3869,
      "step": 15601
    },
    {
      "epoch": 0.9772780657385803,
      "grad_norm": 0.8560425397735254,
      "learning_rate": 1.3478428495536933e-08,
      "loss": 0.3934,
      "step": 15602
    },
    {
      "epoch": 0.9773407037379226,
      "grad_norm": 0.9361531458583241,
      "learning_rate": 1.3404098137759225e-08,
      "loss": 0.3611,
      "step": 15603
    },
    {
      "epoch": 0.977403341737265,
      "grad_norm": 0.9034510572696007,
      "learning_rate": 1.3329973029194698e-08,
      "loss": 0.4027,
      "step": 15604
    },
    {
      "epoch": 0.9774659797366072,
      "grad_norm": 0.878136481889851,
      "learning_rate": 1.325605317289369e-08,
      "loss": 0.3539,
      "step": 15605
    },
    {
      "epoch": 0.9775286177359496,
      "grad_norm": 1.0591924827334729,
      "learning_rate": 1.3182338571899322e-08,
      "loss": 0.4058,
      "step": 15606
    },
    {
      "epoch": 0.9775912557352918,
      "grad_norm": 0.9588534355235336,
      "learning_rate": 1.3108829229245279e-08,
      "loss": 0.377,
      "step": 15607
    },
    {
      "epoch": 0.9776538937346341,
      "grad_norm": 0.8834177479042336,
      "learning_rate": 1.3035525147957473e-08,
      "loss": 0.3786,
      "step": 15608
    },
    {
      "epoch": 0.9777165317339764,
      "grad_norm": 0.876007709925796,
      "learning_rate": 1.2962426331052935e-08,
      "loss": 0.3679,
      "step": 15609
    },
    {
      "epoch": 0.9777791697333187,
      "grad_norm": 0.8842457028728595,
      "learning_rate": 1.2889532781540925e-08,
      "loss": 0.382,
      "step": 15610
    },
    {
      "epoch": 0.977841807732661,
      "grad_norm": 0.8740232980589311,
      "learning_rate": 1.2816844502421267e-08,
      "loss": 0.3666,
      "step": 15611
    },
    {
      "epoch": 0.9779044457320033,
      "grad_norm": 0.8960843747571566,
      "learning_rate": 1.274436149668601e-08,
      "loss": 0.3867,
      "step": 15612
    },
    {
      "epoch": 0.9779670837313457,
      "grad_norm": 0.7850856661139668,
      "learning_rate": 1.2672083767318322e-08,
      "loss": 0.3775,
      "step": 15613
    },
    {
      "epoch": 0.9780297217306879,
      "grad_norm": 0.9085293435061709,
      "learning_rate": 1.2600011317293603e-08,
      "loss": 0.4047,
      "step": 15614
    },
    {
      "epoch": 0.9780923597300302,
      "grad_norm": 0.832881731092479,
      "learning_rate": 1.2528144149577814e-08,
      "loss": 0.3292,
      "step": 15615
    },
    {
      "epoch": 0.9781549977293725,
      "grad_norm": 0.8708891428015757,
      "learning_rate": 1.24564822671297e-08,
      "loss": 0.3239,
      "step": 15616
    },
    {
      "epoch": 0.9782176357287148,
      "grad_norm": 0.9168026675385441,
      "learning_rate": 1.2385025672898565e-08,
      "loss": 0.3804,
      "step": 15617
    },
    {
      "epoch": 0.9782802737280571,
      "grad_norm": 0.7776715817357603,
      "learning_rate": 1.2313774369824838e-08,
      "loss": 0.3719,
      "step": 15618
    },
    {
      "epoch": 0.9783429117273994,
      "grad_norm": 0.8571590431524423,
      "learning_rate": 1.2242728360842283e-08,
      "loss": 0.3975,
      "step": 15619
    },
    {
      "epoch": 0.9784055497267418,
      "grad_norm": 0.9133696877367593,
      "learning_rate": 1.2171887648875224e-08,
      "loss": 0.3897,
      "step": 15620
    },
    {
      "epoch": 0.978468187726084,
      "grad_norm": 0.8529467756398598,
      "learning_rate": 1.2101252236838556e-08,
      "loss": 0.3828,
      "step": 15621
    },
    {
      "epoch": 0.9785308257254264,
      "grad_norm": 0.8555327199668895,
      "learning_rate": 1.2030822127640506e-08,
      "loss": 0.4106,
      "step": 15622
    },
    {
      "epoch": 0.9785934637247686,
      "grad_norm": 0.9070382109357943,
      "learning_rate": 1.1960597324179313e-08,
      "loss": 0.3439,
      "step": 15623
    },
    {
      "epoch": 0.9786561017241109,
      "grad_norm": 0.8923716832710369,
      "learning_rate": 1.1890577829345996e-08,
      "loss": 0.4382,
      "step": 15624
    },
    {
      "epoch": 0.9787187397234532,
      "grad_norm": 0.8068648762141828,
      "learning_rate": 1.1820763646022137e-08,
      "loss": 0.3947,
      "step": 15625
    },
    {
      "epoch": 0.9787813777227955,
      "grad_norm": 0.9559946065678869,
      "learning_rate": 1.1751154777081552e-08,
      "loss": 0.4281,
      "step": 15626
    },
    {
      "epoch": 0.9788440157221379,
      "grad_norm": 0.924798318882835,
      "learning_rate": 1.168175122538917e-08,
      "loss": 0.3849,
      "step": 15627
    },
    {
      "epoch": 0.9789066537214801,
      "grad_norm": 0.8798314198646873,
      "learning_rate": 1.1612552993801596e-08,
      "loss": 0.3705,
      "step": 15628
    },
    {
      "epoch": 0.9789692917208225,
      "grad_norm": 0.8538727970230279,
      "learning_rate": 1.154356008516766e-08,
      "loss": 0.3856,
      "step": 15629
    },
    {
      "epoch": 0.9790319297201647,
      "grad_norm": 0.9111834446903977,
      "learning_rate": 1.1474772502326759e-08,
      "loss": 0.3947,
      "step": 15630
    },
    {
      "epoch": 0.9790945677195071,
      "grad_norm": 0.904330173735385,
      "learning_rate": 1.140619024810996e-08,
      "loss": 0.3672,
      "step": 15631
    },
    {
      "epoch": 0.9791572057188493,
      "grad_norm": 0.8227269002406956,
      "learning_rate": 1.1337813325340008e-08,
      "loss": 0.3493,
      "step": 15632
    },
    {
      "epoch": 0.9792198437181916,
      "grad_norm": 0.8978305511933745,
      "learning_rate": 1.1269641736832426e-08,
      "loss": 0.3543,
      "step": 15633
    },
    {
      "epoch": 0.979282481717534,
      "grad_norm": 0.8585198609071293,
      "learning_rate": 1.1201675485391639e-08,
      "loss": 0.3512,
      "step": 15634
    },
    {
      "epoch": 0.9793451197168762,
      "grad_norm": 0.9296273521574732,
      "learning_rate": 1.1133914573815963e-08,
      "loss": 0.4369,
      "step": 15635
    },
    {
      "epoch": 0.9794077577162186,
      "grad_norm": 0.7921724020046694,
      "learning_rate": 1.1066359004894279e-08,
      "loss": 0.3381,
      "step": 15636
    },
    {
      "epoch": 0.9794703957155608,
      "grad_norm": 0.8762220002671272,
      "learning_rate": 1.099900878140714e-08,
      "loss": 0.3464,
      "step": 15637
    },
    {
      "epoch": 0.9795330337149032,
      "grad_norm": 0.9189152918465481,
      "learning_rate": 1.0931863906127327e-08,
      "loss": 0.41,
      "step": 15638
    },
    {
      "epoch": 0.9795956717142454,
      "grad_norm": 0.8515549822919135,
      "learning_rate": 1.0864924381817631e-08,
      "loss": 0.3548,
      "step": 15639
    },
    {
      "epoch": 0.9796583097135877,
      "grad_norm": 0.8324989381447386,
      "learning_rate": 1.0798190211233628e-08,
      "loss": 0.405,
      "step": 15640
    },
    {
      "epoch": 0.9797209477129301,
      "grad_norm": 0.885558003673698,
      "learning_rate": 1.073166139712256e-08,
      "loss": 0.4274,
      "step": 15641
    },
    {
      "epoch": 0.9797835857122723,
      "grad_norm": 0.9912207783986884,
      "learning_rate": 1.0665337942221687e-08,
      "loss": 0.3826,
      "step": 15642
    },
    {
      "epoch": 0.9798462237116147,
      "grad_norm": 0.9080975702256681,
      "learning_rate": 1.0599219849262154e-08,
      "loss": 0.3975,
      "step": 15643
    },
    {
      "epoch": 0.9799088617109569,
      "grad_norm": 0.6036427308774976,
      "learning_rate": 1.0533307120964564e-08,
      "loss": 0.4326,
      "step": 15644
    },
    {
      "epoch": 0.9799714997102993,
      "grad_norm": 0.8223244130040985,
      "learning_rate": 1.0467599760042302e-08,
      "loss": 0.36,
      "step": 15645
    },
    {
      "epoch": 0.9800341377096415,
      "grad_norm": 0.8981063380968806,
      "learning_rate": 1.0402097769199315e-08,
      "loss": 0.3728,
      "step": 15646
    },
    {
      "epoch": 0.9800967757089839,
      "grad_norm": 0.8481180343840593,
