{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 7416,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 6.961409543494689,
      "learning_rate": 4.4843049327354265e-08,
      "loss": 0.8591,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.068084644073217,
      "learning_rate": 8.968609865470853e-08,
      "loss": 1.099,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.941786739776187,
      "learning_rate": 1.345291479820628e-07,
      "loss": 1.0761,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 9.844383322574918,
      "learning_rate": 1.7937219730941706e-07,
      "loss": 1.1539,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.09350044210741,
      "learning_rate": 2.242152466367713e-07,
      "loss": 0.9987,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.592161425352053,
      "learning_rate": 2.690582959641256e-07,
      "loss": 1.039,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.459160420196962,
      "learning_rate": 3.1390134529147985e-07,
      "loss": 1.2088,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.468300267755148,
      "learning_rate": 3.587443946188341e-07,
      "loss": 0.9747,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 10.725265287675171,
      "learning_rate": 4.0358744394618834e-07,
      "loss": 1.1631,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.614116795270108,
      "learning_rate": 4.484304932735426e-07,
      "loss": 0.8901,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.888840674966046,
      "learning_rate": 4.932735426008969e-07,
      "loss": 1.094,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.206303377815786,
      "learning_rate": 5.381165919282512e-07,
      "loss": 1.0112,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.792146338016799,
      "learning_rate": 5.829596412556054e-07,
      "loss": 0.7926,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.8311357062959,
      "learning_rate": 6.278026905829597e-07,
      "loss": 0.965,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.939883264359239,
      "learning_rate": 6.72645739910314e-07,
      "loss": 1.0251,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.400161933235832,
      "learning_rate": 7.174887892376682e-07,
      "loss": 0.7672,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 7.621402600324801,
      "learning_rate": 7.623318385650225e-07,
      "loss": 0.85,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.136979902527208,
      "learning_rate": 8.071748878923767e-07,
      "loss": 0.7351,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.527326714055755,
      "learning_rate": 8.520179372197309e-07,
      "loss": 0.9788,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.905855395979164,
      "learning_rate": 8.968609865470852e-07,
      "loss": 0.7862,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.7159391121966365,
      "learning_rate": 9.417040358744395e-07,
      "loss": 0.9564,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.8938294796245176,
      "learning_rate": 9.865470852017938e-07,
      "loss": 0.7469,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.158361979606944,
      "learning_rate": 1.031390134529148e-06,
      "loss": 0.8207,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.687072887480626,
      "learning_rate": 1.0762331838565023e-06,
      "loss": 0.8532,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.889524073390678,
      "learning_rate": 1.1210762331838566e-06,
      "loss": 0.8989,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.768646204745344,
      "learning_rate": 1.1659192825112108e-06,
      "loss": 0.7812,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.450274844398882,
      "learning_rate": 1.2107623318385651e-06,
      "loss": 0.7407,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.35880205830437,
      "learning_rate": 1.2556053811659194e-06,
      "loss": 0.8789,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6434319724699327,
      "learning_rate": 1.3004484304932734e-06,
      "loss": 0.7134,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.491302933344519,
      "learning_rate": 1.345291479820628e-06,
      "loss": 0.6748,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.486955338343044,
      "learning_rate": 1.3901345291479822e-06,
      "loss": 0.6111,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1220225135847337,
      "learning_rate": 1.4349775784753365e-06,
      "loss": 0.57,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.418863625050277,
      "learning_rate": 1.4798206278026907e-06,
      "loss": 0.6577,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.254919857449252,
      "learning_rate": 1.524663677130045e-06,
      "loss": 0.5638,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1827366992385073,
      "learning_rate": 1.5695067264573993e-06,
      "loss": 0.5709,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.6307313821819416,
      "learning_rate": 1.6143497757847533e-06,
      "loss": 0.492,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.095405927759096,
      "learning_rate": 1.6591928251121078e-06,
      "loss": 0.591,
      "step": 37
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.629493475882606,
      "learning_rate": 1.7040358744394619e-06,
      "loss": 0.5228,
      "step": 38
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.739268094489607,
      "learning_rate": 1.7488789237668164e-06,
      "loss": 0.6156,
      "step": 39
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.418246360799165,
      "learning_rate": 1.7937219730941704e-06,
      "loss": 0.5223,
      "step": 40
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.188004255868745,
      "learning_rate": 1.838565022421525e-06,
      "loss": 0.4484,
      "step": 41
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.605795229116246,
      "learning_rate": 1.883408071748879e-06,
      "loss": 0.4042,
      "step": 42
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7151507534770265,
      "learning_rate": 1.9282511210762332e-06,
      "loss": 0.3793,
      "step": 43
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.746434734814048,
      "learning_rate": 1.9730941704035875e-06,
      "loss": 0.4389,
      "step": 44
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2295817103355273,
      "learning_rate": 2.0179372197309418e-06,
      "loss": 0.5164,
      "step": 45
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.962318967375179,
      "learning_rate": 2.062780269058296e-06,
      "loss": 0.3965,
      "step": 46
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7677124619961733,
      "learning_rate": 2.1076233183856503e-06,
      "loss": 0.4133,
      "step": 47
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4849910302839833,
      "learning_rate": 2.1524663677130046e-06,
      "loss": 0.4494,
      "step": 48
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9730215264257955,
      "learning_rate": 2.197309417040359e-06,
      "loss": 0.4307,
      "step": 49
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6068354909105764,
      "learning_rate": 2.242152466367713e-06,
      "loss": 0.3631,
      "step": 50
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0922184334079565,
      "learning_rate": 2.2869955156950674e-06,
      "loss": 0.4451,
      "step": 51
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8591269434613555,
      "learning_rate": 2.3318385650224217e-06,
      "loss": 0.3658,
      "step": 52
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2797960321742563,
      "learning_rate": 2.376681614349776e-06,
      "loss": 0.3505,
      "step": 53
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.596905622131219,
      "learning_rate": 2.4215246636771302e-06,
      "loss": 0.5093,
      "step": 54
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.700171031134967,
      "learning_rate": 2.4663677130044845e-06,
      "loss": 0.2772,
      "step": 55
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1163525959543166,
      "learning_rate": 2.5112107623318388e-06,
      "loss": 0.2988,
      "step": 56
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.496120876159893,
      "learning_rate": 2.556053811659193e-06,
      "loss": 0.3441,
      "step": 57
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.546834612729583,
      "learning_rate": 2.600896860986547e-06,
      "loss": 0.3386,
      "step": 58
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2419601237122335,
      "learning_rate": 2.6457399103139016e-06,
      "loss": 0.2807,
      "step": 59
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3107924512252596,
      "learning_rate": 2.690582959641256e-06,
      "loss": 0.313,
      "step": 60
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.496453228752521,
      "learning_rate": 2.73542600896861e-06,
      "loss": 0.3055,
      "step": 61
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0463382787433124,
      "learning_rate": 2.7802690582959644e-06,
      "loss": 0.3541,
      "step": 62
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.690907094947088,
      "learning_rate": 2.8251121076233182e-06,
      "loss": 0.3353,
      "step": 63
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1264772682160493,
      "learning_rate": 2.869955156950673e-06,
      "loss": 0.2437,
      "step": 64
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8401565711421655,
      "learning_rate": 2.9147982062780272e-06,
      "loss": 0.2736,
      "step": 65
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.221969204817513,
      "learning_rate": 2.9596412556053815e-06,
      "loss": 0.2217,
      "step": 66
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1095757292501194,
      "learning_rate": 3.0044843049327353e-06,
      "loss": 0.2711,
      "step": 67
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0277920994978915,
      "learning_rate": 3.04932735426009e-06,
      "loss": 0.3734,
      "step": 68
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8289773855576463,
      "learning_rate": 3.0941704035874443e-06,
      "loss": 0.3792,
      "step": 69
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5028507825166764,
      "learning_rate": 3.1390134529147986e-06,
      "loss": 0.2905,
      "step": 70
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.897239618094786,
      "learning_rate": 3.1838565022421524e-06,
      "loss": 0.3039,
      "step": 71
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4808480664321655,
      "learning_rate": 3.2286995515695067e-06,
      "loss": 0.4212,
      "step": 72
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5149257788658907,
      "learning_rate": 3.2735426008968614e-06,
      "loss": 0.2721,
      "step": 73
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4600353525625276,
      "learning_rate": 3.3183856502242157e-06,
      "loss": 0.2782,
      "step": 74
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.111174897963261,
      "learning_rate": 3.36322869955157e-06,
      "loss": 0.2229,
      "step": 75
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9992985602119315,
      "learning_rate": 3.4080717488789238e-06,
      "loss": 0.3687,
      "step": 76
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4686620697217605,
      "learning_rate": 3.4529147982062785e-06,
      "loss": 0.3111,
      "step": 77
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.814584129116945,
      "learning_rate": 3.4977578475336327e-06,
      "loss": 0.3572,
      "step": 78
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7459466690987364,
      "learning_rate": 3.542600896860987e-06,
      "loss": 0.3403,
      "step": 79
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3057967553595926,
      "learning_rate": 3.587443946188341e-06,
      "loss": 0.2565,
      "step": 80
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5626531722547643,
      "learning_rate": 3.632286995515695e-06,
      "loss": 0.2246,
      "step": 81
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.360419973333545,
      "learning_rate": 3.67713004484305e-06,
      "loss": 0.2656,
      "step": 82
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.453904543177983,
      "learning_rate": 3.721973094170404e-06,
      "loss": 0.2447,
      "step": 83
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8835718492060942,
      "learning_rate": 3.766816143497758e-06,
      "loss": 0.2396,
      "step": 84
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9444137072583336,
      "learning_rate": 3.8116591928251122e-06,
      "loss": 0.3215,
      "step": 85
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7751327018071743,
      "learning_rate": 3.8565022421524665e-06,
      "loss": 0.3939,
      "step": 86
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3860572916595557,
      "learning_rate": 3.901345291479821e-06,
      "loss": 0.2778,
      "step": 87
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6806292080717355,
      "learning_rate": 3.946188340807175e-06,
      "loss": 0.2364,
      "step": 88
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.081092879300505,
      "learning_rate": 3.991031390134529e-06,
      "loss": 0.3548,
      "step": 89
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.543568588808836,
      "learning_rate": 4.0358744394618836e-06,
      "loss": 0.2606,
      "step": 90
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2190104932864942,
      "learning_rate": 4.080717488789238e-06,
      "loss": 0.2614,
      "step": 91
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4159374908237456,
      "learning_rate": 4.125560538116592e-06,
      "loss": 0.2797,
      "step": 92
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6082523444748706,
      "learning_rate": 4.170403587443946e-06,
      "loss": 0.2433,
      "step": 93
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7818833291542697,
      "learning_rate": 4.215246636771301e-06,
      "loss": 0.3054,
      "step": 94
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9645064726475108,
      "learning_rate": 4.260089686098655e-06,
      "loss": 0.2821,
      "step": 95
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.663596686478882,
      "learning_rate": 4.304932735426009e-06,
      "loss": 0.3153,
      "step": 96
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6104707094091344,
      "learning_rate": 4.3497757847533635e-06,
      "loss": 0.3359,
      "step": 97
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8603439030442848,
      "learning_rate": 4.394618834080718e-06,
      "loss": 0.2436,
      "step": 98
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6621420800562126,
      "learning_rate": 4.439461883408072e-06,
      "loss": 0.3327,
      "step": 99
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0925493123087064,
      "learning_rate": 4.484304932735426e-06,
      "loss": 0.2192,
      "step": 100
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8071009490243366,
      "learning_rate": 4.5291479820627806e-06,
      "loss": 0.3077,
      "step": 101
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0571185565408006,
      "learning_rate": 4.573991031390135e-06,
      "loss": 0.273,
      "step": 102
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6497470961537766,
      "learning_rate": 4.618834080717489e-06,
      "loss": 0.2297,
      "step": 103
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3540212264446443,
      "learning_rate": 4.663677130044843e-06,
      "loss": 0.3003,
      "step": 104
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1615276173877853,
      "learning_rate": 4.708520179372198e-06,
      "loss": 0.3016,
      "step": 105
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.568472143549945,
      "learning_rate": 4.753363228699552e-06,
      "loss": 0.2282,
      "step": 106
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.437880086425516,
      "learning_rate": 4.798206278026906e-06,
      "loss": 0.3083,
      "step": 107
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8185885523088223,
      "learning_rate": 4.8430493273542605e-06,
      "loss": 0.3101,
      "step": 108
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2116078289171206,
      "learning_rate": 4.887892376681615e-06,
      "loss": 0.3381,
      "step": 109
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.953589416411613,
      "learning_rate": 4.932735426008969e-06,
      "loss": 0.335,
      "step": 110
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9084195525703147,
      "learning_rate": 4.977578475336323e-06,
      "loss": 0.2497,
      "step": 111
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2077006114218745,
      "learning_rate": 5.0224215246636775e-06,
      "loss": 0.3045,
      "step": 112
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7196869276496902,
      "learning_rate": 5.067264573991032e-06,
      "loss": 0.3098,
      "step": 113
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8003794492223448,
      "learning_rate": 5.112107623318386e-06,
      "loss": 0.2376,
      "step": 114
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.541887240072164,
      "learning_rate": 5.156950672645741e-06,
      "loss": 0.2303,
      "step": 115
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.775535699771745,
      "learning_rate": 5.201793721973094e-06,
      "loss": 0.2962,
      "step": 116
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2369783940136547,
      "learning_rate": 5.246636771300448e-06,
      "loss": 0.3025,
      "step": 117
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.936039697479903,
      "learning_rate": 5.291479820627803e-06,
      "loss": 0.2821,
      "step": 118
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.295683882205222,
      "learning_rate": 5.3363228699551574e-06,
      "loss": 0.2634,
      "step": 119
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.463269896680059,
      "learning_rate": 5.381165919282512e-06,
      "loss": 0.2511,
      "step": 120
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9155294041839315,
      "learning_rate": 5.426008968609866e-06,
      "loss": 0.2486,
      "step": 121
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.149292071012473,
      "learning_rate": 5.47085201793722e-06,
      "loss": 0.29,
      "step": 122
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8986799260999123,
      "learning_rate": 5.5156950672645745e-06,
      "loss": 0.303,
      "step": 123
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1657464085233005,
      "learning_rate": 5.560538116591929e-06,
      "loss": 0.1877,
      "step": 124
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7711689261146724,
      "learning_rate": 5.605381165919282e-06,
      "loss": 0.2675,
      "step": 125
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.8914018818273926,
      "learning_rate": 5.6502242152466365e-06,
      "loss": 0.387,
      "step": 126
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.481214637729105,
      "learning_rate": 5.695067264573992e-06,
      "loss": 0.269,
      "step": 127
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7842042412253984,
      "learning_rate": 5.739910313901346e-06,
      "loss": 0.2433,
      "step": 128
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3026826941412377,
      "learning_rate": 5.7847533632287e-06,
      "loss": 0.2672,
      "step": 129
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4022390833270566,
      "learning_rate": 5.8295964125560544e-06,
      "loss": 0.2869,
      "step": 130
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7904132628693517,
      "learning_rate": 5.874439461883409e-06,
      "loss": 0.259,
      "step": 131
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8044619424831707,
      "learning_rate": 5.919282511210763e-06,
      "loss": 0.25,
      "step": 132
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4255913568856267,
      "learning_rate": 5.964125560538116e-06,
      "loss": 0.265,
      "step": 133
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0291063511899736,
      "learning_rate": 6.008968609865471e-06,
      "loss": 0.3023,
      "step": 134
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3883695202049116,
      "learning_rate": 6.053811659192825e-06,
      "loss": 0.3329,
      "step": 135
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2425213260821186,
      "learning_rate": 6.09865470852018e-06,
      "loss": 0.3011,
      "step": 136
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.577468197202622,
      "learning_rate": 6.143497757847534e-06,
      "loss": 0.2857,
      "step": 137
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9505023405746649,
      "learning_rate": 6.188340807174889e-06,
      "loss": 0.2512,
      "step": 138
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9174410504185457,
      "learning_rate": 6.233183856502243e-06,
      "loss": 0.3387,
      "step": 139
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.659139544780493,
      "learning_rate": 6.278026905829597e-06,
      "loss": 0.2404,
      "step": 140
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4558934220466284,
      "learning_rate": 6.322869955156951e-06,
      "loss": 0.2488,
      "step": 141
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.467739413080413,
      "learning_rate": 6.367713004484305e-06,
      "loss": 0.2952,
      "step": 142
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.702802622313577,
      "learning_rate": 6.412556053811659e-06,
      "loss": 0.334,
      "step": 143
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.2429223067638295,
      "learning_rate": 6.457399103139013e-06,
      "loss": 0.2968,
      "step": 144
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.615223526042839,
      "learning_rate": 6.5022421524663685e-06,
      "loss": 0.3092,
      "step": 145
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5360609876376374,
      "learning_rate": 6.547085201793723e-06,
      "loss": 0.2025,
      "step": 146
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7282131885833096,
      "learning_rate": 6.591928251121077e-06,
      "loss": 0.2747,
      "step": 147
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6363072940619485,
      "learning_rate": 6.636771300448431e-06,
      "loss": 0.1785,
      "step": 148
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0409717817954833,
      "learning_rate": 6.681614349775786e-06,
      "loss": 0.2736,
      "step": 149
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6137213891497093,
      "learning_rate": 6.72645739910314e-06,
      "loss": 0.2592,
      "step": 150
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.640043553290742,
      "learning_rate": 6.771300448430493e-06,
      "loss": 0.1957,
      "step": 151
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8345339737183508,
      "learning_rate": 6.8161434977578476e-06,
      "loss": 0.2164,
      "step": 152
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4326610388544014,
      "learning_rate": 6.860986547085202e-06,
      "loss": 0.1953,
      "step": 153
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4478610105960303,
      "learning_rate": 6.905829596412557e-06,
      "loss": 0.3009,
      "step": 154
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3105090364615077,
      "learning_rate": 6.950672645739911e-06,
      "loss": 0.2315,
      "step": 155
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1987814846843543,
      "learning_rate": 6.9955156950672655e-06,
      "loss": 0.2538,
      "step": 156
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1768223083464213,
      "learning_rate": 7.04035874439462e-06,
      "loss": 0.2331,
      "step": 157
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3479652613152466,
      "learning_rate": 7.085201793721974e-06,
      "loss": 0.2877,
      "step": 158
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8276359584731827,
      "learning_rate": 7.1300448430493275e-06,
      "loss": 0.1907,
      "step": 159
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1714420546352255,
      "learning_rate": 7.174887892376682e-06,
      "loss": 0.1761,
      "step": 160
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9223107789084368,
      "learning_rate": 7.219730941704036e-06,
      "loss": 0.2236,
      "step": 161
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.520605201217695,
      "learning_rate": 7.26457399103139e-06,
      "loss": 0.3357,
      "step": 162
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.9097719869420424,
      "learning_rate": 7.309417040358745e-06,
      "loss": 0.2593,
      "step": 163
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.665940728919098,
      "learning_rate": 7.3542600896861e-06,
      "loss": 0.2504,
      "step": 164
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7444838082590635,
      "learning_rate": 7.399103139013454e-06,
      "loss": 0.3103,
      "step": 165
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8187393339692315,
      "learning_rate": 7.443946188340808e-06,
      "loss": 0.2677,
      "step": 166
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.844815553592905,
      "learning_rate": 7.4887892376681625e-06,
      "loss": 0.3531,
      "step": 167
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.7668500648529997,
      "learning_rate": 7.533632286995516e-06,
      "loss": 0.3038,
      "step": 168
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1052511655238337,
      "learning_rate": 7.57847533632287e-06,
      "loss": 0.224,
      "step": 169
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0415114310460765,
      "learning_rate": 7.6233183856502244e-06,
      "loss": 0.256,
      "step": 170
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.191152471820726,
      "learning_rate": 7.668161434977579e-06,
      "loss": 0.2239,
      "step": 171
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2401681133610962,
      "learning_rate": 7.713004484304933e-06,
      "loss": 0.344,
      "step": 172
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.050543909404076,
      "learning_rate": 7.757847533632287e-06,
      "loss": 0.2066,
      "step": 173
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5733128556297302,
      "learning_rate": 7.802690582959642e-06,
      "loss": 0.3081,
      "step": 174
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.5121724141014234,
      "learning_rate": 7.847533632286996e-06,
      "loss": 0.2907,
      "step": 175
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6801017120514927,
      "learning_rate": 7.89237668161435e-06,
      "loss": 0.2174,
      "step": 176
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.499366090232887,
      "learning_rate": 7.937219730941704e-06,
      "loss": 0.273,
      "step": 177
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9568557909043316,
      "learning_rate": 7.982062780269059e-06,
      "loss": 0.2662,
      "step": 178
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2088414107310737,
      "learning_rate": 8.026905829596413e-06,
      "loss": 0.2331,
      "step": 179
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6586286777710597,
      "learning_rate": 8.071748878923767e-06,
      "loss": 0.2327,
      "step": 180
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8215508582413624,
      "learning_rate": 8.116591928251121e-06,
      "loss": 0.2277,
      "step": 181
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.654059183135072,
      "learning_rate": 8.161434977578476e-06,
      "loss": 0.2532,
      "step": 182
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.289756691551343,
      "learning_rate": 8.20627802690583e-06,
      "loss": 0.2594,
      "step": 183
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4906872230177646,
      "learning_rate": 8.251121076233184e-06,
      "loss": 0.2947,
      "step": 184
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9832491736935196,
      "learning_rate": 8.295964125560539e-06,
      "loss": 0.2262,
      "step": 185
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.430939187625257,
      "learning_rate": 8.340807174887893e-06,
      "loss": 0.3593,
      "step": 186
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5461961843246408,
      "learning_rate": 8.385650224215247e-06,
      "loss": 0.2304,
      "step": 187
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0420044991618953,
      "learning_rate": 8.430493273542601e-06,
      "loss": 0.2617,
      "step": 188
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8740054401534418,
      "learning_rate": 8.475336322869956e-06,
      "loss": 0.2424,
      "step": 189
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.822324084291057,
      "learning_rate": 8.52017937219731e-06,
      "loss": 0.2629,
      "step": 190
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7853180985661834,
      "learning_rate": 8.565022421524664e-06,
      "loss": 0.2517,
      "step": 191
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3183986305253357,
      "learning_rate": 8.609865470852018e-06,
      "loss": 0.2764,
      "step": 192
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6789634213745477,
      "learning_rate": 8.654708520179373e-06,
      "loss": 0.3098,
      "step": 193
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4444426052756074,
      "learning_rate": 8.699551569506727e-06,
      "loss": 0.3149,
      "step": 194
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.292089656469051,
      "learning_rate": 8.744394618834081e-06,
      "loss": 0.2926,
      "step": 195
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9203836657183857,
      "learning_rate": 8.789237668161435e-06,
      "loss": 0.2272,
      "step": 196
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8171435529592985,
      "learning_rate": 8.83408071748879e-06,
      "loss": 0.2755,
      "step": 197
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9031160499986202,
      "learning_rate": 8.878923766816144e-06,
      "loss": 0.2606,
      "step": 198
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9530622775833888,
      "learning_rate": 8.923766816143498e-06,
      "loss": 0.2449,
      "step": 199
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8573170046143632,
      "learning_rate": 8.968609865470853e-06,
      "loss": 0.1901,
      "step": 200
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.6874023778401095,
      "learning_rate": 9.013452914798207e-06,
      "loss": 0.3546,
      "step": 201
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.77592713845347,
      "learning_rate": 9.058295964125561e-06,
      "loss": 0.1726,
      "step": 202
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8105018174732679,
      "learning_rate": 9.103139013452915e-06,
      "loss": 0.2174,
      "step": 203
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.794961898159075,
      "learning_rate": 9.14798206278027e-06,
      "loss": 0.2589,
      "step": 204
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4385366865865028,
      "learning_rate": 9.192825112107624e-06,
      "loss": 0.2057,
      "step": 205
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8309089344658787,
      "learning_rate": 9.237668161434978e-06,
      "loss": 0.2681,
      "step": 206
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6736911383932145,
      "learning_rate": 9.282511210762332e-06,
      "loss": 0.2702,
      "step": 207
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9314495912494731,
      "learning_rate": 9.327354260089687e-06,
      "loss": 0.2222,
      "step": 208
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5325444478730132,
      "learning_rate": 9.372197309417041e-06,
      "loss": 0.2779,
      "step": 209
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5443990700131134,
      "learning_rate": 9.417040358744395e-06,
      "loss": 0.2265,
      "step": 210
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.888794336782419,
      "learning_rate": 9.46188340807175e-06,
      "loss": 0.3335,
      "step": 211
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.954889568901051,
      "learning_rate": 9.506726457399104e-06,
      "loss": 0.257,
      "step": 212
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8308611161918678,
      "learning_rate": 9.551569506726458e-06,
      "loss": 0.2576,
      "step": 213
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0759907623622724,
      "learning_rate": 9.596412556053812e-06,
      "loss": 0.2163,
      "step": 214
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8176402875834878,
      "learning_rate": 9.641255605381167e-06,
      "loss": 0.229,
      "step": 215
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.44157232659259,
      "learning_rate": 9.686098654708521e-06,
      "loss": 0.2632,
      "step": 216
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7942102162349531,
      "learning_rate": 9.730941704035875e-06,
      "loss": 0.2699,
      "step": 217
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.231235980210026,
      "learning_rate": 9.77578475336323e-06,
      "loss": 0.2432,
      "step": 218
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6398446274201475,
      "learning_rate": 9.820627802690584e-06,
      "loss": 0.2061,
      "step": 219
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6349659132551297,
      "learning_rate": 9.865470852017938e-06,
      "loss": 0.2447,
      "step": 220
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5550928457536628,
      "learning_rate": 9.910313901345292e-06,
      "loss": 0.1976,
      "step": 221
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.181677200264685,
      "learning_rate": 9.955156950672647e-06,
      "loss": 0.2927,
      "step": 222
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6120243112253345,
      "learning_rate": 1e-05,
      "loss": 0.1586,
      "step": 223
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9924114535776216,
      "learning_rate": 9.999999523108452e-06,
      "loss": 0.2887,
      "step": 224
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0894719193365305,
      "learning_rate": 9.999998092433895e-06,
      "loss": 0.2744,
      "step": 225
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5989556139791503,
      "learning_rate": 9.999995707976604e-06,
      "loss": 0.2127,
      "step": 226
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1262758013819947,
      "learning_rate": 9.999992369737033e-06,
      "loss": 0.2917,
      "step": 227
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5905678764346425,
      "learning_rate": 9.999988077715818e-06,
      "loss": 0.214,
      "step": 228
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.15835165506946,
      "learning_rate": 9.999982831913782e-06,
      "loss": 0.2837,
      "step": 229
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7842417291879207,
      "learning_rate": 9.99997663233192e-06,
      "loss": 0.2883,
      "step": 230
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0653189680375243,
      "learning_rate": 9.999969478971417e-06,
      "loss": 0.2661,
      "step": 231
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8459535640565652,
      "learning_rate": 9.99996137183364e-06,
      "loss": 0.3145,
      "step": 232
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7749214911155626,
      "learning_rate": 9.999952310920131e-06,
      "loss": 0.2597,
      "step": 233
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5128693360122796,
      "learning_rate": 9.999942296232621e-06,
      "loss": 0.183,
      "step": 234
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7433859225248813,
      "learning_rate": 9.99993132777302e-06,
      "loss": 0.2389,
      "step": 235
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.585073179242758,
      "learning_rate": 9.99991940554342e-06,
      "loss": 0.1926,
      "step": 236
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8802756284747646,
      "learning_rate": 9.999906529546095e-06,
      "loss": 0.2862,
      "step": 237
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5384526797782068,
      "learning_rate": 9.999892699783503e-06,
      "loss": 0.2781,
      "step": 238
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8642545331989269,
      "learning_rate": 9.999877916258281e-06,
      "loss": 0.3026,
      "step": 239
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.04361780170625,
      "learning_rate": 9.999862178973247e-06,
      "loss": 0.3555,
      "step": 240
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9043164162627273,
      "learning_rate": 9.999845487931408e-06,
      "loss": 0.2813,
      "step": 241
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1751266435095817,
      "learning_rate": 9.999827843135943e-06,
      "loss": 0.3021,
      "step": 242
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3220494514857908,
      "learning_rate": 9.999809244590218e-06,
      "loss": 0.1975,
      "step": 243
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6315256961242937,
      "learning_rate": 9.999789692297784e-06,
      "loss": 0.2055,
      "step": 244
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8473766816186759,
      "learning_rate": 9.99976918626237e-06,
      "loss": 0.2116,
      "step": 245
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5653854329017143,
      "learning_rate": 9.999747726487886e-06,
      "loss": 0.2939,
      "step": 246
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0260234726995585,
      "learning_rate": 9.999725312978425e-06,
      "loss": 0.2578,
      "step": 247
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3419959701078072,
      "learning_rate": 9.999701945738265e-06,
      "loss": 0.1868,
      "step": 248
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4331765765822781,
      "learning_rate": 9.999677624771863e-06,
      "loss": 0.168,
      "step": 249
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5875474454318506,
      "learning_rate": 9.999652350083857e-06,
      "loss": 0.2768,
      "step": 250
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8908075270161602,
      "learning_rate": 9.99962612167907e-06,
      "loss": 0.3171,
      "step": 251
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0341180351908017,
      "learning_rate": 9.999598939562504e-06,
      "loss": 0.1806,
      "step": 252
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.19610614525716,
      "learning_rate": 9.999570803739345e-06,
      "loss": 0.2998,
      "step": 253
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.402794713998738,
      "learning_rate": 9.99954171421496e-06,
      "loss": 0.2432,
      "step": 254
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.443677642576484,
      "learning_rate": 9.999511670994896e-06,
      "loss": 0.235,
      "step": 255
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7974153153026966,
      "learning_rate": 9.999480674084888e-06,
      "loss": 0.372,
      "step": 256
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2856386450969266,
      "learning_rate": 9.999448723490843e-06,
      "loss": 0.2879,
      "step": 257
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2183505841820113,
      "learning_rate": 9.999415819218861e-06,
      "loss": 0.2792,
      "step": 258
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.433519945852409,
      "learning_rate": 9.999381961275216e-06,
      "loss": 0.2019,
      "step": 259
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.57799090847337,
      "learning_rate": 9.99934714966637e-06,
      "loss": 0.2473,
      "step": 260
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.496041380836511,
      "learning_rate": 9.99931138439896e-06,
      "loss": 0.2297,
      "step": 261
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9745405493926677,
      "learning_rate": 9.999274665479809e-06,
      "loss": 0.2005,
      "step": 262
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9655756245615328,
      "learning_rate": 9.999236992915922e-06,
      "loss": 0.3022,
      "step": 263
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9461083237394081,
      "learning_rate": 9.999198366714485e-06,
      "loss": 0.3281,
      "step": 264
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.899600839348888,
      "learning_rate": 9.999158786882867e-06,
      "loss": 0.2318,
      "step": 265
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2288886890359425,
      "learning_rate": 9.999118253428617e-06,
      "loss": 0.1953,
      "step": 266
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6640935984124947,
      "learning_rate": 9.999076766359468e-06,
      "loss": 0.2347,
      "step": 267
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.544789589801104,
      "learning_rate": 9.999034325683333e-06,
      "loss": 0.2402,
      "step": 268
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.881110241137654,
      "learning_rate": 9.998990931408308e-06,
      "loss": 0.2095,
      "step": 269
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8184328507696852,
      "learning_rate": 9.998946583542672e-06,
      "loss": 0.313,
      "step": 270
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7721575965896967,
      "learning_rate": 9.998901282094883e-06,
      "loss": 0.3206,
      "step": 271
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.139067096516454,
      "learning_rate": 9.998855027073586e-06,
      "loss": 0.2649,
      "step": 272
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4537166662870942,
      "learning_rate": 9.998807818487599e-06,
      "loss": 0.1638,
      "step": 273
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.495237901826271,
      "learning_rate": 9.998759656345932e-06,
      "loss": 0.2586,
      "step": 274
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5806342598303946,
      "learning_rate": 9.998710540657768e-06,
      "loss": 0.1744,
      "step": 275
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4512517605741175,
      "learning_rate": 9.99866047143248e-06,
      "loss": 0.1907,
      "step": 276
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4472971853731498,
      "learning_rate": 9.998609448679616e-06,
      "loss": 0.2577,
      "step": 277
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6395978441901917,
      "learning_rate": 9.998557472408911e-06,
      "loss": 0.2502,
      "step": 278
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.70537504309961,
      "learning_rate": 9.99850454263028e-06,
      "loss": 0.2755,
      "step": 279
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8898276285007134,
      "learning_rate": 9.99845065935382e-06,
      "loss": 0.2898,
      "step": 280
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9504813519265933,
      "learning_rate": 9.998395822589806e-06,
      "loss": 0.3641,
      "step": 281
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4579689203817288,
      "learning_rate": 9.998340032348703e-06,
      "loss": 0.2369,
      "step": 282
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0055649136156424,
      "learning_rate": 9.998283288641151e-06,
      "loss": 0.273,
      "step": 283
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8462271798284726,
      "learning_rate": 9.998225591477974e-06,
      "loss": 0.2876,
      "step": 284
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6589213308191293,
      "learning_rate": 9.99816694087018e-06,
      "loss": 0.2603,
      "step": 285
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5182852409983394,
      "learning_rate": 9.998107336828954e-06,
      "loss": 0.2427,
      "step": 286
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2635381391566969,
      "learning_rate": 9.99804677936567e-06,
      "loss": 0.2676,
      "step": 287
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7797379434569023,
      "learning_rate": 9.997985268491874e-06,
      "loss": 0.1931,
      "step": 288
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.707134688319711,
      "learning_rate": 9.997922804219306e-06,
      "loss": 0.2405,
      "step": 289
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1521884398002293,
      "learning_rate": 9.997859386559876e-06,
      "loss": 0.1743,
      "step": 290
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3890897992669649,
      "learning_rate": 9.997795015525686e-06,
      "loss": 0.2339,
      "step": 291
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.352852933551124,
      "learning_rate": 9.997729691129012e-06,
      "loss": 0.2378,
      "step": 292
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.071963399059025,
      "learning_rate": 9.997663413382315e-06,
      "loss": 0.3095,
      "step": 293
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.207985718454822,
      "learning_rate": 9.997596182298241e-06,
      "loss": 0.3492,
      "step": 294
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6634815658840405,
      "learning_rate": 9.997527997889611e-06,
      "loss": 0.2711,
      "step": 295
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6254177297715333,
      "learning_rate": 9.997458860169434e-06,
      "loss": 0.2297,
      "step": 296
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5309648035970025,
      "learning_rate": 9.9973887691509e-06,
      "loss": 0.2252,
      "step": 297
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3698161173226695,
      "learning_rate": 9.997317724847374e-06,
      "loss": 0.2925,
      "step": 298
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8259868019911907,
      "learning_rate": 9.997245727272412e-06,
      "loss": 0.3152,
      "step": 299
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9052739844820998,
      "learning_rate": 9.997172776439747e-06,
      "loss": 0.3229,
      "step": 300
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.582574565974949,
      "learning_rate": 9.997098872363297e-06,
      "loss": 0.3965,
      "step": 301
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.55261765202827,
      "learning_rate": 9.997024015057156e-06,
      "loss": 0.2669,
      "step": 302
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7879625771782097,
      "learning_rate": 9.996948204535606e-06,
      "loss": 0.2599,
      "step": 303
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7051280891993243,
      "learning_rate": 9.996871440813108e-06,
      "loss": 0.2746,
      "step": 304
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7556892722099415,
      "learning_rate": 9.996793723904306e-06,
      "loss": 0.2813,
      "step": 305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7318041224805814,
      "learning_rate": 9.996715053824022e-06,
      "loss": 0.2794,
      "step": 306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8637752792292228,
      "learning_rate": 9.996635430587267e-06,
      "loss": 0.2781,
      "step": 307
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9445128637961269,
      "learning_rate": 9.996554854209225e-06,
      "loss": 0.3044,
      "step": 308
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.209350183363098,
      "learning_rate": 9.99647332470527e-06,
      "loss": 0.2262,
      "step": 309
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.055116622389113,
      "learning_rate": 9.996390842090955e-06,
      "loss": 0.3078,
      "step": 310
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6803359154522632,
      "learning_rate": 9.996307406382009e-06,
      "loss": 0.2933,
      "step": 311
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.700049683806981,
      "learning_rate": 9.996223017594354e-06,
      "loss": 0.242,
      "step": 312
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6184090046934805,
      "learning_rate": 9.996137675744083e-06,
      "loss": 0.2567,
      "step": 313
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.768717035633087,
      "learning_rate": 9.996051380847479e-06,
      "loss": 0.2846,
      "step": 314
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.506581835755855,
      "learning_rate": 9.995964132921001e-06,
      "loss": 0.2256,
      "step": 315
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5509361607647292,
      "learning_rate": 9.995875931981294e-06,
      "loss": 0.2504,
      "step": 316
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0046303664066105,
      "learning_rate": 9.99578677804518e-06,
      "loss": 0.2905,
      "step": 317
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2070283476168322,
      "learning_rate": 9.995696671129668e-06,
      "loss": 0.2727,
      "step": 318
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8153143604319486,
      "learning_rate": 9.995605611251948e-06,
      "loss": 0.3081,
      "step": 319
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6066158495210843,
      "learning_rate": 9.995513598429387e-06,
      "loss": 0.2139,
      "step": 320
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.041787649402546,
      "learning_rate": 9.995420632679538e-06,
      "loss": 0.2993,
      "step": 321
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8437722164867407,
      "learning_rate": 9.995326714020136e-06,
      "loss": 0.2722,
      "step": 322
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1564868117872524,
      "learning_rate": 9.995231842469095e-06,
      "loss": 0.2624,
      "step": 323
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.9723940992061613,
      "learning_rate": 9.995136018044513e-06,
      "loss": 0.271,
      "step": 324
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.73038802804571,
      "learning_rate": 9.99503924076467e-06,
      "loss": 0.3289,
      "step": 325
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3328778703589965,
      "learning_rate": 9.994941510648026e-06,
      "loss": 0.1773,
      "step": 326
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6209515821269351,
      "learning_rate": 9.994842827713225e-06,
      "loss": 0.2061,
      "step": 327
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7506618273105239,
      "learning_rate": 9.99474319197909e-06,
      "loss": 0.2574,
      "step": 328
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4575830430626537,
      "learning_rate": 9.994642603464626e-06,
      "loss": 0.2454,
      "step": 329
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.638166981644136,
      "learning_rate": 9.994541062189025e-06,
      "loss": 0.2373,
      "step": 330
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7840616443987694,
      "learning_rate": 9.994438568171652e-06,
      "loss": 0.2253,
      "step": 331
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7411236928775584,
      "learning_rate": 9.99433512143206e-06,
      "loss": 0.2354,
      "step": 332
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.362306936140375,
      "learning_rate": 9.994230721989983e-06,
      "loss": 0.163,
      "step": 333
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1521866105067113,
      "learning_rate": 9.994125369865337e-06,
      "loss": 0.142,
      "step": 334
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8375526417948076,
      "learning_rate": 9.994019065078217e-06,
      "loss": 0.3063,
      "step": 335
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9764586607380783,
      "learning_rate": 9.993911807648901e-06,
      "loss": 0.3668,
      "step": 336
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.75831008854968,
      "learning_rate": 9.993803597597849e-06,
      "loss": 0.3088,
      "step": 337
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.639406224603388,
      "learning_rate": 9.993694434945705e-06,
      "loss": 0.3125,
      "step": 338
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8479663534416089,
      "learning_rate": 9.993584319713289e-06,
      "loss": 0.3031,
      "step": 339
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5617648713058614,
      "learning_rate": 9.993473251921608e-06,
      "loss": 0.2496,
      "step": 340
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.996336714011638,
      "learning_rate": 9.99336123159185e-06,
      "loss": 0.3139,
      "step": 341
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6949488742587895,
      "learning_rate": 9.993248258745383e-06,
      "loss": 0.2414,
      "step": 342
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6476285878172074,
      "learning_rate": 9.993134333403756e-06,
      "loss": 0.3006,
      "step": 343
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.30734707045587,
      "learning_rate": 9.993019455588701e-06,
      "loss": 0.2147,
      "step": 344
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6513939331967316,
      "learning_rate": 9.992903625322135e-06,
      "loss": 0.2724,
      "step": 345
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8524042976967978,
      "learning_rate": 9.99278684262615e-06,
      "loss": 0.3233,
      "step": 346
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7447000415836578,
      "learning_rate": 9.992669107523025e-06,
      "loss": 0.2581,
      "step": 347
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.358548463706941,
      "learning_rate": 9.992550420035217e-06,
      "loss": 0.2084,
      "step": 348
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.521740265393964,
      "learning_rate": 9.992430780185368e-06,
      "loss": 0.2611,
      "step": 349
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5537110131216487,
      "learning_rate": 9.992310187996298e-06,
      "loss": 0.2591,
      "step": 350
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7653938139011018,
      "learning_rate": 9.992188643491013e-06,
      "loss": 0.3445,
      "step": 351
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.34380075127815,
      "learning_rate": 9.992066146692695e-06,
      "loss": 0.1779,
      "step": 352
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1494611403386057,
      "learning_rate": 9.991942697624717e-06,
      "loss": 0.1782,
      "step": 353
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.252133650727571,
      "learning_rate": 9.991818296310623e-06,
      "loss": 0.1997,
      "step": 354
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8128143028682625,
      "learning_rate": 9.991692942774144e-06,
      "loss": 0.2481,
      "step": 355
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.101730431468222,
      "learning_rate": 9.991566637039193e-06,
      "loss": 0.235,
      "step": 356
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3603617279791294,
      "learning_rate": 9.991439379129864e-06,
      "loss": 0.2196,
      "step": 357
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4624778934961298,
      "learning_rate": 9.991311169070433e-06,
      "loss": 0.168,
      "step": 358
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5326217606696262,
      "learning_rate": 9.991182006885354e-06,
      "loss": 0.2587,
      "step": 359
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4234728273118735,
      "learning_rate": 9.991051892599267e-06,
      "loss": 0.2502,
      "step": 360
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3182319467343533,
      "learning_rate": 9.990920826236993e-06,
      "loss": 0.2065,
      "step": 361
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1516566116161606,
      "learning_rate": 9.990788807823533e-06,
      "loss": 0.2377,
      "step": 362
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4170919394074935,
      "learning_rate": 9.99065583738407e-06,
      "loss": 0.1954,
      "step": 363
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.035379683136883,
      "learning_rate": 9.990521914943969e-06,
      "loss": 0.3012,
      "step": 364
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.474186274063641,
      "learning_rate": 9.990387040528778e-06,
      "loss": 0.2573,
      "step": 365
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6081366475768557,
      "learning_rate": 9.990251214164225e-06,
      "loss": 0.2499,
      "step": 366
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.440626134556924,
      "learning_rate": 9.990114435876217e-06,
      "loss": 0.2328,
      "step": 367
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5317837617027603,
      "learning_rate": 9.98997670569085e-06,
      "loss": 0.274,
      "step": 368
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6328400047238645,
      "learning_rate": 9.989838023634392e-06,
      "loss": 0.2855,
      "step": 369
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.407754480578645,
      "learning_rate": 9.9896983897333e-06,
      "loss": 0.2446,
      "step": 370
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.469179966886001,
      "learning_rate": 9.989557804014211e-06,
      "loss": 0.22,
      "step": 371
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0233906923939395,
      "learning_rate": 9.989416266503941e-06,
      "loss": 0.3793,
      "step": 372
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0152417285517032,
      "learning_rate": 9.98927377722949e-06,
      "loss": 0.1754,
      "step": 373
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.514607854209359,
      "learning_rate": 9.989130336218039e-06,
      "loss": 0.2553,
      "step": 374
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3766219248903173,
      "learning_rate": 9.98898594349695e-06,
      "loss": 0.1337,
      "step": 375
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7427332165408644,
      "learning_rate": 9.988840599093765e-06,
      "loss": 0.2234,
      "step": 376
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.464126128979691,
      "learning_rate": 9.988694303036213e-06,
      "loss": 0.372,
      "step": 377
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8563217736273507,
      "learning_rate": 9.988547055352198e-06,
      "loss": 0.266,
      "step": 378
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6781914700162943,
      "learning_rate": 9.98839885606981e-06,
      "loss": 0.2673,
      "step": 379
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.155751408018473,
      "learning_rate": 9.988249705217319e-06,
      "loss": 0.2218,
      "step": 380
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2773435423663768,
      "learning_rate": 9.988099602823176e-06,
      "loss": 0.1695,
      "step": 381
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9207287047121115,
      "learning_rate": 9.987948548916013e-06,
      "loss": 0.3124,
      "step": 382
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5929039925829125,
      "learning_rate": 9.987796543524646e-06,
      "loss": 0.3003,
      "step": 383
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9606189878064246,
      "learning_rate": 9.987643586678072e-06,
      "loss": 0.2778,
      "step": 384
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7029876998328461,
      "learning_rate": 9.987489678405466e-06,
      "loss": 0.2792,
      "step": 385
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4537220626857696,
      "learning_rate": 9.98733481873619e-06,
      "loss": 0.2473,
      "step": 386
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1914177287456573,
      "learning_rate": 9.987179007699781e-06,
      "loss": 0.1775,
      "step": 387
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7656774263089312,
      "learning_rate": 9.987022245325962e-06,
      "loss": 0.2705,
      "step": 388
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8449305478584697,
      "learning_rate": 9.986864531644638e-06,
      "loss": 0.3358,
      "step": 389
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2221327230096954,
      "learning_rate": 9.986705866685895e-06,
      "loss": 0.1285,
      "step": 390
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2390381429424178,
      "learning_rate": 9.986546250479996e-06,
      "loss": 0.2001,
      "step": 391
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1219064062050794,
      "learning_rate": 9.98638568305739e-06,
      "loss": 0.1694,
      "step": 392
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0434242859240044,
      "learning_rate": 9.986224164448706e-06,
      "loss": 0.2479,
      "step": 393
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4206821661978786,
      "learning_rate": 9.986061694684757e-06,
      "loss": 0.2526,
      "step": 394
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1448459715427264,
      "learning_rate": 9.985898273796532e-06,
      "loss": 0.1625,
      "step": 395
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2150347157734798,
      "learning_rate": 9.985733901815208e-06,
      "loss": 0.2268,
      "step": 396
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1167541044676215,
      "learning_rate": 9.985568578772138e-06,
      "loss": 0.1519,
      "step": 397
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7994698084375245,
      "learning_rate": 9.985402304698857e-06,
      "loss": 0.3021,
      "step": 398
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6883930387820392,
      "learning_rate": 9.985235079627087e-06,
      "loss": 0.31,
      "step": 399
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.27019975776995,
      "learning_rate": 9.985066903588723e-06,
      "loss": 0.1557,
      "step": 400
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5216059402625692,
      "learning_rate": 9.98489777661585e-06,
      "loss": 0.2125,
      "step": 401
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7880525842769537,
      "learning_rate": 9.984727698740724e-06,
      "loss": 0.2476,
      "step": 402
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.491605120773047,
      "learning_rate": 9.984556669995794e-06,
      "loss": 0.2284,
      "step": 403
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7641898404649257,
      "learning_rate": 9.984384690413684e-06,
      "loss": 0.2288,
      "step": 404
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5285823958284155,
      "learning_rate": 9.984211760027199e-06,
      "loss": 0.2295,
      "step": 405
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6109222577575069,
      "learning_rate": 9.984037878869325e-06,
      "loss": 0.219,
      "step": 406
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0085774867017783,
      "learning_rate": 9.983863046973235e-06,
      "loss": 0.1425,
      "step": 407
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.554847726109499,
      "learning_rate": 9.983687264372275e-06,
      "loss": 0.1908,
      "step": 408
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.152047171319707,
      "learning_rate": 9.98351053109998e-06,
      "loss": 0.2942,
      "step": 409
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.066256378616708,
      "learning_rate": 9.98333284719006e-06,
      "loss": 0.2479,
      "step": 410
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.022025242828952,
      "learning_rate": 9.983154212676416e-06,
      "loss": 0.2129,
      "step": 411
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3658526855021984,
      "learning_rate": 9.982974627593116e-06,
      "loss": 0.2098,
      "step": 412
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3928705764527871,
      "learning_rate": 9.982794091974421e-06,
      "loss": 0.2562,
      "step": 413
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6005524270682647,
      "learning_rate": 9.982612605854767e-06,
      "loss": 0.3131,
      "step": 414
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5340081343998748,
      "learning_rate": 9.982430169268775e-06,
      "loss": 0.2521,
      "step": 415
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.331724170308008,
      "learning_rate": 9.982246782251247e-06,
      "loss": 0.2752,
      "step": 416
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2612078982758885,
      "learning_rate": 9.982062444837167e-06,
      "loss": 0.2485,
      "step": 417
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2405080280199146,
      "learning_rate": 9.981877157061693e-06,
      "loss": 0.1841,
      "step": 418
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2642511665521114,
      "learning_rate": 9.981690918960172e-06,
      "loss": 0.2496,
      "step": 419
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6540447782558483,
      "learning_rate": 9.981503730568134e-06,
      "loss": 0.2935,
      "step": 420
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6286383466255596,
      "learning_rate": 9.98131559192128e-06,
      "loss": 0.2903,
      "step": 421
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2922032449360228,
      "learning_rate": 9.981126503055505e-06,
      "loss": 0.2639,
      "step": 422
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5183591043592868,
      "learning_rate": 9.980936464006874e-06,
      "loss": 0.218,
      "step": 423
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1031400598277206,
      "learning_rate": 9.980745474811641e-06,
      "loss": 0.2002,
      "step": 424
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5059057939953393,
      "learning_rate": 9.980553535506239e-06,
      "loss": 0.2592,
      "step": 425
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.177892190164666,
      "learning_rate": 9.980360646127279e-06,
      "loss": 0.2481,
      "step": 426
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.641722270532745,
      "learning_rate": 9.980166806711558e-06,
      "loss": 0.2715,
      "step": 427
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5102139419222766,
      "learning_rate": 9.97997201729605e-06,
      "loss": 0.2832,
      "step": 428
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.611984875148702,
      "learning_rate": 9.979776277917915e-06,
      "loss": 0.1804,
      "step": 429
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4047601988190053,
      "learning_rate": 9.97957958861449e-06,
      "loss": 0.2372,
      "step": 430
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1786204771970885,
      "learning_rate": 9.979381949423297e-06,
      "loss": 0.1579,
      "step": 431
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4740441251110188,
      "learning_rate": 9.979183360382034e-06,
      "loss": 0.2291,
      "step": 432
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.116895865719068,
      "learning_rate": 9.978983821528583e-06,
      "loss": 0.207,
      "step": 433
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6510876547081634,
      "learning_rate": 9.97878333290101e-06,
      "loss": 0.2465,
      "step": 434
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.276415310418171,
      "learning_rate": 9.978581894537558e-06,
      "loss": 0.2243,
      "step": 435
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5398265955213237,
      "learning_rate": 9.978379506476654e-06,
      "loss": 0.3386,
      "step": 436
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2817260223377647,
      "learning_rate": 9.978176168756903e-06,
      "loss": 0.2605,
      "step": 437
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7296186117811552,
      "learning_rate": 9.977971881417095e-06,
      "loss": 0.2382,
      "step": 438
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3699895998128955,
      "learning_rate": 9.977766644496195e-06,
      "loss": 0.2833,
      "step": 439
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4034811491452013,
      "learning_rate": 9.977560458033359e-06,
      "loss": 0.2597,
      "step": 440
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.117434377455661,
      "learning_rate": 9.977353322067915e-06,
      "loss": 0.3616,
      "step": 441
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2819590256552735,
      "learning_rate": 9.977145236639377e-06,
      "loss": 0.2597,
      "step": 442
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4133385458608474,
      "learning_rate": 9.976936201787437e-06,
      "loss": 0.2563,
      "step": 443
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.381463036135683,
      "learning_rate": 9.976726217551971e-06,
      "loss": 0.2519,
      "step": 444
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3151395500871566,
      "learning_rate": 9.976515283973035e-06,
      "loss": 0.2464,
      "step": 445
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4536996616551807,
      "learning_rate": 9.976303401090866e-06,
      "loss": 0.255,
      "step": 446
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.388091261090952,
      "learning_rate": 9.97609056894588e-06,
      "loss": 0.2371,
      "step": 447
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.645823837453223,
      "learning_rate": 9.97587678757868e-06,
      "loss": 0.214,
      "step": 448
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0984934033674487,
      "learning_rate": 9.975662057030042e-06,
      "loss": 0.2035,
      "step": 449
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.193863086904732,
      "learning_rate": 9.97544637734093e-06,
      "loss": 0.2145,
      "step": 450
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.588966043524015,
      "learning_rate": 9.975229748552485e-06,
      "loss": 0.3193,
      "step": 451
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0055603616412143,
      "learning_rate": 9.975012170706033e-06,
      "loss": 0.1494,
      "step": 452
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.709236419005986,
      "learning_rate": 9.974793643843076e-06,
      "loss": 0.3301,
      "step": 453
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.36226472460251,
      "learning_rate": 9.9745741680053e-06,
      "loss": 0.2932,
      "step": 454
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4454367350170994,
      "learning_rate": 9.97435374323457e-06,
      "loss": 0.2431,
      "step": 455
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4603359622000458,
      "learning_rate": 9.974132369572936e-06,
      "loss": 0.2377,
      "step": 456
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3531408852287579,
      "learning_rate": 9.973910047062625e-06,
      "loss": 0.2206,
      "step": 457
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3199208151859918,
      "learning_rate": 9.973686775746045e-06,
      "loss": 0.2347,
      "step": 458
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2521765787653636,
      "learning_rate": 9.97346255566579e-06,
      "loss": 0.2047,
      "step": 459
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9106461542220722,
      "learning_rate": 9.973237386864629e-06,
      "loss": 0.2291,
      "step": 460
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2809074564827772,
      "learning_rate": 9.973011269385518e-06,
      "loss": 0.2565,
      "step": 461
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3204804400413626,
      "learning_rate": 9.972784203271583e-06,
      "loss": 0.211,
      "step": 462
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0926110049507076,
      "learning_rate": 9.972556188566147e-06,
      "loss": 0.2181,
      "step": 463
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2669929613658588,
      "learning_rate": 9.972327225312698e-06,
      "loss": 0.2489,
      "step": 464
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8559552917620891,
      "learning_rate": 9.972097313554918e-06,
      "loss": 0.2104,
      "step": 465
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9329313515659348,
      "learning_rate": 9.971866453336662e-06,
      "loss": 0.1433,
      "step": 466
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5368873904697071,
      "learning_rate": 9.971634644701966e-06,
      "loss": 0.2737,
      "step": 467
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3699973020832892,
      "learning_rate": 9.971401887695052e-06,
      "loss": 0.2097,
      "step": 468
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7073666526604245,
      "learning_rate": 9.971168182360318e-06,
      "loss": 0.2758,
      "step": 469
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3860170386719612,
      "learning_rate": 9.970933528742347e-06,
      "loss": 0.2205,
      "step": 470
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.573131827218781,
      "learning_rate": 9.970697926885899e-06,
      "loss": 0.2727,
      "step": 471
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5555894224059688,
      "learning_rate": 9.970461376835916e-06,
      "loss": 0.1817,
      "step": 472
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4680983537503904,
      "learning_rate": 9.970223878637523e-06,
      "loss": 0.2096,
      "step": 473
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.507254938052605,
      "learning_rate": 9.969985432336023e-06,
      "loss": 0.229,
      "step": 474
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5866275403294732,
      "learning_rate": 9.969746037976905e-06,
      "loss": 0.2247,
      "step": 475
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4529661280095152,
      "learning_rate": 9.969505695605828e-06,
      "loss": 0.2137,
      "step": 476
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0636582165570403,
      "learning_rate": 9.969264405268646e-06,
      "loss": 0.2012,
      "step": 477
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3735163108264759,
      "learning_rate": 9.969022167011381e-06,
      "loss": 0.2305,
      "step": 478
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4661228700839977,
      "learning_rate": 9.968778980880247e-06,
      "loss": 0.2493,
      "step": 479
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7203417845597087,
      "learning_rate": 9.96853484692163e-06,
      "loss": 0.2644,
      "step": 480
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3676651456259614,
      "learning_rate": 9.9682897651821e-06,
      "loss": 0.2339,
      "step": 481
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6933553921730415,
      "learning_rate": 9.968043735708411e-06,
      "loss": 0.2501,
      "step": 482
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6970584606866173,
      "learning_rate": 9.96779675854749e-06,
      "loss": 0.2606,
      "step": 483
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1816048186471906,
      "learning_rate": 9.967548833746451e-06,
      "loss": 0.2167,
      "step": 484
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7136313108990782,
      "learning_rate": 9.967299961352592e-06,
      "loss": 0.2697,
      "step": 485
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6199191796460857,
      "learning_rate": 9.967050141413382e-06,
      "loss": 0.2772,
      "step": 486
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6219118871290195,
      "learning_rate": 9.966799373976478e-06,
      "loss": 0.3001,
      "step": 487
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1634006590478403,
      "learning_rate": 9.966547659089711e-06,
      "loss": 0.1875,
      "step": 488
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.299979436068115,
      "learning_rate": 9.966294996801105e-06,
      "loss": 0.2022,
      "step": 489
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4213481875229708,
      "learning_rate": 9.966041387158851e-06,
      "loss": 0.3289,
      "step": 490
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0544732025970767,
      "learning_rate": 9.96578683021133e-06,
      "loss": 0.1975,
      "step": 491
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.148722424342399,
      "learning_rate": 9.965531326007099e-06,
      "loss": 0.2003,
      "step": 492
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9773312609636016,
      "learning_rate": 9.965274874594896e-06,
      "loss": 0.2025,
      "step": 493
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.028542094479082,
      "learning_rate": 9.965017476023645e-06,
      "loss": 0.2589,
      "step": 494
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8126563775567521,
      "learning_rate": 9.96475913034244e-06,
      "loss": 0.299,
      "step": 495
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7055442025194985,
      "learning_rate": 9.96449983760057e-06,
      "loss": 0.2542,
      "step": 496
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4314125162513158,
      "learning_rate": 9.96423959784749e-06,
      "loss": 0.2274,
      "step": 497
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7205350111831672,
      "learning_rate": 9.963978411132845e-06,
      "loss": 0.2848,
      "step": 498
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.637242686472228,
      "learning_rate": 9.963716277506457e-06,
      "loss": 0.2671,
      "step": 499
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.490526491486953,
      "learning_rate": 9.963453197018332e-06,
      "loss": 0.1934,
      "step": 500
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.032335260866634,
      "learning_rate": 9.963189169718655e-06,
      "loss": 0.201,
      "step": 501
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.8524871404809244,
      "learning_rate": 9.962924195657785e-06,
      "loss": 0.2873,
      "step": 502
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3167279083703323,
      "learning_rate": 9.962658274886275e-06,
      "loss": 0.1732,
      "step": 503
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4881878216880058,
      "learning_rate": 9.962391407454849e-06,
      "loss": 0.2577,
      "step": 504
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4956928345698226,
      "learning_rate": 9.962123593414409e-06,
      "loss": 0.2003,
      "step": 505
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4687985924520885,
      "learning_rate": 9.961854832816048e-06,
      "loss": 0.2482,
      "step": 506
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3652822416433232,
      "learning_rate": 9.961585125711031e-06,
      "loss": 0.2663,
      "step": 507
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2719514587663154,
      "learning_rate": 9.961314472150807e-06,
      "loss": 0.2284,
      "step": 508
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4808677787452937,
      "learning_rate": 9.961042872187005e-06,
      "loss": 0.2432,
      "step": 509
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.398236444640137,
      "learning_rate": 9.960770325871435e-06,
      "loss": 0.2082,
      "step": 510
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1146556069681315,
      "learning_rate": 9.960496833256086e-06,
      "loss": 0.1852,
      "step": 511
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1135849827760493,
      "learning_rate": 9.960222394393131e-06,
      "loss": 0.2325,
      "step": 512
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.085126437637428,
      "learning_rate": 9.959947009334918e-06,
      "loss": 0.2364,
      "step": 513
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7384445522749252,
      "learning_rate": 9.95967067813398e-06,
      "loss": 0.2873,
      "step": 514
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1498139816481436,
      "learning_rate": 9.959393400843027e-06,
      "loss": 0.2057,
      "step": 515
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2179986739689934,
      "learning_rate": 9.959115177514955e-06,
      "loss": 0.2135,
      "step": 516
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6523932505417487,
      "learning_rate": 9.958836008202835e-06,
      "loss": 0.2427,
      "step": 517
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2435937148022085,
      "learning_rate": 9.958555892959919e-06,
      "loss": 0.1392,
      "step": 518
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3894258046683345,
      "learning_rate": 9.958274831839642e-06,
      "loss": 0.262,
      "step": 519
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7427854051888891,
      "learning_rate": 9.957992824895621e-06,
      "loss": 0.2672,
      "step": 520
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.37627548031114,
      "learning_rate": 9.957709872181648e-06,
      "loss": 0.2754,
      "step": 521
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9932317688280896,
      "learning_rate": 9.957425973751698e-06,
      "loss": 0.3028,
      "step": 522
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1840227778603012,
      "learning_rate": 9.957141129659926e-06,
      "loss": 0.2292,
      "step": 523
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0970031046528939,
      "learning_rate": 9.95685533996067e-06,
      "loss": 0.1987,
      "step": 524
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3240672988036166,
      "learning_rate": 9.956568604708444e-06,
      "loss": 0.234,
      "step": 525
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2284800942808367,
      "learning_rate": 9.956280923957945e-06,
      "loss": 0.171,
      "step": 526
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6356880551847488,
      "learning_rate": 9.955992297764054e-06,
      "loss": 0.2785,
      "step": 527
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3335816849469193,
      "learning_rate": 9.955702726181823e-06,
      "loss": 0.1748,
      "step": 528
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3896312850944919,
      "learning_rate": 9.955412209266492e-06,
      "loss": 0.3433,
      "step": 529
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6778711046511512,
      "learning_rate": 9.955120747073478e-06,
      "loss": 0.236,
      "step": 530
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.47102785880093,
      "learning_rate": 9.954828339658382e-06,
      "loss": 0.257,
      "step": 531
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.789727597617072,
      "learning_rate": 9.95453498707698e-06,
      "loss": 0.2661,
      "step": 532
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1662090199199446,
      "learning_rate": 9.95424068938523e-06,
      "loss": 0.2263,
      "step": 533
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7545615515629476,
      "learning_rate": 9.953945446639275e-06,
      "loss": 0.3044,
      "step": 534
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.018484957496278,
      "learning_rate": 9.953649258895432e-06,
      "loss": 0.1651,
      "step": 535
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.469276267757643,
      "learning_rate": 9.953352126210202e-06,
      "loss": 0.2948,
      "step": 536
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4309809376229912,
      "learning_rate": 9.953054048640263e-06,
      "loss": 0.313,
      "step": 537
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3512718411972158,
      "learning_rate": 9.952755026242478e-06,
      "loss": 0.2292,
      "step": 538
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3653234568109998,
      "learning_rate": 9.952455059073884e-06,
      "loss": 0.2299,
      "step": 539
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2966123135198702,
      "learning_rate": 9.952154147191706e-06,
      "loss": 0.2337,
      "step": 540
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6657235058104731,
      "learning_rate": 9.951852290653341e-06,
      "loss": 0.2517,
      "step": 541
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4625906095348107,
      "learning_rate": 9.951549489516373e-06,
      "loss": 0.22,
      "step": 542
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0920449971412483,
      "learning_rate": 9.951245743838563e-06,
      "loss": 0.1703,
      "step": 543
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2135801366959844,
      "learning_rate": 9.950941053677849e-06,
      "loss": 0.2434,
      "step": 544
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5356452664497233,
      "learning_rate": 9.950635419092358e-06,
      "loss": 0.2989,
      "step": 545
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.123904002911888,
      "learning_rate": 9.950328840140388e-06,
      "loss": 0.1793,
      "step": 546
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2409260663613944,
      "learning_rate": 9.950021316880422e-06,
      "loss": 0.2321,
      "step": 547
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3440762418459185,
      "learning_rate": 9.949712849371123e-06,
      "loss": 0.2486,
      "step": 548
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1638955785293899,
      "learning_rate": 9.94940343767133e-06,
      "loss": 0.2338,
      "step": 549
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3768457213991032,
      "learning_rate": 9.949093081840071e-06,
      "loss": 0.2442,
      "step": 550
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3519649303852872,
      "learning_rate": 9.948781781936543e-06,
      "loss": 0.2096,
      "step": 551
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.963785626582525,
      "learning_rate": 9.94846953802013e-06,
      "loss": 0.2677,
      "step": 552
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3412188590579937,
      "learning_rate": 9.948156350150398e-06,
      "loss": 0.2048,
      "step": 553
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2142208317758985,
      "learning_rate": 9.947842218387088e-06,
      "loss": 0.2215,
      "step": 554
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.9163946585718983,
      "learning_rate": 9.947527142790119e-06,
      "loss": 0.3431,
      "step": 555
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2230991066168264,
      "learning_rate": 9.947211123419596e-06,
      "loss": 0.154,
      "step": 556
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3291539805102057,
      "learning_rate": 9.946894160335803e-06,
      "loss": 0.1959,
      "step": 557
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.272482063481536,
      "learning_rate": 9.946576253599203e-06,
      "loss": 0.2036,
      "step": 558
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4387337915251277,
      "learning_rate": 9.946257403270437e-06,
      "loss": 0.2276,
      "step": 559
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0219178736890568,
      "learning_rate": 9.94593760941033e-06,
      "loss": 0.2273,
      "step": 560
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.134101677144825,
      "learning_rate": 9.945616872079881e-06,
      "loss": 0.2114,
      "step": 561
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0686433405663007,
      "learning_rate": 9.945295191340276e-06,
      "loss": 0.1692,
      "step": 562
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8580189611736475,
      "learning_rate": 9.944972567252878e-06,
      "loss": 0.3008,
      "step": 563
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1982399843380407,
      "learning_rate": 9.944648999879227e-06,
      "loss": 0.2597,
      "step": 564
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2333019405511845,
      "learning_rate": 9.94432448928105e-06,
      "loss": 0.2528,
      "step": 565
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8045828951814397,
      "learning_rate": 9.943999035520244e-06,
      "loss": 0.1814,
      "step": 566
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4347831574074827,
      "learning_rate": 9.943672638658896e-06,
      "loss": 0.227,
      "step": 567
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4393444787471719,
      "learning_rate": 9.943345298759266e-06,
      "loss": 0.2224,
      "step": 568
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5172413594070782,
      "learning_rate": 9.943017015883797e-06,
      "loss": 0.2434,
      "step": 569
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0791953084669463,
      "learning_rate": 9.942687790095111e-06,
      "loss": 0.1857,
      "step": 570
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6134547124222949,
      "learning_rate": 9.94235762145601e-06,
      "loss": 0.2739,
      "step": 571
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5857039074817099,
      "learning_rate": 9.942026510029476e-06,
      "loss": 0.2553,
      "step": 572
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4053712953584911,
      "learning_rate": 9.94169445587867e-06,
      "loss": 0.2061,
      "step": 573
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2680271455162422,
      "learning_rate": 9.941361459066935e-06,
      "loss": 0.2002,
      "step": 574
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6640549762203467,
      "learning_rate": 9.94102751965779e-06,
      "loss": 0.2518,
      "step": 575
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8288877834505906,
      "learning_rate": 9.940692637714939e-06,
      "loss": 0.2779,
      "step": 576
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.199146349292955,
      "learning_rate": 9.94035681330226e-06,
      "loss": 0.2326,
      "step": 577
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.194685108787481,
      "learning_rate": 9.940020046483817e-06,
      "loss": 0.2248,
      "step": 578
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9392705998846206,
      "learning_rate": 9.939682337323847e-06,
      "loss": 0.1207,
      "step": 579
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7605655254820953,
      "learning_rate": 9.939343685886775e-06,
      "loss": 0.3155,
      "step": 580
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1717188672102792,
      "learning_rate": 9.939004092237197e-06,
      "loss": 0.2233,
      "step": 581
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4074979840465214,
      "learning_rate": 9.938663556439891e-06,
      "loss": 0.2521,
      "step": 582
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4613185352825728,
      "learning_rate": 9.938322078559822e-06,
      "loss": 0.2531,
      "step": 583
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2812936397922998,
      "learning_rate": 9.937979658662125e-06,
      "loss": 0.2641,
      "step": 584
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3314233533704027,
      "learning_rate": 9.937636296812122e-06,
      "loss": 0.2051,
      "step": 585
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.035155842878196,
      "learning_rate": 9.937291993075308e-06,
      "loss": 0.1797,
      "step": 586
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5443165005378425,
      "learning_rate": 9.936946747517363e-06,
      "loss": 0.2216,
      "step": 587
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2093464749917844,
      "learning_rate": 9.936600560204146e-06,
      "loss": 0.2524,
      "step": 588
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4371020078945902,
      "learning_rate": 9.936253431201691e-06,
      "loss": 0.2556,
      "step": 589
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4247386404357398,
      "learning_rate": 9.935905360576221e-06,
      "loss": 0.2454,
      "step": 590
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7405100771624604,
      "learning_rate": 9.935556348394128e-06,
      "loss": 0.2496,
      "step": 591
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.340030876508446,
      "learning_rate": 9.93520639472199e-06,
      "loss": 0.3087,
      "step": 592
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5768981352327542,
      "learning_rate": 9.93485549962656e-06,
      "loss": 0.2861,
      "step": 593
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.48982081910507,
      "learning_rate": 9.934503663174779e-06,
      "loss": 0.2766,
      "step": 594
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2627258465017708,
      "learning_rate": 9.93415088543376e-06,
      "loss": 0.2366,
      "step": 595
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.19478387460378,
      "learning_rate": 9.933797166470795e-06,
      "loss": 0.1812,
      "step": 596
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1436793171443547,
      "learning_rate": 9.933442506353362e-06,
      "loss": 0.2666,
      "step": 597
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.285177779397736,
      "learning_rate": 9.933086905149113e-06,
      "loss": 0.1944,
      "step": 598
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2334164441743987,
      "learning_rate": 9.93273036292588e-06,
      "loss": 0.2499,
      "step": 599
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.215831493762732,
      "learning_rate": 9.932372879751677e-06,
      "loss": 0.2008,
      "step": 600
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3597562975554311,
      "learning_rate": 9.932014455694699e-06,
      "loss": 0.2015,
      "step": 601
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6916139492949407,
      "learning_rate": 9.931655090823311e-06,
      "loss": 0.2819,
      "step": 602
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0725272288898027,
      "learning_rate": 9.931294785206072e-06,
      "loss": 0.2054,
      "step": 603
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.570469431017226,
      "learning_rate": 9.93093353891171e-06,
      "loss": 0.2634,
      "step": 604
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2210578341257852,
      "learning_rate": 9.93057135200913e-06,
      "loss": 0.2446,
      "step": 605
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4828591456438873,
      "learning_rate": 9.930208224567429e-06,
      "loss": 0.3157,
      "step": 606
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8447957113626772,
      "learning_rate": 9.929844156655872e-06,
      "loss": 0.244,
      "step": 607
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.44278316798067,
      "learning_rate": 9.929479148343909e-06,
      "loss": 0.2402,
      "step": 608
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3537252045293708,
      "learning_rate": 9.929113199701163e-06,
      "loss": 0.255,
      "step": 609
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8292581825301466,
      "learning_rate": 9.928746310797448e-06,
      "loss": 0.3316,
      "step": 610
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7406541812701393,
      "learning_rate": 9.928378481702748e-06,
      "loss": 0.3367,
      "step": 611
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1585764439036026,
      "learning_rate": 9.928009712487227e-06,
      "loss": 0.1832,
      "step": 612
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0131021697123965,
      "learning_rate": 9.927640003221233e-06,
      "loss": 0.2053,
      "step": 613
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.587452217961807,
      "learning_rate": 9.927269353975288e-06,
      "loss": 0.2492,
      "step": 614
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5450176202870771,
      "learning_rate": 9.926897764820095e-06,
      "loss": 0.231,
      "step": 615
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.276331991312613,
      "learning_rate": 9.926525235826541e-06,
      "loss": 0.213,
      "step": 616
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3490089278801147,
      "learning_rate": 9.926151767065687e-06,
      "loss": 0.2519,
      "step": 617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3577766268488236,
      "learning_rate": 9.925777358608772e-06,
      "loss": 0.2294,
      "step": 618
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3965352474514439,
      "learning_rate": 9.92540201052722e-06,
      "loss": 0.3029,
      "step": 619
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4668988675565036,
      "learning_rate": 9.92502572289263e-06,
      "loss": 0.2809,
      "step": 620
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4182785897197276,
      "learning_rate": 9.924648495776781e-06,
      "loss": 0.235,
      "step": 621
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7879353436026295,
      "learning_rate": 9.924270329251631e-06,
      "loss": 0.2526,
      "step": 622
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4340170727350245,
      "learning_rate": 9.92389122338932e-06,
      "loss": 0.2627,
      "step": 623
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4495712620456707,
      "learning_rate": 9.923511178262162e-06,
      "loss": 0.1976,
      "step": 624
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3201328963369183,
      "learning_rate": 9.923130193942654e-06,
      "loss": 0.1879,
      "step": 625
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9826596222639931,
      "learning_rate": 9.922748270503472e-06,
      "loss": 0.1525,
      "step": 626
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4123216031515593,
      "learning_rate": 9.922365408017474e-06,
      "loss": 0.2217,
      "step": 627
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.411351842093732,
      "learning_rate": 9.921981606557687e-06,
      "loss": 0.2963,
      "step": 628
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4566795523655254,
      "learning_rate": 9.921596866197328e-06,
      "loss": 0.2664,
      "step": 629
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0666586395188855,
      "learning_rate": 9.921211187009785e-06,
      "loss": 0.181,
      "step": 630
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7419180123921258,
      "learning_rate": 9.920824569068632e-06,
      "loss": 0.3363,
      "step": 631
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8262413847865508,
      "learning_rate": 9.920437012447619e-06,
      "loss": 0.1857,
      "step": 632
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2688735718624455,
      "learning_rate": 9.920048517220674e-06,
      "loss": 0.1995,
      "step": 633
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0837230218705434,
      "learning_rate": 9.919659083461906e-06,
      "loss": 0.2147,
      "step": 634
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6014147321786254,
      "learning_rate": 9.9192687112456e-06,
      "loss": 0.2672,
      "step": 635
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1902261119324566,
      "learning_rate": 9.918877400646223e-06,
      "loss": 0.2272,
      "step": 636
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4727548812306739,
      "learning_rate": 9.918485151738421e-06,
      "loss": 0.2753,
      "step": 637
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1861806379776234,
      "learning_rate": 9.918091964597017e-06,
      "loss": 0.2193,
      "step": 638
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4378238717315506,
      "learning_rate": 9.917697839297016e-06,
      "loss": 0.2722,
      "step": 639
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3734205429704252,
      "learning_rate": 9.917302775913597e-06,
      "loss": 0.2204,
      "step": 640
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.075251672002481,
      "learning_rate": 9.916906774522124e-06,
      "loss": 0.2302,
      "step": 641
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9313942800041353,
      "learning_rate": 9.916509835198134e-06,
      "loss": 0.2033,
      "step": 642
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5138440003517304,
      "learning_rate": 9.916111958017347e-06,
      "loss": 0.3453,
      "step": 643
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4876970346769995,
      "learning_rate": 9.915713143055663e-06,
      "loss": 0.2611,
      "step": 644
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9019295013580393,
      "learning_rate": 9.915313390389154e-06,
      "loss": 0.132,
      "step": 645
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.232873461010102,
      "learning_rate": 9.91491270009408e-06,
      "loss": 0.211,
      "step": 646
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2242105870174862,
      "learning_rate": 9.914511072246874e-06,
      "loss": 0.2195,
      "step": 647
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1723127250279795,
      "learning_rate": 9.914108506924146e-06,
      "loss": 0.1623,
      "step": 648
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2847645249050796,
      "learning_rate": 9.913705004202691e-06,
      "loss": 0.2093,
      "step": 649
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2100818796137633,
      "learning_rate": 9.913300564159478e-06,
      "loss": 0.218,
      "step": 650
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4117834035698307,
      "learning_rate": 9.912895186871661e-06,
      "loss": 0.3043,
      "step": 651
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2024525718147812,
      "learning_rate": 9.912488872416564e-06,
      "loss": 0.2557,
      "step": 652
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0937660466324213,
      "learning_rate": 9.912081620871694e-06,
      "loss": 0.1867,
      "step": 653
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.335671212454067,
      "learning_rate": 9.91167343231474e-06,
      "loss": 0.221,
      "step": 654
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.500778918794191,
      "learning_rate": 9.911264306823564e-06,
      "loss": 0.3304,
      "step": 655
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.447722629723532,
      "learning_rate": 9.910854244476212e-06,
      "loss": 0.2379,
      "step": 656
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3577451368808762,
      "learning_rate": 9.910443245350903e-06,
      "loss": 0.2562,
      "step": 657
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.239221737193145,
      "learning_rate": 9.91003130952604e-06,
      "loss": 0.2219,
      "step": 658
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.575373664520173,
      "learning_rate": 9.909618437080204e-06,
      "loss": 0.2487,
      "step": 659
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8713135806072339,
      "learning_rate": 9.909204628092148e-06,
      "loss": 0.2889,
      "step": 660
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.994423933015644,
      "learning_rate": 9.908789882640812e-06,
      "loss": 0.1127,
      "step": 661
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3432365402120863,
      "learning_rate": 9.908374200805313e-06,
      "loss": 0.1852,
      "step": 662
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2326705030986878,
      "learning_rate": 9.907957582664942e-06,
      "loss": 0.2637,
      "step": 663
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3836664989685823,
      "learning_rate": 9.907540028299173e-06,
      "loss": 0.2398,
      "step": 664
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0078539116714875,
      "learning_rate": 9.907121537787659e-06,
      "loss": 0.2084,
      "step": 665
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2166824977817328,
      "learning_rate": 9.906702111210225e-06,
      "loss": 0.2963,
      "step": 666
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.098280563864795,
      "learning_rate": 9.906281748646884e-06,
      "loss": 0.1964,
      "step": 667
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.408726454010625,
      "learning_rate": 9.905860450177822e-06,
      "loss": 0.2575,
      "step": 668
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5695165704550953,
      "learning_rate": 9.905438215883403e-06,
      "loss": 0.2718,
      "step": 669
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2913982625894043,
      "learning_rate": 9.905015045844173e-06,
      "loss": 0.2343,
      "step": 670
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1942448893602566,
      "learning_rate": 9.904590940140853e-06,
      "loss": 0.2422,
      "step": 671
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1770714221306793,
      "learning_rate": 9.904165898854343e-06,
      "loss": 0.1896,
      "step": 672
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.273254560065629,
      "learning_rate": 9.903739922065724e-06,
      "loss": 0.2349,
      "step": 673
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3524070450472117,
      "learning_rate": 9.903313009856254e-06,
      "loss": 0.2356,
      "step": 674
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2385749457678545,
      "learning_rate": 9.902885162307371e-06,
      "loss": 0.2297,
      "step": 675
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3373734351580295,
      "learning_rate": 9.902456379500686e-06,
      "loss": 0.2105,
      "step": 676
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.146534546096548,
      "learning_rate": 9.902026661517994e-06,
      "loss": 0.2157,
      "step": 677
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.122941485953434,
      "learning_rate": 9.901596008441267e-06,
      "loss": 0.1907,
      "step": 678
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6369360526090062,
      "learning_rate": 9.901164420352654e-06,
      "loss": 0.2686,
      "step": 679
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3165986867588195,
      "learning_rate": 9.900731897334483e-06,
      "loss": 0.2493,
      "step": 680
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7038606771997948,
      "learning_rate": 9.900298439469262e-06,
      "loss": 0.342,
      "step": 681
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8460932150549558,
      "learning_rate": 9.899864046839675e-06,
      "loss": 0.1534,
      "step": 682
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.8854574392536674,
      "learning_rate": 9.899428719528585e-06,
      "loss": 0.336,
      "step": 683
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4589351511335327,
      "learning_rate": 9.898992457619034e-06,
      "loss": 0.2372,
      "step": 684
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5950511263503862,
      "learning_rate": 9.898555261194243e-06,
      "loss": 0.2335,
      "step": 685
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.523853849303764,
      "learning_rate": 9.898117130337609e-06,
      "loss": 0.2555,
      "step": 686
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3528811583845057,
      "learning_rate": 9.897678065132707e-06,
      "loss": 0.2355,
      "step": 687
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.379029208812336,
      "learning_rate": 9.897238065663295e-06,
      "loss": 0.2273,
      "step": 688
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9525968053535399,
      "learning_rate": 9.896797132013304e-06,
      "loss": 0.2211,
      "step": 689
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.356113540178845,
      "learning_rate": 9.896355264266842e-06,
      "loss": 0.288,
      "step": 690
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1505961146115422,
      "learning_rate": 9.895912462508203e-06,
      "loss": 0.243,
      "step": 691
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2359804736830924,
      "learning_rate": 9.895468726821853e-06,
      "loss": 0.2562,
      "step": 692
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0749254954782916,
      "learning_rate": 9.895024057292434e-06,
      "loss": 0.1717,
      "step": 693
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1625281888079761,
      "learning_rate": 9.894578454004776e-06,
      "loss": 0.209,
      "step": 694
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1234877898849043,
      "learning_rate": 9.894131917043875e-06,
      "loss": 0.226,
      "step": 695
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4026176130918955,
      "learning_rate": 9.893684446494915e-06,
      "loss": 0.2329,
      "step": 696
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3071488509371818,
      "learning_rate": 9.893236042443251e-06,
      "loss": 0.2288,
      "step": 697
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.740814301506615,
      "learning_rate": 9.89278670497442e-06,
      "loss": 0.3211,
      "step": 698
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6100080140773747,
      "learning_rate": 9.892336434174137e-06,
      "loss": 0.2779,
      "step": 699
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5739983799372084,
      "learning_rate": 9.891885230128293e-06,
      "loss": 0.27,
      "step": 700
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1492332588535001,
      "learning_rate": 9.89143309292296e-06,
      "loss": 0.1918,
      "step": 701
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.491662962273359,
      "learning_rate": 9.890980022644383e-06,
      "loss": 0.3117,
      "step": 702
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3329187067395991,
      "learning_rate": 9.890526019378991e-06,
      "loss": 0.2777,
      "step": 703
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7786210401757554,
      "learning_rate": 9.890071083213387e-06,
      "loss": 0.278,
      "step": 704
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1918447793002338,
      "learning_rate": 9.889615214234353e-06,
      "loss": 0.2169,
      "step": 705
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.355991216031387,
      "learning_rate": 9.889158412528849e-06,
      "loss": 0.224,
      "step": 706
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7468971003444869,
      "learning_rate": 9.888700678184013e-06,
      "loss": 0.1564,
      "step": 707
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1221905921363289,
      "learning_rate": 9.88824201128716e-06,
      "loss": 0.223,
      "step": 708
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8490894560787874,
      "learning_rate": 9.887782411925788e-06,
      "loss": 0.1373,
      "step": 709
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9779229138751871,
      "learning_rate": 9.887321880187562e-06,
      "loss": 0.1333,
      "step": 710
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4421078925966384,
      "learning_rate": 9.886860416160335e-06,
      "loss": 0.3177,
      "step": 711
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4125661216948424,
      "learning_rate": 9.886398019932136e-06,
      "loss": 0.234,
      "step": 712
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5627678831897098,
      "learning_rate": 9.885934691591167e-06,
      "loss": 0.2612,
      "step": 713
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5013963468130973,
      "learning_rate": 9.885470431225814e-06,
      "loss": 0.2698,
      "step": 714
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4210244586614647,
      "learning_rate": 9.885005238924635e-06,
      "loss": 0.2711,
      "step": 715
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.063439474004361,
      "learning_rate": 9.884539114776368e-06,
      "loss": 0.1798,
      "step": 716
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.175692096695271,
      "learning_rate": 9.884072058869933e-06,
      "loss": 0.1912,
      "step": 717
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3167100356972794,
      "learning_rate": 9.88360407129442e-06,
      "loss": 0.201,
      "step": 718
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.584476397648018,
      "learning_rate": 9.883135152139102e-06,
      "loss": 0.2799,
      "step": 719
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4643707722289512,
      "learning_rate": 9.882665301493431e-06,
      "loss": 0.1715,
      "step": 720
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0940643757664703,
      "learning_rate": 9.882194519447033e-06,
      "loss": 0.1753,
      "step": 721
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1531218924434834,
      "learning_rate": 9.881722806089708e-06,
      "loss": 0.2136,
      "step": 722
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4725858477133837,
      "learning_rate": 9.881250161511446e-06,
      "loss": 0.2608,
      "step": 723
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0346830734273653,
      "learning_rate": 9.8807765858024e-06,
      "loss": 0.1849,
      "step": 724
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3717132730800796,
      "learning_rate": 9.880302079052914e-06,
      "loss": 0.2337,
      "step": 725
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3638876844782337,
      "learning_rate": 9.879826641353498e-06,
      "loss": 0.2531,
      "step": 726
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4574979789104439,
      "learning_rate": 9.879350272794849e-06,
      "loss": 0.2993,
      "step": 727
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4412365176858606,
      "learning_rate": 9.878872973467834e-06,
      "loss": 0.2448,
      "step": 728
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1973796680268303,
      "learning_rate": 9.878394743463503e-06,
      "loss": 0.2077,
      "step": 729
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0449311810702868,
      "learning_rate": 9.877915582873083e-06,
      "loss": 0.186,
      "step": 730
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1701745930825294,
      "learning_rate": 9.877435491787975e-06,
      "loss": 0.2207,
      "step": 731
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.746230543867163,
      "learning_rate": 9.87695447029976e-06,
      "loss": 0.218,
      "step": 732
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5795486736361055,
      "learning_rate": 9.876472518500195e-06,
      "loss": 0.2475,
      "step": 733
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.633086625412564,
      "learning_rate": 9.875989636481216e-06,
      "loss": 0.2407,
      "step": 734
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4946800023750137,
      "learning_rate": 9.875505824334937e-06,
      "loss": 0.3,
      "step": 735
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.171043540317,
      "learning_rate": 9.875021082153648e-06,
      "loss": 0.2136,
      "step": 736
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6834981915821736,
      "learning_rate": 9.874535410029816e-06,
      "loss": 0.2928,
      "step": 737
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.155357657780537,
      "learning_rate": 9.874048808056086e-06,
      "loss": 0.2238,
      "step": 738
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4309808424213013,
      "learning_rate": 9.873561276325281e-06,
      "loss": 0.1915,
      "step": 739
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7160905626446037,
      "learning_rate": 9.873072814930402e-06,
      "loss": 0.2826,
      "step": 740
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4317567977084604,
      "learning_rate": 9.872583423964626e-06,
      "loss": 0.2546,
      "step": 741
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4395553787853244,
      "learning_rate": 9.872093103521306e-06,
      "loss": 0.3131,
      "step": 742
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.43303550526125,
      "learning_rate": 9.871601853693975e-06,
      "loss": 0.3014,
      "step": 743
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4092518769463647,
      "learning_rate": 9.871109674576342e-06,
      "loss": 0.2628,
      "step": 744
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2856634384919887,
      "learning_rate": 9.870616566262295e-06,
      "loss": 0.2752,
      "step": 745
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6120764314440594,
      "learning_rate": 9.870122528845894e-06,
      "loss": 0.2925,
      "step": 746
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8888909676092348,
      "learning_rate": 9.869627562421381e-06,
      "loss": 0.1082,
      "step": 747
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9450232927683705,
      "learning_rate": 9.869131667083176e-06,
      "loss": 0.1426,
      "step": 748
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5870156634758588,
      "learning_rate": 9.868634842925875e-06,
      "loss": 0.2521,
      "step": 749
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2202504166149544,
      "learning_rate": 9.868137090044248e-06,
      "loss": 0.2453,
      "step": 750
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3532718129332717,
      "learning_rate": 9.867638408533244e-06,
      "loss": 0.2679,
      "step": 751
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0248696358791918,
      "learning_rate": 9.867138798487993e-06,
      "loss": 0.246,
      "step": 752
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2940752547195293,
      "learning_rate": 9.866638260003797e-06,
      "loss": 0.2097,
      "step": 753
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4539954761012326,
      "learning_rate": 9.866136793176138e-06,
      "loss": 0.2765,
      "step": 754
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0392805415060045,
      "learning_rate": 9.865634398100671e-06,
      "loss": 0.1533,
      "step": 755
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3003708443437,
      "learning_rate": 9.865131074873235e-06,
      "loss": 0.2946,
      "step": 756
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0253105350755423,
      "learning_rate": 9.864626823589842e-06,
      "loss": 0.214,
      "step": 757
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8298477995316634,
      "learning_rate": 9.864121644346679e-06,
      "loss": 0.2083,
      "step": 758
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0169813901537461,
      "learning_rate": 9.863615537240114e-06,
      "loss": 0.3004,
      "step": 759
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.002662225988702,
      "learning_rate": 9.86310850236669e-06,
      "loss": 0.143,
      "step": 760
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.05250154510306,
      "learning_rate": 9.862600539823124e-06,
      "loss": 0.2086,
      "step": 761
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2578054528912948,
      "learning_rate": 9.862091649706321e-06,
      "loss": 0.2648,
      "step": 762
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4830180443986873,
      "learning_rate": 9.861581832113346e-06,
      "loss": 0.2861,
      "step": 763
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9054683190027109,
      "learning_rate": 9.861071087141456e-06,
      "loss": 0.1909,
      "step": 764
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0699163733010635,
      "learning_rate": 9.860559414888077e-06,
      "loss": 0.177,
      "step": 765
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7554394257283226,
      "learning_rate": 9.860046815450815e-06,
      "loss": 0.3066,
      "step": 766
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4837517215547054,