      "learning_rate": 1.0336801151132336e-08,
      "loss": 0.3725,
      "step": 15647
    },
    {
      "epoch": 0.9801594137083262,
      "grad_norm": 0.9330074878375821,
      "learning_rate": 1.0271709908529214e-08,
      "loss": 0.3631,
      "step": 15648
    },
    {
      "epoch": 0.9802220517076684,
      "grad_norm": 0.7887449897315326,
      "learning_rate": 1.0206824044068365e-08,
      "loss": 0.3658,
      "step": 15649
    },
    {
      "epoch": 0.9802846897070108,
      "grad_norm": 0.8552279148086137,
      "learning_rate": 1.0142143560420425e-08,
      "loss": 0.3373,
      "step": 15650
    },
    {
      "epoch": 0.980347327706353,
      "grad_norm": 0.8362413674076916,
      "learning_rate": 1.0077668460248269e-08,
      "loss": 0.3968,
      "step": 15651
    },
    {
      "epoch": 0.9804099657056954,
      "grad_norm": 0.9266330645285011,
      "learning_rate": 1.0013398746205328e-08,
      "loss": 0.4053,
      "step": 15652
    },
    {
      "epoch": 0.9804726037050376,
      "grad_norm": 0.9042584235646881,
      "learning_rate": 9.949334420937262e-09,
      "loss": 0.3971,
      "step": 15653
    },
    {
      "epoch": 0.98053524170438,
      "grad_norm": 0.8505433148831163,
      "learning_rate": 9.885475487080853e-09,
      "loss": 0.3455,
      "step": 15654
    },
    {
      "epoch": 0.9805978797037223,
      "grad_norm": 0.8986593424531159,
      "learning_rate": 9.821821947264554e-09,
      "loss": 0.4265,
      "step": 15655
    },
    {
      "epoch": 0.9806605177030646,
      "grad_norm": 0.9165511496350117,
      "learning_rate": 9.758373804107935e-09,
      "loss": 0.3662,
      "step": 15656
    },
    {
      "epoch": 0.9807231557024069,
      "grad_norm": 0.8809559216625207,
      "learning_rate": 9.695131060222796e-09,
      "loss": 0.3695,
      "step": 15657
    },
    {
      "epoch": 0.9807857937017491,
      "grad_norm": 0.8726399305906581,
      "learning_rate": 9.63209371821261e-09,
      "loss": 0.3803,
      "step": 15658
    },
    {
      "epoch": 0.9808484317010915,
      "grad_norm": 0.8386661272588318,
      "learning_rate": 9.569261780671413e-09,
      "loss": 0.3558,
      "step": 15659
    },
    {
      "epoch": 0.9809110697004337,
      "grad_norm": 0.8529367193034719,
      "learning_rate": 9.506635250186025e-09,
      "loss": 0.3774,
      "step": 15660
    },
    {
      "epoch": 0.9809737076997761,
      "grad_norm": 0.889459451132399,
      "learning_rate": 9.444214129333828e-09,
      "loss": 0.406,
      "step": 15661
    },
    {
      "epoch": 0.9810363456991184,
      "grad_norm": 0.8915189010667778,
      "learning_rate": 9.38199842068388e-09,
      "loss": 0.3817,
      "step": 15662
    },
    {
      "epoch": 0.9810989836984607,
      "grad_norm": 0.8268781095474886,
      "learning_rate": 9.319988126797463e-09,
      "loss": 0.3894,
      "step": 15663
    },
    {
      "epoch": 0.981161621697803,
      "grad_norm": 0.9796150500479677,
      "learning_rate": 9.258183250226426e-09,
      "loss": 0.4038,
      "step": 15664
    },
    {
      "epoch": 0.9812242596971452,
      "grad_norm": 0.9300560942956713,
      "learning_rate": 9.196583793514291e-09,
      "loss": 0.4273,
      "step": 15665
    },
    {
      "epoch": 0.9812868976964876,
      "grad_norm": 0.8145899349026362,
      "learning_rate": 9.135189759197915e-09,
      "loss": 0.3731,
      "step": 15666
    },
    {
      "epoch": 0.9813495356958298,
      "grad_norm": 0.9515651635406651,
      "learning_rate": 9.074001149803058e-09,
      "loss": 0.4388,
      "step": 15667
    },
    {
      "epoch": 0.9814121736951722,
      "grad_norm": 0.869825538050117,
      "learning_rate": 9.013017967848813e-09,
      "loss": 0.3936,
      "step": 15668
    },
    {
      "epoch": 0.9814748116945144,
      "grad_norm": 0.823536682965763,
      "learning_rate": 8.952240215845398e-09,
      "loss": 0.3443,
      "step": 15669
    },
    {
      "epoch": 0.9815374496938568,
      "grad_norm": 0.8620898040940296,
      "learning_rate": 8.891667896294143e-09,
      "loss": 0.3745,
      "step": 15670
    },
    {
      "epoch": 0.9816000876931991,
      "grad_norm": 0.9497539657337036,
      "learning_rate": 8.831301011688054e-09,
      "loss": 0.393,
      "step": 15671
    },
    {
      "epoch": 0.9816627256925414,
      "grad_norm": 0.8878284720224674,
      "learning_rate": 8.77113956451181e-09,
      "loss": 0.4035,
      "step": 15672
    },
    {
      "epoch": 0.9817253636918837,
      "grad_norm": 0.8874282268134375,
      "learning_rate": 8.711183557241765e-09,
      "loss": 0.3561,
      "step": 15673
    },
    {
      "epoch": 0.9817880016912259,
      "grad_norm": 0.902420374375134,
      "learning_rate": 8.651432992346498e-09,
      "loss": 0.3544,
      "step": 15674
    },
    {
      "epoch": 0.9818506396905683,
      "grad_norm": 0.893008877038261,
      "learning_rate": 8.591887872284044e-09,
      "loss": 0.3928,
      "step": 15675
    },
    {
      "epoch": 0.9819132776899105,
      "grad_norm": 0.8463650288791528,
      "learning_rate": 8.53254819950633e-09,
      "loss": 0.3663,
      "step": 15676
    },
    {
      "epoch": 0.9819759156892529,
      "grad_norm": 0.826556738610992,
      "learning_rate": 8.473413976455292e-09,
      "loss": 0.3655,
      "step": 15677
    },
    {
      "epoch": 0.9820385536885952,
      "grad_norm": 0.8671281443621475,
      "learning_rate": 8.414485205564538e-09,
      "loss": 0.4183,
      "step": 15678
    },
    {
      "epoch": 0.9821011916879375,
      "grad_norm": 0.8257724485554745,
      "learning_rate": 8.35576188926046e-09,
      "loss": 0.3488,
      "step": 15679
    },
    {
      "epoch": 0.9821638296872798,
      "grad_norm": 0.8270268719301973,
      "learning_rate": 8.297244029960017e-09,
      "loss": 0.3553,
      "step": 15680
    },
    {
      "epoch": 0.9822264676866221,
      "grad_norm": 0.9168854943473149,
      "learning_rate": 8.23893163007128e-09,
      "loss": 0.3968,
      "step": 15681
    },
    {
      "epoch": 0.9822891056859644,
      "grad_norm": 0.8056851845141761,
      "learning_rate": 8.180824691994549e-09,
      "loss": 0.3659,
      "step": 15682
    },
    {
      "epoch": 0.9823517436853066,
      "grad_norm": 0.9478005563804751,
      "learning_rate": 8.122923218121803e-09,
      "loss": 0.3662,
      "step": 15683
    },
    {
      "epoch": 0.982414381684649,
      "grad_norm": 0.843159692498125,
      "learning_rate": 8.06522721083558e-09,
      "loss": 0.3973,
      "step": 15684
    },
    {
      "epoch": 0.9824770196839913,
      "grad_norm": 0.8800040781782155,
      "learning_rate": 8.007736672511756e-09,
      "loss": 0.364,
      "step": 15685
    },
    {
      "epoch": 0.9825396576833336,
      "grad_norm": 0.884877081834232,
      "learning_rate": 7.950451605515664e-09,
      "loss": 0.404,
      "step": 15686
    },
    {
      "epoch": 0.9826022956826759,
      "grad_norm": 0.9056476659731542,
      "learning_rate": 7.893372012205413e-09,
      "loss": 0.4128,
      "step": 15687
    },
    {
      "epoch": 0.9826649336820182,
      "grad_norm": 0.8596870122272942,
      "learning_rate": 7.836497894930795e-09,
      "loss": 0.37,
      "step": 15688
    },
    {
      "epoch": 0.9827275716813605,
      "grad_norm": 0.8950752307603416,
      "learning_rate": 7.779829256032711e-09,
      "loss": 0.4156,
      "step": 15689
    },
    {
      "epoch": 0.9827902096807029,
      "grad_norm": 0.6377708219106584,
      "learning_rate": 7.723366097843188e-09,
      "loss": 0.42,
      "step": 15690
    },
    {
      "epoch": 0.9828528476800451,
      "grad_norm": 0.8796676369755636,
      "learning_rate": 7.667108422686475e-09,
      "loss": 0.4173,
      "step": 15691
    },
    {
      "epoch": 0.9829154856793874,
      "grad_norm": 0.8671558182823077,
      "learning_rate": 7.6110562328785e-09,
      "loss": 0.3591,
      "step": 15692
    },
    {
      "epoch": 0.9829781236787297,
      "grad_norm": 0.8873482495928546,
      "learning_rate": 7.555209530725749e-09,
      "loss": 0.3691,
      "step": 15693
    },
    {
      "epoch": 0.983040761678072,
      "grad_norm": 0.8968811091560122,
      "learning_rate": 7.499568318527495e-09,
      "loss": 0.3566,
      "step": 15694
    },
    {
      "epoch": 0.9831033996774143,
      "grad_norm": 0.8748620524341113,
      "learning_rate": 7.444132598573572e-09,
      "loss": 0.3967,
      "step": 15695
    },
    {
      "epoch": 0.9831660376767566,
      "grad_norm": 0.8527753523600466,
      "learning_rate": 7.38890237314549e-09,
      "loss": 0.3728,
      "step": 15696
    },
    {
      "epoch": 0.983228675676099,
      "grad_norm": 0.8457947559916966,
      "learning_rate": 7.33387764451754e-09,
      "loss": 0.3778,
      "step": 15697
    },
    {
      "epoch": 0.9832913136754412,
      "grad_norm": 0.8321121480296865,
      "learning_rate": 7.279058414953466e-09,
      "loss": 0.3555,
      "step": 15698
    },
    {
      "epoch": 0.9833539516747835,
      "grad_norm": 0.9059040174930564,
      "learning_rate": 7.224444686709798e-09,
      "loss": 0.4044,
      "step": 15699
    },
    {
      "epoch": 0.9834165896741258,
      "grad_norm": 0.870829289661591,
      "learning_rate": 7.17003646203529e-09,
      "loss": 0.3611,
      "step": 15700
    },
    {
      "epoch": 0.9834792276734681,
      "grad_norm": 0.8081086689837998,
      "learning_rate": 7.115833743168155e-09,
      "loss": 0.3615,
      "step": 15701
    },
    {
      "epoch": 0.9835418656728104,
      "grad_norm": 0.6434305177331117,
      "learning_rate": 7.061836532340494e-09,
      "loss": 0.4512,
      "step": 15702
    },
    {
      "epoch": 0.9836045036721527,
      "grad_norm": 0.9300188495597695,
      "learning_rate": 7.0080448317738635e-09,
      "loss": 0.3692,
      "step": 15703
    },
    {
      "epoch": 0.983667141671495,
      "grad_norm": 0.9048127442966296,
      "learning_rate": 6.9544586436831595e-09,
      "loss": 0.3494,
      "step": 15704
    },
    {
      "epoch": 0.9837297796708373,
      "grad_norm": 0.8591749115982118,
      "learning_rate": 6.9010779702738396e-09,
      "loss": 0.3948,
      "step": 15705
    },
    {
      "epoch": 0.9837924176701797,
      "grad_norm": 0.953186398334508,
      "learning_rate": 6.847902813743035e-09,
      "loss": 0.4029,
      "step": 15706
    },
    {
      "epoch": 0.9838550556695219,
      "grad_norm": 0.8962330834975352,
      "learning_rate": 6.794933176278995e-09,
      "loss": 0.3954,
      "step": 15707
    },
    {
      "epoch": 0.9839176936688642,
      "grad_norm": 0.7974492078912802,
      "learning_rate": 6.742169060062198e-09,
      "loss": 0.3773,
      "step": 15708
    },
    {
      "epoch": 0.9839803316682065,
      "grad_norm": 0.9099982042914235,
      "learning_rate": 6.689610467264795e-09,
      "loss": 0.4016,
      "step": 15709
    },
    {
      "epoch": 0.9840429696675488,
      "grad_norm": 0.9074747419752186,
      "learning_rate": 6.637257400049501e-09,
      "loss": 0.4065,
      "step": 15710
    },
    {
      "epoch": 0.9841056076668911,
      "grad_norm": 0.8659101929468286,
      "learning_rate": 6.585109860571814e-09,
      "loss": 0.3493,
      "step": 15711
    },
    {
      "epoch": 0.9841682456662334,
      "grad_norm": 0.9433390146961821,
      "learning_rate": 6.5331678509777955e-09,
      "loss": 0.3766,
      "step": 15712
    },
    {
      "epoch": 0.9842308836655758,
      "grad_norm": 0.8808238406167979,
      "learning_rate": 6.4814313734051785e-09,
      "loss": 0.389,
      "step": 15713
    },
    {
      "epoch": 0.984293521664918,
      "grad_norm": 0.7860007934356348,
      "learning_rate": 6.4299004299839266e-09,
      "loss": 0.3869,
      "step": 15714
    },
    {
      "epoch": 0.9843561596642604,
      "grad_norm": 0.828435842332852,
      "learning_rate": 6.3785750228351205e-09,
      "loss": 0.3896,
      "step": 15715
    },
    {
      "epoch": 0.9844187976636026,
      "grad_norm": 0.816223034010542,
      "learning_rate": 6.327455154070406e-09,
      "loss": 0.4009,
      "step": 15716
    },
    {
      "epoch": 0.9844814356629449,
      "grad_norm": 0.8984023621330234,
      "learning_rate": 6.2765408257947636e-09,
      "loss": 0.4115,
      "step": 15717
    },
    {
      "epoch": 0.9845440736622872,
      "grad_norm": 0.8674962058669398,
      "learning_rate": 6.225832040103741e-09,
      "loss": 0.3713,
      "step": 15718
    },
    {
      "epoch": 0.9846067116616295,
      "grad_norm": 0.8885875435694954,
      "learning_rate": 6.175328799084001e-09,
      "loss": 0.3944,
      "step": 15719
    },
    {
      "epoch": 0.9846693496609719,
      "grad_norm": 0.8441575867712644,
      "learning_rate": 6.125031104814994e-09,
      "loss": 0.3788,
      "step": 15720
    },
    {
      "epoch": 0.9847319876603141,
      "grad_norm": 0.9152786840080539,
      "learning_rate": 6.074938959366172e-09,
      "loss": 0.3703,
      "step": 15721
    },
    {
      "epoch": 0.9847946256596565,
      "grad_norm": 0.9141340008286434,
      "learning_rate": 6.025052364800332e-09,
      "loss": 0.3843,
      "step": 15722
    },
    {
      "epoch": 0.9848572636589987,
      "grad_norm": 0.8777916509678151,
      "learning_rate": 5.975371323169721e-09,
      "loss": 0.3723,
      "step": 15723
    },
    {
      "epoch": 0.984919901658341,
      "grad_norm": 0.8048099505460024,
      "learning_rate": 5.9258958365199234e-09,
      "loss": 0.3242,
      "step": 15724
    },
    {
      "epoch": 0.9849825396576833,
      "grad_norm": 0.8741928431597806,
      "learning_rate": 5.876625906887645e-09,
      "loss": 0.3287,
      "step": 15725
    },
    {
      "epoch": 0.9850451776570256,
      "grad_norm": 0.8433878962811131,
      "learning_rate": 5.827561536300152e-09,
      "loss": 0.3702,
      "step": 15726
    },
    {
      "epoch": 0.985107815656368,
      "grad_norm": 0.9299252542170622,
      "learning_rate": 5.778702726776941e-09,
      "loss": 0.4068,
      "step": 15727
    },
    {
      "epoch": 0.9851704536557102,
      "grad_norm": 0.951320112795922,
      "learning_rate": 5.730049480329181e-09,
      "loss": 0.3954,
      "step": 15728
    },
    {
      "epoch": 0.9852330916550526,
      "grad_norm": 1.042472639574018,
      "learning_rate": 5.6816017989597125e-09,
      "loss": 0.3811,
      "step": 15729
    },
    {
      "epoch": 0.9852957296543948,
      "grad_norm": 0.943603198416904,
      "learning_rate": 5.6333596846630536e-09,
      "loss": 0.4155,
      "step": 15730
    },
    {
      "epoch": 0.9853583676537372,
      "grad_norm": 0.8424967744584754,
      "learning_rate": 5.585323139423726e-09,
      "loss": 0.3795,
      "step": 15731
    },
    {
      "epoch": 0.9854210056530794,
      "grad_norm": 0.7972387661869396,
      "learning_rate": 5.537492165219594e-09,
      "loss": 0.339,
      "step": 15732
    },
    {
      "epoch": 0.9854836436524217,
      "grad_norm": 0.830750046308477,
      "learning_rate": 5.4898667640190806e-09,
      "loss": 0.369,
      "step": 15733
    },
    {
      "epoch": 0.9855462816517641,
      "grad_norm": 0.8109750245674223,
      "learning_rate": 5.442446937783396e-09,
      "loss": 0.3506,
      "step": 15734
    },
    {
      "epoch": 0.9856089196511063,
      "grad_norm": 0.9470091268509853,
      "learning_rate": 5.395232688463203e-09,
      "loss": 0.3994,
      "step": 15735
    },
    {
      "epoch": 0.9856715576504487,
      "grad_norm": 0.8565121105313072,
      "learning_rate": 5.348224018001946e-09,
      "loss": 0.3804,
      "step": 15736
    },
    {
      "epoch": 0.9857341956497909,
      "grad_norm": 0.9266249768624247,
      "learning_rate": 5.301420928335299e-09,
      "loss": 0.3887,
      "step": 15737
    },
    {
      "epoch": 0.9857968336491333,
      "grad_norm": 0.9410806489314764,
      "learning_rate": 5.254823421388944e-09,
      "loss": 0.437,
      "step": 15738