      "learning_rate": 9.85953328892745e-06,
      "loss": 0.2069,
      "step": 767
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3506371063912843,
      "learning_rate": 9.859018835415944e-06,
      "loss": 0.2625,
      "step": 768
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1620344653599775,
      "learning_rate": 9.858503455014428e-06,
      "loss": 0.2345,
      "step": 769
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6941691416434994,
      "learning_rate": 9.857987147821216e-06,
      "loss": 0.2888,
      "step": 770
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7607932617007285,
      "learning_rate": 9.857469913934796e-06,
      "loss": 0.1361,
      "step": 771
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0012161839914138,
      "learning_rate": 9.856951753453836e-06,
      "loss": 0.2228,
      "step": 772
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.940854364497359,
      "learning_rate": 9.856432666477178e-06,
      "loss": 0.1932,
      "step": 773
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9955982956777913,
      "learning_rate": 9.855912653103844e-06,
      "loss": 0.1719,
      "step": 774
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0543869179550516,
      "learning_rate": 9.855391713433023e-06,
      "loss": 0.1891,
      "step": 775
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9513353273754006,
      "learning_rate": 9.85486984756409e-06,
      "loss": 0.1705,
      "step": 776
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5063972073935952,
      "learning_rate": 9.8543470555966e-06,
      "loss": 0.3096,
      "step": 777
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.259242066828801,
      "learning_rate": 9.853823337630272e-06,
      "loss": 0.2131,
      "step": 778
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1561683403175882,
      "learning_rate": 9.853298693765013e-06,
      "loss": 0.2195,
      "step": 779
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1840971540717589,
      "learning_rate": 9.8527731241009e-06,
      "loss": 0.2173,
      "step": 780
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.969902528255275,
      "learning_rate": 9.85224662873819e-06,
      "loss": 0.2412,
      "step": 781
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2273182799551192,
      "learning_rate": 9.851719207777317e-06,
      "loss": 0.2538,
      "step": 782
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.080744683364228,
      "learning_rate": 9.851190861318887e-06,
      "loss": 0.1891,
      "step": 783
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3500168406403443,
      "learning_rate": 9.850661589463686e-06,
      "loss": 0.2381,
      "step": 784
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2214630303456726,
      "learning_rate": 9.850131392312677e-06,
      "loss": 0.2096,
      "step": 785
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5101176022979277,
      "learning_rate": 9.849600269967e-06,
      "loss": 0.2365,
      "step": 786
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.286184034991567,
      "learning_rate": 9.849068222527967e-06,
      "loss": 0.2373,
      "step": 787
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0533398670439795,
      "learning_rate": 9.848535250097072e-06,
      "loss": 0.1577,
      "step": 788
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0015523844747476,
      "learning_rate": 9.848001352775982e-06,
      "loss": 0.2164,
      "step": 789
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8957864777970989,
      "learning_rate": 9.847466530666543e-06,
      "loss": 0.1425,
      "step": 790
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3355858604784832,
      "learning_rate": 9.846930783870774e-06,
      "loss": 0.2214,
      "step": 791
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.25510345041101,
      "learning_rate": 9.846394112490871e-06,
      "loss": 0.2318,
      "step": 792
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4266719292047687,
      "learning_rate": 9.845856516629213e-06,
      "loss": 0.247,
      "step": 793
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0812169237731417,
      "learning_rate": 9.845317996388344e-06,
      "loss": 0.1638,
      "step": 794
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6502057875948142,
      "learning_rate": 9.844778551870992e-06,
      "loss": 0.2321,
      "step": 795
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4217383838341269,
      "learning_rate": 9.844238183180062e-06,
      "loss": 0.2641,
      "step": 796
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2575093615267035,
      "learning_rate": 9.84369689041863e-06,
      "loss": 0.2517,
      "step": 797
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0987775412675593,
      "learning_rate": 9.843154673689953e-06,
      "loss": 0.2556,
      "step": 798
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0005285011072222,
      "learning_rate": 9.842611533097463e-06,
      "loss": 0.2598,
      "step": 799
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4755578440436372,
      "learning_rate": 9.842067468744765e-06,
      "loss": 0.2343,
      "step": 800
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1436999502354823,
      "learning_rate": 9.841522480735644e-06,
      "loss": 0.2712,
      "step": 801
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2223999137877746,
      "learning_rate": 9.84097656917406e-06,
      "loss": 0.207,
      "step": 802
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1551097329773554,
      "learning_rate": 9.840429734164153e-06,
      "loss": 0.2203,
      "step": 803
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.544712268942903,
      "learning_rate": 9.839881975810231e-06,
      "loss": 0.2584,
      "step": 804
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8885199426899517,
      "learning_rate": 9.839333294216782e-06,
      "loss": 0.1984,
      "step": 805
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.212914452679541,
      "learning_rate": 9.838783689488473e-06,
      "loss": 0.1849,
      "step": 806
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.401991235653526,
      "learning_rate": 9.838233161730144e-06,
      "loss": 0.2886,
      "step": 807
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.662763325262415,
      "learning_rate": 9.837681711046812e-06,
      "loss": 0.2401,
      "step": 808
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1519168289997064,
      "learning_rate": 9.837129337543668e-06,
      "loss": 0.2296,
      "step": 809
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.158796796764017,
      "learning_rate": 9.836576041326085e-06,
      "loss": 0.2053,
      "step": 810
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1710758607349494,
      "learning_rate": 9.836021822499605e-06,
      "loss": 0.1699,
      "step": 811
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1656773113660077,
      "learning_rate": 9.835466681169949e-06,
      "loss": 0.2284,
      "step": 812
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1963113871504714,
      "learning_rate": 9.834910617443014e-06,
      "loss": 0.257,
      "step": 813
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2648441550957625,
      "learning_rate": 9.834353631424873e-06,
      "loss": 0.2584,
      "step": 814
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1860266223724936,
      "learning_rate": 9.833795723221776e-06,
      "loss": 0.2172,
      "step": 815
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4647908947115373,
      "learning_rate": 9.833236892940145e-06,
      "loss": 0.2721,
      "step": 816
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8971858964142536,
      "learning_rate": 9.832677140686584e-06,
      "loss": 0.2073,
      "step": 817
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8320070041735548,
      "learning_rate": 9.832116466567867e-06,
      "loss": 0.3094,
      "step": 818
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9734784684759342,
      "learning_rate": 9.831554870690945e-06,
      "loss": 0.1967,
      "step": 819
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2066272272865906,
      "learning_rate": 9.830992353162951e-06,
      "loss": 0.2482,
      "step": 820
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0712785566948935,
      "learning_rate": 9.830428914091185e-06,
      "loss": 0.2653,
      "step": 821
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.092713270337398,
      "learning_rate": 9.829864553583126e-06,
      "loss": 0.1757,
      "step": 822
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.188739125067949,
      "learning_rate": 9.829299271746433e-06,
      "loss": 0.1831,
      "step": 823
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.107767761866914,
      "learning_rate": 9.828733068688936e-06,
      "loss": 0.2038,
      "step": 824
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3932322563323003,
      "learning_rate": 9.828165944518641e-06,
      "loss": 0.2558,
      "step": 825
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.602701236695193,
      "learning_rate": 9.827597899343731e-06,
      "loss": 0.2528,
      "step": 826
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5483633312155973,
      "learning_rate": 9.827028933272564e-06,
      "loss": 0.2741,
      "step": 827
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2595605344972771,
      "learning_rate": 9.826459046413675e-06,
      "loss": 0.2243,
      "step": 828
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2738226875842265,
      "learning_rate": 9.825888238875772e-06,
      "loss": 0.2405,
      "step": 829
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.232813211967336,
      "learning_rate": 9.825316510767744e-06,
      "loss": 0.2061,
      "step": 830
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8708702154268229,
      "learning_rate": 9.824743862198647e-06,
      "loss": 0.156,
      "step": 831
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7697069487193199,
      "learning_rate": 9.824170293277722e-06,
      "loss": 0.1702,
      "step": 832
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0269331444443028,
      "learning_rate": 9.823595804114379e-06,
      "loss": 0.2274,
      "step": 833
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.3441793313584478,
      "learning_rate": 9.823020394818203e-06,
      "loss": 0.2539,
      "step": 834
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1219864621176474,
      "learning_rate": 9.822444065498964e-06,
      "loss": 0.2114,
      "step": 835
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5348095754362354,
      "learning_rate": 9.821866816266595e-06,
      "loss": 0.2415,
      "step": 836
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0384560607647009,
      "learning_rate": 9.821288647231208e-06,
      "loss": 0.2024,
      "step": 837
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.240711652529904,
      "learning_rate": 9.8207095585031e-06,
      "loss": 0.265,
      "step": 838
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.200714928517479,
      "learning_rate": 9.820129550192731e-06,
      "loss": 0.2241,
      "step": 839
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.082931142967991,
      "learning_rate": 9.819548622410744e-06,
      "loss": 0.2484,
      "step": 840
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5197698625198457,
      "learning_rate": 9.818966775267951e-06,
      "loss": 0.2639,
      "step": 841
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1842021432539755,
      "learning_rate": 9.818384008875347e-06,
      "loss": 0.263,
      "step": 842
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4047087277327763,
      "learning_rate": 9.817800323344097e-06,
      "loss": 0.2919,
      "step": 843
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2213172698031447,
      "learning_rate": 9.817215718785543e-06,
      "loss": 0.2173,
      "step": 844
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.193541635181449,
      "learning_rate": 9.816630195311202e-06,
      "loss": 0.2072,
      "step": 845
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.143664025867426,
      "learning_rate": 9.816043753032766e-06,
      "loss": 0.2431,
      "step": 846
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6792810241016678,
      "learning_rate": 9.815456392062104e-06,
      "loss": 0.3261,
      "step": 847
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1828813782502552,
      "learning_rate": 9.814868112511258e-06,
      "loss": 0.2736,
      "step": 848
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.071191895121823,
      "learning_rate": 9.814278914492447e-06,
      "loss": 0.2476,
      "step": 849
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7559637224922314,
      "learning_rate": 9.813688798118066e-06,
      "loss": 0.1669,
      "step": 850
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9801631326379127,
      "learning_rate": 9.813097763500678e-06,
      "loss": 0.2286,
      "step": 851
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2499603650846947,
      "learning_rate": 9.812505810753034e-06,
      "loss": 0.1989,
      "step": 852
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8842321667236669,
      "learning_rate": 9.811912939988048e-06,
      "loss": 0.1448,
      "step": 853
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9351363078679387,
      "learning_rate": 9.811319151318815e-06,
      "loss": 0.1705,
      "step": 854
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6170599638403864,
      "learning_rate": 9.810724444858604e-06,
      "loss": 0.2814,
      "step": 855
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3209947376836961,
      "learning_rate": 9.810128820720861e-06,
      "loss": 0.2616,
      "step": 856
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.393112412771032,
      "learning_rate": 9.809532279019203e-06,
      "loss": 0.216,
      "step": 857
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1163247525303877,
      "learning_rate": 9.808934819867423e-06,
      "loss": 0.2198,
      "step": 858
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0806463986698724,
      "learning_rate": 9.808336443379495e-06,
      "loss": 0.2466,
      "step": 859
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8010884241627552,
      "learning_rate": 9.807737149669562e-06,
      "loss": 0.1953,
      "step": 860
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1554627175843275,
      "learning_rate": 9.80713693885194e-06,
      "loss": 0.187,
      "step": 861
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7409948323291893,
      "learning_rate": 9.806535811041126e-06,
      "loss": 0.0877,
      "step": 862
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.289278690538077,
      "learning_rate": 9.805933766351787e-06,
      "loss": 0.2252,
      "step": 863
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.199904611162512,
      "learning_rate": 9.80533080489877e-06,
      "loss": 0.2468,
      "step": 864
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7409285219789339,
      "learning_rate": 9.804726926797092e-06,
      "loss": 0.1803,
      "step": 865
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.287679436974793,
      "learning_rate": 9.804122132161946e-06,
      "loss": 0.201,
      "step": 866
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1389185639084853,
      "learning_rate": 9.803516421108704e-06,
      "loss": 0.22,
      "step": 867
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.258888144967439,
      "learning_rate": 9.802909793752903e-06,
      "loss": 0.2444,
      "step": 868
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1681034775431713,
      "learning_rate": 9.802302250210268e-06,
      "loss": 0.2244,
      "step": 869
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0420843480823854,
      "learning_rate": 9.801693790596688e-06,
      "loss": 0.1994,
      "step": 870
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3536168320724469,
      "learning_rate": 9.80108441502823e-06,
      "loss": 0.2707,
      "step": 871
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0176590752762766,
      "learning_rate": 9.800474123621141e-06,
      "loss": 0.241,
      "step": 872
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1653980761953597,
      "learning_rate": 9.799862916491834e-06,
      "loss": 0.2152,
      "step": 873
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5365545265624991,
      "learning_rate": 9.799250793756902e-06,
      "loss": 0.3097,
      "step": 874
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.238844234980156,
      "learning_rate": 9.798637755533112e-06,
      "loss": 0.1747,
      "step": 875
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3679210895417944,
      "learning_rate": 9.798023801937406e-06,
      "loss": 0.2637,
      "step": 876
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9003740919391385,
      "learning_rate": 9.797408933086896e-06,
      "loss": 0.21,
      "step": 877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3394407126552694,
      "learning_rate": 9.796793149098876e-06,
      "loss": 0.2774,
      "step": 878
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.504199413751013,
      "learning_rate": 9.796176450090808e-06,
      "loss": 0.2513,
      "step": 879
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2512253174107395,
      "learning_rate": 9.795558836180336e-06,
      "loss": 0.2472,
      "step": 880
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9124681054381032,
      "learning_rate": 9.794940307485268e-06,
      "loss": 0.2144,
      "step": 881
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1111154569435775,
      "learning_rate": 9.794320864123597e-06,
      "loss": 0.1822,
      "step": 882
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.424794709786109,
      "learning_rate": 9.793700506213484e-06,
      "loss": 0.2458,
      "step": 883
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9637698398180545,
      "learning_rate": 9.793079233873266e-06,
      "loss": 0.1494,
      "step": 884
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2108863153546066,
      "learning_rate": 9.792457047221457e-06,
      "loss": 0.1779,
      "step": 885
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.263648541641633,
      "learning_rate": 9.79183394637674e-06,
      "loss": 0.2196,
      "step": 886
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3205428572766154,
      "learning_rate": 9.791209931457978e-06,
      "loss": 0.2423,
      "step": 887
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4952871923081208,
      "learning_rate": 9.790585002584206e-06,
      "loss": 0.2521,
      "step": 888
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0746756690598378,
      "learning_rate": 9.789959159874632e-06,
      "loss": 0.2286,
      "step": 889
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.406025524250865,
      "learning_rate": 9.789332403448639e-06,
      "loss": 0.2834,
      "step": 890
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.373034527547828,
      "learning_rate": 9.788704733425787e-06,
      "loss": 0.2483,
      "step": 891
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2156262392762067,
      "learning_rate": 9.788076149925807e-06,
      "loss": 0.2509,
      "step": 892
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1731824490871774,
      "learning_rate": 9.787446653068607e-06,
      "loss": 0.2632,
      "step": 893
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9510618425930021,
      "learning_rate": 9.786816242974267e-06,
      "loss": 0.1572,
      "step": 894
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.106089084668461,
      "learning_rate": 9.786184919763039e-06,
      "loss": 0.2064,
      "step": 895
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9917485123901718,
      "learning_rate": 9.785552683555357e-06,
      "loss": 0.1704,
      "step": 896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4838167878331008,
      "learning_rate": 9.78491953447182e-06,
      "loss": 0.2311,
      "step": 897
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2159156668043185,
      "learning_rate": 9.784285472633208e-06,
      "loss": 0.2057,
      "step": 898
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2129663933264117,
      "learning_rate": 9.783650498160472e-06,
      "loss": 0.2639,
      "step": 899
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1245449020040288,
      "learning_rate": 9.783014611174737e-06,
      "loss": 0.1945,
      "step": 900
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2982282725613732,
      "learning_rate": 9.782377811797302e-06,
      "loss": 0.2238,
      "step": 901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3822367836242817,
      "learning_rate": 9.781740100149643e-06,
      "loss": 0.2831,
      "step": 902
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3814354882044346,
      "learning_rate": 9.781101476353406e-06,
      "loss": 0.235,
      "step": 903
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4718602757829804,
      "learning_rate": 9.780461940530411e-06,
      "loss": 0.2178,
      "step": 904
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0976476185950974,
      "learning_rate": 9.779821492802657e-06,
      "loss": 0.2163,
      "step": 905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2666236548734706,
      "learning_rate": 9.779180133292313e-06,
      "loss": 0.2304,
      "step": 906
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4155081023971583,
      "learning_rate": 9.778537862121722e-06,
      "loss": 0.2637,
      "step": 907
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3379259876292087,
      "learning_rate": 9.7778946794134e-06,
      "loss": 0.2283,
      "step": 908
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1669581822274064,
      "learning_rate": 9.777250585290041e-06,
      "loss": 0.2285,
      "step": 909
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2777471697908045,
      "learning_rate": 9.776605579874508e-06,
      "loss": 0.2595,
      "step": 910
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9449617591786018,
      "learning_rate": 9.775959663289843e-06,
      "loss": 0.1945,
      "step": 911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.233866542530544,
      "learning_rate": 9.775312835659254e-06,
      "loss": 0.208,
      "step": 912
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.062821607517932,
      "learning_rate": 9.774665097106133e-06,
      "loss": 0.1976,
      "step": 913
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9668099889536191,
      "learning_rate": 9.774016447754037e-06,
      "loss": 0.2151,
      "step": 914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2473935576485717,
      "learning_rate": 9.773366887726701e-06,
      "loss": 0.253,
      "step": 915
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0570060466683904,
      "learning_rate": 9.772716417148032e-06,
      "loss": 0.1939,
      "step": 916
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4695428095874645,
      "learning_rate": 9.772065036142115e-06,
      "loss": 0.261,
      "step": 917
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9507937485354795,
      "learning_rate": 9.771412744833202e-06,
      "loss": 0.2256,
      "step": 918
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5058998485172894,
      "learning_rate": 9.77075954334572e-06,
      "loss": 0.2404,
      "step": 919
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9673202745864136,
      "learning_rate": 9.770105431804278e-06,
      "loss": 0.1417,
      "step": 920
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2248565671907512,
      "learning_rate": 9.769450410333646e-06,
      "loss": 0.2564,
      "step": 921
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.007215697227723,
      "learning_rate": 9.768794479058776e-06,
      "loss": 0.2507,
      "step": 922
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3384354490879686,
      "learning_rate": 9.768137638104793e-06,
      "loss": 0.1923,
      "step": 923
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.131798157426711,
      "learning_rate": 9.76747988759699e-06,
      "loss": 0.2083,
      "step": 924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3943147741091138,
      "learning_rate": 9.76682122766084e-06,
      "loss": 0.2408,
      "step": 925
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0155290576271687,
      "learning_rate": 9.766161658421986e-06,
      "loss": 0.166,
      "step": 926
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1854554786395386,
      "learning_rate": 9.765501180006246e-06,
      "loss": 0.1922,
      "step": 927
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4516584876698508,
      "learning_rate": 9.76483979253961e-06,
      "loss": 0.305,
      "step": 928
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2804253635435459,
      "learning_rate": 9.76417749614824e-06,
      "loss": 0.2412,
      "step": 929
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9816431087333565,
      "learning_rate": 9.763514290958476e-06,
      "loss": 0.1854,
      "step": 930
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9461225050509303,
      "learning_rate": 9.762850177096828e-06,
      "loss": 0.2107,
      "step": 931
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0161991894204472,
      "learning_rate": 9.76218515468998e-06,
      "loss": 0.2262,
      "step": 932
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0522249070480922,
      "learning_rate": 9.761519223864789e-06,
      "loss": 0.2229,
      "step": 933
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2271895634404208,
      "learning_rate": 9.760852384748288e-06,
      "loss": 0.2174,
      "step": 934
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2156561510736084,
      "learning_rate": 9.760184637467678e-06,
      "loss": 0.2372,
      "step": 935
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4637896036886333,
      "learning_rate": 9.759515982150338e-06,
      "loss": 0.3021,
      "step": 936
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3182802344965028,
      "learning_rate": 9.758846418923818e-06,
      "loss": 0.2753,
      "step": 937
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.257038988888129,
      "learning_rate": 9.75817594791584e-06,
      "loss": 0.2585,
      "step": 938
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0008764055138328,
      "learning_rate": 9.757504569254305e-06,
      "loss": 0.2837,
      "step": 939
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1696833468889771,
      "learning_rate": 9.75683228306728e-06,
      "loss": 0.1776,
      "step": 940
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6825642457408924,
      "learning_rate": 9.756159089483006e-06,
      "loss": 0.1596,
      "step": 941
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.057846852035875,
      "learning_rate": 9.755484988629901e-06,
      "loss": 0.1615,
      "step": 942
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2012263861791013,
      "learning_rate": 9.754809980636557e-06,
      "loss": 0.1944,
      "step": 943
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1853019440323378,
      "learning_rate": 9.754134065631733e-06,
      "loss": 0.2051,
      "step": 944
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8413821630531813,
      "learning_rate": 9.753457243744366e-06,
      "loss": 0.1536,
      "step": 945
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2698890502929812,
      "learning_rate": 9.752779515103563e-06,
      "loss": 0.1938,
      "step": 946
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1031617841954355,
      "learning_rate": 9.752100879838604e-06,
      "loss": 0.2102,
      "step": 947
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.026323523227093,
      "learning_rate": 9.751421338078946e-06,
      "loss": 0.1589,
      "step": 948
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2170136071098372,
      "learning_rate": 9.750740889954215e-06,
      "loss": 0.2057,
      "step": 949
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4289051545051177,
      "learning_rate": 9.750059535594211e-06,
      "loss": 0.2359,
      "step": 950
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3628091260997732,
      "learning_rate": 9.749377275128906e-06,
      "loss": 0.2159,
      "step": 951
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.319536677477234,
      "learning_rate": 9.748694108688446e-06,
      "loss": 0.2404,
      "step": 952
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4389475574857147,
      "learning_rate": 9.748010036403151e-06,
      "loss": 0.2416,
      "step": 953
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2417447963432469,
      "learning_rate": 9.747325058403512e-06,
      "loss": 0.2456,
      "step": 954
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9977698423398847,
      "learning_rate": 9.746639174820191e-06,
      "loss": 0.1457,
      "step": 955
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1785482101248568,
      "learning_rate": 9.745952385784025e-06,
      "loss": 0.1999,
      "step": 956
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2507402053912173,
      "learning_rate": 9.745264691426027e-06,
      "loss": 0.2051,
      "step": 957
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.050373772005299,
      "learning_rate": 9.744576091877375e-06,
      "loss": 0.1739,
      "step": 958
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3566177138457165,
      "learning_rate": 9.743886587269428e-06,
      "loss": 0.214,
      "step": 959
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3217776955259406,
      "learning_rate": 9.743196177733709e-06,
      "loss": 0.251,
      "step": 960
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5290523638948337,
      "learning_rate": 9.742504863401923e-06,
      "loss": 0.2901,
      "step": 961
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3211695397373946,
      "learning_rate": 9.741812644405939e-06,
      "loss": 0.1842,
      "step": 962
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8958521675544511,
      "learning_rate": 9.741119520877804e-06,
      "loss": 0.1119,
      "step": 963
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.27148202666826,
      "learning_rate": 9.740425492949737e-06,
      "loss": 0.2224,
      "step": 964
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1245190771152203,
      "learning_rate": 9.739730560754125e-06,
      "loss": 0.1952,
      "step": 965
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1393773542864605,
      "learning_rate": 9.739034724423535e-06,
      "loss": 0.1893,
      "step": 966
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.559109215612849,
      "learning_rate": 9.738337984090699e-06,
      "loss": 0.259,
      "step": 967
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7983358895919931,
      "learning_rate": 9.737640339888527e-06,
      "loss": 0.156,
      "step": 968
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3039252362626712,
      "learning_rate": 9.736941791950097e-06,
      "loss": 0.2144,
      "step": 969
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.171530814243301,
      "learning_rate": 9.736242340408665e-06,
      "loss": 0.198,
      "step": 970
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.260244965677875,
      "learning_rate": 9.735541985397654e-06,
      "loss": 0.2778,
      "step": 971
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1326433473588755,
      "learning_rate": 9.73484072705066e-06,
      "loss": 0.2346,
      "step": 972
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0145472705277858,
      "learning_rate": 9.734138565501455e-06,
      "loss": 0.165,
      "step": 973
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9508044762452798,
      "learning_rate": 9.733435500883982e-06,
      "loss": 0.1918,
      "step": 974
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0911061797576933,
      "learning_rate": 9.73273153333235e-06,
      "loss": 0.2078,
      "step": 975
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9512423707426831,
      "learning_rate": 9.73202666298085e-06,
      "loss": 0.1745,
      "step": 976
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2348568077118414,
      "learning_rate": 9.73132088996394e-06,
      "loss": 0.2703,
      "step": 977
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.450752959964214,
      "learning_rate": 9.730614214416251e-06,
      "loss": 0.1878,
      "step": 978
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.027328876612728,
      "learning_rate": 9.729906636472584e-06,
      "loss": 0.2085,
      "step": 979
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2550651447210104,
      "learning_rate": 9.729198156267915e-06,
      "loss": 0.2329,
      "step": 980
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2544193131665407,
      "learning_rate": 9.728488773937394e-06,
      "loss": 0.2412,
      "step": 981
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0816037526033715,
      "learning_rate": 9.727778489616338e-06,
      "loss": 0.223,
      "step": 982
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.003609851622092,
      "learning_rate": 9.727067303440238e-06,
      "loss": 0.2003,
      "step": 983
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8700162682064565,
      "learning_rate": 9.726355215544758e-06,
      "loss": 0.1337,
      "step": 984
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9469457222492842,
      "learning_rate": 9.725642226065735e-06,
      "loss": 0.1598,
      "step": 985
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1783659354800284,
      "learning_rate": 9.724928335139174e-06,
      "loss": 0.2245,
      "step": 986
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7058645124873606,
      "learning_rate": 9.724213542901256e-06,
      "loss": 0.2791,
      "step": 987
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9957354935155431,
      "learning_rate": 9.723497849488331e-06,
      "loss": 0.1567,
      "step": 988
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2752089749538313,
      "learning_rate": 9.722781255036924e-06,
      "loss": 0.2548,
      "step": 989
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0701268431126436,
      "learning_rate": 9.722063759683728e-06,
      "loss": 0.191,
      "step": 990
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.132315377283826,
      "learning_rate": 9.721345363565614e-06,
      "loss": 0.2387,
      "step": 991
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2255486120656973,
      "learning_rate": 9.720626066819616e-06,
      "loss": 0.2241,
      "step": 992
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2666622916266597,
      "learning_rate": 9.719905869582947e-06,
      "loss": 0.2435,
      "step": 993
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9715254336011716,
      "learning_rate": 9.719184771992989e-06,
      "loss": 0.2036,
      "step": 994
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2172737665302815,
      "learning_rate": 9.718462774187297e-06,
      "loss": 0.2358,
      "step": 995
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2305014223774375,
      "learning_rate": 9.717739876303596e-06,
      "loss": 0.2055,
      "step": 996
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0968907591994232,
      "learning_rate": 9.717016078479782e-06,
      "loss": 0.2357,
      "step": 997
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1421367810087129,
      "learning_rate": 9.716291380853929e-06,
      "loss": 0.1928,
      "step": 998
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1747326583658761,
      "learning_rate": 9.715565783564272e-06,
      "loss": 0.2512,
      "step": 999
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3058131680070366,
      "learning_rate": 9.714839286749227e-06,
      "loss": 0.2154,
      "step": 1000
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.15948288444323,
      "learning_rate": 9.714111890547377e-06,
      "loss": 0.1765,
      "step": 1001
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1891269750276225,
      "learning_rate": 9.713383595097479e-06,
      "loss": 0.2326,
      "step": 1002
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1290810002785425,
      "learning_rate": 9.712654400538457e-06,
      "loss": 0.169,
      "step": 1003
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2892014103498604,
      "learning_rate": 9.711924307009414e-06,
      "loss": 0.2111,
      "step": 1004
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.197118596960319,
      "learning_rate": 9.711193314649618e-06,
      "loss": 0.227,
      "step": 1005
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2227381394997423,
      "learning_rate": 9.71046142359851e-06,
      "loss": 0.2163,
      "step": 1006
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1958185386785332,
      "learning_rate": 9.709728633995704e-06,
      "loss": 0.2015,
      "step": 1007
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9557314441033037,
      "learning_rate": 9.708994945980983e-06,
      "loss": 0.1864,
      "step": 1008
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9548185436495796,
      "learning_rate": 9.708260359694303e-06,
      "loss": 0.193,
      "step": 1009
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0842099440211261,
      "learning_rate": 9.707524875275796e-06,
      "loss": 0.1946,
      "step": 1010
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9210583003036165,
      "learning_rate": 9.706788492865753e-06,
      "loss": 0.176,
      "step": 1011
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.495003267106425,
      "learning_rate": 9.70605121260465e-06,
      "loss": 0.2768,
      "step": 1012
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.407185405310799,
      "learning_rate": 9.705313034633123e-06,
      "loss": 0.2287,
      "step": 1013
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.041477945536473,
      "learning_rate": 9.704573959091989e-06,
      "loss": 0.1869,
      "step": 1014
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7991879403727413,
      "learning_rate": 9.70383398612223e-06,
      "loss": 0.1591,
      "step": 1015
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.154677799020606,
      "learning_rate": 9.703093115865e-06,
      "loss": 0.2197,
      "step": 1016
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0661693894018154,
      "learning_rate": 9.702351348461624e-06,
      "loss": 0.2371,
      "step": 1017
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3540584962101443,
      "learning_rate": 9.7016086840536e-06,
      "loss": 0.2413,
      "step": 1018
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9273946337523115,
      "learning_rate": 9.7008651227826e-06,
      "loss": 0.1905,
      "step": 1019
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.851235795705078,
      "learning_rate": 9.700120664790456e-06,
      "loss": 0.1759,
      "step": 1020
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.368274937699009,
      "learning_rate": 9.699375310219182e-06,
      "loss": 0.3157,
      "step": 1021
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5333554326693353,
      "learning_rate": 9.69862905921096e-06,
      "loss": 0.3077,
      "step": 1022
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0651338696855408,
      "learning_rate": 9.697881911908141e-06,
      "loss": 0.2076,
      "step": 1023
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2696309353612496,
      "learning_rate": 9.69713386845325e-06,
      "loss": 0.2757,
      "step": 1024
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7858681698772676,
      "learning_rate": 9.696384928988979e-06,
      "loss": 0.168,
      "step": 1025
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2807881468736741,
      "learning_rate": 9.695635093658195e-06,
      "loss": 0.2921,
      "step": 1026
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3593957685474511,
      "learning_rate": 9.694884362603933e-06,
      "loss": 0.2383,
      "step": 1027
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1141293436430735,
      "learning_rate": 9.6941327359694e-06,
      "loss": 0.1906,
      "step": 1028
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.996447091802241,
      "learning_rate": 9.693380213897974e-06,
      "loss": 0.1685,
      "step": 1029
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4300124129844893,
      "learning_rate": 9.692626796533203e-06,
      "loss": 0.2353,
      "step": 1030
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0843082953409242,
      "learning_rate": 9.691872484018807e-06,
      "loss": 0.145,
      "step": 1031
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4332564992900834,
      "learning_rate": 9.691117276498675e-06,
      "loss": 0.2821,
      "step": 1032
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.439966184825289,
      "learning_rate": 9.690361174116871e-06,
      "loss": 0.2474,
      "step": 1033
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2126280203982789,
      "learning_rate": 9.689604177017623e-06,
      "loss": 0.2172,
      "step": 1034
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9951212997777158,
      "learning_rate": 9.688846285345334e-06,
      "loss": 0.2112,
      "step": 1035
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6605901567663932,
      "learning_rate": 9.68808749924458e-06,
      "loss": 0.254,
      "step": 1036
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1385312758597268,
      "learning_rate": 9.6873278188601e-06,
      "loss": 0.2077,
      "step": 1037
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6533924211742017,
      "learning_rate": 9.686567244336809e-06,
      "loss": 0.1594,
      "step": 1038
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9423383102881253,
      "learning_rate": 9.685805775819794e-06,
      "loss": 0.2056,
      "step": 1039
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3222088233691038,
      "learning_rate": 9.685043413454309e-06,
      "loss": 0.2054,
      "step": 1040
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2361435842748218,
      "learning_rate": 9.684280157385777e-06,
      "loss": 0.2044,
      "step": 1041
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1479849706260885,
      "learning_rate": 9.6835160077598e-06,
      "loss": 0.1898,
      "step": 1042
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2204179738239862,
      "learning_rate": 9.682750964722139e-06,
      "loss": 0.2126,
      "step": 1043
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.191400635620407,
      "learning_rate": 9.681985028418733e-06,
      "loss": 0.226,
      "step": 1044
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2117485524237865,
      "learning_rate": 9.68121819899569e-06,
      "loss": 0.2214,
      "step": 1045
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.263817370373972,
      "learning_rate": 9.680450476599288e-06,
      "loss": 0.2106,
      "step": 1046
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1509589921323664,
      "learning_rate": 9.679681861375974e-06,
      "loss": 0.2274,
      "step": 1047
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1717580826603413,
      "learning_rate": 9.678912353472367e-06,
      "loss": 0.2297,
      "step": 1048
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1381281169912367,
      "learning_rate": 9.678141953035255e-06,
      "loss": 0.2384,
      "step": 1049
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1502519140535346,
      "learning_rate": 9.677370660211599e-06,
      "loss": 0.2553,
      "step": 1050
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9323126498846268,
      "learning_rate": 9.676598475148525e-06,
      "loss": 0.1404,
      "step": 1051
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1019815728941744,
      "learning_rate": 9.675825397993336e-06,
      "loss": 0.2174,
      "step": 1052
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.060933607378649,
      "learning_rate": 9.675051428893499e-06,
      "loss": 0.1817,
      "step": 1053
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.196814687062691,
      "learning_rate": 9.674276567996655e-06,
      "loss": 0.2253,
      "step": 1054
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.980535231569342,
      "learning_rate": 9.673500815450612e-06,
      "loss": 0.1644,
      "step": 1055
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9403965107740396,
      "learning_rate": 9.672724171403353e-06,
      "loss": 0.1736,
      "step": 1056
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1468239963032874,
      "learning_rate": 9.671946636003025e-06,
      "loss": 0.1727,
      "step": 1057
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9646642863335593,
      "learning_rate": 9.671168209397951e-06,
      "loss": 0.129,
      "step": 1058
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9411456480082613,
      "learning_rate": 9.670388891736619e-06,
      "loss": 0.1123,
      "step": 1059
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4568307691292408,
      "learning_rate": 9.669608683167688e-06,
      "loss": 0.2429,
      "step": 1060
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.317543325774873,
      "learning_rate": 9.66882758383999e-06,
      "loss": 0.2369,
      "step": 1061
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3857008940508297,
      "learning_rate": 9.668045593902526e-06,
      "loss": 0.1898,
      "step": 1062
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9374683182173494,
      "learning_rate": 9.667262713504462e-06,
      "loss": 0.1904,
      "step": 1063
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3298445907356944,
      "learning_rate": 9.666478942795141e-06,
      "loss": 0.2614,
      "step": 1064
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0358601707257902,
      "learning_rate": 9.66569428192407e-06,
      "loss": 0.2323,
      "step": 1065
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4773712958550964,
      "learning_rate": 9.66490873104093e-06,
      "loss": 0.2465,
      "step": 1066
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9298322349758408,
      "learning_rate": 9.664122290295568e-06,
      "loss": 0.2154,
      "step": 1067
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3128770156509715,
      "learning_rate": 9.663334959838004e-06,
      "loss": 0.246,
      "step": 1068
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9227606039712569,
      "learning_rate": 9.662546739818427e-06,
      "loss": 0.166,
      "step": 1069
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0017074248239382,
      "learning_rate": 9.661757630387197e-06,
      "loss": 0.1774,
      "step": 1070
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3456789286595758,
      "learning_rate": 9.660967631694837e-06,
      "loss": 0.222,
      "step": 1071
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.389175242445496,
      "learning_rate": 9.66017674389205e-06,
      "loss": 0.2228,
      "step": 1072
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0711348420094444,
      "learning_rate": 9.659384967129697e-06,
      "loss": 0.2108,
      "step": 1073
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1753401107045678,
      "learning_rate": 9.65859230155882e-06,
      "loss": 0.2552,
      "step": 1074
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2282517688145893,
      "learning_rate": 9.657798747330622e-06,
      "loss": 0.1795,
      "step": 1075
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2765197164439281,
      "learning_rate": 9.65700430459648e-06,
      "loss": 0.2155,
      "step": 1076
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.286428884728882,
      "learning_rate": 9.656208973507939e-06,
      "loss": 0.2226,
      "step": 1077
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9398344652364614,
      "learning_rate": 9.655412754216714e-06,
      "loss": 0.2332,
      "step": 1078
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7618204014170626,
      "learning_rate": 9.654615646874688e-06,
      "loss": 0.1455,
      "step": 1079
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3334148024938561,
      "learning_rate": 9.653817651633917e-06,
      "loss": 0.2631,
      "step": 1080
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8275839503860144,
      "learning_rate": 9.65301876864662e-06,
      "loss": 0.146,
      "step": 1081
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0333670664282937,
      "learning_rate": 9.652218998065193e-06,
      "loss": 0.1933,
      "step": 1082
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.953891702214587,
      "learning_rate": 9.651418340042195e-06,
      "loss": 0.1859,
      "step": 1083
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9863329055686668,
      "learning_rate": 9.650616794730358e-06,
      "loss": 0.1704,
      "step": 1084
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4697676330241034,
      "learning_rate": 9.649814362282582e-06,
      "loss": 0.287,
      "step": 1085
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8057640066703509,
      "learning_rate": 9.649011042851934e-06,
      "loss": 0.1682,
      "step": 1086
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1945104427606248,
      "learning_rate": 9.648206836591658e-06,
      "loss": 0.2231,
      "step": 1087
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.181247126916834,
      "learning_rate": 9.647401743655156e-06,
      "loss": 0.2209,
      "step": 1088
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8948900479012942,
      "learning_rate": 9.646595764196008e-06,
      "loss": 0.1786,
      "step": 1089
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.461964396200488,
      "learning_rate": 9.645788898367959e-06,
      "loss": 0.2838,
      "step": 1090
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3812613471671893,
      "learning_rate": 9.644981146324923e-06,
      "loss": 0.2737,
      "step": 1091
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0736571783090652,
      "learning_rate": 9.644172508220986e-06,
      "loss": 0.2156,
      "step": 1092
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.02370578463875,
      "learning_rate": 9.6433629842104e-06,
      "loss": 0.2293,
      "step": 1093
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2484897476828807,
      "learning_rate": 9.642552574447587e-06,
      "loss": 0.239,
      "step": 1094
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.363928508165632,
      "learning_rate": 9.641741279087138e-06,
      "loss": 0.2519,
      "step": 1095
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1717482802607935,
      "learning_rate": 9.640929098283813e-06,
      "loss": 0.2748,
      "step": 1096
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2624343923250165,
      "learning_rate": 9.640116032192542e-06,
      "loss": 0.2806,
      "step": 1097
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.107929779089859,
      "learning_rate": 9.639302080968421e-06,
      "loss": 0.2122,
      "step": 1098
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0673894770115837,
      "learning_rate": 9.638487244766718e-06,
      "loss": 0.1841,
      "step": 1099
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.331812314870078,
      "learning_rate": 9.637671523742867e-06,
      "loss": 0.2589,
      "step": 1100
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0740935287024393,
      "learning_rate": 9.636854918052474e-06,
      "loss": 0.1818,
      "step": 1101
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1409708495515327,
      "learning_rate": 9.63603742785131e-06,
      "loss": 0.2251,
      "step": 1102
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7849481191889477,
      "learning_rate": 9.635219053295317e-06,
      "loss": 0.1437,
      "step": 1103
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8381744766641654,
      "learning_rate": 9.634399794540607e-06,
      "loss": 0.1993,
      "step": 1104
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2585710683605937,
      "learning_rate": 9.633579651743456e-06,
      "loss": 0.2095,
      "step": 1105
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9295452518760953,
      "learning_rate": 9.632758625060317e-06,
      "loss": 0.1973,
      "step": 1106
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3496689935773964,
      "learning_rate": 9.6319367146478e-06,
      "loss": 0.1945,
      "step": 1107
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9370847278454489,
      "learning_rate": 9.631113920662693e-06,
      "loss": 0.1424,
      "step": 1108
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0077328210611567,
      "learning_rate": 9.630290243261951e-06,
      "loss": 0.1674,
      "step": 1109
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.22663873410408,
      "learning_rate": 9.629465682602692e-06,
      "loss": 0.2463,
      "step": 1110
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2190325871713967,
      "learning_rate": 9.62864023884221e-06,
      "loss": 0.2225,
      "step": 1111
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2521482160669737,
      "learning_rate": 9.627813912137961e-06,
      "loss": 0.2363,
      "step": 1112
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.007657500532078,
      "learning_rate": 9.626986702647573e-06,
      "loss": 0.1939,
      "step": 1113
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.507306946492507,
      "learning_rate": 9.626158610528844e-06,
      "loss": 0.2493,
      "step": 1114
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.265076730608744,
      "learning_rate": 9.625329635939736e-06,
      "loss": 0.1873,
      "step": 1115
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0423688261066402,
      "learning_rate": 9.624499779038382e-06,
      "loss": 0.1613,
      "step": 1116
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3081869604580505,
      "learning_rate": 9.623669039983082e-06,
      "loss": 0.2334,
      "step": 1117
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0290613379143476,
      "learning_rate": 9.622837418932304e-06,
      "loss": 0.1741,
      "step": 1118
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2956490738772053,
      "learning_rate": 9.622004916044689e-06,
      "loss": 0.2339,
      "step": 1119
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.122511907940771,
      "learning_rate": 9.621171531479038e-06,
      "loss": 0.2258,
      "step": 1120
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9485507819442724,
      "learning_rate": 9.620337265394328e-06,
      "loss": 0.1933,
      "step": 1121
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9187191224146155,
      "learning_rate": 9.6195021179497e-06,
      "loss": 0.1942,
      "step": 1122
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9388905660898165,
      "learning_rate": 9.618666089304463e-06,
      "loss": 0.1444,
      "step": 1123
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.272403016925933,
      "learning_rate": 9.617829179618094e-06,
      "loss": 0.2142,
      "step": 1124
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2261398376817425,
      "learning_rate": 9.61699138905024e-06,
      "loss": 0.2442,
      "step": 1125
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7676726460402571,
      "learning_rate": 9.616152717760718e-06,
      "loss": 0.1406,
      "step": 1126
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1891428583031687,
      "learning_rate": 9.615313165909507e-06,
      "loss": 0.2242,
      "step": 1127
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2589626804370138,
      "learning_rate": 9.614472733656756e-06,
      "loss": 0.1748,
      "step": 1128
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2477201336789072,
      "learning_rate": 9.613631421162786e-06,
      "loss": 0.2565,
      "step": 1129
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1320053879678584,
      "learning_rate": 9.612789228588082e-06,
      "loss": 0.2435,
      "step": 1130
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2149398108545109,
      "learning_rate": 9.611946156093298e-06,
      "loss": 0.198,
      "step": 1131
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1103558749799733,
      "learning_rate": 9.611102203839254e-06,
      "loss": 0.2259,
      "step": 1132
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3114409970976793,
      "learning_rate": 9.610257371986939e-06,
      "loss": 0.1933,
      "step": 1133
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3205481625531033,
      "learning_rate": 9.609411660697515e-06,
      "loss": 0.2232,
      "step": 1134
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.150517183965438,
      "learning_rate": 9.608565070132301e-06,
      "loss": 0.2067,
      "step": 1135
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2886181428742804,
      "learning_rate": 9.607717600452796e-06,
      "loss": 0.2507,
      "step": 1136
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2205286027797577,
      "learning_rate": 9.606869251820654e-06,
      "loss": 0.2149,
      "step": 1137
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5080366893023396,
      "learning_rate": 9.606020024397707e-06,
      "loss": 0.2732,
      "step": 1138
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.612942114421112,
      "learning_rate": 9.60516991834595e-06,
      "loss": 0.2884,
      "step": 1139
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1736802873624017,
      "learning_rate": 9.604318933827546e-06,
      "loss": 0.2205,
      "step": 1140
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0511748412453092,
      "learning_rate": 9.603467071004826e-06,
      "loss": 0.258,
      "step": 1141
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7485823950562321,
      "learning_rate": 9.60261433004029e-06,
      "loss": 0.1764,
      "step": 1142
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.830953394130942,
      "learning_rate": 9.601760711096601e-06,
      "loss": 0.1924,
      "step": 1143
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7769635942376537,
      "learning_rate": 9.600906214336596e-06,
      "loss": 0.1475,
      "step": 1144
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6024804249783293,
      "learning_rate": 9.600050839923274e-06,
      "loss": 0.2696,
      "step": 1145
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9424553875042163,
      "learning_rate": 9.599194588019803e-06,
      "loss": 0.1104,
      "step": 1146
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9136125548331003,
      "learning_rate": 9.598337458789518e-06,
      "loss": 0.2333,
      "step": 1147
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0620346451149478,
      "learning_rate": 9.597479452395923e-06,
      "loss": 0.2128,
      "step": 1148
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.503369834751904,
      "learning_rate": 9.59662056900269e-06,
      "loss": 0.2896,
      "step": 1149
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1686489370250432,
      "learning_rate": 9.595760808773656e-06,
      "loss": 0.2381,
      "step": 1150
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0561995087935456,
      "learning_rate": 9.594900171872824e-06,
      "loss": 0.1653,
      "step": 1151
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2786601533029027,
      "learning_rate": 9.594038658464369e-06,
      "loss": 0.2571,
      "step": 1152
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3732019238916107,
      "learning_rate": 9.593176268712626e-06,
      "loss": 0.2827,
      "step": 1153
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7625454388022433,
      "learning_rate": 9.592313002782105e-06,
      "loss": 0.118,
      "step": 1154
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.332617884504222,
      "learning_rate": 9.591448860837481e-06,
      "loss": 0.2654,
      "step": 1155
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0852506726600877,
      "learning_rate": 9.590583843043591e-06,
      "loss": 0.2346,
      "step": 1156
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1306802305922279,
      "learning_rate": 9.589717949565447e-06,
      "loss": 0.1839,
      "step": 1157
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0436838721985429,
      "learning_rate": 9.588851180568219e-06,
      "loss": 0.1798,
      "step": 1158
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1606320148538645,
      "learning_rate": 9.587983536217254e-06,
      "loss": 0.2446,
      "step": 1159
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8910815432364031,
      "learning_rate": 9.587115016678056e-06,
      "loss": 0.1616,
      "step": 1160
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2409080482740522,
      "learning_rate": 9.586245622116304e-06,
      "loss": 0.2381,
      "step": 1161
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3754486225804794,
      "learning_rate": 9.585375352697841e-06,
      "loss": 0.2732,
      "step": 1162
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8874894145105835,
      "learning_rate": 9.584504208588676e-06,
      "loss": 0.1986,
      "step": 1163
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2904501063591796,
      "learning_rate": 9.583632189954984e-06,
      "loss": 0.1995,
      "step": 1164
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0841422014660527,
      "learning_rate": 9.58275929696311e-06,
      "loss": 0.2076,
      "step": 1165
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2366742455912014,
      "learning_rate": 9.581885529779565e-06,
      "loss": 0.1946,
      "step": 1166
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.329732326807086,
      "learning_rate": 9.581010888571022e-06,
      "loss": 0.2725,
      "step": 1167
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9193609051332604,
      "learning_rate": 9.580135373504329e-06,
      "loss": 0.2018,
      "step": 1168
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2031463627541326,
      "learning_rate": 9.579258984746493e-06,
      "loss": 0.2622,
      "step": 1169
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9628977623325746,
      "learning_rate": 9.578381722464693e-06,
      "loss": 0.2112,
      "step": 1170
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1827385325824264,
      "learning_rate": 9.577503586826273e-06,
      "loss": 0.2019,
      "step": 1171
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.988483598878411,
      "learning_rate": 9.576624577998742e-06,
      "loss": 0.2126,
      "step": 1172
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2009479391157616,
      "learning_rate": 9.575744696149777e-06,
      "loss": 0.1792,
      "step": 1173
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8911087960323849,
      "learning_rate": 9.57486394144722e-06,
      "loss": 0.0925,
      "step": 1174
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8841993379951325,
      "learning_rate": 9.573982314059082e-06,
      "loss": 0.161,
      "step": 1175
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.195730955197461,
      "learning_rate": 9.573099814153541e-06,
      "loss": 0.1889,
      "step": 1176
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2660932986371676,
      "learning_rate": 9.572216441898937e-06,
      "loss": 0.241,
      "step": 1177
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8975349864219017,
      "learning_rate": 9.57133219746378e-06,
      "loss": 0.2537,
      "step": 1178
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4019651766461163,
      "learning_rate": 9.570447081016746e-06,
      "loss": 0.246,
      "step": 1179
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.332356412132944,
      "learning_rate": 9.569561092726676e-06,
      "loss": 0.2303,
      "step": 1180
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2350372706886266,
      "learning_rate": 9.568674232762578e-06,
      "loss": 0.1765,
      "step": 1181
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1132395081821318,
      "learning_rate": 9.567786501293628e-06,
      "loss": 0.1933,
      "step": 1182
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1112037739761818,
      "learning_rate": 9.566897898489165e-06,
      "loss": 0.1725,
      "step": 1183
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1456869152498839,
      "learning_rate": 9.566008424518696e-06,
      "loss": 0.2347,
      "step": 1184
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3266776695053482,
      "learning_rate": 9.565118079551896e-06,
      "loss": 0.2207,
      "step": 1185
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4163991496145405,
      "learning_rate": 9.5642268637586e-06,
      "loss": 0.2735,
      "step": 1186
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8547791656359769,
      "learning_rate": 9.563334777308819e-06,
      "loss": 0.2054,
      "step": 1187
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2827578768680523,
      "learning_rate": 9.56244182037272e-06,
      "loss": 0.1815,
      "step": 1188
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3105318694060875,
      "learning_rate": 9.561547993120642e-06,
      "loss": 0.2212,
      "step": 1189
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2758431858877899,
      "learning_rate": 9.560653295723088e-06,
      "loss": 0.2979,
      "step": 1190
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9732365465609929,
      "learning_rate": 9.559757728350728e-06,
      "loss": 0.1916,
      "step": 1191
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.916624608462773,
      "learning_rate": 9.558861291174397e-06,
      "loss": 0.1518,
      "step": 1192
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1890458142133697,
      "learning_rate": 9.557963984365097e-06,
      "loss": 0.2023,
      "step": 1193
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1382104945799958,
      "learning_rate": 9.557065808093994e-06,
      "loss": 0.2461,
      "step": 1194
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2309075233911893,
      "learning_rate": 9.556166762532422e-06,
      "loss": 0.2408,
      "step": 1195
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.93238723906668,
      "learning_rate": 9.555266847851881e-06,
      "loss": 0.1901,
      "step": 1196
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.118775341971752,
      "learning_rate": 9.554366064224032e-06,
      "loss": 0.1738,
      "step": 1197
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9333945866797656,
      "learning_rate": 9.55346441182071e-06,
      "loss": 0.1787,
      "step": 1198
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9603617560606039,
      "learning_rate": 9.552561890813907e-06,
      "loss": 0.1439,
      "step": 1199
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.301470499826816,
      "learning_rate": 9.551658501375787e-06,
      "loss": 0.2385,
      "step": 1200
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3238337270269869,
      "learning_rate": 9.55075424367868e-06,
      "loss": 0.2216,
      "step": 1201
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9787082848257809,
      "learning_rate": 9.549849117895074e-06,
      "loss": 0.2062,
      "step": 1202
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0397382637455896,
      "learning_rate": 9.54894312419763e-06,
      "loss": 0.2081,
      "step": 1203
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.256640869578597,
      "learning_rate": 9.548036262759172e-06,
      "loss": 0.2186,
      "step": 1204
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1060298605596957,
      "learning_rate": 9.547128533752693e-06,
      "loss": 0.1961,
      "step": 1205
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1137539924866184,
      "learning_rate": 9.546219937351344e-06,
      "loss": 0.1655,
      "step": 1206
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1129511486675097,
      "learning_rate": 9.545310473728447e-06,
      "loss": 0.2434,
      "step": 1207
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0078928606108777,
      "learning_rate": 9.54440014305749e-06,
      "loss": 0.1507,
      "step": 1208
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7866304939962052,
      "learning_rate": 9.543488945512121e-06,
      "loss": 0.2085,
      "step": 1209
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2074730519554566,
      "learning_rate": 9.542576881266162e-06,
      "loss": 0.205,
      "step": 1210
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8731586874153628,
      "learning_rate": 9.541663950493591e-06,
      "loss": 0.1499,
      "step": 1211
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4535340300532613,
      "learning_rate": 9.540750153368556e-06,
      "loss": 0.2777,
      "step": 1212
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1220003768123943,
      "learning_rate": 9.539835490065373e-06,
      "loss": 0.1722,
      "step": 1213
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9293172636826897,
      "learning_rate": 9.538919960758518e-06,
      "loss": 0.2165,
      "step": 1214
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.717767948535582,
      "learning_rate": 9.538003565622631e-06,
      "loss": 0.1464,
      "step": 1215
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2406611386746451,
      "learning_rate": 9.537086304832527e-06,
      "loss": 0.1873,
      "step": 1216
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2156588189106918,
      "learning_rate": 9.536168178563174e-06,
      "loss": 0.1893,
      "step": 1217
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8670583141941463,
      "learning_rate": 9.535249186989715e-06,
      "loss": 0.1505,
      "step": 1218
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2456195580812153,
      "learning_rate": 9.53432933028745e-06,
      "loss": 0.2158,
      "step": 1219
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2614697455449384,
      "learning_rate": 9.533408608631851e-06,
      "loss": 0.2438,
      "step": 1220
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3493355674377996,
      "learning_rate": 9.53248702219855e-06,
      "loss": 0.2278,
      "step": 1221
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5957818940648346,
      "learning_rate": 9.531564571163346e-06,
      "loss": 0.0824,
      "step": 1222
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0804789206940453,
      "learning_rate": 9.530641255702201e-06,
      "loss": 0.2029,
      "step": 1223
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8469354983254268,
      "learning_rate": 9.529717075991247e-06,
      "loss": 0.1601,
      "step": 1224
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0113073195165758,
      "learning_rate": 9.528792032206773e-06,
      "loss": 0.1764,
      "step": 1225
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.267525467298154,
      "learning_rate": 9.527866124525242e-06,
      "loss": 0.2751,
      "step": 1226
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2334409523425514,
      "learning_rate": 9.526939353123274e-06,
      "loss": 0.1773,
      "step": 1227
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3631160607982924,
      "learning_rate": 9.526011718177657e-06,
      "loss": 0.2335,
      "step": 1228
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8135715870974289,
      "learning_rate": 9.525083219865345e-06,
      "loss": 0.239,
      "step": 1229
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2681547933166373,
      "learning_rate": 9.524153858363452e-06,
      "loss": 0.2378,
      "step": 1230
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1566594306675793,
      "learning_rate": 9.523223633849264e-06,
      "loss": 0.2192,
      "step": 1231
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9816764184940558,
      "learning_rate": 9.522292546500225e-06,
      "loss": 0.1815,
      "step": 1232
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4757155973825244,
      "learning_rate": 9.521360596493946e-06,
      "loss": 0.2151,
      "step": 1233
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0479133066838784,
      "learning_rate": 9.520427784008206e-06,
      "loss": 0.1298,
      "step": 1234
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2207224037624997,
      "learning_rate": 9.519494109220939e-06,
      "loss": 0.2178,
      "step": 1235
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1355437706837639,
      "learning_rate": 9.518559572310256e-06,
      "loss": 0.2018,
      "step": 1236
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.226501876774218,
      "learning_rate": 9.51762417345442e-06,
      "loss": 0.2836,
      "step": 1237
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2589881509290475,
      "learning_rate": 9.51668791283187e-06,
      "loss": 0.2633,
      "step": 1238
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2047667148502224,
      "learning_rate": 9.515750790621201e-06,
      "loss": 0.1603,
      "step": 1239
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0600494264355125,
      "learning_rate": 9.514812807001174e-06,
      "loss": 0.2492,
      "step": 1240
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1604624262350882,
      "learning_rate": 9.51387396215072e-06,
      "loss": 0.248,
      "step": 1241
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1132782973380941,
      "learning_rate": 9.512934256248927e-06,
      "loss": 0.1866,
      "step": 1242
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3643811330056275,
      "learning_rate": 9.51199368947505e-06,
      "loss": 0.2322,
      "step": 1243
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0023400245131475,
      "learning_rate": 9.511052262008507e-06,
      "loss": 0.1354,
      "step": 1244
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2171583129300434,
      "learning_rate": 9.510109974028885e-06,
      "loss": 0.1706,
      "step": 1245
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1418114862171553,
      "learning_rate": 9.509166825715929e-06,
      "loss": 0.2676,
      "step": 1246
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4037161666347713,
      "learning_rate": 9.508222817249554e-06,
      "loss": 0.2704,
      "step": 1247
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1212998726399668,
      "learning_rate": 9.507277948809833e-06,
      "loss": 0.2061,
      "step": 1248
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5139757886853946,
      "learning_rate": 9.506332220577005e-06,
      "loss": 0.3237,
      "step": 1249
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4252448651094596,
      "learning_rate": 9.505385632731477e-06,
      "loss": 0.2565,
      "step": 1250
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.188770774749532,
      "learning_rate": 9.504438185453812e-06,
      "loss": 0.2199,
      "step": 1251
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.246840475938388,
      "learning_rate": 9.503489878924749e-06,
      "loss": 0.2475,
      "step": 1252
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.258458838261731,
      "learning_rate": 9.502540713325178e-06,
      "loss": 0.2013,
      "step": 1253
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1803411937495225,
      "learning_rate": 9.501590688836161e-06,
      "loss": 0.2006,
      "step": 1254
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.074117413298923,
      "learning_rate": 9.50063980563892e-06,
      "loss": 0.2209,
      "step": 1255
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0576948345794406,
      "learning_rate": 9.499688063914845e-06,
      "loss": 0.189,
      "step": 1256
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.366206019574188,
      "learning_rate": 9.498735463845486e-06,
      "loss": 0.2656,
      "step": 1257
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8367510786001132,
      "learning_rate": 9.497782005612555e-06,
      "loss": 0.1821,
      "step": 1258
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0547450207674944,
      "learning_rate": 9.496827689397933e-06,
      "loss": 0.1891,
      "step": 1259
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3281487196663453,
      "learning_rate": 9.495872515383661e-06,
      "loss": 0.22,
      "step": 1260
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1931805709169878,
      "learning_rate": 9.494916483751948e-06,
      "loss": 0.2292,
      "step": 1261
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.190429146867582,
      "learning_rate": 9.493959594685158e-06,
      "loss": 0.216,
      "step": 1262
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1641512939998748,
      "learning_rate": 9.493001848365828e-06,
      "loss": 0.2395,
      "step": 1263
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8590596009316729,
      "learning_rate": 9.492043244976652e-06,
      "loss": 0.1347,
      "step": 1264
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.396173402516045,
      "learning_rate": 9.491083784700494e-06,
      "loss": 0.2496,
      "step": 1265
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.071488111153972,
      "learning_rate": 9.49012346772037e-06,
      "loss": 0.1833,
      "step": 1266
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7557918296621187,
      "learning_rate": 9.489162294219476e-06,
      "loss": 0.1276,
      "step": 1267
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3201148710200241,
      "learning_rate": 9.488200264381155e-06,
      "loss": 0.2503,
      "step": 1268
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1817747437793256,
      "learning_rate": 9.487237378388924e-06,
      "loss": 0.2193,
      "step": 1269
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.465753168830945,
      "learning_rate": 9.486273636426459e-06,
      "loss": 0.2513,
      "step": 1270
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2888153592671217,
      "learning_rate": 9.4853090386776e-06,
      "loss": 0.1965,
      "step": 1271
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4262373215542445,
      "learning_rate": 9.484343585326349e-06,
      "loss": 0.2728,
      "step": 1272
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1838141194723395,
      "learning_rate": 9.483377276556876e-06,
      "loss": 0.188,
      "step": 1273
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0814583714173214,
      "learning_rate": 9.48241011255351e-06,
      "loss": 0.1941,
      "step": 1274
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3408041244275855,
      "learning_rate": 9.481442093500738e-06,
      "loss": 0.2715,
      "step": 1275
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.183992329860293,
      "learning_rate": 9.480473219583227e-06,
      "loss": 0.1776,
      "step": 1276
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2227876370853041,
      "learning_rate": 9.479503490985786e-06,
      "loss": 0.1952,
      "step": 1277
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0745458675231347,
      "learning_rate": 9.478532907893403e-06,
      "loss": 0.1534,
      "step": 1278
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0444229169451351,
      "learning_rate": 9.47756147049122e-06,
      "loss": 0.1527,
      "step": 1279
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0378638115433003,
      "learning_rate": 9.476589178964548e-06,
      "loss": 0.1469,
      "step": 1280
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1254333526734879,
      "learning_rate": 9.475616033498855e-06,
      "loss": 0.2028,
      "step": 1281
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3880828240974858,
      "learning_rate": 9.474642034279778e-06,
      "loss": 0.2649,
      "step": 1282
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.932313714615273,
      "learning_rate": 9.473667181493111e-06,
      "loss": 0.221,
      "step": 1283
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.054437678899713,
      "learning_rate": 9.472691475324816e-06,
      "loss": 0.2174,
      "step": 1284
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1197854244450405,
      "learning_rate": 9.471714915961015e-06,
      "loss": 0.192,
      "step": 1285
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1231208230220715,
      "learning_rate": 9.47073750358799e-06,
      "loss": 0.2048,
      "step": 1286
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7640312112234083,
      "learning_rate": 9.469759238392192e-06,
      "loss": 0.156,
      "step": 1287
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1534497251578217,
      "learning_rate": 9.46878012056023e-06,
      "loss": 0.206,
      "step": 1288
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9763029554570702,
      "learning_rate": 9.46780015027888e-06,
      "loss": 0.194,
      "step": 1289
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1772949888143258,
      "learning_rate": 9.466819327735076e-06,
      "loss": 0.255,
      "step": 1290
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2895104874750818,
      "learning_rate": 9.465837653115916e-06,
      "loss": 0.2381,
      "step": 1291
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9183365522803708,
      "learning_rate": 9.46485512660866e-06,
      "loss": 0.1361,
      "step": 1292
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4469690557016106,
      "learning_rate": 9.463871748400734e-06,
      "loss": 0.2646,
      "step": 1293
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9987004330807961,
      "learning_rate": 9.462887518679722e-06,
      "loss": 0.2046,
      "step": 1294
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.000827125202876,
      "learning_rate": 9.461902437633374e-06,
      "loss": 0.1816,
      "step": 1295
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0006403762610545,
      "learning_rate": 9.460916505449598e-06,
      "loss": 0.2129,
      "step": 1296
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3623446086305673,
      "learning_rate": 9.45992972231647e-06,
      "loss": 0.1984,
      "step": 1297
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0663855119826988,
      "learning_rate": 9.458942088422223e-06,
      "loss": 0.215,
      "step": 1298
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1908652701590956,
      "learning_rate": 9.457953603955257e-06,
      "loss": 0.2188,
      "step": 1299
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2291552627243827,
      "learning_rate": 9.45696426910413e-06,
      "loss": 0.2266,
      "step": 1300
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0449755274051664,
      "learning_rate": 9.455974084057564e-06,
      "loss": 0.1969,
      "step": 1301
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2749205153211525,
      "learning_rate": 9.454983049004448e-06,
      "loss": 0.2532,
      "step": 1302
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.312484380921904,
      "learning_rate": 9.45399116413382e-06,
      "loss": 0.2761,
      "step": 1303
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.965773501644433,
      "learning_rate": 9.452998429634896e-06,
      "loss": 0.2191,
      "step": 1304
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0342458562020738,
      "learning_rate": 9.452004845697045e-06,
      "loss": 0.1886,
      "step": 1305
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0778722268713659,
      "learning_rate": 9.451010412509796e-06,
      "loss": 0.2236,
      "step": 1306
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1220508514640701,
      "learning_rate": 9.450015130262848e-06,
      "loss": 0.2301,
      "step": 1307
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3077897430058296,
      "learning_rate": 9.449018999146058e-06,
      "loss": 0.2215,
      "step": 1308
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.017601901023007,
      "learning_rate": 9.448022019349441e-06,
      "loss": 0.1711,
      "step": 1309
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2303985657369778,
      "learning_rate": 9.447024191063179e-06,
      "loss": 0.2065,
      "step": 1310
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.065178304903982,
      "learning_rate": 9.446025514477615e-06,
      "loss": 0.1802,
      "step": 1311
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9152899306754386,
      "learning_rate": 9.445025989783254e-06,
      "loss": 0.222,
      "step": 1312
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2774730013156146,
      "learning_rate": 9.44402561717076e-06,
      "loss": 0.2308,
      "step": 1313
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.196692227343428,
      "learning_rate": 9.443024396830961e-06,
      "loss": 0.2091,
      "step": 1314
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1471203990063439,
      "learning_rate": 9.442022328954848e-06,
      "loss": 0.2202,
      "step": 1315
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.46989321066294,
      "learning_rate": 9.44101941373357e-06,
      "loss": 0.2879,
      "step": 1316
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7453370771669499,
      "learning_rate": 9.44001565135844e-06,
      "loss": 0.1792,
      "step": 1317
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1165493510540023,
      "learning_rate": 9.439011042020934e-06,
      "loss": 0.1985,
      "step": 1318
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9617976217890614,
      "learning_rate": 9.438005585912687e-06,
      "loss": 0.1939,
      "step": 1319
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9814035128469566,
      "learning_rate": 9.436999283225497e-06,
      "loss": 0.1772,
      "step": 1320
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3091382084549652,
      "learning_rate": 9.435992134151319e-06,
      "loss": 0.1855,
      "step": 1321
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.047718291210123,
      "learning_rate": 9.43498413888228e-06,
      "loss": 0.2247,
      "step": 1322
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1954986300364343,
      "learning_rate": 9.433975297610657e-06,
      "loss": 0.2062,
      "step": 1323
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7724632292834033,
      "learning_rate": 9.432965610528896e-06,
      "loss": 0.16,
      "step": 1324
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9495153348509582,
      "learning_rate": 9.431955077829596e-06,
      "loss": 0.2001,
      "step": 1325
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8530716164033434,
      "learning_rate": 9.430943699705531e-06,
      "loss": 0.14,
      "step": 1326
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.160185288576289,
      "learning_rate": 9.429931476349621e-06,
      "loss": 0.2219,
      "step": 1327
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.044572332259646,
      "learning_rate": 9.428918407954959e-06,
      "loss": 0.1698,
      "step": 1328
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3509907544400608,
      "learning_rate": 9.427904494714791e-06,
      "loss": 0.222,
      "step": 1329
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9826549139048116,
      "learning_rate": 9.42688973682253e-06,
      "loss": 0.2254,
      "step": 1330
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2831019090207385,
      "learning_rate": 9.425874134471748e-06,
      "loss": 0.2337,
      "step": 1331
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7867169115187318,
      "learning_rate": 9.424857687856177e-06,
      "loss": 0.2099,
      "step": 1332
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9799146431161838,
      "learning_rate": 9.423840397169711e-06,
      "loss": 0.2259,
      "step": 1333
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2401579544574053,
      "learning_rate": 9.422822262606403e-06,
      "loss": 0.2345,
      "step": 1334
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2175589818755919,
      "learning_rate": 9.421803284360472e-06,
      "loss": 0.2124,
      "step": 1335
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2379111907003066,
      "learning_rate": 9.420783462626294e-06,
      "loss": 0.2068,
      "step": 1336
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1213833944311444,
      "learning_rate": 9.419762797598404e-06,
      "loss": 0.2174,
      "step": 1337
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8353186280369437,
      "learning_rate": 9.418741289471505e-06,
      "loss": 0.1776,
      "step": 1338
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1434792743099975,
      "learning_rate": 9.417718938440454e-06,
      "loss": 0.1929,
      "step": 1339
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.898059292139711,
      "learning_rate": 9.41669574470027e-06,
      "loss": 0.2039,
      "step": 1340
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.067415795889318,
      "learning_rate": 9.415671708446139e-06,
      "loss": 0.2127,
      "step": 1341
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9298681812762737,
      "learning_rate": 9.414646829873397e-06,
      "loss": 0.1685,
      "step": 1342
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2274297835645902,
      "learning_rate": 9.41362110917755e-06,
      "loss": 0.2569,
      "step": 1343
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0688644385678754,
      "learning_rate": 9.412594546554257e-06,
      "loss": 0.2064,
      "step": 1344
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9456053379909806,
      "learning_rate": 9.411567142199345e-06,
      "loss": 0.1788,
      "step": 1345
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0480112315953471,
      "learning_rate": 9.410538896308799e-06,
      "loss": 0.1492,
      "step": 1346
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0458523819267527,
      "learning_rate": 9.409509809078761e-06,
      "loss": 0.1983,
      "step": 1347
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6531864764107355,
      "learning_rate": 9.408479880705538e-06,
      "loss": 0.1444,
      "step": 1348
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2200668029224015,
      "learning_rate": 9.407449111385596e-06,
      "loss": 0.2148,
      "step": 1349
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6683979689319439,
      "learning_rate": 9.406417501315559e-06,
      "loss": 0.1177,
      "step": 1350
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0114290606470093,
      "learning_rate": 9.405385050692213e-06,
      "loss": 0.1603,
      "step": 1351
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.323446424933365,
      "learning_rate": 9.404351759712509e-06,
      "loss": 0.2489,
      "step": 1352
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1651874870976917,
      "learning_rate": 9.40331762857355e-06,
      "loss": 0.2225,
      "step": 1353
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0863994380675386,
      "learning_rate": 9.402282657472605e-06,
      "loss": 0.1835,
      "step": 1354
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.265904483649397,
      "learning_rate": 9.401246846607101e-06,
      "loss": 0.1914,
      "step": 1355
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1772677785218657,
      "learning_rate": 9.400210196174626e-06,
      "loss": 0.2028,
      "step": 1356
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0479202504151102,
      "learning_rate": 9.39917270637293e-06,
      "loss": 0.2039,
      "step": 1357
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0738800219245066,
      "learning_rate": 9.398134377399916e-06,
      "loss": 0.1626,
      "step": 1358
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.38679200397081,
      "learning_rate": 9.397095209453657e-06,
      "loss": 0.2529,
      "step": 1359
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8315418187087015,
      "learning_rate": 9.39605520273238e-06,
      "loss": 0.1387,
      "step": 1360
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3335374890830352,
      "learning_rate": 9.395014357434472e-06,
      "loss": 0.192,
      "step": 1361
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9926515361930728,
      "learning_rate": 9.393972673758482e-06,
      "loss": 0.1701,
      "step": 1362
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9599779216582695,
      "learning_rate": 9.392930151903118e-06,
      "loss": 0.2065,
      "step": 1363
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2312987980528967,
      "learning_rate": 9.391886792067247e-06,
      "loss": 0.2031,
      "step": 1364
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9135015068916923,
      "learning_rate": 9.390842594449897e-06,
      "loss": 0.1704,
      "step": 1365
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1146729671723443,
      "learning_rate": 9.389797559250257e-06,
      "loss": 0.2196,
      "step": 1366
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9898532899813796,
      "learning_rate": 9.388751686667674e-06,
      "loss": 0.2002,
      "step": 1367
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4150636690251337,
      "learning_rate": 9.387704976901654e-06,
      "loss": 0.2687,
      "step": 1368
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1073353375084933,
      "learning_rate": 9.386657430151864e-06,
      "loss": 0.1936,
      "step": 1369
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8192450284298489,
      "learning_rate": 9.385609046618132e-06,
      "loss": 0.1967,
      "step": 1370
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0186746820774528,
      "learning_rate": 9.384559826500441e-06,
      "loss": 0.1832,
      "step": 1371
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3577984999649868,
      "learning_rate": 9.38350976999894e-06,
      "loss": 0.261,
      "step": 1372
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1559700076029154,
      "learning_rate": 9.382458877313933e-06,
      "loss": 0.2373,
      "step": 1373
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.193362364854207,
      "learning_rate": 9.381407148645885e-06,
      "loss": 0.1896,
      "step": 1374
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9417756543783571,
      "learning_rate": 9.380354584195419e-06,
      "loss": 0.1662,
      "step": 1375
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8407304748133302,
      "learning_rate": 9.379301184163321e-06,
      "loss": 0.2,
      "step": 1376
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0321581781479057,
      "learning_rate": 9.37824694875053e-06,
      "loss": 0.1887,
      "step": 1377
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9503765447870464,
      "learning_rate": 9.377191878158153e-06,
      "loss": 0.1865,
      "step": 1378
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8824765117006864,
      "learning_rate": 9.37613597258745e-06,
      "loss": 0.1717,
      "step": 1379
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2134470789987946,
      "learning_rate": 9.375079232239841e-06,
      "loss": 0.2536,
      "step": 1380
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1066167707987136,
      "learning_rate": 9.374021657316906e-06,
      "loss": 0.1451,
      "step": 1381
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7892429218553653,
      "learning_rate": 9.372963248020385e-06,
      "loss": 0.1567,
      "step": 1382
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9244167988713291,
      "learning_rate": 9.371904004552179e-06,
      "loss": 0.182,
      "step": 1383
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1247757172092325,
      "learning_rate": 9.370843927114342e-06,
      "loss": 0.2085,
      "step": 1384
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.518861098449288,
      "learning_rate": 9.369783015909091e-06,
      "loss": 0.2338,
      "step": 1385
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.879380615805842,
      "learning_rate": 9.368721271138804e-06,
      "loss": 0.1949,
      "step": 1386
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2165336370192013,
      "learning_rate": 9.367658693006017e-06,
      "loss": 0.2301,
      "step": 1387
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8402258784828598,
      "learning_rate": 9.36659528171342e-06,
      "loss": 0.1479,
      "step": 1388
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0787272604799072,
      "learning_rate": 9.365531037463866e-06,
      "loss": 0.2418,
      "step": 1389
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1124918775863204,
      "learning_rate": 9.364465960460372e-06,
      "loss": 0.234,
      "step": 1390
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0518881476987085,
      "learning_rate": 9.363400050906102e-06,
      "loss": 0.1993,
      "step": 1391
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1784561199017143,
      "learning_rate": 9.362333309004387e-06,
      "loss": 0.1975,
      "step": 1392
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1725914410680207,
      "learning_rate": 9.361265734958719e-06,
      "loss": 0.2364,
      "step": 1393
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0793480811987783,
      "learning_rate": 9.36019732897274e-06,
      "loss": 0.2104,
      "step": 1394
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1988204162624065,
      "learning_rate": 9.359128091250258e-06,
      "loss": 0.2395,
      "step": 1395
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9820427637522303,
      "learning_rate": 9.358058021995236e-06,
      "loss": 0.2337,
      "step": 1396
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0176308653089274,
      "learning_rate": 9.356987121411798e-06,
      "loss": 0.2087,
      "step": 1397
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9011558639008038,
      "learning_rate": 9.355915389704224e-06,
      "loss": 0.1822,
      "step": 1398
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8141417231083634,
      "learning_rate": 9.354842827076955e-06,
      "loss": 0.0955,
      "step": 1399
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.285450697716775,
      "learning_rate": 9.35376943373459e-06,
      "loss": 0.2534,
      "step": 1400
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.014195600701642,
      "learning_rate": 9.352695209881883e-06,
      "loss": 0.1814,
      "step": 1401
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1940259815316645,
      "learning_rate": 9.351620155723753e-06,
      "loss": 0.2302,
      "step": 1402
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.098156741687081,
      "learning_rate": 9.35054427146527e-06,
      "loss": 0.2072,
      "step": 1403
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0666109952590201,
      "learning_rate": 9.34946755731167e-06,
      "loss": 0.2173,
      "step": 1404
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1533726973528506,
      "learning_rate": 9.34839001346834e-06,
      "loss": 0.1667,
      "step": 1405
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.80090677438408,
      "learning_rate": 9.34731164014083e-06,
      "loss": 0.194,
      "step": 1406
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0462866326412439,
      "learning_rate": 9.346232437534848e-06,
      "loss": 0.2093,
      "step": 1407
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8836908236450278,
      "learning_rate": 9.345152405856257e-06,
      "loss": 0.1829,
      "step": 1408
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1611138336085767,
      "learning_rate": 9.344071545311082e-06,
      "loss": 0.2438,
      "step": 1409
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0376402115577898,
      "learning_rate": 9.342989856105502e-06,
      "loss": 0.1692,
      "step": 1410
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1248958809118115,
      "learning_rate": 9.34190733844586e-06,
      "loss": 0.2018,
      "step": 1411
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9432288564211504,
      "learning_rate": 9.340823992538647e-06,
      "loss": 0.1911,
      "step": 1412
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0098091514659708,
      "learning_rate": 9.339739818590526e-06,
      "loss": 0.1461,
      "step": 1413
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9916106804922664,
      "learning_rate": 9.338654816808306e-06,
      "loss": 0.155,
      "step": 1414
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1056666631855174,
      "learning_rate": 9.337568987398959e-06,
      "loss": 0.1781,
      "step": 1415
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1233792108377931,
      "learning_rate": 9.336482330569616e-06,
      "loss": 0.2121,
      "step": 1416
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.117179746035687,
      "learning_rate": 9.335394846527559e-06,
      "loss": 0.213,
      "step": 1417
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8189464760452472,
      "learning_rate": 9.334306535480237e-06,
      "loss": 0.2013,
      "step": 1418
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0223864348414202,
      "learning_rate": 9.333217397635255e-06,
      "loss": 0.2242,
      "step": 1419
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4267561211107003,
      "learning_rate": 9.332127433200366e-06,
      "loss": 0.2931,
      "step": 1420
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7233540512927386,
      "learning_rate": 9.331036642383493e-06,
      "loss": 0.1328,
      "step": 1421
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.105557245598516,
      "learning_rate": 9.329945025392711e-06,
      "loss": 0.201,
      "step": 1422
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.032154774483685,
      "learning_rate": 9.328852582436252e-06,
      "loss": 0.1795,
      "step": 1423
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2256741970481615,
      "learning_rate": 9.327759313722506e-06,
      "loss": 0.2498,
      "step": 1424
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.108223479243863,
      "learning_rate": 9.326665219460025e-06,
      "loss": 0.2346,
      "step": 1425
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1521235820543951,
      "learning_rate": 9.32557029985751e-06,
      "loss": 0.1866,
      "step": 1426
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4342270070100993,
      "learning_rate": 9.324474555123827e-06,
      "loss": 0.2495,
      "step": 1427
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1781587420026451,
      "learning_rate": 9.323377985467997e-06,
      "loss": 0.2036,
      "step": 1428
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0458754278225715,
      "learning_rate": 9.322280591099197e-06,
      "loss": 0.2225,
      "step": 1429
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2163620164959257,
      "learning_rate": 9.32118237222676e-06,
      "loss": 0.215,
      "step": 1430
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5506525727095646,
      "learning_rate": 9.320083329060183e-06,
      "loss": 0.1756,
      "step": 1431
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7336673036105852,
      "learning_rate": 9.318983461809112e-06,
      "loss": 0.1087,
      "step": 1432
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.135412560313661,
      "learning_rate": 9.317882770683356e-06,
      "loss": 0.1871,
      "step": 1433
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.035614041080044,
      "learning_rate": 9.31678125589288e-06,
      "loss": 0.1549,
      "step": 1434
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9467089464593578,
      "learning_rate": 9.315678917647802e-06,
      "loss": 0.1953,
      "step": 1435
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0869166707737419,
      "learning_rate": 9.3145757561584e-06,
      "loss": 0.1927,
      "step": 1436
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9029920088881618,
      "learning_rate": 9.313471771635115e-06,
      "loss": 0.1602,
      "step": 1437
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7819406102217891,
      "learning_rate": 9.312366964288532e-06,
      "loss": 0.1388,
      "step": 1438
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2064241316243431,
      "learning_rate": 9.311261334329407e-06,
      "loss": 0.2481,
      "step": 1439
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1483740809539507,
      "learning_rate": 9.310154881968644e-06,
      "loss": 0.2325,
      "step": 1440
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3956731157403215,
      "learning_rate": 9.309047607417303e-06,
      "loss": 0.2298,
      "step": 1441
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3479740024859597,
      "learning_rate": 9.307939510886607e-06,
      "loss": 0.2794,
      "step": 1442
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4005919299431968,
      "learning_rate": 9.306830592587932e-06,
      "loss": 0.292,
      "step": 1443
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0348960776233804,
      "learning_rate": 9.305720852732811e-06,
      "loss": 0.1954,
      "step": 1444
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3564956036231,
      "learning_rate": 9.304610291532936e-06,
      "loss": 0.2469,
      "step": 1445
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1481882661372314,
      "learning_rate": 9.303498909200152e-06,
      "loss": 0.2169,
      "step": 1446
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0851299227960347,
      "learning_rate": 9.302386705946465e-06,
      "loss": 0.2376,
      "step": 1447
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0103811494054087,
      "learning_rate": 9.301273681984032e-06,
      "loss": 0.222,
      "step": 1448
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3603348672481066,
      "learning_rate": 9.30015983752517e-06,
      "loss": 0.2646,
      "step": 1449
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.357057166028174,
      "learning_rate": 9.299045172782356e-06,
      "loss": 0.2627,
      "step": 1450
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2032986124004126,
      "learning_rate": 9.297929687968216e-06,
      "loss": 0.2115,
      "step": 1451
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1690541548057256,
      "learning_rate": 9.296813383295535e-06,
      "loss": 0.208,
      "step": 1452
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3489356111233854,
      "learning_rate": 9.29569625897726e-06,
      "loss": 0.2267,
      "step": 1453
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1269042276837147,
      "learning_rate": 9.294578315226488e-06,
      "loss": 0.2406,
      "step": 1454
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1341300489118624,
      "learning_rate": 9.293459552256472e-06,
      "loss": 0.2395,
      "step": 1455
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0983926936213642,
      "learning_rate": 9.292339970280624e-06,
      "loss": 0.1633,
      "step": 1456
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0619820731874734,
      "learning_rate": 9.291219569512514e-06,
      "loss": 0.2141,
      "step": 1457
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1884771907148304,
      "learning_rate": 9.290098350165866e-06,
      "loss": 0.2073,
      "step": 1458
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1053333921613302,
      "learning_rate": 9.288976312454557e-06,
      "loss": 0.2137,
      "step": 1459
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0685843013684573,
      "learning_rate": 9.287853456592625e-06,
      "loss": 0.2382,
      "step": 1460
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0467676460029007,
      "learning_rate": 9.286729782794261e-06,
      "loss": 0.2281,
      "step": 1461
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2481124092015607,
      "learning_rate": 9.285605291273816e-06,
      "loss": 0.1586,
      "step": 1462
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.067062650866249,
      "learning_rate": 9.28447998224579e-06,
      "loss": 0.2194,
      "step": 1463
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9914626440797804,
      "learning_rate": 9.283353855924848e-06,
      "loss": 0.1972,
      "step": 1464
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.997483680359301,
      "learning_rate": 9.282226912525803e-06,
      "loss": 0.1645,
      "step": 1465
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.202528892051415,
      "learning_rate": 9.281099152263626e-06,
      "loss": 0.2102,
      "step": 1466
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6830996218593023,
      "learning_rate": 9.279970575353447e-06,
      "loss": 0.1127,
      "step": 1467
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5524209458759173,
      "learning_rate": 9.27884118201055e-06,
      "loss": 0.2597,
      "step": 1468
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3585944758009245,
      "learning_rate": 9.277710972450371e-06,
      "loss": 0.2423,
      "step": 1469
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.99674785610715,
      "learning_rate": 9.276579946888507e-06,
      "loss": 0.1696,
      "step": 1470
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1411090903144085,
      "learning_rate": 9.275448105540709e-06,
      "loss": 0.2311,
      "step": 1471
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2568637502454902,
      "learning_rate": 9.274315448622884e-06,
      "loss": 0.2421,
      "step": 1472
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2271374196529397,
      "learning_rate": 9.273181976351091e-06,
      "loss": 0.2324,
      "step": 1473
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.061633215919352,
      "learning_rate": 9.27204768894155e-06,
      "loss": 0.218,
      "step": 1474
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0123112063172717,
      "learning_rate": 9.270912586610632e-06,
      "loss": 0.2069,
      "step": 1475
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0417620347663756,
      "learning_rate": 9.269776669574866e-06,
      "loss": 0.1956,
      "step": 1476
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.228347392323195,
      "learning_rate": 9.268639938050937e-06,
      "loss": 0.2458,
      "step": 1477
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0949542009115703,
      "learning_rate": 9.26750239225568e-06,
      "loss": 0.2406,
      "step": 1478
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9655796821824766,
      "learning_rate": 9.266364032406095e-06,
      "loss": 0.2149,
      "step": 1479
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3889475446098083,
      "learning_rate": 9.265224858719327e-06,
      "loss": 0.2018,
      "step": 1480
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.278893994751683,
      "learning_rate": 9.264084871412684e-06,
      "loss": 0.2401,
      "step": 1481
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2071314287889525,
      "learning_rate": 9.262944070703623e-06,
      "loss": 0.2334,
      "step": 1482
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7924856171946325,
      "learning_rate": 9.261802456809763e-06,
      "loss": 0.1687,
      "step": 1483
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.388515284947204,
      "learning_rate": 9.260660029948872e-06,
      "loss": 0.183,
      "step": 1484
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2957531360175716,
      "learning_rate": 9.259516790338876e-06,
      "loss": 0.2283,
      "step": 1485
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8402747141795965,
      "learning_rate": 9.258372738197856e-06,
      "loss": 0.1579,
      "step": 1486
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0815791474298202,
      "learning_rate": 9.257227873744047e-06,
      "loss": 0.1795,
      "step": 1487
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2174924385904429,
      "learning_rate": 9.25608219719584e-06,
      "loss": 0.1666,
      "step": 1488
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4542752798493963,
      "learning_rate": 9.254935708771779e-06,
      "loss": 0.2946,
      "step": 1489
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3279757214896084,
      "learning_rate": 9.253788408690564e-06,
      "loss": 0.2593,
      "step": 1490
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1599835398436547,
      "learning_rate": 9.252640297171053e-06,
      "loss": 0.2292,
      "step": 1491
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0787593260397648,
      "learning_rate": 9.251491374432256e-06,
      "loss": 0.1921,
      "step": 1492
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0001353484063347,
      "learning_rate": 9.250341640693332e-06,
      "loss": 0.1838,
      "step": 1493
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4474775407621767,
      "learning_rate": 9.249191096173606e-06,
      "loss": 0.213,
      "step": 1494
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8008577839529504,
      "learning_rate": 9.248039741092551e-06,
      "loss": 0.1858,
      "step": 1495
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.412944114565353,
      "learning_rate": 9.246887575669794e-06,
      "loss": 0.3169,
      "step": 1496
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9085813238258992,
      "learning_rate": 9.245734600125118e-06,
      "loss": 0.1648,
      "step": 1497
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.062201542350698,
      "learning_rate": 9.244580814678464e-06,
      "loss": 0.2326,
      "step": 1498
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.036612523344992,
      "learning_rate": 9.243426219549918e-06,
      "loss": 0.187,
      "step": 1499
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2754024714383907,
      "learning_rate": 9.242270814959732e-06,
      "loss": 0.285,
      "step": 1500
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9809321133667331,
      "learning_rate": 9.241114601128307e-06,
      "loss": 0.1185,
      "step": 1501
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9163020199571161,
      "learning_rate": 9.239957578276197e-06,
      "loss": 0.1969,
      "step": 1502
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0117550402273363,
      "learning_rate": 9.238799746624111e-06,
      "loss": 0.1664,
      "step": 1503
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0145394921514055,
      "learning_rate": 9.237641106392913e-06,
      "loss": 0.2028,
      "step": 1504
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.77726701230136,
      "learning_rate": 9.236481657803624e-06,
      "loss": 0.162,
      "step": 1505
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.991207913163313,
      "learning_rate": 9.235321401077412e-06,
      "loss": 0.1992,
      "step": 1506
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1679195524127397,
      "learning_rate": 9.23416033643561e-06,
      "loss": 0.245,
      "step": 1507
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6920723924781512,
      "learning_rate": 9.232998464099691e-06,
      "loss": 0.1618,
      "step": 1508
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2457056435578018,
      "learning_rate": 9.231835784291297e-06,
      "loss": 0.2616,
      "step": 1509
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1843597443559066,
      "learning_rate": 9.230672297232211e-06,
      "loss": 0.2547,
      "step": 1510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.352660037686944,
      "learning_rate": 9.22950800314438e-06,
      "loss": 0.2892,
      "step": 1511
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0153723469288776,
      "learning_rate": 9.228342902249901e-06,
      "loss": 0.1706,
      "step": 1512
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0956617447586077,
      "learning_rate": 9.22717699477102e-06,
      "loss": 0.2138,
      "step": 1513
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2507607641191312,
      "learning_rate": 9.226010280930146e-06,
      "loss": 0.2213,
      "step": 1514
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0453841588862651,
      "learning_rate": 9.224842760949836e-06,
      "loss": 0.1786,
      "step": 1515
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1921653031415262,
      "learning_rate": 9.223674435052803e-06,
      "loss": 0.2396,
      "step": 1516
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1977794038756189,
      "learning_rate": 9.22250530346191e-06,
      "loss": 0.1819,
      "step": 1517
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.032771760646993,
      "learning_rate": 9.221335366400181e-06,
      "loss": 0.2278,
      "step": 1518
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1668377196794233,
      "learning_rate": 9.220164624090786e-06,
      "loss": 0.2145,
      "step": 1519
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0151769164096038,
      "learning_rate": 9.218993076757054e-06,
      "loss": 0.2304,
      "step": 1520
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7297381173272683,
      "learning_rate": 9.217820724622463e-06,
      "loss": 0.1444,
      "step": 1521
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8911880239532375,
      "learning_rate": 9.216647567910649e-06,
      "loss": 0.2031,
      "step": 1522
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4295469270888457,
      "learning_rate": 9.215473606845398e-06,
      "loss": 0.2145,
      "step": 1523
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0863938514123614,
      "learning_rate": 9.214298841650651e-06,
      "loss": 0.1804,
      "step": 1524
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6585070355999181,
      "learning_rate": 9.2131232725505e-06,
      "loss": 0.16,
      "step": 1525
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9349115268058642,