    },
    {
      "epoch": 0.9858594716484755,
      "grad_norm": 0.8737259984570497,
      "learning_rate": 5.208431499081345e-09,
      "loss": 0.3614,
      "step": 15739
    },
    {
      "epoch": 0.9859221096478179,
      "grad_norm": 0.8734879561305627,
      "learning_rate": 5.162245163321533e-09,
      "loss": 0.3307,
      "step": 15740
    },
    {
      "epoch": 0.9859847476471602,
      "grad_norm": 0.81905799540973,
      "learning_rate": 5.116264416011318e-09,
      "loss": 0.3379,
      "step": 15741
    },
    {
      "epoch": 0.9860473856465024,
      "grad_norm": 0.9291448023777427,
      "learning_rate": 5.070489259041966e-09,
      "loss": 0.3975,
      "step": 15742
    },
    {
      "epoch": 0.9861100236458448,
      "grad_norm": 0.8570474496260845,
      "learning_rate": 5.024919694299191e-09,
      "loss": 0.3987,
      "step": 15743
    },
    {
      "epoch": 0.986172661645187,
      "grad_norm": 0.9201394796121775,
      "learning_rate": 4.979555723657048e-09,
      "loss": 0.4051,
      "step": 15744
    },
    {
      "epoch": 0.9862352996445294,
      "grad_norm": 0.8720829858999735,
      "learning_rate": 4.934397348984043e-09,
      "loss": 0.3806,
      "step": 15745
    },
    {
      "epoch": 0.9862979376438716,
      "grad_norm": 0.8793711108022035,
      "learning_rate": 4.889444572138135e-09,
      "loss": 0.3754,
      "step": 15746
    },
    {
      "epoch": 0.986360575643214,
      "grad_norm": 0.8690142296787807,
      "learning_rate": 4.844697394970066e-09,
      "loss": 0.3693,
      "step": 15747
    },
    {
      "epoch": 0.9864232136425563,
      "grad_norm": 0.8912050348289832,
      "learning_rate": 4.800155819321139e-09,
      "loss": 0.3936,
      "step": 15748
    },
    {
      "epoch": 0.9864858516418985,
      "grad_norm": 0.7748203496513177,
      "learning_rate": 4.755819847025445e-09,
      "loss": 0.3261,
      "step": 15749
    },
    {
      "epoch": 0.9865484896412409,
      "grad_norm": 0.9164196365113978,
      "learning_rate": 4.711689479906523e-09,
      "loss": 0.4272,
      "step": 15750
    },
    {
      "epoch": 0.9866111276405831,
      "grad_norm": 0.9050374387545496,
      "learning_rate": 4.667764719782364e-09,
      "loss": 0.3941,
      "step": 15751
    },
    {
      "epoch": 0.9866737656399255,
      "grad_norm": 0.923408800587226,
      "learning_rate": 4.6240455684598566e-09,
      "loss": 0.4132,
      "step": 15752
    },
    {
      "epoch": 0.9867364036392677,
      "grad_norm": 0.820784166818434,
      "learning_rate": 4.580532027739226e-09,
      "loss": 0.3755,
      "step": 15753
    },
    {
      "epoch": 0.9867990416386101,
      "grad_norm": 0.7871304758126157,
      "learning_rate": 4.537224099410709e-09,
      "loss": 0.3498,
      "step": 15754
    },
    {
      "epoch": 0.9868616796379523,
      "grad_norm": 0.8612473031284489,
      "learning_rate": 4.4941217852567666e-09,
      "loss": 0.3792,
      "step": 15755
    },
    {
      "epoch": 0.9869243176372947,
      "grad_norm": 0.9187713638200249,
      "learning_rate": 4.451225087052647e-09,
      "loss": 0.417,
      "step": 15756
    },
    {
      "epoch": 0.986986955636637,
      "grad_norm": 0.8526022489813601,
      "learning_rate": 4.408534006563047e-09,
      "loss": 0.3936,
      "step": 15757
    },
    {
      "epoch": 0.9870495936359792,
      "grad_norm": 0.8461719893880868,
      "learning_rate": 4.3660485455460085e-09,
      "loss": 0.3923,
      "step": 15758
    },
    {
      "epoch": 0.9871122316353216,
      "grad_norm": 0.5895226919409384,
      "learning_rate": 4.3237687057490205e-09,
      "loss": 0.4514,
      "step": 15759
    },
    {
      "epoch": 0.9871748696346638,
      "grad_norm": 0.9109686183319761,
      "learning_rate": 4.281694488912913e-09,
      "loss": 0.3626,
      "step": 15760
    },
    {
      "epoch": 0.9872375076340062,
      "grad_norm": 0.8402383206271725,
      "learning_rate": 4.239825896769634e-09,
      "loss": 0.3949,
      "step": 15761
    },
    {
      "epoch": 0.9873001456333484,
      "grad_norm": 0.8950614234154214,
      "learning_rate": 4.198162931042804e-09,
      "loss": 0.3822,
      "step": 15762
    },
    {
      "epoch": 0.9873627836326908,
      "grad_norm": 0.9261086686947163,
      "learning_rate": 4.156705593446608e-09,
      "loss": 0.4034,
      "step": 15763
    },
    {
      "epoch": 0.9874254216320331,
      "grad_norm": 0.9035513330898188,
      "learning_rate": 4.115453885687459e-09,
      "loss": 0.361,
      "step": 15764
    },
    {
      "epoch": 0.9874880596313754,
      "grad_norm": 0.9114182779973193,
      "learning_rate": 4.074407809463443e-09,
      "loss": 0.3809,
      "step": 15765
    },
    {
      "epoch": 0.9875506976307177,
      "grad_norm": 0.8649086594205042,
      "learning_rate": 4.03356736646432e-09,
      "loss": 0.3493,
      "step": 15766
    },
    {
      "epoch": 0.9876133356300599,
      "grad_norm": 0.8325483153292591,
      "learning_rate": 3.992932558370966e-09,
      "loss": 0.3413,
      "step": 15767
    },
    {
      "epoch": 0.9876759736294023,
      "grad_norm": 0.8411840660583597,
      "learning_rate": 3.9525033868553775e-09,
      "loss": 0.3832,
      "step": 15768
    },
    {
      "epoch": 0.9877386116287445,
      "grad_norm": 0.9143449156635342,
      "learning_rate": 3.912279853582334e-09,
      "loss": 0.4281,
      "step": 15769
    },
    {
      "epoch": 0.9878012496280869,
      "grad_norm": 0.9337943666846527,
      "learning_rate": 3.8722619602071795e-09,
      "loss": 0.3826,
      "step": 15770
    },
    {
      "epoch": 0.9878638876274292,
      "grad_norm": 0.8206452217726375,
      "learning_rate": 3.832449708377484e-09,
      "loss": 0.3611,
      "step": 15771
    },
    {
      "epoch": 0.9879265256267715,
      "grad_norm": 0.902565349038071,
      "learning_rate": 3.7928430997308256e-09,
      "loss": 0.3943,
      "step": 15772
    },
    {
      "epoch": 0.9879891636261138,
      "grad_norm": 0.8291243094527064,
      "learning_rate": 3.7534421358981224e-09,
      "loss": 0.3762,
      "step": 15773
    },
    {
      "epoch": 0.988051801625456,
      "grad_norm": 0.6298659533638411,
      "learning_rate": 3.71424681850141e-09,
      "loss": 0.4651,
      "step": 15774
    },
    {
      "epoch": 0.9881144396247984,
      "grad_norm": 0.8426397692691254,
      "learning_rate": 3.6752571491532883e-09,
      "loss": 0.403,
      "step": 15775
    },
    {
      "epoch": 0.9881770776241406,
      "grad_norm": 0.6047479154031327,
      "learning_rate": 3.6364731294591393e-09,
      "loss": 0.4435,
      "step": 15776
    },
    {
      "epoch": 0.988239715623483,
      "grad_norm": 0.8520292184818723,
      "learning_rate": 3.5978947610154634e-09,
      "loss": 0.3426,
      "step": 15777
    },
    {
      "epoch": 0.9883023536228253,
      "grad_norm": 0.9057964253422695,
      "learning_rate": 3.559522045409325e-09,
      "loss": 0.4003,
      "step": 15778
    },
    {
      "epoch": 0.9883649916221676,
      "grad_norm": 0.8746718887717504,
      "learning_rate": 3.5213549842211257e-09,
      "loss": 0.3623,
      "step": 15779
    },
    {
      "epoch": 0.9884276296215099,
      "grad_norm": 0.9954812706533137,
      "learning_rate": 3.4833935790207217e-09,
      "loss": 0.4053,
      "step": 15780
    },
    {
      "epoch": 0.9884902676208522,
      "grad_norm": 0.8677829435917213,
      "learning_rate": 3.445637831371862e-09,
      "loss": 0.3903,
      "step": 15781
    },
    {
      "epoch": 0.9885529056201945,
      "grad_norm": 0.8570345514246759,
      "learning_rate": 3.408087742827748e-09,
      "loss": 0.3599,
      "step": 15782
    },
    {
      "epoch": 0.9886155436195367,
      "grad_norm": 0.833219055011574,
      "learning_rate": 3.370743314933811e-09,
      "loss": 0.3909,
      "step": 15783
    },
    {
      "epoch": 0.9886781816188791,
      "grad_norm": 0.8753715999556108,
      "learning_rate": 3.333604549227709e-09,
      "loss": 0.4046,
      "step": 15784
    },
    {
      "epoch": 0.9887408196182214,
      "grad_norm": 0.833313072697003,
      "learning_rate": 3.2966714472376647e-09,
      "loss": 0.3956,
      "step": 15785
    },
    {
      "epoch": 0.9888034576175637,
      "grad_norm": 0.8987306396861439,
      "learning_rate": 3.2599440104846837e-09,
      "loss": 0.4178,
      "step": 15786
    },
    {
      "epoch": 0.988866095616906,
      "grad_norm": 0.6631079013220555,
      "learning_rate": 3.223422240479224e-09,
      "loss": 0.4306,
      "step": 15787
    },
    {
      "epoch": 0.9889287336162483,
      "grad_norm": 0.9364052991697798,
      "learning_rate": 3.1871061387256376e-09,
      "loss": 0.4217,
      "step": 15788
    },
    {
      "epoch": 0.9889913716155906,
      "grad_norm": 0.8354459506959928,
      "learning_rate": 3.15099570671773e-09,
      "loss": 0.3451,
      "step": 15789
    },
    {
      "epoch": 0.989054009614933,
      "grad_norm": 0.8830894652108198,
      "learning_rate": 3.115090945942645e-09,
      "loss": 0.3669,
      "step": 15790
    },
    {
      "epoch": 0.9891166476142752,
      "grad_norm": 0.8581580897101101,
      "learning_rate": 3.079391857877534e-09,
      "loss": 0.3897,
      "step": 15791
    },
    {
      "epoch": 0.9891792856136175,
      "grad_norm": 0.8598571408349418,
      "learning_rate": 3.0438984439923325e-09,
      "loss": 0.3525,
      "step": 15792
    },
    {
      "epoch": 0.9892419236129598,
      "grad_norm": 0.809475342345139,
      "learning_rate": 3.008610705748094e-09,
      "loss": 0.3745,
      "step": 15793
    },
    {
      "epoch": 0.9893045616123021,
      "grad_norm": 0.8491539186852464,
      "learning_rate": 2.973528644596435e-09,
      "loss": 0.3759,
      "step": 15794
    },
    {
      "epoch": 0.9893671996116444,
      "grad_norm": 0.8822505104196008,
      "learning_rate": 2.9386522619817557e-09,
      "loss": 0.3879,
      "step": 15795
    },
    {
      "epoch": 0.9894298376109867,
      "grad_norm": 0.8714860457429187,
      "learning_rate": 2.90398155934013e-09,
      "loss": 0.3496,
      "step": 15796
    },
    {
      "epoch": 0.989492475610329,
      "grad_norm": 0.9018800508241017,
      "learning_rate": 2.8695165380981936e-09,
      "loss": 0.4002,
      "step": 15797
    },
    {
      "epoch": 0.9895551136096713,
      "grad_norm": 0.8835032326813157,
      "learning_rate": 2.835257199673702e-09,
      "loss": 0.3576,
      "step": 15798
    },
    {
      "epoch": 0.9896177516090137,
      "grad_norm": 0.6641592371270497,
      "learning_rate": 2.8012035454783037e-09,
      "loss": 0.4515,
      "step": 15799
    },
    {
      "epoch": 0.9896803896083559,
      "grad_norm": 0.938917026809371,
      "learning_rate": 2.767355576912545e-09,
      "loss": 0.4148,
      "step": 15800
    },
    {
      "epoch": 0.9897430276076982,
      "grad_norm": 0.8431167217867624,
      "learning_rate": 2.7337132953697555e-09,
      "loss": 0.3678,
      "step": 15801
    },
    {
      "epoch": 0.9898056656070405,
      "grad_norm": 0.8346776399682705,
      "learning_rate": 2.7002767022354935e-09,
      "loss": 0.3783,
      "step": 15802
    },
    {
      "epoch": 0.9898683036063828,
      "grad_norm": 0.9062714902129058,
      "learning_rate": 2.667045798884771e-09,
      "loss": 0.3845,
      "step": 15803
    },
    {
      "epoch": 0.9899309416057251,
      "grad_norm": 0.9662813106228351,
      "learning_rate": 2.6340205866864922e-09,
      "loss": 0.4166,
      "step": 15804
    },
    {
      "epoch": 0.9899935796050674,
      "grad_norm": 0.8376426042189882,
      "learning_rate": 2.6012010669990153e-09,
      "loss": 0.3991,
      "step": 15805
    },
    {
      "epoch": 0.9900562176044098,
      "grad_norm": 0.8462475320189239,
      "learning_rate": 2.5685872411740363e-09,
      "loss": 0.3572,
      "step": 15806
    },
    {
      "epoch": 0.990118855603752,
      "grad_norm": 0.8340570003029285,
      "learning_rate": 2.5361791105532608e-09,
      "loss": 0.3912,
      "step": 15807
    },
    {
      "epoch": 0.9901814936030943,
      "grad_norm": 0.8817979307877356,
      "learning_rate": 2.503976676471176e-09,
      "loss": 0.38,
      "step": 15808
    },
    {
      "epoch": 0.9902441316024366,
      "grad_norm": 0.9592019195147049,
      "learning_rate": 2.4719799402528333e-09,
      "loss": 0.3861,
      "step": 15809
    },
    {
      "epoch": 0.9903067696017789,
      "grad_norm": 0.8166214539113525,
      "learning_rate": 2.4401889032155125e-09,
      "loss": 0.3632,
      "step": 15810
    },
    {
      "epoch": 0.9903694076011212,
      "grad_norm": 0.9203647560976765,
      "learning_rate": 2.408603566667056e-09,
      "loss": 0.3733,
      "step": 15811
    },
    {
      "epoch": 0.9904320456004635,
      "grad_norm": 0.9020678572040455,
      "learning_rate": 2.3772239319086454e-09,
      "loss": 0.3941,
      "step": 15812
    },
    {
      "epoch": 0.9904946835998059,
      "grad_norm": 0.9353743037919008,
      "learning_rate": 2.34605000023147e-09,
      "loss": 0.3941,
      "step": 15813
    },
    {
      "epoch": 0.9905573215991481,
      "grad_norm": 0.7761563911233004,
      "learning_rate": 2.3150817729178375e-09,
      "loss": 0.3475,
      "step": 15814
    },
    {
      "epoch": 0.9906199595984905,
      "grad_norm": 0.8643438236504578,
      "learning_rate": 2.2843192512433943e-09,
      "loss": 0.3979,
      "step": 15815
    },
    {
      "epoch": 0.9906825975978327,
      "grad_norm": 0.8873630074576946,
      "learning_rate": 2.2537624364743494e-09,
      "loss": 0.4075,
      "step": 15816
    },
    {
      "epoch": 0.990745235597175,
      "grad_norm": 0.9062747866156118,
      "learning_rate": 2.223411329867475e-09,
      "loss": 0.4299,
      "step": 15817
    },
    {
      "epoch": 0.9908078735965173,
      "grad_norm": 0.8642805400551309,
      "learning_rate": 2.1932659326728833e-09,
      "loss": 0.3674,
      "step": 15818
    },
    {
      "epoch": 0.9908705115958596,
      "grad_norm": 0.8767588610884383,
      "learning_rate": 2.1633262461312476e-09,
      "loss": 0.4295,
      "step": 15819
    },
    {
      "epoch": 0.990933149595202,
      "grad_norm": 0.827033500009648,
      "learning_rate": 2.133592271474916e-09,
      "loss": 0.3896,
      "step": 15820
    },
    {
      "epoch": 0.9909957875945442,
      "grad_norm": 0.9128378793926395,
      "learning_rate": 2.1040640099267984e-09,
      "loss": 0.396,
      "step": 15821
    },
    {
      "epoch": 0.9910584255938866,
      "grad_norm": 0.8578925127570678,
      "learning_rate": 2.0747414627037e-09,
      "loss": 0.3576,
      "step": 15822
    },
    {
      "epoch": 0.9911210635932288,
      "grad_norm": 0.8530033755673532,
      "learning_rate": 2.0456246310118776e-09,
      "loss": 0.3748,
      "step": 15823
    },
    {
      "epoch": 0.9911837015925712,
      "grad_norm": 0.8607885979199853,
      "learning_rate": 2.0167135160492623e-09,
      "loss": 0.3939,
      "step": 15824
    },
    {
      "epoch": 0.9912463395919134,
      "grad_norm": 0.9350233211839329,
      "learning_rate": 1.988008119007123e-09,
      "loss": 0.3947,
      "step": 15825
    },
    {
      "epoch": 0.9913089775912557,
      "grad_norm": 0.5429249956489645,
      "learning_rate": 1.9595084410656275e-09,
      "loss": 0.4506,
      "step": 15826
    },
    {
      "epoch": 0.991371615590598,
      "grad_norm": 0.8896771092025705,
      "learning_rate": 1.9312144833988356e-09,
      "loss": 0.3802,
      "step": 15827
    },
    {
      "epoch": 0.9914342535899403,
      "grad_norm": 0.8610015648315894,
      "learning_rate": 1.9031262471708168e-09,
      "loss": 0.4042,
      "step": 15828
    },
    {
      "epoch": 0.9914968915892827,
      "grad_norm": 0.8644374704742545,
      "learning_rate": 1.875243733537868e-09,
      "loss": 0.386,
      "step": 15829
    },
    {
      "epoch": 0.9915595295886249,