      "learning_rate": 9.2119468997692e-06,
      "loss": 0.2074,
      "step": 1526
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7250598654587629,
      "learning_rate": 9.210769723531146e-06,
      "loss": 0.1224,
      "step": 1527
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.143791457099049,
      "learning_rate": 9.209591744060893e-06,
      "loss": 0.2088,
      "step": 1528
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3786990676010165,
      "learning_rate": 9.208412961583146e-06,
      "loss": 0.2447,
      "step": 1529
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.125852670757636,
      "learning_rate": 9.207233376322772e-06,
      "loss": 0.2234,
      "step": 1530
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0889978739534267,
      "learning_rate": 9.206052988504778e-06,
      "loss": 0.2096,
      "step": 1531
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.107429994247639,
      "learning_rate": 9.204871798354336e-06,
      "loss": 0.149,
      "step": 1532
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9539491794290632,
      "learning_rate": 9.203689806096763e-06,
      "loss": 0.1936,
      "step": 1533
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.058257004324509,
      "learning_rate": 9.202507011957531e-06,
      "loss": 0.1617,
      "step": 1534
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9264789632283602,
      "learning_rate": 9.201323416162268e-06,
      "loss": 0.1806,
      "step": 1535
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1554469784085288,
      "learning_rate": 9.200139018936753e-06,
      "loss": 0.25,
      "step": 1536
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1760356701209589,
      "learning_rate": 9.198953820506915e-06,
      "loss": 0.2292,
      "step": 1537
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2719932284194915,
      "learning_rate": 9.197767821098841e-06,
      "loss": 0.2292,
      "step": 1538
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9769969255195149,
      "learning_rate": 9.196581020938767e-06,
      "loss": 0.2163,
      "step": 1539
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9255479970934233,
      "learning_rate": 9.195393420253082e-06,
      "loss": 0.2294,
      "step": 1540
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0891444047088443,
      "learning_rate": 9.194205019268332e-06,
      "loss": 0.176,
      "step": 1541
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7313597526813154,
      "learning_rate": 9.193015818211209e-06,
      "loss": 0.1676,
      "step": 1542
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2646857260496218,
      "learning_rate": 9.191825817308561e-06,
      "loss": 0.2066,
      "step": 1543
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9920022070468745,
      "learning_rate": 9.190635016787392e-06,
      "loss": 0.168,
      "step": 1544
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9954050825764155,
      "learning_rate": 9.189443416874852e-06,
      "loss": 0.1934,
      "step": 1545
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9568026384042294,
      "learning_rate": 9.188251017798248e-06,
      "loss": 0.2173,
      "step": 1546
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0155756099014628,
      "learning_rate": 9.187057819785039e-06,
      "loss": 0.2257,
      "step": 1547
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8900524993057212,
      "learning_rate": 9.18586382306283e-06,
      "loss": 0.1841,
      "step": 1548
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0139633435572508,
      "learning_rate": 9.18466902785939e-06,
      "loss": 0.1489,
      "step": 1549
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0916149635803472,
      "learning_rate": 9.183473434402631e-06,
      "loss": 0.2103,
      "step": 1550
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9056649715429842,
      "learning_rate": 9.182277042920622e-06,
      "loss": 0.1825,
      "step": 1551
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2523603072498686,
      "learning_rate": 9.181079853641581e-06,
      "loss": 0.3,
      "step": 1552
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3069531770117335,
      "learning_rate": 9.17988186679388e-06,
      "loss": 0.1707,
      "step": 1553
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8887887824826743,
      "learning_rate": 9.178683082606045e-06,
      "loss": 0.1489,
      "step": 1554
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.034008377655116,
      "learning_rate": 9.17748350130675e-06,
      "loss": 0.1826,
      "step": 1555
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0665972223031897,
      "learning_rate": 9.176283123124823e-06,
      "loss": 0.2066,
      "step": 1556
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1053041083521946,
      "learning_rate": 9.175081948289244e-06,
      "loss": 0.1984,
      "step": 1557
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8047395507716162,
      "learning_rate": 9.173879977029146e-06,
      "loss": 0.1762,
      "step": 1558
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9204690988605108,
      "learning_rate": 9.172677209573813e-06,
      "loss": 0.1483,
      "step": 1559
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0725299598409954,
      "learning_rate": 9.17147364615268e-06,
      "loss": 0.1883,
      "step": 1560
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3420305941357362,
      "learning_rate": 9.170269286995336e-06,
      "loss": 0.2486,
      "step": 1561
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.92476836599532,
      "learning_rate": 9.16906413233152e-06,
      "loss": 0.1506,
      "step": 1562
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.342283669120384,
      "learning_rate": 9.167858182391121e-06,
      "loss": 0.2742,
      "step": 1563
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1506160989225764,
      "learning_rate": 9.166651437404184e-06,
      "loss": 0.2048,
      "step": 1564
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8678215340494866,
      "learning_rate": 9.165443897600905e-06,
      "loss": 0.1566,
      "step": 1565
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.34815728049161,
      "learning_rate": 9.164235563211628e-06,
      "loss": 0.2302,
      "step": 1566
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9018699570832143,
      "learning_rate": 9.16302643446685e-06,
      "loss": 0.165,
      "step": 1567
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0892486909749626,
      "learning_rate": 9.161816511597224e-06,
      "loss": 0.1952,
      "step": 1568
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9282420622543063,
      "learning_rate": 9.160605794833548e-06,
      "loss": 0.155,
      "step": 1569
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3208469642775909,
      "learning_rate": 9.159394284406775e-06,
      "loss": 0.2494,
      "step": 1570
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8421285165090833,
      "learning_rate": 9.158181980548006e-06,
      "loss": 0.1472,
      "step": 1571
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9547993556009755,
      "learning_rate": 9.1569688834885e-06,
      "loss": 0.1712,
      "step": 1572
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9227599170124827,
      "learning_rate": 9.155754993459662e-06,
      "loss": 0.1771,
      "step": 1573
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.889515728208581,
      "learning_rate": 9.154540310693048e-06,
      "loss": 0.1513,
      "step": 1574
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2994128229710484,
      "learning_rate": 9.15332483542037e-06,
      "loss": 0.2505,
      "step": 1575
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3941551430689056,
      "learning_rate": 9.152108567873485e-06,
      "loss": 0.2854,
      "step": 1576
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.063348272885827,
      "learning_rate": 9.150891508284405e-06,
      "loss": 0.2015,
      "step": 1577
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6520257605960184,
      "learning_rate": 9.149673656885292e-06,
      "loss": 0.1484,
      "step": 1578
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3077789835928049,
      "learning_rate": 9.14845501390846e-06,
      "loss": 0.2415,
      "step": 1579
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9483367858698818,
      "learning_rate": 9.14723557958637e-06,
      "loss": 0.2022,
      "step": 1580
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1234036203973645,
      "learning_rate": 9.146015354151642e-06,
      "loss": 0.2112,
      "step": 1581
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.023458632407832,
      "learning_rate": 9.144794337837039e-06,
      "loss": 0.1793,
      "step": 1582
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2281308611547879,
      "learning_rate": 9.14357253087548e-06,
      "loss": 0.2044,
      "step": 1583
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9172663928973426,
      "learning_rate": 9.142349933500032e-06,
      "loss": 0.1448,
      "step": 1584
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1135701143010492,
      "learning_rate": 9.141126545943913e-06,
      "loss": 0.2311,
      "step": 1585
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9265857512066183,
      "learning_rate": 9.139902368440491e-06,
      "loss": 0.2025,
      "step": 1586
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7613838750039221,
      "learning_rate": 9.138677401223288e-06,
      "loss": 0.1405,
      "step": 1587
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.017339311414063,
      "learning_rate": 9.137451644525975e-06,
      "loss": 0.1818,
      "step": 1588
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9707496341302665,
      "learning_rate": 9.136225098582369e-06,
      "loss": 0.1869,
      "step": 1589
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.283264927566998,
      "learning_rate": 9.134997763626448e-06,
      "loss": 0.1668,
      "step": 1590
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2094704855322616,
      "learning_rate": 9.133769639892332e-06,
      "loss": 0.2447,
      "step": 1591
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7111338152662267,
      "learning_rate": 9.13254072761429e-06,
      "loss": 0.1876,
      "step": 1592
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9994489411968169,
      "learning_rate": 9.13131102702675e-06,
      "loss": 0.2242,
      "step": 1593
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1818575782640182,
      "learning_rate": 9.130080538364283e-06,
      "loss": 0.2312,
      "step": 1594
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.00253945587966,
      "learning_rate": 9.128849261861615e-06,
      "loss": 0.1544,
      "step": 1595
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.091956056308726,
      "learning_rate": 9.127617197753615e-06,
      "loss": 0.2115,
      "step": 1596
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2567975015434925,
      "learning_rate": 9.126384346275315e-06,
      "loss": 0.2477,
      "step": 1597
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9268823948931293,
      "learning_rate": 9.125150707661882e-06,
      "loss": 0.2155,
      "step": 1598
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1833535557873154,
      "learning_rate": 9.123916282148648e-06,
      "loss": 0.2198,
      "step": 1599
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9306418012061235,
      "learning_rate": 9.12268106997108e-06,
      "loss": 0.2227,
      "step": 1600
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9105445777926785,
      "learning_rate": 9.121445071364811e-06,
      "loss": 0.1902,
      "step": 1601
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0554519078304228,
      "learning_rate": 9.12020828656561e-06,
      "loss": 0.2145,
      "step": 1602
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9444092966975481,
      "learning_rate": 9.118970715809404e-06,
      "loss": 0.1907,
      "step": 1603
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.037022618330147,
      "learning_rate": 9.117732359332267e-06,
      "loss": 0.1522,
      "step": 1604
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0624352071387126,
      "learning_rate": 9.116493217370425e-06,
      "loss": 0.1856,
      "step": 1605
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2713046608890548,
      "learning_rate": 9.115253290160252e-06,
      "loss": 0.2798,
      "step": 1606
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8734310376925507,
      "learning_rate": 9.114012577938273e-06,
      "loss": 0.1262,
      "step": 1607
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8722204972213148,
      "learning_rate": 9.112771080941162e-06,
      "loss": 0.1518,
      "step": 1608
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9130812959259934,
      "learning_rate": 9.11152879940574e-06,
      "loss": 0.1531,
      "step": 1609
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3544448381382297,
      "learning_rate": 9.110285733568983e-06,
      "loss": 0.2537,
      "step": 1610
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7928535412133841,
      "learning_rate": 9.109041883668014e-06,
      "loss": 0.1206,
      "step": 1611
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4162297160012143,
      "learning_rate": 9.107797249940105e-06,
      "loss": 0.2209,
      "step": 1612
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9136038302597628,
      "learning_rate": 9.106551832622678e-06,
      "loss": 0.1944,
      "step": 1613
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1665382479296638,
      "learning_rate": 9.105305631953306e-06,
      "loss": 0.2244,
      "step": 1614
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8983300212275168,
      "learning_rate": 9.104058648169706e-06,
      "loss": 0.1875,
      "step": 1615
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.912476907913707,
      "learning_rate": 9.102810881509754e-06,
      "loss": 0.2207,
      "step": 1616
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0805817817700083,
      "learning_rate": 9.101562332211467e-06,
      "loss": 0.2084,
      "step": 1617
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.343770812533494,
      "learning_rate": 9.100313000513013e-06,
      "loss": 0.2621,
      "step": 1618
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9379267213928356,
      "learning_rate": 9.099062886652712e-06,
      "loss": 0.1757,
      "step": 1619
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.837849028982546,
      "learning_rate": 9.09781199086903e-06,
      "loss": 0.1791,
      "step": 1620
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0883387716541542,
      "learning_rate": 9.096560313400584e-06,
      "loss": 0.2308,
      "step": 1621
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9374119003452883,
      "learning_rate": 9.095307854486141e-06,
      "loss": 0.1362,
      "step": 1622
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.016157972237035,
      "learning_rate": 9.094054614364617e-06,
      "loss": 0.1811,
      "step": 1623
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9278455014416334,
      "learning_rate": 9.092800593275073e-06,
      "loss": 0.1855,
      "step": 1624
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9137227419966132,
      "learning_rate": 9.091545791456723e-06,
      "loss": 0.2177,
      "step": 1625
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.036739168446271,
      "learning_rate": 9.090290209148926e-06,
      "loss": 0.2102,
      "step": 1626
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9805996587175271,
      "learning_rate": 9.089033846591198e-06,
      "loss": 0.195,
      "step": 1627
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2239230577003823,
      "learning_rate": 9.087776704023193e-06,
      "loss": 0.2206,
      "step": 1628
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8533110896917859,
      "learning_rate": 9.086518781684724e-06,
      "loss": 0.1769,
      "step": 1629
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2907361790185108,
      "learning_rate": 9.085260079815745e-06,
      "loss": 0.2685,
      "step": 1630
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5732095693371093,
      "learning_rate": 9.084000598656365e-06,
      "loss": 0.134,
      "step": 1631
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9004932746526683,
      "learning_rate": 9.082740338446833e-06,
      "loss": 0.1619,
      "step": 1632
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.104062670699642,
      "learning_rate": 9.081479299427557e-06,
      "loss": 0.2272,
      "step": 1633
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9303202908122238,
      "learning_rate": 9.080217481839084e-06,
      "loss": 0.1628,
      "step": 1634
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.073285497233567,
      "learning_rate": 9.07895488592212e-06,
      "loss": 0.2213,
      "step": 1635
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9312347480275127,
      "learning_rate": 9.077691511917508e-06,
      "loss": 0.2135,
      "step": 1636
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0550920605294838,
      "learning_rate": 9.076427360066247e-06,
      "loss": 0.1892,
      "step": 1637
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.348289882775555,
      "learning_rate": 9.075162430609484e-06,
      "loss": 0.252,
      "step": 1638
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3306911448458103,
      "learning_rate": 9.073896723788509e-06,
      "loss": 0.2476,
      "step": 1639
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8946768124950375,
      "learning_rate": 9.072630239844766e-06,
      "loss": 0.1922,
      "step": 1640
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7606023224135805,
      "learning_rate": 9.071362979019846e-06,
      "loss": 0.1951,
      "step": 1641
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9060080365525509,
      "learning_rate": 9.070094941555487e-06,
      "loss": 0.2019,
      "step": 1642
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.131297500733499,
      "learning_rate": 9.068826127693576e-06,
      "loss": 0.2187,
      "step": 1643
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0135871331928492,
      "learning_rate": 9.067556537676146e-06,
      "loss": 0.1977,
      "step": 1644
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1364504517439749,
      "learning_rate": 9.06628617174538e-06,
      "loss": 0.2628,
      "step": 1645
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9031845283433102,
      "learning_rate": 9.065015030143609e-06,
      "loss": 0.157,
      "step": 1646
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0262464734761978,
      "learning_rate": 9.063743113113312e-06,
      "loss": 0.1937,
      "step": 1647
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2385567951688508,
      "learning_rate": 9.062470420897115e-06,
      "loss": 0.2217,
      "step": 1648
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8894587791133882,
      "learning_rate": 9.061196953737794e-06,
      "loss": 0.1688,
      "step": 1649
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9210339349152251,
      "learning_rate": 9.059922711878271e-06,
      "loss": 0.1795,
      "step": 1650
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0317153572503333,
      "learning_rate": 9.058647695561615e-06,
      "loss": 0.2276,
      "step": 1651
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0999947326355357,
      "learning_rate": 9.057371905031043e-06,
      "loss": 0.2318,
      "step": 1652
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1128058522363629,
      "learning_rate": 9.05609534052992e-06,
      "loss": 0.241,
      "step": 1653
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3208764781744762,
      "learning_rate": 9.054818002301764e-06,
      "loss": 0.2457,
      "step": 1654
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7762426434722949,
      "learning_rate": 9.053539890590231e-06,
      "loss": 0.1546,
      "step": 1655
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9800432751187034,
      "learning_rate": 9.052261005639132e-06,
      "loss": 0.1798,
      "step": 1656
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.041569796792994,
      "learning_rate": 9.05098134769242e-06,
      "loss": 0.2048,
      "step": 1657
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8603717385001127,
      "learning_rate": 9.0497009169942e-06,
      "loss": 0.2004,
      "step": 1658
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.27703549308256,
      "learning_rate": 9.048419713788724e-06,
      "loss": 0.269,
      "step": 1659
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1514517915833544,
      "learning_rate": 9.047137738320385e-06,
      "loss": 0.2553,
      "step": 1660
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4410717355223455,
      "learning_rate": 9.045854990833734e-06,
      "loss": 0.3021,
      "step": 1661
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9592655763708217,
      "learning_rate": 9.04457147157346e-06,
      "loss": 0.1424,
      "step": 1662
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3038397436504692,
      "learning_rate": 9.043287180784405e-06,
      "loss": 0.2445,
      "step": 1663
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5619736230225016,
      "learning_rate": 9.042002118711555e-06,
      "loss": 0.3153,
      "step": 1664
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1656889790933496,
      "learning_rate": 9.040716285600043e-06,
      "loss": 0.1967,
      "step": 1665
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9898476906497795,
      "learning_rate": 9.039429681695153e-06,
      "loss": 0.2176,
      "step": 1666
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6083598363933727,
      "learning_rate": 9.03814230724231e-06,
      "loss": 0.3286,
      "step": 1667
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8116882313098569,
      "learning_rate": 9.036854162487092e-06,
      "loss": 0.1889,
      "step": 1668
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3090562569557114,
      "learning_rate": 9.035565247675218e-06,
      "loss": 0.2685,
      "step": 1669
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9131405218324317,
      "learning_rate": 9.03427556305256e-06,
      "loss": 0.1978,
      "step": 1670
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.403551693409035,
      "learning_rate": 9.032985108865134e-06,
      "loss": 0.2282,
      "step": 1671
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1387621673303578,
      "learning_rate": 9.031693885359099e-06,
      "loss": 0.1951,
      "step": 1672
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0577291974501468,
      "learning_rate": 9.030401892780767e-06,
      "loss": 0.1989,
      "step": 1673
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1810830996910942,
      "learning_rate": 9.029109131376595e-06,
      "loss": 0.2446,
      "step": 1674
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0443640551966644,
      "learning_rate": 9.027815601393184e-06,
      "loss": 0.2045,
      "step": 1675
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1298762245493001,
      "learning_rate": 9.026521303077284e-06,
      "loss": 0.2076,
      "step": 1676
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0080256815383242,
      "learning_rate": 9.025226236675792e-06,
      "loss": 0.2124,
      "step": 1677
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.271611665571337,
      "learning_rate": 9.02393040243575e-06,
      "loss": 0.2044,
      "step": 1678
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0421769777744485,
      "learning_rate": 9.022633800604345e-06,
      "loss": 0.1791,
      "step": 1679
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0667323177065713,
      "learning_rate": 9.021336431428916e-06,
      "loss": 0.2073,
      "step": 1680
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2434461170594993,
      "learning_rate": 9.020038295156942e-06,
      "loss": 0.2143,
      "step": 1681
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.091785917136842,
      "learning_rate": 9.018739392036051e-06,
      "loss": 0.1987,
      "step": 1682
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8788647673599841,
      "learning_rate": 9.017439722314018e-06,
      "loss": 0.1383,
      "step": 1683
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9130165643816887,
      "learning_rate": 9.016139286238765e-06,
      "loss": 0.1728,
      "step": 1684
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6299251636380091,
      "learning_rate": 9.014838084058358e-06,
      "loss": 0.1121,
      "step": 1685
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2400783395006718,
      "learning_rate": 9.01353611602101e-06,
      "loss": 0.2913,
      "step": 1686
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0287409614362961,
      "learning_rate": 9.012233382375076e-06,
      "loss": 0.1995,
      "step": 1687
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0932939447740333,
      "learning_rate": 9.010929883369068e-06,
      "loss": 0.2069,
      "step": 1688
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9295852283749378,
      "learning_rate": 9.009625619251631e-06,
      "loss": 0.169,
      "step": 1689
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9869629832871204,
      "learning_rate": 9.008320590271567e-06,
      "loss": 0.1707,
      "step": 1690
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0136131254602732,
      "learning_rate": 9.007014796677815e-06,
      "loss": 0.2145,
      "step": 1691
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7226709202851849,
      "learning_rate": 9.005708238719466e-06,
      "loss": 0.1345,
      "step": 1692
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8467489927296917,
      "learning_rate": 9.004400916645755e-06,
      "loss": 0.1471,
      "step": 1693
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8572863126166351,
      "learning_rate": 9.00309283070606e-06,
      "loss": 0.1557,
      "step": 1694
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8752086441157216,
      "learning_rate": 9.001783981149907e-06,
      "loss": 0.1735,
      "step": 1695
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0289172055014502,
      "learning_rate": 9.000474368226972e-06,
      "loss": 0.2262,
      "step": 1696
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9593185115054929,
      "learning_rate": 8.999163992187068e-06,
      "loss": 0.175,
      "step": 1697
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.873557036347934,
      "learning_rate": 8.997852853280159e-06,
      "loss": 0.1979,
      "step": 1698
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0493897988597436,
      "learning_rate": 8.996540951756354e-06,
      "loss": 0.1742,
      "step": 1699
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.365144570399527,
      "learning_rate": 8.995228287865907e-06,
      "loss": 0.2572,
      "step": 1700
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0580859624821533,
      "learning_rate": 8.993914861859216e-06,
      "loss": 0.2438,
      "step": 1701
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0691545076268338,
      "learning_rate": 8.992600673986828e-06,
      "loss": 0.235,
      "step": 1702
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8654326544616091,
      "learning_rate": 8.99128572449943e-06,
      "loss": 0.1358,
      "step": 1703
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.096105673009714,
      "learning_rate": 8.98997001364786e-06,
      "loss": 0.2291,
      "step": 1704
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8704918258579404,
      "learning_rate": 8.988653541683098e-06,
      "loss": 0.1918,
      "step": 1705
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1349566503199058,
      "learning_rate": 8.98733630885627e-06,
      "loss": 0.2093,
      "step": 1706
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0959350080402757,
      "learning_rate": 8.986018315418643e-06,
      "loss": 0.1891,
      "step": 1707
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8020375342466055,
      "learning_rate": 8.98469956162164e-06,
      "loss": 0.1718,
      "step": 1708
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.993906909341975,
      "learning_rate": 8.983380047716815e-06,
      "loss": 0.2127,
      "step": 1709
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.975853827045483,
      "learning_rate": 8.982059773955879e-06,
      "loss": 0.1926,
      "step": 1710
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9444088510734382,
      "learning_rate": 8.98073874059068e-06,
      "loss": 0.1891,
      "step": 1711
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2415364240238003,
      "learning_rate": 8.979416947873217e-06,
      "loss": 0.2174,
      "step": 1712
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8793771998850708,
      "learning_rate": 8.978094396055626e-06,
      "loss": 0.15,
      "step": 1713
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1458465420405384,
      "learning_rate": 8.976771085390197e-06,
      "loss": 0.1505,
      "step": 1714
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7257605689769333,
      "learning_rate": 8.975447016129358e-06,
      "loss": 0.1331,
      "step": 1715
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.90159187306588,
      "learning_rate": 8.974122188525685e-06,
      "loss": 0.1686,
      "step": 1716
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.872478248719135,
      "learning_rate": 8.972796602831897e-06,
      "loss": 0.226,
      "step": 1717
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2480180312914106,
      "learning_rate": 8.971470259300858e-06,
      "loss": 0.2406,
      "step": 1718
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.211797070543537,
      "learning_rate": 8.970143158185576e-06,
      "loss": 0.202,
      "step": 1719
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1159376023177094,
      "learning_rate": 8.968815299739206e-06,
      "loss": 0.2004,
      "step": 1720
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.124144008312983,
      "learning_rate": 8.967486684215048e-06,
      "loss": 0.2127,
      "step": 1721
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7163225401869178,
      "learning_rate": 8.96615731186654e-06,
      "loss": 0.1402,
      "step": 1722
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.158452181459997,
      "learning_rate": 8.964827182947268e-06,
      "loss": 0.2265,
      "step": 1723
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8879215377853503,
      "learning_rate": 8.963496297710967e-06,
      "loss": 0.1911,
      "step": 1724
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1549441820873323,
      "learning_rate": 8.96216465641151e-06,
      "loss": 0.1942,
      "step": 1725
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2045666010152052,
      "learning_rate": 8.960832259302914e-06,
      "loss": 0.1988,
      "step": 1726
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0150747378201295,
      "learning_rate": 8.95949910663935e-06,
      "loss": 0.1712,
      "step": 1727
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.054759872877421,
      "learning_rate": 8.958165198675116e-06,
      "loss": 0.221,
      "step": 1728
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2959290473577743,
      "learning_rate": 8.95683053566467e-06,
      "loss": 0.2347,
      "step": 1729
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3781736247461709,
      "learning_rate": 8.955495117862607e-06,
      "loss": 0.272,
      "step": 1730
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9639719271945713,
      "learning_rate": 8.954158945523665e-06,
      "loss": 0.147,
      "step": 1731
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8991365230704169,
      "learning_rate": 8.952822018902731e-06,
      "loss": 0.19,
      "step": 1732
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2374752413004233,
      "learning_rate": 8.95148433825483e-06,
      "loss": 0.183,
      "step": 1733
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1701562426685816,
      "learning_rate": 8.950145903835132e-06,
      "loss": 0.2271,
      "step": 1734
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7864694602478843,
      "learning_rate": 8.948806715898957e-06,
      "loss": 0.2167,
      "step": 1735
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2159161236131628,
      "learning_rate": 8.947466774701759e-06,
      "loss": 0.1951,
      "step": 1736
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.126971441932858,
      "learning_rate": 8.946126080499142e-06,
      "loss": 0.2351,
      "step": 1737
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8726634675307856,
      "learning_rate": 8.944784633546855e-06,
      "loss": 0.1347,
      "step": 1738
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8113588119567229,
      "learning_rate": 8.943442434100784e-06,
      "loss": 0.143,
      "step": 1739
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2422220630915788,
      "learning_rate": 8.942099482416965e-06,
      "loss": 0.2742,
      "step": 1740
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9803385270471321,
      "learning_rate": 8.940755778751576e-06,
      "loss": 0.1924,
      "step": 1741
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.244847463192927,
      "learning_rate": 8.939411323360934e-06,
      "loss": 0.2528,
      "step": 1742
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9238350965221264,
      "learning_rate": 8.938066116501505e-06,
      "loss": 0.2186,
      "step": 1743
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2672046399069874,
      "learning_rate": 8.936720158429895e-06,
      "loss": 0.2644,
      "step": 1744
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8324730357913355,
      "learning_rate": 8.935373449402856e-06,
      "loss": 0.1782,
      "step": 1745
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9551310847202754,
      "learning_rate": 8.93402598967728e-06,
      "loss": 0.2005,
      "step": 1746
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5950512395785004,
      "learning_rate": 8.932677779510204e-06,
      "loss": 0.1695,
      "step": 1747
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9744679459291212,
      "learning_rate": 8.93132881915881e-06,
      "loss": 0.1859,
      "step": 1748
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.290194305188234,
      "learning_rate": 8.929979108880417e-06,
      "loss": 0.2465,
      "step": 1749
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0488683721714607,
      "learning_rate": 8.928628648932496e-06,
      "loss": 0.2157,
      "step": 1750
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0595587497454866,
      "learning_rate": 8.927277439572653e-06,
      "loss": 0.1868,
      "step": 1751
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9239610722600683,
      "learning_rate": 8.925925481058641e-06,
      "loss": 0.1469,
      "step": 1752
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.005423841454979,
      "learning_rate": 8.924572773648355e-06,
      "loss": 0.2021,
      "step": 1753
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8131904597505547,
      "learning_rate": 8.923219317599833e-06,
      "loss": 0.1518,
      "step": 1754
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9714592147511968,
      "learning_rate": 8.921865113171257e-06,
      "loss": 0.1648,
      "step": 1755
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.94210685658369,
      "learning_rate": 8.920510160620948e-06,
      "loss": 0.1628,
      "step": 1756
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2378089881405154,
      "learning_rate": 8.919154460207373e-06,
      "loss": 0.2416,
      "step": 1757
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0193037789996382,
      "learning_rate": 8.917798012189142e-06,
      "loss": 0.1384,
      "step": 1758
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1112441824183277,
      "learning_rate": 8.916440816825006e-06,
      "loss": 0.1885,
      "step": 1759
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.245622822552052,
      "learning_rate": 8.915082874373856e-06,
      "loss": 0.2422,
      "step": 1760
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.429886089399871,
      "learning_rate": 8.913724185094734e-06,
      "loss": 0.2795,
      "step": 1761
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1396893973097424,
      "learning_rate": 8.912364749246813e-06,
      "loss": 0.1998,
      "step": 1762
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8863257942787386,
      "learning_rate": 8.911004567089421e-06,
      "loss": 0.2137,
      "step": 1763
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0796681603424005,
      "learning_rate": 8.909643638882017e-06,
      "loss": 0.2222,
      "step": 1764
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.056746154459645,
      "learning_rate": 8.908281964884206e-06,
      "loss": 0.1892,
      "step": 1765
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1864278409701805,
      "learning_rate": 8.90691954535574e-06,
      "loss": 0.2477,
      "step": 1766
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.124040906727328,
      "learning_rate": 8.90555638055651e-06,
      "loss": 0.2342,
      "step": 1767
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9498599383521321,
      "learning_rate": 8.904192470746546e-06,
      "loss": 0.1787,
      "step": 1768
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9600969735677135,
      "learning_rate": 8.902827816186024e-06,
      "loss": 0.2171,
      "step": 1769
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3073673183844057,
      "learning_rate": 8.90146241713526e-06,
      "loss": 0.2115,
      "step": 1770
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0959309449109231,
      "learning_rate": 8.900096273854712e-06,
      "loss": 0.2237,
      "step": 1771
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.235952499760344,
      "learning_rate": 8.898729386604985e-06,
      "loss": 0.2607,
      "step": 1772
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1339341934683678,
      "learning_rate": 8.897361755646816e-06,
      "loss": 0.2234,
      "step": 1773
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8440200841492308,
      "learning_rate": 8.895993381241095e-06,
      "loss": 0.1547,
      "step": 1774
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9098059200146356,
      "learning_rate": 8.894624263648846e-06,
      "loss": 0.1691,
      "step": 1775
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9078250414345342,
      "learning_rate": 8.893254403131238e-06,
      "loss": 0.1883,
      "step": 1776
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9731823607225754,
      "learning_rate": 8.89188379994958e-06,
      "loss": 0.2027,
      "step": 1777
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2211337818984407,
      "learning_rate": 8.890512454365324e-06,
      "loss": 0.2202,
      "step": 1778
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.903602475836941,
      "learning_rate": 8.889140366640063e-06,
      "loss": 0.1903,
      "step": 1779
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9515947191488106,
      "learning_rate": 8.887767537035532e-06,
      "loss": 0.2065,
      "step": 1780
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8897690626362017,
      "learning_rate": 8.886393965813608e-06,
      "loss": 0.1972,
      "step": 1781
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.866287751698666,
      "learning_rate": 8.88501965323631e-06,
      "loss": 0.1717,
      "step": 1782
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9688229332849452,
      "learning_rate": 8.883644599565793e-06,
      "loss": 0.224,
      "step": 1783
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7637537868613037,
      "learning_rate": 8.882268805064363e-06,
      "loss": 0.1401,
      "step": 1784
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0480354178462314,
      "learning_rate": 8.880892269994455e-06,
      "loss": 0.1747,
      "step": 1785
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2409286853010952,
      "learning_rate": 8.879514994618659e-06,
      "loss": 0.2266,
      "step": 1786
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7918691557617817,
      "learning_rate": 8.878136979199697e-06,
      "loss": 0.1988,
      "step": 1787
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.300292322793181,
      "learning_rate": 8.876758224000432e-06,
      "loss": 0.2379,
      "step": 1788
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1551226806324038,
      "learning_rate": 8.875378729283875e-06,
      "loss": 0.2377,
      "step": 1789
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0675824649413215,
      "learning_rate": 8.87399849531317e-06,
      "loss": 0.1841,
      "step": 1790
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9631439194818042,
      "learning_rate": 8.872617522351607e-06,
      "loss": 0.168,
      "step": 1791
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2401773522494002,
      "learning_rate": 8.871235810662618e-06,
      "loss": 0.2066,
      "step": 1792
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1223680402064242,
      "learning_rate": 8.869853360509771e-06,
      "loss": 0.2366,
      "step": 1793
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8400429973478517,
      "learning_rate": 8.868470172156778e-06,
      "loss": 0.1458,
      "step": 1794
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9675288250369417,
      "learning_rate": 8.867086245867492e-06,
      "loss": 0.1687,
      "step": 1795
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.842657201019928,
      "learning_rate": 8.865701581905905e-06,
      "loss": 0.1188,
      "step": 1796
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4207706898255597,
      "learning_rate": 8.864316180536151e-06,
      "loss": 0.2974,
      "step": 1797
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9302073220861108,
      "learning_rate": 8.862930042022509e-06,
      "loss": 0.1388,
      "step": 1798
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9828915779217465,
      "learning_rate": 8.861543166629384e-06,
      "loss": 0.1649,
      "step": 1799
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9953106489722987,
      "learning_rate": 8.860155554621343e-06,
      "loss": 0.181,
      "step": 1800
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1739925784673515,
      "learning_rate": 8.858767206263075e-06,
      "loss": 0.195,
      "step": 1801
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7442932122132846,
      "learning_rate": 8.857378121819417e-06,
      "loss": 0.13,
      "step": 1802
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9310780129335169,
      "learning_rate": 8.85598830155535e-06,
      "loss": 0.1507,
      "step": 1803
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9818395174880251,
      "learning_rate": 8.854597745735989e-06,
      "loss": 0.2031,
      "step": 1804
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1219888163785094,
      "learning_rate": 8.853206454626591e-06,
      "loss": 0.2086,
      "step": 1805
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.308051696459148,
      "learning_rate": 8.851814428492554e-06,
      "loss": 0.2782,
      "step": 1806
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8612334726654575,
      "learning_rate": 8.850421667599419e-06,
      "loss": 0.1191,
      "step": 1807
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1941704410438456,
      "learning_rate": 8.84902817221286e-06,
      "loss": 0.1757,
      "step": 1808
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.008020334311649,
      "learning_rate": 8.8476339425987e-06,
      "loss": 0.1856,
      "step": 1809
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9837201377864982,
      "learning_rate": 8.846238979022894e-06,
      "loss": 0.1837,
      "step": 1810
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0575807730584905,
      "learning_rate": 8.84484328175154e-06,
      "loss": 0.1945,
      "step": 1811
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9819032925120229,
      "learning_rate": 8.843446851050882e-06,
      "loss": 0.2166,
      "step": 1812
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9520000938741624,
      "learning_rate": 8.842049687187293e-06,
      "loss": 0.2344,
      "step": 1813
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0291645268988383,
      "learning_rate": 8.840651790427293e-06,
      "loss": 0.1653,
      "step": 1814
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0234696864912818,
      "learning_rate": 8.83925316103754e-06,
      "loss": 0.1889,
      "step": 1815
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9856259821571862,
      "learning_rate": 8.837853799284833e-06,
      "loss": 0.1474,
      "step": 1816
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8990512697823295,
      "learning_rate": 8.836453705436107e-06,
      "loss": 0.2191,
      "step": 1817
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.78236210230986,
      "learning_rate": 8.835052879758441e-06,
      "loss": 0.1246,
      "step": 1818
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.887687347271875,
      "learning_rate": 8.83365132251905e-06,
      "loss": 0.1676,
      "step": 1819
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1715786342969685,
      "learning_rate": 8.832249033985293e-06,
      "loss": 0.1736,
      "step": 1820
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0249348035090398,
      "learning_rate": 8.830846014424666e-06,
      "loss": 0.2133,
      "step": 1821
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0762346183682316,
      "learning_rate": 8.829442264104802e-06,
      "loss": 0.2289,
      "step": 1822
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0823730659476294,
      "learning_rate": 8.828037783293474e-06,
      "loss": 0.2118,
      "step": 1823
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1533912980562724,
      "learning_rate": 8.826632572258602e-06,
      "loss": 0.2533,
      "step": 1824
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9961405516995437,
      "learning_rate": 8.825226631268234e-06,
      "loss": 0.1683,
      "step": 1825
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.782749525657312,
      "learning_rate": 8.823819960590564e-06,
      "loss": 0.1664,
      "step": 1826
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7776663888697151,
      "learning_rate": 8.822412560493925e-06,
      "loss": 0.1551,
      "step": 1827
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.372468759344486,
      "learning_rate": 8.821004431246786e-06,
      "loss": 0.2106,
      "step": 1828
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0753389145784056,
      "learning_rate": 8.819595573117758e-06,
      "loss": 0.178,
      "step": 1829
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4009344964318713,
      "learning_rate": 8.81818598637559e-06,
      "loss": 0.2484,
      "step": 1830
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.132554327123729,
      "learning_rate": 8.81677567128917e-06,
      "loss": 0.2014,
      "step": 1831
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2080122424617399,
      "learning_rate": 8.815364628127525e-06,
      "loss": 0.2777,
      "step": 1832
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0545562913434603,
      "learning_rate": 8.81395285715982e-06,
      "loss": 0.182,
      "step": 1833
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2514752288493536,
      "learning_rate": 8.812540358655361e-06,
      "loss": 0.2494,
      "step": 1834
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2469238162075873,
      "learning_rate": 8.811127132883591e-06,
      "loss": 0.2306,
      "step": 1835
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1433497458377526,
      "learning_rate": 8.809713180114092e-06,
      "loss": 0.2331,
      "step": 1836
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.219632968426858,
      "learning_rate": 8.808298500616583e-06,
      "loss": 0.2503,
      "step": 1837
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9384817878844491,
      "learning_rate": 8.806883094660928e-06,
      "loss": 0.1822,
      "step": 1838
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7207512832691771,
      "learning_rate": 8.80546696251712e-06,
      "loss": 0.1556,
      "step": 1839
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1440072117917917,
      "learning_rate": 8.8040501044553e-06,
      "loss": 0.2039,
      "step": 1840
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.137254648695283,
      "learning_rate": 8.80263252074574e-06,
      "loss": 0.2115,
      "step": 1841
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1085309625018394,
      "learning_rate": 8.801214211658852e-06,
      "loss": 0.2261,
      "step": 1842
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7731984425067134,
      "learning_rate": 8.799795177465195e-06,
      "loss": 0.1477,
      "step": 1843
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.076346605049355,
      "learning_rate": 8.798375418435452e-06,
      "loss": 0.2098,
      "step": 1844
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8995769106676461,
      "learning_rate": 8.796954934840453e-06,
      "loss": 0.2196,
      "step": 1845
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.032966874227555,
      "learning_rate": 8.795533726951166e-06,
      "loss": 0.2127,
      "step": 1846
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9855966631030988,
      "learning_rate": 8.794111795038697e-06,
      "loss": 0.1655,
      "step": 1847
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8607357735504841,
      "learning_rate": 8.792689139374287e-06,
      "loss": 0.1531,
      "step": 1848
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0798402818449206,
      "learning_rate": 8.791265760229315e-06,
      "loss": 0.2444,
      "step": 1849
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0956873045187245,
      "learning_rate": 8.789841657875305e-06,
      "loss": 0.2498,
      "step": 1850
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0851445544848286,
      "learning_rate": 8.78841683258391e-06,
      "loss": 0.1862,
      "step": 1851
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7674252446038955,
      "learning_rate": 8.786991284626924e-06,
      "loss": 0.1496,
      "step": 1852
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7710955445694133,
      "learning_rate": 8.785565014276285e-06,
      "loss": 0.145,
      "step": 1853
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9364335854804174,
      "learning_rate": 8.784138021804058e-06,
      "loss": 0.1471,
      "step": 1854
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1903757625118065,
      "learning_rate": 8.782710307482456e-06,
      "loss": 0.2402,
      "step": 1855
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0631449535236093,
      "learning_rate": 8.78128187158382e-06,
      "loss": 0.2252,
      "step": 1856
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9147860470743994,
      "learning_rate": 8.779852714380636e-06,
      "loss": 0.1809,
      "step": 1857
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1328373490358221,
      "learning_rate": 8.778422836145529e-06,
      "loss": 0.2065,
      "step": 1858
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0917665395246432,
      "learning_rate": 8.77699223715125e-06,
      "loss": 0.2455,
      "step": 1859
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8142121591064512,
      "learning_rate": 8.775560917670701e-06,
      "loss": 0.1802,
      "step": 1860
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9526160718749868,
      "learning_rate": 8.774128877976914e-06,
      "loss": 0.1903,
      "step": 1861
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8683997736997215,
      "learning_rate": 8.77269611834306e-06,
      "loss": 0.1776,
      "step": 1862
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8851277310855235,
      "learning_rate": 8.771262639042446e-06,
      "loss": 0.1668,
      "step": 1863
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9529730888369812,
      "learning_rate": 8.76982844034852e-06,
      "loss": 0.1756,
      "step": 1864
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2787862705744721,
      "learning_rate": 8.768393522534864e-06,
      "loss": 0.2023,
      "step": 1865
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3880572232091226,
      "learning_rate": 8.766957885875198e-06,
      "loss": 0.2594,
      "step": 1866
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8833737871155324,
      "learning_rate": 8.76552153064338e-06,
      "loss": 0.1647,
      "step": 1867
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0682583149178926,
      "learning_rate": 8.764084457113403e-06,
      "loss": 0.2022,
      "step": 1868
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3907524870210932,
      "learning_rate": 8.762646665559398e-06,
      "loss": 0.2586,
      "step": 1869
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8120993716140003,
      "learning_rate": 8.761208156255634e-06,
      "loss": 0.1344,
      "step": 1870
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0321271546514192,
      "learning_rate": 8.759768929476517e-06,
      "loss": 0.2085,
      "step": 1871
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.81804944960183,
      "learning_rate": 8.758328985496588e-06,
      "loss": 0.1313,
      "step": 1872
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1861084804274171,
      "learning_rate": 8.756888324590526e-06,
      "loss": 0.2204,
      "step": 1873
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9807124566694407,
      "learning_rate": 8.755446947033145e-06,
      "loss": 0.1749,
      "step": 1874
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8498154919278662,
      "learning_rate": 8.754004853099403e-06,
      "loss": 0.1233,
      "step": 1875
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.437072159005827,
      "learning_rate": 8.752562043064382e-06,
      "loss": 0.2477,
      "step": 1876
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9034275390149012,
      "learning_rate": 8.751118517203311e-06,
      "loss": 0.1841,
      "step": 1877
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1118832155311673,
      "learning_rate": 8.749674275791552e-06,
      "loss": 0.2179,
      "step": 1878
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0008249642888791,
      "learning_rate": 8.748229319104602e-06,
      "loss": 0.1973,
      "step": 1879
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0336907687200303,
      "learning_rate": 8.746783647418097e-06,
      "loss": 0.1965,
      "step": 1880
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2096776607841795,
      "learning_rate": 8.74533726100781e-06,
      "loss": 0.2379,
      "step": 1881
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2151504022711428,
      "learning_rate": 8.74389016014965e-06,
      "loss": 0.2143,
      "step": 1882
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9545738241743099,
      "learning_rate": 8.742442345119655e-06,
      "loss": 0.2272,
      "step": 1883
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1151360766048921,
      "learning_rate": 8.74099381619401e-06,
      "loss": 0.2131,
      "step": 1884
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0590814644581865,
      "learning_rate": 8.73954457364903e-06,
      "loss": 0.1907,
      "step": 1885
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.446541782891458,
      "learning_rate": 8.738094617761168e-06,
      "loss": 0.2566,
      "step": 1886
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.004459850897038,
      "learning_rate": 8.736643948807013e-06,
      "loss": 0.1722,
      "step": 1887
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9459364052886919,
      "learning_rate": 8.735192567063292e-06,
      "loss": 0.1839,
      "step": 1888
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8841846674886797,
      "learning_rate": 8.73374047280686e-06,
      "loss": 0.1882,
      "step": 1889
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0066674939560427,
      "learning_rate": 8.732287666314716e-06,
      "loss": 0.1951,
      "step": 1890
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9812475604665867,
      "learning_rate": 8.730834147863993e-06,
      "loss": 0.1816,
      "step": 1891
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.049854654326019,
      "learning_rate": 8.729379917731962e-06,
      "loss": 0.2451,
      "step": 1892
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8348921425263717,
      "learning_rate": 8.727924976196023e-06,
      "loss": 0.154,
      "step": 1893
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8724115884227754,
      "learning_rate": 8.726469323533715e-06,
      "loss": 0.1799,
      "step": 1894
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2488001813695329,
      "learning_rate": 8.725012960022719e-06,
      "loss": 0.2779,
      "step": 1895
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0428179929107515,
      "learning_rate": 8.723555885940839e-06,
      "loss": 0.1582,
      "step": 1896
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0431182978422335,
      "learning_rate": 8.722098101566029e-06,
      "loss": 0.2217,
      "step": 1897
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2766099546200196,
      "learning_rate": 8.720639607176365e-06,
      "loss": 0.2443,
      "step": 1898
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0604157596062975,
      "learning_rate": 8.719180403050065e-06,
      "loss": 0.1826,
      "step": 1899
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1081557475579054,
      "learning_rate": 8.717720489465484e-06,
      "loss": 0.2362,
      "step": 1900
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1792533709549378,
      "learning_rate": 8.716259866701111e-06,
      "loss": 0.2094,
      "step": 1901
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.055889991862084,
      "learning_rate": 8.714798535035567e-06,
      "loss": 0.1728,
      "step": 1902
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0006001492307426,
      "learning_rate": 8.713336494747611e-06,
      "loss": 0.1988,
      "step": 1903
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9868181080040117,
      "learning_rate": 8.711873746116139e-06,
      "loss": 0.1923,
      "step": 1904
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8987411290605517,
      "learning_rate": 8.710410289420178e-06,
      "loss": 0.19,
      "step": 1905
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1723215829490277,
      "learning_rate": 8.708946124938893e-06,
      "loss": 0.2687,
      "step": 1906
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.084765681103905,
      "learning_rate": 8.707481252951582e-06,
      "loss": 0.1864,
      "step": 1907
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0491588838906274,
      "learning_rate": 8.706015673737679e-06,
      "loss": 0.2005,
      "step": 1908
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.130862352649595,
      "learning_rate": 8.704549387576754e-06,
      "loss": 0.2186,
      "step": 1909
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1814252044334612,
      "learning_rate": 8.703082394748511e-06,
      "loss": 0.2081,
      "step": 1910
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9149890794380442,
      "learning_rate": 8.701614695532787e-06,
      "loss": 0.1571,
      "step": 1911
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.097814711263668,
      "learning_rate": 8.700146290209557e-06,
      "loss": 0.1868,
      "step": 1912
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1641004236142978,
      "learning_rate": 8.698677179058929e-06,
      "loss": 0.1847,
      "step": 1913
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7327648340289978,
      "learning_rate": 8.697207362361145e-06,
      "loss": 0.1499,
      "step": 1914
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9559676258055121,
      "learning_rate": 8.69573684039658e-06,
      "loss": 0.1411,
      "step": 1915
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9437080183340508,
      "learning_rate": 8.69426561344575e-06,
      "loss": 0.2142,
      "step": 1916
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0563519026090458,
      "learning_rate": 8.6927936817893e-06,
      "loss": 0.1912,
      "step": 1917
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9486421198814015,
      "learning_rate": 8.691321045708007e-06,
      "loss": 0.1472,
      "step": 1918
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2924385449116031,
      "learning_rate": 8.689847705482793e-06,
      "loss": 0.2572,
      "step": 1919
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0347564169723253,
      "learning_rate": 8.6883736613947e-06,
      "loss": 0.2219,
      "step": 1920
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.028930330809374,
      "learning_rate": 8.686898913724917e-06,
      "loss": 0.1997,
      "step": 1921
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0721960554394292,
      "learning_rate": 8.685423462754761e-06,
      "loss": 0.1983,
      "step": 1922
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8900614353534536,
      "learning_rate": 8.683947308765681e-06,
      "loss": 0.1667,
      "step": 1923
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3181732470026806,
      "learning_rate": 8.682470452039268e-06,
      "loss": 0.2285,
      "step": 1924
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0588762854099099,
      "learning_rate": 8.680992892857237e-06,
      "loss": 0.2167,
      "step": 1925
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8641685474378329,
      "learning_rate": 8.679514631501447e-06,
      "loss": 0.1337,
      "step": 1926
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9297675419077986,
      "learning_rate": 8.678035668253882e-06,
      "loss": 0.1639,
      "step": 1927
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0745648250543616,
      "learning_rate": 8.676556003396665e-06,
      "loss": 0.2162,
      "step": 1928
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0394002832353888,
      "learning_rate": 8.675075637212055e-06,
      "loss": 0.2324,
      "step": 1929
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.052048676507404,
      "learning_rate": 8.67359456998244e-06,
      "loss": 0.2149,
      "step": 1930
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8435604812344915,
      "learning_rate": 8.67211280199034e-06,
      "loss": 0.1526,
      "step": 1931
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1599228388798308,
      "learning_rate": 8.670630333518415e-06,
      "loss": 0.2624,
      "step": 1932
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1861912082637356,
      "learning_rate": 8.669147164849456e-06,
      "loss": 0.1739,
      "step": 1933
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9537810313852728,
      "learning_rate": 8.66766329626639e-06,
      "loss": 0.1912,
      "step": 1934
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7773047013699285,
      "learning_rate": 8.666178728052267e-06,
      "loss": 0.1276,
      "step": 1935
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.767773770712797,
      "learning_rate": 8.664693460490283e-06,
      "loss": 0.121,
      "step": 1936
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9446821880588713,
      "learning_rate": 8.663207493863764e-06,
      "loss": 0.2096,
      "step": 1937
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2561640882535028,
      "learning_rate": 8.661720828456166e-06,
      "loss": 0.1546,
      "step": 1938
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2121716214993652,
      "learning_rate": 8.66023346455108e-06,
      "loss": 0.2555,
      "step": 1939
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2426588328091588,
      "learning_rate": 8.658745402432233e-06,
      "loss": 0.2201,
      "step": 1940
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9600979345145125,
      "learning_rate": 8.657256642383477e-06,
      "loss": 0.2382,
      "step": 1941
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9394460171049835,
      "learning_rate": 8.65576718468881e-06,
      "loss": 0.1684,
      "step": 1942
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9448011875693647,
      "learning_rate": 8.654277029632351e-06,
      "loss": 0.1855,
      "step": 1943
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9783526008781825,
      "learning_rate": 8.652786177498361e-06,
      "loss": 0.1945,
      "step": 1944
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8433159021410657,
      "learning_rate": 8.651294628571222e-06,
      "loss": 0.1987,
      "step": 1945
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.186117646086067,
      "learning_rate": 8.649802383135466e-06,
      "loss": 0.2137,
      "step": 1946
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0116799432590853,
      "learning_rate": 8.648309441475745e-06,
      "loss": 0.1986,
      "step": 1947
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1603419090399503,
      "learning_rate": 8.646815803876847e-06,
      "loss": 0.174,
      "step": 1948
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0894123043761723,
      "learning_rate": 8.645321470623693e-06,
      "loss": 0.1972,
      "step": 1949
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0324607885720332,
      "learning_rate": 8.643826442001338e-06,
      "loss": 0.2016,
      "step": 1950
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9993722265041053,
      "learning_rate": 8.642330718294968e-06,
      "loss": 0.1709,
      "step": 1951
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7815244223447455,
      "learning_rate": 8.640834299789902e-06,
      "loss": 0.1503,
      "step": 1952
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8769201550753022,
      "learning_rate": 8.639337186771593e-06,
      "loss": 0.1706,
      "step": 1953
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0683517486433702,
      "learning_rate": 8.637839379525623e-06,
      "loss": 0.1745,
      "step": 1954
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.975789593955614,
      "learning_rate": 8.636340878337713e-06,
      "loss": 0.2135,
      "step": 1955
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.067047346306158,
      "learning_rate": 8.634841683493705e-06,
      "loss": 0.2474,
      "step": 1956
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9407841850405558,
      "learning_rate": 8.633341795279586e-06,
      "loss": 0.1382,
      "step": 1957
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9364733516242136,
      "learning_rate": 8.63184121398147e-06,
      "loss": 0.1892,
      "step": 1958
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9712729304765322,
      "learning_rate": 8.630339939885598e-06,
      "loss": 0.2025,
      "step": 1959
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9572674783973716,
      "learning_rate": 8.628837973278352e-06,
      "loss": 0.1624,
      "step": 1960
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9110067316133976,
      "learning_rate": 8.62733531444624e-06,
      "loss": 0.1993,
      "step": 1961
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2090757221040953,
      "learning_rate": 8.625831963675906e-06,
      "loss": 0.2754,
      "step": 1962
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9427877264373946,
      "learning_rate": 8.624327921254123e-06,
      "loss": 0.201,
      "step": 1963
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9128591564033987,
      "learning_rate": 8.622823187467796e-06,
      "loss": 0.1501,
      "step": 1964
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9672717438655641,
      "learning_rate": 8.621317762603966e-06,
      "loss": 0.1931,
      "step": 1965
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7478757123069414,
      "learning_rate": 8.619811646949801e-06,
      "loss": 0.1496,
      "step": 1966
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7289749322797419,
      "learning_rate": 8.618304840792599e-06,
      "loss": 0.1425,
      "step": 1967
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0022097262991652,
      "learning_rate": 8.616797344419799e-06,
      "loss": 0.2116,
      "step": 1968
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.28730974633379,
      "learning_rate": 8.615289158118964e-06,
      "loss": 0.1739,
      "step": 1969
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9733871915714983,
      "learning_rate": 8.613780282177789e-06,
      "loss": 0.1739,
      "step": 1970
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0655021264087063,
      "learning_rate": 8.612270716884104e-06,
      "loss": 0.2126,
      "step": 1971
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8247128256540801,
      "learning_rate": 8.610760462525866e-06,
      "loss": 0.1888,
      "step": 1972
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0089086875287983,
      "learning_rate": 8.609249519391169e-06,
      "loss": 0.2052,
      "step": 1973
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0677706674673546,
      "learning_rate": 8.607737887768232e-06,
      "loss": 0.169,
      "step": 1974
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0543912617421036,
      "learning_rate": 8.606225567945412e-06,
      "loss": 0.2111,
      "step": 1975
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0363167373524964,
      "learning_rate": 8.604712560211195e-06,
      "loss": 0.1893,
      "step": 1976
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1224431226546217,
      "learning_rate": 8.603198864854192e-06,
      "loss": 0.2441,
      "step": 1977
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1950979460651014,
      "learning_rate": 8.601684482163157e-06,
      "loss": 0.2205,
      "step": 1978
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2166305909591952,
      "learning_rate": 8.600169412426962e-06,
      "loss": 0.2011,
      "step": 1979
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0030397557716961,
      "learning_rate": 8.598653655934621e-06,
      "loss": 0.1916,
      "step": 1980
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0109710745038214,
      "learning_rate": 8.597137212975273e-06,
      "loss": 0.2411,
      "step": 1981
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1673513637569877,
      "learning_rate": 8.595620083838189e-06,
      "loss": 0.1876,
      "step": 1982
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1041943001430568,
      "learning_rate": 8.594102268812772e-06,
      "loss": 0.2009,
      "step": 1983
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.201295831217475,
      "learning_rate": 8.592583768188556e-06,
      "loss": 0.2097,
      "step": 1984
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0165686723734277,
      "learning_rate": 8.591064582255205e-06,
      "loss": 0.2141,
      "step": 1985
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6859249389128687,
      "learning_rate": 8.589544711302512e-06,
      "loss": 0.1451,
      "step": 1986
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0731701166616474,
      "learning_rate": 8.588024155620405e-06,
      "loss": 0.214,
      "step": 1987
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8742680880597794,
      "learning_rate": 8.586502915498938e-06,
      "loss": 0.205,
      "step": 1988
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9843910551220715,
      "learning_rate": 8.584980991228298e-06,
      "loss": 0.171,
      "step": 1989
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1124452285903583,
      "learning_rate": 8.583458383098803e-06,
      "loss": 0.2261,
      "step": 1990
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8641976497855357,
      "learning_rate": 8.581935091400898e-06,
      "loss": 0.1941,
      "step": 1991
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.377612764852815,
      "learning_rate": 8.580411116425167e-06,
      "loss": 0.2453,
      "step": 1992
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1121492712162273,
      "learning_rate": 8.578886458462312e-06,
      "loss": 0.2018,
      "step": 1993
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.868504321229707,
      "learning_rate": 8.577361117803174e-06,
      "loss": 0.1581,
      "step": 1994
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.211883475194471,
      "learning_rate": 8.575835094738722e-06,
      "loss": 0.2708,
      "step": 1995
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1382046812552828,
      "learning_rate": 8.574308389560055e-06,
      "loss": 0.2213,
      "step": 1996
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0428991502414753,
      "learning_rate": 8.572781002558402e-06,
      "loss": 0.2478,
      "step": 1997
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9065203317041074,
      "learning_rate": 8.571252934025122e-06,
      "loss": 0.147,
      "step": 1998
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.894437592196295,
      "learning_rate": 8.569724184251705e-06,
      "loss": 0.1581,
      "step": 1999
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.994408458538145,
      "learning_rate": 8.568194753529769e-06,
      "loss": 0.1882,
      "step": 2000
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1498868051839164,
      "learning_rate": 8.566664642151061e-06,
      "loss": 0.1746,
      "step": 2001
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9466504893824678,
      "learning_rate": 8.565133850407466e-06,
      "loss": 0.1698,
      "step": 2002
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9295187407268806,
      "learning_rate": 8.563602378590987e-06,
      "loss": 0.2448,
      "step": 2003
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9484081477821281,
      "learning_rate": 8.562070226993764e-06,
      "loss": 0.1447,
      "step": 2004
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8568356678545455,
      "learning_rate": 8.560537395908067e-06,
      "loss": 0.1586,
      "step": 2005
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8146675552109632,
      "learning_rate": 8.55900388562629e-06,
      "loss": 0.183,
      "step": 2006
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.136452491548966,
      "learning_rate": 8.557469696440963e-06,
      "loss": 0.2342,
      "step": 2007
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3164102096800907,
      "learning_rate": 8.555934828644743e-06,
      "loss": 0.2111,
      "step": 2008
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0685643835260215,
      "learning_rate": 8.554399282530414e-06,
      "loss": 0.23,
      "step": 2009
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9089065263163626,
      "learning_rate": 8.552863058390895e-06,
      "loss": 0.1559,
      "step": 2010
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1787899167080997,
      "learning_rate": 8.551326156519227e-06,
      "loss": 0.2347,
      "step": 2011
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6607855089409741,
      "learning_rate": 8.549788577208587e-06,
      "loss": 0.1663,
      "step": 2012
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3091561227500434,
      "learning_rate": 8.548250320752276e-06,
      "loss": 0.2513,
      "step": 2013
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2410518924005798,
      "learning_rate": 8.54671138744373e-06,
      "loss": 0.2189,
      "step": 2014
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0823444376603235,
      "learning_rate": 8.545171777576506e-06,
      "loss": 0.2096,
      "step": 2015
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3242166353640032,
      "learning_rate": 8.543631491444302e-06,
      "loss": 0.2631,
      "step": 2016
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6270691885352067,
      "learning_rate": 8.54209052934093e-06,
      "loss": 0.1268,
      "step": 2017
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0555907929290167,
      "learning_rate": 8.540548891560341e-06,
      "loss": 0.1849,
      "step": 2018
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.027233791010255,
      "learning_rate": 8.539006578396616e-06,
      "loss": 0.1475,
      "step": 2019
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6816821754057859,
      "learning_rate": 8.537463590143958e-06,
      "loss": 0.135,
      "step": 2020
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.906843750415452,
      "learning_rate": 8.535919927096703e-06,
      "loss": 0.225,
      "step": 2021
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1773169306260034,
      "learning_rate": 8.534375589549315e-06,
      "loss": 0.2386,
      "step": 2022
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3516612463332225,
      "learning_rate": 8.532830577796389e-06,
      "loss": 0.2542,
      "step": 2023
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0661947811454242,
      "learning_rate": 8.531284892132643e-06,
      "loss": 0.2109,
      "step": 2024
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2295638840830654,
      "learning_rate": 8.529738532852928e-06,
      "loss": 0.2236,
      "step": 2025
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9583619106103646,
      "learning_rate": 8.52819150025222e-06,
      "loss": 0.213,
      "step": 2026
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9743675238617364,
      "learning_rate": 8.526643794625628e-06,
      "loss": 0.1763,
      "step": 2027
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.007182873655389,
      "learning_rate": 8.52509541626839e-06,
      "loss": 0.1962,
      "step": 2028
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9184664125563002,
      "learning_rate": 8.523546365475865e-06,
      "loss": 0.1725,
      "step": 2029
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0996527145852992,
      "learning_rate": 8.521996642543544e-06,
      "loss": 0.2059,
      "step": 2030
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9537688313346405,
      "learning_rate": 8.52044624776705e-06,
      "loss": 0.1924,
      "step": 2031
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8416750653537654,
      "learning_rate": 8.518895181442133e-06,
      "loss": 0.1952,
      "step": 2032
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0710569471983904,
      "learning_rate": 8.517343443864663e-06,
      "loss": 0.1756,
      "step": 2033
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9830754652782011,
      "learning_rate": 8.515791035330647e-06,
      "loss": 0.1685,
      "step": 2034
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7985965655100863,
      "learning_rate": 8.51423795613622e-06,
      "loss": 0.2072,
      "step": 2035
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1895126242532097,
      "learning_rate": 8.512684206577638e-06,
      "loss": 0.2056,
      "step": 2036
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3455448758693533,
      "learning_rate": 8.51112978695129e-06,
      "loss": 0.257,
      "step": 2037
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.224906104691444,
      "learning_rate": 8.509574697553695e-06,
      "loss": 0.2635,
      "step": 2038
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8814652851647444,
      "learning_rate": 8.508018938681493e-06,
      "loss": 0.1572,
      "step": 2039
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9537238536317335,
      "learning_rate": 8.506462510631456e-06,
      "loss": 0.2155,
      "step": 2040
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1775442658689665,
      "learning_rate": 8.504905413700484e-06,
      "loss": 0.1812,
      "step": 2041
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8891481434014032,
      "learning_rate": 8.503347648185602e-06,
      "loss": 0.1823,
      "step": 2042
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2306856285310386,
      "learning_rate": 8.501789214383967e-06,
      "loss": 0.1866,
      "step": 2043
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8751652789540881,
      "learning_rate": 8.500230112592856e-06,
      "loss": 0.1902,
      "step": 2044
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1031770629966293,
      "learning_rate": 8.498670343109682e-06,
      "loss": 0.2309,
      "step": 2045
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0852301300940856,
      "learning_rate": 8.49710990623198e-06,
      "loss": 0.2058,
      "step": 2046
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7939582315932869,
      "learning_rate": 8.495548802257415e-06,
      "loss": 0.1644,
      "step": 2047
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2004857535959648,
      "learning_rate": 8.493987031483774e-06,
      "loss": 0.1791,
      "step": 2048
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9220256070986392,
      "learning_rate": 8.49242459420898e-06,
      "loss": 0.1692,
      "step": 2049
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1306981433521022,
      "learning_rate": 8.490861490731074e-06,
      "loss": 0.213,
      "step": 2050
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7740709431478887,
      "learning_rate": 8.48929772134823e-06,
      "loss": 0.1376,
      "step": 2051
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1308567145863244,
      "learning_rate": 8.48773328635875e-06,
      "loss": 0.2293,
      "step": 2052
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9487192206204771,
      "learning_rate": 8.486168186061057e-06,
      "loss": 0.2113,
      "step": 2053
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4272480071539562,
      "learning_rate": 8.484602420753705e-06,
      "loss": 0.2246,
      "step": 2054
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0978506131285124,
      "learning_rate": 8.483035990735374e-06,
      "loss": 0.1976,
      "step": 2055
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0198435792990055,
      "learning_rate": 8.48146889630487e-06,
      "loss": 0.2092,
      "step": 2056
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7929565993118171,
      "learning_rate": 8.479901137761128e-06,
      "loss": 0.1848,
      "step": 2057
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7824889683430029,
      "learning_rate": 8.47833271540321e-06,
      "loss": 0.1788,
      "step": 2058
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0353188576930867,
      "learning_rate": 8.4767636295303e-06,
      "loss": 0.2116,
      "step": 2059
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0253872907651664,
      "learning_rate": 8.475193880441715e-06,
      "loss": 0.2112,
      "step": 2060
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8974391024289713,
      "learning_rate": 8.473623468436888e-06,
      "loss": 0.2103,
      "step": 2061
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0175711128258655,
      "learning_rate": 8.472052393815394e-06,
      "loss": 0.1921,
      "step": 2062
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1698863977422596,
      "learning_rate": 8.47048065687692e-06,
      "loss": 0.1913,
      "step": 2063
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2919025748927975,
      "learning_rate": 8.468908257921288e-06,
      "loss": 0.2578,
      "step": 2064
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.030631987505418,
      "learning_rate": 8.467335197248442e-06,
      "loss": 0.1729,
      "step": 2065
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9823951551021479,
      "learning_rate": 8.465761475158456e-06,
      "loss": 0.1663,
      "step": 2066
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6912203852332345,
      "learning_rate": 8.464187091951524e-06,
      "loss": 0.1495,
      "step": 2067
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.005505771682938,
      "learning_rate": 8.462612047927974e-06,
      "loss": 0.2236,
      "step": 2068
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9854057787743936,
      "learning_rate": 8.461036343388254e-06,
      "loss": 0.1968,
      "step": 2069
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0842812639358081,
      "learning_rate": 8.45945997863294e-06,
      "loss": 0.2443,
      "step": 2070
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1883505325115087,
      "learning_rate": 8.457882953962735e-06,
      "loss": 0.224,
      "step": 2071
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8504447556465586,
      "learning_rate": 8.456305269678466e-06,
      "loss": 0.1842,
      "step": 2072
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6446930259409397,
      "learning_rate": 8.45472692608109e-06,
      "loss": 0.1148,
      "step": 2073
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3042396875181173,
      "learning_rate": 8.453147923471681e-06,
      "loss": 0.2458,
      "step": 2074
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8516262467781878,
      "learning_rate": 8.451568262151447e-06,
      "loss": 0.1511,
      "step": 2075
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.198409002029336,
      "learning_rate": 8.449987942421718e-06,
      "loss": 0.2305,
      "step": 2076
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8181944075925733,
      "learning_rate": 8.448406964583953e-06,
      "loss": 0.1326,
      "step": 2077
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0781378172678426,
      "learning_rate": 8.446825328939731e-06,
      "loss": 0.2217,
      "step": 2078
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.204379558193334,
      "learning_rate": 8.445243035790762e-06,
      "loss": 0.2787,
      "step": 2079
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.607389295061974,
      "learning_rate": 8.443660085438876e-06,
      "loss": 0.1581,
      "step": 2080
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0162316707615406,
      "learning_rate": 8.442076478186033e-06,
      "loss": 0.1745,
      "step": 2081
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8971027455879684,
      "learning_rate": 8.440492214334318e-06,
      "loss": 0.1876,
      "step": 2082
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1619001731420728,
      "learning_rate": 8.438907294185936e-06,
      "loss": 0.1953,
      "step": 2083
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9030736676129841,
      "learning_rate": 8.437321718043223e-06,
      "loss": 0.1476,
      "step": 2084
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0311547837973922,
      "learning_rate": 8.43573548620864e-06,
      "loss": 0.1975,
      "step": 2085
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0010982724599153,
      "learning_rate": 8.434148598984769e-06,
      "loss": 0.1857,
      "step": 2086
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9635341413349177,
      "learning_rate": 8.432561056674318e-06,
      "loss": 0.1771,
      "step": 2087
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0153793569343534,
      "learning_rate": 8.430972859580124e-06,
      "loss": 0.2004,
      "step": 2088
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9917662340450992,
      "learning_rate": 8.429384008005146e-06,
      "loss": 0.162,
      "step": 2089
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4968991991235743,
      "learning_rate": 8.427794502252466e-06,
      "loss": 0.3151,
      "step": 2090
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9533342554082688,
      "learning_rate": 8.426204342625295e-06,
      "loss": 0.1857,
      "step": 2091
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1639844035678464,
      "learning_rate": 8.424613529426964e-06,
      "loss": 0.196,
      "step": 2092
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9316628958281509,
      "learning_rate": 8.423022062960934e-06,
      "loss": 0.1961,
      "step": 2093
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0092498680004418,
      "learning_rate": 8.421429943530786e-06,
      "loss": 0.2343,
      "step": 2094
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0292648275758112,
      "learning_rate": 8.419837171440226e-06,
      "loss": 0.1882,
      "step": 2095
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1290195325447587,
      "learning_rate": 8.418243746993087e-06,
      "loss": 0.2062,
      "step": 2096
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1402199320248103,
      "learning_rate": 8.416649670493326e-06,
      "loss": 0.1929,
      "step": 2097
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9069918480382964,
      "learning_rate": 8.415054942245025e-06,
      "loss": 0.2142,
      "step": 2098
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.032761758193651,
      "learning_rate": 8.413459562552385e-06,
      "loss": 0.2159,
      "step": 2099
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1708123867692994,
      "learning_rate": 8.411863531719739e-06,
      "loss": 0.2348,
      "step": 2100
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9549449088917619,
      "learning_rate": 8.410266850051539e-06,
      "loss": 0.1868,
      "step": 2101
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6826461289342649,
      "learning_rate": 8.40866951785236e-06,
      "loss": 0.1421,
      "step": 2102
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9760367114584734,
      "learning_rate": 8.407071535426907e-06,
      "loss": 0.1851,
      "step": 2103
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9229048858117024,
      "learning_rate": 8.405472903080007e-06,
      "loss": 0.1899,
      "step": 2104
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2138508107119301,
      "learning_rate": 8.403873621116608e-06,
      "loss": 0.2524,
      "step": 2105
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3601868910060384,
      "learning_rate": 8.40227368984178e-06,
      "loss": 0.2072,
      "step": 2106
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9368571852907944,
      "learning_rate": 8.400673109560724e-06,
      "loss": 0.1863,
      "step": 2107
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2759749999119687,
      "learning_rate": 8.399071880578763e-06,
      "loss": 0.2047,
      "step": 2108
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3927423787247357,
      "learning_rate": 8.397470003201339e-06,
      "loss": 0.275,
      "step": 2109
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0006318089621322,
      "learning_rate": 8.395867477734023e-06,
      "loss": 0.2095,
      "step": 2110
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3845631014312223,
      "learning_rate": 8.394264304482505e-06,
      "loss": 0.2565,
      "step": 2111
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1291132255885314,
      "learning_rate": 8.3926604837526e-06,
      "loss": 0.1672,
      "step": 2112
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1879100933448277,
      "learning_rate": 8.391056015850251e-06,
      "loss": 0.1969,
      "step": 2113
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.157212056844958,
      "learning_rate": 8.38945090108152e-06,
      "loss": 0.1953,
      "step": 2114
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9121009280373578,
      "learning_rate": 8.387845139752592e-06,
      "loss": 0.1784,
      "step": 2115
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0503942952325562,
      "learning_rate": 8.386238732169777e-06,
      "loss": 0.2226,
      "step": 2116
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8481931279410719,
      "learning_rate": 8.384631678639508e-06,
      "loss": 0.2121,
      "step": 2117
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0845021726862571,
      "learning_rate": 8.383023979468341e-06,
      "loss": 0.2349,
      "step": 2118
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3135883770682046,
      "learning_rate": 8.381415634962957e-06,
      "loss": 0.2246,
      "step": 2119
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.177236133785637,
      "learning_rate": 8.379806645430155e-06,
      "loss": 0.228,
      "step": 2120
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1743999799511216,
      "learning_rate": 8.378197011176863e-06,
      "loss": 0.2299,
      "step": 2121
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0301760326431628,
      "learning_rate": 8.376586732510127e-06,
      "loss": 0.2352,
      "step": 2122
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9816770757315344,
      "learning_rate": 8.374975809737121e-06,
      "loss": 0.2095,
      "step": 2123
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9478696137284703,
      "learning_rate": 8.373364243165138e-06,
      "loss": 0.2008,
      "step": 2124
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0875572578828445,
      "learning_rate": 8.371752033101594e-06,
      "loss": 0.1933,
      "step": 2125
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9980537759770316,
      "learning_rate": 8.370139179854032e-06,
      "loss": 0.2175,
      "step": 2126
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9334440613287731,
      "learning_rate": 8.36852568373011e-06,
      "loss": 0.1801,
      "step": 2127
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2995319118279243,
      "learning_rate": 8.366911545037617e-06,
      "loss": 0.2527,
      "step": 2128
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0504647150006032,
      "learning_rate": 8.365296764084458e-06,
      "loss": 0.1483,
      "step": 2129
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1688104086618796,
      "learning_rate": 8.363681341178665e-06,
      "loss": 0.216,
      "step": 2130
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0654965907335177,
      "learning_rate": 8.362065276628387e-06,
      "loss": 0.2468,
      "step": 2131
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1184838988447174,
      "learning_rate": 8.360448570741903e-06,
      "loss": 0.1713,
      "step": 2132
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9377381946019561,
      "learning_rate": 8.358831223827611e-06,
      "loss": 0.1949,
      "step": 2133
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3360003846949062,
      "learning_rate": 8.357213236194027e-06,
      "loss": 0.2802,
      "step": 2134
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2956091077720964,
      "learning_rate": 8.355594608149794e-06,
      "loss": 0.2756,
      "step": 2135
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8561303617245075,
      "learning_rate": 8.353975340003677e-06,
      "loss": 0.1839,
      "step": 2136
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1084994214187498,
      "learning_rate": 8.352355432064563e-06,
      "loss": 0.2635,
      "step": 2137
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6413410704482605,
      "learning_rate": 8.350734884641457e-06,
      "loss": 0.1375,
      "step": 2138
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.896328992123526,
      "learning_rate": 8.349113698043492e-06,
      "loss": 0.1982,
      "step": 2139
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1717254158867774,
      "learning_rate": 8.347491872579916e-06,
      "loss": 0.2114,
      "step": 2140
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9292744891810452,
      "learning_rate": 8.34586940856011e-06,
      "loss": 0.2102,
      "step": 2141
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1171861906924359,
      "learning_rate": 8.344246306293564e-06,
      "loss": 0.2226,
      "step": 2142
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0930157714729685,
      "learning_rate": 8.342622566089897e-06,
      "loss": 0.2342,
      "step": 2143
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2593278094074942,
      "learning_rate": 8.340998188258849e-06,
      "loss": 0.2195,
      "step": 2144
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1544048223347412,
      "learning_rate": 8.33937317311028e-06,
      "loss": 0.2222,
      "step": 2145
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9408133225311955,
      "learning_rate": 8.337747520954174e-06,
      "loss": 0.1643,
      "step": 2146
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0459033573133925,
      "learning_rate": 8.336121232100631e-06,
      "loss": 0.2027,
      "step": 2147
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7560717398783314,
      "learning_rate": 8.334494306859882e-06,
      "loss": 0.1413,
      "step": 2148
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0329919634914546,
      "learning_rate": 8.332866745542267e-06,
      "loss": 0.2488,
      "step": 2149
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9357437778574349,
      "learning_rate": 8.33123854845826e-06,
      "loss": 0.2094,
      "step": 2150
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8628132895545215,
      "learning_rate": 8.329609715918448e-06,
      "loss": 0.1509,
      "step": 2151
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2977898536219303,
      "learning_rate": 8.327980248233541e-06,
      "loss": 0.2377,
      "step": 2152
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0105442371702493,
      "learning_rate": 8.326350145714372e-06,
      "loss": 0.205,
      "step": 2153
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1534626907685037,
      "learning_rate": 8.324719408671893e-06,
      "loss": 0.2225,
      "step": 2154
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.095552389974664,
      "learning_rate": 8.32308803741718e-06,
      "loss": 0.2437,
      "step": 2155
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0364710847571041,
      "learning_rate": 8.321456032261423e-06,
      "loss": 0.173,
      "step": 2156
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9266516557682718,
      "learning_rate": 8.319823393515943e-06,
      "loss": 0.193,
      "step": 2157
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0508847823003926,
      "learning_rate": 8.318190121492174e-06,
      "loss": 0.1974,
      "step": 2158
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.11304544034215,
      "learning_rate": 8.316556216501674e-06,
      "loss": 0.2162,
      "step": 2159
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0514993703075277,
      "learning_rate": 8.31492167885612e-06,
      "loss": 0.2244,
      "step": 2160
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0838646458291894,
      "learning_rate": 8.313286508867314e-06,
      "loss": 0.1417,
      "step": 2161
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1047922122552807,
      "learning_rate": 8.311650706847172e-06,
      "loss": 0.1906,
      "step": 2162
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9452268967783143,
      "learning_rate": 8.310014273107735e-06,
      "loss": 0.2163,
      "step": 2163
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9063587016980882,
      "learning_rate": 8.308377207961166e-06,
      "loss": 0.2081,
      "step": 2164
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.46291218696199904,
      "learning_rate": 8.306739511719745e-06,
      "loss": 0.1256,
      "step": 2165
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.11655754717091,
      "learning_rate": 8.30510118469587e-06,
      "loss": 0.2524,
      "step": 2166
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2280038776555644,
      "learning_rate": 8.303462227202069e-06,
      "loss": 0.2394,
      "step": 2167
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6362539345600617,
      "learning_rate": 8.301822639550976e-06,
      "loss": 0.1531,
      "step": 2168
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9883033482091266,
      "learning_rate": 8.300182422055358e-06,
      "loss": 0.2011,
      "step": 2169
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9805711880479415,
      "learning_rate": 8.298541575028099e-06,
      "loss": 0.1635,
      "step": 2170
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.600871032284938,
      "learning_rate": 8.296900098782197e-06,
      "loss": 0.1501,
      "step": 2171
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2716774357818486,
      "learning_rate": 8.295257993630778e-06,
      "loss": 0.2316,
      "step": 2172
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0672648696201783,
      "learning_rate": 8.293615259887085e-06,
      "loss": 0.2252,
      "step": 2173
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7528412316884567,
      "learning_rate": 8.291971897864475e-06,
      "loss": 0.1994,
      "step": 2174
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2641631015209451,
      "learning_rate": 8.290327907876437e-06,
      "loss": 0.2243,
      "step": 2175
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0194503091081621,
      "learning_rate": 8.288683290236567e-06,
      "loss": 0.1862,
      "step": 2176
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8671374328766818,
      "learning_rate": 8.287038045258591e-06,
      "loss": 0.1279,
      "step": 2177
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1406651779533552,
      "learning_rate": 8.285392173256348e-06,
      "loss": 0.1967,
      "step": 2178
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.027605555908166,
      "learning_rate": 8.283745674543798e-06,
      "loss": 0.1675,
      "step": 2179
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1386709601053016,
      "learning_rate": 8.282098549435026e-06,
      "loss": 0.195,
      "step": 2180
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.97179890238985,
      "learning_rate": 8.280450798244227e-06,
      "loss": 0.1875,
      "step": 2181
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8629062351789024,
      "learning_rate": 8.278802421285726e-06,
      "loss": 0.1507,
      "step": 2182
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8251024530739629,
      "learning_rate": 8.277153418873956e-06,
      "loss": 0.1639,
      "step": 2183
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0076829897083344,
      "learning_rate": 8.275503791323476e-06,
      "loss": 0.1667,
      "step": 2184
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9235913960296773,
      "learning_rate": 8.273853538948969e-06,
      "loss": 0.1507,
      "step": 2185
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2343223398150562,
      "learning_rate": 8.272202662065225e-06,
      "loss": 0.2585,
      "step": 2186
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8771120730814458,
      "learning_rate": 8.270551160987161e-06,
      "loss": 0.142,
      "step": 2187
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.075361585575739,
      "learning_rate": 8.268899036029816e-06,
      "loss": 0.2387,
      "step": 2188
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9200708567116188,
      "learning_rate": 8.26724628750834e-06,
      "loss": 0.1446,
      "step": 2189
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.204593667340598,
      "learning_rate": 8.265592915738006e-06,
      "loss": 0.1939,
      "step": 2190
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.079451342683517,
      "learning_rate": 8.263938921034206e-06,
      "loss": 0.2049,
      "step": 2191
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3314767249926893,
      "learning_rate": 8.26228430371245e-06,
      "loss": 0.2524,
      "step": 2192
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.027950821765022,
      "learning_rate": 8.26062906408837e-06,
      "loss": 0.1862,
      "step": 2193
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0696516968922556,
      "learning_rate": 8.25897320247771e-06,
      "loss": 0.2141,
      "step": 2194
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9268463422850302,
      "learning_rate": 8.25731671919634e-06,
      "loss": 0.1785,
      "step": 2195
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9568213186108524,
      "learning_rate": 8.255659614560241e-06,
      "loss": 0.1664,
      "step": 2196
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7855335029884315,
      "learning_rate": 8.25400188888552e-06,
      "loss": 0.1346,
      "step": 2197
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1003398841709715,
      "learning_rate": 8.252343542488397e-06,
      "loss": 0.2323,
      "step": 2198
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0404913236378173,
      "learning_rate": 8.250684575685214e-06,
      "loss": 0.1926,
      "step": 2199
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9276251519125138,
      "learning_rate": 8.249024988792433e-06,
      "loss": 0.2053,
      "step": 2200
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6964253064516434,
      "learning_rate": 8.247364782126625e-06,
      "loss": 0.1524,
      "step": 2201
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0201232542444862,
      "learning_rate": 8.24570395600449e-06,
      "loss": 0.2491,
      "step": 2202
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9647316368238521,
      "learning_rate": 8.244042510742838e-06,
      "loss": 0.1661,
      "step": 2203
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8401868570761994,
      "learning_rate": 8.242380446658602e-06,
      "loss": 0.1342,
      "step": 2204
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9377789220931526,
      "learning_rate": 8.240717764068835e-06,
      "loss": 0.231,
      "step": 2205
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1547510107759391,
      "learning_rate": 8.239054463290701e-06,
      "loss": 0.1934,
      "step": 2206
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.148402310315943,
      "learning_rate": 8.237390544641485e-06,
      "loss": 0.2161,
      "step": 2207
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0783584387212832,
      "learning_rate": 8.235726008438594e-06,
      "loss": 0.1838,
      "step": 2208
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6728763474999111,
      "learning_rate": 8.234060854999546e-06,
      "loss": 0.1338,
      "step": 2209
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0256516301263843,
      "learning_rate": 8.232395084641982e-06,
      "loss": 0.1528,
      "step": 2210
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7140594613291507,
      "learning_rate": 8.230728697683658e-06,
      "loss": 0.1651,
      "step": 2211
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7403565266502637,
      "learning_rate": 8.229061694442451e-06,
      "loss": 0.1834,
      "step": 2212
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8781943557943832,
      "learning_rate": 8.227394075236347e-06,
      "loss": 0.1308,
      "step": 2213
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9632484691072696,
      "learning_rate": 8.22572584038346e-06,
      "loss": 0.1681,
      "step": 2214
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9865672605209224,
      "learning_rate": 8.224056990202016e-06,
      "loss": 0.1585,
      "step": 2215
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.914609306796269,
      "learning_rate": 8.222387525010357e-06,
      "loss": 0.1526,
      "step": 2216
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9916696019494328,
      "learning_rate": 8.22071744512695e-06,
      "loss": 0.2009,
      "step": 2217
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1813677726654375,
      "learning_rate": 8.219046750870367e-06,
      "loss": 0.2165,
      "step": 2218
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1370880225356776,
      "learning_rate": 8.21737544255931e-06,
      "loss": 0.1836,
      "step": 2219
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.54660953112403,
      "learning_rate": 8.215703520512588e-06,
      "loss": 0.3344,
      "step": 2220
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9144064997880809,
      "learning_rate": 8.214030985049131e-06,
      "loss": 0.1977,
      "step": 2221
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9517972335106583,
      "learning_rate": 8.21235783648799e-06,
      "loss": 0.1229,
      "step": 2222
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1743576627434067,
      "learning_rate": 8.210684075148325e-06,
      "loss": 0.2278,
      "step": 2223
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8607666934059566,
      "learning_rate": 8.20900970134942e-06,
      "loss": 0.1833,
      "step": 2224
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5880782276347337,
      "learning_rate": 8.207334715410671e-06,
      "loss": 0.1439,
      "step": 2225
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8751033401120727,
      "learning_rate": 8.205659117651594e-06,
      "loss": 0.1484,
      "step": 2226
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0477296152432438,
      "learning_rate": 8.203982908391821e-06,
      "loss": 0.2107,
      "step": 2227
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.06918575689938,
      "learning_rate": 8.202306087951098e-06,
      "loss": 0.2067,
      "step": 2228
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1483728147898322,
      "learning_rate": 8.20062865664929e-06,
      "loss": 0.1955,
      "step": 2229
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9731235502300674,
      "learning_rate": 8.198950614806378e-06,
      "loss": 0.1819,
      "step": 2230
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.766139183068051,
      "learning_rate": 8.197271962742463e-06,
      "loss": 0.1721,
      "step": 2231
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7798250414378838,
      "learning_rate": 8.195592700777755e-06,
      "loss": 0.1091,
      "step": 2232
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9589274117442927,
      "learning_rate": 8.193912829232585e-06,
      "loss": 0.1664,
      "step": 2233
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.812140528118853,
      "learning_rate": 8.1922323484274e-06,
      "loss": 0.156,
      "step": 2234
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0299781729968773,
      "learning_rate": 8.190551258682761e-06,
      "loss": 0.2161,
      "step": 2235
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1274118719317008,
      "learning_rate": 8.188869560319353e-06,
      "loss": 0.1999,
      "step": 2236
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6314067135791346,
      "learning_rate": 8.187187253657965e-06,
      "loss": 0.1386,
      "step": 2237
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9813418178525695,
      "learning_rate": 8.18550433901951e-06,
      "loss": 0.2126,
      "step": 2238
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0693064080366987,
      "learning_rate": 8.183820816725015e-06,
      "loss": 0.2138,
      "step": 2239
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0156369026062158,
      "learning_rate": 8.182136687095625e-06,
      "loss": 0.2357,
      "step": 2240
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1322877290276059,
      "learning_rate": 8.180451950452595e-06,
      "loss": 0.2096,
      "step": 2241
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0856564508924773,
      "learning_rate": 8.178766607117303e-06,
      "loss": 0.1944,
      "step": 2242
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9933243616832047,
      "learning_rate": 8.177080657411238e-06,
      "loss": 0.1857,
      "step": 2243
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7110954796848553,
      "learning_rate": 8.175394101656007e-06,
      "loss": 0.1443,
      "step": 2244
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.903329915722252,
      "learning_rate": 8.173706940173331e-06,
      "loss": 0.1521,
      "step": 2245
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2084116418671134,
      "learning_rate": 8.172019173285046e-06,
      "loss": 0.2563,
      "step": 2246
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.124100392018233,
      "learning_rate": 8.170330801313106e-06,
      "loss": 0.1769,
      "step": 2247
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9997579735117957,
      "learning_rate": 8.16864182457958e-06,
      "loss": 0.2105,
      "step": 2248
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.838953834420982,
      "learning_rate": 8.16695224340665e-06,
      "loss": 0.1167,
      "step": 2249
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8022931399414237,
      "learning_rate": 8.165262058116617e-06,
      "loss": 0.1419,
      "step": 2250
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9496443464727633,
      "learning_rate": 8.16357126903189e-06,
      "loss": 0.182,
      "step": 2251
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.213728402372826,
      "learning_rate": 8.161879876475006e-06,
      "loss": 0.2355,
      "step": 2252
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9830721799428025,
      "learning_rate": 8.160187880768602e-06,
      "loss": 0.1819,
      "step": 2253
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1724173131103592,
      "learning_rate": 8.15849528223544e-06,
      "loss": 0.2182,
      "step": 2254
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9418261562593166,
      "learning_rate": 8.156802081198396e-06,
      "loss": 0.2115,
      "step": 2255
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.057334616417378,
      "learning_rate": 8.155108277980456e-06,
      "loss": 0.1848,
      "step": 2256
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8250724999191061,
      "learning_rate": 8.153413872904728e-06,
      "loss": 0.1884,
      "step": 2257
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.001520118704844,
      "learning_rate": 8.151718866294429e-06,
      "loss": 0.1916,
      "step": 2258
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0537827331386813,
      "learning_rate": 8.150023258472893e-06,
      "loss": 0.2099,
      "step": 2259
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0813018022532674,
      "learning_rate": 8.148327049763569e-06,
      "loss": 0.1562,
      "step": 2260
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8069208663461622,
      "learning_rate": 8.146630240490018e-06,
      "loss": 0.2044,
      "step": 2261
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9459465749749674,
      "learning_rate": 8.14493283097592e-06,
      "loss": 0.2164,
      "step": 2262
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.84640600809592,
      "learning_rate": 8.143234821545063e-06,
      "loss": 0.1931,
      "step": 2263
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9519293490357946,
      "learning_rate": 8.14153621252136e-06,
      "loss": 0.227,
      "step": 2264
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0084275134722331,
      "learning_rate": 8.139837004228828e-06,
      "loss": 0.1993,
      "step": 2265
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1504973097506954,
      "learning_rate": 8.138137196991602e-06,
      "loss": 0.1814,
      "step": 2266
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1267452024544347,
      "learning_rate": 8.136436791133932e-06,
      "loss": 0.2006,
      "step": 2267
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8861560806565022,
      "learning_rate": 8.13473578698018e-06,
      "loss": 0.1969,
      "step": 2268
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0763100572931472,
      "learning_rate": 8.133034184854828e-06,
      "loss": 0.1674,
      "step": 2269
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9844768190056123,
      "learning_rate": 8.131331985082466e-06,
      "loss": 0.1822,
      "step": 2270
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.235466299357091,
      "learning_rate": 8.129629187987799e-06,
      "loss": 0.2015,
      "step": 2271
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.060476478597213,
      "learning_rate": 8.127925793895647e-06,
      "loss": 0.1992,
      "step": 2272
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0310623923537072,
      "learning_rate": 8.126221803130944e-06,
      "loss": 0.2116,
      "step": 2273
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1994031270474312,
      "learning_rate": 8.124517216018738e-06,
      "loss": 0.2211,
      "step": 2274
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8271171869109314,
      "learning_rate": 8.122812032884188e-06,
      "loss": 0.1795,
      "step": 2275
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0834028244877305,
      "learning_rate": 8.121106254052571e-06,
      "loss": 0.186,
      "step": 2276
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0151854003796195,
      "learning_rate": 8.119399879849277e-06,
      "loss": 0.1585,
      "step": 2277
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8269065084743129,
      "learning_rate": 8.117692910599805e-06,
      "loss": 0.1746,
      "step": 2278
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9140710048085571,
      "learning_rate": 8.115985346629774e-06,
      "loss": 0.1376,
      "step": 2279
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0229528488935287,
      "learning_rate": 8.114277188264909e-06,
      "loss": 0.1979,
      "step": 2280
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8182465394457685,
      "learning_rate": 8.112568435831057e-06,
      "loss": 0.1588,
      "step": 2281
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8713364449779797,
      "learning_rate": 8.11085908965417e-06,
      "loss": 0.1513,
      "step": 2282
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.950597702958833,
      "learning_rate": 8.10914915006032e-06,