      "grad_norm": 0.8833409181714443,
      "learning_rate": 1.8475669436474052e-09,
      "loss": 0.3557,
      "step": 15830
    },
    {
      "epoch": 0.9916221675879673,
      "grad_norm": 0.8108464013480067,
      "learning_rate": 1.8200958786390722e-09,
      "loss": 0.3658,
      "step": 15831
    },
    {
      "epoch": 0.9916848055873095,
      "grad_norm": 0.9154807836716566,
      "learning_rate": 1.7928305396430757e-09,
      "loss": 0.3817,
      "step": 15832
    },
    {
      "epoch": 0.9917474435866518,
      "grad_norm": 0.9550420713415126,
      "learning_rate": 1.7657709277818513e-09,
      "loss": 0.3855,
      "step": 15833
    },
    {
      "epoch": 0.9918100815859942,
      "grad_norm": 0.8599477570927531,
      "learning_rate": 1.738917044169508e-09,
      "loss": 0.3416,
      "step": 15834
    },
    {
      "epoch": 0.9918727195853364,
      "grad_norm": 0.7677895967690527,
      "learning_rate": 1.7122688899107176e-09,
      "loss": 0.3609,
      "step": 15835
    },
    {
      "epoch": 0.9919353575846788,
      "grad_norm": 0.900446743906251,
      "learning_rate": 1.6858264661029356e-09,
      "loss": 0.4283,
      "step": 15836
    },
    {
      "epoch": 0.991997995584021,
      "grad_norm": 0.5624683025814874,
      "learning_rate": 1.6595897738341803e-09,
      "loss": 0.4334,
      "step": 15837
    },
    {
      "epoch": 0.9920606335833634,
      "grad_norm": 0.7901362385848802,
      "learning_rate": 1.6335588141846992e-09,
      "loss": 0.3494,
      "step": 15838
    },
    {
      "epoch": 0.9921232715827056,
      "grad_norm": 0.8609981334671423,
      "learning_rate": 1.607733588225302e-09,
      "loss": 0.348,
      "step": 15839
    },
    {
      "epoch": 0.992185909582048,
      "grad_norm": 0.8393292119427442,
      "learning_rate": 1.5821140970190275e-09,
      "loss": 0.3511,
      "step": 15840
    },
    {
      "epoch": 0.9922485475813903,
      "grad_norm": 0.7887472406999779,
      "learning_rate": 1.5567003416211424e-09,
      "loss": 0.3369,
      "step": 15841
    },
    {
      "epoch": 0.9923111855807325,
      "grad_norm": 0.9328521182277406,
      "learning_rate": 1.531492323077477e-09,
      "loss": 0.4052,
      "step": 15842
    },
    {
      "epoch": 0.9923738235800749,
      "grad_norm": 0.9094711926459806,
      "learning_rate": 1.5064900424249794e-09,
      "loss": 0.3976,
      "step": 15843
    },
    {
      "epoch": 0.9924364615794171,
      "grad_norm": 0.8401269177400257,
      "learning_rate": 1.4816935006928268e-09,
      "loss": 0.4002,
      "step": 15844
    },
    {
      "epoch": 0.9924990995787595,
      "grad_norm": 0.895430832598951,
      "learning_rate": 1.4571026989024239e-09,
      "loss": 0.3651,
      "step": 15845
    },
    {
      "epoch": 0.9925617375781017,
      "grad_norm": 0.7973102548298426,
      "learning_rate": 1.4327176380651842e-09,
      "loss": 0.3202,
      "step": 15846
    },
    {
      "epoch": 0.9926243755774441,
      "grad_norm": 0.8433919468686695,
      "learning_rate": 1.4085383191853042e-09,
      "loss": 0.4005,
      "step": 15847
    },
    {
      "epoch": 0.9926870135767863,
      "grad_norm": 0.8626739151282595,
      "learning_rate": 1.3845647432575438e-09,
      "loss": 0.3996,
      "step": 15848
    },
    {
      "epoch": 0.9927496515761287,
      "grad_norm": 0.8870753854843522,
      "learning_rate": 1.3607969112694464e-09,
      "loss": 0.4203,
      "step": 15849
    },
    {
      "epoch": 0.992812289575471,
      "grad_norm": 0.8517407461024064,
      "learning_rate": 1.3372348241980083e-09,
      "loss": 0.3828,
      "step": 15850
    },
    {
      "epoch": 0.9928749275748132,
      "grad_norm": 0.9046268895178802,
      "learning_rate": 1.3138784830141195e-09,
      "loss": 0.394,
      "step": 15851
    },
    {
      "epoch": 0.9929375655741556,
      "grad_norm": 0.9590993493815972,
      "learning_rate": 1.2907278886786779e-09,
      "loss": 0.3904,
      "step": 15852
    },
    {
      "epoch": 0.9930002035734978,
      "grad_norm": 0.877204983130624,
      "learning_rate": 1.26778304214481e-09,
      "loss": 0.3548,
      "step": 15853
    },
    {
      "epoch": 0.9930628415728402,
      "grad_norm": 0.9061024452937938,
      "learning_rate": 1.2450439443567608e-09,
      "loss": 0.3995,
      "step": 15854
    },
    {
      "epoch": 0.9931254795721824,
      "grad_norm": 0.6408349839480836,
      "learning_rate": 1.222510596250448e-09,
      "loss": 0.44,
      "step": 15855
    },
    {
      "epoch": 0.9931881175715248,
      "grad_norm": 0.7931307495084391,
      "learning_rate": 1.200182998753463e-09,
      "loss": 0.3555,
      "step": 15856
    },
    {
      "epoch": 0.9932507555708671,
      "grad_norm": 0.8863794856453198,
      "learning_rate": 1.1780611527845155e-09,
      "loss": 0.4252,
      "step": 15857
    },
    {
      "epoch": 0.9933133935702093,
      "grad_norm": 0.8949367436310165,
      "learning_rate": 1.156145059254543e-09,
      "loss": 0.3962,
      "step": 15858
    },
    {
      "epoch": 0.9933760315695517,
      "grad_norm": 0.9082998449227617,
      "learning_rate": 1.1344347190656024e-09,
      "loss": 0.4331,
      "step": 15859
    },
    {
      "epoch": 0.9934386695688939,
      "grad_norm": 0.8504021984310107,
      "learning_rate": 1.1129301331114229e-09,
      "loss": 0.367,
      "step": 15860
    },
    {
      "epoch": 0.9935013075682363,
      "grad_norm": 0.936198474146538,
      "learning_rate": 1.091631302276297e-09,
      "loss": 0.3863,
      "step": 15861
    },
    {
      "epoch": 0.9935639455675785,
      "grad_norm": 0.861719912288722,
      "learning_rate": 1.0705382274378561e-09,
      "loss": 0.3564,
      "step": 15862
    },
    {
      "epoch": 0.9936265835669209,
      "grad_norm": 0.852072962622377,
      "learning_rate": 1.0496509094637397e-09,
      "loss": 0.3559,
      "step": 15863
    },
    {
      "epoch": 0.9936892215662632,
      "grad_norm": 0.876843869292355,
      "learning_rate": 1.0289693492138153e-09,
      "loss": 0.4105,
      "step": 15864
    },
    {
      "epoch": 0.9937518595656055,
      "grad_norm": 0.8354435259540086,
      "learning_rate": 1.008493547539624e-09,
      "loss": 0.3768,
      "step": 15865
    },
    {
      "epoch": 0.9938144975649478,
      "grad_norm": 0.8649256712297377,
      "learning_rate": 9.8822350528327e-10,
      "loss": 0.3834,
      "step": 15866
    },
    {
      "epoch": 0.99387713556429,
      "grad_norm": 0.889539304557597,
      "learning_rate": 9.681592232801962e-10,
      "loss": 0.3648,
      "step": 15867
    },
    {
      "epoch": 0.9939397735636324,
      "grad_norm": 0.800924294373022,
      "learning_rate": 9.48300702354743e-10,
      "loss": 0.372,
      "step": 15868
    },
    {
      "epoch": 0.9940024115629746,
      "grad_norm": 0.9188041158144797,
      "learning_rate": 9.286479433257e-10,
      "loss": 0.4096,
      "step": 15869
    },
    {
      "epoch": 0.994065049562317,
      "grad_norm": 0.9650696879235219,
      "learning_rate": 9.092009470013097e-10,
      "loss": 0.3727,
      "step": 15870
    },
    {
      "epoch": 0.9941276875616593,
      "grad_norm": 0.8671018426007665,
      "learning_rate": 8.899597141814875e-10,
      "loss": 0.3619,
      "step": 15871
    },
    {
      "epoch": 0.9941903255610016,
      "grad_norm": 0.8890605365170589,
      "learning_rate": 8.70924245659488e-10,
      "loss": 0.4185,
      "step": 15872
    },
    {
      "epoch": 0.9942529635603439,
      "grad_norm": 0.7773171759868001,
      "learning_rate": 8.520945422180183e-10,
      "loss": 0.3544,
      "step": 15873
    },
    {
      "epoch": 0.9943156015596862,
      "grad_norm": 0.9134803748102297,
      "learning_rate": 8.33470604632014e-10,
      "loss": 0.3668,
      "step": 15874
    },
    {
      "epoch": 0.9943782395590285,
      "grad_norm": 0.9520278810026416,
      "learning_rate": 8.150524336680843e-10,
      "loss": 0.4017,
      "step": 15875
    },
    {