      "loss": 0.1809,
      "step": 2283
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9336158360753143,
      "learning_rate": 8.107438617375687e-06,
      "loss": 0.1326,
      "step": 2284
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0020485437097177,
      "learning_rate": 8.105727491926568e-06,
      "loss": 0.2028,
      "step": 2285
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9294228417072202,
      "learning_rate": 8.10401577403937e-06,
      "loss": 0.2101,
      "step": 2286
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0551059935350728,
      "learning_rate": 8.102303464040615e-06,
      "loss": 0.1721,
      "step": 2287
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8002456182535358,
      "learning_rate": 8.100590562256941e-06,
      "loss": 0.1718,
      "step": 2288
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.685898954815334,
      "learning_rate": 8.09887706901509e-06,
      "loss": 0.1881,
      "step": 2289
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9223903361797338,
      "learning_rate": 8.097162984641927e-06,
      "loss": 0.1012,
      "step": 2290
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0657433509121568,
      "learning_rate": 8.095448309464417e-06,
      "loss": 0.2535,
      "step": 2291
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7552544638302718,
      "learning_rate": 8.093733043809656e-06,
      "loss": 0.1296,
      "step": 2292
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0508381071221844,
      "learning_rate": 8.092017188004837e-06,
      "loss": 0.2258,
      "step": 2293
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8986133331940236,
      "learning_rate": 8.090300742377269e-06,
      "loss": 0.1528,
      "step": 2294
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6762325784355255,
      "learning_rate": 8.088583707254379e-06,
      "loss": 0.1496,
      "step": 2295
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0918771696850438,
      "learning_rate": 8.086866082963702e-06,
      "loss": 0.1697,
      "step": 2296
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8786738060378387,
      "learning_rate": 8.085147869832884e-06,
      "loss": 0.1807,
      "step": 2297
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1191804226890123,
      "learning_rate": 8.083429068189688e-06,
      "loss": 0.2155,
      "step": 2298
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.341506085910093,
      "learning_rate": 8.081709678361986e-06,
      "loss": 0.312,
      "step": 2299
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2956057374765844,
      "learning_rate": 8.079989700677762e-06,
      "loss": 0.2137,
      "step": 2300
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0679446230736978,
      "learning_rate": 8.078269135465114e-06,
      "loss": 0.1658,
      "step": 2301
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8978706476749345,
      "learning_rate": 8.076547983052252e-06,
      "loss": 0.2017,
      "step": 2302
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8762232580870323,
      "learning_rate": 8.074826243767497e-06,
      "loss": 0.1929,
      "step": 2303
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2023799051507404,
      "learning_rate": 8.07310391793928e-06,
      "loss": 0.2257,
      "step": 2304
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8930059827167441,
      "learning_rate": 8.07138100589615e-06,
      "loss": 0.1609,
      "step": 2305
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.062902497664649,
      "learning_rate": 8.069657507966759e-06,
      "loss": 0.2483,
      "step": 2306
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.205468803581842,
      "learning_rate": 8.067933424479882e-06,
      "loss": 0.2504,
      "step": 2307
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0787037869005491,
      "learning_rate": 8.066208755764391e-06,
      "loss": 0.2036,
      "step": 2308
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.162873988601957,
      "learning_rate": 8.064483502149285e-06,
      "loss": 0.2216,
      "step": 2309
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1635037976133231,
      "learning_rate": 8.062757663963665e-06,
      "loss": 0.1649,
      "step": 2310
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.035798078766979,
      "learning_rate": 8.061031241536745e-06,
      "loss": 0.2032,
      "step": 2311
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.114509863269153,
      "learning_rate": 8.059304235197853e-06,
      "loss": 0.2279,
      "step": 2312
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0494255806148676,
      "learning_rate": 8.057576645276428e-06,
      "loss": 0.2005,
      "step": 2313
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9523983752832498,
      "learning_rate": 8.055848472102015e-06,
      "loss": 0.1826,
      "step": 2314
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0752978697612667,
      "learning_rate": 8.054119716004279e-06,
      "loss": 0.1732,
      "step": 2315
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7778857077184407,
      "learning_rate": 8.05239037731299e-06,
      "loss": 0.1735,
      "step": 2316
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.706122550806684,
      "learning_rate": 8.050660456358031e-06,
      "loss": 0.1329,
      "step": 2317
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9050315004334504,
      "learning_rate": 8.048929953469394e-06,
      "loss": 0.2022,
      "step": 2318
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0546961326635733,
      "learning_rate": 8.047198868977187e-06,
      "loss": 0.1789,
      "step": 2319
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8795432928772389,
      "learning_rate": 8.045467203211623e-06,
      "loss": 0.1523,
      "step": 2320
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1423464595753097,
      "learning_rate": 8.04373495650303e-06,
      "loss": 0.2373,
      "step": 2321
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2220184216622463,
      "learning_rate": 8.042002129181844e-06,
      "loss": 0.2425,
      "step": 2322
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8469105471789272,
      "learning_rate": 8.040268721578619e-06,
      "loss": 0.1132,
      "step": 2323
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.021835515055419,
      "learning_rate": 8.038534734024008e-06,
      "loss": 0.2258,
      "step": 2324
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9587618183238347,
      "learning_rate": 8.036800166848781e-06,
      "loss": 0.2198,
      "step": 2325
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.075340271406541,
      "learning_rate": 8.03506502038382e-06,
      "loss": 0.2288,
      "step": 2326
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6019555696347267,
      "learning_rate": 8.033329294960116e-06,
      "loss": 0.1051,
      "step": 2327
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.053513761483285,
      "learning_rate": 8.031592990908769e-06,
      "loss": 0.2033,
      "step": 2328
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9490452921000121,
      "learning_rate": 8.02985610856099e-06,
      "loss": 0.1382,
      "step": 2329
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.094057100785234,
      "learning_rate": 8.028118648248102e-06,
      "loss": 0.2374,
      "step": 2330
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8460114292550422,
      "learning_rate": 8.026380610301537e-06,
      "loss": 0.1802,
      "step": 2331
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8565540692119484,
      "learning_rate": 8.024641995052837e-06,
      "loss": 0.1708,
      "step": 2332
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0548190483370201,
      "learning_rate": 8.022902802833655e-06,
      "loss": 0.2069,
      "step": 2333
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8647164956064042,
      "learning_rate": 8.02116303397575e-06,
      "loss": 0.1465,
      "step": 2334
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9850540199981374,
      "learning_rate": 8.019422688811e-06,
      "loss": 0.2614,
      "step": 2335
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3592602964900535,
      "learning_rate": 8.017681767671382e-06,
      "loss": 0.2582,
      "step": 2336
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1340579184908186,
      "learning_rate": 8.015940270888992e-06,
      "loss": 0.2205,
      "step": 2337
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.076511436789397,
      "learning_rate": 8.01419819879603e-06,
      "loss": 0.1516,
      "step": 2338
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0307741744241292,
      "learning_rate": 8.01245555172481e-06,
      "loss": 0.2178,
      "step": 2339
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8162032751032494,
      "learning_rate": 8.010712330007752e-06,
      "loss": 0.1367,
      "step": 2340
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9088378723933039,
      "learning_rate": 8.008968533977386e-06,
      "loss": 0.1502,
      "step": 2341
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1280857538919171,
      "learning_rate": 8.007224163966353e-06,
      "loss": 0.2112,
      "step": 2342
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0862308970500194,
      "learning_rate": 8.005479220307405e-06,
      "loss": 0.2046,
      "step": 2343
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3863605349702721,
      "learning_rate": 8.003733703333402e-06,
      "loss": 0.2646,
      "step": 2344
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9706236513545669,
      "learning_rate": 8.00198761337731e-06,
      "loss": 0.2118,
      "step": 2345
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9871121520876748,
      "learning_rate": 8.000240950772208e-06,
      "loss": 0.1787,
      "step": 2346
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7819099274809233,
      "learning_rate": 7.998493715851283e-06,
      "loss": 0.1564,
      "step": 2347
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8724080813077161,
      "learning_rate": 7.996745908947834e-06,
      "loss": 0.1769,
      "step": 2348
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2031448214622709,
      "learning_rate": 7.994997530395266e-06,
      "loss": 0.2669,
      "step": 2349
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8780463009364584,
      "learning_rate": 7.993248580527093e-06,
      "loss": 0.2003,
      "step": 2350
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9711861978962493,
      "learning_rate": 7.991499059676937e-06,
      "loss": 0.1751,
      "step": 2351
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1273470995505042,
      "learning_rate": 7.989748968178536e-06,
      "loss": 0.2431,
      "step": 2352
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0611702101247467,
      "learning_rate": 7.987998306365725e-06,
      "loss": 0.2012,
      "step": 2353
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9225746742320041,
      "learning_rate": 7.986247074572457e-06,
      "loss": 0.2135,
      "step": 2354
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7666784183249867,
      "learning_rate": 7.984495273132795e-06,
      "loss": 0.1356,
      "step": 2355
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8857298692556056,
      "learning_rate": 7.982742902380901e-06,
      "loss": 0.1745,
      "step": 2356
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0413502694543366,
      "learning_rate": 7.980989962651054e-06,
      "loss": 0.1803,
      "step": 2357
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8361358731319376,
      "learning_rate": 7.979236454277637e-06,
      "loss": 0.137,
      "step": 2358
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9238188085228582,
      "learning_rate": 7.977482377595147e-06,
      "loss": 0.1766,
      "step": 2359
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9224114790572554,
      "learning_rate": 7.975727732938183e-06,
      "loss": 0.1404,
      "step": 2360
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9194154670896808,
      "learning_rate": 7.973972520641454e-06,
      "loss": 0.1854,
      "step": 2361
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6780653290262262,
      "learning_rate": 7.972216741039781e-06,
      "loss": 0.1223,
      "step": 2362
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8782508209568837,
      "learning_rate": 7.970460394468088e-06,
      "loss": 0.1621,
      "step": 2363
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9262643382055853,
      "learning_rate": 7.968703481261414e-06,
      "loss": 0.1461,
      "step": 2364
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.730793152333362,
      "learning_rate": 7.966946001754899e-06,
      "loss": 0.1729,
      "step": 2365
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0646549153935234,
      "learning_rate": 7.965187956283791e-06,
      "loss": 0.2345,
      "step": 2366
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3440770745420596,
      "learning_rate": 7.963429345183455e-06,
      "loss": 0.2437,
      "step": 2367
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.079135911373999,
      "learning_rate": 7.961670168789353e-06,
      "loss": 0.1833,
      "step": 2368
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8013734107679173,
      "learning_rate": 7.95991042743706e-06,
      "loss": 0.1649,
      "step": 2369
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1632545482994772,
      "learning_rate": 7.95815012146226e-06,
      "loss": 0.2167,
      "step": 2370
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1449168561195702,
      "learning_rate": 7.956389251200742e-06,
      "loss": 0.2141,
      "step": 2371
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0665605408437047,
      "learning_rate": 7.954627816988405e-06,
      "loss": 0.2012,
      "step": 2372
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0630924267351818,
      "learning_rate": 7.952865819161253e-06,
      "loss": 0.2267,
      "step": 2373
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.364600007342432,
      "learning_rate": 7.951103258055398e-06,
      "loss": 0.2459,
      "step": 2374
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.098092256682196,
      "learning_rate": 7.949340134007061e-06,
      "loss": 0.2045,
      "step": 2375
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8059605805311861,
      "learning_rate": 7.947576447352571e-06,
      "loss": 0.1748,
      "step": 2376
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8900642732706798,
      "learning_rate": 7.945812198428361e-06,
      "loss": 0.2014,
      "step": 2377
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9777438876192389,
      "learning_rate": 7.944047387570974e-06,
      "loss": 0.2243,
      "step": 2378
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8400652998211581,
      "learning_rate": 7.94228201511706e-06,
      "loss": 0.1356,
      "step": 2379
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0358595069992182,
      "learning_rate": 7.940516081403373e-06,
      "loss": 0.1934,
      "step": 2380
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1807714733736934,
      "learning_rate": 7.938749586766778e-06,
      "loss": 0.23,
      "step": 2381
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9246765439690922,
      "learning_rate": 7.936982531544248e-06,
      "loss": 0.2262,
      "step": 2382
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1946005861886884,
      "learning_rate": 7.935214916072856e-06,
      "loss": 0.2211,
      "step": 2383
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2115717190331012,
      "learning_rate": 7.93344674068979e-06,
      "loss": 0.2157,
      "step": 2384
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8679539857092076,
      "learning_rate": 7.93167800573234e-06,
      "loss": 0.1702,
      "step": 2385
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0388874595553887,
      "learning_rate": 7.929908711537902e-06,
      "loss": 0.2076,
      "step": 2386
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0734999866485089,
      "learning_rate": 7.928138858443984e-06,
      "loss": 0.195,
      "step": 2387
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0982874320957747,
      "learning_rate": 7.926368446788192e-06,
      "loss": 0.2459,
      "step": 2388
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.857893707861552,
      "learning_rate": 7.92459747690825e-06,
      "loss": 0.1578,
      "step": 2389
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9246761479352688,
      "learning_rate": 7.92282594914198e-06,
      "loss": 0.17,
      "step": 2390
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.876576562840626,
      "learning_rate": 7.921053863827311e-06,
      "loss": 0.2148,
      "step": 2391
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8330634864455151,
      "learning_rate": 7.919281221302279e-06,
      "loss": 0.1767,
      "step": 2392
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8507800714330225,
      "learning_rate": 7.917508021905033e-06,
      "loss": 0.1756,
      "step": 2393
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9635351455820275,
      "learning_rate": 7.915734265973818e-06,
      "loss": 0.2405,
      "step": 2394
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.995774633850403,
      "learning_rate": 7.913959953846989e-06,
      "loss": 0.1982,
      "step": 2395
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.322627333630704,
      "learning_rate": 7.912185085863009e-06,
      "loss": 0.2548,
      "step": 2396
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0782000776442393,
      "learning_rate": 7.910409662360448e-06,
      "loss": 0.1819,
      "step": 2397
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0835774968155412,
      "learning_rate": 7.908633683677976e-06,
      "loss": 0.2026,
      "step": 2398
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8758797004249126,
      "learning_rate": 7.906857150154377e-06,
      "loss": 0.1612,
      "step": 2399
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1798346223098863,
      "learning_rate": 7.905080062128532e-06,
      "loss": 0.2192,
      "step": 2400
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1152870256172602,
      "learning_rate": 7.903302419939436e-06,
      "loss": 0.242,
      "step": 2401
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0363314680886466,
      "learning_rate": 7.901524223926183e-06,
      "loss": 0.1709,
      "step": 2402
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8856845752633457,
      "learning_rate": 7.899745474427978e-06,
      "loss": 0.1769,
      "step": 2403
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8133988891410937,
      "learning_rate": 7.897966171784128e-06,
      "loss": 0.1444,
      "step": 2404
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.007848332468057,
      "learning_rate": 7.896186316334047e-06,
      "loss": 0.1799,
      "step": 2405
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8777306739989317,
      "learning_rate": 7.894405908417256e-06,
      "loss": 0.1858,
      "step": 2406
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1212644400814125,
      "learning_rate": 7.892624948373375e-06,
      "loss": 0.1715,
      "step": 2407
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1128796950923365,
      "learning_rate": 7.890843436542138e-06,
      "loss": 0.2235,
      "step": 2408
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.30961176536662,
      "learning_rate": 7.88906137326338e-06,
      "loss": 0.2296,
      "step": 2409
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.041836199233003,
      "learning_rate": 7.88727875887704e-06,
      "loss": 0.1866,
      "step": 2410
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.085873013193842,
      "learning_rate": 7.885495593723163e-06,
      "loss": 0.1861,
      "step": 2411
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2958323815465607,
      "learning_rate": 7.8837118781419e-06,
      "loss": 0.2487,
      "step": 2412
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0399667242019999,
      "learning_rate": 7.881927612473509e-06,
      "loss": 0.1882,
      "step": 2413
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7861500464382131,
      "learning_rate": 7.880142797058348e-06,
      "loss": 0.1713,
      "step": 2414
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9544938127528406,
      "learning_rate": 7.87835743223688e-06,
      "loss": 0.1975,
      "step": 2415
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.770325580265094,
      "learning_rate": 7.876571518349682e-06,
      "loss": 0.143,
      "step": 2416
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9653385614901939,
      "learning_rate": 7.874785055737423e-06,
      "loss": 0.168,
      "step": 2417
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.359363866876407,
      "learning_rate": 7.872998044740885e-06,
      "loss": 0.2481,
      "step": 2418
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7732545603607058,
      "learning_rate": 7.871210485700951e-06,
      "loss": 0.1155,
      "step": 2419
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9649554525538463,
      "learning_rate": 7.86942237895861e-06,
      "loss": 0.1834,
      "step": 2420
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8836140144086708,
      "learning_rate": 7.867633724854956e-06,
      "loss": 0.1338,
      "step": 2421
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9654234300512068,
      "learning_rate": 7.865844523731187e-06,
      "loss": 0.178,
      "step": 2422
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.310502499659962,
      "learning_rate": 7.864054775928603e-06,
      "loss": 0.2716,
      "step": 2423
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6593168715655524,
      "learning_rate": 7.862264481788611e-06,
      "loss": 0.1438,
      "step": 2424
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.164295588456566,
      "learning_rate": 7.860473641652724e-06,
      "loss": 0.2481,
      "step": 2425
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2452690442280814,
      "learning_rate": 7.85868225586255e-06,
      "loss": 0.2451,
      "step": 2426
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2856937980412066,
      "learning_rate": 7.856890324759817e-06,
      "loss": 0.2424,
      "step": 2427
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.031847928911392,
      "learning_rate": 7.855097848686341e-06,
      "loss": 0.1917,
      "step": 2428
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1224779322492175,
      "learning_rate": 7.853304827984051e-06,
      "loss": 0.2252,
      "step": 2429
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9316150177559973,
      "learning_rate": 7.851511262994976e-06,
      "loss": 0.1541,
      "step": 2430
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8562839689554356,
      "learning_rate": 7.849717154061251e-06,
      "loss": 0.1818,
      "step": 2431
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9779225566524464,
      "learning_rate": 7.847922501525116e-06,
      "loss": 0.1886,
      "step": 2432
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1815189638810069,
      "learning_rate": 7.846127305728912e-06,
      "loss": 0.258,
      "step": 2433
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8194029884223379,
      "learning_rate": 7.844331567015083e-06,
      "loss": 0.1618,
      "step": 2434
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8702348737614874,
      "learning_rate": 7.842535285726179e-06,
      "loss": 0.1549,
      "step": 2435
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7491022052123658,
      "learning_rate": 7.840738462204853e-06,
      "loss": 0.1481,
      "step": 2436
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9417048621843412,
      "learning_rate": 7.83894109679386e-06,
      "loss": 0.2116,
      "step": 2437
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0390017376825342,
      "learning_rate": 7.837143189836061e-06,
      "loss": 0.2019,
      "step": 2438
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8601324744269842,
      "learning_rate": 7.835344741674415e-06,
      "loss": 0.1431,
      "step": 2439
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1263137880611092,
      "learning_rate": 7.833545752651993e-06,
      "loss": 0.2028,
      "step": 2440
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9948550994010357,
      "learning_rate": 7.83174622311196e-06,
      "loss": 0.1856,
      "step": 2441
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6580242504923062,
      "learning_rate": 7.829946153397592e-06,
      "loss": 0.131,
      "step": 2442
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.180366312068499,
      "learning_rate": 7.82814554385226e-06,
      "loss": 0.1874,
      "step": 2443
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8961627144953566,
      "learning_rate": 7.826344394819443e-06,
      "loss": 0.162,
      "step": 2444
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.072847049721341,
      "learning_rate": 7.824542706642724e-06,
      "loss": 0.2431,
      "step": 2445
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0991997437991172,
      "learning_rate": 7.822740479665786e-06,
      "loss": 0.1866,
      "step": 2446
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7879630480765036,
      "learning_rate": 7.820937714232418e-06,
      "loss": 0.1429,
      "step": 2447
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2823115464355592,
      "learning_rate": 7.819134410686505e-06,
      "loss": 0.2042,
      "step": 2448
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8585351912113849,
      "learning_rate": 7.817330569372043e-06,
      "loss": 0.1615,
      "step": 2449
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9708759188821241,
      "learning_rate": 7.815526190633124e-06,
      "loss": 0.2139,
      "step": 2450
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9014206615267103,
      "learning_rate": 7.813721274813946e-06,
      "loss": 0.2002,
      "step": 2451
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8781584006723137,
      "learning_rate": 7.81191582225881e-06,
      "loss": 0.2308,
      "step": 2452
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2886630193981286,
      "learning_rate": 7.810109833312116e-06,
      "loss": 0.2479,
      "step": 2453
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9824594627943716,
      "learning_rate": 7.808303308318371e-06,
      "loss": 0.2109,
      "step": 2454
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8041269171736151,
      "learning_rate": 7.806496247622177e-06,
      "loss": 0.1653,
      "step": 2455
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0361157348056695,
      "learning_rate": 7.804688651568247e-06,
      "loss": 0.2104,
      "step": 2456
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7954897603283365,
      "learning_rate": 7.80288052050139e-06,
      "loss": 0.1723,
      "step": 2457
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9374887656538512,
      "learning_rate": 7.80107185476652e-06,
      "loss": 0.1753,
      "step": 2458
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9988695309556025,
      "learning_rate": 7.799262654708653e-06,
      "loss": 0.2163,
      "step": 2459
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1854913497685624,
      "learning_rate": 7.797452920672902e-06,
      "loss": 0.1923,
      "step": 2460
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0419563787207728,
      "learning_rate": 7.79564265300449e-06,
      "loss": 0.1895,
      "step": 2461
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9349675776844178,
      "learning_rate": 7.793831852048734e-06,
      "loss": 0.2049,
      "step": 2462
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8004645074821134,
      "learning_rate": 7.79202051815106e-06,
      "loss": 0.1809,
      "step": 2463
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0267446778740834,
      "learning_rate": 7.790208651656987e-06,
      "loss": 0.2442,
      "step": 2464
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1536855438143556,
      "learning_rate": 7.788396252912145e-06,
      "loss": 0.1884,
      "step": 2465
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3533316708917116,
      "learning_rate": 7.78658332226226e-06,
      "loss": 0.2455,
      "step": 2466
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9499356237530023,
      "learning_rate": 7.784769860053159e-06,
      "loss": 0.2064,
      "step": 2467
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47378579721710057,
      "learning_rate": 7.782955866630774e-06,
      "loss": 0.1129,
      "step": 2468
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9835908834264799,
      "learning_rate": 7.781141342341133e-06,
      "loss": 0.1872,
      "step": 2469
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1579940062304819,
      "learning_rate": 7.779326287530373e-06,
      "loss": 0.2701,
      "step": 2470
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9682420808355241,
      "learning_rate": 7.777510702544724e-06,
      "loss": 0.1305,
      "step": 2471
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1596157892822583,
      "learning_rate": 7.775694587730523e-06,
      "loss": 0.2176,
      "step": 2472
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9137857395443876,
      "learning_rate": 7.773877943434205e-06,
      "loss": 0.1829,
      "step": 2473
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9883442688172587,
      "learning_rate": 7.772060770002308e-06,
      "loss": 0.1933,
      "step": 2474
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5980186989299213,
      "learning_rate": 7.770243067781469e-06,
      "loss": 0.1355,
      "step": 2475
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9600509703899313,
      "learning_rate": 7.768424837118424e-06,
      "loss": 0.2112,
      "step": 2476
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9939658712215663,
      "learning_rate": 7.766606078360017e-06,
      "loss": 0.2051,
      "step": 2477
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.0415575628258684,
      "learning_rate": 7.764786791853186e-06,
      "loss": 0.1862,
      "step": 2478
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8762371063755043,
      "learning_rate": 7.762966977944973e-06,
      "loss": 0.1658,
      "step": 2479
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8200253622766234,
      "learning_rate": 7.761146636982519e-06,
      "loss": 0.1752,
      "step": 2480
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9059075406472024,
      "learning_rate": 7.759325769313067e-06,
      "loss": 0.1797,
      "step": 2481
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8094196182599945,
      "learning_rate": 7.757504375283957e-06,
      "loss": 0.1921,
      "step": 2482
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9849032873751608,
      "learning_rate": 7.755682455242633e-06,
      "loss": 0.2204,
      "step": 2483
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1001115290016754,
      "learning_rate": 7.75386000953664e-06,
      "loss": 0.2091,
      "step": 2484
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.205398953900868,
      "learning_rate": 7.752037038513622e-06,
      "loss": 0.2292,
      "step": 2485
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8101228537554984,
      "learning_rate": 7.75021354252132e-06,
      "loss": 0.1969,
      "step": 2486
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8939558279673628,
      "learning_rate": 7.748389521907577e-06,
      "loss": 0.1607,
      "step": 2487
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9562953946297904,
      "learning_rate": 7.746564977020344e-06,
      "loss": 0.2412,
      "step": 2488
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.238919997073506,
      "learning_rate": 7.744739908207656e-06,
      "loss": 0.2395,
      "step": 2489
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1140703436645192,
      "learning_rate": 7.742914315817664e-06,
      "loss": 0.185,
      "step": 2490
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8930378043819944,
      "learning_rate": 7.741088200198607e-06,
      "loss": 0.1884,
      "step": 2491
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0059960383454092,
      "learning_rate": 7.739261561698832e-06,
      "loss": 0.1869,
      "step": 2492
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9719317329139144,
      "learning_rate": 7.73743440066678e-06,
      "loss": 0.222,
      "step": 2493
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9562090029074628,
      "learning_rate": 7.735606717450997e-06,
      "loss": 0.1777,
      "step": 2494
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2855231228690032,
      "learning_rate": 7.73377851240012e-06,
      "loss": 0.2286,
      "step": 2495
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8089886642491668,
      "learning_rate": 7.7319497858629e-06,
      "loss": 0.1379,
      "step": 2496
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8640550098811857,
      "learning_rate": 7.73012053818817e-06,
      "loss": 0.173,
      "step": 2497
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.975382125194293,
      "learning_rate": 7.728290769724876e-06,
      "loss": 0.1813,
      "step": 2498
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.073200586785624,
      "learning_rate": 7.726460480822056e-06,
      "loss": 0.2006,
      "step": 2499
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.999287784824526,
      "learning_rate": 7.724629671828854e-06,
      "loss": 0.1787,
      "step": 2500
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9606467072950612,
      "learning_rate": 7.722798343094505e-06,
      "loss": 0.1795,
      "step": 2501
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1447331574703805,
      "learning_rate": 7.720966494968346e-06,
      "loss": 0.2321,
      "step": 2502
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0243077474869624,
      "learning_rate": 7.719134127799816e-06,
      "loss": 0.2035,
      "step": 2503
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.111762978453211,
      "learning_rate": 7.717301241938454e-06,
      "loss": 0.2222,
      "step": 2504
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7328318129312789,
      "learning_rate": 7.71546783773389e-06,
      "loss": 0.1522,
      "step": 2505
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7357700770475664,
      "learning_rate": 7.713633915535861e-06,
      "loss": 0.1701,
      "step": 2506
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6482438673463664,
      "learning_rate": 7.7117994756942e-06,
      "loss": 0.1475,
      "step": 2507
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7675869469893206,
      "learning_rate": 7.709964518558838e-06,
      "loss": 0.1589,
      "step": 2508
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9246215895707957,
      "learning_rate": 7.708129044479803e-06,
      "loss": 0.2047,
      "step": 2509
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.04740206502982,
      "learning_rate": 7.706293053807228e-06,
      "loss": 0.1984,
      "step": 2510
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0223840910741568,
      "learning_rate": 7.704456546891337e-06,
      "loss": 0.2114,
      "step": 2511
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1702729816678852,
      "learning_rate": 7.702619524082458e-06,
      "loss": 0.2127,
      "step": 2512
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2106842197786745,
      "learning_rate": 7.700781985731013e-06,
      "loss": 0.1856,
      "step": 2513
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.05742884850698,
      "learning_rate": 7.698943932187529e-06,
      "loss": 0.1629,
      "step": 2514
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9543189824954512,
      "learning_rate": 7.69710536380262e-06,
      "loss": 0.1658,
      "step": 2515
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.931006677028311,
      "learning_rate": 7.69526628092701e-06,
      "loss": 0.1789,
      "step": 2516
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9665479026613295,
      "learning_rate": 7.693426683911516e-06,
      "loss": 0.1331,
      "step": 2517
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.209468372287929,
      "learning_rate": 7.691586573107052e-06,
      "loss": 0.2349,
      "step": 2518
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0077037192925118,
      "learning_rate": 7.68974594886463e-06,
      "loss": 0.2031,
      "step": 2519
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8565203666738854,
      "learning_rate": 7.687904811535364e-06,
      "loss": 0.1976,
      "step": 2520
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.100673202083589,
      "learning_rate": 7.686063161470463e-06,
      "loss": 0.2424,
      "step": 2521
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9843797711327092,
      "learning_rate": 7.684220999021234e-06,
      "loss": 0.1626,
      "step": 2522
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0694359829884565,
      "learning_rate": 7.682378324539079e-06,
      "loss": 0.2256,
      "step": 2523
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7594254125221145,
      "learning_rate": 7.680535138375503e-06,
      "loss": 0.1359,
      "step": 2524
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9914422327557196,
      "learning_rate": 7.678691440882104e-06,
      "loss": 0.1795,
      "step": 2525
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9188916301211716,
      "learning_rate": 7.676847232410582e-06,
      "loss": 0.2012,
      "step": 2526
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0503455759255365,
      "learning_rate": 7.675002513312727e-06,
      "loss": 0.1516,
      "step": 2527
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0184026136579512,
      "learning_rate": 7.673157283940438e-06,
      "loss": 0.1897,
      "step": 2528
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.240114721121259,
      "learning_rate": 7.671311544645703e-06,
      "loss": 0.2263,
      "step": 2529
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0715436494192168,
      "learning_rate": 7.669465295780606e-06,
      "loss": 0.2006,
      "step": 2530
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.277136500757425,
      "learning_rate": 7.667618537697332e-06,
      "loss": 0.265,
      "step": 2531
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9510023216211945,
      "learning_rate": 7.665771270748164e-06,
      "loss": 0.1278,
      "step": 2532
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6893838613359251,
      "learning_rate": 7.66392349528548e-06,
      "loss": 0.1868,
      "step": 2533
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1528883679483053,
      "learning_rate": 7.662075211661754e-06,
      "loss": 0.2378,
      "step": 2534
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9263070456631116,
      "learning_rate": 7.66022642022956e-06,
      "loss": 0.183,
      "step": 2535
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9432598708294494,
      "learning_rate": 7.658377121341564e-06,
      "loss": 0.1496,
      "step": 2536
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.07269210671933,
      "learning_rate": 7.656527315350537e-06,
      "loss": 0.211,
      "step": 2537
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9576703157438666,
      "learning_rate": 7.65467700260934e-06,
      "loss": 0.1944,
      "step": 2538
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8313235216984847,
      "learning_rate": 7.65282618347093e-06,
      "loss": 0.1741,
      "step": 2539
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6107254389504282,
      "learning_rate": 7.650974858288364e-06,
      "loss": 0.0949,
      "step": 2540
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.876466824418133,
      "learning_rate": 7.649123027414798e-06,
      "loss": 0.1882,
      "step": 2541
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.331145621185843,
      "learning_rate": 7.647270691203476e-06,
      "loss": 0.2622,
      "step": 2542
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8132225013980352,
      "learning_rate": 7.645417850007745e-06,
      "loss": 0.1595,
      "step": 2543
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.194206073998705,
      "learning_rate": 7.643564504181048e-06,
      "loss": 0.2148,
      "step": 2544
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0384158068720195,
      "learning_rate": 7.641710654076923e-06,
      "loss": 0.1786,
      "step": 2545
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0498450185285226,
      "learning_rate": 7.639856300049003e-06,
      "loss": 0.1814,
      "step": 2546
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9108888163836882,
      "learning_rate": 7.638001442451019e-06,
      "loss": 0.1616,
      "step": 2547
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.876002484252067,
      "learning_rate": 7.636146081636797e-06,
      "loss": 0.1753,
      "step": 2548
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9401535284846605,
      "learning_rate": 7.634290217960258e-06,
      "loss": 0.1383,
      "step": 2549
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9151891918955153,
      "learning_rate": 7.632433851775422e-06,
      "loss": 0.2064,
      "step": 2550
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8338386369618562,
      "learning_rate": 7.630576983436404e-06,
      "loss": 0.1826,
      "step": 2551
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8272508605159983,
      "learning_rate": 7.628719613297414e-06,
      "loss": 0.1644,
      "step": 2552
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7178547010180232,
      "learning_rate": 7.626861741712755e-06,
      "loss": 0.1957,
      "step": 2553
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0128268462691266,
      "learning_rate": 7.625003369036832e-06,
      "loss": 0.1944,
      "step": 2554
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1889103136996255,
      "learning_rate": 7.623144495624136e-06,
      "loss": 0.2347,
      "step": 2555
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8464214105221063,
      "learning_rate": 7.621285121829265e-06,
      "loss": 0.1684,
      "step": 2556
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0335943302274293,
      "learning_rate": 7.6194252480069045e-06,
      "loss": 0.184,
      "step": 2557
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.98082314055913,
      "learning_rate": 7.617564874511838e-06,
      "loss": 0.1962,
      "step": 2558
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0580456606074862,
      "learning_rate": 7.615704001698945e-06,
      "loss": 0.2451,
      "step": 2559
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1633424605025564,
      "learning_rate": 7.613842629923198e-06,
      "loss": 0.2612,
      "step": 2560
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0140774725516295,
      "learning_rate": 7.611980759539668e-06,
      "loss": 0.1979,
      "step": 2561
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0550838689990871,
      "learning_rate": 7.610118390903516e-06,
      "loss": 0.1756,
      "step": 2562
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7981994426478345,
      "learning_rate": 7.608255524370003e-06,
      "loss": 0.1762,
      "step": 2563
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9620810987387195,
      "learning_rate": 7.606392160294482e-06,
      "loss": 0.1868,
      "step": 2564
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7018145732163338,
      "learning_rate": 7.604528299032404e-06,
      "loss": 0.1336,
      "step": 2565
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8391379780983679,
      "learning_rate": 7.602663940939311e-06,
      "loss": 0.182,
      "step": 2566
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.868863975097558,
      "learning_rate": 7.6007990863708425e-06,
      "loss": 0.1623,
      "step": 2567
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9993743731898556,
      "learning_rate": 7.598933735682734e-06,
      "loss": 0.1897,
      "step": 2568
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9438514593779093,
      "learning_rate": 7.59706788923081e-06,
      "loss": 0.1729,
      "step": 2569
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8399660724030051,
      "learning_rate": 7.595201547370994e-06,
      "loss": 0.1697,
      "step": 2570
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.047928184905739,
      "learning_rate": 7.593334710459303e-06,
      "loss": 0.2097,
      "step": 2571
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9105123551228184,
      "learning_rate": 7.591467378851849e-06,
      "loss": 0.1166,
      "step": 2572
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.115158614275299,
      "learning_rate": 7.589599552904838e-06,
      "loss": 0.1892,
      "step": 2573
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0174802669537286,
      "learning_rate": 7.5877312329745716e-06,
      "loss": 0.2179,
      "step": 2574
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5441411414479186,
      "learning_rate": 7.5858624194174405e-06,
      "loss": 0.1544,
      "step": 2575
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.180834028711788,
      "learning_rate": 7.583993112589936e-06,
      "loss": 0.254,
      "step": 2576
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9386216527378254,
      "learning_rate": 7.58212331284864e-06,
      "loss": 0.1998,
      "step": 2577
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8712095677125203,
      "learning_rate": 7.58025302055023e-06,
      "loss": 0.1754,
      "step": 2578
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0638647082924035,
      "learning_rate": 7.578382236051475e-06,
      "loss": 0.2037,
      "step": 2579
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1141024608596102,
      "learning_rate": 7.57651095970924e-06,
      "loss": 0.193,
      "step": 2580
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0274854679701937,
      "learning_rate": 7.574639191880484e-06,
      "loss": 0.1559,
      "step": 2581
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9601825319530732,
      "learning_rate": 7.57276693292226e-06,
      "loss": 0.1929,
      "step": 2582
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9691596642201638,
      "learning_rate": 7.570894183191712e-06,
      "loss": 0.1762,
      "step": 2583
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.178211482974723,
      "learning_rate": 7.5690209430460796e-06,
      "loss": 0.2046,
      "step": 2584
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9295834643493733,
      "learning_rate": 7.567147212842696e-06,
      "loss": 0.2085,
      "step": 2585
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1300474099619744,
      "learning_rate": 7.565272992938988e-06,
      "loss": 0.1832,
      "step": 2586
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1065597797776203,
      "learning_rate": 7.563398283692476e-06,
      "loss": 0.2165,
      "step": 2587
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0194839714007786,
      "learning_rate": 7.561523085460772e-06,
      "loss": 0.2294,
      "step": 2588
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4386574438949562,
      "learning_rate": 7.559647398601583e-06,
      "loss": 0.2714,
      "step": 2589
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9339560502056736,
      "learning_rate": 7.557771223472708e-06,
      "loss": 0.1884,
      "step": 2590
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9678356966229497,
      "learning_rate": 7.555894560432042e-06,
      "loss": 0.2285,
      "step": 2591
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9143553629843998,
      "learning_rate": 7.554017409837568e-06,
      "loss": 0.1906,
      "step": 2592
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9094212983032927,
      "learning_rate": 7.5521397720473686e-06,
      "loss": 0.1748,
      "step": 2593
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0366436712368505,
      "learning_rate": 7.55026164741961e-06,
      "loss": 0.194,
      "step": 2594
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.040735690519662,
      "learning_rate": 7.548383036312562e-06,
      "loss": 0.1928,
      "step": 2595
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8730764345811022,
      "learning_rate": 7.546503939084579e-06,
      "loss": 0.1573,
      "step": 2596
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9159069061819339,
      "learning_rate": 7.544624356094114e-06,
      "loss": 0.1638,
      "step": 2597
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9789703359615398,
      "learning_rate": 7.542744287699709e-06,
      "loss": 0.203,
      "step": 2598
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9084077706314987,
      "learning_rate": 7.5408637342599985e-06,
      "loss": 0.1921,
      "step": 2599
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1678234074291303,
      "learning_rate": 7.538982696133711e-06,
      "loss": 0.2358,
      "step": 2600
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8439686827284687,
      "learning_rate": 7.537101173679666e-06,
      "loss": 0.1361,
      "step": 2601
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.892220325827293,
      "learning_rate": 7.535219167256777e-06,
      "loss": 0.1863,
      "step": 2602
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.085641616118169,
      "learning_rate": 7.533336677224052e-06,
      "loss": 0.2005,
      "step": 2603
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7900143796401147,
      "learning_rate": 7.5314537039405845e-06,
      "loss": 0.156,
      "step": 2604
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8510175523316104,
      "learning_rate": 7.5295702477655645e-06,
      "loss": 0.1729,
      "step": 2605
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0385889819201126,
      "learning_rate": 7.527686309058275e-06,
      "loss": 0.2239,
      "step": 2606
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9185186059073092,
      "learning_rate": 7.525801888178092e-06,
      "loss": 0.1697,
      "step": 2607
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0684996563701876,
      "learning_rate": 7.523916985484476e-06,
      "loss": 0.2369,
      "step": 2608
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0601243042219135,
      "learning_rate": 7.522031601336989e-06,
      "loss": 0.1807,
      "step": 2609
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7682957881539859,
      "learning_rate": 7.520145736095279e-06,
      "loss": 0.1309,
      "step": 2610
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1431996573673888,
      "learning_rate": 7.518259390119086e-06,
      "loss": 0.1952,
      "step": 2611
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1547805498478023,
      "learning_rate": 7.516372563768245e-06,
      "loss": 0.2543,
      "step": 2612
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9550897761116337,
      "learning_rate": 7.5144852574026795e-06,
      "loss": 0.1711,
      "step": 2613
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1845490838761463,
      "learning_rate": 7.512597471382407e-06,
      "loss": 0.2194,
      "step": 2614
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2203786834387687,
      "learning_rate": 7.5107092060675344e-06,
      "loss": 0.2548,
      "step": 2615
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8838482905758169,
      "learning_rate": 7.5088204618182605e-06,
      "loss": 0.179,
      "step": 2616
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1673088095845232,
      "learning_rate": 7.506931238994874e-06,
      "loss": 0.2095,
      "step": 2617
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8705596296202388,
      "learning_rate": 7.50504153795776e-06,
      "loss": 0.1546,
      "step": 2618
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1176213864981543,
      "learning_rate": 7.503151359067391e-06,
      "loss": 0.2487,
      "step": 2619
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8976275374931851,
      "learning_rate": 7.5012607026843285e-06,
      "loss": 0.1284,
      "step": 2620
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1229061196841248,
      "learning_rate": 7.49936956916923e-06,
      "loss": 0.2343,
      "step": 2621
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0080071042698007,
      "learning_rate": 7.497477958882842e-06,
      "loss": 0.1977,
      "step": 2622
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0956943288520335,
      "learning_rate": 7.495585872185999e-06,
      "loss": 0.1962,
      "step": 2623
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5996312025599517,
      "learning_rate": 7.493693309439632e-06,
      "loss": 0.1823,
      "step": 2624
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8617816885565124,
      "learning_rate": 7.49180027100476e-06,
      "loss": 0.1205,
      "step": 2625
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.072888112315538,
      "learning_rate": 7.48990675724249e-06,
      "loss": 0.2404,
      "step": 2626
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1813452293401616,
      "learning_rate": 7.4880127685140235e-06,
      "loss": 0.2427,
      "step": 2627
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9918270552104747,
      "learning_rate": 7.486118305180651e-06,
      "loss": 0.1963,
      "step": 2628
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0121884147973859,
      "learning_rate": 7.484223367603756e-06,
      "loss": 0.18,
      "step": 2629
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.940394237770484,
      "learning_rate": 7.4823279561448084e-06,
      "loss": 0.1772,
      "step": 2630
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1976762776029029,
      "learning_rate": 7.480432071165371e-06,
      "loss": 0.2134,
      "step": 2631
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8629697001646478,
      "learning_rate": 7.478535713027096e-06,
      "loss": 0.1702,
      "step": 2632
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.937806260569262,
      "learning_rate": 7.476638882091726e-06,
      "loss": 0.2156,
      "step": 2633
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0177175766746294,
      "learning_rate": 7.474741578721096e-06,
      "loss": 0.2663,
      "step": 2634
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9719514674203783,
      "learning_rate": 7.472843803277127e-06,
      "loss": 0.1967,
      "step": 2635
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.988372687042928,
      "learning_rate": 7.4709455561218326e-06,
      "loss": 0.216,
      "step": 2636
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0427128649289727,
      "learning_rate": 7.469046837617317e-06,
      "loss": 0.1827,
      "step": 2637
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0299670364643276,
      "learning_rate": 7.4671476481257735e-06,
      "loss": 0.1781,
      "step": 2638
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9359275926224101,
      "learning_rate": 7.465247988009483e-06,
      "loss": 0.1782,
      "step": 2639
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8590425076333492,
      "learning_rate": 7.4633478576308215e-06,
      "loss": 0.1356,
      "step": 2640
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0392574998344555,
      "learning_rate": 7.461447257352248e-06,
      "loss": 0.2054,
      "step": 2641
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6095565101615245,
      "learning_rate": 7.459546187536317e-06,
      "loss": 0.1642,
      "step": 2642
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8665951128038286,
      "learning_rate": 7.4576446485456685e-06,
      "loss": 0.1842,
      "step": 2643
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9310647190406239,
      "learning_rate": 7.455742640743035e-06,
      "loss": 0.2072,
      "step": 2644
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7071920475707153,
      "learning_rate": 7.453840164491237e-06,
      "loss": 0.1623,
      "step": 2645
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7085179005563707,
      "learning_rate": 7.451937220153184e-06,
      "loss": 0.1002,
      "step": 2646
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9480402329426247,
      "learning_rate": 7.450033808091875e-06,
      "loss": 0.2103,
      "step": 2647
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.189882415968913,
      "learning_rate": 7.448129928670398e-06,
      "loss": 0.2023,
      "step": 2648
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1590569803886348,
      "learning_rate": 7.446225582251931e-06,
      "loss": 0.2547,
      "step": 2649
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9269129268625055,
      "learning_rate": 7.444320769199741e-06,
      "loss": 0.2276,
      "step": 2650
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9200451653064249,
      "learning_rate": 7.442415489877185e-06,
      "loss": 0.1465,
      "step": 2651
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9142825964743402,
      "learning_rate": 7.440509744647705e-06,
      "loss": 0.1617,
      "step": 2652
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1524469593005653,
      "learning_rate": 7.438603533874838e-06,
      "loss": 0.2208,
      "step": 2653
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7951901338314104,
      "learning_rate": 7.436696857922202e-06,
      "loss": 0.1382,
      "step": 2654
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8366729372350902,
      "learning_rate": 7.434789717153511e-06,
      "loss": 0.1449,
      "step": 2655
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8451299783316879,
      "learning_rate": 7.4328821119325626e-06,
      "loss": 0.1756,
      "step": 2656
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.215529346473779,
      "learning_rate": 7.430974042623247e-06,
      "loss": 0.2188,
      "step": 2657
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8145132734898775,
      "learning_rate": 7.429065509589542e-06,
      "loss": 0.1312,
      "step": 2658
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0450251483880602,
      "learning_rate": 7.4271565131955095e-06,
      "loss": 0.2146,
      "step": 2659
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1057627756843322,
      "learning_rate": 7.4252470538053055e-06,
      "loss": 0.2056,
      "step": 2660
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9327537691834462,
      "learning_rate": 7.4233371317831724e-06,
      "loss": 0.1889,
      "step": 2661
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0580558438645953,
      "learning_rate": 7.42142674749344e-06,
      "loss": 0.1799,
      "step": 2662
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9108535474492828,
      "learning_rate": 7.419515901300525e-06,
      "loss": 0.1745,
      "step": 2663
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.011387211307849,
      "learning_rate": 7.417604593568937e-06,
      "loss": 0.1925,
      "step": 2664
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9297701707330855,
      "learning_rate": 7.415692824663268e-06,
      "loss": 0.1782,
      "step": 2665
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8586044735486273,
      "learning_rate": 7.4137805949482015e-06,
      "loss": 0.1751,
      "step": 2666
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7387672942194006,
      "learning_rate": 7.411867904788508e-06,
      "loss": 0.1029,
      "step": 2667
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5927426200859575,
      "learning_rate": 7.409954754549047e-06,
      "loss": 0.1503,
      "step": 2668
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2459352474757537,
      "learning_rate": 7.408041144594762e-06,
      "loss": 0.2391,
      "step": 2669
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.003457646383541,
      "learning_rate": 7.406127075290689e-06,
      "loss": 0.1548,
      "step": 2670
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9950229489911964,
      "learning_rate": 7.404212547001947e-06,
      "loss": 0.1806,
      "step": 2671
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1000395128001494,
      "learning_rate": 7.4022975600937495e-06,
      "loss": 0.2231,
      "step": 2672
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9348175619357479,
      "learning_rate": 7.400382114931387e-06,
      "loss": 0.1921,
      "step": 2673
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9483779726318815,
      "learning_rate": 7.3984662118802476e-06,
      "loss": 0.1815,
      "step": 2674
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8079023278176141,
      "learning_rate": 7.396549851305801e-06,
      "loss": 0.1832,
      "step": 2675
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.829226325825035,
      "learning_rate": 7.394633033573606e-06,
      "loss": 0.0962,
      "step": 2676
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8811599127993811,
      "learning_rate": 7.39271575904931e-06,
      "loss": 0.1639,
      "step": 2677
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0218311878295687,
      "learning_rate": 7.390798028098642e-06,
      "loss": 0.1972,
      "step": 2678
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9791714054033586,
      "learning_rate": 7.388879841087423e-06,
      "loss": 0.2034,
      "step": 2679
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.052687023791769,
      "learning_rate": 7.386961198381561e-06,
      "loss": 0.1702,
      "step": 2680
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2486607113159174,
      "learning_rate": 7.385042100347051e-06,
      "loss": 0.2177,
      "step": 2681
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8972531995035304,
      "learning_rate": 7.383122547349971e-06,
      "loss": 0.213,
      "step": 2682
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0894071882315572,
      "learning_rate": 7.38120253975649e-06,
      "loss": 0.238,
      "step": 2683
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0654505217830808,
      "learning_rate": 7.379282077932861e-06,
      "loss": 0.209,
      "step": 2684
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2011635132466219,
      "learning_rate": 7.377361162245426e-06,
      "loss": 0.1969,
      "step": 2685
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8927070653561099,
      "learning_rate": 7.375439793060612e-06,
      "loss": 0.2034,
      "step": 2686
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2019835001932984,
      "learning_rate": 7.3735179707449345e-06,
      "loss": 0.2547,
      "step": 2687
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0678120136642224,
      "learning_rate": 7.37159569566499e-06,
      "loss": 0.165,
      "step": 2688
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2629334168732944,
      "learning_rate": 7.369672968187467e-06,
      "loss": 0.2776,
      "step": 2689
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0780240414125095,
      "learning_rate": 7.3677497886791395e-06,
      "loss": 0.1925,
      "step": 2690
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9636990177683012,
      "learning_rate": 7.365826157506867e-06,
      "loss": 0.1784,
      "step": 2691
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9938288531534171,
      "learning_rate": 7.363902075037593e-06,
      "loss": 0.193,
      "step": 2692
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6085504516766802,
      "learning_rate": 7.361977541638351e-06,
      "loss": 0.1737,
      "step": 2693
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0031558787978274,
      "learning_rate": 7.360052557676256e-06,
      "loss": 0.1642,
      "step": 2694
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.3160336370116197,
      "learning_rate": 7.358127123518512e-06,
      "loss": 0.2513,
      "step": 2695
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8763638163722971,
      "learning_rate": 7.356201239532411e-06,
      "loss": 0.1858,
      "step": 2696
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8460616937347156,
      "learning_rate": 7.354274906085326e-06,
      "loss": 0.1826,
      "step": 2697
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2492336016752206,
      "learning_rate": 7.352348123544718e-06,
      "loss": 0.2221,
      "step": 2698
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9340956185363404,
      "learning_rate": 7.350420892278134e-06,
      "loss": 0.1933,
      "step": 2699
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8012318036069901,
      "learning_rate": 7.348493212653205e-06,
      "loss": 0.1762,
      "step": 2700
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1329743225676785,
      "learning_rate": 7.346565085037649e-06,
      "loss": 0.2316,
      "step": 2701
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8713737314225705,
      "learning_rate": 7.34463650979927e-06,
      "loss": 0.195,
      "step": 2702
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.215901840007839,
      "learning_rate": 7.3427074873059565e-06,
      "loss": 0.1884,
      "step": 2703
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.043664831776346,
      "learning_rate": 7.340778017925682e-06,
      "loss": 0.2306,
      "step": 2704
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0659591615573485,
      "learning_rate": 7.338848102026505e-06,
      "loss": 0.2442,
      "step": 2705
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8956181104476681,
      "learning_rate": 7.336917739976569e-06,
      "loss": 0.1579,
      "step": 2706
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1084214566570678,
      "learning_rate": 7.334986932144106e-06,
      "loss": 0.2179,
      "step": 2707
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9027785613774786,
      "learning_rate": 7.333055678897427e-06,
      "loss": 0.161,
      "step": 2708
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9563652116865602,
      "learning_rate": 7.331123980604935e-06,
      "loss": 0.1697,
      "step": 2709
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9383880512529885,
      "learning_rate": 7.329191837635112e-06,
      "loss": 0.2144,
      "step": 2710
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9010243625838051,
      "learning_rate": 7.327259250356527e-06,
      "loss": 0.192,
      "step": 2711
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.173620739140971,
      "learning_rate": 7.325326219137834e-06,
      "loss": 0.2282,
      "step": 2712
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0881105950119698,
      "learning_rate": 7.323392744347772e-06,
      "loss": 0.2031,
      "step": 2713
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5319768162693915,
      "learning_rate": 7.321458826355165e-06,
      "loss": 0.1093,
      "step": 2714
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0326215125686249,
      "learning_rate": 7.319524465528919e-06,
      "loss": 0.2107,
      "step": 2715
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8686601073940099,
      "learning_rate": 7.3175896622380264e-06,
      "loss": 0.2022,
      "step": 2716
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0046749804651958,
      "learning_rate": 7.315654416851563e-06,
      "loss": 0.1689,
      "step": 2717
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1597134738240868,
      "learning_rate": 7.313718729738691e-06,
      "loss": 0.2287,
      "step": 2718
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0039250971401603,
      "learning_rate": 7.311782601268657e-06,
      "loss": 0.1987,
      "step": 2719
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8185078175876662,
      "learning_rate": 7.309846031810786e-06,
      "loss": 0.1778,
      "step": 2720
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1540555545885667,
      "learning_rate": 7.3079090217344965e-06,
      "loss": 0.1929,
      "step": 2721
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7936481355298155,
      "learning_rate": 7.305971571409282e-06,
      "loss": 0.1548,
      "step": 2722
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0821870267562106,
      "learning_rate": 7.3040336812047266e-06,
      "loss": 0.1578,
      "step": 2723
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6118649227202061,
      "learning_rate": 7.302095351490494e-06,
      "loss": 0.1376,
      "step": 2724
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0865644822533227,
      "learning_rate": 7.300156582636333e-06,
      "loss": 0.1773,
      "step": 2725
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8664032367578841,
      "learning_rate": 7.298217375012078e-06,
      "loss": 0.196,
      "step": 2726
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0083995878762835,
      "learning_rate": 7.296277728987646e-06,
      "loss": 0.1936,
      "step": 2727
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8180917164235116,
      "learning_rate": 7.294337644933036e-06,
      "loss": 0.1532,
      "step": 2728
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7553023933183168,
      "learning_rate": 7.292397123218331e-06,
      "loss": 0.1336,
      "step": 2729
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8630473947228512,
      "learning_rate": 7.290456164213703e-06,
      "loss": 0.1402,
      "step": 2730
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0236945597128244,
      "learning_rate": 7.288514768289399e-06,
      "loss": 0.2022,
      "step": 2731
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9376513467265303,
      "learning_rate": 7.2865729358157525e-06,
      "loss": 0.1398,
      "step": 2732
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8698204350713303,
      "learning_rate": 7.284630667163182e-06,
      "loss": 0.1755,
      "step": 2733
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2695084710492324,
      "learning_rate": 7.282687962702186e-06,
      "loss": 0.2712,
      "step": 2734
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9791852103069936,
      "learning_rate": 7.280744822803354e-06,
      "loss": 0.1649,
      "step": 2735
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0470760898277545,
      "learning_rate": 7.2788012478373485e-06,
      "loss": 0.1629,
      "step": 2736
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9877600544879925,
      "learning_rate": 7.276857238174919e-06,
      "loss": 0.2083,
      "step": 2737
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.117237266092997,
      "learning_rate": 7.274912794186901e-06,
      "loss": 0.231,
      "step": 2738
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0488783028268864,
      "learning_rate": 7.272967916244207e-06,
      "loss": 0.2339,
      "step": 2739
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9629223250136234,
      "learning_rate": 7.2710226047178365e-06,
      "loss": 0.1412,
      "step": 2740
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.138001985788664,
      "learning_rate": 7.269076859978871e-06,
      "loss": 0.1741,
      "step": 2741
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.064022334743869,
      "learning_rate": 7.2671306823984735e-06,
      "loss": 0.2137,
      "step": 2742
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1860953225508009,
      "learning_rate": 7.265184072347892e-06,
      "loss": 0.2244,
      "step": 2743
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.922972681029589,
      "learning_rate": 7.263237030198452e-06,
      "loss": 0.1954,
      "step": 2744
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.853011478695407,
      "learning_rate": 7.261289556321568e-06,
      "loss": 0.1733,
      "step": 2745
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5584877527973753,
      "learning_rate": 7.2593416510887315e-06,
      "loss": 0.061,
      "step": 2746
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1782317943747938,
      "learning_rate": 7.257393314871519e-06,
      "loss": 0.2158,
      "step": 2747
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0486316731373484,
      "learning_rate": 7.255444548041589e-06,
      "loss": 0.2071,
      "step": 2748
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0880253570475447,
      "learning_rate": 7.253495350970681e-06,
      "loss": 0.2317,
      "step": 2749
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8652014969424182,
      "learning_rate": 7.251545724030616e-06,
      "loss": 0.1329,
      "step": 2750
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0012366374789825,
      "learning_rate": 7.249595667593301e-06,
      "loss": 0.1917,
      "step": 2751
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0347779136617856,
      "learning_rate": 7.247645182030721e-06,
      "loss": 0.1979,
      "step": 2752
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8519484463376209,
      "learning_rate": 7.245694267714944e-06,
      "loss": 0.1688,
      "step": 2753
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.888454717560395,
      "learning_rate": 7.2437429250181176e-06,
      "loss": 0.2073,
      "step": 2754
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.919418845263843,
      "learning_rate": 7.241791154312478e-06,
      "loss": 0.2175,
      "step": 2755
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2105743836358647,
      "learning_rate": 7.239838955970334e-06,
      "loss": 0.2471,
      "step": 2756
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9854519528984398,
      "learning_rate": 7.237886330364081e-06,
      "loss": 0.163,
      "step": 2757
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8434060377395088,
      "learning_rate": 7.235933277866199e-06,
      "loss": 0.2062,
      "step": 2758
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.801148253053177,
      "learning_rate": 7.233979798849242e-06,
      "loss": 0.2152,
      "step": 2759
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2377591861193946,
      "learning_rate": 7.23202589368585e-06,
      "loss": 0.2283,
      "step": 2760
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0436460856549974,
      "learning_rate": 7.230071562748744e-06,
      "loss": 0.2089,
      "step": 2761
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9751278395948033,
      "learning_rate": 7.228116806410724e-06,
      "loss": 0.1838,
      "step": 2762
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8676482934145258,
      "learning_rate": 7.2261616250446755e-06,
      "loss": 0.1723,
      "step": 2763
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1258572357180048,
      "learning_rate": 7.2242060190235595e-06,
      "loss": 0.2306,
      "step": 2764
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8403180800865869,
      "learning_rate": 7.222249988720422e-06,
      "loss": 0.149,
      "step": 2765
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.798826997520314,
      "learning_rate": 7.220293534508388e-06,
      "loss": 0.1623,
      "step": 2766
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7741823776863653,
      "learning_rate": 7.218336656760665e-06,
      "loss": 0.1718,
      "step": 2767
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.134801839652925,
      "learning_rate": 7.21637935585054e-06,
      "loss": 0.2516,
      "step": 2768
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9307998581259873,
      "learning_rate": 7.214421632151381e-06,
      "loss": 0.182,
      "step": 2769
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9005250334228009,
      "learning_rate": 7.212463486036638e-06,
      "loss": 0.189,
      "step": 2770
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7913732667796961,
      "learning_rate": 7.210504917879838e-06,
      "loss": 0.1202,
      "step": 2771
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0188081260976039,
      "learning_rate": 7.208545928054593e-06,
      "loss": 0.2049,
      "step": 2772
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.856023487802795,
      "learning_rate": 7.206586516934592e-06,
      "loss": 0.1718,
      "step": 2773
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7758001558136339,
      "learning_rate": 7.204626684893604e-06,
      "loss": 0.1609,
      "step": 2774
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8878690489166294,
      "learning_rate": 7.202666432305484e-06,
      "loss": 0.1956,
      "step": 2775
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1032415434084213,
      "learning_rate": 7.200705759544161e-06,
      "loss": 0.1775,
      "step": 2776
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.074081021756156,
      "learning_rate": 7.198744666983647e-06,
      "loss": 0.2409,
      "step": 2777
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7767089501299029,
      "learning_rate": 7.196783154998033e-06,
      "loss": 0.1204,
      "step": 2778
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9086291781981162,
      "learning_rate": 7.194821223961488e-06,
      "loss": 0.1961,
      "step": 2779
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0298667781206245,
      "learning_rate": 7.192858874248266e-06,
      "loss": 0.1673,
      "step": 2780
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9351582019539361,
      "learning_rate": 7.190896106232699e-06,
      "loss": 0.1979,
      "step": 2781
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0890907562292997,
      "learning_rate": 7.1889329202891956e-06,
      "loss": 0.2229,
      "step": 2782
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7587788165515045,
      "learning_rate": 7.186969316792249e-06,
      "loss": 0.1987,
      "step": 2783
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0255569032810912,
      "learning_rate": 7.185005296116426e-06,
      "loss": 0.2279,
      "step": 2784
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9858384904547729,
      "learning_rate": 7.183040858636381e-06,
      "loss": 0.2104,
      "step": 2785
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8846652280864897,
      "learning_rate": 7.181076004726839e-06,
      "loss": 0.1741,
      "step": 2786
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.556424178365562,
      "learning_rate": 7.179110734762613e-06,
      "loss": 0.1478,
      "step": 2787
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0168726503288958,
      "learning_rate": 7.177145049118588e-06,
      "loss": 0.1704,
      "step": 2788
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.825857242061566,
      "learning_rate": 7.175178948169734e-06,
      "loss": 0.1472,
      "step": 2789
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0014680941382086,
      "learning_rate": 7.173212432291094e-06,
      "loss": 0.1691,
      "step": 2790
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7998621605864867,
      "learning_rate": 7.171245501857799e-06,
      "loss": 0.1672,
      "step": 2791
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9210752601488422,
      "learning_rate": 7.169278157245052e-06,
      "loss": 0.1598,
      "step": 2792
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.984752022541704,
      "learning_rate": 7.167310398828137e-06,
      "loss": 0.1862,
      "step": 2793
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.086644149445141,
      "learning_rate": 7.1653422269824145e-06,
      "loss": 0.1649,
      "step": 2794
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1799648414837443,
      "learning_rate": 7.163373642083329e-06,
      "loss": 0.2698,
      "step": 2795
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7729011761438327,
      "learning_rate": 7.1614046445064e-06,
      "loss": 0.1326,
      "step": 2796
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0821751389848597,
      "learning_rate": 7.159435234627228e-06,
      "loss": 0.1795,
      "step": 2797
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8317269274413615,
      "learning_rate": 7.157465412821489e-06,
      "loss": 0.1404,
      "step": 2798
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6391754175153338,
      "learning_rate": 7.155495179464942e-06,
      "loss": 0.1097,
      "step": 2799
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1180526082457025,
      "learning_rate": 7.153524534933422e-06,
      "loss": 0.1954,
      "step": 2800
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0744716088156947,
      "learning_rate": 7.15155347960284e-06,
      "loss": 0.1868,
      "step": 2801
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9555889436631148,
      "learning_rate": 7.149582013849188e-06,
      "loss": 0.1825,
      "step": 2802
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.12863015007667,
      "learning_rate": 7.1476101380485385e-06,
      "loss": 0.2075,
      "step": 2803
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9877433234813672,
      "learning_rate": 7.145637852577039e-06,
      "loss": 0.2038,
      "step": 2804
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.778052805408713,
      "learning_rate": 7.143665157810915e-06,
      "loss": 0.1217,
      "step": 2805
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0982075865548544,
      "learning_rate": 7.141692054126471e-06,
      "loss": 0.1939,
      "step": 2806
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9504200155091456,
      "learning_rate": 7.139718541900092e-06,
      "loss": 0.1934,
      "step": 2807
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7898779068252555,
      "learning_rate": 7.137744621508236e-06,
      "loss": 0.1518,
      "step": 2808
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0224221739190253,
      "learning_rate": 7.135770293327442e-06,
      "loss": 0.211,
      "step": 2809
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8099584618949003,
      "learning_rate": 7.133795557734328e-06,
      "loss": 0.1414,
      "step": 2810
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7658183556844463,
      "learning_rate": 7.1318204151055845e-06,
      "loss": 0.1249,
      "step": 2811
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.027533614789803,
      "learning_rate": 7.129844865817985e-06,
      "loss": 0.1396,
      "step": 2812
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9030160185312495,
      "learning_rate": 7.127868910248379e-06,
      "loss": 0.176,
      "step": 2813
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.051055711081802,
      "learning_rate": 7.125892548773692e-06,
      "loss": 0.1961,
      "step": 2814
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6462817239216077,
      "learning_rate": 7.1239157817709305e-06,
      "loss": 0.1369,
      "step": 2815
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9505198612567022,
      "learning_rate": 7.121938609617172e-06,
      "loss": 0.1583,
      "step": 2816
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6790097296434044,
      "learning_rate": 7.119961032689578e-06,
      "loss": 0.1118,
      "step": 2817
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0541508734680638,
      "learning_rate": 7.117983051365383e-06,
      "loss": 0.2218,
      "step": 2818
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1762786719431721,
      "learning_rate": 7.116004666021901e-06,
      "loss": 0.216,
      "step": 2819
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.989249192189872,
      "learning_rate": 7.114025877036521e-06,
      "loss": 0.1701,
      "step": 2820
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9328224224485978,
      "learning_rate": 7.112046684786711e-06,
      "loss": 0.1803,
      "step": 2821
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.993176370718209,
      "learning_rate": 7.110067089650014e-06,
      "loss": 0.2287,
      "step": 2822
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0185224442678311,
      "learning_rate": 7.108087092004055e-06,
      "loss": 0.2133,
      "step": 2823
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1139947017788174,
      "learning_rate": 7.106106692226524e-06,
      "loss": 0.2159,
      "step": 2824
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9044509267413214,
      "learning_rate": 7.104125890695202e-06,
      "loss": 0.1839,
      "step": 2825
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1066428048181574,
      "learning_rate": 7.102144687787938e-06,
      "loss": 0.2008,
      "step": 2826
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1504339430860684,
      "learning_rate": 7.1001630838826586e-06,
      "loss": 0.1682,
      "step": 2827
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9700492029440559,
      "learning_rate": 7.098181079357369e-06,
      "loss": 0.1854,
      "step": 2828
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8170891740808521,
      "learning_rate": 7.096198674590149e-06,
      "loss": 0.1964,
      "step": 2829
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.84965904751481,
      "learning_rate": 7.094215869959156e-06,
      "loss": 0.1815,
      "step": 2830
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9371978547907245,
      "learning_rate": 7.092232665842622e-06,
      "loss": 0.1775,
      "step": 2831
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1499387756174637,
      "learning_rate": 7.090249062618857e-06,
      "loss": 0.2169,
      "step": 2832
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8132025131645627,
      "learning_rate": 7.088265060666248e-06,
      "loss": 0.1625,
      "step": 2833
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1101108388000513,
      "learning_rate": 7.086280660363254e-06,
      "loss": 0.2594,
      "step": 2834
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.970081632859375,
      "learning_rate": 7.084295862088411e-06,
      "loss": 0.1942,
      "step": 2835
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.813404403069584,
      "learning_rate": 7.082310666220338e-06,
      "loss": 0.1365,
      "step": 2836
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.129870665620912,
      "learning_rate": 7.08032507313772e-06,
      "loss": 0.2402,
      "step": 2837
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8400915460274699,
      "learning_rate": 7.078339083219326e-06,
      "loss": 0.1758,
      "step": 2838
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.741939149347158,
      "learning_rate": 7.076352696843991e-06,
      "loss": 0.1583,
      "step": 2839
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8369529016931115,
      "learning_rate": 7.074365914390635e-06,
      "loss": 0.1461,
      "step": 2840
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.337776723759854,
      "learning_rate": 7.072378736238248e-06,
      "loss": 0.2083,
      "step": 2841
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1259350838398972,
      "learning_rate": 7.070391162765901e-06,
      "loss": 0.1974,
      "step": 2842
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9584881293509669,
      "learning_rate": 7.068403194352732e-06,
      "loss": 0.2256,
      "step": 2843
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9477145862029993,
      "learning_rate": 7.066414831377964e-06,
      "loss": 0.1702,
      "step": 2844
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.889255040360095,
      "learning_rate": 7.064426074220886e-06,
      "loss": 0.1417,
      "step": 2845
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9420488382907629,
      "learning_rate": 7.062436923260869e-06,
      "loss": 0.2064,
      "step": 2846
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8989272085046384,
      "learning_rate": 7.060447378877355e-06,
      "loss": 0.1583,
      "step": 2847
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1716611281239868,
      "learning_rate": 7.058457441449866e-06,
      "loss": 0.2516,
      "step": 2848
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9675166465901386,
      "learning_rate": 7.056467111357993e-06,
      "loss": 0.1709,
      "step": 2849
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0346359160140688,
      "learning_rate": 7.054476388981405e-06,
      "loss": 0.238,
      "step": 2850
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4584853158872184,
      "learning_rate": 7.052485274699846e-06,
      "loss": 0.3076,
      "step": 2851
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1393295004309818,
      "learning_rate": 7.050493768893134e-06,
      "loss": 0.1631,
      "step": 2852
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7674628908317231,
      "learning_rate": 7.048501871941162e-06,
      "loss": 0.1744,
      "step": 2853
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.880584332822167,
      "learning_rate": 7.046509584223898e-06,
      "loss": 0.2052,
      "step": 2854
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8070513095180055,
      "learning_rate": 7.044516906121383e-06,
      "loss": 0.1731,
      "step": 2855
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6885669016341891,
      "learning_rate": 7.042523838013735e-06,
      "loss": 0.108,
      "step": 2856
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9217754862347629,
      "learning_rate": 7.040530380281143e-06,
      "loss": 0.1671,
      "step": 2857
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8636524456549796,
      "learning_rate": 7.038536533303872e-06,
      "loss": 0.1315,
      "step": 2858
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0658283510525652,
      "learning_rate": 7.036542297462265e-06,
      "loss": 0.2262,
      "step": 2859
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7175160708565577,
      "learning_rate": 7.034547673136732e-06,
      "loss": 0.1226,
      "step": 2860
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9805406438942226,
      "learning_rate": 7.0325526607077655e-06,
      "loss": 0.2068,
      "step": 2861
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1180522990040893,
      "learning_rate": 7.030557260555922e-06,
      "loss": 0.1739,
      "step": 2862
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.118433914176929,
      "learning_rate": 7.028561473061837e-06,
      "loss": 0.2718,
      "step": 2863
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6396540916319086,
      "learning_rate": 7.026565298606225e-06,
      "loss": 0.1642,
      "step": 2864
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7081923472164935,
      "learning_rate": 7.024568737569867e-06,
      "loss": 0.1417,
      "step": 2865
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.93850699025464,
      "learning_rate": 7.02257179033362e-06,
      "loss": 0.1701,
      "step": 2866
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9833184400010689,
      "learning_rate": 7.020574457278415e-06,
      "loss": 0.1927,
      "step": 2867
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.877230376376586,
      "learning_rate": 7.018576738785257e-06,
      "loss": 0.1635,
      "step": 2868
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9393283849737951,
      "learning_rate": 7.016578635235224e-06,
      "loss": 0.16,
      "step": 2869
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9939834480529591,
      "learning_rate": 7.014580147009468e-06,
      "loss": 0.1711,
      "step": 2870
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8760258837471128,
      "learning_rate": 7.0125812744892115e-06,
      "loss": 0.1686,
      "step": 2871
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0448838889030816,
      "learning_rate": 7.010582018055755e-06,
      "loss": 0.18,
      "step": 2872
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.919708924601207,
      "learning_rate": 7.0085823780904686e-06,
      "loss": 0.1706,
      "step": 2873
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8481001388344491,
      "learning_rate": 7.006582354974798e-06,
      "loss": 0.1347,
      "step": 2874
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9539333328141497,
      "learning_rate": 7.004581949090259e-06,
      "loss": 0.1829,
      "step": 2875
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8341259559261798,
      "learning_rate": 7.0025811608184455e-06,
      "loss": 0.1644,
      "step": 2876
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7910408779009669,
      "learning_rate": 7.000579990541018e-06,
      "loss": 0.1325,
      "step": 2877
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1647263590579806,
      "learning_rate": 6.998578438639714e-06,
      "loss": 0.2263,
      "step": 2878
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9871788933486256,
      "learning_rate": 6.996576505496343e-06,
      "loss": 0.1895,
      "step": 2879
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8692491822281139,
      "learning_rate": 6.994574191492787e-06,
      "loss": 0.1562,
      "step": 2880
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3034315834724852,
      "learning_rate": 6.9925714970110004e-06,
      "loss": 0.2972,
      "step": 2881
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9329556778081658,
      "learning_rate": 6.9905684224330105e-06,
      "loss": 0.1345,
      "step": 2882
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9438017013342324,
      "learning_rate": 6.988564968140917e-06,
      "loss": 0.2564,
      "step": 2883
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8979809361139897,
      "learning_rate": 6.9865611345168925e-06,
      "loss": 0.1554,
      "step": 2884
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1246530243337567,
      "learning_rate": 6.984556921943182e-06,
      "loss": 0.241,
      "step": 2885
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0119557251660558,
      "learning_rate": 6.982552330802099e-06,
      "loss": 0.1954,
      "step": 2886
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6141870874103676,
      "learning_rate": 6.980547361476038e-06,
      "loss": 0.1195,
      "step": 2887
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8194115048596414,
      "learning_rate": 6.978542014347455e-06,
      "loss": 0.1648,
      "step": 2888
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8412103290858174,
      "learning_rate": 6.976536289798887e-06,
      "loss": 0.1687,
      "step": 2889
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1062817569658148,
      "learning_rate": 6.974530188212935e-06,
      "loss": 0.1737,
      "step": 2890
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7864506657385418,
      "learning_rate": 6.972523709972282e-06,
      "loss": 0.1954,
      "step": 2891
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.027787022420036,
      "learning_rate": 6.970516855459672e-06,
      "loss": 0.2233,
      "step": 2892
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7401855640069765,
      "learning_rate": 6.9685096250579285e-06,
      "loss": 0.0964,
      "step": 2893
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8324107826310021,
      "learning_rate": 6.966502019149943e-06,
      "loss": 0.1566,
      "step": 2894
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8452283238561933,
      "learning_rate": 6.96449403811868e-06,
      "loss": 0.1243,
      "step": 2895
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9165541232586821,
      "learning_rate": 6.9624856823471744e-06,
      "loss": 0.1541,
      "step": 2896
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0645909805726692,
      "learning_rate": 6.960476952218534e-06,
      "loss": 0.2146,
      "step": 2897
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9414614311718337,
      "learning_rate": 6.958467848115937e-06,
      "loss": 0.1456,
      "step": 2898
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.926994627224411,
      "learning_rate": 6.956458370422633e-06,
      "loss": 0.2041,
      "step": 2899
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.055531593361823,
      "learning_rate": 6.954448519521946e-06,
      "loss": 0.1935,
      "step": 2900
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7523909029697885,
      "learning_rate": 6.952438295797263e-06,
      "loss": 0.1699,
      "step": 2901
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0094516059182563,
      "learning_rate": 6.950427699632053e-06,
      "loss": 0.1743,
      "step": 2902
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7877864259291313,
      "learning_rate": 6.948416731409846e-06,
      "loss": 0.1689,
      "step": 2903
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8057412781629671,
      "learning_rate": 6.9464053915142505e-06,
      "loss": 0.2074,
      "step": 2904
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0179582644553282,
      "learning_rate": 6.944393680328943e-06,
      "loss": 0.157,
      "step": 2905
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8376374286140917,
      "learning_rate": 6.942381598237668e-06,
      "loss": 0.1687,
      "step": 2906
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2517122417528066,
      "learning_rate": 6.940369145624246e-06,
      "loss": 0.2199,
      "step": 2907
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1882674718139281,
      "learning_rate": 6.938356322872565e-06,
      "loss": 0.2508,
      "step": 2908
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2064464014460519,
      "learning_rate": 6.936343130366584e-06,
      "loss": 0.1984,
      "step": 2909
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1546286514481559,
      "learning_rate": 6.934329568490333e-06,
      "loss": 0.1998,
      "step": 2910
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1324126170230275,
      "learning_rate": 6.932315637627912e-06,
      "loss": 0.2611,
      "step": 2911
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1401017054023181,
      "learning_rate": 6.930301338163492e-06,
      "loss": 0.2523,
      "step": 2912
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9933525314945522,
      "learning_rate": 6.928286670481313e-06,
      "loss": 0.1907,
      "step": 2913
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1065107877657505,
      "learning_rate": 6.926271634965689e-06,
      "loss": 0.2612,
      "step": 2914
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.929530038909208,
      "learning_rate": 6.924256232000997e-06,
      "loss": 0.1703,
      "step": 2915
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9653478057768324,
      "learning_rate": 6.922240461971692e-06,
      "loss": 0.1605,
      "step": 2916
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.731206766045542,
      "learning_rate": 6.920224325262295e-06,
      "loss": 0.1276,
      "step": 2917
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.023316894150658,
      "learning_rate": 6.9182078222573946e-06,
      "loss": 0.1878,
      "step": 2918
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0568490380109097,
      "learning_rate": 6.916190953341655e-06,
      "loss": 0.1678,
      "step": 2919
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.899038714875204,
      "learning_rate": 6.914173718899806e-06,
      "loss": 0.1736,
      "step": 2920
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1317873222020984,
      "learning_rate": 6.91215611931665e-06,
      "loss": 0.1663,
      "step": 2921
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7414298663674406,
      "learning_rate": 6.910138154977055e-06,
      "loss": 0.1423,
      "step": 2922
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1464918269092335,
      "learning_rate": 6.908119826265965e-06,
      "loss": 0.2,
      "step": 2923
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9123432236325203,
      "learning_rate": 6.906101133568386e-06,
      "loss": 0.1601,
      "step": 2924
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9179272703222926,
      "learning_rate": 6.904082077269396e-06,
      "loss": 0.1869,
      "step": 2925
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0750666378523566,
      "learning_rate": 6.902062657754146e-06,
      "loss": 0.2021,
      "step": 2926
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0583145954619708,
      "learning_rate": 6.900042875407854e-06,
      "loss": 0.191,
      "step": 2927
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.042590289503633,
      "learning_rate": 6.898022730615806e-06,
      "loss": 0.2275,
      "step": 2928
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0953014149067084,
      "learning_rate": 6.896002223763357e-06,
      "loss": 0.1724,
      "step": 2929
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8321575693539154,
      "learning_rate": 6.893981355235933e-06,
      "loss": 0.1473,
      "step": 2930
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9845530108149722,
      "learning_rate": 6.891960125419028e-06,
      "loss": 0.205,
      "step": 2931
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9733599977196662,
      "learning_rate": 6.889938534698205e-06,
      "loss": 0.154,
      "step": 2932
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8920965491359834,
      "learning_rate": 6.8879165834590955e-06,
      "loss": 0.1787,
      "step": 2933
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9212048032528944,
      "learning_rate": 6.885894272087401e-06,
      "loss": 0.207,
      "step": 2934
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7664378406562191,
      "learning_rate": 6.88387160096889e-06,
      "loss": 0.1404,
      "step": 2935
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9394642699440696,
      "learning_rate": 6.881848570489399e-06,
      "loss": 0.1808,
      "step": 2936
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.117116742614876,
      "learning_rate": 6.879825181034837e-06,
      "loss": 0.1786,
      "step": 2937
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8278171551980723,
      "learning_rate": 6.877801432991179e-06,
      "loss": 0.162,
      "step": 2938
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8707528097508889,
      "learning_rate": 6.875777326744466e-06,
      "loss": 0.1595,
      "step": 2939
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1267357970394287,
      "learning_rate": 6.873752862680812e-06,
      "loss": 0.25,
      "step": 2940
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7841947268558386,
      "learning_rate": 6.871728041186394e-06,
      "loss": 0.1904,
      "step": 2941
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8066871071538689,
      "learning_rate": 6.869702862647462e-06,
      "loss": 0.2043,
      "step": 2942
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2029568169129454,
      "learning_rate": 6.867677327450333e-06,
      "loss": 0.2773,
      "step": 2943
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2089548359321944,
      "learning_rate": 6.86565143598139e-06,
      "loss": 0.2066,
      "step": 2944
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0737955625112643,
      "learning_rate": 6.863625188627086e-06,
      "loss": 0.252,
      "step": 2945
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.277628799328283,
      "learning_rate": 6.861598585773941e-06,
      "loss": 0.2596,
      "step": 2946
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.247528570719736,
      "learning_rate": 6.859571627808542e-06,
      "loss": 0.2395,
      "step": 2947
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1984305463614513,
      "learning_rate": 6.8575443151175435e-06,
      "loss": 0.2499,
      "step": 2948
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0630063599987645,
      "learning_rate": 6.855516648087672e-06,
      "loss": 0.2051,
      "step": 2949
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0785230138100959,
      "learning_rate": 6.853488627105717e-06,
      "loss": 0.1789,
      "step": 2950
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9544349496276712,
      "learning_rate": 6.851460252558538e-06,
      "loss": 0.2119,
      "step": 2951
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8823248608840746,
      "learning_rate": 6.849431524833059e-06,
      "loss": 0.1947,
      "step": 2952
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1070445494670988,
      "learning_rate": 6.8474024443162735e-06,
      "loss": 0.2168,
      "step": 2953
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8212053230484287,
      "learning_rate": 6.845373011395242e-06,
      "loss": 0.1915,
      "step": 2954
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2452558469833603,
      "learning_rate": 6.843343226457095e-06,
      "loss": 0.2495,
      "step": 2955
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8350448156365683,
      "learning_rate": 6.841313089889023e-06,
      "loss": 0.1854,
      "step": 2956
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8095924917861316,
      "learning_rate": 6.839282602078291e-06,
      "loss": 0.1607,
      "step": 2957
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7413163142299141,
      "learning_rate": 6.837251763412227e-06,
      "loss": 0.1321,
      "step": 2958
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9643295845123688,
      "learning_rate": 6.835220574278226e-06,
      "loss": 0.2052,
      "step": 2959
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8131501535441809,
      "learning_rate": 6.833189035063754e-06,
      "loss": 0.1241,
      "step": 2960
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8746853003732196,
      "learning_rate": 6.831157146156339e-06,
      "loss": 0.1548,
      "step": 2961
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8375224400253533,
      "learning_rate": 6.829124907943574e-06,
      "loss": 0.2028,
      "step": 2962
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0405662663422648,
      "learning_rate": 6.8270923208131266e-06,
      "loss": 0.1785,
      "step": 2963
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0703558494197594,
      "learning_rate": 6.825059385152722e-06,
      "loss": 0.2144,
      "step": 2964
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9421078255885694,
      "learning_rate": 6.8230261013501595e-06,
      "loss": 0.1492,
      "step": 2965
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6692618870621301,
      "learning_rate": 6.8209924697933015e-06,
      "loss": 0.1574,
      "step": 2966
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9487285719444921,
      "learning_rate": 6.818958490870074e-06,
      "loss": 0.1874,
      "step": 2967
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7104009706208959,
      "learning_rate": 6.8169241649684745e-06,
      "loss": 0.1442,
      "step": 2968
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8181998588986615,
      "learning_rate": 6.814889492476563e-06,
      "loss": 0.1369,
      "step": 2969
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9236266648588737,
      "learning_rate": 6.812854473782466e-06,
      "loss": 0.2001,
      "step": 2970
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.85946903493574,
      "learning_rate": 6.810819109274378e-06,
      "loss": 0.1465,
      "step": 2971
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9158532040761805,
      "learning_rate": 6.8087833993405584e-06,
      "loss": 0.1724,
      "step": 2972
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.766436283723811,
      "learning_rate": 6.806747344369331e-06,
      "loss": 0.1821,
      "step": 2973
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.761649141837633,
      "learning_rate": 6.804710944749088e-06,
      "loss": 0.1664,
      "step": 2974
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9615800262667212,
      "learning_rate": 6.8026742008682855e-06,
      "loss": 0.1774,
      "step": 2975
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.174871544361149,
      "learning_rate": 6.800637113115447e-06,
      "loss": 0.2022,
      "step": 2976
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9850914930807271,
      "learning_rate": 6.798599681879158e-06,
      "loss": 0.2016,
      "step": 2977
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0458121334795112,
      "learning_rate": 6.796561907548074e-06,
      "loss": 0.1926,
      "step": 2978
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.285775352787307,
      "learning_rate": 6.794523790510914e-06,
      "loss": 0.219,
      "step": 2979
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5841518793607494,
      "learning_rate": 6.7924853311564606e-06,
      "loss": 0.0992,
      "step": 2980
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6383140835166666,
      "learning_rate": 6.7904465298735645e-06,
      "loss": 0.1364,
      "step": 2981
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.166691541845335,
      "learning_rate": 6.788407387051142e-06,
      "loss": 0.237,
      "step": 2982
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6802958828412328,
      "learning_rate": 6.786367903078172e-06,
      "loss": 0.1625,
      "step": 2983
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2097546111382464,
      "learning_rate": 6.784328078343698e-06,
      "loss": 0.2436,
      "step": 2984
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0694658114999311,
      "learning_rate": 6.782287913236834e-06,
      "loss": 0.1702,
      "step": 2985
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9117792311727552,
      "learning_rate": 6.780247408146749e-06,
      "loss": 0.1752,
      "step": 2986
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9305753342230348,
      "learning_rate": 6.778206563462687e-06,
      "loss": 0.1243,
      "step": 2987
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9986068371416488,
      "learning_rate": 6.776165379573952e-06,
      "loss": 0.1658,
      "step": 2988
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8556606970389373,
      "learning_rate": 6.774123856869914e-06,
      "loss": 0.1514,
      "step": 2989
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9970678330310709,
      "learning_rate": 6.772081995740005e-06,
      "loss": 0.186,
      "step": 2990
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.119860311378756,
      "learning_rate": 6.770039796573726e-06,
      "loss": 0.1942,
      "step": 2991
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1243302541543778,
      "learning_rate": 6.767997259760638e-06,
      "loss": 0.1801,
      "step": 2992
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7905715954099999,
      "learning_rate": 6.76595438569037e-06,
      "loss": 0.1386,
      "step": 2993
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9734907664699802,
      "learning_rate": 6.763911174752611e-06,
      "loss": 0.1523,
      "step": 2994
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6362326796257797,
      "learning_rate": 6.761867627337121e-06,
      "loss": 0.1568,
      "step": 2995
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1337440944033976,
      "learning_rate": 6.759823743833717e-06,
      "loss": 0.2207,
      "step": 2996
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.114691917700905,
      "learning_rate": 6.7577795246322844e-06,
      "loss": 0.2137,
      "step": 2997
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9947965469478168,
      "learning_rate": 6.755734970122771e-06,
      "loss": 0.2162,
      "step": 2998
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8846923540542035,
      "learning_rate": 6.753690080695191e-06,
      "loss": 0.1778,
      "step": 2999
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9392159712679259,
      "learning_rate": 6.751644856739619e-06,
      "loss": 0.191,
      "step": 3000
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8113072543948711,
      "learning_rate": 6.749599298646194e-06,
      "loss": 0.1708,
      "step": 3001
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.865778935668425,
      "learning_rate": 6.747553406805122e-06,
      "loss": 0.2109,
      "step": 3002
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0526157083314227,
      "learning_rate": 6.7455071816066684e-06,
      "loss": 0.2082,
      "step": 3003
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9210919764190071,
      "learning_rate": 6.743460623441165e-06,
      "loss": 0.162,
      "step": 3004
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1733098495534304,
      "learning_rate": 6.741413732699008e-06,
      "loss": 0.219,
      "step": 3005
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9209955209949069,
      "learning_rate": 6.739366509770653e-06,
      "loss": 0.1732,
      "step": 3006
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7518408105939033,
      "learning_rate": 6.7373189550466225e-06,
      "loss": 0.119,
      "step": 3007
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9877969225774159,
      "learning_rate": 6.735271068917501e-06,
      "loss": 0.1598,
      "step": 3008
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0054963748448682,
      "learning_rate": 6.733222851773936e-06,
      "loss": 0.1793,
      "step": 3009
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9789385138449563,
      "learning_rate": 6.731174304006639e-06,
      "loss": 0.1911,
      "step": 3010
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9767866526534039,
      "learning_rate": 6.7291254260063834e-06,
      "loss": 0.2146,
      "step": 3011
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0601824227108267,
      "learning_rate": 6.727076218164006e-06,
      "loss": 0.1893,
      "step": 3012
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8476407948737271,
      "learning_rate": 6.725026680870408e-06,
      "loss": 0.1663,
      "step": 3013
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.872910910536,
      "learning_rate": 6.722976814516551e-06,
      "loss": 0.1414,
      "step": 3014
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0320594542964807,
      "learning_rate": 6.720926619493462e-06,
      "loss": 0.1772,
      "step": 3015
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8575952230239846,
      "learning_rate": 6.718876096192227e-06,
      "loss": 0.2103,
      "step": 3016
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.026575439565605,
      "learning_rate": 6.7168252450039995e-06,
      "loss": 0.2229,
      "step": 3017
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.852995007619263,
      "learning_rate": 6.7147740663199915e-06,
      "loss": 0.1554,
      "step": 3018
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0170994145803802,
      "learning_rate": 6.712722560531477e-06,
      "loss": 0.1728,
      "step": 3019
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9133571944333275,
      "learning_rate": 6.7106707280297974e-06,
      "loss": 0.157,
      "step": 3020
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8279952386447424,
      "learning_rate": 6.708618569206353e-06,
      "loss": 0.1983,
      "step": 3021
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7067678449263621,
      "learning_rate": 6.706566084452606e-06,
      "loss": 0.1686,
      "step": 3022
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8687815703140261,
      "learning_rate": 6.704513274160083e-06,
      "loss": 0.0997,
      "step": 3023
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8409064513204234,
      "learning_rate": 6.702460138720368e-06,
      "loss": 0.1668,
      "step": 3024
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0277330861220255,
      "learning_rate": 6.700406678525112e-06,
      "loss": 0.215,
      "step": 3025
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.815628860162803,
      "learning_rate": 6.698352893966024e-06,
      "loss": 0.1507,
      "step": 3026
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6447312410092424,
      "learning_rate": 6.696298785434882e-06,
      "loss": 0.0989,
      "step": 3027
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7591551999948479,
      "learning_rate": 6.694244353323515e-06,
      "loss": 0.1612,
      "step": 3028
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8322987307531645,
      "learning_rate": 6.6921895980238235e-06,
      "loss": 0.151,
      "step": 3029
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8665384232727259,
      "learning_rate": 6.6901345199277654e-06,
      "loss": 0.154,
      "step": 3030
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9829853119133568,
      "learning_rate": 6.68807911942736e-06,
      "loss": 0.1735,
      "step": 3031
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.052781063938126,
      "learning_rate": 6.686023396914685e-06,
      "loss": 0.2465,
      "step": 3032
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.823553537011791,
      "learning_rate": 6.6839673527818885e-06,
      "loss": 0.1271,
      "step": 3033
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.183682889146376,
      "learning_rate": 6.681910987421171e-06,
      "loss": 0.2422,
      "step": 3034
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1038292405504577,
      "learning_rate": 6.6798543012247995e-06,
      "loss": 0.2116,
      "step": 3035
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7603105969540861,
      "learning_rate": 6.6777972945851e-06,
      "loss": 0.1274,
      "step": 3036
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9096959970608113,
      "learning_rate": 6.675739967894459e-06,
      "loss": 0.1939,
      "step": 3037
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0984147219450568,
      "learning_rate": 6.673682321545327e-06,
      "loss": 0.2195,
      "step": 3038
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1918852920321084,
      "learning_rate": 6.671624355930213e-06,
      "loss": 0.2065,
      "step": 3039
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8923877238172716,
      "learning_rate": 6.669566071441689e-06,
      "loss": 0.1601,
      "step": 3040
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8093376093984918,
      "learning_rate": 6.667507468472383e-06,
      "loss": 0.1589,
      "step": 3041
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9620745317995796,
      "learning_rate": 6.665448547414989e-06,
      "loss": 0.1565,
      "step": 3042
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8302782451028086,
      "learning_rate": 6.663389308662259e-06,
      "loss": 0.121,
      "step": 3043
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0164228874551233,
      "learning_rate": 6.661329752607009e-06,
      "loss": 0.2625,
      "step": 3044
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0075950466742682,
      "learning_rate": 6.65926987964211e-06,
      "loss": 0.19,
      "step": 3045
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1246571145313804,
      "learning_rate": 6.6572096901605e-06,
      "loss": 0.1953,
      "step": 3046
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8080117946579581,
      "learning_rate": 6.65514918455517e-06,
      "loss": 0.1394,
      "step": 3047
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9763179521569438,
      "learning_rate": 6.653088363219176e-06,
      "loss": 0.1276,
      "step": 3048
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0841816232013195,
      "learning_rate": 6.651027226545632e-06,
      "loss": 0.2294,
      "step": 3049
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9941460346052649,
      "learning_rate": 6.648965774927718e-06,
      "loss": 0.1795,
      "step": 3050
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0205581860352413,
      "learning_rate": 6.646904008758665e-06,
      "loss": 0.16,
      "step": 3051
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.999116724118137,
      "learning_rate": 6.6448419284317724e-06,
      "loss": 0.1655,
      "step": 3052
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7336061821796505,
      "learning_rate": 6.642779534340392e-06,
      "loss": 0.1794,
      "step": 3053
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7483707492800282,
      "learning_rate": 6.640716826877942e-06,
      "loss": 0.1264,
      "step": 3054
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8022215437295233,
      "learning_rate": 6.638653806437896e-06,
      "loss": 0.1562,
      "step": 3055
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8070237671602603,
      "learning_rate": 6.636590473413787e-06,
      "loss": 0.1389,
      "step": 3056
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9454622740243044,
      "learning_rate": 6.634526828199215e-06,
      "loss": 0.1232,
      "step": 3057
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6794980668659582,
      "learning_rate": 6.632462871187828e-06,
      "loss": 0.1051,
      "step": 3058
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3513143506105916,
      "learning_rate": 6.630398602773343e-06,
      "loss": 0.2614,
      "step": 3059
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9299288828350463,
      "learning_rate": 6.62833402334953e-06,
      "loss": 0.1616,
      "step": 3060
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8487399598034975,
      "learning_rate": 6.626269133310225e-06,
      "loss": 0.1431,
      "step": 3061
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9327027928806635,
      "learning_rate": 6.624203933049317e-06,
      "loss": 0.1914,
      "step": 3062
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0267771777727248,
      "learning_rate": 6.622138422960757e-06,
      "loss": 0.2149,
      "step": 3063
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8495181354812115,
      "learning_rate": 6.620072603438554e-06,
      "loss": 0.1597,
      "step": 3064
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0092282568647621,
      "learning_rate": 6.618006474876777e-06,
      "loss": 0.2309,
      "step": 3065
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2464759878972203,
      "learning_rate": 6.615940037669556e-06,
      "loss": 0.2731,
      "step": 3066
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7394008210723984,
      "learning_rate": 6.613873292211075e-06,
      "loss": 0.1687,
      "step": 3067
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0411981196525806,
      "learning_rate": 6.61180623889558e-06,
      "loss": 0.2074,
      "step": 3068
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6723385587942562,
      "learning_rate": 6.609738878117375e-06,
      "loss": 0.1012,
      "step": 3069
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.982299292429497,
      "learning_rate": 6.607671210270825e-06,
      "loss": 0.1167,
      "step": 3070
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.077047381238657,
      "learning_rate": 6.605603235750347e-06,
      "loss": 0.2173,
      "step": 3071
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7403292922914633,
      "learning_rate": 6.603534954950424e-06,
      "loss": 0.1457,
      "step": 3072
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9024872702704261,
      "learning_rate": 6.601466368265594e-06,
      "loss": 0.146,
      "step": 3073
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1887648840455118,
      "learning_rate": 6.599397476090453e-06,
      "loss": 0.2579,
      "step": 3074
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9022487276526236,
      "learning_rate": 6.597328278819656e-06,
      "loss": 0.1816,
      "step": 3075
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2543591397034792,
      "learning_rate": 6.5952587768479144e-06,
      "loss": 0.2208,
      "step": 3076
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7368583334943896,
      "learning_rate": 6.5931889705700035e-06,
      "loss": 0.1164,
      "step": 3077
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7210733045168832,
      "learning_rate": 6.591118860380749e-06,
      "loss": 0.1097,
      "step": 3078
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8690324109290877,
      "learning_rate": 6.589048446675039e-06,
      "loss": 0.1732,
      "step": 3079
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9207789749019689,
      "learning_rate": 6.586977729847819e-06,
      "loss": 0.1352,
      "step": 3080
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.11996486674473,
      "learning_rate": 6.584906710294091e-06,
      "loss": 0.1892,
      "step": 3081
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7029128308466434,
      "learning_rate": 6.582835388408916e-06,
      "loss": 0.1797,
      "step": 3082
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2667043908027833,
      "learning_rate": 6.580763764587413e-06,
      "loss": 0.2033,
      "step": 3083
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0988051521094353,
      "learning_rate": 6.57869183922476e-06,
      "loss": 0.2053,
      "step": 3084
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9595496511353603,
      "learning_rate": 6.576619612716188e-06,
      "loss": 0.2144,
      "step": 3085
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.122062178089135,
      "learning_rate": 6.574547085456986e-06,
      "loss": 0.1984,
      "step": 3086
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0903514947869972,
      "learning_rate": 6.572474257842505e-06,
      "loss": 0.216,
      "step": 3087
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7404673971277069,
      "learning_rate": 6.570401130268149e-06,
      "loss": 0.1199,
      "step": 3088
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1490859753640137,
      "learning_rate": 6.568327703129383e-06,
      "loss": 0.2018,
      "step": 3089
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0664369992867675,
      "learning_rate": 6.566253976821727e-06,
      "loss": 0.1567,
      "step": 3090
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2102328779050444,
      "learning_rate": 6.564179951740755e-06,
      "loss": 0.2677,
      "step": 3091
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1449624343249254,
      "learning_rate": 6.562105628282105e-06,
      "loss": 0.2047,
      "step": 3092
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.034287200447329,
      "learning_rate": 6.5600310068414645e-06,
      "loss": 0.1834,
      "step": 3093
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9128639337966944,
      "learning_rate": 6.557956087814582e-06,
      "loss": 0.173,
      "step": 3094
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1981185354578547,
      "learning_rate": 6.555880871597262e-06,
      "loss": 0.2893,
      "step": 3095
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0378468531253984,
      "learning_rate": 6.5538053585853675e-06,
      "loss": 0.1732,
      "step": 3096
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9059727783602584,
      "learning_rate": 6.5517295491748156e-06,
      "loss": 0.1393,
      "step": 3097
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8135352457100078,
      "learning_rate": 6.54965344376158e-06,
      "loss": 0.1286,
      "step": 3098
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6981272573325141,
      "learning_rate": 6.547577042741691e-06,
      "loss": 0.1231,
      "step": 3099
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9277897284640803,
      "learning_rate": 6.545500346511237e-06,
      "loss": 0.19,
      "step": 3100
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7219501191161909,
      "learning_rate": 6.54342335546636e-06,
      "loss": 0.1165,
      "step": 3101
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.91798715992262,
      "learning_rate": 6.541346070003264e-06,
      "loss": 0.1691,
      "step": 3102
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1345008245160655,
      "learning_rate": 6.5392684905182e-06,
      "loss": 0.2647,
      "step": 3103
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0150583204392196,
      "learning_rate": 6.537190617407481e-06,
      "loss": 0.2069,
      "step": 3104
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8799646467853974,
      "learning_rate": 6.535112451067477e-06,
      "loss": 0.1421,
      "step": 3105
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9975182551088886,
      "learning_rate": 6.533033991894611e-06,
      "loss": 0.2068,
      "step": 3106
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8601102252767973,
      "learning_rate": 6.530955240285363e-06,
      "loss": 0.1209,
      "step": 3107
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9486104553120993,
      "learning_rate": 6.528876196636269e-06,
      "loss": 0.1452,
      "step": 3108
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.669236983589855,
      "learning_rate": 6.526796861343919e-06,
      "loss": 0.1083,
      "step": 3109
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.843903289842928,
      "learning_rate": 6.52471723480496e-06,
      "loss": 0.1853,
      "step": 3110
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1105002875056555,
      "learning_rate": 6.522637317416096e-06,
      "loss": 0.1788,
      "step": 3111
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1154644017555144,
      "learning_rate": 6.520557109574086e-06,
      "loss": 0.1731,
      "step": 3112
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9149793705074292,
      "learning_rate": 6.5184766116757395e-06,
      "loss": 0.1828,
      "step": 3113
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.315416256418724,
      "learning_rate": 6.516395824117929e-06,
      "loss": 0.2449,
      "step": 3114
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5897856719683376,
      "learning_rate": 6.514314747297577e-06,
      "loss": 0.1892,
      "step": 3115
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7363202344202495,
      "learning_rate": 6.512233381611661e-06,
      "loss": 0.1445,
      "step": 3116
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9619264098219557,
      "learning_rate": 6.510151727457218e-06,
      "loss": 0.1822,
      "step": 3117
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7629273649985496,
      "learning_rate": 6.508069785231337e-06,
      "loss": 0.1351,
      "step": 3118
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1457145939117543,
      "learning_rate": 6.50598755533116e-06,
      "loss": 0.2144,
      "step": 3119
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9702039425430392,
      "learning_rate": 6.5039050381538885e-06,
      "loss": 0.2159,
      "step": 3120
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.974481151521882,
      "learning_rate": 6.501822234096775e-06,
      "loss": 0.1904,
      "step": 3121
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1051738228320551,
      "learning_rate": 6.499739143557129e-06,
      "loss": 0.1661,
      "step": 3122
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8783267186236835,
      "learning_rate": 6.497655766932315e-06,
      "loss": 0.15,
      "step": 3123
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8328408529756012,
      "learning_rate": 6.495572104619748e-06,
      "loss": 0.178,
      "step": 3124
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0262172016194633,
      "learning_rate": 6.493488157016902e-06,
      "loss": 0.2029,
      "step": 3125
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9145841654124343,
      "learning_rate": 6.491403924521304e-06,
      "loss": 0.1664,
      "step": 3126
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0540380610790152,
      "learning_rate": 6.489319407530534e-06,
      "loss": 0.2608,
      "step": 3127
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8157890300327244,
      "learning_rate": 6.487234606442229e-06,
      "loss": 0.1712,
      "step": 3128
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1537489368689404,
      "learning_rate": 6.4851495216540775e-06,
      "loss": 0.2353,
      "step": 3129
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.043304312279388,
      "learning_rate": 6.483064153563823e-06,
      "loss": 0.2176,
      "step": 3130
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9719012995334656,
      "learning_rate": 6.480978502569265e-06,
      "loss": 0.2101,
      "step": 3131
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9941662181695049,
      "learning_rate": 6.4788925690682535e-06,
      "loss": 0.1958,
      "step": 3132
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0184819384139774,
      "learning_rate": 6.476806353458693e-06,
      "loss": 0.1644,
      "step": 3133
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.933720378297325,
      "learning_rate": 6.4747198561385475e-06,
      "loss": 0.1644,
      "step": 3134
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0198794401712061,
      "learning_rate": 6.472633077505825e-06,
      "loss": 0.2173,
      "step": 3135
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7720681560124554,
      "learning_rate": 6.470546017958595e-06,
      "loss": 0.1483,
      "step": 3136
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9234719440508313,
      "learning_rate": 6.468458677894978e-06,
      "loss": 0.2168,
      "step": 3137
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8731125730561147,
      "learning_rate": 6.4663710577131475e-06,
      "loss": 0.1496,
      "step": 3138
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7133018834072532,
      "learning_rate": 6.464283157811332e-06,
      "loss": 0.192,
      "step": 3139
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9478875144991992,
      "learning_rate": 6.4621949785878095e-06,
      "loss": 0.1897,
      "step": 3140
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9234501224793876,
      "learning_rate": 6.460106520440916e-06,
      "loss": 0.1793,
      "step": 3141
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.275388513478614,
      "learning_rate": 6.458017783769039e-06,
      "loss": 0.2261,
      "step": 3142
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9351882337062494,
      "learning_rate": 6.455928768970616e-06,
      "loss": 0.128,
      "step": 3143
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6758761449362894,
      "learning_rate": 6.453839476444144e-06,
      "loss": 0.1116,
      "step": 3144
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.652408591624659,
      "learning_rate": 6.451749906588167e-06,
      "loss": 0.0957,
      "step": 3145
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9902393711918263,
      "learning_rate": 6.449660059801287e-06,
      "loss": 0.1756,
      "step": 3146
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2262582621791105,
      "learning_rate": 6.447569936482153e-06,
      "loss": 0.2533,
      "step": 3147
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6369617198018447,
      "learning_rate": 6.445479537029471e-06,
      "loss": 0.1269,
      "step": 3148
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8571201294891237,
      "learning_rate": 6.443388861841999e-06,
      "loss": 0.214,
      "step": 3149
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9400961394428352,
      "learning_rate": 6.441297911318547e-06,
      "loss": 0.1712,
      "step": 3150
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2155052068391083,
      "learning_rate": 6.439206685857977e-06,
      "loss": 0.2392,
      "step": 3151
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9397064237215956,
      "learning_rate": 6.437115185859205e-06,
      "loss": 0.1667,
      "step": 3152
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7441595639003105,
      "learning_rate": 6.435023411721198e-06,
      "loss": 0.1432,
      "step": 3153
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9702298470033125,
      "learning_rate": 6.432931363842977e-06,
      "loss": 0.142,
      "step": 3154
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0389588059233288,
      "learning_rate": 6.430839042623611e-06,
      "loss": 0.1845,
      "step": 3155
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9474660732826402,
      "learning_rate": 6.428746448462227e-06,
      "loss": 0.2112,
      "step": 3156
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8404674681924468,
      "learning_rate": 6.426653581757999e-06,
      "loss": 0.1571,
      "step": 3157
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0974076042202654,
      "learning_rate": 6.424560442910157e-06,
      "loss": 0.2028,
      "step": 3158
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6893662914329438,
      "learning_rate": 6.42246703231798e-06,
      "loss": 0.1211,
      "step": 3159
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0103800331200004,
      "learning_rate": 6.420373350380801e-06,
      "loss": 0.2142,
      "step": 3160
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8669178276982972,
      "learning_rate": 6.418279397498003e-06,
      "loss": 0.1621,
      "step": 3161
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.149950752686713,
      "learning_rate": 6.416185174069021e-06,
      "loss": 0.2446,
      "step": 3162
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8330580007977934,
      "learning_rate": 6.414090680493343e-06,
      "loss": 0.1887,
      "step": 3163
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1142009063432725,
      "learning_rate": 6.411995917170506e-06,
      "loss": 0.1946,
      "step": 3164
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8317125772656634,
      "learning_rate": 6.4099008845001e-06,
      "loss": 0.178,
      "step": 3165
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8496933442779565,
      "learning_rate": 6.407805582881767e-06,
      "loss": 0.123,
      "step": 3166
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9015630409193259,
      "learning_rate": 6.405710012715201e-06,
      "loss": 0.2069,
      "step": 3167
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7810509124176188,
      "learning_rate": 6.403614174400146e-06,
      "loss": 0.1474,
      "step": 3168
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.005090876531031,
      "learning_rate": 6.401518068336395e-06,
      "loss": 0.1809,
      "step": 3169
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9439538727412466,
      "learning_rate": 6.399421694923794e-06,
      "loss": 0.1459,
      "step": 3170
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.096665581492917,
      "learning_rate": 6.397325054562242e-06,
      "loss": 0.2199,
      "step": 3171
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7173492886723163,
      "learning_rate": 6.395228147651684e-06,
      "loss": 0.1691,
      "step": 3172
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8992081054397256,
      "learning_rate": 6.393130974592122e-06,
      "loss": 0.1695,
      "step": 3173
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.879774215171381,
      "learning_rate": 6.391033535783605e-06,
      "loss": 0.1332,
      "step": 3174
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7958426263139131,
      "learning_rate": 6.388935831626233e-06,
      "loss": 0.1531,
      "step": 3175
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0395378287743053,
      "learning_rate": 6.386837862520157e-06,
      "loss": 0.1972,
      "step": 3176
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8495339211390418,
      "learning_rate": 6.384739628865577e-06,
      "loss": 0.1799,
      "step": 3177
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1973147417487078,
      "learning_rate": 6.382641131062746e-06,
      "loss": 0.22,
      "step": 3178
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7974096219881709,
      "learning_rate": 6.380542369511969e-06,
      "loss": 0.1547,
      "step": 3179
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9149559970354038,
      "learning_rate": 6.378443344613594e-06,
      "loss": 0.166,
      "step": 3180
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7729138486873252,
      "learning_rate": 6.376344056768027e-06,
      "loss": 0.133,
      "step": 3181
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9320895492331024,
      "learning_rate": 6.374244506375721e-06,
      "loss": 0.2099,
      "step": 3182
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9757005024531279,
      "learning_rate": 6.372144693837176e-06,
      "loss": 0.1584,
      "step": 3183
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7500559195082059,
      "learning_rate": 6.37004461955295e-06,
      "loss": 0.1658,
      "step": 3184
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.946418606680315,
      "learning_rate": 6.367944283923643e-06,
      "loss": 0.1924,
      "step": 3185
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9955776210512474,
      "learning_rate": 6.3658436873499075e-06,
      "loss": 0.1468,
      "step": 3186
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8866849383918626,
      "learning_rate": 6.363742830232448e-06,
      "loss": 0.1612,
      "step": 3187
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.041940689948413,
      "learning_rate": 6.361641712972015e-06,
      "loss": 0.2201,
      "step": 3188
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0182517128253659,
      "learning_rate": 6.3595403359694116e-06,
      "loss": 0.2153,
      "step": 3189
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48577499589136536,
      "learning_rate": 6.357438699625491e-06,
      "loss": 0.0922,
      "step": 3190
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0380890331912611,
      "learning_rate": 6.355336804341152e-06,
      "loss": 0.1965,
      "step": 3191
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.098037431184755,
      "learning_rate": 6.3532346505173446e-06,
      "loss": 0.1988,
      "step": 3192
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9535285909672399,
      "learning_rate": 6.351132238555072e-06,
      "loss": 0.2235,
      "step": 3193
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9586077648381052,
      "learning_rate": 6.349029568855378e-06,
      "loss": 0.1942,
      "step": 3194
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8785998059537523,
      "learning_rate": 6.346926641819365e-06,
      "loss": 0.1759,
      "step": 3195
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9040083521858789,
      "learning_rate": 6.3448234578481795e-06,
      "loss": 0.2132,
      "step": 3196
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8513994277572755,
      "learning_rate": 6.342720017343016e-06,
      "loss": 0.1086,
      "step": 3197
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0127605506266084,
      "learning_rate": 6.340616320705122e-06,
      "loss": 0.1928,
      "step": 3198
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8748286185770402,
      "learning_rate": 6.33851236833579e-06,
      "loss": 0.167,
      "step": 3199
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8442037215042029,
      "learning_rate": 6.336408160636362e-06,
      "loss": 0.1632,
      "step": 3200
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7571573309013994,
      "learning_rate": 6.334303698008232e-06,
      "loss": 0.147,
      "step": 3201
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6103247896306828,
      "learning_rate": 6.332198980852839e-06,
      "loss": 0.1227,
      "step": 3202
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0538364909114044,
      "learning_rate": 6.33009400957167e-06,
      "loss": 0.2016,
      "step": 3203
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1418110440038336,
      "learning_rate": 6.3279887845662654e-06,
      "loss": 0.2359,
      "step": 3204
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8874391328612645,
      "learning_rate": 6.3258833062382095e-06,
      "loss": 0.1564,
      "step": 3205
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7384562764113297,
      "learning_rate": 6.323777574989134e-06,
      "loss": 0.1646,
      "step": 3206
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7860922331959276,
      "learning_rate": 6.321671591220726e-06,
      "loss": 0.1584,
      "step": 3207
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9882339050904325,
      "learning_rate": 6.319565355334711e-06,
      "loss": 0.2031,
      "step": 3208
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9070373092173576,
      "learning_rate": 6.31745886773287e-06,
      "loss": 0.1636,
      "step": 3209
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1089091586300088,
      "learning_rate": 6.315352128817028e-06,
      "loss": 0.2171,
      "step": 3210
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.358575334740753,
      "learning_rate": 6.31324513898906e-06,
      "loss": 0.2446,
      "step": 3211
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0614493587244402,
      "learning_rate": 6.311137898650889e-06,
      "loss": 0.2196,
      "step": 3212
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8344592747267526,
      "learning_rate": 6.309030408204485e-06,
      "loss": 0.1681,
      "step": 3213
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6853671376791425,
      "learning_rate": 6.306922668051866e-06,
      "loss": 0.0898,
      "step": 3214
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8140198350958752,
      "learning_rate": 6.304814678595095e-06,
      "loss": 0.1649,
      "step": 3215
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8154762882992758,
      "learning_rate": 6.3027064402362895e-06,
      "loss": 0.1224,
      "step": 3216
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1082399035884154,
      "learning_rate": 6.300597953377603e-06,
      "loss": 0.2424,
      "step": 3217
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.698776638810563,
      "learning_rate": 6.2984892184212485e-06,
      "loss": 0.1321,
      "step": 3218
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0330229169657499,
      "learning_rate": 6.296380235769481e-06,
      "loss": 0.1802,
      "step": 3219
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7941812242908346,
      "learning_rate": 6.2942710058246024e-06,
      "loss": 0.1556,
      "step": 3220
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8781860050548086,
      "learning_rate": 6.29216152898896e-06,
      "loss": 0.1762,
      "step": 3221
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1152958759066267,
      "learning_rate": 6.290051805664954e-06,
      "loss": 0.2214,
      "step": 3222
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0505611252046057,
      "learning_rate": 6.287941836255026e-06,
      "loss": 0.1928,
      "step": 3223
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.022246526100619,
      "learning_rate": 6.285831621161667e-06,
      "loss": 0.1984,
      "step": 3224
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0928716636630766,
      "learning_rate": 6.283721160787413e-06,
      "loss": 0.2173,
      "step": 3225
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1801422693931947,
      "learning_rate": 6.281610455534851e-06,
      "loss": 0.2435,
      "step": 3226
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8526243218292492,
      "learning_rate": 6.27949950580661e-06,
      "loss": 0.1333,
      "step": 3227
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8762899224509515,
      "learning_rate": 6.277388312005368e-06,
      "loss": 0.1614,
      "step": 3228
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8136621310251485,
      "learning_rate": 6.2752768745338515e-06,
      "loss": 0.1584,
      "step": 3229
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4661037769864387,
      "learning_rate": 6.2731651937948275e-06,
      "loss": 0.2837,
      "step": 3230
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8384090450569323,
      "learning_rate": 6.271053270191116e-06,
      "loss": 0.1535,
      "step": 3231
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1515773360088255,
      "learning_rate": 6.26894110412558e-06,
      "loss": 0.1912,
      "step": 3232
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9748707898642016,
      "learning_rate": 6.266828696001127e-06,
      "loss": 0.1657,
      "step": 3233
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.969840420886193,
      "learning_rate": 6.264716046220713e-06,
      "loss": 0.19,
      "step": 3234
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9146987653775734,
      "learning_rate": 6.262603155187343e-06,
      "loss": 0.1914,
      "step": 3235
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.755970438236634,
      "learning_rate": 6.260490023304062e-06,
      "loss": 0.1444,
      "step": 3236
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0530455404770793,
      "learning_rate": 6.258376650973967e-06,
      "loss": 0.1953,
      "step": 3237
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0468988556836076,
      "learning_rate": 6.2562630386001945e-06,
      "loss": 0.1759,
      "step": 3238
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0048260346700242,
      "learning_rate": 6.25414918658593e-06,
      "loss": 0.2248,
      "step": 3239
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7784600736379296,
      "learning_rate": 6.252035095334408e-06,
      "loss": 0.1842,
      "step": 3240
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.117263520778145,
      "learning_rate": 6.249920765248902e-06,
      "loss": 0.2218,
      "step": 3241
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8141482493264715,
      "learning_rate": 6.2478061967327374e-06,
      "loss": 0.1396,
      "step": 3242
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0544008417555442,
      "learning_rate": 6.245691390189281e-06,
      "loss": 0.1967,
      "step": 3243
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.036777434246522,
      "learning_rate": 6.243576346021944e-06,
      "loss": 0.1957,
      "step": 3244
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.907210428274371,
      "learning_rate": 6.241461064634188e-06,
      "loss": 0.2092,
      "step": 3245
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8202021985326615,
      "learning_rate": 6.239345546429516e-06,
      "loss": 0.2088,
      "step": 3246
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9421030026234818,
      "learning_rate": 6.2372297918114766e-06,
      "loss": 0.1584,
      "step": 3247
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2771513306557223,
      "learning_rate": 6.235113801183666e-06,
      "loss": 0.2175,
      "step": 3248
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9777460722568628,
      "learning_rate": 6.23299757494972e-06,
      "loss": 0.2155,
      "step": 3249
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9418388407293709,
      "learning_rate": 6.230881113513324e-06,
      "loss": 0.1428,
      "step": 3250
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9366562075291531,
      "learning_rate": 6.228764417278207e-06,
      "loss": 0.1853,
      "step": 3251
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9238510611002327,
      "learning_rate": 6.226647486648145e-06,
      "loss": 0.1851,
      "step": 3252
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1358496677076082,
      "learning_rate": 6.224530322026953e-06,
      "loss": 0.2264,
      "step": 3253
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8445795314999224,
      "learning_rate": 6.222412923818499e-06,
      "loss": 0.1509,
      "step": 3254
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0928240628292558,
      "learning_rate": 6.220295292426688e-06,
      "loss": 0.2233,
      "step": 3255
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.807558862476007,
      "learning_rate": 6.218177428255468e-06,
      "loss": 0.1593,
      "step": 3256
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.091893390667862,
      "learning_rate": 6.2160593317088425e-06,
      "loss": 0.2033,
      "step": 3257
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0335697964133073,
      "learning_rate": 6.213941003190849e-06,
      "loss": 0.1872,
      "step": 3258
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8752295923552332,
      "learning_rate": 6.211822443105574e-06,
      "loss": 0.2291,
      "step": 3259
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0028598999200582,
      "learning_rate": 6.2097036518571455e-06,
      "loss": 0.1928,
      "step": 3260
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8902364307556637,
      "learning_rate": 6.207584629849736e-06,
      "loss": 0.1318,
      "step": 3261
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.676730178379467,
      "learning_rate": 6.205465377487568e-06,
      "loss": 0.1325,
      "step": 3262
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0604459455284045,
      "learning_rate": 6.203345895174896e-06,
      "loss": 0.226,
      "step": 3263
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9938779881987492,
      "learning_rate": 6.20122618331603e-06,
      "loss": 0.1804,
      "step": 3264
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.811637721543829,
      "learning_rate": 6.199106242315318e-06,
      "loss": 0.1362,
      "step": 3265
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2252292115625805,
      "learning_rate": 6.1969860725771534e-06,
      "loss": 0.2746,
      "step": 3266
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1973550313027639,
      "learning_rate": 6.19486567450597e-06,
      "loss": 0.2543,
      "step": 3267
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1427337608490478,
      "learning_rate": 6.19274504850625e-06,
      "loss": 0.1829,
      "step": 3268
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6934810617580925,
      "learning_rate": 6.1906241949825175e-06,
      "loss": 0.1694,
      "step": 3269
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8417920479019366,
      "learning_rate": 6.1885031143393385e-06,
      "loss": 0.1898,
      "step": 3270
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.907939599129123,
      "learning_rate": 6.186381806981322e-06,
      "loss": 0.1712,
      "step": 3271
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9938534226632566,
      "learning_rate": 6.184260273313122e-06,
      "loss": 0.1889,
      "step": 3272
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8656495916331595,
      "learning_rate": 6.182138513739435e-06,
      "loss": 0.1392,
      "step": 3273
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7593072252560845,
      "learning_rate": 6.180016528665001e-06,
      "loss": 0.1464,
      "step": 3274
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8451109131116119,
      "learning_rate": 6.177894318494605e-06,
      "loss": 0.165,
      "step": 3275
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0367230918314014,
      "learning_rate": 6.175771883633068e-06,
      "loss": 0.1483,
      "step": 3276
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9459202562486363,
      "learning_rate": 6.173649224485263e-06,
      "loss": 0.1947,
      "step": 3277
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.156306073093026,
      "learning_rate": 6.1715263414561e-06,
      "loss": 0.1971,
      "step": 3278
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.978807194383666,
      "learning_rate": 6.16940323495053e-06,
      "loss": 0.1689,
      "step": 3279
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9377976077008201,
      "learning_rate": 6.167279905373552e-06,
      "loss": 0.1619,
      "step": 3280
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2181670783949765,
      "learning_rate": 6.165156353130205e-06,
      "loss": 0.229,
      "step": 3281
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.959410654098788,
      "learning_rate": 6.1630325786255714e-06,
      "loss": 0.2091,
      "step": 3282
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0828707901862433,
      "learning_rate": 6.160908582264774e-06,
      "loss": 0.1983,
      "step": 3283
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7544465965181328,
      "learning_rate": 6.158784364452981e-06,
      "loss": 0.1259,
      "step": 3284
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8585498771228145,
      "learning_rate": 6.156659925595398e-06,
      "loss": 0.1669,
      "step": 3285
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8356258326594794,
      "learning_rate": 6.154535266097279e-06,
      "loss": 0.1623,
      "step": 3286
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9345842485267477,
      "learning_rate": 6.152410386363914e-06,
      "loss": 0.1983,
      "step": 3287
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8491441595775413,
      "learning_rate": 6.1502852868006405e-06,
      "loss": 0.1462,
      "step": 3288
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0405429724258248,
      "learning_rate": 6.148159967812832e-06,
      "loss": 0.19,
      "step": 3289
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9914221982149378,
      "learning_rate": 6.146034429805909e-06,
      "loss": 0.1713,
      "step": 3290
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8849530232950932,
      "learning_rate": 6.143908673185334e-06,
      "loss": 0.1881,
      "step": 3291
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9651160561799463,
      "learning_rate": 6.141782698356608e-06,
      "loss": 0.1984,
      "step": 3292
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8428327490060369,
      "learning_rate": 6.139656505725274e-06,
      "loss": 0.2204,
      "step": 3293
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0634207486619869,
      "learning_rate": 6.1375300956969155e-06,
      "loss": 0.2225,
      "step": 3294
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.947504243051404,
      "learning_rate": 6.135403468677161e-06,
      "loss": 0.2107,
      "step": 3295
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0071402801850007,
      "learning_rate": 6.133276625071681e-06,
      "loss": 0.1999,
      "step": 3296
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8808533446606809,
      "learning_rate": 6.131149565286181e-06,
      "loss": 0.1452,
      "step": 3297
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0262015458721239,
      "learning_rate": 6.129022289726414e-06,
      "loss": 0.1841,
      "step": 3298
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1915867719121067,
      "learning_rate": 6.126894798798171e-06,
      "loss": 0.2212,
      "step": 3299
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1755675540861534,
      "learning_rate": 6.1247670929072865e-06,
      "loss": 0.2615,
      "step": 3300
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6129509821274504,
      "learning_rate": 6.122639172459632e-06,
      "loss": 0.0969,
      "step": 3301
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7186439506384219,
      "learning_rate": 6.120511037861125e-06,
      "loss": 0.155,
      "step": 3302
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9208245301955248,
      "learning_rate": 6.11838268951772e-06,
      "loss": 0.1577,
      "step": 3303
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0075952199636242,
      "learning_rate": 6.116254127835414e-06,
      "loss": 0.2378,
      "step": 3304
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9416757596762488,
      "learning_rate": 6.114125353220243e-06,
      "loss": 0.1997,
      "step": 3305
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.095988362022771,
      "learning_rate": 6.111996366078287e-06,
      "loss": 0.2312,
      "step": 3306
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1062634398152513,
      "learning_rate": 6.109867166815662e-06,
      "loss": 0.2267,
      "step": 3307
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7353189762814795,
      "learning_rate": 6.107737755838528e-06,
      "loss": 0.1695,
      "step": 3308
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6736289133911205,
      "learning_rate": 6.105608133553085e-06,
      "loss": 0.0942,
      "step": 3309
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0988892944173225,
      "learning_rate": 6.103478300365571e-06,
      "loss": 0.1762,
      "step": 3310
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9982331097239802,
      "learning_rate": 6.1013482566822665e-06,
      "loss": 0.1868,
      "step": 3311
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1033585598295905,
      "learning_rate": 6.0992180029094905e-06,
      "loss": 0.1837,
      "step": 3312
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9230524013037473,
      "learning_rate": 6.097087539453606e-06,
      "loss": 0.191,
      "step": 3313
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8006442531279105,
      "learning_rate": 6.094956866721009e-06,
      "loss": 0.1616,
      "step": 3314
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0082843954955896,
      "learning_rate": 6.0928259851181425e-06,
      "loss": 0.1721,
      "step": 3315
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8367880088812952,
      "learning_rate": 6.090694895051487e-06,
      "loss": 0.1624,
      "step": 3316
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9517207767367818,
      "learning_rate": 6.088563596927558e-06,
      "loss": 0.1464,
      "step": 3317
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0533126685617096,
      "learning_rate": 6.086432091152916e-06,
      "loss": 0.2215,
      "step": 3318
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9148148021045074,
      "learning_rate": 6.084300378134162e-06,
      "loss": 0.1832,
      "step": 3319
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.890135743934907,
      "learning_rate": 6.082168458277934e-06,
      "loss": 0.1397,
      "step": 3320
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7563020776969095,
      "learning_rate": 6.080036331990908e-06,
      "loss": 0.1379,
      "step": 3321
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.209909100923266,
      "learning_rate": 6.077903999679803e-06,
      "loss": 0.2266,
      "step": 3322
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8347384910449455,
      "learning_rate": 6.075771461751373e-06,
      "loss": 0.1092,
      "step": 3323
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8906773496251132,
      "learning_rate": 6.0736387186124165e-06,
      "loss": 0.2051,
      "step": 3324
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9031451536572365,
      "learning_rate": 6.071505770669767e-06,
      "loss": 0.168,
      "step": 3325
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1253120663352307,
      "learning_rate": 6.0693726183303e-06,
      "loss": 0.1972,
      "step": 3326
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9662984308244261,
      "learning_rate": 6.067239262000925e-06,
      "loss": 0.1667,
      "step": 3327
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0568590880684343,
      "learning_rate": 6.065105702088598e-06,
      "loss": 0.2093,
      "step": 3328
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4768588235973381,
      "learning_rate": 6.062971939000306e-06,
      "loss": 0.1162,
      "step": 3329
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7271376836980898,
      "learning_rate": 6.060837973143081e-06,
      "loss": 0.163,
      "step": 3330
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9003031905984398,
      "learning_rate": 6.058703804923991e-06,
      "loss": 0.1516,
      "step": 3331
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2390316680863758,
      "learning_rate": 6.056569434750141e-06,
      "loss": 0.2279,
      "step": 3332
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.987951646498775,
      "learning_rate": 6.054434863028678e-06,
      "loss": 0.1788,
      "step": 3333
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9749979595469361,
      "learning_rate": 6.052300090166783e-06,
      "loss": 0.2129,
      "step": 3334
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2458130090999744,
      "learning_rate": 6.0501651165716805e-06,
      "loss": 0.2349,
      "step": 3335
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8377324767223108,
      "learning_rate": 6.0480299426506305e-06,
      "loss": 0.1617,
      "step": 3336
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0796883814722067,
      "learning_rate": 6.045894568810931e-06,
      "loss": 0.1575,
      "step": 3337
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7923449145015939,
      "learning_rate": 6.0437589954599195e-06,
      "loss": 0.1777,
      "step": 3338
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.790525949052376,
      "learning_rate": 6.041623223004971e-06,
      "loss": 0.1509,
      "step": 3339
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5623605670827303,
      "learning_rate": 6.039487251853497e-06,
      "loss": 0.137,
      "step": 3340
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8631358474132035,
      "learning_rate": 6.037351082412947e-06,
      "loss": 0.1575,
      "step": 3341
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8369383934935082,
      "learning_rate": 6.035214715090813e-06,
      "loss": 0.1711,
      "step": 3342
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0527460644566973,
      "learning_rate": 6.033078150294619e-06,
      "loss": 0.193,
      "step": 3343
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1259633895198504,
      "learning_rate": 6.030941388431929e-06,
      "loss": 0.2406,
      "step": 3344
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9119085219525361,
      "learning_rate": 6.028804429910344e-06,
      "loss": 0.1881,
      "step": 3345
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.106247758885928,
      "learning_rate": 6.026667275137506e-06,
      "loss": 0.2116,
      "step": 3346
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6796696194191811,
      "learning_rate": 6.024529924521086e-06,
      "loss": 0.1591,
      "step": 3347
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8716255751392579,
      "learning_rate": 6.0223923784688025e-06,
      "loss": 0.1544,
      "step": 3348
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7966828345076515,
      "learning_rate": 6.020254637388404e-06,
      "loss": 0.1606,
      "step": 3349
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0461448368220103,
      "learning_rate": 6.018116701687681e-06,
      "loss": 0.211,
      "step": 3350
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9087212703392943,
      "learning_rate": 6.015978571774457e-06,
      "loss": 0.1925,
      "step": 3351
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2270801509199503,
      "learning_rate": 6.013840248056593e-06,
      "loss": 0.2107,
      "step": 3352
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1386582345480818,
      "learning_rate": 6.011701730941993e-06,
      "loss": 0.1966,
      "step": 3353
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1724820084542635,
      "learning_rate": 6.00956302083859e-06,
      "loss": 0.2075,
      "step": 3354
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6908129495302547,
      "learning_rate": 6.007424118154358e-06,
      "loss": 0.1551,
      "step": 3355
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2055353979840329,
      "learning_rate": 6.0052850232973055e-06,
      "loss": 0.2548,
      "step": 3356
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8446655394623233,
      "learning_rate": 6.00314573667548e-06,
      "loss": 0.1494,
      "step": 3357
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6865029532985428,
      "learning_rate": 6.001006258696965e-06,
      "loss": 0.1518,
      "step": 3358
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9859646115012686,
      "learning_rate": 5.99886658976988e-06,
      "loss": 0.1715,
      "step": 3359
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6313181510783413,
      "learning_rate": 5.996726730302382e-06,
      "loss": 0.1222,
      "step": 3360
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7224993636454411,
      "learning_rate": 5.994586680702662e-06,
      "loss": 0.1057,
      "step": 3361
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8797065089029054,
      "learning_rate": 5.992446441378947e-06,
      "loss": 0.2099,
      "step": 3362
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8856091589844136,
      "learning_rate": 5.990306012739507e-06,
      "loss": 0.1794,
      "step": 3363
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.112422567603893,
      "learning_rate": 5.988165395192635e-06,
      "loss": 0.1954,
      "step": 3364
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.975403435551824,
      "learning_rate": 5.9860245891466766e-06,
      "loss": 0.1831,
      "step": 3365
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.25480607765012,
      "learning_rate": 5.983883595009999e-06,
      "loss": 0.2417,
      "step": 3366
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.07932013571442,
      "learning_rate": 5.981742413191012e-06,
      "loss": 0.2012,
      "step": 3367
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.172080856597703,
      "learning_rate": 5.97960104409816e-06,
      "loss": 0.2295,
      "step": 3368
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0427404658680373,
      "learning_rate": 5.977459488139928e-06,
      "loss": 0.1703,
      "step": 3369
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.005622558011316,
      "learning_rate": 5.975317745724824e-06,
      "loss": 0.1878,
      "step": 3370
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.826504269738225,
      "learning_rate": 5.973175817261406e-06,
      "loss": 0.1785,
      "step": 3371
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0656157342927082,
      "learning_rate": 5.971033703158259e-06,
      "loss": 0.1942,
      "step": 3372
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8582720261194278,
      "learning_rate": 5.968891403824005e-06,
      "loss": 0.1619,
      "step": 3373
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8584579104281521,
      "learning_rate": 5.9667489196673e-06,
      "loss": 0.1846,
      "step": 3374
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8812579972450608,
      "learning_rate": 5.964606251096839e-06,
      "loss": 0.1837,
      "step": 3375
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.46964319241777563,
      "learning_rate": 5.962463398521352e-06,
      "loss": 0.1181,
      "step": 3376
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0995901025326582,
      "learning_rate": 5.960320362349602e-06,
      "loss": 0.2252,
      "step": 3377
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9039656002653291,
      "learning_rate": 5.958177142990383e-06,
      "loss": 0.1678,
      "step": 3378
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.078142143884205,
      "learning_rate": 5.956033740852533e-06,
      "loss": 0.2231,
      "step": 3379
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0115649333866812,
      "learning_rate": 5.953890156344917e-06,
      "loss": 0.2078,
      "step": 3380
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7571236754799984,
      "learning_rate": 5.951746389876439e-06,
      "loss": 0.1199,
      "step": 3381
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0831773621972127,
      "learning_rate": 5.949602441856037e-06,
      "loss": 0.1853,
      "step": 3382
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.077967017736181,
      "learning_rate": 5.947458312692685e-06,
      "loss": 0.2256,
      "step": 3383
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7838964518273219,
      "learning_rate": 5.945314002795386e-06,
      "loss": 0.1402,
      "step": 3384
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0168246789638522,
      "learning_rate": 5.9431695125731834e-06,
      "loss": 0.2297,
      "step": 3385
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1366750209201903,
      "learning_rate": 5.941024842435154e-06,
      "loss": 0.204,
      "step": 3386
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.905751476764097,
      "learning_rate": 5.9388799927904065e-06,
      "loss": 0.1456,
      "step": 3387
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8254184780875954,
      "learning_rate": 5.936734964048084e-06,
      "loss": 0.1579,
      "step": 3388
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.232852402382801,
      "learning_rate": 5.934589756617367e-06,
      "loss": 0.248,
      "step": 3389
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.861346015141579,
      "learning_rate": 5.932444370907468e-06,
      "loss": 0.1925,
      "step": 3390
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7959080958853717,
      "learning_rate": 5.93029880732763e-06,
      "loss": 0.1387,
      "step": 3391
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.974713205882901,
      "learning_rate": 5.92815306628714e-06,
      "loss": 0.1759,
      "step": 3392
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7665304076323795,
      "learning_rate": 5.926007148195306e-06,
      "loss": 0.1498,
      "step": 3393
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0439176547709133,
      "learning_rate": 5.923861053461479e-06,
      "loss": 0.1856,
      "step": 3394
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9464594381826548,
      "learning_rate": 5.921714782495042e-06,
      "loss": 0.1662,
      "step": 3395
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1606229869197155,
      "learning_rate": 5.919568335705406e-06,
      "loss": 0.234,
      "step": 3396
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8714370032392162,
      "learning_rate": 5.917421713502024e-06,
      "loss": 0.1501,
      "step": 3397
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0085117231184364,
      "learning_rate": 5.915274916294378e-06,
      "loss": 0.1489,
      "step": 3398
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0277831506818609,
      "learning_rate": 5.913127944491982e-06,
      "loss": 0.2049,
      "step": 3399
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1072906931107513,
      "learning_rate": 5.910980798504387e-06,
      "loss": 0.1971,
      "step": 3400
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.174948373592888,
      "learning_rate": 5.908833478741174e-06,
      "loss": 0.1654,
      "step": 3401
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8569093777644314,
      "learning_rate": 5.906685985611956e-06,
      "loss": 0.1155,
      "step": 3402
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.896487264301202,
      "learning_rate": 5.9045383195263875e-06,
      "loss": 0.1578,
      "step": 3403
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9111904935233098,
      "learning_rate": 5.9023904808941456e-06,
      "loss": 0.1792,
      "step": 3404
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2068602476397683,
      "learning_rate": 5.900242470124946e-06,
      "loss": 0.2156,
      "step": 3405
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0187064813004947,
      "learning_rate": 5.898094287628536e-06,
      "loss": 0.1733,
      "step": 3406
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.206483445024679,
      "learning_rate": 5.895945933814695e-06,
      "loss": 0.2133,
      "step": 3407
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8002046889662504,
      "learning_rate": 5.893797409093237e-06,
      "loss": 0.1699,
      "step": 3408
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.152532995683105,
      "learning_rate": 5.891648713874006e-06,
      "loss": 0.2068,
      "step": 3409
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0174355993760777,
      "learning_rate": 5.889499848566882e-06,
      "loss": 0.2109,
      "step": 3410
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.945765294753759,
      "learning_rate": 5.887350813581772e-06,
      "loss": 0.1225,
      "step": 3411
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8979048334893208,
      "learning_rate": 5.885201609328621e-06,
      "loss": 0.1729,
      "step": 3412
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9649419404678922,
      "learning_rate": 5.883052236217402e-06,
      "loss": 0.2008,
      "step": 3413
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9774624926271177,
      "learning_rate": 5.880902694658124e-06,
      "loss": 0.1634,
      "step": 3414
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.944022093926397,
      "learning_rate": 5.878752985060826e-06,
      "loss": 0.2112,
      "step": 3415
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0409287926688275,
      "learning_rate": 5.87660310783558e-06,
      "loss": 0.1682,
      "step": 3416
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8826181799019032,
      "learning_rate": 5.874453063392488e-06,
      "loss": 0.1855,
      "step": 3417
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6332883384192426,
      "learning_rate": 5.872302852141684e-06,
      "loss": 0.1356,
      "step": 3418
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1492705512768928,
      "learning_rate": 5.870152474493338e-06,
      "loss": 0.2152,
      "step": 3419
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0209834107862066,
      "learning_rate": 5.868001930857647e-06,
      "loss": 0.1903,
      "step": 3420
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.832543458960036,
      "learning_rate": 5.865851221644842e-06,
      "loss": 0.1644,
      "step": 3421
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0130905976110307,
      "learning_rate": 5.863700347265184e-06,
      "loss": 0.2182,
      "step": 3422
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.986883211102315,
      "learning_rate": 5.861549308128968e-06,
      "loss": 0.1526,
      "step": 3423
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9745417863152303,
      "learning_rate": 5.859398104646518e-06,
      "loss": 0.1785,
      "step": 3424
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8796932929504703,
      "learning_rate": 5.857246737228191e-06,
      "loss": 0.1858,
      "step": 3425
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2282872135063234,
      "learning_rate": 5.855095206284373e-06,
      "loss": 0.2,
      "step": 3426
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7609843425897445,
      "learning_rate": 5.852943512225486e-06,
      "loss": 0.1014,
      "step": 3427
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1201751468037568,
      "learning_rate": 5.850791655461977e-06,
      "loss": 0.1907,
      "step": 3428
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1219976404566703,
      "learning_rate": 5.848639636404327e-06,
      "loss": 0.1784,
      "step": 3429
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3295595309451895,
      "learning_rate": 5.846487455463049e-06,
      "loss": 0.2374,
      "step": 3430
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9031893786804409,
      "learning_rate": 5.844335113048686e-06,
      "loss": 0.1992,
      "step": 3431
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0104859606776093,
      "learning_rate": 5.84218260957181e-06,
      "loss": 0.1799,
      "step": 3432
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9610803000226441,
      "learning_rate": 5.8400299454430275e-06,
      "loss": 0.218,
      "step": 3433
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9674117964998696,
      "learning_rate": 5.837877121072971e-06,
      "loss": 0.1808,
      "step": 3434
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9152081319901623,
      "learning_rate": 5.835724136872307e-06,
      "loss": 0.1239,
      "step": 3435
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1017368095735576,
      "learning_rate": 5.833570993251732e-06,
      "loss": 0.192,
      "step": 3436
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9452022223160129,
      "learning_rate": 5.831417690621972e-06,
      "loss": 0.2092,
      "step": 3437
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3397028462663014,
      "learning_rate": 5.829264229393784e-06,
      "loss": 0.232,
      "step": 3438
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8516797931249739,
      "learning_rate": 5.827110609977956e-06,
      "loss": 0.1621,
      "step": 3439
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6796373735735737,
      "learning_rate": 5.824956832785303e-06,
      "loss": 0.1052,
      "step": 3440
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6973093236612845,
      "learning_rate": 5.822802898226672e-06,
      "loss": 0.0876,
      "step": 3441
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4420564962796288,
      "learning_rate": 5.820648806712943e-06,
      "loss": 0.1051,
      "step": 3442
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9989974211910538,
      "learning_rate": 5.818494558655021e-06,
      "loss": 0.1588,
      "step": 3443
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9414351017342197,
      "learning_rate": 5.816340154463844e-06,
      "loss": 0.1848,
      "step": 3444
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.053950370544144,
      "learning_rate": 5.814185594550378e-06,
      "loss": 0.227,
      "step": 3445
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9555570914021073,
      "learning_rate": 5.812030879325621e-06,
      "loss": 0.2126,
      "step": 3446
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8261788758088368,
      "learning_rate": 5.809876009200598e-06,
      "loss": 0.1686,
      "step": 3447
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0161307547515817,
      "learning_rate": 5.807720984586365e-06,
      "loss": 0.2288,
      "step": 3448
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7652965973486272,
      "learning_rate": 5.805565805894008e-06,
      "loss": 0.1144,
      "step": 3449
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.268288047150215,
      "learning_rate": 5.803410473534641e-06,
      "loss": 0.22,
      "step": 3450
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8874696309029468,
      "learning_rate": 5.801254987919407e-06,
      "loss": 0.1448,
      "step": 3451
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9118862006062535,
      "learning_rate": 5.799099349459479e-06,
      "loss": 0.1593,
      "step": 3452
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9883780123556914,
      "learning_rate": 5.796943558566061e-06,
      "loss": 0.1875,
      "step": 3453
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9168870402365901,
      "learning_rate": 5.794787615650385e-06,
      "loss": 0.1468,
      "step": 3454
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0651034336346525,
      "learning_rate": 5.7926315211237106e-06,
      "loss": 0.1798,
      "step": 3455
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7640150744641651,
      "learning_rate": 5.790475275397325e-06,
      "loss": 0.1276,
      "step": 3456
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9717278698413699,
      "learning_rate": 5.788318878882548e-06,
      "loss": 0.178,
      "step": 3457
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.95918783654714,
      "learning_rate": 5.786162331990727e-06,
      "loss": 0.1876,
      "step": 3458
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0583575613688803,
      "learning_rate": 5.784005635133236e-06,
      "loss": 0.1994,
      "step": 3459
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9978603621660672,
      "learning_rate": 5.781848788721481e-06,
      "loss": 0.214,
      "step": 3460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9068990319223217,
      "learning_rate": 5.7796917931668935e-06,
      "loss": 0.183,
      "step": 3461
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8576320988105547,
      "learning_rate": 5.777534648880937e-06,
      "loss": 0.1232,
      "step": 3462
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9242432098619917,
      "learning_rate": 5.775377356275098e-06,
      "loss": 0.18,
      "step": 3463
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1420291410840144,
      "learning_rate": 5.773219915760895e-06,
      "loss": 0.2143,
      "step": 3464
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9821646769287372,
      "learning_rate": 5.771062327749875e-06,
      "loss": 0.2108,
      "step": 3465
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.132330467662958,
      "learning_rate": 5.768904592653611e-06,
      "loss": 0.21,
      "step": 3466
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0367406959968277,
      "learning_rate": 5.766746710883707e-06,
      "loss": 0.2163,
      "step": 3467
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.069365472975599,
      "learning_rate": 5.764588682851792e-06,
      "loss": 0.2063,
      "step": 3468
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4059140781470958,
      "learning_rate": 5.762430508969524e-06,
      "loss": 0.2294,
      "step": 3469
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8690181201396947,
      "learning_rate": 5.7602721896485904e-06,
      "loss": 0.1404,
      "step": 3470
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1767129945936163,
      "learning_rate": 5.758113725300703e-06,
      "loss": 0.2304,
      "step": 3471
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0449196273100205,
      "learning_rate": 5.755955116337605e-06,
      "loss": 0.1663,
      "step": 3472
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8802353986374349,
      "learning_rate": 5.753796363171064e-06,
      "loss": 0.1401,
      "step": 3473
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9828952353613137,
      "learning_rate": 5.751637466212876e-06,
      "loss": 0.1768,
      "step": 3474
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0202902730213153,
      "learning_rate": 5.7494784258748645e-06,
      "loss": 0.1622,
      "step": 3475
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0494429595613133,
      "learning_rate": 5.7473192425688836e-06,
      "loss": 0.182,
      "step": 3476
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2112512140926057,
      "learning_rate": 5.745159916706812e-06,
      "loss": 0.2316,
      "step": 3477
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8499543257192338,
      "learning_rate": 5.743000448700552e-06,
      "loss": 0.1744,
      "step": 3478
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9798604614172947,
      "learning_rate": 5.7408408389620374e-06,
      "loss": 0.1954,
      "step": 3479
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.740024776831007,
      "learning_rate": 5.73868108790323e-06,
      "loss": 0.1811,
      "step": 3480
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8411731337569097,
      "learning_rate": 5.736521195936112e-06,
      "loss": 0.1642,
      "step": 3481
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9827423000380965,
      "learning_rate": 5.734361163472702e-06,
      "loss": 0.1428,
      "step": 3482
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0718480288965377,
      "learning_rate": 5.732200990925041e-06,
      "loss": 0.1754,
      "step": 3483
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0104395877320353,
      "learning_rate": 5.730040678705192e-06,
      "loss": 0.2213,
      "step": 3484
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8563384272814186,
      "learning_rate": 5.727880227225251e-06,
      "loss": 0.1581,
      "step": 3485
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9713886251759807,
      "learning_rate": 5.725719636897339e-06,
      "loss": 0.2047,
      "step": 3486
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7865898213250888,
      "learning_rate": 5.723558908133602e-06,
      "loss": 0.1424,
      "step": 3487
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0459945121199914,
      "learning_rate": 5.721398041346214e-06,
      "loss": 0.2041,
      "step": 3488
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9818061042867089,
      "learning_rate": 5.719237036947374e-06,
      "loss": 0.1833,
      "step": 3489
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9383920255726135,
      "learning_rate": 5.717075895349306e-06,
      "loss": 0.1783,
      "step": 3490
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9153731933072574,
      "learning_rate": 5.714914616964266e-06,
      "loss": 0.1674,
      "step": 3491
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0002359131655705,
      "learning_rate": 5.7127532022045305e-06,
      "loss": 0.1623,
      "step": 3492
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0672114663469872,
      "learning_rate": 5.710591651482403e-06,
      "loss": 0.2196,
      "step": 3493
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1761748709537494,
      "learning_rate": 5.7084299652102145e-06,
      "loss": 0.1666,
      "step": 3494
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9682970959143533,
      "learning_rate": 5.706268143800321e-06,
      "loss": 0.1886,
      "step": 3495
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0405008994674394,
      "learning_rate": 5.704106187665103e-06,
      "loss": 0.1921,
      "step": 3496
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9823288383911443,
      "learning_rate": 5.701944097216968e-06,
      "loss": 0.1859,
      "step": 3497
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2018100143826287,
      "learning_rate": 5.699781872868351e-06,
      "loss": 0.2162,
      "step": 3498
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0653882520636286,
      "learning_rate": 5.6976195150317096e-06,
      "loss": 0.2067,
      "step": 3499
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.006553990943175,
      "learning_rate": 5.695457024119527e-06,
      "loss": 0.1979,
      "step": 3500
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9987735884761776,
      "learning_rate": 5.693294400544314e-06,
      "loss": 0.2431,
      "step": 3501
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1472477857111023,
      "learning_rate": 5.691131644718604e-06,
      "loss": 0.1962,
      "step": 3502
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.118904099982939,
      "learning_rate": 5.6889687570549565e-06,
      "loss": 0.2026,
      "step": 3503
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0023062985956346,
      "learning_rate": 5.68680573796596e-06,
      "loss": 0.1666,
      "step": 3504
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7332298206860999,
      "learning_rate": 5.684642587864222e-06,
      "loss": 0.1377,
      "step": 3505
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.872897716315616,
      "learning_rate": 5.682479307162378e-06,
      "loss": 0.1795,
      "step": 3506
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0498055256281973,
      "learning_rate": 5.680315896273087e-06,
      "loss": 0.2436,
      "step": 3507
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7710224397211426,
      "learning_rate": 5.678152355609038e-06,
      "loss": 0.1479,
      "step": 3508
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7713355257680096,
      "learning_rate": 5.6759886855829364e-06,
      "loss": 0.1228,
      "step": 3509
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.024104911431396,
      "learning_rate": 5.673824886607519e-06,
      "loss": 0.2385,
      "step": 3510
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0452074848857618,
      "learning_rate": 5.6716609590955426e-06,
      "loss": 0.2264,
      "step": 3511
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9575451708936266,
      "learning_rate": 5.669496903459793e-06,
      "loss": 0.1708,
      "step": 3512
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9148167455801122,
      "learning_rate": 5.667332720113078e-06,
      "loss": 0.167,
      "step": 3513
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9583671232379307,
      "learning_rate": 5.665168409468227e-06,
      "loss": 0.1463,
      "step": 3514
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9675381448157362,
      "learning_rate": 5.663003971938102e-06,
      "loss": 0.1839,
      "step": 3515
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8912918651877492,
      "learning_rate": 5.660839407935578e-06,
      "loss": 0.1565,
      "step": 3516
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9098675355042668,
      "learning_rate": 5.658674717873562e-06,
      "loss": 0.1542,
      "step": 3517
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8562216402505095,
      "learning_rate": 5.656509902164985e-06,
      "loss": 0.1687,
      "step": 3518
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8166566002912837,
      "learning_rate": 5.654344961222798e-06,
      "loss": 0.1579,
      "step": 3519
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8779603615139941,
      "learning_rate": 5.652179895459976e-06,
      "loss": 0.1615,
      "step": 3520
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9783442337998776,
      "learning_rate": 5.650014705289523e-06,
      "loss": 0.1674,
      "step": 3521
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8359311917075843,
      "learning_rate": 5.647849391124463e-06,
      "loss": 0.1732,
      "step": 3522
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.750251097841947,
      "learning_rate": 5.645683953377842e-06,
      "loss": 0.1439,
      "step": 3523
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3414916255840943,
      "learning_rate": 5.643518392462734e-06,
      "loss": 0.2307,
      "step": 3524
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8661671365989168,
      "learning_rate": 5.641352708792231e-06,
      "loss": 0.1508,
      "step": 3525
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1720625715883748,
      "learning_rate": 5.639186902779454e-06,
      "loss": 0.2176,
      "step": 3526
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2107444841320558,
      "learning_rate": 5.637020974837544e-06,
      "loss": 0.2586,
      "step": 3527
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1563562794642448,
      "learning_rate": 5.634854925379667e-06,
      "loss": 0.2224,
      "step": 3528
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3538301554865513,
      "learning_rate": 5.632688754819009e-06,
      "loss": 0.2352,
      "step": 3529
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6698463280136914,
      "learning_rate": 5.630522463568784e-06,
      "loss": 0.0969,
      "step": 3530
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5825995724031865,
      "learning_rate": 5.628356052042225e-06,
      "loss": 0.182,
      "step": 3531
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0570636785900962,
      "learning_rate": 5.6261895206525895e-06,
      "loss": 0.2011,
      "step": 3532
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0346633675418975,
      "learning_rate": 5.624022869813157e-06,
      "loss": 0.1854,
      "step": 3533
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7079188610213504,
      "learning_rate": 5.621856099937232e-06,
      "loss": 0.1534,
      "step": 3534
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9565611353916732,
      "learning_rate": 5.619689211438139e-06,
      "loss": 0.1768,
      "step": 3535
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0409529826638535,
      "learning_rate": 5.617522204729228e-06,
      "loss": 0.16,
      "step": 3536
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8864473133418783,
      "learning_rate": 5.615355080223867e-06,
      "loss": 0.224,
      "step": 3537
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8605921679849728,
      "learning_rate": 5.613187838335451e-06,
      "loss": 0.1604,
      "step": 3538
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0585880444371085,
      "learning_rate": 5.611020479477398e-06,
      "loss": 0.173,
      "step": 3539
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1705045711506876,
      "learning_rate": 5.608853004063142e-06,
      "loss": 0.1437,
      "step": 3540
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6917047579809597,
      "learning_rate": 5.6066854125061454e-06,
      "loss": 0.1492,
      "step": 3541
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0525906773472247,
      "learning_rate": 5.604517705219889e-06,
      "loss": 0.2108,
      "step": 3542
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5690129273301658,
      "learning_rate": 5.60234988261788e-06,
      "loss": 0.1008,
      "step": 3543
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0554368231834732,
      "learning_rate": 5.600181945113643e-06,
      "loss": 0.2277,
      "step": 3544
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.219987602931922,
      "learning_rate": 5.5980138931207275e-06,
      "loss": 0.186,
      "step": 3545
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6476746608851938,
      "learning_rate": 5.5958457270527035e-06,
      "loss": 0.1195,
      "step": 3546
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.863197560501459,
      "learning_rate": 5.593677447323164e-06,
      "loss": 0.1515,
      "step": 3547
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0093536425221958,
      "learning_rate": 5.591509054345717e-06,
      "loss": 0.2203,
      "step": 3548
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2076913122341653,
      "learning_rate": 5.589340548534006e-06,
      "loss": 0.2349,
      "step": 3549
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9770073218949222,
      "learning_rate": 5.587171930301683e-06,
      "loss": 0.2134,
      "step": 3550
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9633065094530128,
      "learning_rate": 5.585003200062428e-06,
      "loss": 0.2161,
      "step": 3551
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1004270477163367,
      "learning_rate": 5.5828343582299394e-06,
      "loss": 0.2013,
      "step": 3552
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9172014899471503,
      "learning_rate": 5.580665405217939e-06,
      "loss": 0.1759,
      "step": 3553
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0785858772116448,
      "learning_rate": 5.5784963414401685e-06,
      "loss": 0.1881,
      "step": 3554
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9529210909153586,
      "learning_rate": 5.576327167310391e-06,
      "loss": 0.1577,
      "step": 3555
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1016918245233671,
      "learning_rate": 5.574157883242393e-06,
      "loss": 0.2161,
      "step": 3556
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8375801257149773,
      "learning_rate": 5.571988489649977e-06,
      "loss": 0.1568,
      "step": 3557
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7903594821874299,
      "learning_rate": 5.56981898694697e-06,
      "loss": 0.1231,
      "step": 3558
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9616428325672117,
      "learning_rate": 5.567649375547218e-06,
      "loss": 0.199,
      "step": 3559
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0260416540005144,
      "learning_rate": 5.565479655864592e-06,
      "loss": 0.1592,
      "step": 3560
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8274255003576445,
      "learning_rate": 5.563309828312977e-06,
      "loss": 0.1441,
      "step": 3561
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.873425399756485,
      "learning_rate": 5.561139893306285e-06,
      "loss": 0.1654,
      "step": 3562
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.076896211540425,
      "learning_rate": 5.558969851258442e-06,
      "loss": 0.2073,
      "step": 3563
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7734796029700642,
      "learning_rate": 5.556799702583401e-06,
      "loss": 0.1732,
      "step": 3564
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9450725383948545,
      "learning_rate": 5.554629447695129e-06,
      "loss": 0.1438,
      "step": 3565
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9867616177966336,
      "learning_rate": 5.55245908700762e-06,
      "loss": 0.1274,
      "step": 3566
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.150831198088945,
      "learning_rate": 5.550288620934883e-06,
      "loss": 0.2226,
      "step": 3567
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1526198774374081,
      "learning_rate": 5.548118049890948e-06,
      "loss": 0.1491,
      "step": 3568
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8210917956231526,
      "learning_rate": 5.545947374289867e-06,
      "loss": 0.192,
      "step": 3569
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9097368862353339,
      "learning_rate": 5.543776594545711e-06,
      "loss": 0.1675,
      "step": 3570
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.994815847288984,
      "learning_rate": 5.541605711072569e-06,
      "loss": 0.1621,
      "step": 3571
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.90247976551592,
      "learning_rate": 5.539434724284554e-06,
      "loss": 0.1753,
      "step": 3572
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0932026141492677,
      "learning_rate": 5.537263634595793e-06,
      "loss": 0.1748,
      "step": 3573
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0586148184682165,
      "learning_rate": 5.5350924424204376e-06,
      "loss": 0.1795,
      "step": 3574
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0978524270490575,
      "learning_rate": 5.532921148172657e-06,
      "loss": 0.2164,
      "step": 3575
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7344578482323675,
      "learning_rate": 5.530749752266638e-06,
      "loss": 0.1357,
      "step": 3576
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0858664837549474,
      "learning_rate": 5.528578255116593e-06,
      "loss": 0.1707,
      "step": 3577
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8141633051537962,
      "learning_rate": 5.526406657136745e-06,
      "loss": 0.1826,
      "step": 3578
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.953478881051072,
      "learning_rate": 5.524234958741342e-06,
      "loss": 0.1829,
      "step": 3579
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2412610533315311,
      "learning_rate": 5.522063160344652e-06,
      "loss": 0.2241,
      "step": 3580
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7641880083277478,
      "learning_rate": 5.5198912623609565e-06,
      "loss": 0.1526,
      "step": 3581
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8451557498550003,
      "learning_rate": 5.517719265204562e-06,
      "loss": 0.1702,
      "step": 3582
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0683457251297876,
      "learning_rate": 5.51554716928979e-06,
      "loss": 0.2204,
      "step": 3583
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7782831607068591,
      "learning_rate": 5.513374975030984e-06,
      "loss": 0.1166,
      "step": 3584
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6882473252756633,
      "learning_rate": 5.511202682842504e-06,
      "loss": 0.1231,
      "step": 3585
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.731788561585918,
      "learning_rate": 5.509030293138726e-06,
      "loss": 0.1519,
      "step": 3586
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8978042605443396,
      "learning_rate": 5.50685780633405e-06,
      "loss": 0.2028,
      "step": 3587
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.101982744367731,
      "learning_rate": 5.504685222842893e-06,
      "loss": 0.2011,
      "step": 3588
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9894173848913205,
      "learning_rate": 5.5025125430796886e-06,
      "loss": 0.193,
      "step": 3589
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.874257003577039,
      "learning_rate": 5.50033976745889e-06,
      "loss": 0.1716,
      "step": 3590
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9554569606149926,
      "learning_rate": 5.49816689639497e-06,
      "loss": 0.1345,
      "step": 3591
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9690039012165691,
      "learning_rate": 5.495993930302415e-06,
      "loss": 0.2328,
      "step": 3592
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7868152736489606,
      "learning_rate": 5.493820869595735e-06,
      "loss": 0.1647,
      "step": 3593
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0677968468792531,
      "learning_rate": 5.491647714689455e-06,
      "loss": 0.1914,
      "step": 3594
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9199206398593596,
      "learning_rate": 5.4894744659981205e-06,
      "loss": 0.1807,
      "step": 3595
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9858784024893309,
      "learning_rate": 5.487301123936291e-06,
      "loss": 0.1808,
      "step": 3596
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0113468206303298,
      "learning_rate": 5.485127688918544e-06,
      "loss": 0.1922,
      "step": 3597
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1257980626667028,
      "learning_rate": 5.482954161359481e-06,
      "loss": 0.2273,
      "step": 3598
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1095887966812763,
      "learning_rate": 5.480780541673714e-06,
      "loss": 0.1972,
      "step": 3599
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7861910991763744,
      "learning_rate": 5.478606830275877e-06,
      "loss": 0.1727,
      "step": 3600
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.823223273321448,
      "learning_rate": 5.476433027580618e-06,
      "loss": 0.1025,
      "step": 3601
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.771205468139475,
      "learning_rate": 5.474259134002605e-06,
      "loss": 0.1306,
      "step": 3602
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9217017574273735,
      "learning_rate": 5.4720851499565246e-06,
      "loss": 0.1909,
      "step": 3603
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8471343248957393,
      "learning_rate": 5.469911075857073e-06,
      "loss": 0.1802,
      "step": 3604
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9858553612101635,
      "learning_rate": 5.467736912118976e-06,
      "loss": 0.1903,
      "step": 3605
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0232034184852092,
      "learning_rate": 5.465562659156965e-06,
      "loss": 0.1862,
      "step": 3606
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9546579788799846,
      "learning_rate": 5.463388317385797e-06,
      "loss": 0.2129,
      "step": 3607
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9828461738442722,
      "learning_rate": 5.461213887220239e-06,
      "loss": 0.1434,
      "step": 3608
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.136746714894649,
      "learning_rate": 5.45903936907508e-06,
      "loss": 0.2442,
      "step": 3609
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.913525818024531,
      "learning_rate": 5.456864763365121e-06,
      "loss": 0.1826,
      "step": 3610
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0639287550147942,
      "learning_rate": 5.454690070505184e-06,
      "loss": 0.2229,
      "step": 3611
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0898767119185941,
      "learning_rate": 5.4525152909101065e-06,
      "loss": 0.2151,
      "step": 3612
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6596871108882049,
      "learning_rate": 5.450340424994742e-06,
      "loss": 0.0761,
      "step": 3613
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1620018399351548,
      "learning_rate": 5.4481654731739604e-06,
      "loss": 0.2256,
      "step": 3614
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2004976997978007,
      "learning_rate": 5.445990435862648e-06,
      "loss": 0.247,
      "step": 3615
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2953551982854867,
      "learning_rate": 5.4438153134757075e-06,
      "loss": 0.1979,
      "step": 3616
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1004544514733288,
      "learning_rate": 5.441640106428059e-06,
      "loss": 0.2173,
      "step": 3617
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9157738947571697,
      "learning_rate": 5.439464815134635e-06,
      "loss": 0.1735,
      "step": 3618
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8834613291696289,
      "learning_rate": 5.437289440010391e-06,
      "loss": 0.1813,
      "step": 3619
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.917445412228481,
      "learning_rate": 5.435113981470289e-06,
      "loss": 0.2123,
      "step": 3620
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.001847692815748,
      "learning_rate": 5.432938439929314e-06,
      "loss": 0.1834,
      "step": 3621
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9311612093486108,
      "learning_rate": 5.430762815802467e-06,
      "loss": 0.193,
      "step": 3622
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0095907547155138,
      "learning_rate": 5.428587109504763e-06,
      "loss": 0.1906,
      "step": 3623
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8904411173708802,
      "learning_rate": 5.426411321451229e-06,
      "loss": 0.1827,
      "step": 3624
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.95546484305356,
      "learning_rate": 5.424235452056914e-06,
      "loss": 0.1916,
      "step": 3625
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8915963164099806,
      "learning_rate": 5.422059501736876e-06,
      "loss": 0.1683,
      "step": 3626
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.902984218582121,
      "learning_rate": 5.419883470906196e-06,
      "loss": 0.1799,
      "step": 3627
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.013722642294324,
      "learning_rate": 5.417707359979964e-06,
      "loss": 0.1396,
      "step": 3628
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9526072924273594,
      "learning_rate": 5.415531169373287e-06,
      "loss": 0.1862,
      "step": 3629
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1492951153146773,
      "learning_rate": 5.41335489950129e-06,
      "loss": 0.2321,
      "step": 3630
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9263173518589936,
      "learning_rate": 5.41117855077911e-06,
      "loss": 0.1991,
      "step": 3631
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9393622572011336,
      "learning_rate": 5.409002123621898e-06,
      "loss": 0.1253,
      "step": 3632
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0788788497014692,
      "learning_rate": 5.4068256184448245e-06,
      "loss": 0.1899,
      "step": 3633
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7325966349210554,
      "learning_rate": 5.404649035663071e-06,
      "loss": 0.1411,
      "step": 3634
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1083288581585722,
      "learning_rate": 5.402472375691835e-06,
      "loss": 0.2095,
      "step": 3635
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.825090232017793,
      "learning_rate": 5.40029563894633e-06,
      "loss": 0.1585,
      "step": 3636
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8654264730502506,
      "learning_rate": 5.398118825841781e-06,
      "loss": 0.1895,
      "step": 3637
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8677014734881309,
      "learning_rate": 5.395941936793432e-06,
      "loss": 0.1913,
      "step": 3638
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9238181196349637,
      "learning_rate": 5.393764972216537e-06,
      "loss": 0.1694,
      "step": 3639
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8607202517942578,
      "learning_rate": 5.391587932526366e-06,
      "loss": 0.1615,
      "step": 3640
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1336732293067702,
      "learning_rate": 5.389410818138206e-06,
      "loss": 0.1825,
      "step": 3641
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.111317686288547,
      "learning_rate": 5.387233629467354e-06,
      "loss": 0.247,
      "step": 3642
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9329492786221599,
      "learning_rate": 5.385056366929121e-06,
      "loss": 0.1915,
      "step": 3643
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0109906236856956,
      "learning_rate": 5.3828790309388404e-06,
      "loss": 0.1855,
      "step": 3644
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8784089201853924,
      "learning_rate": 5.3807016219118494e-06,
      "loss": 0.139,
      "step": 3645
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8642312616730615,
      "learning_rate": 5.378524140263504e-06,
      "loss": 0.1415,
      "step": 3646
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8109232383264376,
      "learning_rate": 5.3763465864091736e-06,
      "loss": 0.1561,
      "step": 3647
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9199860566185367,
      "learning_rate": 5.37416896076424e-06,
      "loss": 0.1925,
      "step": 3648
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0143839731739712,
      "learning_rate": 5.3719912637440995e-06,
      "loss": 0.196,
      "step": 3649
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6673311928997985,
      "learning_rate": 5.369813495764164e-06,
      "loss": 0.132,
      "step": 3650
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8274403623802135,
      "learning_rate": 5.367635657239856e-06,
      "loss": 0.1591,
      "step": 3651
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9311121873508922,
      "learning_rate": 5.365457748586612e-06,
      "loss": 0.1952,
      "step": 3652
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0232417178665827,
      "learning_rate": 5.363279770219885e-06,
      "loss": 0.1705,
      "step": 3653
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2514753980658904,
      "learning_rate": 5.3611017225551356e-06,
      "loss": 0.2134,
      "step": 3654
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0356607703978868,
      "learning_rate": 5.358923606007842e-06,
      "loss": 0.1886,
      "step": 3655
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7772378604546438,
      "learning_rate": 5.356745420993496e-06,
      "loss": 0.1378,
      "step": 3656
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.967248535294205,
      "learning_rate": 5.354567167927598e-06,
      "loss": 0.1642,
      "step": 3657
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9944168615049148,
      "learning_rate": 5.352388847225666e-06,
      "loss": 0.1788,
      "step": 3658
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0562655031963255,
      "learning_rate": 5.350210459303229e-06,
      "loss": 0.1878,
      "step": 3659
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.103791685514638,
      "learning_rate": 5.348032004575827e-06,
      "loss": 0.2362,
      "step": 3660
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9687019007317865,
      "learning_rate": 5.345853483459018e-06,
      "loss": 0.146,
      "step": 3661
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0309800366024215,
      "learning_rate": 5.3436748963683685e-06,
      "loss": 0.2167,
      "step": 3662
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9517773809528843,
      "learning_rate": 5.341496243719456e-06,
      "loss": 0.1981,
      "step": 3663
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0066908507659251,
      "learning_rate": 5.339317525927875e-06,
      "loss": 0.1887,
      "step": 3664
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9379421897708567,
      "learning_rate": 5.337138743409229e-06,
      "loss": 0.1673,
      "step": 3665
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9654508273396873,
      "learning_rate": 5.334959896579136e-06,
      "loss": 0.1692,
      "step": 3666
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8836846997517235,
      "learning_rate": 5.3327809858532256e-06,
      "loss": 0.1808,
      "step": 3667
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7502306371591951,
      "learning_rate": 5.33060201164714e-06,
      "loss": 0.1621,
      "step": 3668
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8971966074958593,
      "learning_rate": 5.328422974376532e-06,
      "loss": 0.1834,
      "step": 3669
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.99092551374883,
      "learning_rate": 5.326243874457068e-06,
      "loss": 0.1652,
      "step": 3670
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8206882418584414,
      "learning_rate": 5.324064712304423e-06,
      "loss": 0.1789,
      "step": 3671
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0086199857677698,
      "learning_rate": 5.3218854883342906e-06,
      "loss": 0.2028,
      "step": 3672
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1661750999284348,
      "learning_rate": 5.31970620296237e-06,
      "loss": 0.2369,
      "step": 3673
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8977260028841173,
      "learning_rate": 5.317526856604376e-06,
      "loss": 0.1753,
      "step": 3674
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8885258793688319,
      "learning_rate": 5.315347449676032e-06,
      "loss": 0.1552,
      "step": 3675
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8894292777642367,
      "learning_rate": 5.3131679825930725e-06,
      "loss": 0.1678,
      "step": 3676
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7592331570893116,
      "learning_rate": 5.310988455771249e-06,
      "loss": 0.1683,
      "step": 3677
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1359499394120067,
      "learning_rate": 5.308808869626319e-06,
      "loss": 0.218,
      "step": 3678
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6283760831144379,
      "learning_rate": 5.306629224574052e-06,
      "loss": 0.1129,
      "step": 3679
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8952604335278628,
      "learning_rate": 5.304449521030231e-06,
      "loss": 0.176,
      "step": 3680
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9356140174534662,
      "learning_rate": 5.3022697594106485e-06,
      "loss": 0.1718,
      "step": 3681
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.674183184663138,
      "learning_rate": 5.300089940131109e-06,
      "loss": 0.1315,
      "step": 3682
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9472304814852959,
      "learning_rate": 5.2979100636074254e-06,
      "loss": 0.1637,
      "step": 3683
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7847372566783003,
      "learning_rate": 5.295730130255427e-06,
      "loss": 0.0962,
      "step": 3684
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1484451253030608,
      "learning_rate": 5.293550140490948e-06,
      "loss": 0.2179,
      "step": 3685
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9499879374695065,
      "learning_rate": 5.291370094729837e-06,
      "loss": 0.1607,
      "step": 3686
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0789223456496657,
      "learning_rate": 5.28918999338795e-06,
      "loss": 0.1813,
      "step": 3687
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0056431545858957,
      "learning_rate": 5.287009836881157e-06,
      "loss": 0.1965,
      "step": 3688
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.882661956032725,
      "learning_rate": 5.28482962562534e-06,
      "loss": 0.1627,
      "step": 3689
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0031708833227877,
      "learning_rate": 5.282649360036385e-06,
      "loss": 0.1893,
      "step": 3690
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9451564291994551,
      "learning_rate": 5.2804690405301935e-06,
      "loss": 0.1414,
      "step": 3691
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9293478902113413,
      "learning_rate": 5.278288667522677e-06,
      "loss": 0.1423,
      "step": 3692
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7224619315206502,
      "learning_rate": 5.276108241429754e-06,
      "loss": 0.1589,
      "step": 3693
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7333638817543263,
      "learning_rate": 5.273927762667357e-06,
      "loss": 0.1626,
      "step": 3694
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7005870033118,
      "learning_rate": 5.271747231651425e-06,
      "loss": 0.1223,
      "step": 3695
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0564115746875755,
      "learning_rate": 5.269566648797911e-06,
      "loss": 0.1803,
      "step": 3696
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.968857120763289,
      "learning_rate": 5.267386014522773e-06,
      "loss": 0.1863,
      "step": 3697
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9884706711598452,
      "learning_rate": 5.265205329241984e-06,
      "loss": 0.1822,
      "step": 3698
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7072500858540849,
      "learning_rate": 5.2630245933715205e-06,
      "loss": 0.1101,
      "step": 3699
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0749306004075125,
      "learning_rate": 5.260843807327377e-06,
      "loss": 0.1995,
      "step": 3700
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7996933110290201,
      "learning_rate": 5.25866297152555e-06,
      "loss": 0.1733,
      "step": 3701
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7803833501486902,
      "learning_rate": 5.256482086382048e-06,
      "loss": 0.1633,
      "step": 3702
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0708962356231315,
      "learning_rate": 5.254301152312892e-06,
      "loss": 0.2159,
      "step": 3703
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7492188855353958,
      "learning_rate": 5.252120169734106e-06,
      "loss": 0.1117,
      "step": 3704
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1496295171836661,
      "learning_rate": 5.249939139061729e-06,
      "loss": 0.2188,
      "step": 3705
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.269230452951328,
      "learning_rate": 5.247758060711807e-06,
      "loss": 0.2236,
      "step": 3706
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0604161548717466,
      "learning_rate": 5.245576935100395e-06,
      "loss": 0.1618,
      "step": 3707
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0870712949226224,
      "learning_rate": 5.243395762643557e-06,
      "loss": 0.221,
      "step": 3708
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.046995467391783,
      "learning_rate": 5.241214543757367e-06,
      "loss": 0.2187,
      "step": 3709
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8722785257107022,
      "learning_rate": 5.2390332788579046e-06,
      "loss": 0.2145,
      "step": 3710
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8330654294192307,
      "learning_rate": 5.236851968361262e-06,
      "loss": 0.1539,
      "step": 3711
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9803912558800459,
      "learning_rate": 5.234670612683539e-06,
      "loss": 0.1935,
      "step": 3712
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0134989963033771,
      "learning_rate": 5.232489212240843e-06,
      "loss": 0.2017,
      "step": 3713
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8288271377475809,
      "learning_rate": 5.2303077674492905e-06,
      "loss": 0.1533,
      "step": 3714
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0449657593864183,
      "learning_rate": 5.2281262787250075e-06,
      "loss": 0.1936,
      "step": 3715
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8936409430176847,
      "learning_rate": 5.2259447464841265e-06,
      "loss": 0.1768,
      "step": 3716
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.925984654368913,
      "learning_rate": 5.22376317114279e-06,
      "loss": 0.1789,
      "step": 3717
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2102594811799199,
      "learning_rate": 5.221581553117145e-06,
      "loss": 0.2347,
      "step": 3718
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.165983991328409,
      "learning_rate": 5.219399892823354e-06,
      "loss": 0.1878,
      "step": 3719
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0632121966358203,
      "learning_rate": 5.2172181906775805e-06,
      "loss": 0.1838,
      "step": 3720
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.837140372269511,
      "learning_rate": 5.215036447096e-06,
      "loss": 0.1526,
      "step": 3721
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9096107734033061,
      "learning_rate": 5.2128546624947915e-06,
      "loss": 0.1714,
      "step": 3722
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9441005146533967,
      "learning_rate": 5.210672837290151e-06,
      "loss": 0.1812,
      "step": 3723
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7969846321285017,
      "learning_rate": 5.208490971898269e-06,
      "loss": 0.1464,
      "step": 3724
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0092825384628463,
      "learning_rate": 5.206309066735354e-06,
      "loss": 0.1783,
      "step": 3725
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0802591936825787,
      "learning_rate": 5.20412712221762e-06,
      "loss": 0.1995,
      "step": 3726
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1109621951403017,
      "learning_rate": 5.2019451387612854e-06,
      "loss": 0.1868,
      "step": 3727
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1196502780326685,
      "learning_rate": 5.199763116782579e-06,
      "loss": 0.1991,
      "step": 3728
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8885200865369395,
      "learning_rate": 5.197581056697736e-06,
      "loss": 0.1969,
      "step": 3729
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0949503668631662,
      "learning_rate": 5.195398958922999e-06,
      "loss": 0.2354,
      "step": 3730
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8010768803967515,
      "learning_rate": 5.193216823874616e-06,
      "loss": 0.1395,
      "step": 3731
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1284662349481616,
      "learning_rate": 5.191034651968847e-06,
      "loss": 0.1985,
      "step": 3732
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8432873269386263,
      "learning_rate": 5.188852443621951e-06,
      "loss": 0.1308,
      "step": 3733
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9923910375440489,
      "learning_rate": 5.186670199250202e-06,
      "loss": 0.1937,
      "step": 3734
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9210543274045925,
      "learning_rate": 5.184487919269877e-06,
      "loss": 0.2218,
      "step": 3735
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7174372479587027,
      "learning_rate": 5.18230560409726e-06,
      "loss": 0.1566,
      "step": 3736
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7996729318732365,
      "learning_rate": 5.180123254148642e-06,
      "loss": 0.1473,
      "step": 3737
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.928917185638948,
      "learning_rate": 5.177940869840321e-06,
      "loss": 0.1987,
      "step": 3738
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1613328954202478,
      "learning_rate": 5.175758451588601e-06,
      "loss": 0.2023,
      "step": 3739
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9902834827358428,
      "learning_rate": 5.173575999809795e-06,
      "loss": 0.2217,
      "step": 3740
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9499255124929148,
      "learning_rate": 5.171393514920216e-06,
      "loss": 0.1738,
      "step": 3741
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9702897431425609,
      "learning_rate": 5.1692109973361895e-06,
      "loss": 0.2137,
      "step": 3742
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1608040448497112,
      "learning_rate": 5.167028447474045e-06,
      "loss": 0.1524,
      "step": 3743
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9125324933125507,
      "learning_rate": 5.164845865750118e-06,
      "loss": 0.1794,
      "step": 3744
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.079290013514648,
      "learning_rate": 5.162663252580752e-06,
      "loss": 0.1954,
      "step": 3745
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8890295834206469,
      "learning_rate": 5.160480608382293e-06,
      "loss": 0.1377,
      "step": 3746
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9680142387011001,
      "learning_rate": 5.158297933571098e-06,
      "loss": 0.205,
      "step": 3747
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2407246404578403,
      "learning_rate": 5.156115228563522e-06,
      "loss": 0.2283,
      "step": 3748
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8526313160584734,
      "learning_rate": 5.153932493775934e-06,
      "loss": 0.1528,
      "step": 3749
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1975366958051747,
      "learning_rate": 5.151749729624701e-06,
      "loss": 0.1682,
      "step": 3750
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.143725427951856,
      "learning_rate": 5.149566936526205e-06,
      "loss": 0.2234,
      "step": 3751
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7777380370529448,
      "learning_rate": 5.147384114896825e-06,
      "loss": 0.141,
      "step": 3752
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.906392766769436,
      "learning_rate": 5.1452012651529506e-06,
      "loss": 0.1465,
      "step": 3753
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8872235815541994,
      "learning_rate": 5.143018387710972e-06,
      "loss": 0.1452,
      "step": 3754
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0779168977171776,
      "learning_rate": 5.140835482987291e-06,
      "loss": 0.2064,
      "step": 3755
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7556475922923319,
      "learning_rate": 5.138652551398308e-06,
      "loss": 0.1604,
      "step": 3756
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9220340326586427,
      "learning_rate": 5.136469593360434e-06,
      "loss": 0.1587,
      "step": 3757
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.000216532678795,
      "learning_rate": 5.13428660929008e-06,
      "loss": 0.1489,
      "step": 3758
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9745263769473879,
      "learning_rate": 5.132103599603668e-06,
      "loss": 0.1742,
      "step": 3759
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9203353122412775,
      "learning_rate": 5.1299205647176184e-06,
      "loss": 0.1723,
      "step": 3760
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1459578451592884,
      "learning_rate": 5.127737505048361e-06,
      "loss": 0.2186,
      "step": 3761
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9788207694932425,
      "learning_rate": 5.125554421012332e-06,
      "loss": 0.1935,
      "step": 3762
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1293794339222851,
      "learning_rate": 5.123371313025964e-06,
      "loss": 0.2445,
      "step": 3763
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8468951992218279,
      "learning_rate": 5.121188181505701e-06,
      "loss": 0.1789,
      "step": 3764
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8511899100651229,
      "learning_rate": 5.11900502686799e-06,
      "loss": 0.1112,
      "step": 3765
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8150212487547686,
      "learning_rate": 5.116821849529283e-06,
      "loss": 0.1364,
      "step": 3766
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1384247348524583,
      "learning_rate": 5.114638649906034e-06,
      "loss": 0.1952,
      "step": 3767
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9921294357015834,
      "learning_rate": 5.112455428414704e-06,
      "loss": 0.1817,
      "step": 3768
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9069453896081027,
      "learning_rate": 5.110272185471758e-06,
      "loss": 0.1863,
      "step": 3769
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8783263837582249,
      "learning_rate": 5.108088921493661e-06,
      "loss": 0.1968,
      "step": 3770
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7735576089171402,
      "learning_rate": 5.105905636896889e-06,
      "loss": 0.1313,
      "step": 3771
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.23797015787439,
      "learning_rate": 5.103722332097912e-06,
      "loss": 0.2512,
      "step": 3772
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7130477471189424,
      "learning_rate": 5.1015390075132155e-06,
      "loss": 0.1377,
      "step": 3773
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.848521292031544,
      "learning_rate": 5.099355663559281e-06,
      "loss": 0.1353,
      "step": 3774
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6120815964746659,
      "learning_rate": 5.0971723006525965e-06,
      "loss": 0.1413,
      "step": 3775
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8920882927959981,
      "learning_rate": 5.094988919209652e-06,
      "loss": 0.157,
      "step": 3776
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7132816950138048,
      "learning_rate": 5.0928055196469426e-06,
      "loss": 0.1521,
      "step": 3777
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7651111317849348,
      "learning_rate": 5.090622102380966e-06,
      "loss": 0.1261,
      "step": 3778
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9878036755759267,
      "learning_rate": 5.088438667828223e-06,
      "loss": 0.2148,
      "step": 3779
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0562284268517312,
      "learning_rate": 5.086255216405219e-06,
      "loss": 0.2493,
      "step": 3780
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9240016249332471,
      "learning_rate": 5.084071748528462e-06,
      "loss": 0.169,
      "step": 3781
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.155340908162839,
      "learning_rate": 5.081888264614462e-06,
      "loss": 0.2524,
      "step": 3782
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.735232596917756,
      "learning_rate": 5.079704765079733e-06,
      "loss": 0.1198,
      "step": 3783
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9793150313883279,
      "learning_rate": 5.077521250340791e-06,
      "loss": 0.1999,
      "step": 3784
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9258968967574076,
      "learning_rate": 5.07533772081416e-06,
      "loss": 0.2022,
      "step": 3785
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1438794534701744,
      "learning_rate": 5.07315417691636e-06,
      "loss": 0.1987,
      "step": 3786
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.952465190342942,
      "learning_rate": 5.070970619063915e-06,
      "loss": 0.1341,
      "step": 3787
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0929583721354423,
      "learning_rate": 5.068787047673356e-06,
      "loss": 0.2026,
      "step": 3788
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9194558924745534,
      "learning_rate": 5.06660346316121e-06,
      "loss": 0.1823,
      "step": 3789
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7250770611759731,
      "learning_rate": 5.0644198659440145e-06,
      "loss": 0.1112,
      "step": 3790
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.561716501723813,
      "learning_rate": 5.062236256438303e-06,
      "loss": 0.0948,
      "step": 3791
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0591681716700405,
      "learning_rate": 5.060052635060613e-06,
      "loss": 0.2446,
      "step": 3792
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9239851682394968,
      "learning_rate": 5.057869002227485e-06,
      "loss": 0.2056,
      "step": 3793
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.910095183679671,
      "learning_rate": 5.055685358355464e-06,
      "loss": 0.1916,
      "step": 3794
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9918258352961842,
      "learning_rate": 5.05350170386109e-06,
      "loss": 0.1829,
      "step": 3795
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9333594347896143,