      "epoch": 0.9944408775583707,
      "grad_norm": 0.8493976032568664,
      "learning_rate": 7.968400300845114e-10,
      "loss": 0.3803,
      "step": 15876
    },
    {
      "epoch": 0.9945035155577131,
      "grad_norm": 0.8416761219054876,
      "learning_rate": 7.78833394631251e-10,
      "loss": 0.3667,
      "step": 15877
    },
    {
      "epoch": 0.9945661535570554,
      "grad_norm": 0.8403776541426486,
      "learning_rate": 7.610325280488218e-10,
      "loss": 0.371,
      "step": 15878
    },
    {
      "epoch": 0.9946287915563977,
      "grad_norm": 0.8670656686686318,
      "learning_rate": 7.434374310705261e-10,
      "loss": 0.3887,
      "step": 15879
    },
    {
      "epoch": 0.99469142955574,
      "grad_norm": 0.7656283494808234,
      "learning_rate": 7.260481044202294e-10,
      "loss": 0.3238,
      "step": 15880
    },
    {
      "epoch": 0.9947540675550823,
      "grad_norm": 0.915908528853314,
      "learning_rate": 7.088645488134705e-10,
      "loss": 0.4109,
      "step": 15881
    },
    {
      "epoch": 0.9948167055544246,
      "grad_norm": 0.862428134132297,
      "learning_rate": 6.918867649574612e-10,
      "loss": 0.4025,
      "step": 15882
    },
    {
      "epoch": 0.9948793435537668,
      "grad_norm": 0.9060615681161723,
      "learning_rate": 6.751147535516423e-10,
      "loss": 0.3786,
      "step": 15883
    },
    {
      "epoch": 0.9949419815531092,
      "grad_norm": 0.6845570748085618,
      "learning_rate": 6.585485152860171e-10,
      "loss": 0.4628,
      "step": 15884
    },
    {
      "epoch": 0.9950046195524515,
      "grad_norm": 0.5967242406131008,
      "learning_rate": 6.42188050842818e-10,
      "loss": 0.4274,
      "step": 15885
    },
    {
      "epoch": 0.9950672575517938,
      "grad_norm": 0.7755071539676966,
      "learning_rate": 6.260333608948399e-10,
      "loss": 0.3483,
      "step": 15886
    },
    {
      "epoch": 0.9951298955511361,
      "grad_norm": 0.9481888784863586,
      "learning_rate": 6.100844461071065e-10,
      "loss": 0.4247,
      "step": 15887
    },
    {
      "epoch": 0.9951925335504784,
      "grad_norm": 0.8937275030460762,
      "learning_rate": 5.943413071363147e-10,
      "loss": 0.4007,
      "step": 15888
    },
    {
      "epoch": 0.9952551715498207,
      "grad_norm": 0.6354198814017397,
      "learning_rate": 5.788039446302795e-10,
      "loss": 0.4484,
      "step": 15889
    },
    {
      "epoch": 0.995317809549163,
      "grad_norm": 0.8873598868902375,
      "learning_rate": 5.634723592284897e-10,
      "loss": 0.3904,
      "step": 15890
    },
    {
      "epoch": 0.9953804475485053,
      "grad_norm": 0.8713032447427973,
      "learning_rate": 5.483465515621067e-10,
      "loss": 0.3711,
      "step": 15891
    },
    {
      "epoch": 0.9954430855478476,
      "grad_norm": 0.6443182019052921,
      "learning_rate": 5.334265222539658e-10,
      "loss": 0.416,
      "step": 15892
    },
    {
      "epoch": 0.9955057235471899,
      "grad_norm": 0.8386117865867101,
      "learning_rate": 5.187122719180204e-10,
      "loss": 0.4047,
      "step": 15893
    },
    {
      "epoch": 0.9955683615465322,
      "grad_norm": 0.9170268242975125,
      "learning_rate": 5.042038011598971e-10,
      "loss": 0.4149,
      "step": 15894
    },
    {
      "epoch": 0.9956309995458745,
      "grad_norm": 0.872388306268055,
      "learning_rate": 4.899011105763407e-10,
      "loss": 0.4066,
      "step": 15895
    },
    {
      "epoch": 0.9956936375452168,
      "grad_norm": 0.9087889929451237,
      "learning_rate": 4.758042007563246e-10,
      "loss": 0.3757,
      "step": 15896
    },
    {
      "epoch": 0.9957562755445591,
      "grad_norm": 0.8473053199216071,
      "learning_rate": 4.619130722804954e-10,
      "loss": 0.371,
      "step": 15897
    },
    {
      "epoch": 0.9958189135439014,
      "grad_norm": 0.8803476769833103,
      "learning_rate": 4.482277257200629e-10,
      "loss": 0.4368,
      "step": 15898
    },
    {
      "epoch": 0.9958815515432438,
      "grad_norm": 0.9885195642595558,
      "learning_rate": 4.3474816163902033e-10,
      "loss": 0.4124,
      "step": 15899
    },
    {
      "epoch": 0.995944189542586,
      "grad_norm": 0.873460731724712,
      "learning_rate": 4.214743805908139e-10,
      "loss": 0.3726,
      "step": 15900
    },
    {
      "epoch": 0.9960068275419283,
      "grad_norm": 0.902600009462044,
      "learning_rate": 4.084063831233387e-10,
      "loss": 0.3823,
      "step": 15901
    },
    {
      "epoch": 0.9960694655412706,
      "grad_norm": 0.8957632060248677,
      "learning_rate": 3.9554416977394263e-10,
      "loss": 0.3701,
      "step": 15902
    },
    {
      "epoch": 0.9961321035406129,
      "grad_norm": 0.8336326507065966,
      "learning_rate": 3.8288774107164694e-10,
      "loss": 0.3584,
      "step": 15903
    },
    {
      "epoch": 0.9961947415399552,
      "grad_norm": 0.825635988139314,
      "learning_rate": 3.704370975377014e-10,
      "loss": 0.3956,
      "step": 15904
    },
    {
      "epoch": 0.9962573795392975,
      "grad_norm": 0.892402440139362,
      "learning_rate": 3.581922396844739e-10,
      "loss": 0.4156,
      "step": 15905
    },
    {
      "epoch": 0.9963200175386399,
      "grad_norm": 0.9341090873158647,
      "learning_rate": 3.4615316801600575e-10,
      "loss": 0.397,
      "step": 15906
    },
    {
      "epoch": 0.9963826555379821,
      "grad_norm": 0.8477132451023782,
      "learning_rate": 3.3431988302801144e-10,
      "loss": 0.4034,
      "step": 15907
    },
    {
      "epoch": 0.9964452935373245,
      "grad_norm": 0.9389648247817085,
      "learning_rate": 3.226923852073238e-10,
      "loss": 0.3756,
      "step": 15908
    },
    {
      "epoch": 0.9965079315366667,
      "grad_norm": 0.8999460786625836,
      "learning_rate": 3.11270675032449e-10,
      "loss": 0.3585,
      "step": 15909
    },
    {
      "epoch": 0.996570569536009,
      "grad_norm": 0.9780194891730372,
      "learning_rate": 3.0005475297356647e-10,
      "loss": 0.4496,
      "step": 15910
    },
    {
      "epoch": 0.9966332075353513,
      "grad_norm": 0.9141230604709402,
      "learning_rate": 2.8904461949308405e-10,
      "loss": 0.3397,
      "step": 15911
    },
    {
      "epoch": 0.9966958455346936,
      "grad_norm": 0.9132930374692361,
      "learning_rate": 2.7824027504286253e-10,
      "loss": 0.4403,
      "step": 15912
    },
    {
      "epoch": 0.996758483534036,
      "grad_norm": 0.8494022497390158,
      "learning_rate": 2.6764172006865654e-10,
      "loss": 0.3125,
      "step": 15913
    },
    {
      "epoch": 0.9968211215333782,
      "grad_norm": 0.8507949387116837,
      "learning_rate": 2.572489550062285e-10,
      "loss": 0.3774,
      "step": 15914
    },
    {
      "epoch": 0.9968837595327206,
      "grad_norm": 0.8893871451376961,
      "learning_rate": 2.470619802830143e-10,
      "loss": 0.3821,
      "step": 15915
    },
    {
      "epoch": 0.9969463975320628,
      "grad_norm": 0.8487261968105335,
      "learning_rate": 2.3708079631923344e-10,
      "loss": 0.3833,
      "step": 15916
    },
    {
      "epoch": 0.9970090355314051,
      "grad_norm": 0.9499569412759221,
      "learning_rate": 2.273054035251132e-10,
      "loss": 0.3834,
      "step": 15917
    },
    {
      "epoch": 0.9970716735307474,
      "grad_norm": 0.8989291962079233,
      "learning_rate": 2.1773580230255442e-10,
      "loss": 0.3387,
      "step": 15918
    },
    {
      "epoch": 0.9971343115300897,
      "grad_norm": 0.9183905928142329,
      "learning_rate": 2.0837199304679646e-10,
      "loss": 0.3986,
      "step": 15919
    },
    {
      "epoch": 0.997196949529432,
      "grad_norm": 0.8033370585039532,
      "learning_rate": 1.9921397614197645e-10,
      "loss": 0.3624,
      "step": 15920
    },
    {
      "epoch": 0.9972595875287743,
      "grad_norm": 0.8936073567062089,
      "learning_rate": 1.902617519655703e-10,