      "learning_rate": 5.051318039160913e-06,
      "loss": 0.1986,
      "step": 3796
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0220792205052367,
      "learning_rate": 5.04913436467148e-06,
      "loss": 0.1743,
      "step": 3797
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9156249494242872,
      "learning_rate": 5.046950680809341e-06,
      "loss": 0.1502,
      "step": 3798
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1598782644666075,
      "learning_rate": 5.04476698799105e-06,
      "loss": 0.2143,
      "step": 3799
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0080687912066868,
      "learning_rate": 5.042583286633158e-06,
      "loss": 0.1428,
      "step": 3800
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0173780467143883,
      "learning_rate": 5.040399577152224e-06,
      "loss": 0.1394,
      "step": 3801
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7082613800838057,
      "learning_rate": 5.038215859964802e-06,
      "loss": 0.1444,
      "step": 3802
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8471817688986188,
      "learning_rate": 5.036032135487452e-06,
      "loss": 0.1885,
      "step": 3803
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9600291022842052,
      "learning_rate": 5.033848404136734e-06,
      "loss": 0.1616,
      "step": 3804
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8172052669151572,
      "learning_rate": 5.031664666329209e-06,
      "loss": 0.1664,
      "step": 3805
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1941085359999977,
      "learning_rate": 5.029480922481438e-06,
      "loss": 0.1946,
      "step": 3806
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.056428766836335,
      "learning_rate": 5.0272971730099865e-06,
      "loss": 0.154,
      "step": 3807
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8342434733433041,
      "learning_rate": 5.02511341833142e-06,
      "loss": 0.1741,
      "step": 3808
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3862723687186698,
      "learning_rate": 5.022929658862302e-06,
      "loss": 0.2148,
      "step": 3809
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.121466693619795,
      "learning_rate": 5.020745895019199e-06,
      "loss": 0.2209,
      "step": 3810
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9486743015538583,
      "learning_rate": 5.01856212721868e-06,
      "loss": 0.1749,
      "step": 3811
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0468454423294558,
      "learning_rate": 5.0163783558773104e-06,
      "loss": 0.2424,
      "step": 3812
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9140835784280135,
      "learning_rate": 5.014194581411663e-06,
      "loss": 0.1866,
      "step": 3813
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6697742984396392,
      "learning_rate": 5.0120108042383045e-06,
      "loss": 0.1678,
      "step": 3814
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9000300708090622,
      "learning_rate": 5.009827024773806e-06,
      "loss": 0.1411,
      "step": 3815
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9538338036268852,
      "learning_rate": 5.0076432434347375e-06,
      "loss": 0.1917,
      "step": 3816
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0396754732741675,
      "learning_rate": 5.00545946063767e-06,
      "loss": 0.1918,
      "step": 3817
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8175653621383379,
      "learning_rate": 5.003275676799173e-06,
      "loss": 0.1417,
      "step": 3818
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9727316629977372,
      "learning_rate": 5.00109189233582e-06,
      "loss": 0.1915,
      "step": 3819
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7638908161671571,
      "learning_rate": 4.99890810766418e-06,
      "loss": 0.1433,
      "step": 3820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8335404455816515,
      "learning_rate": 4.996724323200827e-06,
      "loss": 0.1518,
      "step": 3821
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8354504910341995,
      "learning_rate": 4.994540539362331e-06,
      "loss": 0.1585,
      "step": 3822
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.171101342695506,
      "learning_rate": 4.992356756565264e-06,
      "loss": 0.1981,
      "step": 3823
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.49215697852619444,
      "learning_rate": 4.990172975226195e-06,
      "loss": 0.1228,
      "step": 3824
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2023057132169046,
      "learning_rate": 4.987989195761696e-06,
      "loss": 0.2238,
      "step": 3825
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5834259218510427,
      "learning_rate": 4.985805418588339e-06,
      "loss": 0.0965,
      "step": 3826
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8290979221449061,
      "learning_rate": 4.983621644122691e-06,
      "loss": 0.1589,
      "step": 3827
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9231875261876668,
      "learning_rate": 4.981437872781323e-06,
      "loss": 0.1895,
      "step": 3828
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7912589633855552,
      "learning_rate": 4.979254104980803e-06,
      "loss": 0.1633,
      "step": 3829
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.664169241479277,
      "learning_rate": 4.9770703411377005e-06,
      "loss": 0.1274,
      "step": 3830
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8119496474317474,
      "learning_rate": 4.974886581668581e-06,
      "loss": 0.1582,
      "step": 3831
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9496594922505179,
      "learning_rate": 4.9727028269900135e-06,
      "loss": 0.1853,
      "step": 3832
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8643991226422142,
      "learning_rate": 4.970519077518563e-06,
      "loss": 0.2108,
      "step": 3833
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8888097753339053,
      "learning_rate": 4.968335333670792e-06,
      "loss": 0.161,
      "step": 3834
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6702436629217124,
      "learning_rate": 4.966151595863267e-06,
      "loss": 0.0704,
      "step": 3835
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8828507512999598,
      "learning_rate": 4.963967864512549e-06,
      "loss": 0.16,
      "step": 3836
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7685815868248181,
      "learning_rate": 4.9617841400351994e-06,
      "loss": 0.1684,
      "step": 3837
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3278373412241034,
      "learning_rate": 4.9596004228477775e-06,
      "loss": 0.2339,
      "step": 3838
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2252368507395326,
      "learning_rate": 4.9574167133668425e-06,
      "loss": 0.1669,
      "step": 3839
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.819353301957595,
      "learning_rate": 4.9552330120089535e-06,
      "loss": 0.1684,
      "step": 3840
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.918986272952228,
      "learning_rate": 4.953049319190662e-06,
      "loss": 0.1619,
      "step": 3841
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.282288024319792,
      "learning_rate": 4.95086563532852e-06,
      "loss": 0.2363,
      "step": 3842
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1785767737122321,
      "learning_rate": 4.948681960839088e-06,
      "loss": 0.1969,
      "step": 3843
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8795750540825493,
      "learning_rate": 4.946498296138911e-06,
      "loss": 0.1285,
      "step": 3844
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9809214891555782,
      "learning_rate": 4.944314641644537e-06,
      "loss": 0.2023,
      "step": 3845
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8960981582942757,
      "learning_rate": 4.942130997772515e-06,
      "loss": 0.215,
      "step": 3846
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8239846397827746,
      "learning_rate": 4.939947364939388e-06,
      "loss": 0.1283,
      "step": 3847
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1177728299646266,
      "learning_rate": 4.937763743561699e-06,
      "loss": 0.2586,
      "step": 3848
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6966964141432771,
      "learning_rate": 4.935580134055986e-06,
      "loss": 0.1059,
      "step": 3849
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9881611671501764,
      "learning_rate": 4.933396536838791e-06,
      "loss": 0.2216,
      "step": 3850
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.905856216067459,
      "learning_rate": 4.931212952326646e-06,
      "loss": 0.1816,
      "step": 3851
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2350319586946417,
      "learning_rate": 4.929029380936087e-06,
      "loss": 0.2421,
      "step": 3852
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8690823772037858,
      "learning_rate": 4.926845823083643e-06,
      "loss": 0.1481,
      "step": 3853
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8874150227745957,
      "learning_rate": 4.92466227918584e-06,
      "loss": 0.1573,
      "step": 3854
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8554332111186558,
      "learning_rate": 4.922478749659208e-06,
      "loss": 0.1696,
      "step": 3855
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7882162099885286,
      "learning_rate": 4.920295234920269e-06,
      "loss": 0.1242,
      "step": 3856
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9158407549519675,
      "learning_rate": 4.91811173538554e-06,
      "loss": 0.1771,
      "step": 3857
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8213150441271284,
      "learning_rate": 4.91592825147154e-06,
      "loss": 0.1489,
      "step": 3858
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9216459569130909,
      "learning_rate": 4.913744783594783e-06,
      "loss": 0.1803,
      "step": 3859
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8033096340138559,
      "learning_rate": 4.911561332171779e-06,
      "loss": 0.1902,
      "step": 3860
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8523357117268346,
      "learning_rate": 4.909377897619036e-06,
      "loss": 0.1615,
      "step": 3861
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9353447964802794,
      "learning_rate": 4.907194480353059e-06,
      "loss": 0.1701,
      "step": 3862
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9897257452273038,
      "learning_rate": 4.9050110807903495e-06,
      "loss": 0.2112,
      "step": 3863
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0565965899154968,
      "learning_rate": 4.902827699347406e-06,
      "loss": 0.2284,
      "step": 3864
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9842966954383243,
      "learning_rate": 4.900644336440719e-06,
      "loss": 0.1784,
      "step": 3865
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0288535978056468,
      "learning_rate": 4.8984609924867845e-06,
      "loss": 0.189,
      "step": 3866
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9740721617788877,
      "learning_rate": 4.896277667902089e-06,
      "loss": 0.2244,
      "step": 3867
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9279381827293477,
      "learning_rate": 4.894094363103114e-06,
      "loss": 0.138,
      "step": 3868
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1279175366977847,
      "learning_rate": 4.89191107850634e-06,
      "loss": 0.1645,
      "step": 3869
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0306371738153002,
      "learning_rate": 4.889727814528245e-06,
      "loss": 0.1963,
      "step": 3870
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.845423663031792,
      "learning_rate": 4.8875445715852974e-06,
      "loss": 0.1431,
      "step": 3871
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.06912939847565,
      "learning_rate": 4.885361350093968e-06,
      "loss": 0.159,
      "step": 3872
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7664634001361564,
      "learning_rate": 4.883178150470719e-06,
      "loss": 0.1188,
      "step": 3873
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0879843258024027,
      "learning_rate": 4.8809949731320124e-06,
      "loss": 0.2073,
      "step": 3874
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9997711372013901,
      "learning_rate": 4.878811818494301e-06,
      "loss": 0.2016,
      "step": 3875
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7308478655475236,
      "learning_rate": 4.876628686974038e-06,
      "loss": 0.1245,
      "step": 3876
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8351028527851251,
      "learning_rate": 4.87444557898767e-06,
      "loss": 0.1313,
      "step": 3877
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8774215195067292,
      "learning_rate": 4.872262494951639e-06,
      "loss": 0.1502,
      "step": 3878
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0109846043409234,
      "learning_rate": 4.870079435282382e-06,
      "loss": 0.1723,
      "step": 3879
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8719588432564775,
      "learning_rate": 4.8678964003963336e-06,
      "loss": 0.1955,
      "step": 3880
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7222409609414056,
      "learning_rate": 4.8657133907099215e-06,
      "loss": 0.1677,
      "step": 3881
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8908594117517278,
      "learning_rate": 4.863530406639568e-06,
      "loss": 0.2171,
      "step": 3882
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7760040396997208,
      "learning_rate": 4.861347448601694e-06,
      "loss": 0.1613,
      "step": 3883
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8974476138476687,
      "learning_rate": 4.859164517012711e-06,
      "loss": 0.1569,
      "step": 3884
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1518958204108947,
      "learning_rate": 4.856981612289029e-06,
      "loss": 0.2268,
      "step": 3885
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0619909983189082,
      "learning_rate": 4.854798734847052e-06,
      "loss": 0.1883,
      "step": 3886
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0001360228366571,
      "learning_rate": 4.852615885103175e-06,
      "loss": 0.2275,
      "step": 3887
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8872152488709741,
      "learning_rate": 4.850433063473795e-06,
      "loss": 0.1599,
      "step": 3888
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.837091214744943,
      "learning_rate": 4.848250270375298e-06,
      "loss": 0.1044,
      "step": 3889
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7167698345522614,
      "learning_rate": 4.846067506224068e-06,
      "loss": 0.1589,
      "step": 3890
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8407250864937743,
      "learning_rate": 4.843884771436479e-06,
      "loss": 0.1412,
      "step": 3891
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8554436321969415,
      "learning_rate": 4.841702066428904e-06,
      "loss": 0.1816,
      "step": 3892
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6856523374574385,
      "learning_rate": 4.839519391617708e-06,
      "loss": 0.1211,
      "step": 3893
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9310956211449959,
      "learning_rate": 4.83733674741925e-06,
      "loss": 0.1769,
      "step": 3894
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9322646242543944,
      "learning_rate": 4.835154134249883e-06,
      "loss": 0.1735,
      "step": 3895
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7801369193934391,
      "learning_rate": 4.832971552525957e-06,
      "loss": 0.1356,
      "step": 3896
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9306880129704641,
      "learning_rate": 4.830789002663813e-06,
      "loss": 0.1523,
      "step": 3897
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7746034376360663,
      "learning_rate": 4.828606485079787e-06,
      "loss": 0.1273,
      "step": 3898
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7581867379849405,
      "learning_rate": 4.826424000190207e-06,
      "loss": 0.1308,
      "step": 3899
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0522613901580788,
      "learning_rate": 4.8242415484113995e-06,
      "loss": 0.1884,
      "step": 3900
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8064230197129383,
      "learning_rate": 4.82205913015968e-06,
      "loss": 0.1345,
      "step": 3901
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7779891322021222,
      "learning_rate": 4.819876745851359e-06,
      "loss": 0.1476,
      "step": 3902
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.126807938196983,
      "learning_rate": 4.817694395902741e-06,
      "loss": 0.2002,
      "step": 3903
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0695085496808847,
      "learning_rate": 4.815512080730125e-06,
      "loss": 0.1893,
      "step": 3904
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0269167445390677,
      "learning_rate": 4.813329800749799e-06,
      "loss": 0.1441,
      "step": 3905
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7329744240657049,
      "learning_rate": 4.811147556378051e-06,
      "loss": 0.1391,
      "step": 3906
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9205876881580598,
      "learning_rate": 4.808965348031156e-06,
      "loss": 0.1909,
      "step": 3907
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0097374705866675,
      "learning_rate": 4.806783176125385e-06,
      "loss": 0.2278,
      "step": 3908
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.195305329629581,
      "learning_rate": 4.8046010410770025e-06,
      "loss": 0.1945,
      "step": 3909
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9222578623671897,
      "learning_rate": 4.8024189433022635e-06,
      "loss": 0.2043,
      "step": 3910
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0489907398001412,
      "learning_rate": 4.800236883217421e-06,
      "loss": 0.1838,
      "step": 3911
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0039734674458207,
      "learning_rate": 4.798054861238715e-06,
      "loss": 0.1916,
      "step": 3912
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9018080623348476,
      "learning_rate": 4.795872877782381e-06,
      "loss": 0.1778,
      "step": 3913
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.690714923498507,
      "learning_rate": 4.7936909332646465e-06,
      "loss": 0.1359,
      "step": 3914
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9832511893317845,
      "learning_rate": 4.791509028101732e-06,
      "loss": 0.1679,
      "step": 3915
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8707693437230548,
      "learning_rate": 4.789327162709852e-06,
      "loss": 0.1451,
      "step": 3916
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.978804276263669,
      "learning_rate": 4.787145337505209e-06,
      "loss": 0.1633,
      "step": 3917
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1533126938669376,
      "learning_rate": 4.7849635529040025e-06,
      "loss": 0.2131,
      "step": 3918
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8609182321384063,
      "learning_rate": 4.782781809322421e-06,
      "loss": 0.1321,
      "step": 3919
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5563422856951902,
      "learning_rate": 4.780600107176648e-06,
      "loss": 0.0982,
      "step": 3920
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0122362703245025,
      "learning_rate": 4.778418446882855e-06,
      "loss": 0.1718,
      "step": 3921
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8132614666926253,
      "learning_rate": 4.776236828857211e-06,
      "loss": 0.1428,
      "step": 3922
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5526081679855167,
      "learning_rate": 4.774055253515875e-06,
      "loss": 0.127,
      "step": 3923
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8701672349767257,
      "learning_rate": 4.771873721274994e-06,
      "loss": 0.1609,
      "step": 3924
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9486011271070752,
      "learning_rate": 4.769692232550711e-06,
      "loss": 0.2094,
      "step": 3925
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8940973243291737,
      "learning_rate": 4.767510787759158e-06,
      "loss": 0.1758,
      "step": 3926
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9997571213025298,
      "learning_rate": 4.765329387316463e-06,
      "loss": 0.1769,
      "step": 3927
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9610928325188381,
      "learning_rate": 4.7631480316387395e-06,
      "loss": 0.17,
      "step": 3928
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8105875399896512,
      "learning_rate": 4.760966721142097e-06,
      "loss": 0.1702,
      "step": 3929
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7184382961834598,
      "learning_rate": 4.758785456242636e-06,
      "loss": 0.1073,
      "step": 3930
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8388050098996329,
      "learning_rate": 4.7566042373564445e-06,
      "loss": 0.1256,
      "step": 3931
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.778159926874574,
      "learning_rate": 4.754423064899605e-06,
      "loss": 0.1607,
      "step": 3932
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9129985058065117,
      "learning_rate": 4.752241939288193e-06,
      "loss": 0.2019,
      "step": 3933
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2180707685643726,
      "learning_rate": 4.7500608609382715e-06,
      "loss": 0.2343,
      "step": 3934
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8363574385680551,
      "learning_rate": 4.7478798302658944e-06,
      "loss": 0.175,
      "step": 3935
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7747051367198476,
      "learning_rate": 4.74569884768711e-06,
      "loss": 0.1227,
      "step": 3936
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.825484052809516,
      "learning_rate": 4.7435179136179525e-06,
      "loss": 0.2104,
      "step": 3937
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.141624365481997,
      "learning_rate": 4.741337028474452e-06,
      "loss": 0.193,
      "step": 3938
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8373712843838721,
      "learning_rate": 4.739156192672624e-06,
      "loss": 0.0907,
      "step": 3939
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8158017141345678,
      "learning_rate": 4.736975406628481e-06,
      "loss": 0.1521,
      "step": 3940
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9228094092804496,
      "learning_rate": 4.734794670758019e-06,
      "loss": 0.1609,
      "step": 3941
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8285708789042918,
      "learning_rate": 4.73261398547723e-06,
      "loss": 0.1526,
      "step": 3942
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1709764035782442,
      "learning_rate": 4.730433351202089e-06,
      "loss": 0.2283,
      "step": 3943
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8002350601611758,
      "learning_rate": 4.728252768348574e-06,
      "loss": 0.1928,
      "step": 3944
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.085061167796444,
      "learning_rate": 4.726072237332644e-06,
      "loss": 0.17,
      "step": 3945
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.08487666280776,
      "learning_rate": 4.723891758570247e-06,
      "loss": 0.1542,
      "step": 3946
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9051892534954081,
      "learning_rate": 4.721711332477324e-06,
      "loss": 0.1914,
      "step": 3947
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8930246522716515,
      "learning_rate": 4.719530959469807e-06,
      "loss": 0.1775,
      "step": 3948
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0204484269020921,
      "learning_rate": 4.717350639963616e-06,
      "loss": 0.2059,
      "step": 3949
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9597416738162747,
      "learning_rate": 4.715170374374662e-06,
      "loss": 0.1909,
      "step": 3950
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.034549496385055,
      "learning_rate": 4.712990163118844e-06,
      "loss": 0.2302,
      "step": 3951
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9671694330706364,
      "learning_rate": 4.710810006612052e-06,
      "loss": 0.1915,
      "step": 3952
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.686875115775582,
      "learning_rate": 4.708629905270166e-06,
      "loss": 0.1535,
      "step": 3953
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.91762685678245,
      "learning_rate": 4.706449859509055e-06,
      "loss": 0.1828,
      "step": 3954
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8524657677565112,
      "learning_rate": 4.704269869744574e-06,
      "loss": 0.1838,
      "step": 3955
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8948072380999594,
      "learning_rate": 4.7020899363925745e-06,
      "loss": 0.1549,
      "step": 3956
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6935852370605616,
      "learning_rate": 4.699910059868892e-06,
      "loss": 0.1484,
      "step": 3957
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9164110305337375,
      "learning_rate": 4.697730240589352e-06,
      "loss": 0.1571,
      "step": 3958
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8564251540772106,
      "learning_rate": 4.695550478969769e-06,
      "loss": 0.1377,
      "step": 3959
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8496147896821714,
      "learning_rate": 4.6933707754259485e-06,
      "loss": 0.1766,
      "step": 3960
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9251392386733661,
      "learning_rate": 4.691191130373682e-06,
      "loss": 0.1313,
      "step": 3961
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.656558428729138,
      "learning_rate": 4.689011544228752e-06,
      "loss": 0.1364,
      "step": 3962
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9569428186165508,
      "learning_rate": 4.686832017406929e-06,
      "loss": 0.154,
      "step": 3963
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.151582101459163,
      "learning_rate": 4.684652550323972e-06,
      "loss": 0.2111,
      "step": 3964
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6208502537432142,
      "learning_rate": 4.682473143395626e-06,
      "loss": 0.0891,
      "step": 3965
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1200321627292125,
      "learning_rate": 4.68029379703763e-06,
      "loss": 0.1954,
      "step": 3966
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9096408166360085,
      "learning_rate": 4.678114511665709e-06,
      "loss": 0.1318,
      "step": 3967
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8433072798144762,
      "learning_rate": 4.675935287695577e-06,
      "loss": 0.1486,
      "step": 3968
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7623203418848972,
      "learning_rate": 4.673756125542934e-06,
      "loss": 0.1302,
      "step": 3969
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9144559463285483,
      "learning_rate": 4.67157702562347e-06,
      "loss": 0.1443,
      "step": 3970
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8519460747815207,
      "learning_rate": 4.669397988352862e-06,
      "loss": 0.1497,
      "step": 3971
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8002424926756003,
      "learning_rate": 4.667219014146775e-06,
      "loss": 0.1662,
      "step": 3972
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.84366648117451,
      "learning_rate": 4.665040103420865e-06,
      "loss": 0.2015,
      "step": 3973
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0750896246863462,
      "learning_rate": 4.662861256590772e-06,
      "loss": 0.2256,
      "step": 3974
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0145208822731813,
      "learning_rate": 4.660682474072127e-06,
      "loss": 0.1532,
      "step": 3975
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9007004739412044,
      "learning_rate": 4.658503756280546e-06,
      "loss": 0.2014,
      "step": 3976
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7756406183127332,
      "learning_rate": 4.656325103631632e-06,
      "loss": 0.1412,
      "step": 3977
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1894541493237256,
      "learning_rate": 4.654146516540982e-06,
      "loss": 0.2161,
      "step": 3978
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.848021161112219,
      "learning_rate": 4.651967995424173e-06,
      "loss": 0.1476,
      "step": 3979
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7630845014691678,
      "learning_rate": 4.649789540696772e-06,
      "loss": 0.1432,
      "step": 3980
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0862205465161958,
      "learning_rate": 4.647611152774335e-06,
      "loss": 0.1571,
      "step": 3981
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.036127420357128,
      "learning_rate": 4.645432832072404e-06,
      "loss": 0.2073,
      "step": 3982
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9749022739896953,
      "learning_rate": 4.643254579006506e-06,
      "loss": 0.2198,
      "step": 3983
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8560423676473358,
      "learning_rate": 4.641076393992159e-06,
      "loss": 0.1314,
      "step": 3984
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0194804981069274,
      "learning_rate": 4.638898277444866e-06,
      "loss": 0.1955,
      "step": 3985
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0405580915572261,
      "learning_rate": 4.636720229780117e-06,
      "loss": 0.2142,
      "step": 3986
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8658518720618914,
      "learning_rate": 4.63454225141339e-06,
      "loss": 0.1471,
      "step": 3987
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5857244969691943,
      "learning_rate": 4.632364342760145e-06,
      "loss": 0.1045,
      "step": 3988
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9876477633595183,
      "learning_rate": 4.630186504235836e-06,
      "loss": 0.1654,
      "step": 3989
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9704478777980777,
      "learning_rate": 4.6280087362559004e-06,
      "loss": 0.2277,
      "step": 3990
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9200394975248318,
      "learning_rate": 4.625831039235761e-06,
      "loss": 0.1647,
      "step": 3991
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.789408619787891,
      "learning_rate": 4.623653413590827e-06,
      "loss": 0.1259,
      "step": 3992
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4421378371323545,
      "learning_rate": 4.621475859736498e-06,
      "loss": 0.2243,
      "step": 3993
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8805704069820712,
      "learning_rate": 4.619298378088152e-06,
      "loss": 0.1556,
      "step": 3994
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9487497213141076,
      "learning_rate": 4.617120969061161e-06,
      "loss": 0.1979,
      "step": 3995
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8584474035430331,
      "learning_rate": 4.61494363307088e-06,
      "loss": 0.1663,
      "step": 3996
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9812679458656697,
      "learning_rate": 4.61276637053265e-06,
      "loss": 0.154,
      "step": 3997
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6311868090961761,
      "learning_rate": 4.6105891818617976e-06,
      "loss": 0.152,
      "step": 3998
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.912480596398402,
      "learning_rate": 4.608412067473637e-06,
      "loss": 0.1546,
      "step": 3999
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9884926788821717,
      "learning_rate": 4.606235027783466e-06,
      "loss": 0.1826,
      "step": 4000
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2319768451719282,
      "learning_rate": 4.60405806320657e-06,
      "loss": 0.2191,
      "step": 4001
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8344184811739532,
      "learning_rate": 4.6018811741582194e-06,
      "loss": 0.153,
      "step": 4002
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4296759528869725,
      "learning_rate": 4.599704361053672e-06,
      "loss": 0.2204,
      "step": 4003
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.00314379566071,
      "learning_rate": 4.597527624308166e-06,
      "loss": 0.188,
      "step": 4004
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0838807374529467,
      "learning_rate": 4.595350964336931e-06,
      "loss": 0.1729,
      "step": 4005
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7771085431627314,
      "learning_rate": 4.593174381555176e-06,
      "loss": 0.1488,
      "step": 4006
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2037287836327564,
      "learning_rate": 4.590997876378103e-06,
      "loss": 0.2149,
      "step": 4007
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8815845498533571,
      "learning_rate": 4.588821449220892e-06,
      "loss": 0.0974,
      "step": 4008
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9360401780098738,
      "learning_rate": 4.586645100498711e-06,
      "loss": 0.1919,
      "step": 4009
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0413401041512087,
      "learning_rate": 4.584468830626715e-06,
      "loss": 0.191,
      "step": 4010
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8521986721832869,
      "learning_rate": 4.582292640020037e-06,
      "loss": 0.1628,
      "step": 4011
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8698653303107923,
      "learning_rate": 4.580116529093805e-06,
      "loss": 0.148,
      "step": 4012
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6259978829698777,
      "learning_rate": 4.577940498263124e-06,
      "loss": 0.1131,
      "step": 4013
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9091311548318229,
      "learning_rate": 4.575764547943087e-06,
      "loss": 0.1763,
      "step": 4014
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8869144228550706,
      "learning_rate": 4.573588678548772e-06,
      "loss": 0.1565,
      "step": 4015
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2896149328545616,
      "learning_rate": 4.571412890495239e-06,
      "loss": 0.2172,
      "step": 4016
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2176540061192198,
      "learning_rate": 4.569237184197534e-06,
      "loss": 0.2123,
      "step": 4017
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.710202093276334,
      "learning_rate": 4.567061560070687e-06,
      "loss": 0.1626,
      "step": 4018
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0601108740801823,
      "learning_rate": 4.564886018529714e-06,
      "loss": 0.2425,
      "step": 4019
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8956149030683478,
      "learning_rate": 4.5627105599896126e-06,
      "loss": 0.1672,
      "step": 4020
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0704603689730194,
      "learning_rate": 4.5605351848653665e-06,
      "loss": 0.2132,
      "step": 4021
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9411502925122591,
      "learning_rate": 4.5583598935719416e-06,
      "loss": 0.1501,
      "step": 4022
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2342280117433357,
      "learning_rate": 4.556184686524293e-06,
      "loss": 0.2331,
      "step": 4023
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1207839596553553,
      "learning_rate": 4.5540095641373535e-06,
      "loss": 0.2243,
      "step": 4024
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9410658471447971,
      "learning_rate": 4.55183452682604e-06,
      "loss": 0.1644,
      "step": 4025
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8680724627406653,
      "learning_rate": 4.5496595750052595e-06,
      "loss": 0.1657,
      "step": 4026
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9823308432112777,
      "learning_rate": 4.547484709089895e-06,
      "loss": 0.174,
      "step": 4027
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8499308210466723,
      "learning_rate": 4.545309929494817e-06,
      "loss": 0.1652,
      "step": 4028
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.924271049703917,
      "learning_rate": 4.543135236634881e-06,
      "loss": 0.2032,
      "step": 4029
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.743848513086163,
      "learning_rate": 4.540960630924923e-06,
      "loss": 0.1598,
      "step": 4030
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0212506762756084,
      "learning_rate": 4.5387861127797625e-06,
      "loss": 0.2116,
      "step": 4031
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6010012224961518,
      "learning_rate": 4.5366116826142045e-06,
      "loss": 0.1248,
      "step": 4032
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1908766260894346,
      "learning_rate": 4.534437340843035e-06,
      "loss": 0.2708,
      "step": 4033
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8290303909564103,
      "learning_rate": 4.532263087881025e-06,
      "loss": 0.1112,
      "step": 4034
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4611738477111491,
      "learning_rate": 4.530088924142927e-06,
      "loss": 0.0934,
      "step": 4035
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8513272386388259,
      "learning_rate": 4.527914850043478e-06,
      "loss": 0.1736,
      "step": 4036
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9838473054440287,
      "learning_rate": 4.525740865997396e-06,
      "loss": 0.1688,
      "step": 4037
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1382452705436044,
      "learning_rate": 4.523566972419383e-06,
      "loss": 0.2355,
      "step": 4038
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9535398031508988,
      "learning_rate": 4.521393169724124e-06,
      "loss": 0.1676,
      "step": 4039
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8754414034418576,
      "learning_rate": 4.519219458326288e-06,
      "loss": 0.1552,
      "step": 4040
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.213750850031794,
      "learning_rate": 4.517045838640521e-06,
      "loss": 0.2415,
      "step": 4041
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0646903186835386,
      "learning_rate": 4.514872311081458e-06,
      "loss": 0.1736,
      "step": 4042
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1345127328588112,
      "learning_rate": 4.512698876063713e-06,
      "loss": 0.2213,
      "step": 4043
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8770933454780391,
      "learning_rate": 4.51052553400188e-06,
      "loss": 0.1874,
      "step": 4044
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9781070597347284,
      "learning_rate": 4.508352285310545e-06,
      "loss": 0.2123,
      "step": 4045
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1899006760999102,
      "learning_rate": 4.5061791304042664e-06,
      "loss": 0.231,
      "step": 4046
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6514754645011095,
      "learning_rate": 4.504006069697586e-06,
      "loss": 0.1015,
      "step": 4047
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7135806317198017,
      "learning_rate": 4.501833103605032e-06,
      "loss": 0.1401,
      "step": 4048
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8036282012724237,
      "learning_rate": 4.499660232541112e-06,
      "loss": 0.1306,
      "step": 4049
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7320451886726567,
      "learning_rate": 4.497487456920313e-06,
      "loss": 0.0987,
      "step": 4050
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1606528760063328,
      "learning_rate": 4.495314777157109e-06,
      "loss": 0.1983,
      "step": 4051
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8239643162407717,
      "learning_rate": 4.493142193665952e-06,
      "loss": 0.1767,
      "step": 4052
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0674063098011812,
      "learning_rate": 4.4909697068612765e-06,
      "loss": 0.2433,
      "step": 4053
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0942715865190606,
      "learning_rate": 4.4887973171574994e-06,
      "loss": 0.2112,
      "step": 4054
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9280973661799354,
      "learning_rate": 4.486625024969018e-06,
      "loss": 0.1865,
      "step": 4055
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7742259707084385,
      "learning_rate": 4.48445283071021e-06,
      "loss": 0.1711,
      "step": 4056
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8807524525074715,
      "learning_rate": 4.4822807347954385e-06,
      "loss": 0.1547,
      "step": 4057
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8585319803567698,
      "learning_rate": 4.480108737639044e-06,
      "loss": 0.1574,
      "step": 4058
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1707702966288704,
      "learning_rate": 4.4779368396553505e-06,
      "loss": 0.214,
      "step": 4059
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9488487322160449,
      "learning_rate": 4.4757650412586594e-06,
      "loss": 0.1605,
      "step": 4060
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7100787750552442,
      "learning_rate": 4.4735933428632575e-06,
      "loss": 0.1399,
      "step": 4061
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7656167150022092,
      "learning_rate": 4.47142174488341e-06,
      "loss": 0.1423,
      "step": 4062
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8259217139523918,
      "learning_rate": 4.469250247733365e-06,
      "loss": 0.1951,
      "step": 4063
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8915774171501172,
      "learning_rate": 4.467078851827346e-06,
      "loss": 0.1634,
      "step": 4064
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9790131138045057,
      "learning_rate": 4.464907557579565e-06,
      "loss": 0.1824,
      "step": 4065
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1060879137307627,
      "learning_rate": 4.46273636540421e-06,
      "loss": 0.2469,
      "step": 4066
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8190436448632614,
      "learning_rate": 4.460565275715447e-06,
      "loss": 0.144,
      "step": 4067
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8297031286720372,
      "learning_rate": 4.458394288927431e-06,
      "loss": 0.1658,
      "step": 4068
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8148795311610765,
      "learning_rate": 4.45622340545429e-06,
      "loss": 0.1274,
      "step": 4069
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8815679485791238,
      "learning_rate": 4.454052625710134e-06,
      "loss": 0.1771,
      "step": 4070
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8392646544445846,
      "learning_rate": 4.4518819501090534e-06,
      "loss": 0.1718,
      "step": 4071
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2386216182143632,
      "learning_rate": 4.449711379065119e-06,
      "loss": 0.196,
      "step": 4072
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8666623255465813,
      "learning_rate": 4.447540912992382e-06,
      "loss": 0.1935,
      "step": 4073
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9212923115593503,
      "learning_rate": 4.4453705523048714e-06,
      "loss": 0.1598,
      "step": 4074
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8066923934859511,
      "learning_rate": 4.443200297416601e-06,
      "loss": 0.1296,
      "step": 4075
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.173032827648446,
      "learning_rate": 4.4410301487415596e-06,
      "loss": 0.2146,
      "step": 4076
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9098098046078548,
      "learning_rate": 4.438860106693717e-06,
      "loss": 0.1441,
      "step": 4077
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9043190334816884,
      "learning_rate": 4.436690171687024e-06,
      "loss": 0.1752,
      "step": 4078
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8307129158130062,
      "learning_rate": 4.434520344135409e-06,
      "loss": 0.1823,
      "step": 4079
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8929591439607125,
      "learning_rate": 4.432350624452783e-06,
      "loss": 0.1785,
      "step": 4080
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0408125219348519,
      "learning_rate": 4.430181013053032e-06,
      "loss": 0.1609,
      "step": 4081
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9297187360258055,
      "learning_rate": 4.428011510350025e-06,
      "loss": 0.154,
      "step": 4082
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7433803096602359,
      "learning_rate": 4.425842116757609e-06,
      "loss": 0.1529,
      "step": 4083
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9717912414507244,
      "learning_rate": 4.4236728326896096e-06,
      "loss": 0.1872,
      "step": 4084
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8394959094484026,
      "learning_rate": 4.421503658559832e-06,
      "loss": 0.1743,
      "step": 4085
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4534737426942952,
      "learning_rate": 4.419334594782062e-06,
      "loss": 0.2925,
      "step": 4086
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9105984452734304,
      "learning_rate": 4.417165641770063e-06,
      "loss": 0.1903,
      "step": 4087
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0238123248831859,
      "learning_rate": 4.414996799937575e-06,
      "loss": 0.1351,
      "step": 4088
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2862331839588483,
      "learning_rate": 4.412828069698317e-06,
      "loss": 0.2205,
      "step": 4089
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0727312412957997,
      "learning_rate": 4.410659451465995e-06,
      "loss": 0.2015,
      "step": 4090
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1601423032698672,
      "learning_rate": 4.408490945654282e-06,
      "loss": 0.1821,
      "step": 4091
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9093717169883975,
      "learning_rate": 4.406322552676838e-06,
      "loss": 0.1582,
      "step": 4092
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8728885079446868,
      "learning_rate": 4.404154272947298e-06,
      "loss": 0.1539,
      "step": 4093
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8511054975265999,
      "learning_rate": 4.401986106879273e-06,
      "loss": 0.1169,
      "step": 4094
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0234634134188116,
      "learning_rate": 4.3998180548863576e-06,
      "loss": 0.21,
      "step": 4095
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2232808696113255,
      "learning_rate": 4.397650117382122e-06,
      "loss": 0.184,
      "step": 4096
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9202050380194406,
      "learning_rate": 4.395482294780112e-06,
      "loss": 0.1423,
      "step": 4097
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9066138048309804,
      "learning_rate": 4.393314587493857e-06,
      "loss": 0.1723,
      "step": 4098
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7230675468656245,
      "learning_rate": 4.3911469959368605e-06,
      "loss": 0.0809,
      "step": 4099
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7561007991005488,
      "learning_rate": 4.388979520522605e-06,
      "loss": 0.1164,
      "step": 4100
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.910139865294084,
      "learning_rate": 4.3868121616645495e-06,
      "loss": 0.1868,
      "step": 4101
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8391222067305387,
      "learning_rate": 4.384644919776134e-06,
      "loss": 0.1295,
      "step": 4102
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8018865821232832,
      "learning_rate": 4.3824777952707736e-06,
      "loss": 0.1527,
      "step": 4103
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7188985436334914,
      "learning_rate": 4.3803107885618615e-06,
      "loss": 0.0894,
      "step": 4104
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6060783364462642,
      "learning_rate": 4.378143900062769e-06,
      "loss": 0.1408,
      "step": 4105
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8690236803189219,
      "learning_rate": 4.375977130186844e-06,
      "loss": 0.1912,
      "step": 4106
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9334437571599191,
      "learning_rate": 4.373810479347412e-06,
      "loss": 0.1772,
      "step": 4107
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9566805439908534,
      "learning_rate": 4.371643947957777e-06,
      "loss": 0.1707,
      "step": 4108
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8567539723634645,
      "learning_rate": 4.369477536431218e-06,
      "loss": 0.1267,
      "step": 4109
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7966280572862711,
      "learning_rate": 4.367311245180993e-06,
      "loss": 0.1296,
      "step": 4110
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7822870647851352,
      "learning_rate": 4.365145074620336e-06,
      "loss": 0.1827,
      "step": 4111
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9100850345146164,
      "learning_rate": 4.362979025162456e-06,
      "loss": 0.1651,
      "step": 4112
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8190390069536524,
      "learning_rate": 4.360813097220546e-06,
      "loss": 0.1542,
      "step": 4113
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9233266994912741,
      "learning_rate": 4.358647291207769e-06,
      "loss": 0.1944,
      "step": 4114
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2784655416391657,
      "learning_rate": 4.3564816075372675e-06,
      "loss": 0.2786,
      "step": 4115
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9185110289972842,
      "learning_rate": 4.354316046622159e-06,
      "loss": 0.1596,
      "step": 4116
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8450131800236639,
      "learning_rate": 4.35215060887554e-06,
      "loss": 0.1685,
      "step": 4117
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8418572110329651,
      "learning_rate": 4.349985294710478e-06,
      "loss": 0.1565,
      "step": 4118
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7106367375044466,
      "learning_rate": 4.347820104540026e-06,
      "loss": 0.174,
      "step": 4119
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7994350939269989,
      "learning_rate": 4.345655038777206e-06,
      "loss": 0.1674,
      "step": 4120
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0649610569256458,
      "learning_rate": 4.343490097835018e-06,
      "loss": 0.1913,
      "step": 4121
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0890211867676458,
      "learning_rate": 4.34132528212644e-06,
      "loss": 0.2145,
      "step": 4122
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2334438403606458,
      "learning_rate": 4.339160592064424e-06,
      "loss": 0.2073,
      "step": 4123
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0511043255951023,
      "learning_rate": 4.336996028061901e-06,
      "loss": 0.1889,
      "step": 4124
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9951541062244113,
      "learning_rate": 4.334831590531773e-06,
      "loss": 0.185,
      "step": 4125
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1191612501267645,
      "learning_rate": 4.3326672798869236e-06,
      "loss": 0.2358,
      "step": 4126
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1551304217625893,
      "learning_rate": 4.330503096540208e-06,
      "loss": 0.2372,
      "step": 4127
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0458751992033648,
      "learning_rate": 4.328339040904458e-06,
      "loss": 0.1853,
      "step": 4128
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9467668000196772,
      "learning_rate": 4.326175113392483e-06,
      "loss": 0.1865,
      "step": 4129
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8415132581654822,
      "learning_rate": 4.324011314417064e-06,
      "loss": 0.1762,
      "step": 4130
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9918147203127362,
      "learning_rate": 4.321847644390964e-06,
      "loss": 0.2097,
      "step": 4131
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1165682946278441,
      "learning_rate": 4.3196841037269135e-06,
      "loss": 0.1775,
      "step": 4132
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.220155310058599,
      "learning_rate": 4.317520692837625e-06,
      "loss": 0.2123,
      "step": 4133
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0481163972883267,
      "learning_rate": 4.315357412135779e-06,
      "loss": 0.1836,
      "step": 4134
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0451749854735635,
      "learning_rate": 4.313194262034041e-06,
      "loss": 0.1799,
      "step": 4135
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9921384299744181,
      "learning_rate": 4.3110312429450435e-06,
      "loss": 0.1638,
      "step": 4136
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.060426636121653,
      "learning_rate": 4.308868355281397e-06,
      "loss": 0.2068,
      "step": 4137
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7714851251276741,
      "learning_rate": 4.306705599455687e-06,
      "loss": 0.1114,
      "step": 4138
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9875981941134427,
      "learning_rate": 4.304542975880474e-06,
      "loss": 0.174,
      "step": 4139
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9680101521124876,
      "learning_rate": 4.302380484968293e-06,
      "loss": 0.1451,
      "step": 4140
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9293515932174983,
      "learning_rate": 4.30021812713165e-06,
      "loss": 0.1947,
      "step": 4141
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8566131032725971,
      "learning_rate": 4.2980559027830336e-06,
      "loss": 0.1654,
      "step": 4142
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.784295075098625,
      "learning_rate": 4.295893812334899e-06,
      "loss": 0.1212,
      "step": 4143
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9253288024538547,
      "learning_rate": 4.293731856199682e-06,
      "loss": 0.1388,
      "step": 4144
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.294931748706023,
      "learning_rate": 4.2915700347897854e-06,
      "loss": 0.2606,
      "step": 4145
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9634026751431582,
      "learning_rate": 4.2894083485175974e-06,
      "loss": 0.1502,
      "step": 4146
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.089398376493181,
      "learning_rate": 4.28724679779547e-06,
      "loss": 0.2056,
      "step": 4147
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1726861012422092,
      "learning_rate": 4.2850853830357345e-06,
      "loss": 0.2124,
      "step": 4148
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1134002541294197,
      "learning_rate": 4.282924104650694e-06,
      "loss": 0.216,
      "step": 4149
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7213529751742481,
      "learning_rate": 4.280762963052628e-06,
      "loss": 0.171,
      "step": 4150
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7949120331478577,
      "learning_rate": 4.278601958653788e-06,
      "loss": 0.1044,
      "step": 4151
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8616212886328642,
      "learning_rate": 4.2764410918664e-06,
      "loss": 0.1738,
      "step": 4152
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.006976092714174,
      "learning_rate": 4.274280363102663e-06,
      "loss": 0.1873,
      "step": 4153
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8536662913922969,
      "learning_rate": 4.27211977277475e-06,
      "loss": 0.1153,
      "step": 4154
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9698801349946972,
      "learning_rate": 4.269959321294809e-06,
      "loss": 0.1958,
      "step": 4155
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9552656297363844,
      "learning_rate": 4.267799009074961e-06,
      "loss": 0.1577,
      "step": 4156
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8510012443720653,
      "learning_rate": 4.265638836527297e-06,
      "loss": 0.1724,
      "step": 4157
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6031416670002322,
      "learning_rate": 4.263478804063888e-06,
      "loss": 0.166,
      "step": 4158
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9978147630216729,
      "learning_rate": 4.261318912096772e-06,
      "loss": 0.1404,
      "step": 4159
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1972300468722703,
      "learning_rate": 4.259159161037964e-06,
      "loss": 0.2221,
      "step": 4160
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7393346487078943,
      "learning_rate": 4.25699955129945e-06,
      "loss": 0.1555,
      "step": 4161
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7291015108458768,
      "learning_rate": 4.25484008329319e-06,
      "loss": 0.1198,
      "step": 4162
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9894766357695457,
      "learning_rate": 4.252680757431117e-06,
      "loss": 0.1808,
      "step": 4163
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9833411935168395,
      "learning_rate": 4.250521574125136e-06,
      "loss": 0.2069,
      "step": 4164
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8182886320611911,
      "learning_rate": 4.248362533787127e-06,
      "loss": 0.108,
      "step": 4165
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8367755570677949,
      "learning_rate": 4.24620363682894e-06,
      "loss": 0.1883,
      "step": 4166
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2225577630086737,
      "learning_rate": 4.244044883662398e-06,
      "loss": 0.2509,
      "step": 4167
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1914350845987183,
      "learning_rate": 4.241886274699298e-06,
      "loss": 0.2176,
      "step": 4168
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0687047057034496,
      "learning_rate": 4.239727810351411e-06,
      "loss": 0.2176,
      "step": 4169
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0185437280927194,
      "learning_rate": 4.237569491030477e-06,
      "loss": 0.1927,
      "step": 4170
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8867003796993029,
      "learning_rate": 4.235411317148209e-06,
      "loss": 0.1859,
      "step": 4171
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0881688304173232,
      "learning_rate": 4.233253289116295e-06,
      "loss": 0.2072,
      "step": 4172
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7137406022400554,
      "learning_rate": 4.231095407346391e-06,
      "loss": 0.1609,
      "step": 4173
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0302464737756754,
      "learning_rate": 4.228937672250128e-06,
      "loss": 0.1957,
      "step": 4174
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1415431297964815,
      "learning_rate": 4.226780084239107e-06,
      "loss": 0.2056,
      "step": 4175
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2220959921835037,
      "learning_rate": 4.224622643724905e-06,
      "loss": 0.1913,
      "step": 4176
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8064137585119698,
      "learning_rate": 4.222465351119065e-06,
      "loss": 0.1374,
      "step": 4177
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7035770668925161,
      "learning_rate": 4.220308206833107e-06,
      "loss": 0.1215,
      "step": 4178
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7735099807850028,
      "learning_rate": 4.2181512112785195e-06,
      "loss": 0.1237,
      "step": 4179
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.930293463997304,
      "learning_rate": 4.215994364866765e-06,
      "loss": 0.2046,
      "step": 4180
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0820761952927465,
      "learning_rate": 4.213837668009274e-06,
      "loss": 0.1648,
      "step": 4181
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8589396725033536,
      "learning_rate": 4.211681121117452e-06,
      "loss": 0.1571,
      "step": 4182
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0653636171353231,
      "learning_rate": 4.209524724602676e-06,
      "loss": 0.1782,
      "step": 4183
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9855245189070144,
      "learning_rate": 4.207368478876291e-06,
      "loss": 0.1641,
      "step": 4184
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7712460160152794,
      "learning_rate": 4.205212384349615e-06,
      "loss": 0.1068,
      "step": 4185
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.890889697199211,
      "learning_rate": 4.2030564414339395e-06,
      "loss": 0.1658,
      "step": 4186
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9768495230978815,
      "learning_rate": 4.2009006505405225e-06,
      "loss": 0.2154,
      "step": 4187
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8946853066804379,
      "learning_rate": 4.1987450120805955e-06,
      "loss": 0.1705,
      "step": 4188
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.946277786606894,
      "learning_rate": 4.1965895264653625e-06,
      "loss": 0.1868,
      "step": 4189
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6289851755347592,
      "learning_rate": 4.194434194105993e-06,
      "loss": 0.1054,
      "step": 4190
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6337960196603084,
      "learning_rate": 4.192279015413635e-06,
      "loss": 0.1027,
      "step": 4191
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8567601266210736,
      "learning_rate": 4.1901239907994025e-06,
      "loss": 0.0982,
      "step": 4192
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2904871783516103,
      "learning_rate": 4.187969120674381e-06,
      "loss": 0.2679,
      "step": 4193
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.125705117216332,
      "learning_rate": 4.185814405449623e-06,
      "loss": 0.2176,
      "step": 4194
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0064307380227573,
      "learning_rate": 4.183659845536158e-06,
      "loss": 0.195,
      "step": 4195
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9269246737410933,
      "learning_rate": 4.181505441344981e-06,
      "loss": 0.1431,
      "step": 4196
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1061247978287434,
      "learning_rate": 4.179351193287059e-06,
      "loss": 0.1856,
      "step": 4197
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9177950508044852,
      "learning_rate": 4.177197101773329e-06,
      "loss": 0.1415,
      "step": 4198
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.887271568986184,
      "learning_rate": 4.175043167214699e-06,
      "loss": 0.2028,
      "step": 4199
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7078550037916349,
      "learning_rate": 4.172889390022046e-06,
      "loss": 0.1578,
      "step": 4200
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0805452163821907,
      "learning_rate": 4.170735770606217e-06,
      "loss": 0.1733,
      "step": 4201
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8943839649941758,
      "learning_rate": 4.168582309378028e-06,
      "loss": 0.1841,
      "step": 4202
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9278391334849861,
      "learning_rate": 4.166429006748268e-06,
      "loss": 0.1422,
      "step": 4203
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9249586476838173,
      "learning_rate": 4.164275863127694e-06,
      "loss": 0.1649,
      "step": 4204
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2126928488028028,
      "learning_rate": 4.16212287892703e-06,
      "loss": 0.2375,
      "step": 4205
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9612736433143728,
      "learning_rate": 4.159970054556974e-06,
      "loss": 0.1819,
      "step": 4206
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1255270205590455,
      "learning_rate": 4.157817390428191e-06,
      "loss": 0.2403,
      "step": 4207
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9902206430364914,
      "learning_rate": 4.155664886951315e-06,
      "loss": 0.1964,
      "step": 4208
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9251439992722739,
      "learning_rate": 4.153512544536952e-06,
      "loss": 0.1751,
      "step": 4209
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.872256771175488,
      "learning_rate": 4.151360363595675e-06,
      "loss": 0.1658,
      "step": 4210
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.208241801949881,
      "learning_rate": 4.149208344538026e-06,
      "loss": 0.2129,
      "step": 4211
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.04595434535565,
      "learning_rate": 4.147056487774517e-06,
      "loss": 0.2094,
      "step": 4212
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0631927236337988,
      "learning_rate": 4.144904793715626e-06,
      "loss": 0.2147,
      "step": 4213
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9614890754833446,
      "learning_rate": 4.142753262771809e-06,
      "loss": 0.1677,
      "step": 4214
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1309573344753328,
      "learning_rate": 4.1406018953534825e-06,
      "loss": 0.2012,
      "step": 4215
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8550574802755017,
      "learning_rate": 4.138450691871033e-06,
      "loss": 0.1539,
      "step": 4216
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8526788694383064,
      "learning_rate": 4.136299652734818e-06,
      "loss": 0.132,
      "step": 4217
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.333834725826035,
      "learning_rate": 4.13414877835516e-06,
      "loss": 0.2352,
      "step": 4218
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1685036675148126,
      "learning_rate": 4.131998069142354e-06,
      "loss": 0.2247,
      "step": 4219
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.743967777113806,
      "learning_rate": 4.129847525506664e-06,
      "loss": 0.1392,
      "step": 4220
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7827536233701168,
      "learning_rate": 4.127697147858317e-06,
      "loss": 0.1421,
      "step": 4221
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9864448819699293,
      "learning_rate": 4.125546936607515e-06,
      "loss": 0.1973,
      "step": 4222
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9179879398121278,
      "learning_rate": 4.123396892164422e-06,
      "loss": 0.154,
      "step": 4223
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9886951239492435,
      "learning_rate": 4.121247014939174e-06,
      "loss": 0.2097,
      "step": 4224
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1182820209439202,
      "learning_rate": 4.119097305341877e-06,
      "loss": 0.2009,
      "step": 4225
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0301983387990012,
      "learning_rate": 4.116947763782599e-06,
      "loss": 0.1358,
      "step": 4226
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9993318764926071,
      "learning_rate": 4.114798390671381e-06,
      "loss": 0.1792,
      "step": 4227
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0719993577673506,
      "learning_rate": 4.112649186418229e-06,
      "loss": 0.2223,
      "step": 4228
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8498983007756447,
      "learning_rate": 4.11050015143312e-06,
      "loss": 0.1407,
      "step": 4229
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7083319149853353,
      "learning_rate": 4.1083512861259945e-06,
      "loss": 0.1351,
      "step": 4230
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9045948426990083,
      "learning_rate": 4.106202590906764e-06,
      "loss": 0.1922,
      "step": 4231
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1608746129639385,
      "learning_rate": 4.104054066185306e-06,
      "loss": 0.2145,
      "step": 4232
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0378795019151072,
      "learning_rate": 4.101905712371466e-06,
      "loss": 0.1576,
      "step": 4233
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1643210733695855,
      "learning_rate": 4.099757529875057e-06,
      "loss": 0.2302,
      "step": 4234
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1164009343307735,
      "learning_rate": 4.097609519105855e-06,
      "loss": 0.2376,
      "step": 4235
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3408624791048491,
      "learning_rate": 4.095461680473613e-06,
      "loss": 0.2544,
      "step": 4236
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.737710779797675,
      "learning_rate": 4.093314014388043e-06,
      "loss": 0.1169,
      "step": 4237
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0733480546170082,
      "learning_rate": 4.091166521258828e-06,
      "loss": 0.1688,
      "step": 4238
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8538967475213206,
      "learning_rate": 4.089019201495616e-06,
      "loss": 0.1537,
      "step": 4239
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2106769958762,
      "learning_rate": 4.08687205550802e-06,
      "loss": 0.2297,
      "step": 4240
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8679750105358881,
      "learning_rate": 4.084725083705624e-06,
      "loss": 0.1458,
      "step": 4241
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.084293791677736,
      "learning_rate": 4.0825782864979765e-06,
      "loss": 0.2162,
      "step": 4242
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.040634754710091,
      "learning_rate": 4.080431664294595e-06,
      "loss": 0.2003,
      "step": 4243
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0580864806194143,
      "learning_rate": 4.078285217504961e-06,
      "loss": 0.1916,
      "step": 4244
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7261735961732284,
      "learning_rate": 4.076138946538523e-06,
      "loss": 0.1546,
      "step": 4245
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0607420150003017,
      "learning_rate": 4.073992851804695e-06,
      "loss": 0.2091,
      "step": 4246
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7748094919206441,
      "learning_rate": 4.071846933712862e-06,
      "loss": 0.1459,
      "step": 4247
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0898261167286223,
      "learning_rate": 4.069701192672369e-06,
      "loss": 0.2319,
      "step": 4248
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0885720090618063,
      "learning_rate": 4.067555629092534e-06,
      "loss": 0.2113,
      "step": 4249
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9823298484011916,
      "learning_rate": 4.0654102433826345e-06,
      "loss": 0.1714,
      "step": 4250
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9278810086666247,
      "learning_rate": 4.063265035951917e-06,
      "loss": 0.1535,
      "step": 4251
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9060001509791987,
      "learning_rate": 4.061120007209595e-06,
      "loss": 0.1691,
      "step": 4252
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9269308387371156,
      "learning_rate": 4.058975157564847e-06,
      "loss": 0.1613,
      "step": 4253
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8292464864496741,
      "learning_rate": 4.056830487426817e-06,
      "loss": 0.1755,
      "step": 4254
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7819177023027984,
      "learning_rate": 4.0546859972046155e-06,
      "loss": 0.1482,
      "step": 4255
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2537309577212188,
      "learning_rate": 4.052541687307318e-06,
      "loss": 0.2233,
      "step": 4256
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9456157896626027,
      "learning_rate": 4.050397558143965e-06,
      "loss": 0.166,
      "step": 4257
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.874768215762556,
      "learning_rate": 4.048253610123562e-06,
      "loss": 0.1594,
      "step": 4258
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8639673860975546,
      "learning_rate": 4.046109843655084e-06,
      "loss": 0.1747,
      "step": 4259
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0378116066865217,
      "learning_rate": 4.043966259147468e-06,
      "loss": 0.1818,
      "step": 4260
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8696019583047706,
      "learning_rate": 4.0418228570096174e-06,
      "loss": 0.1853,
      "step": 4261
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7017695856377557,
      "learning_rate": 4.0396796376503995e-06,
      "loss": 0.13,
      "step": 4262
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9339363795184776,
      "learning_rate": 4.037536601478649e-06,
      "loss": 0.1746,
      "step": 4263
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0205860592333624,
      "learning_rate": 4.035393748903162e-06,
      "loss": 0.1942,
      "step": 4264
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9187630681679945,
      "learning_rate": 4.033251080332702e-06,
      "loss": 0.1447,
      "step": 4265
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0094850464406286,
      "learning_rate": 4.031108596175998e-06,
      "loss": 0.1801,
      "step": 4266
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9219462685693404,
      "learning_rate": 4.028966296841744e-06,
      "loss": 0.1927,
      "step": 4267
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1687809878401463,
      "learning_rate": 4.026824182738597e-06,
      "loss": 0.1717,
      "step": 4268
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8896347974989808,
      "learning_rate": 4.024682254275176e-06,
      "loss": 0.1802,
      "step": 4269
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7377228799190884,
      "learning_rate": 4.022540511860075e-06,
      "loss": 0.1506,
      "step": 4270
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2753933069929435,
      "learning_rate": 4.0203989559018396e-06,
      "loss": 0.204,
      "step": 4271
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8838197768004928,
      "learning_rate": 4.0182575868089895e-06,
      "loss": 0.1661,
      "step": 4272
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9541817011824509,
      "learning_rate": 4.016116404990003e-06,
      "loss": 0.1707,
      "step": 4273
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1560388261911985,
      "learning_rate": 4.013975410853325e-06,
      "loss": 0.226,
      "step": 4274
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0264929107021035,
      "learning_rate": 4.0118346048073656e-06,
      "loss": 0.1576,
      "step": 4275
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6864703715611559,
      "learning_rate": 4.009693987260496e-06,
      "loss": 0.1281,
      "step": 4276
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.045446454584618,
      "learning_rate": 4.007553558621054e-06,
      "loss": 0.1782,
      "step": 4277
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.884685069169878,
      "learning_rate": 4.00541331929734e-06,
      "loss": 0.1954,
      "step": 4278
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0342211055316315,
      "learning_rate": 4.003273269697619e-06,
      "loss": 0.1802,
      "step": 4279
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.602777805988345,
      "learning_rate": 4.001133410230119e-06,
      "loss": 0.1093,
      "step": 4280
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.309743549341959,
      "learning_rate": 3.998993741303035e-06,
      "loss": 0.2056,
      "step": 4281
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0215130020801706,
      "learning_rate": 3.99685426332452e-06,
      "loss": 0.1738,
      "step": 4282
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8717393824833066,
      "learning_rate": 3.994714976702695e-06,
      "loss": 0.1498,
      "step": 4283
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6082546377672297,
      "learning_rate": 3.9925758818456436e-06,
      "loss": 0.1258,
      "step": 4284
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7654086571678702,
      "learning_rate": 3.990436979161411e-06,
      "loss": 0.1044,
      "step": 4285
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9158785328091376,
      "learning_rate": 3.988298269058009e-06,
      "loss": 0.1598,
      "step": 4286
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.987609245276838,
      "learning_rate": 3.986159751943408e-06,
      "loss": 0.1803,
      "step": 4287
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9462328741268768,
      "learning_rate": 3.984021428225546e-06,
      "loss": 0.1343,
      "step": 4288
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9998912109767663,
      "learning_rate": 3.9818832983123215e-06,
      "loss": 0.1858,
      "step": 4289
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0325526380204775,
      "learning_rate": 3.979745362611598e-06,
      "loss": 0.2002,
      "step": 4290
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7783522462477583,
      "learning_rate": 3.9776076215311975e-06,
      "loss": 0.1083,
      "step": 4291
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.029685819483452,
      "learning_rate": 3.975470075478915e-06,
      "loss": 0.1796,
      "step": 4292
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0450909182613894,
      "learning_rate": 3.973332724862497e-06,
      "loss": 0.1605,
      "step": 4293
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7466294458932318,
      "learning_rate": 3.971195570089657e-06,
      "loss": 0.1426,
      "step": 4294
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0204528926741319,
      "learning_rate": 3.9690586115680725e-06,
      "loss": 0.2043,
      "step": 4295
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0932758574710444,
      "learning_rate": 3.966921849705383e-06,
      "loss": 0.2045,
      "step": 4296
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9760282748949191,
      "learning_rate": 3.964785284909188e-06,
      "loss": 0.174,
      "step": 4297
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1197683567651049,
      "learning_rate": 3.962648917587053e-06,
      "loss": 0.2089,
      "step": 4298
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0153846857327227,
      "learning_rate": 3.960512748146505e-06,
      "loss": 0.2231,
      "step": 4299
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1300732952473194,
      "learning_rate": 3.958376776995031e-06,
      "loss": 0.1981,
      "step": 4300
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8421186219261809,
      "learning_rate": 3.956241004540081e-06,
      "loss": 0.1596,
      "step": 4301
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9562758227105759,
      "learning_rate": 3.95410543118907e-06,
      "loss": 0.1649,
      "step": 4302
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0277270805932626,
      "learning_rate": 3.95197005734937e-06,
      "loss": 0.2239,
      "step": 4303
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0423714290241082,
      "learning_rate": 3.94983488342832e-06,
      "loss": 0.166,
      "step": 4304
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.049547593067017,
      "learning_rate": 3.947699909833218e-06,
      "loss": 0.1694,
      "step": 4305
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7664576564245964,
      "learning_rate": 3.9455651369713246e-06,
      "loss": 0.1468,
      "step": 4306
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8572318796250151,
      "learning_rate": 3.94343056524986e-06,
      "loss": 0.1593,
      "step": 4307
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8844059486163716,
      "learning_rate": 3.941296195076011e-06,
      "loss": 0.156,
      "step": 4308
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9805527451499938,
      "learning_rate": 3.939162026856921e-06,
      "loss": 0.1652,
      "step": 4309
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8609219995939573,
      "learning_rate": 3.937028060999696e-06,
      "loss": 0.1429,
      "step": 4310
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0427528041432963,
      "learning_rate": 3.934894297911404e-06,
      "loss": 0.227,
      "step": 4311
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8119318405515041,
      "learning_rate": 3.932760737999077e-06,
      "loss": 0.1303,
      "step": 4312
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6693403507997321,
      "learning_rate": 3.930627381669703e-06,
      "loss": 0.1331,
      "step": 4313
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.362416561045485,
      "learning_rate": 3.928494229330233e-06,
      "loss": 0.2397,
      "step": 4314
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2590251441993785,
      "learning_rate": 3.9263612813875835e-06,
      "loss": 0.2213,
      "step": 4315
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2539309767364777,
      "learning_rate": 3.9242285382486275e-06,
      "loss": 0.2161,
      "step": 4316
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9358855406920246,
      "learning_rate": 3.922096000320199e-06,
      "loss": 0.1475,
      "step": 4317
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1556718858877635,
      "learning_rate": 3.919963668009093e-06,
      "loss": 0.2528,
      "step": 4318
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9046784105484051,
      "learning_rate": 3.9178315417220684e-06,
      "loss": 0.153,
      "step": 4319
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7610395338627288,
      "learning_rate": 3.915699621865839e-06,
      "loss": 0.1612,
      "step": 4320
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0275703229240063,
      "learning_rate": 3.913567908847085e-06,
      "loss": 0.214,
      "step": 4321
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7897235010122751,
      "learning_rate": 3.9114364030724446e-06,
      "loss": 0.1496,
      "step": 4322
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.996369238849826,
      "learning_rate": 3.909305104948516e-06,
      "loss": 0.1899,
      "step": 4323
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7536907347742022,
      "learning_rate": 3.907174014881858e-06,
      "loss": 0.1143,
      "step": 4324
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2467441571725923,
      "learning_rate": 3.905043133278991e-06,
      "loss": 0.2089,
      "step": 4325
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6771452182861245,
      "learning_rate": 3.902912460546395e-06,
      "loss": 0.1399,
      "step": 4326
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9681263652489857,
      "learning_rate": 3.9007819970905095e-06,
      "loss": 0.2063,
      "step": 4327
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0129340410876682,
      "learning_rate": 3.898651743317735e-06,
      "loss": 0.2166,
      "step": 4328
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9320811197834922,
      "learning_rate": 3.896521699634431e-06,
      "loss": 0.165,
      "step": 4329
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0777103487242672,
      "learning_rate": 3.894391866446917e-06,
      "loss": 0.2084,
      "step": 4330
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7457118381794521,
      "learning_rate": 3.892262244161473e-06,
      "loss": 0.1563,
      "step": 4331
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9813617376673082,
      "learning_rate": 3.89013283318434e-06,
      "loss": 0.1526,
      "step": 4332
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9566108207079296,
      "learning_rate": 3.888003633921716e-06,
      "loss": 0.1847,
      "step": 4333
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7629111886823083,
      "learning_rate": 3.885874646779759e-06,
      "loss": 0.1124,
      "step": 4334
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7237974208251365,
      "learning_rate": 3.883745872164588e-06,
      "loss": 0.1214,
      "step": 4335
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8195315978424395,
      "learning_rate": 3.88161731048228e-06,
      "loss": 0.1854,
      "step": 4336
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.091040009199511,
      "learning_rate": 3.879488962138875e-06,
      "loss": 0.2121,
      "step": 4337
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8312332072644686,
      "learning_rate": 3.877360827540367e-06,
      "loss": 0.1655,
      "step": 4338
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9145977387692709,
      "learning_rate": 3.875232907092715e-06,
      "loss": 0.1379,
      "step": 4339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.930743684841806,
      "learning_rate": 3.87310520120183e-06,
      "loss": 0.19,
      "step": 4340
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.019461174775143,
      "learning_rate": 3.870977710273588e-06,
      "loss": 0.1885,
      "step": 4341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9570020211337245,
      "learning_rate": 3.868850434713821e-06,
      "loss": 0.1857,
      "step": 4342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8763609182615707,
      "learning_rate": 3.866723374928322e-06,
      "loss": 0.1589,
      "step": 4343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7951626454630002,
      "learning_rate": 3.86459653132284e-06,
      "loss": 0.1441,
      "step": 4344
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1047139952974843,
      "learning_rate": 3.862469904303087e-06,
      "loss": 0.2462,
      "step": 4345
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0648137395163366,
      "learning_rate": 3.860343494274729e-06,
      "loss": 0.179,
      "step": 4346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8008961703247325,
      "learning_rate": 3.858217301643393e-06,
      "loss": 0.1647,
      "step": 4347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.896981207075646,
      "learning_rate": 3.8560913268146664e-06,
      "loss": 0.1627,
      "step": 4348
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.110358597877956,
      "learning_rate": 3.85396557019409e-06,
      "loss": 0.2025,
      "step": 4349
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.086913851395784,
      "learning_rate": 3.851840032187169e-06,
      "loss": 0.2342,
      "step": 4350
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0678556076209735,
      "learning_rate": 3.849714713199362e-06,
      "loss": 0.157,
      "step": 4351
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0083109904489493,
      "learning_rate": 3.8475896136360876e-06,
      "loss": 0.1947,
      "step": 4352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9013780974241382,
      "learning_rate": 3.8454647339027225e-06,
      "loss": 0.1276,
      "step": 4353
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.241728486750486,
      "learning_rate": 3.8433400744046035e-06,
      "loss": 0.229,
      "step": 4354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7984764583538554,
      "learning_rate": 3.841215635547021e-06,
      "loss": 0.1771,
      "step": 4355
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0424173464922903,
      "learning_rate": 3.839091417735228e-06,
      "loss": 0.2025,
      "step": 4356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6104221074752221,
      "learning_rate": 3.836967421374431e-06,
      "loss": 0.1179,
      "step": 4357
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1333987416793896,
      "learning_rate": 3.834843646869797e-06,
      "loss": 0.2042,
      "step": 4358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9379317587905673,
      "learning_rate": 3.832720094626449e-06,
      "loss": 0.2058,
      "step": 4359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6089141225165466,
      "learning_rate": 3.830596765049471e-06,
      "loss": 0.1058,
      "step": 4360
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9612654948431042,
      "learning_rate": 3.828473658543902e-06,
      "loss": 0.1711,
      "step": 4361
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6895698018751334,
      "learning_rate": 3.826350775514739e-06,
      "loss": 0.1502,
      "step": 4362
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8025653941494961,
      "learning_rate": 3.824228116366933e-06,
      "loss": 0.1333,
      "step": 4363
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6485388846390948,
      "learning_rate": 3.8221056815053966e-06,
      "loss": 0.1259,
      "step": 4364
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8448730437153577,
      "learning_rate": 3.8199834713349995e-06,
      "loss": 0.1274,
      "step": 4365
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8629960959858458,
      "learning_rate": 3.8178614862605666e-06,
      "loss": 0.1658,
      "step": 4366
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9787737186563928,
      "learning_rate": 3.81573972668688e-06,
      "loss": 0.142,
      "step": 4367
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7838068830547757,
      "learning_rate": 3.813618193018681e-06,
      "loss": 0.1265,
      "step": 4368
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1265947429829317,
      "learning_rate": 3.8114968856606644e-06,
      "loss": 0.1786,
      "step": 4369
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6884550410794877,
      "learning_rate": 3.8093758050174833e-06,
      "loss": 0.1595,
      "step": 4370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9506723351868047,
      "learning_rate": 3.80725495149375e-06,
      "loss": 0.1481,
      "step": 4371
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1858330603745277,
      "learning_rate": 3.805134325494031e-06,
      "loss": 0.2093,
      "step": 4372
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8121047944787255,
      "learning_rate": 3.8030139274228482e-06,
      "loss": 0.1602,
      "step": 4373
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6616328853151616,
      "learning_rate": 3.8008937576846826e-06,
      "loss": 0.1117,
      "step": 4374
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7762314122844813,
      "learning_rate": 3.7987738166839705e-06,
      "loss": 0.1075,
      "step": 4375
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7978710968400071,
      "learning_rate": 3.796654104825105e-06,
      "loss": 0.1426,
      "step": 4376
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0249015597028943,
      "learning_rate": 3.7945346225124345e-06,
      "loss": 0.1792,
      "step": 4377
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0114177233775752,
      "learning_rate": 3.792415370150264e-06,
      "loss": 0.1687,
      "step": 4378
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9128591590129569,
      "learning_rate": 3.7902963481428566e-06,
      "loss": 0.2018,
      "step": 4379
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8165488894415546,
      "learning_rate": 3.788177556894429e-06,
      "loss": 0.1305,
      "step": 4380
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9883363558048808,
      "learning_rate": 3.786058996809151e-06,
      "loss": 0.1908,
      "step": 4381
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1365464212106637,
      "learning_rate": 3.783940668291158e-06,
      "loss": 0.1901,
      "step": 4382
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1404514480563013,
      "learning_rate": 3.7818225717445316e-06,
      "loss": 0.2058,
      "step": 4383
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1129470563321358,
      "learning_rate": 3.779704707573314e-06,
      "loss": 0.1812,
      "step": 4384
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9550327046781285,
      "learning_rate": 3.7775870761815016e-06,
      "loss": 0.154,
      "step": 4385
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.852461900926676,
      "learning_rate": 3.775469677973047e-06,
      "loss": 0.1198,
      "step": 4386
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7979021743045372,
      "learning_rate": 3.773352513351857e-06,
      "loss": 0.1411,
      "step": 4387
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8966331418071114,
      "learning_rate": 3.7712355827217946e-06,
      "loss": 0.1673,
      "step": 4388
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1429355612818262,
      "learning_rate": 3.769118886486679e-06,
      "loss": 0.1648,
      "step": 4389
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9853186194581216,
      "learning_rate": 3.7670024250502836e-06,
      "loss": 0.1318,
      "step": 4390
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0504913357284047,
      "learning_rate": 3.7648861988163376e-06,
      "loss": 0.1763,
      "step": 4391
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7817560271003773,
      "learning_rate": 3.7627702081885243e-06,
      "loss": 0.1708,
      "step": 4392
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9054835621938473,
      "learning_rate": 3.760654453570485e-06,
      "loss": 0.1504,
      "step": 4393
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9988285676439191,
      "learning_rate": 3.758538935365813e-06,
      "loss": 0.2048,
      "step": 4394
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.861377987307909,
      "learning_rate": 3.756423653978057e-06,
      "loss": 0.132,
      "step": 4395
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9454423015623813,
      "learning_rate": 3.7543086098107206e-06,
      "loss": 0.1708,
      "step": 4396
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0590714517549784,
      "learning_rate": 3.7521938032672634e-06,
      "loss": 0.209,
      "step": 4397
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7323216905502147,
      "learning_rate": 3.750079234751098e-06,
      "loss": 0.1412,
      "step": 4398
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8939032014540864,
      "learning_rate": 3.747964904665593e-06,
      "loss": 0.1476,
      "step": 4399
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0430439294381324,
      "learning_rate": 3.74585081341407e-06,
      "loss": 0.1742,
      "step": 4400
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8705340264317661,
      "learning_rate": 3.7437369613998076e-06,
      "loss": 0.1507,
      "step": 4401
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1462817525901825,
      "learning_rate": 3.741623349026035e-06,
      "loss": 0.2135,
      "step": 4402
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1465349334619217,
      "learning_rate": 3.7395099766959398e-06,
      "loss": 0.2087,
      "step": 4403
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9618171696341987,
      "learning_rate": 3.7373968448126574e-06,
      "loss": 0.1679,
      "step": 4404
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9799370337088611,
      "learning_rate": 3.7352839537792874e-06,
      "loss": 0.1962,
      "step": 4405
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8457284413632584,
      "learning_rate": 3.733171303998875e-06,
      "loss": 0.1329,
      "step": 4406
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0581805302951763,
      "learning_rate": 3.731058895874422e-06,
      "loss": 0.2168,
      "step": 4407
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8844634071053589,
      "learning_rate": 3.7289467298088843e-06,
      "loss": 0.14,
      "step": 4408
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7998596245250369,
      "learning_rate": 3.726834806205174e-06,
      "loss": 0.1543,
      "step": 4409
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9295592868120984,
      "learning_rate": 3.72472312546615e-06,
      "loss": 0.2057,
      "step": 4410
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9839983325812759,
      "learning_rate": 3.722611687994633e-06,
      "loss": 0.1686,
      "step": 4411
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0969758523090805,
      "learning_rate": 3.720500494193392e-06,
      "loss": 0.2022,
      "step": 4412
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8652971790423482,
      "learning_rate": 3.718389544465151e-06,
      "loss": 0.107,
      "step": 4413
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1246244523623548,
      "learning_rate": 3.716278839212589e-06,
      "loss": 0.2148,
      "step": 4414
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2935225858308905,
      "learning_rate": 3.7141683788383354e-06,
      "loss": 0.2345,
      "step": 4415
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6921995408167112,
      "learning_rate": 3.712058163744976e-06,
      "loss": 0.1357,
      "step": 4416
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0247927553044356,
      "learning_rate": 3.7099481943350472e-06,
      "loss": 0.1604,
      "step": 4417
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9782041461346989,
      "learning_rate": 3.7078384710110404e-06,
      "loss": 0.1972,
      "step": 4418
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.728102977436545,
      "learning_rate": 3.7057289941753992e-06,
      "loss": 0.1187,
      "step": 4419
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2266249848607977,
      "learning_rate": 3.7036197642305193e-06,
      "loss": 0.1864,
      "step": 4420
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8673612894284716,
      "learning_rate": 3.7015107815787523e-06,
      "loss": 0.1644,
      "step": 4421
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.004667994532613,
      "learning_rate": 3.699402046622398e-06,
      "loss": 0.1553,
      "step": 4422
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8500833747112413,
      "learning_rate": 3.697293559763714e-06,
      "loss": 0.177,
      "step": 4423
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0191450765466155,
      "learning_rate": 3.695185321404906e-06,
      "loss": 0.196,
      "step": 4424
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7963899272908277,
      "learning_rate": 3.693077331948136e-06,
      "loss": 0.1649,
      "step": 4425
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.082095325474534,
      "learning_rate": 3.6909695917955145e-06,
      "loss": 0.1664,
      "step": 4426
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0368060975357833,
      "learning_rate": 3.688862101349111e-06,
      "loss": 0.1745,
      "step": 4427
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7942456868032501,
      "learning_rate": 3.68675486101094e-06,
      "loss": 0.1321,
      "step": 4428
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.994254114113499,
      "learning_rate": 3.684647871182973e-06,
      "loss": 0.1691,
      "step": 4429
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9163162103059245,
      "learning_rate": 3.6825411322671318e-06,
      "loss": 0.1903,
      "step": 4430
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8793965199358442,
      "learning_rate": 3.6804346446652904e-06,
      "loss": 0.1441,
      "step": 4431
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0549612049608847,
      "learning_rate": 3.678328408779276e-06,
      "loss": 0.1824,
      "step": 4432
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8764979411811156,
      "learning_rate": 3.6762224250108674e-06,
      "loss": 0.1671,
      "step": 4433
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7982902111753772,
      "learning_rate": 3.6741166937617935e-06,
      "loss": 0.1375,
      "step": 4434
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.028769820396496,
      "learning_rate": 3.6720112154337375e-06,
      "loss": 0.1945,
      "step": 4435
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.856837345886647,
      "learning_rate": 3.669905990428332e-06,
      "loss": 0.1277,
      "step": 4436
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9271705832557484,
      "learning_rate": 3.6678010191471625e-06,
      "loss": 0.1316,
      "step": 4437
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0004998055770955,
      "learning_rate": 3.665696301991769e-06,
      "loss": 0.1731,
      "step": 4438
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7973974758750759,
      "learning_rate": 3.6635918393636394e-06,
      "loss": 0.1507,
      "step": 4439
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0560277695024831,
      "learning_rate": 3.6614876316642125e-06,
      "loss": 0.2458,
      "step": 4440
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3483038254270716,
      "learning_rate": 3.65938367929488e-06,
      "loss": 0.231,
      "step": 4441
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7392127727252089,
      "learning_rate": 3.6572799826569847e-06,
      "loss": 0.1225,
      "step": 4442
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7292942232325496,
      "learning_rate": 3.655176542151822e-06,
      "loss": 0.1204,
      "step": 4443
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8412470707434573,
      "learning_rate": 3.6530733581806356e-06,
      "loss": 0.1616,
      "step": 4444
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6786400787908524,
      "learning_rate": 3.6509704311446227e-06,
      "loss": 0.1558,
      "step": 4445
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7886541910143329,
      "learning_rate": 3.64886776144493e-06,
      "loss": 0.1322,
      "step": 4446
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.161657849575285,
      "learning_rate": 3.646765349482656e-06,
      "loss": 0.2134,
      "step": 4447
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0255639568542627,
      "learning_rate": 3.6446631956588485e-06,
      "loss": 0.2234,
      "step": 4448
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7376456278593372,
      "learning_rate": 3.6425613003745096e-06,
      "loss": 0.1445,
      "step": 4449
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0738157601614973,
      "learning_rate": 3.640459664030588e-06,
      "loss": 0.1831,
      "step": 4450
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9038256235729651,
      "learning_rate": 3.6383582870279853e-06,
      "loss": 0.1975,
      "step": 4451
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7245501511684216,
      "learning_rate": 3.6362571697675535e-06,
      "loss": 0.1655,
      "step": 4452
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2588851372719416,
      "learning_rate": 3.6341563126500933e-06,
      "loss": 0.2166,
      "step": 4453
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0911587710991606,
      "learning_rate": 3.6320557160763594e-06,
      "loss": 0.1878,
      "step": 4454
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0700135443224643,
      "learning_rate": 3.629955380447051e-06,
      "loss": 0.1784,
      "step": 4455
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4604125703276245,
      "learning_rate": 3.627855306162825e-06,
      "loss": 0.2599,
      "step": 4456
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8425089229012506,
      "learning_rate": 3.6257554936242822e-06,
      "loss": 0.1464,
      "step": 4457
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8552934993045179,
      "learning_rate": 3.6236559432319753e-06,
      "loss": 0.16,
      "step": 4458
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.820730037841644,
      "learning_rate": 3.6215566553864077e-06,
      "loss": 0.1234,
      "step": 4459
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0237349616095686,
      "learning_rate": 3.619457630488032e-06,
      "loss": 0.2224,
      "step": 4460
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8494805584474145,
      "learning_rate": 3.6173588689372536e-06,
      "loss": 0.1692,
      "step": 4461
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9086261420033765,
      "learning_rate": 3.615260371134425e-06,
      "loss": 0.1693,
      "step": 4462
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.206986443224008,
      "learning_rate": 3.613162137479845e-06,
      "loss": 0.1826,
      "step": 4463
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2127496180595394,
      "learning_rate": 3.6110641683737686e-06,
      "loss": 0.1852,
      "step": 4464
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0257720771634098,
      "learning_rate": 3.608966464216396e-06,
      "loss": 0.1796,
      "step": 4465
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.042241899651314,
      "learning_rate": 3.6068690254078785e-06,
      "loss": 0.1704,
      "step": 4466
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0995776951519298,
      "learning_rate": 3.604771852348317e-06,
      "loss": 0.2251,
      "step": 4467
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8083039986789788,
      "learning_rate": 3.602674945437761e-06,
      "loss": 0.1358,
      "step": 4468
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0746366339828586,
      "learning_rate": 3.600578305076208e-06,
      "loss": 0.1987,
      "step": 4469
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8939585565035044,
      "learning_rate": 3.5984819316636077e-06,
      "loss": 0.1698,
      "step": 4470
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9722621146085114,
      "learning_rate": 3.5963858255998553e-06,
      "loss": 0.1966,
      "step": 4471
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8979090042184962,
      "learning_rate": 3.5942899872847986e-06,
      "loss": 0.1537,
      "step": 4472
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7697040462872451,
      "learning_rate": 3.5921944171182327e-06,
      "loss": 0.12,
      "step": 4473
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.269833207718989,
      "learning_rate": 3.590099115499901e-06,
      "loss": 0.233,
      "step": 4474
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1525058373074206,
      "learning_rate": 3.588004082829496e-06,
      "loss": 0.1853,
      "step": 4475
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9301251145749455,
      "learning_rate": 3.58590931950666e-06,
      "loss": 0.195,
      "step": 4476
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9575574665792909,
      "learning_rate": 3.583814825930981e-06,
      "loss": 0.1604,
      "step": 4477
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.881857599469321,
      "learning_rate": 3.581720602501999e-06,
      "loss": 0.1814,
      "step": 4478
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9072136445887775,
      "learning_rate": 3.579626649619201e-06,
      "loss": 0.1921,
      "step": 4479
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8079343654231306,
      "learning_rate": 3.5775329676820223e-06,
      "loss": 0.1228,
      "step": 4480
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.241310297370198,
      "learning_rate": 3.5754395570898456e-06,
      "loss": 0.2303,
      "step": 4481
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0105450487480667,
      "learning_rate": 3.5733464182420013e-06,
      "loss": 0.1735,
      "step": 4482
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9724084122382136,
      "learning_rate": 3.5712535515377746e-06,
      "loss": 0.1295,
      "step": 4483
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8289514753138669,
      "learning_rate": 3.56916095737639e-06,
      "loss": 0.1744,
      "step": 4484
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8006659192608335,
      "learning_rate": 3.567068636157024e-06,
      "loss": 0.1458,
      "step": 4485
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9385725153149054,
      "learning_rate": 3.5649765882788034e-06,
      "loss": 0.1558,
      "step": 4486
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.662360582911537,
      "learning_rate": 3.5628848141407963e-06,
      "loss": 0.1396,
      "step": 4487
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8531207981230648,
      "learning_rate": 3.560793314142024e-06,
      "loss": 0.1766,
      "step": 4488
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7520623718204582,
      "learning_rate": 3.558702088681455e-06,
      "loss": 0.119,
      "step": 4489
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8711978222351363,
      "learning_rate": 3.5566111381580025e-06,
      "loss": 0.1681,
      "step": 4490
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9966460992887938,
      "learning_rate": 3.554520462970531e-06,
      "loss": 0.1918,
      "step": 4491
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8132305722985139,
      "learning_rate": 3.5524300635178492e-06,
      "loss": 0.1591,
      "step": 4492
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7919837502478271,
      "learning_rate": 3.550339940198715e-06,
      "loss": 0.1415,
      "step": 4493
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.820027185819046,
      "learning_rate": 3.5482500934118334e-06,
      "loss": 0.145,
      "step": 4494
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9588757554065278,
      "learning_rate": 3.546160523555857e-06,
      "loss": 0.1863,
      "step": 4495
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8829445516554887,
      "learning_rate": 3.5440712310293853e-06,
      "loss": 0.1496,
      "step": 4496
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9396350272701753,
      "learning_rate": 3.5419822162309634e-06,
      "loss": 0.1946,
      "step": 4497
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9814323983525045,
      "learning_rate": 3.5398934795590857e-06,
      "loss": 0.1447,
      "step": 4498
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8729369274175995,
      "learning_rate": 3.5378050214121918e-06,
      "loss": 0.1383,
      "step": 4499
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6982352181583841,
      "learning_rate": 3.53571684218867e-06,
      "loss": 0.1713,
      "step": 4500
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9207443472087757,
      "learning_rate": 3.533628942286853e-06,
      "loss": 0.169,
      "step": 4501
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0371153833318443,
      "learning_rate": 3.5315413221050225e-06,
      "loss": 0.1639,
      "step": 4502
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.005996970925269,
      "learning_rate": 3.5294539820414063e-06,
      "loss": 0.2086,
      "step": 4503
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1957081947452903,
      "learning_rate": 3.5273669224941776e-06,
      "loss": 0.2428,
      "step": 4504
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9451956512271618,
      "learning_rate": 3.5252801438614537e-06,
      "loss": 0.1532,
      "step": 4505
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0666362952943391,
      "learning_rate": 3.523193646541306e-06,
      "loss": 0.2061,
      "step": 4506
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9201230523817502,
      "learning_rate": 3.5211074309317478e-06,
      "loss": 0.1627,
      "step": 4507
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.995297712777418,
      "learning_rate": 3.5190214974307358e-06,
      "loss": 0.2034,
      "step": 4508
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0091940045185608,
      "learning_rate": 3.5169358464361787e-06,
      "loss": 0.1652,
      "step": 4509
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9876780416796546,
      "learning_rate": 3.5148504783459246e-06,
      "loss": 0.1871,
      "step": 4510
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7211483884967476,
      "learning_rate": 3.512765393557773e-06,
      "loss": 0.1393,
      "step": 4511
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8080020771147695,
      "learning_rate": 3.5106805924694676e-06,
      "loss": 0.175,
      "step": 4512
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8634681218944426,
      "learning_rate": 3.5085960754786986e-06,
      "loss": 0.128,
      "step": 4513
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8166753444743415,
      "learning_rate": 3.5065118429831e-06,
      "loss": 0.1427,
      "step": 4514
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9292296834440131,
      "learning_rate": 3.5044278953802546e-06,
      "loss": 0.1625,
      "step": 4515
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9252885925102377,
      "learning_rate": 3.5023442330676866e-06,
      "loss": 0.1723,
      "step": 4516
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9599160241164099,
      "learning_rate": 3.5002608564428712e-06,
      "loss": 0.193,
      "step": 4517
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.164857903713516,
      "learning_rate": 3.4981777659032255e-06,
      "loss": 0.1634,
      "step": 4518
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9828351949089699,
      "learning_rate": 3.4960949618461128e-06,
      "loss": 0.1867,
      "step": 4519
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7787881539685579,
      "learning_rate": 3.494012444668841e-06,
      "loss": 0.1432,
      "step": 4520
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9507791000133687,
      "learning_rate": 3.491930214768665e-06,
      "loss": 0.167,
      "step": 4521
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8686341149585823,
      "learning_rate": 3.4898482725427827e-06,
      "loss": 0.1733,
      "step": 4522
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0499711529718079,
      "learning_rate": 3.48776661838834e-06,
      "loss": 0.1875,
      "step": 4523
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0175050016758014,
      "learning_rate": 3.4856852527024256e-06,
      "loss": 0.188,
      "step": 4524
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.119662288871847,
      "learning_rate": 3.4836041758820726e-06,
      "loss": 0.1836,
      "step": 4525
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8910567478160418,
      "learning_rate": 3.4815233883242626e-06,
      "loss": 0.1541,
      "step": 4526
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0949918194146493,
      "learning_rate": 3.479442890425915e-06,
      "loss": 0.1866,
      "step": 4527
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9548138338277052,
      "learning_rate": 3.4773626825839033e-06,
      "loss": 0.1445,
      "step": 4528
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0531968370779805,
      "learning_rate": 3.47528276519504e-06,
      "loss": 0.2022,
      "step": 4529
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0287437416129133,
      "learning_rate": 3.4732031386560814e-06,
      "loss": 0.1949,
      "step": 4530
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9648087608713005,
      "learning_rate": 3.4711238033637318e-06,
      "loss": 0.1989,
      "step": 4531
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9917322537094673,
      "learning_rate": 3.4690447597146387e-06,
      "loss": 0.1822,
      "step": 4532
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9200326667232288,
      "learning_rate": 3.4669660081053903e-06,
      "loss": 0.1361,
      "step": 4533
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0411144358844189,
      "learning_rate": 3.4648875489325246e-06,
      "loss": 0.1842,
      "step": 4534
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6892061958947691,
      "learning_rate": 3.4628093825925208e-06,
      "loss": 0.1603,
      "step": 4535
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9500314055168871,
      "learning_rate": 3.460731509481803e-06,
      "loss": 0.1688,
      "step": 4536
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9846723267618007,
      "learning_rate": 3.4586539299967393e-06,
      "loss": 0.1715,
      "step": 4537
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2292051708095493,
      "learning_rate": 3.456576644533639e-06,
      "loss": 0.2399,
      "step": 4538
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9382214013926128,
      "learning_rate": 3.4544996534887643e-06,
      "loss": 0.184,
      "step": 4539
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8764650658769491,
      "learning_rate": 3.4524229572583102e-06,
      "loss": 0.1739,
      "step": 4540
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7779110592598545,
      "learning_rate": 3.450346556238422e-06,
      "loss": 0.1513,
      "step": 4541
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9450908359300433,
      "learning_rate": 3.4482704508251857e-06,
      "loss": 0.1867,
      "step": 4542
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2246611598404553,
      "learning_rate": 3.446194641414633e-06,
      "loss": 0.2121,
      "step": 4543
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9956812924030509,
      "learning_rate": 3.4441191284027385e-06,
      "loss": 0.1973,
      "step": 4544
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9667226214217274,
      "learning_rate": 3.44204391218542e-06,
      "loss": 0.1687,
      "step": 4545
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6551926637563678,
      "learning_rate": 3.4399689931585376e-06,
      "loss": 0.0771,
      "step": 4546
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8225457309235048,
      "learning_rate": 3.437894371717897e-06,
      "loss": 0.1466,
      "step": 4547
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0658106521496067,
      "learning_rate": 3.435820048259246e-06,
      "loss": 0.1858,
      "step": 4548
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.077362310236794,
      "learning_rate": 3.4337460231782738e-06,
      "loss": 0.1994,
      "step": 4549
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9876564536571913,
      "learning_rate": 3.431672296870616e-06,
      "loss": 0.1605,
      "step": 4550
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7742761402369993,
      "learning_rate": 3.4295988697318506e-06,
      "loss": 0.1721,
      "step": 4551
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.014831628215709,
      "learning_rate": 3.427525742157496e-06,
      "loss": 0.2027,
      "step": 4552
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0157819267087174,
      "learning_rate": 3.4254529145430156e-06,
      "loss": 0.1856,
      "step": 4553
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0297482197884502,
      "learning_rate": 3.4233803872838143e-06,
      "loss": 0.1645,
      "step": 4554
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0618733392576813,
      "learning_rate": 3.4213081607752407e-06,
      "loss": 0.1935,
      "step": 4555
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7262816442135156,
      "learning_rate": 3.4192362354125873e-06,
      "loss": 0.1174,
      "step": 4556
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8476900065962328,
      "learning_rate": 3.4171646115910856e-06,
      "loss": 0.1763,
      "step": 4557
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2274938484002913,
      "learning_rate": 3.415093289705911e-06,
      "loss": 0.2513,
      "step": 4558
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0768721066197002,
      "learning_rate": 3.413022270152184e-06,
      "loss": 0.1806,
      "step": 4559
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.662100076404571,
      "learning_rate": 3.4109515533249637e-06,
      "loss": 0.1723,
      "step": 4560
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6450220421246762,
      "learning_rate": 3.4088811396192518e-06,
      "loss": 0.1487,
      "step": 4561
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9498045730907304,
      "learning_rate": 3.406811029429998e-06,
      "loss": 0.1629,
      "step": 4562
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7974376511157453,
      "learning_rate": 3.404741223152086e-06,
      "loss": 0.1407,
      "step": 4563
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8986281443811109,
      "learning_rate": 3.4026717211803454e-06,
      "loss": 0.1853,
      "step": 4564
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8720219820289326,
      "learning_rate": 3.4006025239095487e-06,
      "loss": 0.1744,
      "step": 4565
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0550160862561246,
      "learning_rate": 3.398533631734407e-06,
      "loss": 0.1778,
      "step": 4566
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9759399822355166,
      "learning_rate": 3.3964650450495766e-06,
      "loss": 0.1536,
      "step": 4567
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9519809949132851,
      "learning_rate": 3.394396764249654e-06,
      "loss": 0.1829,
      "step": 4568
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9707480369754697,
      "learning_rate": 3.3923287897291766e-06,
      "loss": 0.1773,
      "step": 4569
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8987976439244042,
      "learning_rate": 3.3902611218826256e-06,
      "loss": 0.1328,
      "step": 4570
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9375632184735514,
      "learning_rate": 3.3881937611044215e-06,
      "loss": 0.1775,
      "step": 4571
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6894913615151235,
      "learning_rate": 3.386126707788926e-06,
      "loss": 0.1322,
      "step": 4572
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8394670154366565,
      "learning_rate": 3.384059962330445e-06,
      "loss": 0.1369,
      "step": 4573
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1641576063161887,
      "learning_rate": 3.3819935251232226e-06,
      "loss": 0.2489,
      "step": 4574
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8991987393924215,
      "learning_rate": 3.379927396561447e-06,
      "loss": 0.1576,
      "step": 4575
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9747255565656484,
      "learning_rate": 3.377861577039245e-06,
      "loss": 0.1773,
      "step": 4576
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1194564171051788,
      "learning_rate": 3.3757960669506846e-06,
      "loss": 0.2077,
      "step": 4577
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9790791056289458,
      "learning_rate": 3.373730866689776e-06,
      "loss": 0.2315,
      "step": 4578
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3183438816796527,
      "learning_rate": 3.371665976650472e-06,
      "loss": 0.2097,
      "step": 4579
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.923497548372654,
      "learning_rate": 3.36960139722666e-06,
      "loss": 0.197,
      "step": 4580
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9145629467808072,
      "learning_rate": 3.367537128812175e-06,
      "loss": 0.1649,
      "step": 4581
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2195707084639895,
      "learning_rate": 3.3654731718007883e-06,
      "loss": 0.2311,
      "step": 4582
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1433895243376289,
      "learning_rate": 3.363409526586212e-06,
      "loss": 0.2306,
      "step": 4583
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2917766630805192,
      "learning_rate": 3.361346193562105e-06,
      "loss": 0.2152,
      "step": 4584
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6525528969927801,
      "learning_rate": 3.35928317312206e-06,
      "loss": 0.1056,
      "step": 4585
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.180738708538722,
      "learning_rate": 3.357220465659609e-06,
      "loss": 0.2023,
      "step": 4586
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9127680294804665,
      "learning_rate": 3.3551580715682292e-06,
      "loss": 0.1642,
      "step": 4587
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0592064115722888,
      "learning_rate": 3.353095991241335e-06,
      "loss": 0.1659,
      "step": 4588
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7986429380523816,
      "learning_rate": 3.351034225072284e-06,
      "loss": 0.137,
      "step": 4589
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8278908961363475,
      "learning_rate": 3.3489727734543686e-06,
      "loss": 0.139,
      "step": 4590
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8320387730050671,
      "learning_rate": 3.3469116367808267e-06,
      "loss": 0.1951,
      "step": 4591
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0277599958162604,
      "learning_rate": 3.3448508154448333e-06,
      "loss": 0.1803,
      "step": 4592
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7807217366472202,
      "learning_rate": 3.342790309839503e-06,
      "loss": 0.1055,
      "step": 4593
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8110272910399833,
      "learning_rate": 3.34073012035789e-06,
      "loss": 0.1774,
      "step": 4594
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6661303118912044,
      "learning_rate": 3.338670247392992e-06,
      "loss": 0.13,
      "step": 4595
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1728342322457213,
      "learning_rate": 3.3366106913377416e-06,
      "loss": 0.2354,
      "step": 4596
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0326021574357824,
      "learning_rate": 3.3345514525850125e-06,
      "loss": 0.1928,
      "step": 4597
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9106358544487076,
      "learning_rate": 3.332492531527619e-06,
      "loss": 0.133,
      "step": 4598
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7321092094031858,
      "learning_rate": 3.330433928558313e-06,
      "loss": 0.1589,
      "step": 4599
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9946287814713226,
      "learning_rate": 3.328375644069788e-06,
      "loss": 0.1945,
      "step": 4600
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7877409897337574,
      "learning_rate": 3.3263176784546735e-06,
      "loss": 0.1603,
      "step": 4601
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9837890652917382,
      "learning_rate": 3.324260032105543e-06,
      "loss": 0.1881,
      "step": 4602
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.803746496720738,
      "learning_rate": 3.322202705414903e-06,
      "loss": 0.1356,
      "step": 4603
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8749032181759611,
      "learning_rate": 3.3201456987752034e-06,
      "loss": 0.1812,
      "step": 4604
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6701665415306419,
      "learning_rate": 3.3180890125788314e-06,
      "loss": 0.1331,
      "step": 4605
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9865490871229168,
      "learning_rate": 3.3160326472181127e-06,
      "loss": 0.1997,
      "step": 4606
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3658752086063302,
      "learning_rate": 3.3139766030853147e-06,
      "loss": 0.2497,
      "step": 4607
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8195093719383653,
      "learning_rate": 3.3119208805726416e-06,
      "loss": 0.1477,
      "step": 4608
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9217987883637799,
      "learning_rate": 3.3098654800722362e-06,
      "loss": 0.1706,
      "step": 4609
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2129340293902025,
      "learning_rate": 3.307810401976177e-06,
      "loss": 0.2327,
      "step": 4610
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8750391608409409,
      "learning_rate": 3.305755646676486e-06,
      "loss": 0.1536,
      "step": 4611
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8983894602547882,
      "learning_rate": 3.3037012145651202e-06,
      "loss": 0.1776,
      "step": 4612
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9071893849070287,
      "learning_rate": 3.3016471060339773e-06,
      "loss": 0.1686,
      "step": 4613
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8391424752787728,
      "learning_rate": 3.299593321474891e-06,
      "loss": 0.1124,
      "step": 4614
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0876608009478774,
      "learning_rate": 3.2975398612796346e-06,
      "loss": 0.1662,
      "step": 4615
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0777583521502876,
      "learning_rate": 3.295486725839919e-06,
      "loss": 0.2103,
      "step": 4616
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2074986321544159,
      "learning_rate": 3.2934339155473938e-06,
      "loss": 0.2523,
      "step": 4617
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2069908600482675,
      "learning_rate": 3.291381430793647e-06,
      "loss": 0.2238,
      "step": 4618
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1741390332002202,
      "learning_rate": 3.2893292719702025e-06,
      "loss": 0.237,
      "step": 4619
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8934270592470233,
      "learning_rate": 3.2872774394685234e-06,
      "loss": 0.1759,
      "step": 4620
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.512355432035999,
      "learning_rate": 3.2852259336800106e-06,
      "loss": 0.0997,
      "step": 4621
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8174303309055896,
      "learning_rate": 3.2831747549960026e-06,
      "loss": 0.1367,
      "step": 4622
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7316293061132229,
      "learning_rate": 3.281123903807775e-06,
      "loss": 0.1205,
      "step": 4623
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9171273038336089,
      "learning_rate": 3.27907338050654e-06,
      "loss": 0.1827,
      "step": 4624
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8944997861914638,
      "learning_rate": 3.27702318548345e-06,
      "loss": 0.1264,
      "step": 4625
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9488221518048677,
      "learning_rate": 3.2749733191295946e-06,
      "loss": 0.1765,
      "step": 4626
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8824570539684398,
      "learning_rate": 3.272923781835996e-06,
      "loss": 0.1527,
      "step": 4627
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9289769157745674,
      "learning_rate": 3.270874573993618e-06,
      "loss": 0.1457,
      "step": 4628
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8297932745124901,
      "learning_rate": 3.2688256959933618e-06,
      "loss": 0.1606,
      "step": 4629
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7616949266201526,
      "learning_rate": 3.2667771482260645e-06,
      "loss": 0.148,
      "step": 4630
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.173681715311864,
      "learning_rate": 3.264728931082499e-06,
      "loss": 0.2068,
      "step": 4631
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8743689494479515,
      "learning_rate": 3.2626810449533787e-06,
      "loss": 0.1314,
      "step": 4632
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.008504857984711,
      "learning_rate": 3.260633490229348e-06,
      "loss": 0.1437,
      "step": 4633
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.141785425342125,
      "learning_rate": 3.258586267300994e-06,
      "loss": 0.1666,
      "step": 4634
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1677927229255276,
      "learning_rate": 3.256539376558836e-06,
      "loss": 0.2524,
      "step": 4635
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8161048497938125,
      "learning_rate": 3.2544928183933332e-06,
      "loss": 0.1212,
      "step": 4636
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1719568917740018,
      "learning_rate": 3.2524465931948806e-06,
      "loss": 0.2168,
      "step": 4637
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.049685953760718,
      "learning_rate": 3.2504007013538088e-06,
      "loss": 0.1827,
      "step": 4638
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9352824492157631,
      "learning_rate": 3.248355143260383e-06,
      "loss": 0.156,
      "step": 4639
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8012241066958157,
      "learning_rate": 3.246309919304811e-06,
      "loss": 0.1284,
      "step": 4640
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0609089949376782,
      "learning_rate": 3.2442650298772298e-06,
      "loss": 0.1582,
      "step": 4641
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8579941623799037,
      "learning_rate": 3.2422204753677172e-06,
      "loss": 0.1672,
      "step": 4642
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0122259781526974,
      "learning_rate": 3.2401762561662846e-06,
      "loss": 0.1988,
      "step": 4643
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9781313079450916,
      "learning_rate": 3.238132372662881e-06,
      "loss": 0.147,
      "step": 4644
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9135127982577672,
      "learning_rate": 3.2360888252473898e-06,
      "loss": 0.1835,
      "step": 4645
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6737067851456214,
      "learning_rate": 3.234045614309632e-06,
      "loss": 0.1241,
      "step": 4646
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.969767821902647,
      "learning_rate": 3.2320027402393634e-06,
      "loss": 0.1672,
      "step": 4647
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7416989508436309,
      "learning_rate": 3.2299602034262754e-06,
      "loss": 0.1557,
      "step": 4648
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9357358296453814,
      "learning_rate": 3.2279180042599965e-06,
      "loss": 0.183,
      "step": 4649
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9622169165867278,
      "learning_rate": 3.2258761431300868e-06,
      "loss": 0.179,
      "step": 4650
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1867241464361895,
      "learning_rate": 3.2238346204260484e-06,
      "loss": 0.1641,
      "step": 4651
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0254252902786325,
      "learning_rate": 3.2217934365373138e-06,
      "loss": 0.1717,
      "step": 4652
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9846442372903196,
      "learning_rate": 3.2197525918532524e-06,
      "loss": 0.206,
      "step": 4653
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1399545041059882,
      "learning_rate": 3.2177120867631683e-06,
      "loss": 0.2179,
      "step": 4654
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5758285043719985,
      "learning_rate": 3.2156719216563037e-06,
      "loss": 0.064,
      "step": 4655
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8716949354551982,
      "learning_rate": 3.2136320969218298e-06,
      "loss": 0.1738,
      "step": 4656
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8591966437610068,
      "learning_rate": 3.2115926129488596e-06,
      "loss": 0.1623,
      "step": 4657
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9320967968744298,
      "learning_rate": 3.2095534701264363e-06,
      "loss": 0.1701,
      "step": 4658
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3896971306776897,
      "learning_rate": 3.207514668843541e-06,
      "loss": 0.1887,
      "step": 4659
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9027084214937802,
      "learning_rate": 3.205476209489089e-06,
      "loss": 0.1455,
      "step": 4660
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9902470570050866,
      "learning_rate": 3.2034380924519283e-06,
      "loss": 0.1474,
      "step": 4661
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5751687605867268,
      "learning_rate": 3.201400318120843e-06,
      "loss": 0.0883,
      "step": 4662
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.907805202910934,
      "learning_rate": 3.199362886884555e-06,
      "loss": 0.1559,
      "step": 4663
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.20419968864583,
      "learning_rate": 3.1973257991317157e-06,
      "loss": 0.2188,
      "step": 4664
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2012780330053823,
      "learning_rate": 3.195289055250913e-06,
      "loss": 0.2206,
      "step": 4665
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1550470062555551,
      "learning_rate": 3.1932526556306697e-06,
      "loss": 0.1897,
      "step": 4666
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9821163239671993,
      "learning_rate": 3.1912166006594432e-06,
      "loss": 0.2205,
      "step": 4667
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7689602517285052,
      "learning_rate": 3.189180890725623e-06,
      "loss": 0.1044,
      "step": 4668
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8391607507124611,
      "learning_rate": 3.1871455262175356e-06,
      "loss": 0.1728,
      "step": 4669
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0054510064736555,
      "learning_rate": 3.185110507523439e-06,
      "loss": 0.1771,
      "step": 4670
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7423220948980407,
      "learning_rate": 3.183075835031527e-06,
      "loss": 0.1333,
      "step": 4671
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0073715944520527,
      "learning_rate": 3.1810415091299266e-06,
      "loss": 0.1763,
      "step": 4672
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.128621795086374,
      "learning_rate": 3.1790075302066993e-06,
      "loss": 0.2009,
      "step": 4673
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7773586165632235,
      "learning_rate": 3.17697389864984e-06,
      "loss": 0.0982,
      "step": 4674
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8929705715401999,
      "learning_rate": 3.174940614847278e-06,
      "loss": 0.1724,
      "step": 4675
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.839676269033263,
      "learning_rate": 3.172907679186875e-06,
      "loss": 0.1075,
      "step": 4676
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7182422297544032,
      "learning_rate": 3.1708750920564266e-06,
      "loss": 0.1476,
      "step": 4677
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9239410154652626,
      "learning_rate": 3.1688428538436634e-06,
      "loss": 0.174,
      "step": 4678
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.932465751424384,
      "learning_rate": 3.166810964936248e-06,
      "loss": 0.1695,
      "step": 4679
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6765082632263673,
      "learning_rate": 3.1647794257217745e-06,
      "loss": 0.106,
      "step": 4680
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9958074602467104,
      "learning_rate": 3.1627482365877747e-06,
      "loss": 0.1484,
      "step": 4681
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9037742544887448,
      "learning_rate": 3.160717397921711e-06,
      "loss": 0.1603,
      "step": 4682
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0312237570634508,
      "learning_rate": 3.1586869101109796e-06,
      "loss": 0.1917,
      "step": 4683
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9883708825961035,
      "learning_rate": 3.156656773542906e-06,
      "loss": 0.1798,
      "step": 4684
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.992635483852042,
      "learning_rate": 3.1546269886047586e-06,
      "loss": 0.1848,
      "step": 4685
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7654255115355624,
      "learning_rate": 3.1525975556837278e-06,
      "loss": 0.1048,
      "step": 4686
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7624053390828998,
      "learning_rate": 3.150568475166943e-06,
      "loss": 0.1288,
      "step": 4687
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8629100268672427,
      "learning_rate": 3.1485397474414637e-06,
      "loss": 0.1453,
      "step": 4688
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9120497335856028,
      "learning_rate": 3.146511372894283e-06,
      "loss": 0.1312,
      "step": 4689
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1637253539741486,
      "learning_rate": 3.1444833519123286e-06,
      "loss": 0.192,
      "step": 4690
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8792423793765689,
      "learning_rate": 3.1424556848824573e-06,
      "loss": 0.1752,
      "step": 4691
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7555037467913729,
      "learning_rate": 3.140428372191461e-06,
      "loss": 0.2004,
      "step": 4692
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2998532857818257,
      "learning_rate": 3.138401414226062e-06,
      "loss": 0.2404,
      "step": 4693
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8808909061276046,
      "learning_rate": 3.136374811372916e-06,
      "loss": 0.1516,
      "step": 4694
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9460629918439518,
      "learning_rate": 3.1343485640186104e-06,
      "loss": 0.1761,
      "step": 4695
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7018433602696402,
      "learning_rate": 3.132322672549668e-06,
      "loss": 0.1305,
      "step": 4696
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8919964621390223,
      "learning_rate": 3.1302971373525385e-06,
      "loss": 0.1554,
      "step": 4697
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.825481481652663,
      "learning_rate": 3.1282719588136072e-06,
      "loss": 0.1432,
      "step": 4698
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3014503830079158,
      "learning_rate": 3.1262471373191904e-06,
      "loss": 0.2389,
      "step": 4699
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9844062586557191,
      "learning_rate": 3.1242226732555355e-06,
      "loss": 0.1952,
      "step": 4700
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8262983021723914,
      "learning_rate": 3.122198567008822e-06,
      "loss": 0.1346,
      "step": 4701
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8414212209562214,
      "learning_rate": 3.120174818965164e-06,
      "loss": 0.1076,
      "step": 4702
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7816012662712774,
      "learning_rate": 3.1181514295106024e-06,
      "loss": 0.1292,
      "step": 4703
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.933007027669192,
      "learning_rate": 3.1161283990311127e-06,
      "loss": 0.1959,
      "step": 4704
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8142592024916373,
      "learning_rate": 3.1141057279126006e-06,
      "loss": 0.127,
      "step": 4705
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8977675565069196,
      "learning_rate": 3.1120834165409066e-06,
      "loss": 0.1747,
      "step": 4706
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0142617059707162,
      "learning_rate": 3.110061465301795e-06,
      "loss": 0.1856,
      "step": 4707
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5787933122528763,
      "learning_rate": 3.1080398745809733e-06,
      "loss": 0.1255,
      "step": 4708
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9028827655613233,
      "learning_rate": 3.1060186447640685e-06,
      "loss": 0.1919,
      "step": 4709
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2166440916401415,
      "learning_rate": 3.1039977762366447e-06,
      "loss": 0.2121,
      "step": 4710
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0305622417918696,
      "learning_rate": 3.1019772693841952e-06,
      "loss": 0.18,
      "step": 4711
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1785041015568778,
      "learning_rate": 3.0999571245921475e-06,
      "loss": 0.2293,
      "step": 4712
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9873133416148526,
      "learning_rate": 3.097937342245855e-06,
      "loss": 0.1838,
      "step": 4713
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8367546086231239,
      "learning_rate": 3.0959179227306056e-06,
      "loss": 0.141,
      "step": 4714
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9011294115088571,
      "learning_rate": 3.093898866431617e-06,
      "loss": 0.1524,
      "step": 4715
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8690168195115833,
      "learning_rate": 3.0918801737340374e-06,
      "loss": 0.1209,
      "step": 4716
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8852269284734271,
      "learning_rate": 3.0898618450229455e-06,
      "loss": 0.1623,
      "step": 4717
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9603474170706017,
      "learning_rate": 3.087843880683351e-06,
      "loss": 0.1798,
      "step": 4718
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8125298717035702,
      "learning_rate": 3.0858262811001944e-06,
      "loss": 0.1355,
      "step": 4719
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0724194260988933,
      "learning_rate": 3.0838090466583458e-06,
      "loss": 0.1462,
      "step": 4720
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9483391604822478,
      "learning_rate": 3.0817921777426063e-06,
      "loss": 0.1816,
      "step": 4721
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9912235115481471,
      "learning_rate": 3.079775674737707e-06,
      "loss": 0.1307,
      "step": 4722
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.102864631891142,
      "learning_rate": 3.0777595380283095e-06,
      "loss": 0.2193,
      "step": 4723
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9221522872250497,
      "learning_rate": 3.0757437679990043e-06,
      "loss": 0.1676,
      "step": 4724
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.059514108826089,
      "learning_rate": 3.073728365034313e-06,
      "loss": 0.1551,
      "step": 4725
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8056517009676307,
      "learning_rate": 3.0717133295186884e-06,
      "loss": 0.1267,
      "step": 4726
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9532885277281021,
      "learning_rate": 3.06969866183651e-06,
      "loss": 0.145,
      "step": 4727
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9391945319040846,
      "learning_rate": 3.06768436237209e-06,
      "loss": 0.1469,
      "step": 4728
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9591246376952401,
      "learning_rate": 3.0656704315096675e-06,
      "loss": 0.1921,
      "step": 4729
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8892942626510202,
      "learning_rate": 3.0636568696334164e-06,
      "loss": 0.126,
      "step": 4730
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6044399870731322,
      "learning_rate": 3.061643677127435e-06,
      "loss": 0.1473,
      "step": 4731
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9800494029939545,
      "learning_rate": 3.059630854375755e-06,
      "loss": 0.1708,
      "step": 4732
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3165675312630576,
      "learning_rate": 3.0576184017623334e-06,
      "loss": 0.2568,
      "step": 4733
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7308396428443287,
      "learning_rate": 3.0556063196710595e-06,
      "loss": 0.1618,
      "step": 4734
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7885930793030683,
      "learning_rate": 3.0535946084857503e-06,
      "loss": 0.1542,
      "step": 4735
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8603305423208258,
      "learning_rate": 3.051583268590156e-06,
      "loss": 0.1693,
      "step": 4736
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8821665873058916,
      "learning_rate": 3.0495723003679493e-06,
      "loss": 0.135,
      "step": 4737
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9575363023584477,
      "learning_rate": 3.0475617042027384e-06,
      "loss": 0.1669,
      "step": 4738
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9822071519773357,
      "learning_rate": 3.0455514804780572e-06,
      "loss": 0.1723,
      "step": 4739
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8440556137190863,
      "learning_rate": 3.043541629577368e-06,
      "loss": 0.1191,
      "step": 4740
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9619401740069229,
      "learning_rate": 3.0415321518840647e-06,
      "loss": 0.195,
      "step": 4741
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0030515299759246,
      "learning_rate": 3.0395230477814677e-06,
      "loss": 0.1434,
      "step": 4742
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0418609164577066,
      "learning_rate": 3.0375143176528264e-06,
      "loss": 0.1712,
      "step": 4743
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9622703821668259,
      "learning_rate": 3.0355059618813216e-06,
      "loss": 0.1338,
      "step": 4744
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8763670983899442,
      "learning_rate": 3.0334979808500583e-06,
      "loss": 0.1438,
      "step": 4745
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0173447446403108,
      "learning_rate": 3.0314903749420723e-06,
      "loss": 0.1774,
      "step": 4746
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1220916482726853,
      "learning_rate": 3.0294831445403285e-06,
      "loss": 0.1647,
      "step": 4747
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8496556456535859,
      "learning_rate": 3.0274762900277196e-06,
      "loss": 0.1409,
      "step": 4748
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0435682996361686,
      "learning_rate": 3.0254698117870664e-06,
      "loss": 0.1889,
      "step": 4749
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.956638168714782,
      "learning_rate": 3.023463710201116e-06,
      "loss": 0.1186,
      "step": 4750
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8938649234206243,
      "learning_rate": 3.021457985652545e-06,
      "loss": 0.1616,
      "step": 4751
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9319665123173214,
      "learning_rate": 3.019452638523963e-06,
      "loss": 0.1686,
      "step": 4752
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0465032874120548,
      "learning_rate": 3.0174476691979005e-06,
      "loss": 0.1822,
      "step": 4753
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6969689309043711,
      "learning_rate": 3.0154430780568196e-06,
      "loss": 0.1448,
      "step": 4754
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9003995500944247,
      "learning_rate": 3.013438865483109e-06,
      "loss": 0.1646,
      "step": 4755
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8606296099847768,
      "learning_rate": 3.0114350318590847e-06,
      "loss": 0.1126,
      "step": 4756
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.082992204586381,
      "learning_rate": 3.009431577566991e-06,
      "loss": 0.1784,
      "step": 4757
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6970237592584401,
      "learning_rate": 3.0074285029890016e-06,
      "loss": 0.1404,
      "step": 4758
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8866239586696696,
      "learning_rate": 3.0054258085072153e-06,
      "loss": 0.1685,
      "step": 4759
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8464996950373507,
      "learning_rate": 3.0034234945036587e-06,
      "loss": 0.167,
      "step": 4760
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.228618416155985,
      "learning_rate": 3.0014215613602884e-06,
      "loss": 0.2002,
      "step": 4761
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8704290531227764,
      "learning_rate": 2.9994200094589844e-06,
      "loss": 0.1633,
      "step": 4762
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2394523241248638,
      "learning_rate": 2.997418839181556e-06,
      "loss": 0.1973,
      "step": 4763
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8064928044798542,
      "learning_rate": 2.995418050909742e-06,
      "loss": 0.1798,
      "step": 4764
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.062920202697664,
      "learning_rate": 2.993417645025204e-06,
      "loss": 0.1433,
      "step": 4765
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8385776645017993,
      "learning_rate": 2.9914176219095327e-06,
      "loss": 0.1085,
      "step": 4766
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0197112439293319,
      "learning_rate": 2.989417981944247e-06,
      "loss": 0.1732,
      "step": 4767
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.804082981864474,
      "learning_rate": 2.98741872551079e-06,
      "loss": 0.1378,
      "step": 4768
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0056812833276678,
      "learning_rate": 2.985419852990534e-06,
      "loss": 0.1835,
      "step": 4769
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.716864732055956,
      "learning_rate": 2.983421364764777e-06,
      "loss": 0.1202,
      "step": 4770
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.084244550595785,
      "learning_rate": 2.9814232612147444e-06,
      "loss": 0.2024,
      "step": 4771
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8658803581171032,
      "learning_rate": 2.979425542721587e-06,
      "loss": 0.1246,
      "step": 4772
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.916954278597525,
      "learning_rate": 2.9774282096663822e-06,
      "loss": 0.1451,
      "step": 4773
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0221929370304814,
      "learning_rate": 2.9754312624301335e-06,
      "loss": 0.2156,
      "step": 4774
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8027431827048073,
      "learning_rate": 2.9734347013937752e-06,
      "loss": 0.1658,
      "step": 4775
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8315427209831505,
      "learning_rate": 2.971438526938163e-06,
      "loss": 0.1248,
      "step": 4776
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9582127288730451,
      "learning_rate": 2.9694427394440808e-06,
      "loss": 0.1722,
      "step": 4777
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0903193866119865,
      "learning_rate": 2.9674473392922366e-06,
      "loss": 0.1886,
      "step": 4778
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8893955154132113,
      "learning_rate": 2.9654523268632683e-06,
      "loss": 0.1465,
      "step": 4779
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9549492086593641,
      "learning_rate": 2.9634577025377365e-06,
      "loss": 0.172,
      "step": 4780
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8807861039043311,
      "learning_rate": 2.961463466696129e-06,
      "loss": 0.1218,
      "step": 4781
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.027094832454673,
      "learning_rate": 2.9594696197188597e-06,
      "loss": 0.1721,
      "step": 4782
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8206977142731728,
      "learning_rate": 2.957476161986268e-06,
      "loss": 0.1606,
      "step": 4783
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8951631196813647,
      "learning_rate": 2.9554830938786195e-06,
      "loss": 0.1533,
      "step": 4784
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1185414486313783,
      "learning_rate": 2.9534904157761035e-06,
      "loss": 0.2101,
      "step": 4785
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.006288592247412,
      "learning_rate": 2.951498128058839e-06,
      "loss": 0.1806,
      "step": 4786
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1030353571482443,
      "learning_rate": 2.949506231106867e-06,
      "loss": 0.2169,
      "step": 4787
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9796649893215852,
      "learning_rate": 2.947514725300155e-06,
      "loss": 0.1763,
      "step": 4788
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.813726248407595,
      "learning_rate": 2.945523611018596e-06,
      "loss": 0.1267,
      "step": 4789
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7249971524991313,
      "learning_rate": 2.9435328886420083e-06,
      "loss": 0.1802,
      "step": 4790
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7732774892790756,
      "learning_rate": 2.9415425585501358e-06,
      "loss": 0.1548,
      "step": 4791
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2141814842452416,
      "learning_rate": 2.9395526211226455e-06,
      "loss": 0.2432,
      "step": 4792
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.028388431257084,
      "learning_rate": 2.9375630767391327e-06,
      "loss": 0.2034,
      "step": 4793
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0034995196157483,
      "learning_rate": 2.9355739257791162e-06,
      "loss": 0.1742,
      "step": 4794
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1598721453309726,
      "learning_rate": 2.9335851686220383e-06,
      "loss": 0.2087,
      "step": 4795
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8173224979188186,
      "learning_rate": 2.9315968056472676e-06,
      "loss": 0.1733,
      "step": 4796
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3188343627103112,
      "learning_rate": 2.9296088372341005e-06,
      "loss": 0.1905,
      "step": 4797
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9068180675228201,
      "learning_rate": 2.927621263761752e-06,
      "loss": 0.1576,
      "step": 4798
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.665957550717257,
      "learning_rate": 2.9256340856093655e-06,
      "loss": 0.1345,
      "step": 4799
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6578799790089187,
      "learning_rate": 2.92364730315601e-06,
      "loss": 0.1321,
      "step": 4800
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8790663317083451,
      "learning_rate": 2.921660916780676e-06,
      "loss": 0.1026,
      "step": 4801
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6276982356140018,
      "learning_rate": 2.9196749268622805e-06,
      "loss": 0.0725,
      "step": 4802
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1303678960663681,
      "learning_rate": 2.917689333779662e-06,
      "loss": 0.2218,
      "step": 4803
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9570048230430263,
      "learning_rate": 2.9157041379115887e-06,
      "loss": 0.1226,
      "step": 4804
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6029698039622101,
      "learning_rate": 2.913719339636748e-06,
      "loss": 0.0908,
      "step": 4805
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8071461689604195,
      "learning_rate": 2.911734939333754e-06,
      "loss": 0.131,
      "step": 4806
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8697785503460298,
      "learning_rate": 2.909750937381144e-06,
      "loss": 0.1537,
      "step": 4807
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7987084658563163,
      "learning_rate": 2.9077673341573794e-06,
      "loss": 0.107,
      "step": 4808
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.909522332604642,
      "learning_rate": 2.9057841300408456e-06,
      "loss": 0.1718,
      "step": 4809
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7867016939766268,
      "learning_rate": 2.9038013254098522e-06,
      "loss": 0.121,
      "step": 4810
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.959135463381658,
      "learning_rate": 2.9018189206426324e-06,
      "loss": 0.1488,
      "step": 4811
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.075103323579178,
      "learning_rate": 2.8998369161173423e-06,
      "loss": 0.1805,
      "step": 4812
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7768557722895444,
      "learning_rate": 2.8978553122120635e-06,
      "loss": 0.1033,
      "step": 4813
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8781246818661913,
      "learning_rate": 2.8958741093047985e-06,
      "loss": 0.1213,
      "step": 4814
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9727553220334444,
      "learning_rate": 2.8938933077734765e-06,
      "loss": 0.162,
      "step": 4815
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7935084646691422,
      "learning_rate": 2.8919129079959474e-06,
      "loss": 0.1669,
      "step": 4816
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0993067399361633,
      "learning_rate": 2.889932910349986e-06,
      "loss": 0.2098,
      "step": 4817
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2947943912878046,
      "learning_rate": 2.8879533152132904e-06,
      "loss": 0.2097,
      "step": 4818
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8688280512115046,
      "learning_rate": 2.8859741229634787e-06,
      "loss": 0.127,
      "step": 4819
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6402263096526185,
      "learning_rate": 2.8839953339781013e-06,
      "loss": 0.138,
      "step": 4820
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0807794460482705,
      "learning_rate": 2.8820169486346193e-06,
      "loss": 0.2171,
      "step": 4821
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9710305423306551,
      "learning_rate": 2.8800389673104244e-06,
      "loss": 0.2059,
      "step": 4822
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0157273312425896,
      "learning_rate": 2.8780613903828305e-06,
      "loss": 0.1486,
      "step": 4823
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0965099272987313,
      "learning_rate": 2.876084218229072e-06,
      "loss": 0.2052,
      "step": 4824
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.836446702287726,
      "learning_rate": 2.874107451226309e-06,
      "loss": 0.1559,
      "step": 4825
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.878137083378054,
      "learning_rate": 2.872131089751623e-06,
      "loss": 0.1671,
      "step": 4826
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.027135409409414,
      "learning_rate": 2.8701551341820165e-06,
      "loss": 0.1768,
      "step": 4827
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0134466071145143,
      "learning_rate": 2.8681795848944185e-06,
      "loss": 0.1507,
      "step": 4828
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.23006419872225,
      "learning_rate": 2.8662044422656753e-06,
      "loss": 0.212,
      "step": 4829
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.827899827946285,
      "learning_rate": 2.8642297066725577e-06,
      "loss": 0.1363,
      "step": 4830
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9714034647235003,
      "learning_rate": 2.8622553784917647e-06,
      "loss": 0.2037,
      "step": 4831
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9299197230262347,
      "learning_rate": 2.860281458099908e-06,
      "loss": 0.1842,
      "step": 4832
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6460089547109259,
      "learning_rate": 2.8583079458735287e-06,
      "loss": 0.1189,
      "step": 4833
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.172698705510818,
      "learning_rate": 2.8563348421890858e-06,
      "loss": 0.2359,
      "step": 4834
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0033515192590712,
      "learning_rate": 2.8543621474229623e-06,
      "loss": 0.1835,
      "step": 4835
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8584835187902697,
      "learning_rate": 2.8523898619514623e-06,
      "loss": 0.2032,
      "step": 4836
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8348646256394,
      "learning_rate": 2.8504179861508143e-06,
      "loss": 0.1586,
      "step": 4837
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9005735993715169,
      "learning_rate": 2.8484465203971635e-06,
      "loss": 0.1805,
      "step": 4838
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9019604243017688,
      "learning_rate": 2.8464754650665815e-06,
      "loss": 0.1814,
      "step": 4839
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7632945770478157,
      "learning_rate": 2.84450482053506e-06,
      "loss": 0.1161,
      "step": 4840
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8225135741914956,
      "learning_rate": 2.84253458717851e-06,
      "loss": 0.1311,
      "step": 4841
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6683193615935838,
      "learning_rate": 2.840564765372773e-06,
      "loss": 0.1092,
      "step": 4842
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8458115034340621,
      "learning_rate": 2.838595355493601e-06,
      "loss": 0.1536,
      "step": 4843
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1178934794828033,
      "learning_rate": 2.836626357916672e-06,
      "loss": 0.2272,
      "step": 4844
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8472580259674335,
      "learning_rate": 2.8346577730175867e-06,
      "loss": 0.1356,
      "step": 4845
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0105084846074728,
      "learning_rate": 2.8326896011718654e-06,
      "loss": 0.1956,
      "step": 4846
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7959298869524345,
      "learning_rate": 2.8307218427549487e-06,
      "loss": 0.1517,
      "step": 4847
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9511320940520632,
      "learning_rate": 2.828754498142201e-06,
      "loss": 0.2027,
      "step": 4848
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7518357207977259,
      "learning_rate": 2.8267875677089053e-06,
      "loss": 0.1668,
      "step": 4849
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9793635173183574,
      "learning_rate": 2.8248210518302675e-06,
      "loss": 0.1686,
      "step": 4850
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9146176217753562,
      "learning_rate": 2.822854950881413e-06,
      "loss": 0.1599,
      "step": 4851
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.743975218666986,
      "learning_rate": 2.820889265237389e-06,
      "loss": 0.1562,
      "step": 4852
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1435401370818297,
      "learning_rate": 2.818923995273162e-06,
      "loss": 0.2269,
      "step": 4853
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8450150930946893,
      "learning_rate": 2.816959141363621e-06,
      "loss": 0.1425,
      "step": 4854
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9921950584478195,
      "learning_rate": 2.814994703883575e-06,
      "loss": 0.1814,
      "step": 4855
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7886715820051566,
      "learning_rate": 2.8130306832077535e-06,
      "loss": 0.1743,
      "step": 4856
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9107799483175354,
      "learning_rate": 2.8110670797108057e-06,
      "loss": 0.1381,
      "step": 4857
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9106638857420376,
      "learning_rate": 2.8091038937673034e-06,
      "loss": 0.1875,
      "step": 4858
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8197867054484369,
      "learning_rate": 2.8071411257517356e-06,
      "loss": 0.1699,
      "step": 4859
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1413741824611612,
      "learning_rate": 2.805178776038514e-06,
      "loss": 0.1477,
      "step": 4860
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0262320467470374,
      "learning_rate": 2.8032168450019703e-06,
      "loss": 0.1656,
      "step": 4861
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6711636723178491,
      "learning_rate": 2.801255333016355e-06,
      "loss": 0.1232,
      "step": 4862
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.065653770566519,
      "learning_rate": 2.7992942404558404e-06,
      "loss": 0.1827,
      "step": 4863
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.077839493639535,
      "learning_rate": 2.797333567694515e-06,
      "loss": 0.2016,
      "step": 4864
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0223102259983146,
      "learning_rate": 2.7953733151063948e-06,
      "loss": 0.1949,
      "step": 4865
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8268838621654786,
      "learning_rate": 2.7934134830654093e-06,
      "loss": 0.1819,
      "step": 4866
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5175993121355847,
      "learning_rate": 2.7914540719454097e-06,
      "loss": 0.0662,
      "step": 4867
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9211252321467628,
      "learning_rate": 2.7894950821201638e-06,
      "loss": 0.1396,
      "step": 4868
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8043077136518327,
      "learning_rate": 2.7875365139633647e-06,
      "loss": 0.1416,
      "step": 4869
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1046337176119685,
      "learning_rate": 2.7855783678486203e-06,
      "loss": 0.1707,
      "step": 4870
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.092899806341238,
      "learning_rate": 2.7836206441494618e-06,
      "loss": 0.181,
      "step": 4871
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.18064151717057,
      "learning_rate": 2.781663343239337e-06,
      "loss": 0.1735,
      "step": 4872
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9011248429938981,
      "learning_rate": 2.779706465491614e-06,
      "loss": 0.1834,
      "step": 4873
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9031445572440274,
      "learning_rate": 2.777750011279581e-06,
      "loss": 0.1511,
      "step": 4874
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9627888897745234,
      "learning_rate": 2.7757939809764413e-06,
      "loss": 0.1934,
      "step": 4875
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0951358067464123,
      "learning_rate": 2.7738383749553254e-06,
      "loss": 0.2187,
      "step": 4876
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0749176552623034,
      "learning_rate": 2.7718831935892753e-06,
      "loss": 0.1593,
      "step": 4877
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0028915598066523,
      "learning_rate": 2.7699284372512563e-06,
      "loss": 0.1685,
      "step": 4878
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9835473543079484,
      "learning_rate": 2.76797410631415e-06,
      "loss": 0.2003,
      "step": 4879
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3698175117798213,
      "learning_rate": 2.7660202011507585e-06,
      "loss": 0.2021,
      "step": 4880
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9747997004958198,
      "learning_rate": 2.7640667221338013e-06,
      "loss": 0.1783,
      "step": 4881
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8997395472062103,
      "learning_rate": 2.7621136696359186e-06,
      "loss": 0.1839,
      "step": 4882
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1065298608959115,
      "learning_rate": 2.7601610440296673e-06,
      "loss": 0.2226,
      "step": 4883
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7886372979369028,
      "learning_rate": 2.7582088456875255e-06,
      "loss": 0.1396,
      "step": 4884
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9154607089372749,
      "learning_rate": 2.756257074981885e-06,
      "loss": 0.1802,
      "step": 4885
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.107841802877891,
      "learning_rate": 2.754305732285058e-06,
      "loss": 0.1832,
      "step": 4886
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.929118655787956,
      "learning_rate": 2.7523548179692806e-06,
      "loss": 0.1594,
      "step": 4887
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8621829802577224,
      "learning_rate": 2.7504043324067e-06,
      "loss": 0.1534,
      "step": 4888
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8875096291707457,
      "learning_rate": 2.7484542759693846e-06,
      "loss": 0.143,
      "step": 4889
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7674300583388913,
      "learning_rate": 2.7465046490293206e-06,
      "loss": 0.1385,
      "step": 4890
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5662707365314065,
      "learning_rate": 2.744555451958413e-06,
      "loss": 0.11,
      "step": 4891
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.930931617103319,
      "learning_rate": 2.7426066851284816e-06,
      "loss": 0.1605,
      "step": 4892
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9009581868697794,
      "learning_rate": 2.7406583489112694e-06,
      "loss": 0.1577,
      "step": 4893
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.031723586198371,
      "learning_rate": 2.7387104436784336e-06,
      "loss": 0.2007,
      "step": 4894
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.135196708512929,
      "learning_rate": 2.7367629698015486e-06,
      "loss": 0.2075,
      "step": 4895
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.029401131757408,
      "learning_rate": 2.7348159276521103e-06,
      "loss": 0.1929,
      "step": 4896
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9617139931323638,
      "learning_rate": 2.7328693176015273e-06,
      "loss": 0.1691,
      "step": 4897
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0602676121517567,
      "learning_rate": 2.7309231400211306e-06,
      "loss": 0.1826,
      "step": 4898
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6933974124064597,
      "learning_rate": 2.728977395282165e-06,
      "loss": 0.1604,
      "step": 4899
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9394910491726669,
      "learning_rate": 2.727032083755795e-06,
      "loss": 0.1781,
      "step": 4900
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0431110133861634,
      "learning_rate": 2.7250872058131013e-06,
      "loss": 0.177,
      "step": 4901
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9125153945728175,
      "learning_rate": 2.7231427618250823e-06,
      "loss": 0.2008,
      "step": 4902
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0722480778287622,
      "learning_rate": 2.7211987521626536e-06,
      "loss": 0.2114,
      "step": 4903
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8298620236612262,
      "learning_rate": 2.7192551771966475e-06,
      "loss": 0.1323,
      "step": 4904
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1552043956475029,
      "learning_rate": 2.7173120372978144e-06,
      "loss": 0.2187,
      "step": 4905
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7392139932655177,
      "learning_rate": 2.7153693328368207e-06,
      "loss": 0.1515,
      "step": 4906
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0431608091097093,
      "learning_rate": 2.713427064184251e-06,
      "loss": 0.202,
      "step": 4907
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9397554828447723,
      "learning_rate": 2.7114852317106023e-06,
      "loss": 0.1523,
      "step": 4908
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8085887811788575,
      "learning_rate": 2.7095438357862972e-06,
      "loss": 0.1637,
      "step": 4909
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7822490232194957,
      "learning_rate": 2.7076028767816674e-06,
      "loss": 0.1161,
      "step": 4910
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9753560693573874,
      "learning_rate": 2.705662355066964e-06,
      "loss": 0.2163,
      "step": 4911
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0290225393703578,
      "learning_rate": 2.7037222710123545e-06,
      "loss": 0.1877,
      "step": 4912
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9038007296892648,
      "learning_rate": 2.7017826249879244e-06,
      "loss": 0.166,
      "step": 4913
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1601220851744785,
      "learning_rate": 2.6998434173636687e-06,
      "loss": 0.1974,
      "step": 4914
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1724047181894401,
      "learning_rate": 2.697904648509509e-06,
      "loss": 0.191,
      "step": 4915
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8597624695773306,
      "learning_rate": 2.695966318795276e-06,
      "loss": 0.1394,
      "step": 4916
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9823322978982176,
      "learning_rate": 2.6940284285907205e-06,
      "loss": 0.1795,
      "step": 4917
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7545007544205657,
      "learning_rate": 2.692090978265506e-06,
      "loss": 0.146,
      "step": 4918
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9219104275125721,
      "learning_rate": 2.6901539681892153e-06,
      "loss": 0.1469,
      "step": 4919
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.062550925235499,
      "learning_rate": 2.688217398731344e-06,
      "loss": 0.1834,
      "step": 4920
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9798978509939383,
      "learning_rate": 2.686281270261309e-06,
      "loss": 0.2156,
      "step": 4921
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.016589851735975,
      "learning_rate": 2.6843455831484376e-06,
      "loss": 0.1714,
      "step": 4922
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1261135174922774,
      "learning_rate": 2.682410337761975e-06,
      "loss": 0.1907,
      "step": 4923
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8303924231503154,
      "learning_rate": 2.6804755344710824e-06,
      "loss": 0.1219,
      "step": 4924
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8432149557552374,
      "learning_rate": 2.6785411736448365e-06,
      "loss": 0.1402,
      "step": 4925
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8295936771079949,
      "learning_rate": 2.676607255652228e-06,
      "loss": 0.133,
      "step": 4926
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0205283687895117,
      "learning_rate": 2.674673780862167e-06,
      "loss": 0.1811,
      "step": 4927
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0886087559976536,
      "learning_rate": 2.672740749643474e-06,
      "loss": 0.2034,
      "step": 4928
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.953343014542083,
      "learning_rate": 2.6708081623648896e-06,
      "loss": 0.1876,
      "step": 4929
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0740419365380711,
      "learning_rate": 2.668876019395066e-06,
      "loss": 0.1954,
      "step": 4930
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8844083826647653,
      "learning_rate": 2.6669443211025736e-06,
      "loss": 0.171,
      "step": 4931
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0259010949252416,
      "learning_rate": 2.6650130678558963e-06,
      "loss": 0.1595,
      "step": 4932
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7414898813220253,
      "learning_rate": 2.663082260023432e-06,
      "loss": 0.1286,
      "step": 4933
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8704693791231196,
      "learning_rate": 2.6611518979734974e-06,
      "loss": 0.1466,
      "step": 4934
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0221241635082021,
      "learning_rate": 2.6592219820743197e-06,
      "loss": 0.1724,
      "step": 4935
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9877515163929151,
      "learning_rate": 2.657292512694045e-06,
      "loss": 0.1456,
      "step": 4936
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.955458565765673,
      "learning_rate": 2.6553634902007306e-06,
      "loss": 0.1673,
      "step": 4937
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5987937095392742,
      "learning_rate": 2.6534349149623523e-06,
      "loss": 0.1129,
      "step": 4938
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0849952744452915,
      "learning_rate": 2.651506787346797e-06,
      "loss": 0.1966,
      "step": 4939
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8270842317955355,
      "learning_rate": 2.649579107721868e-06,
      "loss": 0.1315,
      "step": 4940
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8372803192993726,
      "learning_rate": 2.647651876455284e-06,
      "loss": 0.1538,
      "step": 4941
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4000311811375346,
      "learning_rate": 2.6457250939146734e-06,
      "loss": 0.2651,
      "step": 4942
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.018041156863994,
      "learning_rate": 2.6437987604675903e-06,
      "loss": 0.1981,
      "step": 4943
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9630017397462728,
      "learning_rate": 2.641872876481488e-06,
      "loss": 0.1594,
      "step": 4944
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9461052659333563,
      "learning_rate": 2.639947442323746e-06,
      "loss": 0.1635,
      "step": 4945
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9196950444466561,
      "learning_rate": 2.6380224583616517e-06,
      "loss": 0.1818,
      "step": 4946
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0595362492378826,
      "learning_rate": 2.6360979249624087e-06,
      "loss": 0.2316,
      "step": 4947
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9284966626606888,
      "learning_rate": 2.6341738424931353e-06,
      "loss": 0.1385,
      "step": 4948
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9616744996666852,
      "learning_rate": 2.6322502113208613e-06,
      "loss": 0.1703,
      "step": 4949
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0189651750590845,
      "learning_rate": 2.630327031812534e-06,
      "loss": 0.1561,
      "step": 4950
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8362343989100496,
      "learning_rate": 2.628404304335012e-06,
      "loss": 0.1565,
      "step": 4951
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9298926465590625,
      "learning_rate": 2.6264820292550684e-06,
      "loss": 0.1864,
      "step": 4952
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8428261741174568,
      "learning_rate": 2.624560206939387e-06,
      "loss": 0.1039,
      "step": 4953
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8629670862660379,
      "learning_rate": 2.6226388377545732e-06,
      "loss": 0.1936,
      "step": 4954
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9176221248905747,
      "learning_rate": 2.6207179220671385e-06,
      "loss": 0.1785,
      "step": 4955
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1263552613312258,
      "learning_rate": 2.61879746024351e-06,
      "loss": 0.205,
      "step": 4956
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7201854280511801,
      "learning_rate": 2.6168774526500294e-06,
      "loss": 0.1117,
      "step": 4957
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8978325583679906,
      "learning_rate": 2.6149578996529495e-06,
      "loss": 0.1642,
      "step": 4958
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.107118698110606,
      "learning_rate": 2.613038801618439e-06,
      "loss": 0.179,
      "step": 4959
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7684805110304622,
      "learning_rate": 2.611120158912579e-06,
      "loss": 0.1603,
      "step": 4960
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0038605700813508,
      "learning_rate": 2.6092019719013616e-06,
      "loss": 0.1705,
      "step": 4961
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.856933148338922,
      "learning_rate": 2.6072842409506937e-06,
      "loss": 0.1635,
      "step": 4962
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7265176829528964,
      "learning_rate": 2.605366966426396e-06,
      "loss": 0.1668,
      "step": 4963
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2653035807144446,
      "learning_rate": 2.603450148694201e-06,
      "loss": 0.1927,
      "step": 4964
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6966857026458907,
      "learning_rate": 2.601533788119753e-06,
      "loss": 0.1285,
      "step": 4965
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0300545173263491,
      "learning_rate": 2.599617885068614e-06,
      "loss": 0.1441,
      "step": 4966
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1903023341721095,
      "learning_rate": 2.597702439906252e-06,
      "loss": 0.2118,
      "step": 4967
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1070020978735333,
      "learning_rate": 2.595787452998053e-06,
      "loss": 0.2286,
      "step": 4968
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.015053682614415,
      "learning_rate": 2.5938729247093125e-06,
      "loss": 0.2131,
      "step": 4969
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1977043978898612,
      "learning_rate": 2.5919588554052395e-06,
      "loss": 0.2415,
      "step": 4970
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.005005658363005,
      "learning_rate": 2.590045245450955e-06,
      "loss": 0.1832,
      "step": 4971
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9304965662605854,
      "learning_rate": 2.588132095211493e-06,
      "loss": 0.1844,
      "step": 4972
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8391636061721983,
      "learning_rate": 2.5862194050517998e-06,
      "loss": 0.124,
      "step": 4973
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7993799134104876,
      "learning_rate": 2.584307175336734e-06,
      "loss": 0.1349,
      "step": 4974
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7694278462386184,
      "learning_rate": 2.5823954064310653e-06,
      "loss": 0.1108,
      "step": 4975
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.883512382030539,
      "learning_rate": 2.5804840986994762e-06,
      "loss": 0.173,
      "step": 4976
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0466132217474229,
      "learning_rate": 2.5785732525065623e-06,
      "loss": 0.1983,
      "step": 4977
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9995683366197746,
      "learning_rate": 2.576662868216829e-06,
      "loss": 0.1913,
      "step": 4978
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.942675044402386,
      "learning_rate": 2.5747529461946953e-06,
      "loss": 0.1361,
      "step": 4979
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0678490325454144,
      "learning_rate": 2.5728434868044917e-06,
      "loss": 0.2034,
      "step": 4980
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.855711788036404,
      "learning_rate": 2.5709344904104605e-06,
      "loss": 0.1178,
      "step": 4981
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8802276266636793,
      "learning_rate": 2.569025957376754e-06,
      "loss": 0.1297,
      "step": 4982
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9665482842784512,
      "learning_rate": 2.567117888067439e-06,
      "loss": 0.1771,
      "step": 4983
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8320550189394075,
      "learning_rate": 2.565210282846492e-06,
      "loss": 0.1531,
      "step": 4984
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.967760833103234,
      "learning_rate": 2.5633031420778002e-06,
      "loss": 0.1863,
      "step": 4985
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0895959805449746,
      "learning_rate": 2.5613964661251646e-06,
      "loss": 0.2192,
      "step": 4986
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0655185301352366,
      "learning_rate": 2.5594902553522943e-06,
      "loss": 0.165,
      "step": 4987
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7502761992193296,
      "learning_rate": 2.557584510122815e-06,
      "loss": 0.1461,
      "step": 4988
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0520375055582707,
      "learning_rate": 2.5556792308002583e-06,
      "loss": 0.2106,
      "step": 4989
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5947797109108266,
      "learning_rate": 2.5537744177480706e-06,
      "loss": 0.1071,
      "step": 4990
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9902984039287009,
      "learning_rate": 2.5518700713296043e-06,
      "loss": 0.1804,
      "step": 4991
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6977344637516543,
      "learning_rate": 2.5499661919081277e-06,
      "loss": 0.1192,
      "step": 4992
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8406358734908599,
      "learning_rate": 2.548062779846818e-06,
      "loss": 0.1273,
      "step": 4993
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0328463761284725,
      "learning_rate": 2.546159835508765e-06,
      "loss": 0.1938,
      "step": 4994
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0735827552012194,
      "learning_rate": 2.5442573592569664e-06,
      "loss": 0.2028,
      "step": 4995
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3058994325826796,
      "learning_rate": 2.542355351454333e-06,
      "loss": 0.24,
      "step": 4996
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8918427751451438,
      "learning_rate": 2.540453812463685e-06,
      "loss": 0.1391,
      "step": 4997
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9503791180877594,
      "learning_rate": 2.538552742647752e-06,
      "loss": 0.1535,
      "step": 4998
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0063472982630326,
      "learning_rate": 2.5366521423691793e-06,
      "loss": 0.1726,
      "step": 4999
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1351249585757697,
      "learning_rate": 2.534752011990517e-06,
      "loss": 0.1762,
      "step": 5000
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0802731287496359,
      "learning_rate": 2.532852351874227e-06,
      "loss": 0.1432,
      "step": 5001
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9439284234343366,
      "learning_rate": 2.5309531623826833e-06,
      "loss": 0.1404,
      "step": 5002
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8028148458528475,
      "learning_rate": 2.529054443878168e-06,
      "loss": 0.1589,
      "step": 5003
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5281968179123187,
      "learning_rate": 2.527156196722874e-06,
      "loss": 0.1244,
      "step": 5004
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.169659722048614,
      "learning_rate": 2.5252584212789055e-06,
      "loss": 0.1949,
      "step": 5005
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0478902265960715,
      "learning_rate": 2.5233611179082753e-06,
      "loss": 0.1842,
      "step": 5006
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9589867573984824,
      "learning_rate": 2.5214642869729073e-06,
      "loss": 0.1416,
      "step": 5007
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0676009126138715,
      "learning_rate": 2.5195679288346325e-06,
      "loss": 0.1874,
      "step": 5008
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2304802480481578,
      "learning_rate": 2.517672043855193e-06,
      "loss": 0.1987,
      "step": 5009
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8380452816935902,
      "learning_rate": 2.515776632396245e-06,
      "loss": 0.1742,
      "step": 5010
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1408834865088873,
      "learning_rate": 2.5138816948193494e-06,
      "loss": 0.1813,
      "step": 5011
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7098328813037635,
      "learning_rate": 2.5119872314859777e-06,
      "loss": 0.1192,
      "step": 5012
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1036324943480669,
      "learning_rate": 2.510093242757512e-06,
      "loss": 0.1883,
      "step": 5013
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.128462133429825,
      "learning_rate": 2.508199728995242e-06,
      "loss": 0.2176,
      "step": 5014
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1422569404373972,
      "learning_rate": 2.506306690560368e-06,
      "loss": 0.2189,
      "step": 5015
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0728098403700177,
      "learning_rate": 2.5044141278140014e-06,
      "loss": 0.1842,
      "step": 5016
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7970595932776317,
      "learning_rate": 2.50252204111716e-06,
      "loss": 0.1314,
      "step": 5017
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6157326849927425,
      "learning_rate": 2.500630430830771e-06,
      "loss": 0.1227,
      "step": 5018
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8677606247047926,
      "learning_rate": 2.4987392973156727e-06,
      "loss": 0.1126,
      "step": 5019
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.944757254390419,
      "learning_rate": 2.496848640932611e-06,
      "loss": 0.1893,
      "step": 5020
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9382244652714963,
      "learning_rate": 2.494958462042241e-06,
      "loss": 0.1317,
      "step": 5021
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0756587692628046,
      "learning_rate": 2.4930687610051266e-06,
      "loss": 0.2128,
      "step": 5022
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8470497365216256,
      "learning_rate": 2.4911795381817416e-06,
      "loss": 0.1603,
      "step": 5023
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8475764923468746,
      "learning_rate": 2.4892907939324672e-06,
      "loss": 0.1472,
      "step": 5024
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7460335721487082,
      "learning_rate": 2.487402528617594e-06,
      "loss": 0.1236,
      "step": 5025
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0130219775169587,
      "learning_rate": 2.4855147425973214e-06,
      "loss": 0.1671,
      "step": 5026
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.958345586285766,
      "learning_rate": 2.4836274362317565e-06,
      "loss": 0.1585,
      "step": 5027
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1400144573458444,
      "learning_rate": 2.4817406098809157e-06,
      "loss": 0.1961,
      "step": 5028
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9525985072456087,
      "learning_rate": 2.4798542639047235e-06,
      "loss": 0.1885,
      "step": 5029
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1542850095548496,
      "learning_rate": 2.4779683986630127e-06,
      "loss": 0.2116,
      "step": 5030
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7334920669392312,
      "learning_rate": 2.4760830145155257e-06,
      "loss": 0.1197,
      "step": 5031
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.957624873994848,
      "learning_rate": 2.4741981118219087e-06,
      "loss": 0.1472,
      "step": 5032
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0872731762885373,
      "learning_rate": 2.472313690941724e-06,
      "loss": 0.1945,
      "step": 5033
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6648534687223439,
      "learning_rate": 2.470429752234435e-06,
      "loss": 0.0758,
      "step": 5034
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6854211110486883,
      "learning_rate": 2.468546296059416e-06,
      "loss": 0.0992,
      "step": 5035
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.113776009658678,
      "learning_rate": 2.4666633227759505e-06,
      "loss": 0.2038,
      "step": 5036
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0338249463492504,
      "learning_rate": 2.4647808327432236e-06,
      "loss": 0.1857,
      "step": 5037
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0257189585816513,
      "learning_rate": 2.4628988263203355e-06,
      "loss": 0.2006,
      "step": 5038
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7666094338756401,
      "learning_rate": 2.4610173038662915e-06,
      "loss": 0.1049,
      "step": 5039
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9324259649161197,
      "learning_rate": 2.4591362657400036e-06,
      "loss": 0.1477,
      "step": 5040
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8034754333922486,
      "learning_rate": 2.4572557123002927e-06,
      "loss": 0.1381,
      "step": 5041
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5314159206350622,
      "learning_rate": 2.4553756439058874e-06,
      "loss": 0.0862,
      "step": 5042
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.245609959179011,
      "learning_rate": 2.4534960609154203e-06,
      "loss": 0.1989,
      "step": 5043
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1235561243795296,
      "learning_rate": 2.451616963687438e-06,
      "loss": 0.1475,
      "step": 5044
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9337526265132309,
      "learning_rate": 2.44973835258039e-06,
      "loss": 0.1726,
      "step": 5045
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8488738986074333,
      "learning_rate": 2.447860227952633e-06,
      "loss": 0.1577,
      "step": 5046
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9864744172281046,
      "learning_rate": 2.4459825901624322e-06,
      "loss": 0.1684,
      "step": 5047
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8441573131182548,
      "learning_rate": 2.444105439567959e-06,
      "loss": 0.1527,
      "step": 5048
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9585103022467782,
      "learning_rate": 2.442228776527292e-06,
      "loss": 0.1503,
      "step": 5049
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7500707487113747,
      "learning_rate": 2.440352601398418e-06,
      "loss": 0.1569,
      "step": 5050
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0539816389128671,
      "learning_rate": 2.438476914539229e-06,
      "loss": 0.1937,
      "step": 5051
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7516059262346256,
      "learning_rate": 2.436601716307525e-06,
      "loss": 0.123,
      "step": 5052
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0459280020756099,
      "learning_rate": 2.434727007061012e-06,
      "loss": 0.1714,
      "step": 5053
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8251990321510023,
      "learning_rate": 2.432852787157305e-06,
      "loss": 0.1439,
      "step": 5054
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1241683765097508,
      "learning_rate": 2.430979056953922e-06,
      "loss": 0.1948,
      "step": 5055
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1054064016672984,
      "learning_rate": 2.4291058168082898e-06,
      "loss": 0.2003,
      "step": 5056
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.02437708225197,
      "learning_rate": 2.4272330670777415e-06,
      "loss": 0.1817,
      "step": 5057
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7913649350936726,
      "learning_rate": 2.4253608081195166e-06,
      "loss": 0.1072,
      "step": 5058
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8043152485015671,
      "learning_rate": 2.4234890402907612e-06,
      "loss": 0.1307,
      "step": 5059
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1672378412474331,
      "learning_rate": 2.4216177639485265e-06,
      "loss": 0.2007,
      "step": 5060
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6863391410575154,
      "learning_rate": 2.4197469794497717e-06,
      "loss": 0.0844,
      "step": 5061
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2061577466716975,
      "learning_rate": 2.417876687151361e-06,
      "loss": 0.1776,
      "step": 5062
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9522119257465232,
      "learning_rate": 2.416006887410065e-06,
      "loss": 0.1797,
      "step": 5063
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0740371359176373,
      "learning_rate": 2.4141375805825607e-06,
      "loss": 0.1731,
      "step": 5064
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1729497481940145,
      "learning_rate": 2.41226876702543e-06,
      "loss": 0.1946,
      "step": 5065
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1390606680093514,
      "learning_rate": 2.4104004470951623e-06,
      "loss": 0.1637,
      "step": 5066
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0118070548094733,
      "learning_rate": 2.408532621148152e-06,
      "loss": 0.1548,
      "step": 5067
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8710333982898321,
      "learning_rate": 2.4066652895406983e-06,
      "loss": 0.1437,
      "step": 5068
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2274925460234825,
      "learning_rate": 2.404798452629008e-06,
      "loss": 0.2016,
      "step": 5069
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9341683472655709,
      "learning_rate": 2.402932110769192e-06,
      "loss": 0.1582,
      "step": 5070
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.944447910789672,
      "learning_rate": 2.401066264317268e-06,
      "loss": 0.1704,
      "step": 5071
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0436935554043743,
      "learning_rate": 2.399200913629158e-06,
      "loss": 0.1886,
      "step": 5072
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0749504191875847,
      "learning_rate": 2.3973360590606903e-06,
      "loss": 0.1743,
      "step": 5073
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8572239529721192,
      "learning_rate": 2.3954717009675984e-06,
      "loss": 0.1775,
      "step": 5074
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0766307924036584,
      "learning_rate": 2.39360783970552e-06,
      "loss": 0.1828,
      "step": 5075
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0078245099656848,
      "learning_rate": 2.3917444756300002e-06,
      "loss": 0.1658,
      "step": 5076
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.727297567207007,
      "learning_rate": 2.389881609096485e-06,
      "loss": 0.1136,
      "step": 5077
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8621710283224906,
      "learning_rate": 2.3880192404603338e-06,
      "loss": 0.1375,
      "step": 5078
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0328353484135506,
      "learning_rate": 2.386157370076802e-06,
      "loss": 0.1685,
      "step": 5079
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8750360978279925,
      "learning_rate": 2.3842959983010557e-06,
      "loss": 0.1646,
      "step": 5080
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0475545424155337,
      "learning_rate": 2.3824351254881617e-06,
      "loss": 0.172,
      "step": 5081
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0988304466283993,
      "learning_rate": 2.3805747519930954e-06,
      "loss": 0.2305,
      "step": 5082
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1981055221316899,
      "learning_rate": 2.3787148781707374e-06,
      "loss": 0.2172,
      "step": 5083
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1533842644206684,
      "learning_rate": 2.3768555043758662e-06,
      "loss": 0.2248,
      "step": 5084
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9494068398462501,
      "learning_rate": 2.3749966309631717e-06,
      "loss": 0.1383,
      "step": 5085
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7900901271151252,
      "learning_rate": 2.373138258287247e-06,
      "loss": 0.1725,
      "step": 5086
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9320834329748456,
      "learning_rate": 2.371280386702588e-06,
      "loss": 0.1947,
      "step": 5087
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9406886576628941,
      "learning_rate": 2.369423016563595e-06,
      "loss": 0.174,
      "step": 5088
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8778789621198979,
      "learning_rate": 2.3675661482245773e-06,
      "loss": 0.1933,
      "step": 5089
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7831283934332902,
      "learning_rate": 2.3657097820397417e-06,
      "loss": 0.1302,
      "step": 5090
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.922131107060234,
      "learning_rate": 2.363853918363204e-06,
      "loss": 0.1378,
      "step": 5091
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7248989588391285,
      "learning_rate": 2.361998557548982e-06,
      "loss": 0.1274,
      "step": 5092
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7529272475357218,
      "learning_rate": 2.360143699950998e-06,
      "loss": 0.1358,
      "step": 5093
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0069107050970092,
      "learning_rate": 2.3582893459230776e-06,
      "loss": 0.1777,
      "step": 5094
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9823774865426554,
      "learning_rate": 2.3564354958189522e-06,
      "loss": 0.1479,
      "step": 5095
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7986351648939432,
      "learning_rate": 2.3545821499922557e-06,
      "loss": 0.1594,
      "step": 5096
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9582672122822573,
      "learning_rate": 2.352729308796526e-06,
      "loss": 0.1795,
      "step": 5097
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8787777407716979,
      "learning_rate": 2.3508769725852038e-06,
      "loss": 0.1469,
      "step": 5098
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9471124745118521,
      "learning_rate": 2.349025141711636e-06,
      "loss": 0.1533,
      "step": 5099
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5429932397042518,
      "learning_rate": 2.3471738165290712e-06,
      "loss": 0.1189,
      "step": 5100
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8266176305934444,
      "learning_rate": 2.3453229973906614e-06,
      "loss": 0.1669,
      "step": 5101
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9925605371962564,
      "learning_rate": 2.343472684649464e-06,
      "loss": 0.1771,
      "step": 5102
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.066525959093927,
      "learning_rate": 2.341622878658436e-06,
      "loss": 0.1864,
      "step": 5103
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1210195352494572,
      "learning_rate": 2.3397735797704425e-06,
      "loss": 0.2111,
      "step": 5104
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7813281589620362,
      "learning_rate": 2.3379247883382476e-06,
      "loss": 0.1306,
      "step": 5105
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2977392793612965,
      "learning_rate": 2.336076504714522e-06,
      "loss": 0.238,
      "step": 5106
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.116612436157339,
      "learning_rate": 2.3342287292518377e-06,
      "loss": 0.1934,
      "step": 5107
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0947873966125956,
      "learning_rate": 2.3323814623026696e-06,
      "loss": 0.1422,
      "step": 5108
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.112795829058827,
      "learning_rate": 2.330534704219396e-06,
      "loss": 0.1965,
      "step": 5109
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9438473768329463,
      "learning_rate": 2.328688455354297e-06,
      "loss": 0.1632,
      "step": 5110
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.016264107438513,
      "learning_rate": 2.3268427160595602e-06,
      "loss": 0.1713,
      "step": 5111
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.960341462021817,
      "learning_rate": 2.324997486687271e-06,
      "loss": 0.2051,
      "step": 5112
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0240516053417597,
      "learning_rate": 2.3231527675894205e-06,
      "loss": 0.2039,
      "step": 5113
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8157893026883156,
      "learning_rate": 2.321308559117898e-06,
      "loss": 0.1236,
      "step": 5114
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1264107303909108,
      "learning_rate": 2.3194648616244995e-06,
      "loss": 0.2153,
      "step": 5115
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3270381281993306,
      "learning_rate": 2.317621675460923e-06,
      "loss": 0.1993,
      "step": 5116
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0550879244316786,
      "learning_rate": 2.3157790009787686e-06,
      "loss": 0.1809,
      "step": 5117
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8956920463995288,
      "learning_rate": 2.313936838529538e-06,
      "loss": 0.1281,
      "step": 5118
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0533202113164644,
      "learning_rate": 2.312095188464637e-06,
      "loss": 0.161,
      "step": 5119
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8181804232562605,
      "learning_rate": 2.3102540511353715e-06,
      "loss": 0.147,
      "step": 5120
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0293074893066025,
      "learning_rate": 2.308413426892951e-06,
      "loss": 0.2033,
      "step": 5121
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9117239520495174,
      "learning_rate": 2.3065733160884847e-06,
      "loss": 0.1414,
      "step": 5122
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8623257118781961,
      "learning_rate": 2.3047337190729903e-06,
      "loss": 0.1352,
      "step": 5123
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9716478673365669,
      "learning_rate": 2.3028946361973803e-06,
      "loss": 0.1646,
      "step": 5124
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0648760026144501,
      "learning_rate": 2.3010560678124726e-06,
      "loss": 0.1888,
      "step": 5125
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1313076324853084,
      "learning_rate": 2.299218014268987e-06,
      "loss": 0.2101,
      "step": 5126
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9580658071158289,
      "learning_rate": 2.2973804759175423e-06,
      "loss": 0.1458,
      "step": 5127
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9532111303456124,
      "learning_rate": 2.2955434531086627e-06,
      "loss": 0.1652,
      "step": 5128
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8970308706461048,
      "learning_rate": 2.2937069461927726e-06,
      "loss": 0.1308,
      "step": 5129
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9917669902260314,
      "learning_rate": 2.2918709555201983e-06,
      "loss": 0.1975,
      "step": 5130
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.988467556711384,
      "learning_rate": 2.2900354814411648e-06,
      "loss": 0.1574,
      "step": 5131
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1071305931698558,
      "learning_rate": 2.288200524305802e-06,
      "loss": 0.1762,
      "step": 5132
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9000778413542483,
      "learning_rate": 2.286366084464139e-06,
      "loss": 0.1156,
      "step": 5133
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1825181549487118,
      "learning_rate": 2.2845321622661105e-06,
      "loss": 0.152,
      "step": 5134
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.002968732943295,
      "learning_rate": 2.2826987580615472e-06,
      "loss": 0.1534,
      "step": 5135
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9256462105237258,
      "learning_rate": 2.2808658722001843e-06,
      "loss": 0.1603,
      "step": 5136
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6438485621906281,
      "learning_rate": 2.2790335050316553e-06,
      "loss": 0.0969,
      "step": 5137
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.080375375929712,
      "learning_rate": 2.2772016569054978e-06,
      "loss": 0.1707,
      "step": 5138
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8881551425631201,
      "learning_rate": 2.2753703281711475e-06,
      "loss": 0.1649,
      "step": 5139
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.900662474429067,
      "learning_rate": 2.273539519177944e-06,
      "loss": 0.1746,
      "step": 5140
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7530601353410219,
      "learning_rate": 2.2717092302751252e-06,
      "loss": 0.1304,
      "step": 5141
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1665486209602873,
      "learning_rate": 2.2698794618118315e-06,
      "loss": 0.1836,
      "step": 5142
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5819622603185716,
      "learning_rate": 2.2680502141371024e-06,
      "loss": 0.1387,
      "step": 5143
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1375526963458324,
      "learning_rate": 2.26622148759988e-06,
      "loss": 0.1794,
      "step": 5144
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9564087941045062,
      "learning_rate": 2.2643932825490055e-06,
      "loss": 0.1612,
      "step": 5145
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0146145917495004,
      "learning_rate": 2.2625655993332215e-06,
      "loss": 0.1486,
      "step": 5146
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.240640982924503,
      "learning_rate": 2.26073843830117e-06,
      "loss": 0.1951,
      "step": 5147
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8482580719019635,
      "learning_rate": 2.258911799801394e-06,
      "loss": 0.2029,
      "step": 5148
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8461575199178509,
      "learning_rate": 2.2570856841823384e-06,
      "loss": 0.1369,
      "step": 5149
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0962315198062242,
      "learning_rate": 2.255260091792345e-06,
      "loss": 0.2003,
      "step": 5150
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9390515766605418,
      "learning_rate": 2.2534350229796585e-06,
      "loss": 0.1154,
      "step": 5151
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0075079836268932,
      "learning_rate": 2.251610478092423e-06,
      "loss": 0.186,
      "step": 5152
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7114508789236266,
      "learning_rate": 2.249786457478682e-06,
      "loss": 0.102,
      "step": 5153
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7184812877238694,
      "learning_rate": 2.24796296148638e-06,
      "loss": 0.1522,
      "step": 5154
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0110855473720566,
      "learning_rate": 2.246139990463359e-06,
      "loss": 0.1712,
      "step": 5155
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7878846590671814,
      "learning_rate": 2.244317544757366e-06,
      "loss": 0.1295,
      "step": 5156
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9965085160043814,
      "learning_rate": 2.2424956247160435e-06,
      "loss": 0.1673,
      "step": 5157
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8605602746376789,
      "learning_rate": 2.240674230686934e-06,
      "loss": 0.1843,
      "step": 5158
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.872362048465069,
      "learning_rate": 2.238853363017481e-06,
      "loss": 0.1384,
      "step": 5159
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8188834087233577,
      "learning_rate": 2.2370330220550284e-06,
      "loss": 0.1495,
      "step": 5160
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8002486175780793,
      "learning_rate": 2.235213208146816e-06,
      "loss": 0.1509,
      "step": 5161
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9021615217937132,
      "learning_rate": 2.2333939216399854e-06,
      "loss": 0.1504,
      "step": 5162
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9818770379348042,
      "learning_rate": 2.2315751628815778e-06,
      "loss": 0.2064,
      "step": 5163
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8759056678475724,
      "learning_rate": 2.2297569322185348e-06,
      "loss": 0.1468,
      "step": 5164
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8957949581590213,
      "learning_rate": 2.227939229997695e-06,
      "loss": 0.1453,
      "step": 5165
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6427456929822455,
      "learning_rate": 2.226122056565797e-06,
      "loss": 0.2783,
      "step": 5166
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0141908655576757,
      "learning_rate": 2.2243054122694774e-06,
      "loss": 0.1587,
      "step": 5167
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.812522632106931,
      "learning_rate": 2.222489297455276e-06,
      "loss": 0.1398,
      "step": 5168
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0106492432016099,
      "learning_rate": 2.220673712469628e-06,
      "loss": 0.1785,
      "step": 5169
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1855843959977845,
      "learning_rate": 2.2188586576588666e-06,
      "loss": 0.2148,
      "step": 5170
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.995596222855162,
      "learning_rate": 2.217044133369227e-06,
      "loss": 0.1746,
      "step": 5171
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.933059021988597,
      "learning_rate": 2.215230139946842e-06,
      "loss": 0.1294,
      "step": 5172
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0483737508295283,
      "learning_rate": 2.213416677737741e-06,
      "loss": 0.1903,
      "step": 5173
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9237163894722823,
      "learning_rate": 2.2116037470878554e-06,
      "loss": 0.1625,
      "step": 5174
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49015472028336493,
      "learning_rate": 2.209791348343013e-06,
      "loss": 0.102,
      "step": 5175
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8674382980299368,
      "learning_rate": 2.207979481848942e-06,
      "loss": 0.1163,
      "step": 5176
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1727263321915589,
      "learning_rate": 2.2061681479512687e-06,
      "loss": 0.2241,
      "step": 5177
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8988021870393325,
      "learning_rate": 2.2043573469955116e-06,
      "loss": 0.1317,
      "step": 5178
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.107406486485925,
      "learning_rate": 2.2025470793270986e-06,
      "loss": 0.1751,
      "step": 5179
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9003429385001971,
      "learning_rate": 2.200737345291349e-06,
      "loss": 0.1292,
      "step": 5180
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6569041920213644,
      "learning_rate": 2.1989281452334806e-06,
      "loss": 0.0803,
      "step": 5181
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9257221441552794,
      "learning_rate": 2.1971194794986107e-06,
      "loss": 0.1639,
      "step": 5182
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8386203024312245,
      "learning_rate": 2.1953113484317543e-06,
      "loss": 0.1477,
      "step": 5183
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9114815409037735,
      "learning_rate": 2.193503752377824e-06,
      "loss": 0.149,
      "step": 5184
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0213664731821395,
      "learning_rate": 2.1916966916816316e-06,
      "loss": 0.1621,
      "step": 5185
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1952351703344137,
      "learning_rate": 2.1898901666878853e-06,
      "loss": 0.2326,
      "step": 5186
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7851603745791829,
      "learning_rate": 2.1880841777411915e-06,
      "loss": 0.1124,
      "step": 5187
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.892088859988036,
      "learning_rate": 2.186278725186055e-06,
      "loss": 0.1625,
      "step": 5188
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9854940778802016,
      "learning_rate": 2.1844738093668777e-06,
      "loss": 0.1988,
      "step": 5189
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8632707079473064,
      "learning_rate": 2.182669430627959e-06,
      "loss": 0.1357,
      "step": 5190
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0562654812654555,
      "learning_rate": 2.1808655893134955e-06,
      "loss": 0.1804,
      "step": 5191
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2198153096477928,
      "learning_rate": 2.1790622857675837e-06,
      "loss": 0.2426,
      "step": 5192
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0637791280134588,
      "learning_rate": 2.177259520334214e-06,
      "loss": 0.1929,
      "step": 5193
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.60654151394383,
      "learning_rate": 2.175457293357277e-06,
      "loss": 0.1153,
      "step": 5194
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0569009990405513,
      "learning_rate": 2.1736556051805585e-06,
      "loss": 0.1672,
      "step": 5195
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0170889472585958,
      "learning_rate": 2.1718544561477427e-06,
      "loss": 0.1765,
      "step": 5196
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0131191553742562,
      "learning_rate": 2.1700538466024105e-06,
      "loss": 0.1838,
      "step": 5197
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1632332231193023,
      "learning_rate": 2.1682537768880413e-06,
      "loss": 0.209,
      "step": 5198
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0293797206695703,
      "learning_rate": 2.1664542473480086e-06,
      "loss": 0.1605,
      "step": 5199
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9099613619784896,
      "learning_rate": 2.164655258325584e-06,
      "loss": 0.1999,
      "step": 5200
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8976723182341697,
      "learning_rate": 2.16285681016394e-06,
      "loss": 0.1508,
      "step": 5201
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.265683348626232,
      "learning_rate": 2.1610589032061403e-06,
      "loss": 0.2422,
      "step": 5202
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8295317581508024,
      "learning_rate": 2.1592615377951475e-06,
      "loss": 0.1504,
      "step": 5203
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7809470254550457,
      "learning_rate": 2.1574647142738214e-06,
      "loss": 0.1683,
      "step": 5204
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1309135896854745,
      "learning_rate": 2.1556684329849174e-06,
      "loss": 0.2059,
      "step": 5205
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0342725243436408,
      "learning_rate": 2.1538726942710904e-06,
      "loss": 0.1643,
      "step": 5206
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9081723049490497,
      "learning_rate": 2.1520774984748856e-06,
      "loss": 0.1534,
      "step": 5207
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9882693223839399,
      "learning_rate": 2.1502828459387504e-06,
      "loss": 0.1435,
      "step": 5208
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8511187749380588,
      "learning_rate": 2.1484887370050263e-06,
      "loss": 0.1419,
      "step": 5209
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9831264283688805,
      "learning_rate": 2.1466951720159517e-06,
      "loss": 0.1706,
      "step": 5210
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1530945443461806,
      "learning_rate": 2.1449021513136595e-06,
      "loss": 0.1926,
      "step": 5211
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.610726711005002,
      "learning_rate": 2.1431096752401837e-06,
      "loss": 0.1059,
      "step": 5212
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0398826691506844,
      "learning_rate": 2.141317744137449e-06,
      "loss": 0.1815,
      "step": 5213
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8330419687361401,
      "learning_rate": 2.1395263583472777e-06,
      "loss": 0.1375,
      "step": 5214
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0010854260808537,
      "learning_rate": 2.1377355182113895e-06,
      "loss": 0.1867,
      "step": 5215
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.510746136928644,
      "learning_rate": 2.135945224071398e-06,
      "loss": 0.0592,
      "step": 5216
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0883541881050738,
      "learning_rate": 2.1341554762688145e-06,
      "loss": 0.1675,
      "step": 5217
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6919849382440006,
      "learning_rate": 2.1323662751450445e-06,
      "loss": 0.092,
      "step": 5218
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9951065894943325,
      "learning_rate": 2.1305776210413907e-06,
      "loss": 0.1665,
      "step": 5219
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7701168540477913,
      "learning_rate": 2.1287895142990506e-06,
      "loss": 0.1315,
      "step": 5220
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.101413132589804,
      "learning_rate": 2.127001955259117e-06,
      "loss": 0.2014,
      "step": 5221
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.03601286170246,
      "learning_rate": 2.1252149442625785e-06,
      "loss": 0.1842,
      "step": 5222
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0530853431504639,
      "learning_rate": 2.12342848165032e-06,
      "loss": 0.1832,
      "step": 5223
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8977572010305682,
      "learning_rate": 2.1216425677631196e-06,
      "loss": 0.1189,
      "step": 5224
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9782610301869513,
      "learning_rate": 2.1198572029416544e-06,
      "loss": 0.1961,
      "step": 5225
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.043529878859056,
      "learning_rate": 2.1180723875264932e-06,
      "loss": 0.1848,
      "step": 5226
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1013732971515666,
      "learning_rate": 2.1162881218581007e-06,
      "loss": 0.195,
      "step": 5227
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8941496504241335,
      "learning_rate": 2.1145044062768387e-06,
      "loss": 0.1863,
      "step": 5228
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7576476698147447,
      "learning_rate": 2.1127212411229625e-06,
      "loss": 0.1062,
      "step": 5229
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9466267988173712,
      "learning_rate": 2.1109386267366223e-06,
      "loss": 0.1403,
      "step": 5230
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0293238869550285,
      "learning_rate": 2.109156563457864e-06,
      "loss": 0.1663,
      "step": 5231
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1439182190285688,
      "learning_rate": 2.1073750516266267e-06,
      "loss": 0.2378,
      "step": 5232
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8217742111350279,
      "learning_rate": 2.1055940915827473e-06,
      "loss": 0.1356,
      "step": 5233
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9489767098395485,
      "learning_rate": 2.103813683665953e-06,
      "loss": 0.1629,
      "step": 5234
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9767184461498047,
      "learning_rate": 2.1020338282158725e-06,
      "loss": 0.1738,
      "step": 5235
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6310124445555436,
      "learning_rate": 2.100254525572024e-06,
      "loss": 0.1048,
      "step": 5236
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8655798330292693,
      "learning_rate": 2.098475776073819e-06,
      "loss": 0.1592,
      "step": 5237
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.146051526110434,
      "learning_rate": 2.0966975800605665e-06,
      "loss": 0.1851,
      "step": 5238
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8428867258701347,
      "learning_rate": 2.0949199378714697e-06,
      "loss": 0.1515,
      "step": 5239
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7963409783993155,
      "learning_rate": 2.093142849845626e-06,
      "loss": 0.139,
      "step": 5240
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0955848527579124,
      "learning_rate": 2.0913663163220256e-06,
      "loss": 0.1734,
      "step": 5241
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8720216859568629,
      "learning_rate": 2.089590337639554e-06,
      "loss": 0.147,
      "step": 5242
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1346446940221524,
      "learning_rate": 2.0878149141369926e-06,
      "loss": 0.1515,
      "step": 5243
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8168763246855515,
      "learning_rate": 2.0860400461530133e-06,
      "loss": 0.1248,
      "step": 5244
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7763956036236361,
      "learning_rate": 2.0842657340261836e-06,
      "loss": 0.1509,
      "step": 5245
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9702166752976114,
      "learning_rate": 2.0824919780949675e-06,
      "loss": 0.1852,
      "step": 5246
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1739192904037745,
      "learning_rate": 2.0807187786977196e-06,
      "loss": 0.1976,
      "step": 5247
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0369331175779324,
      "learning_rate": 2.07894613617269e-06,
      "loss": 0.1765,
      "step": 5248
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.086548562229614,
      "learning_rate": 2.0771740508580205e-06,
      "loss": 0.2202,
      "step": 5249
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0117202254394895,
      "learning_rate": 2.0754025230917498e-06,
      "loss": 0.1691,
      "step": 5250
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1970278177469342,
      "learning_rate": 2.0736315532118073e-06,
      "loss": 0.184,
      "step": 5251
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8920700214670607,
      "learning_rate": 2.071861141556018e-06,
      "loss": 0.1983,
      "step": 5252
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0736438891447917,
      "learning_rate": 2.0700912884621006e-06,
      "loss": 0.2071,
      "step": 5253
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8167570164894775,
      "learning_rate": 2.0683219942676634e-06,
      "loss": 0.1697,
      "step": 5254
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9740113411298168,
      "learning_rate": 2.066553259310213e-06,
      "loss": 0.1608,
      "step": 5255
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7849418991734501,
      "learning_rate": 2.0647850839271444e-06,
      "loss": 0.1136,
      "step": 5256
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0352898796803098,
      "learning_rate": 2.063017468455753e-06,
      "loss": 0.1747,
      "step": 5257
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9077111918012932,
      "learning_rate": 2.0612504132332217e-06,
      "loss": 0.1709,
      "step": 5258
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1638459548831896,
      "learning_rate": 2.0594839185966282e-06,
      "loss": 0.1935,
      "step": 5259
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7328953261542221,
      "learning_rate": 2.0577179848829415e-06,
      "loss": 0.12,
      "step": 5260
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7841789009596776,
      "learning_rate": 2.0559526124290268e-06,
      "loss": 0.1536,
      "step": 5261
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9815132744598671,
      "learning_rate": 2.05418780157164e-06,
      "loss": 0.1588,
      "step": 5262
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8221090142948434,
      "learning_rate": 2.05242355264743e-06,
      "loss": 0.1385,
      "step": 5263
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8618266127016659,
      "learning_rate": 2.0506598659929396e-06,
      "loss": 0.1764,
      "step": 5264
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2500024971100976,
      "learning_rate": 2.048896741944603e-06,
      "loss": 0.2384,
      "step": 5265
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0760427993074166,
      "learning_rate": 2.047134180838749e-06,
      "loss": 0.2252,
      "step": 5266
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9023281874688119,
      "learning_rate": 2.045372183011597e-06,
      "loss": 0.1738,
      "step": 5267
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8243776562930734,
      "learning_rate": 2.043610748799259e-06,
      "loss": 0.1476,
      "step": 5268
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8555697313986249,
      "learning_rate": 2.041849878537742e-06,
      "loss": 0.1263,
      "step": 5269
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9700938273876363,
      "learning_rate": 2.0400895725629415e-06,
      "loss": 0.1611,
      "step": 5270
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8002054841546978,
      "learning_rate": 2.03832983121065e-06,
      "loss": 0.1377,
      "step": 5271
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2803784775704683,
      "learning_rate": 2.0365706548165476e-06,
      "loss": 0.2131,
      "step": 5272
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8351147356682314,
      "learning_rate": 2.03481204371621e-06,
      "loss": 0.1628,
      "step": 5273
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9123391068617992,
      "learning_rate": 2.0330539982451037e-06,
      "loss": 0.1562,
      "step": 5274
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8190161243674605,
      "learning_rate": 2.0312965187385875e-06,
      "loss": 0.1471,
      "step": 5275
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8425671254140754,
      "learning_rate": 2.0295396055319123e-06,
      "loss": 0.1658,
      "step": 5276
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0595181298958565,
      "learning_rate": 2.027783258960221e-06,
      "loss": 0.203,
      "step": 5277
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1268614820310185,
      "learning_rate": 2.0260274793585478e-06,
      "loss": 0.1736,
      "step": 5278
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9203087905955695,
      "learning_rate": 2.0242722670618183e-06,
      "loss": 0.1452,
      "step": 5279
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0834300001714667,
      "learning_rate": 2.0225176224048537e-06,
      "loss": 0.1874,
      "step": 5280
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7559041610638716,
      "learning_rate": 2.0207635457223624e-06,
      "loss": 0.1226,
      "step": 5281
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0256709212544945,
      "learning_rate": 2.019010037348947e-06,
      "loss": 0.1626,
      "step": 5282
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0349278421051835,
      "learning_rate": 2.0172570976191013e-06,
      "loss": 0.2077,
      "step": 5283
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7202276765622286,
      "learning_rate": 2.015504726867208e-06,
      "loss": 0.1162,
      "step": 5284
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9846218411458137,
      "learning_rate": 2.0137529254275436e-06,
      "loss": 0.1654,
      "step": 5285
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9379416350026927,
      "learning_rate": 2.0120016936342774e-06,
      "loss": 0.1413,
      "step": 5286
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6919445113794854,
      "learning_rate": 2.0102510318214673e-06,
      "loss": 0.1255,
      "step": 5287
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.048181465115288,
      "learning_rate": 2.0085009403230642e-06,
      "loss": 0.1581,
      "step": 5288
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0325553221663513,
      "learning_rate": 2.006751419472909e-06,
      "loss": 0.1733,
      "step": 5289
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.113486861472338,
      "learning_rate": 2.0050024696047343e-06,
      "loss": 0.2009,
      "step": 5290
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8339606364351033,
      "learning_rate": 2.0032540910521653e-06,
      "loss": 0.147,
      "step": 5291
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9504036200905683,
      "learning_rate": 2.0015062841487167e-06,
      "loss": 0.1327,
      "step": 5292
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8208450201860303,
      "learning_rate": 1.999759049227793e-06,
      "loss": 0.1419,
      "step": 5293
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5489374468204958,
      "learning_rate": 1.998012386622691e-06,
      "loss": 0.0997,
      "step": 5294
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9579334410818509,
      "learning_rate": 1.9962662966665992e-06,
      "loss": 0.1483,
      "step": 5295
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9788628378500015,
      "learning_rate": 1.994520779692594e-06,
      "loss": 0.1504,
      "step": 5296
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9919918870701304,
      "learning_rate": 1.9927758360336465e-06,
      "loss": 0.2337,
      "step": 5297
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9919868064119409,
      "learning_rate": 1.991031466022615e-06,
      "loss": 0.2223,
      "step": 5298
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0236799393365767,
      "learning_rate": 1.9892876699922493e-06,
      "loss": 0.1608,
      "step": 5299
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.733208717862599,
      "learning_rate": 1.987544448275192e-06,
      "loss": 0.1105,
      "step": 5300
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9835954782308561,
      "learning_rate": 1.98580180120397e-06,
      "loss": 0.1801,
      "step": 5301
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9339238707063232,
      "learning_rate": 1.9840597291110084e-06,
      "loss": 0.1957,
      "step": 5302
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9647266537162923,
      "learning_rate": 1.982318232328619e-06,
      "loss": 0.1481,
      "step": 5303
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8344959280350065,
      "learning_rate": 1.980577311189002e-06,
      "loss": 0.1425,
      "step": 5304
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1421081454938207,
      "learning_rate": 1.978836966024251e-06,
      "loss": 0.1501,
      "step": 5305