      "loss": 0.3826,
      "step": 15921
    },
    {
      "epoch": 0.9973222255281167,
      "grad_norm": 0.8316492841846548,
      "learning_rate": 1.8151532088617195e-10,
      "loss": 0.3737,
      "step": 15922
    },
    {
      "epoch": 0.9973848635274589,
      "grad_norm": 0.8131774409945176,
      "learning_rate": 1.7297468326349375e-10,
      "loss": 0.3643,
      "step": 15923
    },
    {
      "epoch": 0.9974475015268013,
      "grad_norm": 0.8141310808931004,
      "learning_rate": 1.6463983944947638e-10,
      "loss": 0.377,
      "step": 15924
    },
    {
      "epoch": 0.9975101395261435,
      "grad_norm": 0.9016881227244774,
      "learning_rate": 1.565107897866236e-10,
      "loss": 0.3875,
      "step": 15925
    },
    {
      "epoch": 0.9975727775254858,
      "grad_norm": 0.8568613763292107,
      "learning_rate": 1.485875346096677e-10,
      "loss": 0.3697,
      "step": 15926
    },
    {
      "epoch": 0.9976354155248282,
      "grad_norm": 0.9134120729000144,
      "learning_rate": 1.408700742450142e-10,
      "loss": 0.3816,
      "step": 15927
    },
    {
      "epoch": 0.9976980535241704,
      "grad_norm": 0.8596580561203769,
      "learning_rate": 1.3335840901018693e-10,
      "loss": 0.3879,
      "step": 15928
    },
    {
      "epoch": 0.9977606915235128,
      "grad_norm": 0.8768099201517378,
      "learning_rate": 1.2605253921438298e-10,
      "loss": 0.3638,
      "step": 15929
    },
    {
      "epoch": 0.997823329522855,
      "grad_norm": 0.8903927025385795,
      "learning_rate": 1.189524651579177e-10,
      "loss": 0.3874,
      "step": 15930
    },
    {
      "epoch": 0.9978859675221974,
      "grad_norm": 0.8336435099951921,
      "learning_rate": 1.1205818713388994e-10,
      "loss": 0.4045,
      "step": 15931
    },
    {
      "epoch": 0.9979486055215396,
      "grad_norm": 0.8743756739415384,
      "learning_rate": 1.0536970542540659e-10,
      "loss": 0.3804,
      "step": 15932
    },
    {
      "epoch": 0.998011243520882,
      "grad_norm": 0.908754327541972,
      "learning_rate": 9.888702030780295e-11,
      "loss": 0.4114,
      "step": 15933
    },
    {
      "epoch": 0.9980738815202242,
      "grad_norm": 0.8771856333154879,
      "learning_rate": 9.261013204808767e-11,
      "loss": 0.3855,
      "step": 15934
    },
    {
      "epoch": 0.9981365195195665,
      "grad_norm": 0.8523308799604129,
      "learning_rate": 8.653904090438758e-11,
      "loss": 0.3801,
      "step": 15935
    },
    {
      "epoch": 0.9981991575189089,
      "grad_norm": 0.9188546690196592,
      "learning_rate": 8.0673747127058e-11,
      "loss": 0.3494,
      "step": 15936
    },
    {
      "epoch": 0.9982617955182511,
      "grad_norm": 0.9227705571814729,
      "learning_rate": 7.50142509570173e-11,
      "loss": 0.4268,
      "step": 15937
    },
    {
      "epoch": 0.9983244335175935,
      "grad_norm": 0.8646767520239018,
      "learning_rate": 6.956055262796746e-11,
      "loss": 0.4147,
      "step": 15938
    },
    {
      "epoch": 0.9983870715169357,
      "grad_norm": 0.842861575965517,
      "learning_rate": 6.431265236306327e-11,
      "loss": 0.3808,
      "step": 15939
    },
    {
      "epoch": 0.9984497095162781,
      "grad_norm": 0.9218101315939139,
      "learning_rate": 5.927055037935337e-11,
      "loss": 0.4192,
      "step": 15940
    },
    {
      "epoch": 0.9985123475156203,
      "grad_norm": 0.8663234260581287,
      "learning_rate": 5.4434246884449423e-11,
      "loss": 0.3535,
      "step": 15941
    },
    {
      "epoch": 0.9985749855149626,
      "grad_norm": 0.873843351933299,
      "learning_rate": 4.980374207652627e-11,
      "loss": 0.4063,
      "step": 15942
    },
    {
      "epoch": 0.998637623514305,
      "grad_norm": 0.8841067709514158,
      "learning_rate": 4.5379036147097375e-11,
      "loss": 0.3705,
      "step": 15943
    },
    {
      "epoch": 0.9987002615136472,
      "grad_norm": 0.8260284517679759,
      "learning_rate": 4.11601292776842e-11,
      "loss": 0.3692,
      "step": 15944
    },
    {
      "epoch": 0.9987628995129896,
      "grad_norm": 0.894007415895992,
      "learning_rate": 3.7147021642591764e-11,
      "loss": 0.3583,
      "step": 15945
    },
    {
      "epoch": 0.9988255375123318,
      "grad_norm": 0.8777327052772096,
      "learning_rate": 3.3339713406133067e-11,
      "loss": 0.3877,
      "step": 15946
    },
    {
      "epoch": 0.9988881755116742,
      "grad_norm": 0.8769129627307914,
      "learning_rate": 2.9738204725404674e-11,
      "loss": 0.3742,
      "step": 15947
    },
    {
      "epoch": 0.9989508135110164,
      "grad_norm": 0.8455403979131051,
      "learning_rate": 2.6342495749176465e-11,
      "loss": 0.3978,
      "step": 15948
    },
    {
      "epoch": 0.9990134515103588,
      "grad_norm": 0.5901786539196062,
      "learning_rate": 2.3152586616226323e-11,
      "loss": 0.4378,
      "step": 15949
    },
    {
      "epoch": 0.9990760895097011,
      "grad_norm": 0.8335914656626412,
      "learning_rate": 2.0168477458670787e-11,
      "loss": 0.3737,
      "step": 15950
    },
    {
      "epoch": 0.9991387275090433,
      "grad_norm": 0.8479999144983759,
      "learning_rate": 1.7390168398634387e-11,
      "loss": 0.3962,
      "step": 15951
    },
    {
      "epoch": 0.9992013655083857,
      "grad_norm": 0.8480530706392897,
      "learning_rate": 1.481765955102521e-11,
      "loss": 0.3801,
      "step": 15952
    },
    {
      "epoch": 0.9992640035077279,
      "grad_norm": 0.7969139932908254,
      "learning_rate": 1.2450951021869551e-11,
      "loss": 0.3684,
      "step": 15953
    },
    {
      "epoch": 0.9993266415070703,
      "grad_norm": 0.8445161769800401,
      "learning_rate": 1.0290042907756815e-11,
      "loss": 0.357,
      "step": 15954
    },
    {
      "epoch": 0.9993892795064125,
      "grad_norm": 0.9293412490739961,
      "learning_rate": 8.334935298615066e-12,
      "loss": 0.4363,
      "step": 15955
    },
    {
      "epoch": 0.9994519175057549,
      "grad_norm": 0.6858112326885529,
      "learning_rate": 6.5856282743803625e-12,
      "loss": 0.4503,
      "step": 15956
    },
    {
      "epoch": 0.9995145555050972,
      "grad_norm": 0.8584014483871996,
      "learning_rate": 5.04212190666209e-12,
      "loss": 0.3943,
      "step": 15957
    },
    {
      "epoch": 0.9995771935044395,
      "grad_norm": 0.8576438971321227,
      "learning_rate": 3.704416259853183e-12,
      "loss": 0.4347,
      "step": 15958
    },
    {
      "epoch": 0.9996398315037818,
      "grad_norm": 0.8606148090560346,
      "learning_rate": 2.572511388909682e-12,
      "loss": 0.376,
      "step": 15959
    },
    {
      "epoch": 0.999702469503124,
      "grad_norm": 0.8634623360478229,
      "learning_rate": 1.6464073399058422e-12,
      "loss": 0.3389,
      "step": 15960
    },
    {
      "epoch": 0.9997651075024664,
      "grad_norm": 0.8731434209772633,
      "learning_rate": 9.261041505892465e-13,
      "loss": 0.3737,
      "step": 15961
    },
    {
      "epoch": 0.9998277455018086,
      "grad_norm": 0.9143078052733866,
      "learning_rate": 4.116018520461396e-13,
      "loss": 0.3844,
      "step": 15962
    },
    {
      "epoch": 0.999890383501151,
      "grad_norm": 0.8724485662472965,
      "learning_rate": 1.0290046426053579e-13,
      "loss": 0.416,
      "step": 15963
    },
    {
      "epoch": 0.9999530215004933,
      "grad_norm": 0.8785715791011841,
      "learning_rate": 0.0,
      "loss": 0.3779,
      "step": 15964
    },
    {
      "epoch": 0.9999530215004933,
      "step": 15964,
      "total_flos": 3.566379079447347e+16,
      "train_loss": 0.41283787322302745,
      "train_runtime": 284241.4636,
      "train_samples_per_second": 14.379,
      "train_steps_per_second": 0.056
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 15964,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.566379079447347e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}