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8491257477772186,
      "learning_rate": 1.977097197166348e-06,
      "loss": 0.1649,
      "step": 5306
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8542633315063157,
      "learning_rate": 1.975358004947165e-06,
      "loss": 0.1804,
      "step": 5307
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1565498750848222,
      "learning_rate": 1.9736193896984646e-06,
      "loss": 0.18,
      "step": 5308
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9684978311827142,
      "learning_rate": 1.9718813517518996e-06,
      "loss": 0.1353,
      "step": 5309
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7253621200522771,
      "learning_rate": 1.970143891439012e-06,
      "loss": 0.0833,
      "step": 5310
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0721993558509795,
      "learning_rate": 1.9684070090912335e-06,
      "loss": 0.1692,
      "step": 5311
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.010311528627763,
      "learning_rate": 1.966670705039886e-06,
      "loss": 0.1688,
      "step": 5312
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9648709116292027,
      "learning_rate": 1.9649349796161817e-06,
      "loss": 0.168,
      "step": 5313
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8838888316031701,
      "learning_rate": 1.963199833151221e-06,
      "loss": 0.129,
      "step": 5314
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6782709416999746,
      "learning_rate": 1.961465265975995e-06,
      "loss": 0.1176,
      "step": 5315
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9546968332019982,
      "learning_rate": 1.9597312784213834e-06,
      "loss": 0.17,
      "step": 5316
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8749465878586437,
      "learning_rate": 1.9579978708181557e-06,
      "loss": 0.1629,
      "step": 5317
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9598452259419247,
      "learning_rate": 1.956265043496972e-06,
      "loss": 0.1441,
      "step": 5318
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7127740613850134,
      "learning_rate": 1.9545327967883794e-06,
      "loss": 0.1269,
      "step": 5319
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8162826308357644,
      "learning_rate": 1.952801131022816e-06,
      "loss": 0.132,
      "step": 5320
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8509184968916198,
      "learning_rate": 1.9510700465306084e-06,
      "loss": 0.1191,
      "step": 5321
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0181565378631932,
      "learning_rate": 1.949339543641972e-06,
      "loss": 0.1278,
      "step": 5322
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8756280587819664,
      "learning_rate": 1.947609622687012e-06,
      "loss": 0.1603,
      "step": 5323
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1701573719205893,
      "learning_rate": 1.9458802839957207e-06,
      "loss": 0.1937,
      "step": 5324
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8362396992818782,
      "learning_rate": 1.944151527897985e-06,
      "loss": 0.1748,
      "step": 5325
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0354479131670675,
      "learning_rate": 1.9424233547235733e-06,
      "loss": 0.2147,
      "step": 5326
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0693667739764026,
      "learning_rate": 1.940695764802147e-06,
      "loss": 0.1715,
      "step": 5327
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.129490777760813,
      "learning_rate": 1.9389687584632554e-06,
      "loss": 0.2406,
      "step": 5328
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7630987507169708,
      "learning_rate": 1.937242336036336e-06,
      "loss": 0.1314,
      "step": 5329
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0023161619976755,
      "learning_rate": 1.9355164978507175e-06,
      "loss": 0.1931,
      "step": 5330
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9671511970772629,
      "learning_rate": 1.933791244235611e-06,
      "loss": 0.1781,
      "step": 5331
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8151386349864533,
      "learning_rate": 1.932066575520122e-06,
      "loss": 0.1644,
      "step": 5332
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0198801931239905,
      "learning_rate": 1.9303424920332426e-06,
      "loss": 0.2088,
      "step": 5333
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9868815360822494,
      "learning_rate": 1.928618994103853e-06,
      "loss": 0.1844,
      "step": 5334
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8527657503713765,
      "learning_rate": 1.9268960820607197e-06,
      "loss": 0.154,
      "step": 5335
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9718021505644555,
      "learning_rate": 1.925173756232504e-06,
      "loss": 0.1699,
      "step": 5336
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9130805863250179,
      "learning_rate": 1.9234520169477484e-06,
      "loss": 0.141,
      "step": 5337
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0228060442827374,
      "learning_rate": 1.9217308645348864e-06,
      "loss": 0.1537,
      "step": 5338
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9254347278192775,
      "learning_rate": 1.9200102993222393e-06,
      "loss": 0.1711,
      "step": 5339
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.949552748696963,
      "learning_rate": 1.9182903216380156e-06,
      "loss": 0.159,
      "step": 5340
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.222887694800065,
      "learning_rate": 1.9165709318103133e-06,
      "loss": 0.2328,
      "step": 5341
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0068035866795517,
      "learning_rate": 1.9148521301671173e-06,
      "loss": 0.1351,
      "step": 5342
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9326670631790344,
      "learning_rate": 1.9131339170363e-06,
      "loss": 0.1749,
      "step": 5343
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1510465501893044,
      "learning_rate": 1.911416292745622e-06,
      "loss": 0.2035,
      "step": 5344
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7838839829117576,
      "learning_rate": 1.9096992576227317e-06,
      "loss": 0.1259,
      "step": 5345
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9565993591412771,
      "learning_rate": 1.907982811995165e-06,
      "loss": 0.1612,
      "step": 5346
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9384915829239686,
      "learning_rate": 1.906266956190345e-06,
      "loss": 0.1502,
      "step": 5347
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8726706958401146,
      "learning_rate": 1.9045516905355826e-06,
      "loss": 0.1195,
      "step": 5348
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8842047133941989,
      "learning_rate": 1.9028370153580761e-06,
      "loss": 0.1438,
      "step": 5349
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9495672725360129,
      "learning_rate": 1.9011229309849111e-06,
      "loss": 0.1401,
      "step": 5350
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9417264392585245,
      "learning_rate": 1.8994094377430605e-06,
      "loss": 0.13,
      "step": 5351
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8657780191659399,
      "learning_rate": 1.8976965359593852e-06,
      "loss": 0.1337,
      "step": 5352
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8961755915593453,
      "learning_rate": 1.8959842259606315e-06,
      "loss": 0.1127,
      "step": 5353
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8968673043411751,
      "learning_rate": 1.8942725080734342e-06,
      "loss": 0.1608,
      "step": 5354
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7618691511198207,
      "learning_rate": 1.892561382624315e-06,
      "loss": 0.124,
      "step": 5355
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8051426497114296,
      "learning_rate": 1.8908508499396821e-06,
      "loss": 0.1229,
      "step": 5356
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9765329372765934,
      "learning_rate": 1.8891409103458298e-06,
      "loss": 0.1746,
      "step": 5357
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1614209139481608,
      "learning_rate": 1.8874315641689433e-06,
      "loss": 0.1888,
      "step": 5358
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7888030207381139,
      "learning_rate": 1.8857228117350917e-06,
      "loss": 0.0969,
      "step": 5359
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.263304422261045,
      "learning_rate": 1.8840146533702281e-06,
      "loss": 0.2088,
      "step": 5360
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1608047090302553,
      "learning_rate": 1.8823070894001955e-06,
      "loss": 0.1719,
      "step": 5361
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1367430240685457,
      "learning_rate": 1.8806001201507245e-06,
      "loss": 0.1574,
      "step": 5362
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9842733680059509,
      "learning_rate": 1.8788937459474294e-06,
      "loss": 0.2035,
      "step": 5363
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7915964174443448,
      "learning_rate": 1.8771879671158132e-06,
      "loss": 0.12,
      "step": 5364
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9341801845188683,
      "learning_rate": 1.8754827839812646e-06,
      "loss": 0.1578,
      "step": 5365
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7648744377262323,
      "learning_rate": 1.8737781968690578e-06,
      "loss": 0.1021,
      "step": 5366
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.345672414325671,
      "learning_rate": 1.8720742061043545e-06,
      "loss": 0.2273,
      "step": 5367
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8347922674002691,
      "learning_rate": 1.8703708120122027e-06,
      "loss": 0.1412,
      "step": 5368
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1877888656630105,
      "learning_rate": 1.868668014917534e-06,
      "loss": 0.1853,
      "step": 5369
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9773247489617155,
      "learning_rate": 1.8669658151451713e-06,
      "loss": 0.1668,
      "step": 5370
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9132367948971447,
      "learning_rate": 1.865264213019819e-06,
      "loss": 0.1661,
      "step": 5371
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1125178856587739,
      "learning_rate": 1.8635632088660694e-06,
      "loss": 0.2488,
      "step": 5372
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.092058812786988,
      "learning_rate": 1.8618628030083996e-06,
      "loss": 0.1853,
      "step": 5373
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.861028066630506,
      "learning_rate": 1.8601629957711736e-06,
      "loss": 0.1976,
      "step": 5374
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7866770316065402,
      "learning_rate": 1.858463787478641e-06,
      "loss": 0.1454,
      "step": 5375
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0065448268292279,
      "learning_rate": 1.8567651784549385e-06,
      "loss": 0.1472,
      "step": 5376
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0082142925520245,
      "learning_rate": 1.8550671690240835e-06,
      "loss": 0.1686,
      "step": 5377
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8863809009780179,
      "learning_rate": 1.8533697595099848e-06,
      "loss": 0.1156,
      "step": 5378
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8751721879883956,
      "learning_rate": 1.8516729502364345e-06,
      "loss": 0.1598,
      "step": 5379
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9231061399215724,
      "learning_rate": 1.8499767415271074e-06,
      "loss": 0.1604,
      "step": 5380
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9323320970002122,
      "learning_rate": 1.8482811337055713e-06,
      "loss": 0.1328,
      "step": 5381
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9372305690183812,
      "learning_rate": 1.8465861270952724e-06,
      "loss": 0.1614,
      "step": 5382
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8735599435618272,
      "learning_rate": 1.844891722019544e-06,
      "loss": 0.1609,
      "step": 5383
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8629125894800985,
      "learning_rate": 1.8431979188016054e-06,
      "loss": 0.1436,
      "step": 5384
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9388482019274271,
      "learning_rate": 1.8415047177645607e-06,
      "loss": 0.1369,
      "step": 5385
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9553333220720569,
      "learning_rate": 1.8398121192314e-06,
      "loss": 0.1678,
      "step": 5386
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7228720751952686,
      "learning_rate": 1.8381201235249962e-06,
      "loss": 0.1487,
      "step": 5387
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8413931767789704,
      "learning_rate": 1.83642873096811e-06,
      "loss": 0.1304,
      "step": 5388
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1904565139417393,
      "learning_rate": 1.834737941883385e-06,
      "loss": 0.1981,
      "step": 5389
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.280228348121079,
      "learning_rate": 1.8330477565933507e-06,
      "loss": 0.2036,
      "step": 5390
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2594449220718116,
      "learning_rate": 1.8313581754204214e-06,
      "loss": 0.2191,
      "step": 5391
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.170869738458693,
      "learning_rate": 1.8296691986868947e-06,
      "loss": 0.1875,
      "step": 5392
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1119545650276978,
      "learning_rate": 1.8279808267149558e-06,
      "loss": 0.214,
      "step": 5393
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.00294364287825,
      "learning_rate": 1.8262930598266714e-06,
      "loss": 0.1892,
      "step": 5394
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0639947492333646,
      "learning_rate": 1.8246058983439952e-06,
      "loss": 0.1498,
      "step": 5395
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6871366144639812,
      "learning_rate": 1.8229193425887637e-06,
      "loss": 0.1353,
      "step": 5396
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8269479362672745,
      "learning_rate": 1.8212333928826986e-06,
      "loss": 0.1224,
      "step": 5397
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6467686611127685,
      "learning_rate": 1.8195480495474065e-06,
      "loss": 0.1001,
      "step": 5398
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0007948539676346,
      "learning_rate": 1.8178633129043776e-06,
      "loss": 0.1759,
      "step": 5399
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8885275144394643,
      "learning_rate": 1.8161791832749863e-06,
      "loss": 0.1302,
      "step": 5400
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0401311793780996,
      "learning_rate": 1.814495660980492e-06,
      "loss": 0.194,
      "step": 5401
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8206311143511128,
      "learning_rate": 1.8128127463420358e-06,
      "loss": 0.1007,
      "step": 5402
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0963129894110348,
      "learning_rate": 1.8111304396806478e-06,
      "loss": 0.1564,
      "step": 5403
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7995167093211596,
      "learning_rate": 1.8094487413172379e-06,
      "loss": 0.1304,
      "step": 5404
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2468767398170977,
      "learning_rate": 1.8077676515726012e-06,
      "loss": 0.2007,
      "step": 5405
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.134145907113633,
      "learning_rate": 1.8060871707674178e-06,
      "loss": 0.1999,
      "step": 5406
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9119617384773656,
      "learning_rate": 1.8044072992222478e-06,
      "loss": 0.1507,
      "step": 5407
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3348615056878388,
      "learning_rate": 1.802728037257539e-06,
      "loss": 0.2352,
      "step": 5408
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1781174410333242,
      "learning_rate": 1.8010493851936223e-06,
      "loss": 0.1871,
      "step": 5409
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8941143045490761,
      "learning_rate": 1.7993713433507115e-06,
      "loss": 0.1507,
      "step": 5410
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.963012396037207,
      "learning_rate": 1.797693912048904e-06,
      "loss": 0.1893,
      "step": 5411
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1095288047957839,
      "learning_rate": 1.796017091608181e-06,
      "loss": 0.1797,
      "step": 5412
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8549753379826608,
      "learning_rate": 1.7943408823484054e-06,
      "loss": 0.122,
      "step": 5413
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.047721072342754,
      "learning_rate": 1.792665284589329e-06,
      "loss": 0.1726,
      "step": 5414
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0945967833304888,
      "learning_rate": 1.7909902986505806e-06,
      "loss": 0.1817,
      "step": 5415
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0425265157849413,
      "learning_rate": 1.789315924851675e-06,
      "loss": 0.2065,
      "step": 5416
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8610060232200819,
      "learning_rate": 1.787642163512011e-06,
      "loss": 0.1269,
      "step": 5417
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9606647165541904,
      "learning_rate": 1.7859690149508696e-06,
      "loss": 0.2041,
      "step": 5418
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8281492129533026,
      "learning_rate": 1.7842964794874135e-06,
      "loss": 0.1358,
      "step": 5419
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6473910054004474,
      "learning_rate": 1.7826245574406915e-06,
      "loss": 0.1609,
      "step": 5420
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7961446192204192,
      "learning_rate": 1.7809532491296332e-06,
      "loss": 0.1545,
      "step": 5421
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.007512520091417,
      "learning_rate": 1.7792825548730513e-06,
      "loss": 0.1336,
      "step": 5422
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8522686896225697,
      "learning_rate": 1.777612474989644e-06,
      "loss": 0.1493,
      "step": 5423
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.011361838350534,
      "learning_rate": 1.7759430097979868e-06,
      "loss": 0.1451,
      "step": 5424
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0789222974700516,
      "learning_rate": 1.7742741596165407e-06,
      "loss": 0.1683,
      "step": 5425
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.253729487296883,
      "learning_rate": 1.772605924763654e-06,
      "loss": 0.207,
      "step": 5426
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7606642516651748,
      "learning_rate": 1.7709383055575514e-06,
      "loss": 0.1235,
      "step": 5427
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7825772703010311,
      "learning_rate": 1.769271302316342e-06,
      "loss": 0.1479,
      "step": 5428
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1599054520153789,
      "learning_rate": 1.7676049153580188e-06,
      "loss": 0.2025,
      "step": 5429
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9852829728043756,
      "learning_rate": 1.7659391450004548e-06,
      "loss": 0.175,
      "step": 5430
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8094354614684406,
      "learning_rate": 1.7642739915614077e-06,
      "loss": 0.1547,
      "step": 5431
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7545588845526919,
      "learning_rate": 1.762609455358516e-06,
      "loss": 0.0985,
      "step": 5432
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0354635576010562,
      "learning_rate": 1.760945536709301e-06,
      "loss": 0.1559,
      "step": 5433
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1567417956550479,
      "learning_rate": 1.7592822359311667e-06,
      "loss": 0.1668,
      "step": 5434
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1672085174680804,
      "learning_rate": 1.7576195533413981e-06,
      "loss": 0.1766,
      "step": 5435
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7467470971973799,
      "learning_rate": 1.7559574892571635e-06,
      "loss": 0.131,
      "step": 5436
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9500435463888054,
      "learning_rate": 1.7542960439955126e-06,
      "loss": 0.1604,
      "step": 5437
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2391174108957301,
      "learning_rate": 1.7526352178733762e-06,
      "loss": 0.221,
      "step": 5438
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9182079286650079,
      "learning_rate": 1.7509750112075692e-06,
      "loss": 0.1754,
      "step": 5439
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0716663594325768,
      "learning_rate": 1.7493154243147858e-06,
      "loss": 0.1375,
      "step": 5440
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0765272277260522,
      "learning_rate": 1.747656457511604e-06,
      "loss": 0.1392,
      "step": 5441
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0319700167968473,
      "learning_rate": 1.7459981111144826e-06,
      "loss": 0.1814,
      "step": 5442
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7886637480103499,
      "learning_rate": 1.7443403854397618e-06,
      "loss": 0.1529,
      "step": 5443
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2076623677161185,
      "learning_rate": 1.7426832808036638e-06,
      "loss": 0.2207,
      "step": 5444
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9509386673433309,
      "learning_rate": 1.7410267975222922e-06,
      "loss": 0.1495,
      "step": 5445
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9602054767810572,
      "learning_rate": 1.7393709359116323e-06,
      "loss": 0.131,
      "step": 5446
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9033005096261524,
      "learning_rate": 1.7377156962875496e-06,
      "loss": 0.143,
      "step": 5447
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0625621679589154,
      "learning_rate": 1.7360610789657944e-06,
      "loss": 0.1829,
      "step": 5448
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6372684838955184,
      "learning_rate": 1.7344070842619948e-06,
      "loss": 0.1161,
      "step": 5449
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9634444178010282,
      "learning_rate": 1.732753712491661e-06,
      "loss": 0.1493,
      "step": 5450
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8615483070364983,
      "learning_rate": 1.7311009639701848e-06,
      "loss": 0.1315,
      "step": 5451
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0988195880025384,
      "learning_rate": 1.7294488390128384e-06,
      "loss": 0.1733,
      "step": 5452
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.005520054751948,
      "learning_rate": 1.727797337934778e-06,
      "loss": 0.1825,
      "step": 5453
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7633987827384069,
      "learning_rate": 1.7261464610510347e-06,
      "loss": 0.1076,
      "step": 5454
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7387304410193812,
      "learning_rate": 1.7244962086765254e-06,
      "loss": 0.1214,
      "step": 5455
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8386256169952521,
      "learning_rate": 1.7228465811260475e-06,
      "loss": 0.1628,
      "step": 5456
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9566476741462138,
      "learning_rate": 1.7211975787142777e-06,
      "loss": 0.1576,
      "step": 5457
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8552766362973051,
      "learning_rate": 1.719549201755773e-06,
      "loss": 0.1553,
      "step": 5458
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9743817772253441,
      "learning_rate": 1.7179014505649755e-06,
      "loss": 0.1953,
      "step": 5459
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9444874071094768,
      "learning_rate": 1.716254325456202e-06,
      "loss": 0.1821,
      "step": 5460
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0923167002585408,
      "learning_rate": 1.714607826743654e-06,
      "loss": 0.1725,
      "step": 5461
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.952472455505045,
      "learning_rate": 1.7129619547414105e-06,
      "loss": 0.1407,
      "step": 5462
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0963534398476074,
      "learning_rate": 1.7113167097634342e-06,
      "loss": 0.2155,
      "step": 5463
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8164158874705063,
      "learning_rate": 1.7096720921235648e-06,
      "loss": 0.1184,
      "step": 5464
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8054809766976814,
      "learning_rate": 1.7080281021355256e-06,
      "loss": 0.1561,
      "step": 5465
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9705310204865646,
      "learning_rate": 1.706384740112917e-06,
      "loss": 0.14,
      "step": 5466
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0724777459505972,
      "learning_rate": 1.7047420063692222e-06,
      "loss": 0.1975,
      "step": 5467
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9497698007344063,
      "learning_rate": 1.7030999012178029e-06,
      "loss": 0.1798,
      "step": 5468
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8342502419295791,
      "learning_rate": 1.7014584249719024e-06,
      "loss": 0.1636,
      "step": 5469
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9821182758757945,
      "learning_rate": 1.6998175779446424e-06,
      "loss": 0.1466,
      "step": 5470
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9019738525741529,
      "learning_rate": 1.6981773604490254e-06,
      "loss": 0.1423,
      "step": 5471
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9152563559663186,
      "learning_rate": 1.6965377727979342e-06,
      "loss": 0.145,
      "step": 5472
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0315619596169243,
      "learning_rate": 1.6948988153041307e-06,
      "loss": 0.1706,
      "step": 5473
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0198951300663672,
      "learning_rate": 1.6932604882802567e-06,
      "loss": 0.1791,
      "step": 5474
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9348215652138215,
      "learning_rate": 1.6916227920388345e-06,
      "loss": 0.1388,
      "step": 5475
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7431968548710054,
      "learning_rate": 1.6899857268922647e-06,
      "loss": 0.1135,
      "step": 5476
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9143702913586842,
      "learning_rate": 1.6883492931528289e-06,
      "loss": 0.1348,
      "step": 5477
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9970812233112853,
      "learning_rate": 1.6867134911326877e-06,
      "loss": 0.1891,
      "step": 5478
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9384317346354392,
      "learning_rate": 1.6850783211438804e-06,
      "loss": 0.1782,
      "step": 5479
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1001379968157268,
      "learning_rate": 1.6834437834983274e-06,
      "loss": 0.2264,
      "step": 5480
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7307114684172865,
      "learning_rate": 1.6818098785078257e-06,
      "loss": 0.1484,
      "step": 5481
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9344541317650762,
      "learning_rate": 1.6801766064840587e-06,
      "loss": 0.1645,
      "step": 5482
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1581846894826273,
      "learning_rate": 1.6785439677385778e-06,
      "loss": 0.1385,
      "step": 5483
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1489096849929483,
      "learning_rate": 1.6769119625828228e-06,
      "loss": 0.2031,
      "step": 5484
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.982095821377364,
      "learning_rate": 1.6752805913281085e-06,
      "loss": 0.1803,
      "step": 5485
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2067290595279863,
      "learning_rate": 1.6736498542856294e-06,
      "loss": 0.1941,
      "step": 5486
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0047829592389081,
      "learning_rate": 1.6720197517664604e-06,
      "loss": 0.1398,
      "step": 5487
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.15305928225377,
      "learning_rate": 1.6703902840815539e-06,
      "loss": 0.1752,
      "step": 5488
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9516659481619251,
      "learning_rate": 1.6687614515417416e-06,
      "loss": 0.2168,
      "step": 5489
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0483501218882803,
      "learning_rate": 1.6671332544577345e-06,
      "loss": 0.1639,
      "step": 5490
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0610141017427652,
      "learning_rate": 1.6655056931401215e-06,
      "loss": 0.1564,
      "step": 5491
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8270610909792203,
      "learning_rate": 1.663878767899369e-06,
      "loss": 0.1344,
      "step": 5492
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0045895550506574,
      "learning_rate": 1.6622524790458273e-06,
      "loss": 0.1499,
      "step": 5493
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.551689764944302,
      "learning_rate": 1.6606268268897202e-06,
      "loss": 0.0801,
      "step": 5494
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.844407426078191,
      "learning_rate": 1.659001811741151e-06,
      "loss": 0.1469,
      "step": 5495
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0956942505532217,
      "learning_rate": 1.6573774339101028e-06,
      "loss": 0.1872,
      "step": 5496
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9115158175073138,
      "learning_rate": 1.6557536937064366e-06,
      "loss": 0.1799,
      "step": 5497
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7484539065891064,
      "learning_rate": 1.6541305914398908e-06,
      "loss": 0.1178,
      "step": 5498
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0333671128011532,
      "learning_rate": 1.652508127420085e-06,
      "loss": 0.1862,
      "step": 5499
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1806714706709942,
      "learning_rate": 1.6508863019565114e-06,
      "loss": 0.2127,
      "step": 5500
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3139308982416964,
      "learning_rate": 1.6492651153585459e-06,
      "loss": 0.2415,
      "step": 5501
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.682756636954985,
      "learning_rate": 1.6476445679354408e-06,
      "loss": 0.1327,
      "step": 5502
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.043175784904232,
      "learning_rate": 1.6460246599963238e-06,
      "loss": 0.1846,
      "step": 5503
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.102911454743628,
      "learning_rate": 1.6444053918502068e-06,
      "loss": 0.1843,
      "step": 5504
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0136337027989546,
      "learning_rate": 1.6427867638059746e-06,
      "loss": 0.1725,
      "step": 5505
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0436292667882043,
      "learning_rate": 1.6411687761723906e-06,
      "loss": 0.2029,
      "step": 5506
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9342564730889761,
      "learning_rate": 1.6395514292580971e-06,
      "loss": 0.1274,
      "step": 5507
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9673205167441755,
      "learning_rate": 1.6379347233716136e-06,
      "loss": 0.1741,
      "step": 5508
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1323444102190257,
      "learning_rate": 1.636318658821337e-06,
      "loss": 0.2095,
      "step": 5509
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8352982749066139,
      "learning_rate": 1.6347032359155435e-06,
      "loss": 0.1689,
      "step": 5510
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9289999106356058,
      "learning_rate": 1.6330884549623844e-06,
      "loss": 0.1234,
      "step": 5511
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8227342329762959,
      "learning_rate": 1.6314743162698903e-06,
      "loss": 0.1569,
      "step": 5512
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6304139715984225,
      "learning_rate": 1.6298608201459692e-06,
      "loss": 0.1017,
      "step": 5513
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.284321275418402,
      "learning_rate": 1.6282479668984064e-06,
      "loss": 0.202,
      "step": 5514
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7373380090481375,
      "learning_rate": 1.626635756834863e-06,
      "loss": 0.1218,
      "step": 5515
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2768852473346564,
      "learning_rate": 1.62502419026288e-06,
      "loss": 0.2195,
      "step": 5516
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1362639833519115,
      "learning_rate": 1.623413267489874e-06,
      "loss": 0.1853,
      "step": 5517
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.047154621570371,
      "learning_rate": 1.6218029888231391e-06,
      "loss": 0.1735,
      "step": 5518
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1270426496368198,
      "learning_rate": 1.6201933545698467e-06,
      "loss": 0.2275,
      "step": 5519
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5647129504044901,
      "learning_rate": 1.6185843650370448e-06,
      "loss": 0.0969,
      "step": 5520
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.124113837138375,
      "learning_rate": 1.61697602053166e-06,
      "loss": 0.1682,
      "step": 5521
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8908526677721254,
      "learning_rate": 1.6153683213604932e-06,
      "loss": 0.1711,
      "step": 5522
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.956939829983241,
      "learning_rate": 1.6137612678302245e-06,
      "loss": 0.1697,
      "step": 5523
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6998761363753127,
      "learning_rate": 1.6121548602474096e-06,
      "loss": 0.1066,
      "step": 5524
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6648704446287486,
      "learning_rate": 1.6105490989184813e-06,
      "loss": 0.119,
      "step": 5525
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9293492675173619,
      "learning_rate": 1.6089439841497485e-06,
      "loss": 0.1625,
      "step": 5526
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9876269026968426,
      "learning_rate": 1.6073395162473997e-06,
      "loss": 0.1674,
      "step": 5527
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6395535729600427,
      "learning_rate": 1.6057356955174964e-06,
      "loss": 0.0844,
      "step": 5528
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.801493152608683,
      "learning_rate": 1.6041325222659805e-06,
      "loss": 0.1193,
      "step": 5529
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3240641542918758,
      "learning_rate": 1.602529996798663e-06,
      "loss": 0.2315,
      "step": 5530
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0454119522016874,
      "learning_rate": 1.6009281194212388e-06,
      "loss": 0.1783,
      "step": 5531
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.077045539841088,
      "learning_rate": 1.5993268904392772e-06,
      "loss": 0.1535,
      "step": 5532
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9727301000594214,
      "learning_rate": 1.5977263101582225e-06,
      "loss": 0.1645,
      "step": 5533
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8747164168925882,
      "learning_rate": 1.5961263788833959e-06,
      "loss": 0.1809,
      "step": 5534
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1405230482328645,
      "learning_rate": 1.5945270969199955e-06,
      "loss": 0.1897,
      "step": 5535
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8783806163066304,
      "learning_rate": 1.5929284645730942e-06,
      "loss": 0.1493,
      "step": 5536
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0328312553917045,
      "learning_rate": 1.5913304821476405e-06,
      "loss": 0.165,
      "step": 5537
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0454692372869767,
      "learning_rate": 1.589733149948463e-06,
      "loss": 0.1671,
      "step": 5538
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7758327226389495,
      "learning_rate": 1.5881364682802618e-06,
      "loss": 0.1539,
      "step": 5539
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0224892774115582,
      "learning_rate": 1.5865404374476156e-06,
      "loss": 0.1757,
      "step": 5540
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0560452256710988,
      "learning_rate": 1.5849450577549763e-06,
      "loss": 0.1717,
      "step": 5541
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9024617933965221,
      "learning_rate": 1.5833503295066744e-06,
      "loss": 0.1555,
      "step": 5542
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.880744951425145,
      "learning_rate": 1.5817562530069135e-06,
      "loss": 0.1382,
      "step": 5543
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7048328101355242,
      "learning_rate": 1.5801628285597759e-06,
      "loss": 0.1418,
      "step": 5544
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7987488367880792,
      "learning_rate": 1.5785700564692163e-06,
      "loss": 0.1376,
      "step": 5545
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8413783461022013,
      "learning_rate": 1.5769779370390692e-06,
      "loss": 0.1169,
      "step": 5546
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.029651895164156,
      "learning_rate": 1.5753864705730382e-06,
      "loss": 0.172,
      "step": 5547
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1538576137785854,
      "learning_rate": 1.5737956573747059e-06,
      "loss": 0.1803,
      "step": 5548
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9432674627611783,
      "learning_rate": 1.5722054977475342e-06,
      "loss": 0.1637,
      "step": 5549
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8959409774455163,
      "learning_rate": 1.570615991994855e-06,
      "loss": 0.131,
      "step": 5550
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0463617303541264,
      "learning_rate": 1.5690271404198765e-06,
      "loss": 0.156,
      "step": 5551
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2061974851011135,
      "learning_rate": 1.5674389433256826e-06,
      "loss": 0.1749,
      "step": 5552
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9877365891002947,
      "learning_rate": 1.5658514010152332e-06,
      "loss": 0.1658,
      "step": 5553
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9293110973778718,
      "learning_rate": 1.564264513791362e-06,
      "loss": 0.2134,
      "step": 5554
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9173628612286365,
      "learning_rate": 1.5626782819567781e-06,
      "loss": 0.1585,
      "step": 5555
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.962001324113383,
      "learning_rate": 1.5610927058140663e-06,
      "loss": 0.1486,
      "step": 5556
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1624011752326804,
      "learning_rate": 1.559507785665685e-06,
      "loss": 0.1696,
      "step": 5557
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8759865584940039,
      "learning_rate": 1.5579235218139682e-06,
      "loss": 0.1407,
      "step": 5558
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8271025241275688,
      "learning_rate": 1.5563399145611258e-06,
      "loss": 0.1251,
      "step": 5559
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9544529717132422,
      "learning_rate": 1.5547569642092397e-06,
      "loss": 0.1707,
      "step": 5560
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9888278611344647,
      "learning_rate": 1.5531746710602701e-06,
      "loss": 0.1614,
      "step": 5561
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0416680824334568,
      "learning_rate": 1.5515930354160485e-06,
      "loss": 0.1543,
      "step": 5562
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8725634249036154,
      "learning_rate": 1.5500120575782823e-06,
      "loss": 0.1329,
      "step": 5563
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1104195833926191,
      "learning_rate": 1.5484317378485548e-06,
      "loss": 0.1865,
      "step": 5564
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0939594770680836,
      "learning_rate": 1.5468520765283207e-06,
      "loss": 0.1488,
      "step": 5565
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8151971076384611,
      "learning_rate": 1.5452730739189125e-06,
      "loss": 0.1317,
      "step": 5566
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.792759132210997,
      "learning_rate": 1.5436947303215344e-06,
      "loss": 0.1431,
      "step": 5567
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.010319955220926,
      "learning_rate": 1.5421170460372659e-06,
      "loss": 0.1973,
      "step": 5568
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7824304138752047,
      "learning_rate": 1.5405400213670612e-06,
      "loss": 0.1234,
      "step": 5569
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.118546803335767,
      "learning_rate": 1.5389636566117478e-06,
      "loss": 0.2131,
      "step": 5570
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0304472791435797,
      "learning_rate": 1.5373879520720264e-06,
      "loss": 0.1809,
      "step": 5571
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6428866752117786,
      "learning_rate": 1.5358129080484763e-06,
      "loss": 0.1435,
      "step": 5572
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8789017747528924,
      "learning_rate": 1.5342385248415453e-06,
      "loss": 0.1561,
      "step": 5573
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.193529561889568,
      "learning_rate": 1.5326648027515583e-06,
      "loss": 0.2206,
      "step": 5574
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7778456083303918,
      "learning_rate": 1.531091742078713e-06,
      "loss": 0.1579,
      "step": 5575
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8606557804844505,
      "learning_rate": 1.529519343123082e-06,
      "loss": 0.1605,
      "step": 5576
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8069643537418828,
      "learning_rate": 1.5279476061846088e-06,
      "loss": 0.1396,
      "step": 5577
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9620302721686195,
      "learning_rate": 1.5263765315631134e-06,
      "loss": 0.1734,
      "step": 5578
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8893656261145784,
      "learning_rate": 1.5248061195582886e-06,
      "loss": 0.1464,
      "step": 5579
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9261914567791036,
      "learning_rate": 1.5232363704697017e-06,
      "loss": 0.1459,
      "step": 5580
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.086310222780151,
      "learning_rate": 1.5216672845967918e-06,
      "loss": 0.1777,
      "step": 5581
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8644090970310923,
      "learning_rate": 1.5200988622388713e-06,
      "loss": 0.1631,
      "step": 5582
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.235735675725106,
      "learning_rate": 1.51853110369513e-06,
      "loss": 0.2307,
      "step": 5583
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0952194450837458,
      "learning_rate": 1.5169640092646275e-06,
      "loss": 0.2012,
      "step": 5584
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7398299296211878,
      "learning_rate": 1.5153975792462966e-06,
      "loss": 0.1179,
      "step": 5585
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0347185627966597,
      "learning_rate": 1.513831813938944e-06,
      "loss": 0.1474,
      "step": 5586
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2622209013377907,
      "learning_rate": 1.5122667136412511e-06,
      "loss": 0.2183,
      "step": 5587
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9466966713189474,
      "learning_rate": 1.5107022786517694e-06,
      "loss": 0.2023,
      "step": 5588
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9791336878466935,
      "learning_rate": 1.5091385092689265e-06,
      "loss": 0.1559,
      "step": 5589
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.06101603680242,
      "learning_rate": 1.5075754057910214e-06,
      "loss": 0.2369,
      "step": 5590
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1955610819353113,
      "learning_rate": 1.5060129685162262e-06,
      "loss": 0.2095,
      "step": 5591
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0451710084567567,
      "learning_rate": 1.5044511977425862e-06,
      "loss": 0.1903,
      "step": 5592
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1678192508754173,
      "learning_rate": 1.5028900937680202e-06,
      "loss": 0.2016,
      "step": 5593
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7890223590441017,
      "learning_rate": 1.501329656890318e-06,
      "loss": 0.1374,
      "step": 5594
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9170023407665784,
      "learning_rate": 1.4997698874071443e-06,
      "loss": 0.1734,
      "step": 5595
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0555325633452421,
      "learning_rate": 1.498210785616035e-06,
      "loss": 0.1823,
      "step": 5596
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9241047222843082,
      "learning_rate": 1.4966523518143987e-06,
      "loss": 0.1511,
      "step": 5597
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0332136986723277,
      "learning_rate": 1.4950945862995176e-06,
      "loss": 0.1918,
      "step": 5598
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7067934023814627,
      "learning_rate": 1.493537489368545e-06,
      "loss": 0.1073,
      "step": 5599
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3075666604691603,
      "learning_rate": 1.4919810613185088e-06,
      "loss": 0.2156,
      "step": 5600
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.135673787936759,
      "learning_rate": 1.4904253024463066e-06,
      "loss": 0.1385,
      "step": 5601
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1021359164254065,
      "learning_rate": 1.4888702130487104e-06,
      "loss": 0.201,
      "step": 5602
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6399577586157662,
      "learning_rate": 1.4873157934223636e-06,
      "loss": 0.096,
      "step": 5603
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7872210984821215,
      "learning_rate": 1.4857620438637804e-06,
      "loss": 0.1551,
      "step": 5604
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.957906227860744,
      "learning_rate": 1.4842089646693536e-06,
      "loss": 0.1831,
      "step": 5605
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9703273195668216,
      "learning_rate": 1.4826565561353389e-06,
      "loss": 0.144,
      "step": 5606
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7493042032582041,
      "learning_rate": 1.4811048185578697e-06,
      "loss": 0.1186,
      "step": 5607
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6747202643314286,
      "learning_rate": 1.47955375223295e-06,
      "loss": 0.1167,
      "step": 5608
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8721626873468166,
      "learning_rate": 1.4780033574564568e-06,
      "loss": 0.1253,
      "step": 5609
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9627178599231091,
      "learning_rate": 1.4764536345241375e-06,
      "loss": 0.1812,
      "step": 5610
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8388705920721807,
      "learning_rate": 1.4749045837316123e-06,
      "loss": 0.1442,
      "step": 5611
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8618166817019361,
      "learning_rate": 1.4733562053743727e-06,
      "loss": 0.1614,
      "step": 5612
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0413587841562717,
      "learning_rate": 1.4718084997477821e-06,
      "loss": 0.2213,
      "step": 5613
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.040488357778776,
      "learning_rate": 1.4702614671470754e-06,
      "loss": 0.2158,
      "step": 5614
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1399917066332172,
      "learning_rate": 1.4687151078673584e-06,
      "loss": 0.1972,
      "step": 5615
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4147503745655832,
      "learning_rate": 1.467169422203612e-06,
      "loss": 0.2306,
      "step": 5616
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1358502844034042,
      "learning_rate": 1.4656244104506844e-06,
      "loss": 0.1796,
      "step": 5617
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1103195022267982,
      "learning_rate": 1.4640800729032973e-06,
      "loss": 0.2047,
      "step": 5618
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9910118593414691,
      "learning_rate": 1.4625364098560429e-06,
      "loss": 0.1873,
      "step": 5619
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0121085147324915,
      "learning_rate": 1.460993421603385e-06,
      "loss": 0.1419,
      "step": 5620
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.10786415179282,
      "learning_rate": 1.459451108439659e-06,
      "loss": 0.2308,
      "step": 5621
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7044331219777688,
      "learning_rate": 1.4579094706590718e-06,
      "loss": 0.1054,
      "step": 5622
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9493912619493424,
      "learning_rate": 1.4563685085557017e-06,
      "loss": 0.1646,
      "step": 5623
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9918273686576888,
      "learning_rate": 1.454828222423495e-06,
      "loss": 0.1548,
      "step": 5624
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0983804197984168,
      "learning_rate": 1.453288612556273e-06,
      "loss": 0.1864,
      "step": 5625
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0648045147064589,
      "learning_rate": 1.451749679247726e-06,
      "loss": 0.1971,
      "step": 5626
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8536868742808928,
      "learning_rate": 1.4502114227914138e-06,
      "loss": 0.1367,
      "step": 5627
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8308743792421408,
      "learning_rate": 1.4486738434807734e-06,
      "loss": 0.1102,
      "step": 5628
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2338921326083185,
      "learning_rate": 1.4471369416091064e-06,
      "loss": 0.2544,
      "step": 5629
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8969279313640397,
      "learning_rate": 1.4456007174695857e-06,
      "loss": 0.134,
      "step": 5630
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9770668390667788,
      "learning_rate": 1.4440651713552577e-06,
      "loss": 0.1108,
      "step": 5631
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7927420891065524,
      "learning_rate": 1.4425303035590372e-06,
      "loss": 0.1753,
      "step": 5632
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9489299400712636,
      "learning_rate": 1.440996114373711e-06,
      "loss": 0.1493,
      "step": 5633
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8917261578576979,
      "learning_rate": 1.439462604091935e-06,
      "loss": 0.1619,
      "step": 5634
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7609235212659502,
      "learning_rate": 1.4379297730062368e-06,
      "loss": 0.1462,
      "step": 5635
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8551429292482775,
      "learning_rate": 1.4363976214090147e-06,
      "loss": 0.157,
      "step": 5636
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9692526868830637,
      "learning_rate": 1.434866149592536e-06,
      "loss": 0.1881,
      "step": 5637
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0973316753119573,
      "learning_rate": 1.4333353578489396e-06,
      "loss": 0.1673,
      "step": 5638
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0825385513582186,
      "learning_rate": 1.4318052464702337e-06,
      "loss": 0.1694,
      "step": 5639
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8149984197755236,
      "learning_rate": 1.4302758157482972e-06,
      "loss": 0.162,
      "step": 5640
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9962131220599628,
      "learning_rate": 1.4287470659748792e-06,
      "loss": 0.1324,
      "step": 5641
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8915416187671057,
      "learning_rate": 1.427218997441599e-06,
      "loss": 0.1251,
      "step": 5642
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0812529006923497,
      "learning_rate": 1.425691610439946e-06,
      "loss": 0.1719,
      "step": 5643
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9362491981693407,
      "learning_rate": 1.4241649052612789e-06,
      "loss": 0.1259,
      "step": 5644
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8424636730629799,
      "learning_rate": 1.4226388821968267e-06,
      "loss": 0.1739,
      "step": 5645
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9562604652735005,
      "learning_rate": 1.4211135415376892e-06,
      "loss": 0.155,
      "step": 5646
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9031468710374682,
      "learning_rate": 1.4195888835748344e-06,
      "loss": 0.1534,
      "step": 5647
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2728602359404124,
      "learning_rate": 1.4180649085991015e-06,
      "loss": 0.2219,
      "step": 5648
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1111606718826983,
      "learning_rate": 1.416541616901197e-06,
      "loss": 0.2128,
      "step": 5649
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8797074975436308,
      "learning_rate": 1.415019008771702e-06,
      "loss": 0.1568,
      "step": 5650
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8814655184396062,
      "learning_rate": 1.4134970845010627e-06,
      "loss": 0.1534,
      "step": 5651
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.959618854986842,
      "learning_rate": 1.4119758443795972e-06,
      "loss": 0.1919,
      "step": 5652
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7698253445927986,
      "learning_rate": 1.4104552886974892e-06,
      "loss": 0.1049,
      "step": 5653
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9110000825562816,
      "learning_rate": 1.4089354177447974e-06,
      "loss": 0.1461,
      "step": 5654
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7366159046721621,
      "learning_rate": 1.4074162318114459e-06,
      "loss": 0.119,
      "step": 5655
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7464123105917371,
      "learning_rate": 1.4058977311872295e-06,
      "loss": 0.1403,
      "step": 5656
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7013735715734631,
      "learning_rate": 1.4043799161618132e-06,
      "loss": 0.1246,
      "step": 5657
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7132618340693992,
      "learning_rate": 1.4028627870247297e-06,
      "loss": 0.0992,
      "step": 5658
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7200297265489327,
      "learning_rate": 1.4013463440653812e-06,
      "loss": 0.1242,
      "step": 5659
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0617393123796535,
      "learning_rate": 1.3998305875730383e-06,
      "loss": 0.1697,
      "step": 5660
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7250957027438037,
      "learning_rate": 1.3983155178368447e-06,
      "loss": 0.1363,
      "step": 5661
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9339822146924116,
      "learning_rate": 1.3968011351458077e-06,
      "loss": 0.2175,
      "step": 5662
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6988704840035852,
      "learning_rate": 1.395287439788806e-06,
      "loss": 0.1294,
      "step": 5663
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9800955001482545,
      "learning_rate": 1.3937744320545871e-06,
      "loss": 0.1773,
      "step": 5664
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9105204826255374,
      "learning_rate": 1.392262112231768e-06,
      "loss": 0.1556,
      "step": 5665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3835248990232372,
      "learning_rate": 1.3907504806088323e-06,
      "loss": 0.2526,
      "step": 5666
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0099776428486726,
      "learning_rate": 1.3892395374741353e-06,
      "loss": 0.1673,
      "step": 5667
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9665290850852246,
      "learning_rate": 1.3877292831158983e-06,
      "loss": 0.137,
      "step": 5668
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8712943672350778,
      "learning_rate": 1.3862197178222143e-06,
      "loss": 0.1593,
      "step": 5669
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8684475796888456,
      "learning_rate": 1.384710841881039e-06,
      "loss": 0.1384,
      "step": 5670
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7722624935808339,
      "learning_rate": 1.3832026555802031e-06,
      "loss": 0.1258,
      "step": 5671
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0462096718908946,
      "learning_rate": 1.3816951592074013e-06,
      "loss": 0.1473,
      "step": 5672
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0720667145425578,
      "learning_rate": 1.3801883530502015e-06,
      "loss": 0.1646,
      "step": 5673
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0972721963892826,
      "learning_rate": 1.3786822373960352e-06,
      "loss": 0.159,
      "step": 5674
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8854877266050669,
      "learning_rate": 1.3771768125322044e-06,
      "loss": 0.1694,
      "step": 5675
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1248657585674666,
      "learning_rate": 1.3756720787458783e-06,
      "loss": 0.1712,
      "step": 5676
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9691120486744056,
      "learning_rate": 1.374168036324095e-06,
      "loss": 0.1557,
      "step": 5677
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1130400492617574,
      "learning_rate": 1.3726646855537606e-06,
      "loss": 0.1846,
      "step": 5678
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6587941779904131,
      "learning_rate": 1.3711620267216492e-06,
      "loss": 0.0776,
      "step": 5679
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0998800752416784,
      "learning_rate": 1.3696600601144034e-06,
      "loss": 0.1406,
      "step": 5680
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0158183768688154,
      "learning_rate": 1.3681587860185325e-06,
      "loss": 0.1287,
      "step": 5681
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9351924039951767,
      "learning_rate": 1.3666582047204146e-06,
      "loss": 0.146,
      "step": 5682
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0956756287615625,
      "learning_rate": 1.365158316506296e-06,
      "loss": 0.2,
      "step": 5683
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7018238401737436,
      "learning_rate": 1.3636591216622902e-06,
      "loss": 0.0994,
      "step": 5684
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9765663848449894,
      "learning_rate": 1.3621606204743782e-06,
      "loss": 0.1549,
      "step": 5685
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8824223612408356,
      "learning_rate": 1.3606628132284088e-06,
      "loss": 0.1667,
      "step": 5686
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8451413505798369,
      "learning_rate": 1.3591657002100995e-06,
      "loss": 0.1255,
      "step": 5687
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8403027281084284,
      "learning_rate": 1.357669281705034e-06,
      "loss": 0.1643,
      "step": 5688
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9881361673906525,
      "learning_rate": 1.356173557998664e-06,
      "loss": 0.1777,
      "step": 5689
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.060683946321515,
      "learning_rate": 1.3546785293763086e-06,
      "loss": 0.1892,
      "step": 5690
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1857659145387929,
      "learning_rate": 1.353184196123155e-06,
      "loss": 0.2204,
      "step": 5691
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.977262425871865,
      "learning_rate": 1.3516905585242568e-06,
      "loss": 0.169,
      "step": 5692
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7982493384346914,
      "learning_rate": 1.3501976168645348e-06,
      "loss": 0.1239,
      "step": 5693
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0247354937286968,
      "learning_rate": 1.3487053714287768e-06,
      "loss": 0.1928,
      "step": 5694
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.852812593073416,
      "learning_rate": 1.3472138225016406e-06,
      "loss": 0.1631,
      "step": 5695
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5204030194361894,
      "learning_rate": 1.3457229703676483e-06,
      "loss": 0.1027,
      "step": 5696
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8480415384857102,
      "learning_rate": 1.34423281531119e-06,
      "loss": 0.1673,
      "step": 5697
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0401528402016536,
      "learning_rate": 1.3427433576165217e-06,
      "loss": 0.1766,
      "step": 5698
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6362944400517881,
      "learning_rate": 1.3412545975677693e-06,
      "loss": 0.1068,
      "step": 5699
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0231105857822622,
      "learning_rate": 1.3397665354489208e-06,
      "loss": 0.1521,
      "step": 5700
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6071695400849098,
      "learning_rate": 1.3382791715438353e-06,
      "loss": 0.1018,
      "step": 5701
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2383091413430833,
      "learning_rate": 1.3367925061362374e-06,
      "loss": 0.2224,
      "step": 5702
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8546941545208087,
      "learning_rate": 1.335306539509718e-06,
      "loss": 0.133,
      "step": 5703
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0363729086290066,
      "learning_rate": 1.333821271947735e-06,
      "loss": 0.1592,
      "step": 5704
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.217536924233559,
      "learning_rate": 1.332336703733612e-06,
      "loss": 0.1899,
      "step": 5705
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9229985913767291,
      "learning_rate": 1.3308528351505433e-06,
      "loss": 0.2144,
      "step": 5706
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0197117648307006,
      "learning_rate": 1.3293696664815852e-06,
      "loss": 0.2049,
      "step": 5707
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8816045680616584,
      "learning_rate": 1.3278871980096608e-06,
      "loss": 0.1447,
      "step": 5708
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2009849465577755,
      "learning_rate": 1.326405430017562e-06,
      "loss": 0.2162,
      "step": 5709
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7288186268013396,
      "learning_rate": 1.324924362787946e-06,
      "loss": 0.1185,
      "step": 5710
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8989670028601034,
      "learning_rate": 1.3234439966033346e-06,
      "loss": 0.1466,
      "step": 5711
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.143571756799725,
      "learning_rate": 1.3219643317461195e-06,
      "loss": 0.2312,
      "step": 5712
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0362370638490492,
      "learning_rate": 1.3204853684985547e-06,
      "loss": 0.1605,
      "step": 5713
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9416623377296394,
      "learning_rate": 1.3190071071427634e-06,
      "loss": 0.1393,
      "step": 5714
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9690402946757679,
      "learning_rate": 1.3175295479607336e-06,
      "loss": 0.169,
      "step": 5715
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1828095383677444,
      "learning_rate": 1.3160526912343192e-06,
      "loss": 0.2237,
      "step": 5716
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.895356957220248,
      "learning_rate": 1.3145765372452402e-06,
      "loss": 0.1588,
      "step": 5717
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6989923770012598,
      "learning_rate": 1.3131010862750838e-06,
      "loss": 0.1251,
      "step": 5718
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.053154395071392,
      "learning_rate": 1.3116263386053002e-06,
      "loss": 0.1668,
      "step": 5719
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6655267388408878,
      "learning_rate": 1.310152294517209e-06,
      "loss": 0.1521,
      "step": 5720
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5724723569781257,
      "learning_rate": 1.3086789542919936e-06,
      "loss": 0.0961,
      "step": 5721
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6418876841328918,
      "learning_rate": 1.3072063182107026e-06,
      "loss": 0.1216,
      "step": 5722
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8260343972094495,
      "learning_rate": 1.3057343865542516e-06,
      "loss": 0.1107,
      "step": 5723
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7498775186017452,
      "learning_rate": 1.304263159603421e-06,
      "loss": 0.1279,
      "step": 5724
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9678020557367353,
      "learning_rate": 1.3027926376388578e-06,
      "loss": 0.2099,
      "step": 5725
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0039054490043784,
      "learning_rate": 1.3013228209410732e-06,
      "loss": 0.1576,
      "step": 5726
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8936079518505844,
      "learning_rate": 1.2998537097904445e-06,
      "loss": 0.1753,
      "step": 5727
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6870570955498208,
      "learning_rate": 1.298385304467213e-06,
      "loss": 0.0862,
      "step": 5728
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8477124578031644,
      "learning_rate": 1.296917605251491e-06,
      "loss": 0.168,
      "step": 5729
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2758245007415772,
      "learning_rate": 1.2954506124232475e-06,
      "loss": 0.1681,
      "step": 5730
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8308163229465512,
      "learning_rate": 1.2939843262623224e-06,
      "loss": 0.156,
      "step": 5731
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2840689816845992,
      "learning_rate": 1.2925187470484203e-06,
      "loss": 0.2161,
      "step": 5732
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9084468225293593,
      "learning_rate": 1.2910538750611095e-06,
      "loss": 0.1655,
      "step": 5733
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9473946240892267,
      "learning_rate": 1.2895897105798238e-06,
      "loss": 0.1835,
      "step": 5734
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9335530707941561,
      "learning_rate": 1.2881262538838624e-06,
      "loss": 0.1398,
      "step": 5735
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7521251725297335,
      "learning_rate": 1.2866635052523902e-06,
      "loss": 0.112,
      "step": 5736
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9405289823443415,
      "learning_rate": 1.2852014649644345e-06,
      "loss": 0.1883,
      "step": 5737
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8890110008495447,
      "learning_rate": 1.2837401332988907e-06,
      "loss": 0.1475,
      "step": 5738
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1765448998899475,
      "learning_rate": 1.2822795105345155e-06,
      "loss": 0.1714,
      "step": 5739
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6976119282258958,
      "learning_rate": 1.2808195969499354e-06,
      "loss": 0.1031,
      "step": 5740
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.201068247001143,
      "learning_rate": 1.279360392823637e-06,
      "loss": 0.183,
      "step": 5741
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8488294843275107,
      "learning_rate": 1.2779018984339724e-06,
      "loss": 0.1735,
      "step": 5742
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7074763858489392,
      "learning_rate": 1.2764441140591605e-06,
      "loss": 0.1303,
      "step": 5743
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1448669523865138,
      "learning_rate": 1.2749870399772824e-06,
      "loss": 0.2024,
      "step": 5744
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.096660117077752,
      "learning_rate": 1.2735306764662847e-06,
      "loss": 0.1899,
      "step": 5745
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1670308393461015,
      "learning_rate": 1.2720750238039804e-06,
      "loss": 0.2298,
      "step": 5746
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8724606208596126,
      "learning_rate": 1.2706200822680414e-06,
      "loss": 0.1237,
      "step": 5747
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0523918780296329,
      "learning_rate": 1.2691658521360084e-06,
      "loss": 0.2287,
      "step": 5748
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1798221604914925,
      "learning_rate": 1.267712333685287e-06,
      "loss": 0.2091,
      "step": 5749
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9486977847115328,
      "learning_rate": 1.2662595271931422e-06,
      "loss": 0.1566,
      "step": 5750
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9640932749986896,
      "learning_rate": 1.2648074329367105e-06,
      "loss": 0.1501,
      "step": 5751
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.839914390047931,
      "learning_rate": 1.2633560511929871e-06,
      "loss": 0.1314,
      "step": 5752
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.865525559295721,
      "learning_rate": 1.2619053822388321e-06,
      "loss": 0.1198,
      "step": 5753
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9623939121876963,
      "learning_rate": 1.2604554263509706e-06,
      "loss": 0.1574,
      "step": 5754
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1700873642642557,
      "learning_rate": 1.2590061838059908e-06,
      "loss": 0.2009,
      "step": 5755
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0647766185671295,
      "learning_rate": 1.2575576548803459e-06,
      "loss": 0.2032,
      "step": 5756
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9545967908547007,
      "learning_rate": 1.2561098398503518e-06,
      "loss": 0.1647,
      "step": 5757
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.547569332732202,
      "learning_rate": 1.2546627389921896e-06,
      "loss": 0.1115,
      "step": 5758
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1559550322012027,
      "learning_rate": 1.2532163525819025e-06,
      "loss": 0.1878,
      "step": 5759
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8410140430879167,
      "learning_rate": 1.2517706808953983e-06,
      "loss": 0.1529,
      "step": 5760
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0615234311262072,
      "learning_rate": 1.2503257242084494e-06,
      "loss": 0.1553,
      "step": 5761
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9657827707122657,
      "learning_rate": 1.2488814827966906e-06,
      "loss": 0.1407,
      "step": 5762
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2209844953084565,
      "learning_rate": 1.2474379569356194e-06,
      "loss": 0.1959,
      "step": 5763
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.112431483463938,
      "learning_rate": 1.2459951469005987e-06,
      "loss": 0.2041,
      "step": 5764
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7924979071202282,
      "learning_rate": 1.2445530529668547e-06,
      "loss": 0.1407,
      "step": 5765
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9393149628701397,
      "learning_rate": 1.2431116754094751e-06,
      "loss": 0.1382,
      "step": 5766
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8663802754703375,
      "learning_rate": 1.2416710145034133e-06,
      "loss": 0.1237,
      "step": 5767
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1349159709871812,
      "learning_rate": 1.2402310705234844e-06,
      "loss": 0.1847,
      "step": 5768
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4947135698562794,
      "learning_rate": 1.238791843744367e-06,
      "loss": 0.1074,
      "step": 5769
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9991133720430295,
      "learning_rate": 1.2373533344406036e-06,
      "loss": 0.1447,
      "step": 5770
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0036467647103433,
      "learning_rate": 1.2359155428865989e-06,
      "loss": 0.1592,
      "step": 5771
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8894971165455631,
      "learning_rate": 1.234478469356622e-06,
      "loss": 0.1748,
      "step": 5772
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8578524020533317,
      "learning_rate": 1.2330421141248016e-06,
      "loss": 0.1169,
      "step": 5773
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7990018760335045,
      "learning_rate": 1.2316064774651353e-06,
      "loss": 0.151,
      "step": 5774
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.863817344342179,
      "learning_rate": 1.2301715596514814e-06,
      "loss": 0.1376,
      "step": 5775
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9777861093657693,
      "learning_rate": 1.2287373609575553e-06,
      "loss": 0.1511,
      "step": 5776
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8980639368679179,
      "learning_rate": 1.2273038816569421e-06,
      "loss": 0.1305,
      "step": 5777
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1526784947414448,
      "learning_rate": 1.2258711220230878e-06,
      "loss": 0.2194,
      "step": 5778
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9345581009230273,
      "learning_rate": 1.2244390823293006e-06,
      "loss": 0.1401,
      "step": 5779
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9845779663302904,
      "learning_rate": 1.223007762848752e-06,
      "loss": 0.1967,
      "step": 5780
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5192636589269721,
      "learning_rate": 1.2215771638544743e-06,
      "loss": 0.0733,
      "step": 5781
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1569218461734108,
      "learning_rate": 1.2201472856193648e-06,
      "loss": 0.2252,
      "step": 5782
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8333565426033752,
      "learning_rate": 1.218718128416182e-06,
      "loss": 0.1516,
      "step": 5783
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.104261750652682,
      "learning_rate": 1.2172896925175455e-06,
      "loss": 0.1627,
      "step": 5784
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1045303133166224,
      "learning_rate": 1.2158619781959418e-06,
      "loss": 0.1784,
      "step": 5785
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9459964389801034,
      "learning_rate": 1.2144349857237159e-06,
      "loss": 0.1784,
      "step": 5786
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1235954821374865,
      "learning_rate": 1.2130087153730757e-06,
      "loss": 0.168,
      "step": 5787
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7743037116178275,
      "learning_rate": 1.2115831674160916e-06,
      "loss": 0.1677,
      "step": 5788
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.96178501967681,
      "learning_rate": 1.2101583421246965e-06,
      "loss": 0.1554,
      "step": 5789
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7926006070067841,
      "learning_rate": 1.2087342397706852e-06,
      "loss": 0.1253,
      "step": 5790
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6722061974720122,
      "learning_rate": 1.207310860625715e-06,
      "loss": 0.0753,
      "step": 5791
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8025428243093825,
      "learning_rate": 1.2058882049613053e-06,
      "loss": 0.1571,
      "step": 5792
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7892043138004476,
      "learning_rate": 1.2044662730488355e-06,
      "loss": 0.1496,
      "step": 5793
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9343514716197824,
      "learning_rate": 1.2030450651595489e-06,
      "loss": 0.2097,
      "step": 5794
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0350889543393462,
      "learning_rate": 1.2016245815645493e-06,
      "loss": 0.1891,
      "step": 5795
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.951850004369731,
      "learning_rate": 1.2002048225348067e-06,
      "loss": 0.1624,
      "step": 5796
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8938291847193152,
      "learning_rate": 1.1987857883411474e-06,
      "loss": 0.1349,
      "step": 5797
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9903595850030387,
      "learning_rate": 1.1973674792542618e-06,
      "loss": 0.1848,
      "step": 5798
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9057001580123292,
      "learning_rate": 1.1959498955447014e-06,
      "loss": 0.1712,
      "step": 5799
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0414981747201142,
      "learning_rate": 1.1945330374828806e-06,
      "loss": 0.2131,
      "step": 5800
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9049625459201293,
      "learning_rate": 1.1931169053390735e-06,
      "loss": 0.1696,
      "step": 5801
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9150623282329752,
      "learning_rate": 1.1917014993834175e-06,
      "loss": 0.1676,
      "step": 5802
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8723303822805556,
      "learning_rate": 1.1902868198859096e-06,
      "loss": 0.1567,
      "step": 5803
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1708962536388647,
      "learning_rate": 1.1888728671164102e-06,
      "loss": 0.2148,
      "step": 5804
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0068797622974566,
      "learning_rate": 1.1874596413446398e-06,
      "loss": 0.1781,
      "step": 5805
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0316478067069812,
      "learning_rate": 1.186047142840181e-06,
      "loss": 0.1881,
      "step": 5806
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7424675624771958,
      "learning_rate": 1.1846353718724767e-06,
      "loss": 0.121,
      "step": 5807
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9805534255704718,
      "learning_rate": 1.1832243287108314e-06,
      "loss": 0.1982,
      "step": 5808
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8768679564057156,
      "learning_rate": 1.1818140136244117e-06,
      "loss": 0.1815,
      "step": 5809
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3471218375449172,
      "learning_rate": 1.1804044268822434e-06,
      "loss": 0.238,
      "step": 5810
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.061153458809499,
      "learning_rate": 1.1789955687532157e-06,
      "loss": 0.1922,
      "step": 5811
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8534359786571267,
      "learning_rate": 1.1775874395060772e-06,
      "loss": 0.1329,
      "step": 5812
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2165316930712184,
      "learning_rate": 1.1761800394094375e-06,
      "loss": 0.1948,
      "step": 5813
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.847456718772052,
      "learning_rate": 1.174773368731768e-06,
      "loss": 0.1408,
      "step": 5814
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0843506442942221,
      "learning_rate": 1.1733674277413999e-06,
      "loss": 0.1418,
      "step": 5815
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9546108774897045,
      "learning_rate": 1.1719622167065264e-06,
      "loss": 0.1425,
      "step": 5816
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7782252589220415,
      "learning_rate": 1.1705577358951986e-06,
      "loss": 0.1224,
      "step": 5817
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8589205907010459,
      "learning_rate": 1.1691539855753342e-06,
      "loss": 0.1542,
      "step": 5818
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9668300706814611,
      "learning_rate": 1.1677509660147056e-06,
      "loss": 0.1473,
      "step": 5819
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9393512522575863,
      "learning_rate": 1.166348677480949e-06,
      "loss": 0.1563,
      "step": 5820
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.936897749729241,
      "learning_rate": 1.1649471202415597e-06,
      "loss": 0.1199,
      "step": 5821
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0388947885023831,
      "learning_rate": 1.1635462945638954e-06,
      "loss": 0.1965,
      "step": 5822
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0392319371137915,
      "learning_rate": 1.1621462007151696e-06,
      "loss": 0.1484,
      "step": 5823
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8827763754402883,
      "learning_rate": 1.1607468389624622e-06,
      "loss": 0.1505,
      "step": 5824
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6994744090262519,
      "learning_rate": 1.1593482095727094e-06,
      "loss": 0.1209,
      "step": 5825
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8702765341675481,
      "learning_rate": 1.1579503128127096e-06,
      "loss": 0.1316,
      "step": 5826
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8645885320366815,
      "learning_rate": 1.1565531489491205e-06,
      "loss": 0.119,
      "step": 5827
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7333128554572194,
      "learning_rate": 1.1551567182484607e-06,
      "loss": 0.1146,
      "step": 5828
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0211739199650232,
      "learning_rate": 1.1537610209771077e-06,
      "loss": 0.1377,
      "step": 5829
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9688796291815543,
      "learning_rate": 1.1523660574013017e-06,
      "loss": 0.1637,
      "step": 5830
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0455254235878852,
      "learning_rate": 1.1509718277871406e-06,
      "loss": 0.1834,
      "step": 5831
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7890213682387811,
      "learning_rate": 1.1495783324005828e-06,
      "loss": 0.0996,
      "step": 5832
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.797061703669105,
      "learning_rate": 1.1481855715074464e-06,
      "loss": 0.1254,
      "step": 5833
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9324160116756445,
      "learning_rate": 1.1467935453734102e-06,
      "loss": 0.165,
      "step": 5834
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9115642611306656,
      "learning_rate": 1.1454022542640125e-06,
      "loss": 0.1572,
      "step": 5835
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0983802011938328,
      "learning_rate": 1.1440116984446504e-06,
      "loss": 0.1941,
      "step": 5836
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5355451425350636,
      "learning_rate": 1.1426218781805826e-06,
      "loss": 0.0923,
      "step": 5837
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0606176120383233,
      "learning_rate": 1.141232793736926e-06,
      "loss": 0.1724,
      "step": 5838
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1166573879777189,
      "learning_rate": 1.1398444453786594e-06,
      "loss": 0.1935,
      "step": 5839
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1076242587094154,
      "learning_rate": 1.1384568333706152e-06,
      "loss": 0.1604,
      "step": 5840
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0888648718696605,
      "learning_rate": 1.1370699579774934e-06,
      "loss": 0.2105,
      "step": 5841
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0013005892479045,
      "learning_rate": 1.1356838194638486e-06,
      "loss": 0.1435,
      "step": 5842
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.075606894742888,
      "learning_rate": 1.1342984180940953e-06,
      "loss": 0.1687,
      "step": 5843
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9718866403765271,
      "learning_rate": 1.132913754132509e-06,
      "loss": 0.1541,
      "step": 5844
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.749724093568597,
      "learning_rate": 1.131529827843223e-06,
      "loss": 0.1417,
      "step": 5845
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6771991488492745,
      "learning_rate": 1.13014663949023e-06,
      "loss": 0.0922,
      "step": 5846
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0796817778996561,
      "learning_rate": 1.1287641893373829e-06,
      "loss": 0.1916,
      "step": 5847
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.726173741966069,
      "learning_rate": 1.1273824776483932e-06,
      "loss": 0.1432,
      "step": 5848
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9388390568112637,
      "learning_rate": 1.1260015046868311e-06,
      "loss": 0.1456,
      "step": 5849
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9413919800169055,
      "learning_rate": 1.1246212707161268e-06,
      "loss": 0.1318,
      "step": 5850
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8817119779686249,
      "learning_rate": 1.1232417759995678e-06,
      "loss": 0.1349,
      "step": 5851
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1293180362280777,
      "learning_rate": 1.121863020800305e-06,
      "loss": 0.1896,
      "step": 5852
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.262226720657168,
      "learning_rate": 1.120485005381342e-06,
      "loss": 0.1885,
      "step": 5853
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.166752624859448,
      "learning_rate": 1.1191077300055458e-06,
      "loss": 0.2165,
      "step": 5854
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0232143332785448,
      "learning_rate": 1.11773119493564e-06,
      "loss": 0.1357,
      "step": 5855
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5703926415739226,
      "learning_rate": 1.1163554004342082e-06,
      "loss": 0.1258,
      "step": 5856
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0465270136038383,
      "learning_rate": 1.114980346763692e-06,
      "loss": 0.184,
      "step": 5857
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9808148627121593,
      "learning_rate": 1.1136060341863924e-06,
      "loss": 0.1575,
      "step": 5858
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1369571312820632,
      "learning_rate": 1.1122324629644687e-06,
      "loss": 0.1951,
      "step": 5859
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0042673649992575,
      "learning_rate": 1.1108596333599386e-06,
      "loss": 0.1629,
      "step": 5860
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8586640741690875,
      "learning_rate": 1.1094875456346781e-06,
      "loss": 0.139,
      "step": 5861
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8054610799747773,
      "learning_rate": 1.1081162000504208e-06,
      "loss": 0.1522,
      "step": 5862
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9372935816769451,
      "learning_rate": 1.1067455968687625e-06,
      "loss": 0.1893,
      "step": 5863
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9970656284576083,
      "learning_rate": 1.1053757363511541e-06,
      "loss": 0.1957,
      "step": 5864
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.724638824178854,
      "learning_rate": 1.104006618758905e-06,
      "loss": 0.1234,
      "step": 5865
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7474735141634685,
      "learning_rate": 1.1026382443531836e-06,
      "loss": 0.1493,
      "step": 5866
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7718741997508572,
      "learning_rate": 1.1012706133950163e-06,
      "loss": 0.1325,
      "step": 5867
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.093994082709672,
      "learning_rate": 1.0999037261452882e-06,
      "loss": 0.16,
      "step": 5868
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1901160280843768,
      "learning_rate": 1.0985375828647432e-06,
      "loss": 0.177,
      "step": 5869
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9579362085854579,
      "learning_rate": 1.0971721838139788e-06,
      "loss": 0.1598,
      "step": 5870
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.051050825072103,
      "learning_rate": 1.0958075292534558e-06,
      "loss": 0.1979,
      "step": 5871
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9599675471844948,
      "learning_rate": 1.0944436194434916e-06,
      "loss": 0.1576,
      "step": 5872
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1791247441765522,
      "learning_rate": 1.0930804546442602e-06,
      "loss": 0.1826,
      "step": 5873
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7611070493746559,
      "learning_rate": 1.0917180351157936e-06,
      "loss": 0.0984,
      "step": 5874
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0403440106054265,
      "learning_rate": 1.0903563611179847e-06,
      "loss": 0.1859,
      "step": 5875
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2284576628613284,
      "learning_rate": 1.0889954329105801e-06,
      "loss": 0.2379,
      "step": 5876
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7365336118734844,
      "learning_rate": 1.0876352507531867e-06,
      "loss": 0.1547,
      "step": 5877
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.689295928351416,
      "learning_rate": 1.0862758149052678e-06,
      "loss": 0.139,
      "step": 5878
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2735366520603069,
      "learning_rate": 1.084917125626145e-06,
      "loss": 0.1983,
      "step": 5879
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8096862405782672,
      "learning_rate": 1.0835591831749963e-06,
      "loss": 0.1055,
      "step": 5880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8802382470497511,
      "learning_rate": 1.0822019878108597e-06,
      "loss": 0.1514,
      "step": 5881
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0512587878839719,
      "learning_rate": 1.0808455397926282e-06,
      "loss": 0.1651,
      "step": 5882
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9652256285157749,
      "learning_rate": 1.0794898393790537e-06,
      "loss": 0.1262,
      "step": 5883
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9948256762817864,
      "learning_rate": 1.078134886828745e-06,
      "loss": 0.1242,
      "step": 5884
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9571670659813182,
      "learning_rate": 1.076780682400168e-06,
      "loss": 0.1303,
      "step": 5885
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8631679913474917,
      "learning_rate": 1.0754272263516463e-06,
      "loss": 0.1434,
      "step": 5886
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0632765573184144,
      "learning_rate": 1.074074518941361e-06,
      "loss": 0.1716,
      "step": 5887
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8847574006688135,
      "learning_rate": 1.0727225604273489e-06,
      "loss": 0.1817,
      "step": 5888
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6823806825191966,
      "learning_rate": 1.0713713510675062e-06,
      "loss": 0.1168,
      "step": 5889
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8713736328655425,
      "learning_rate": 1.0700208911195841e-06,
      "loss": 0.1435,
      "step": 5890
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7808305836305928,
      "learning_rate": 1.0686711808411925e-06,
      "loss": 0.1265,
      "step": 5891
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9298925152638644,
      "learning_rate": 1.0673222204897977e-06,
      "loss": 0.1511,
      "step": 5892
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.865939371099325,
      "learning_rate": 1.0659740103227217e-06,
      "loss": 0.1461,
      "step": 5893
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0952153052610827,
      "learning_rate": 1.0646265505971458e-06,
      "loss": 0.1847,
      "step": 5894
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8539051939543886,
      "learning_rate": 1.0632798415701062e-06,
      "loss": 0.1204,
      "step": 5895
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9393970675081375,
      "learning_rate": 1.061933883498495e-06,
      "loss": 0.2042,
      "step": 5896
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.929663275644126,
      "learning_rate": 1.060588676639066e-06,
      "loss": 0.1495,
      "step": 5897
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8995945870739792,
      "learning_rate": 1.0592442212484256e-06,
      "loss": 0.1624,
      "step": 5898
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8930882696724365,
      "learning_rate": 1.0579005175830354e-06,
      "loss": 0.1377,
      "step": 5899
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.44548806075305,
      "learning_rate": 1.0565575658992173e-06,
      "loss": 0.2541,
      "step": 5900
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9002046664489659,
      "learning_rate": 1.0552153664531473e-06,
      "loss": 0.1585,
      "step": 5901
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0509285517190072,
      "learning_rate": 1.0538739195008596e-06,
      "loss": 0.1534,
      "step": 5902
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.48370839316025094,
      "learning_rate": 1.0525332252982435e-06,
      "loss": 0.098,
      "step": 5903
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8461935158029521,
      "learning_rate": 1.0511932841010457e-06,
      "loss": 0.172,
      "step": 5904
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7321494660075919,
      "learning_rate": 1.0498540961648695e-06,
      "loss": 0.0812,
      "step": 5905
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.82259411662842,
      "learning_rate": 1.0485156617451725e-06,
      "loss": 0.146,
      "step": 5906
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1718636703013035,
      "learning_rate": 1.0471779810972694e-06,
      "loss": 0.1765,
      "step": 5907
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1565981992614995,
      "learning_rate": 1.0458410544763341e-06,
      "loss": 0.2188,
      "step": 5908
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8697678065476009,
      "learning_rate": 1.0445048821373932e-06,
      "loss": 0.1639,
      "step": 5909
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.07137032071878,
      "learning_rate": 1.0431694643353302e-06,
      "loss": 0.1575,
      "step": 5910
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2525357745215289,
      "learning_rate": 1.0418348013248846e-06,
      "loss": 0.202,
      "step": 5911
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9749801744668483,
      "learning_rate": 1.0405008933606524e-06,
      "loss": 0.1858,
      "step": 5912
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3214456008020057,
      "learning_rate": 1.0391677406970857e-06,
      "loss": 0.1998,
      "step": 5913
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1761249173490917,
      "learning_rate": 1.0378353435884918e-06,
      "loss": 0.2163,
      "step": 5914
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8999369495324827,
      "learning_rate": 1.0365037022890346e-06,
      "loss": 0.1641,
      "step": 5915
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7734013467992044,
      "learning_rate": 1.0351728170527342e-06,
      "loss": 0.1408,
      "step": 5916
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.095319514049018,
      "learning_rate": 1.0338426881334634e-06,
      "loss": 0.2171,
      "step": 5917
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0334668665429942,
      "learning_rate": 1.0325133157849537e-06,
      "loss": 0.1984,
      "step": 5918
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9123324824228264,
      "learning_rate": 1.0311847002607938e-06,
      "loss": 0.1533,
      "step": 5919
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.871751999281278,
      "learning_rate": 1.0298568418144244e-06,
      "loss": 0.125,
      "step": 5920
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9763656885903459,
      "learning_rate": 1.0285297406991434e-06,
      "loss": 0.1817,
      "step": 5921
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.118847523591927,
      "learning_rate": 1.0272033971681045e-06,
      "loss": 0.1997,
      "step": 5922
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9324405264882413,
      "learning_rate": 1.0258778114743163e-06,
      "loss": 0.1324,
      "step": 5923
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0506872172411919,
      "learning_rate": 1.024552983870643e-06,
      "loss": 0.193,
      "step": 5924
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0327103779472484,
      "learning_rate": 1.023228914609804e-06,
      "loss": 0.145,
      "step": 5925
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0348546131623266,
      "learning_rate": 1.0219056039443748e-06,
      "loss": 0.1331,
      "step": 5926
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9242147844393043,
      "learning_rate": 1.0205830521267851e-06,
      "loss": 0.179,
      "step": 5927
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7989863482994736,
      "learning_rate": 1.0192612594093205e-06,
      "loss": 0.1558,
      "step": 5928
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.081107548604219,
      "learning_rate": 1.0179402260441224e-06,
      "loss": 0.1977,
      "step": 5929
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.012388431450819,
      "learning_rate": 1.016619952283186e-06,
      "loss": 0.2156,
      "step": 5930
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8263263223600333,
      "learning_rate": 1.0153004383783621e-06,
      "loss": 0.1231,
      "step": 5931
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0464023091239543,
      "learning_rate": 1.0139816845813573e-06,
      "loss": 0.162,
      "step": 5932
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1072575890339782,
      "learning_rate": 1.0126636911437321e-06,
      "loss": 0.1661,
      "step": 5933
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1096521298101838,
      "learning_rate": 1.0113464583169031e-06,
      "loss": 0.1903,
      "step": 5934
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9530689673724995,
      "learning_rate": 1.0100299863521406e-06,
      "loss": 0.1749,
      "step": 5935
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7976603119298444,
      "learning_rate": 1.0087142755005708e-06,
      "loss": 0.1495,
      "step": 5936
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0052411157667935,
      "learning_rate": 1.0073993260131737e-06,
      "loss": 0.1513,
      "step": 5937
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6758637324252647,
      "learning_rate": 1.006085138140785e-06,
      "loss": 0.1373,
      "step": 5938
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1657958566595465,
      "learning_rate": 1.0047717121340944e-06,
      "loss": 0.1884,
      "step": 5939
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2319650836049016,
      "learning_rate": 1.0034590482436474e-06,
      "loss": 0.1968,
      "step": 5940
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9457062403581286,
      "learning_rate": 1.0021471467198406e-06,
      "loss": 0.1524,
      "step": 5941
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1401946958756688,
      "learning_rate": 1.000836007812932e-06,
      "loss": 0.1955,
      "step": 5942
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0155152845379596,
      "learning_rate": 9.995256317730284e-07,
      "loss": 0.1669,
      "step": 5943
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0280314754634117,
      "learning_rate": 9.982160188500923e-07,
      "loss": 0.1416,
      "step": 5944
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8772743244580399,
      "learning_rate": 9.96907169293942e-07,
      "loss": 0.1892,
      "step": 5945
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0430472486108335,
      "learning_rate": 9.955990833542472e-07,
      "loss": 0.2038,
      "step": 5946
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8851921589838498,
      "learning_rate": 9.942917612805352e-07,
      "loss": 0.15,
      "step": 5947
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2313959716891392,
      "learning_rate": 9.929852033221864e-07,
      "loss": 0.2458,
      "step": 5948
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9237341153444951,
      "learning_rate": 9.91679409728435e-07,
      "loss": 0.145,
      "step": 5949
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7844958534672382,
      "learning_rate": 9.9037438074837e-07,
      "loss": 0.1417,
      "step": 5950
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9033866992450642,
      "learning_rate": 9.890701166309347e-07,
      "loss": 0.1502,
      "step": 5951
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1708792775906616,
      "learning_rate": 9.87766617624924e-07,
      "loss": 0.1863,
      "step": 5952
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9218335127254954,
      "learning_rate": 9.86463883978992e-07,
      "loss": 0.1255,
      "step": 5953
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0717870128131457,
      "learning_rate": 9.851619159416426e-07,
      "loss": 0.1628,
      "step": 5954
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6117279217194304,
      "learning_rate": 9.83860713761235e-07,
      "loss": 0.098,
      "step": 5955
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0357502576409476,
      "learning_rate": 9.825602776859816e-07,
      "loss": 0.1508,
      "step": 5956
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7586910473991906,
      "learning_rate": 9.81260607963949e-07,
      "loss": 0.1167,
      "step": 5957
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9194997930359551,
      "learning_rate": 9.799617048430588e-07,
      "loss": 0.1416,
      "step": 5958
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6296320726318815,
      "learning_rate": 9.786635685710843e-07,
      "loss": 0.0858,
      "step": 5959
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8089870576458185,
      "learning_rate": 9.773661993956546e-07,
      "loss": 0.1304,
      "step": 5960
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0531465613110234,
      "learning_rate": 9.760695975642504e-07,
      "loss": 0.1872,
      "step": 5961
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9159895643701134,
      "learning_rate": 9.747737633242095e-07,
      "loss": 0.1443,
      "step": 5962
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1887358750326367,
      "learning_rate": 9.73478696922716e-07,
      "loss": 0.1531,
      "step": 5963
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9653575900760092,
      "learning_rate": 9.721843986068164e-07,
      "loss": 0.1159,
      "step": 5964
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7982212002593821,
      "learning_rate": 9.708908686234059e-07,
      "loss": 0.1218,
      "step": 5965
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2184388432110285,
      "learning_rate": 9.69598107219234e-07,
      "loss": 0.1196,
      "step": 5966
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0398715248434653,
      "learning_rate": 9.683061146409029e-07,
      "loss": 0.1882,
      "step": 5967
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9097435251157673,
      "learning_rate": 9.670148911348688e-07,
      "loss": 0.1464,
      "step": 5968
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1279791779618704,
      "learning_rate": 9.657244369474411e-07,
      "loss": 0.1893,
      "step": 5969
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0401966208505442,
      "learning_rate": 9.644347523247832e-07,
      "loss": 0.161,
      "step": 5970
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1380259764182514,
      "learning_rate": 9.631458375129099e-07,
      "loss": 0.1644,
      "step": 5971
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7906114423514449,
      "learning_rate": 9.618576927576912e-07,
      "loss": 0.1147,
      "step": 5972
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7975807098120996,
      "learning_rate": 9.605703183048487e-07,
      "loss": 0.1114,
      "step": 5973
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6168480363687389,
      "learning_rate": 9.592837143999578e-07,
      "loss": 0.1314,
      "step": 5974
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8406217354628224,
      "learning_rate": 9.579978812884467e-07,
      "loss": 0.125,
      "step": 5975
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.088301057659492,
      "learning_rate": 9.56712819215596e-07,
      "loss": 0.1901,
      "step": 5976
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1149405278530742,
      "learning_rate": 9.554285284265407e-07,
      "loss": 0.1953,
      "step": 5977
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8372186232089885,
      "learning_rate": 9.54145009166268e-07,
      "loss": 0.1251,
      "step": 5978
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8508182005499986,
      "learning_rate": 9.528622616796163e-07,
      "loss": 0.1551,
      "step": 5979
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.128323103747275,
      "learning_rate": 9.515802862112788e-07,
      "loss": 0.219,
      "step": 5980
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0899465035955778,
      "learning_rate": 9.502990830058017e-07,
      "loss": 0.1919,
      "step": 5981
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0512183090602762,
      "learning_rate": 9.490186523075817e-07,
      "loss": 0.1927,
      "step": 5982
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9028428693703211,
      "learning_rate": 9.477389943608701e-07,
      "loss": 0.1571,
      "step": 5983
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0108706862583792,
      "learning_rate": 9.464601094097703e-07,
      "loss": 0.1656,
      "step": 5984
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0056466644153061,
      "learning_rate": 9.451819976982374e-07,
      "loss": 0.1746,
      "step": 5985
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2031474929104626,
      "learning_rate": 9.439046594700791e-07,
      "loss": 0.2262,
      "step": 5986
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.143835149813184,
      "learning_rate": 9.426280949689581e-07,
      "loss": 0.2137,
      "step": 5987
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0947492912729504,
      "learning_rate": 9.413523044383865e-07,
      "loss": 0.1771,
      "step": 5988
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5427639025354125,
      "learning_rate": 9.400772881217296e-07,
      "loss": 0.0708,
      "step": 5989
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9418220585329525,
      "learning_rate": 9.388030462622056e-07,
      "loss": 0.1392,
      "step": 5990
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9917718682158934,
      "learning_rate": 9.375295791028843e-07,
      "loss": 0.1629,
      "step": 5991
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9758608237657274,
      "learning_rate": 9.362568868866895e-07,
      "loss": 0.1607,
      "step": 5992
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0069979060852614,
      "learning_rate": 9.349849698563928e-07,
      "loss": 0.1509,
      "step": 5993
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1786753778286148,
      "learning_rate": 9.337138282546227e-07,
      "loss": 0.2221,
      "step": 5994
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8831466303549397,
      "learning_rate": 9.324434623238571e-07,
      "loss": 0.144,
      "step": 5995
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1357640233587116,
      "learning_rate": 9.311738723064267e-07,
      "loss": 0.1801,
      "step": 5996
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1780040419174371,
      "learning_rate": 9.299050584445135e-07,
      "loss": 0.1619,
      "step": 5997
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0401737869788859,
      "learning_rate": 9.286370209801543e-07,
      "loss": 0.1965,
      "step": 5998
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0747886858354856,
      "learning_rate": 9.273697601552345e-07,
      "loss": 0.1501,
      "step": 5999
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9145389321769791,
      "learning_rate": 9.261032762114924e-07,
      "loss": 0.1534,
      "step": 6000
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8117993626912174,
      "learning_rate": 9.248375693905181e-07,
      "loss": 0.1692,
      "step": 6001
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6575260591928157,
      "learning_rate": 9.235726399337541e-07,
      "loss": 0.0888,
      "step": 6002
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9831596207448686,
      "learning_rate": 9.223084880824934e-07,
      "loss": 0.1754,
      "step": 6003
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.906032581853624,
      "learning_rate": 9.21045114077882e-07,
      "loss": 0.1452,
      "step": 6004
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1388197118748735,
      "learning_rate": 9.197825181609165e-07,
      "loss": 0.1988,
      "step": 6005
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0983003158921338,
      "learning_rate": 9.18520700572445e-07,
      "loss": 0.1975,
      "step": 6006
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1249351209216267,
      "learning_rate": 9.172596615531682e-07,
      "loss": 0.1552,
      "step": 6007
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.072436030736748,
      "learning_rate": 9.159994013436374e-07,
      "loss": 0.2087,
      "step": 6008
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8208199257685567,
      "learning_rate": 9.147399201842555e-07,
      "loss": 0.1427,
      "step": 6009
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9136967844875316,
      "learning_rate": 9.13481218315277e-07,
      "loss": 0.1484,
      "step": 6010
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8049464742644323,
      "learning_rate": 9.122232959768073e-07,
      "loss": 0.1074,
      "step": 6011
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9644688184353029,
      "learning_rate": 9.109661534088043e-07,
      "loss": 0.168,
      "step": 6012
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8246833775525004,
      "learning_rate": 9.097097908510749e-07,
      "loss": 0.1481,
      "step": 6013
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7427663218923876,
      "learning_rate": 9.084542085432796e-07,
      "loss": 0.1755,
      "step": 6014
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.837391528701616,
      "learning_rate": 9.071994067249291e-07,
      "loss": 0.1485,
      "step": 6015
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1615355968376897,
      "learning_rate": 9.059453856353845e-07,
      "loss": 0.1944,
      "step": 6016
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6068571132023078,
      "learning_rate": 9.046921455138591e-07,
      "loss": 0.108,
      "step": 6017
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9292985067367193,
      "learning_rate": 9.034396865994166e-07,
      "loss": 0.1445,
      "step": 6018
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1158968398003324,
      "learning_rate": 9.021880091309704e-07,
      "loss": 0.193,
      "step": 6019
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.199284568299364,
      "learning_rate": 9.009371133472889e-07,
      "loss": 0.235,
      "step": 6020
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9661216131053245,
      "learning_rate": 8.996869994869878e-07,
      "loss": 0.1574,
      "step": 6021
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7761702333638768,
      "learning_rate": 8.984376677885354e-07,
      "loss": 0.1404,
      "step": 6022
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0388808096590618,
      "learning_rate": 8.971891184902476e-07,
      "loss": 0.1759,
      "step": 6023
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.081094005800363,
      "learning_rate": 8.959413518302946e-07,
      "loss": 0.1831,
      "step": 6024
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9353341474122308,
      "learning_rate": 8.946943680466968e-07,
      "loss": 0.1453,
      "step": 6025
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0189121123353349,
      "learning_rate": 8.934481673773238e-07,
      "loss": 0.1448,
      "step": 6026
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9553710078736047,
      "learning_rate": 8.922027500598968e-07,
      "loss": 0.185,
      "step": 6027
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7937125913821367,
      "learning_rate": 8.909581163319875e-07,
      "loss": 0.1373,
      "step": 6028
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8890056266341019,
      "learning_rate": 8.897142664310182e-07,
      "loss": 0.1429,
      "step": 6029
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0783189686827472,
      "learning_rate": 8.88471200594262e-07,
      "loss": 0.196,
      "step": 6030
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2098771319006716,
      "learning_rate": 8.87228919058839e-07,
      "loss": 0.2018,
      "step": 6031
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8452362668315772,
      "learning_rate": 8.859874220617271e-07,
      "loss": 0.1412,
      "step": 6032
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0264610868869166,
      "learning_rate": 8.847467098397472e-07,
      "loss": 0.1855,
      "step": 6033
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6284670447837418,
      "learning_rate": 8.835067826295745e-07,
      "loss": 0.1559,
      "step": 6034
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.367341644300006,
      "learning_rate": 8.822676406677327e-07,
      "loss": 0.2488,
      "step": 6035
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8636221745621318,
      "learning_rate": 8.810292841905965e-07,
      "loss": 0.173,
      "step": 6036
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.928057520413017,
      "learning_rate": 8.797917134343908e-07,
      "loss": 0.1349,
      "step": 6037
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0362616214390674,
      "learning_rate": 8.785549286351902e-07,
      "loss": 0.2107,
      "step": 6038
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9818898815883779,
      "learning_rate": 8.773189300289209e-07,
      "loss": 0.1809,
      "step": 6039
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.688627087503231,
      "learning_rate": 8.760837178513553e-07,
      "loss": 0.1118,
      "step": 6040
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8972340765532953,
      "learning_rate": 8.748492923381191e-07,
      "loss": 0.1707,
      "step": 6041
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.986692203492362,
      "learning_rate": 8.73615653724687e-07,
      "loss": 0.1628,
      "step": 6042
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1985711830494263,
      "learning_rate": 8.723828022463848e-07,
      "loss": 0.2152,
      "step": 6043
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1709379139914136,
      "learning_rate": 8.711507381383871e-07,
      "loss": 0.1972,
      "step": 6044
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.79424658206911,
      "learning_rate": 8.699194616357181e-07,
      "loss": 0.1562,
      "step": 6045
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1521830522061658,
      "learning_rate": 8.686889729732512e-07,
      "loss": 0.2314,
      "step": 6046
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.971942651330073,
      "learning_rate": 8.674592723857106e-07,
      "loss": 0.2042,
      "step": 6047
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9735097871994244,
      "learning_rate": 8.662303601076699e-07,
      "loss": 0.1763,
      "step": 6048
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0252353300450494,
      "learning_rate": 8.650022363735522e-07,
      "loss": 0.1879,
      "step": 6049
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1929219458673261,
      "learning_rate": 8.637749014176305e-07,
      "loss": 0.1968,
      "step": 6050
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0087396437745628,
      "learning_rate": 8.625483554740271e-07,
      "loss": 0.2034,
      "step": 6051
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.326024696037896,
      "learning_rate": 8.613225987767132e-07,
      "loss": 0.2394,
      "step": 6052
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0686379923772296,
      "learning_rate": 8.6009763155951e-07,
      "loss": 0.1788,
      "step": 6053
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1486655032023105,
      "learning_rate": 8.588734540560889e-07,
      "loss": 0.2155,
      "step": 6054
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.847843453150639,
      "learning_rate": 8.576500664999693e-07,
      "loss": 0.0953,
      "step": 6055
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0656747963875648,
      "learning_rate": 8.564274691245206e-07,
      "loss": 0.1665,
      "step": 6056
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7739511703405263,
      "learning_rate": 8.552056621629612e-07,
      "loss": 0.1648,
      "step": 6057
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1450374250871949,
      "learning_rate": 8.539846458483586e-07,
      "loss": 0.1632,
      "step": 6058
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9374580410773327,
      "learning_rate": 8.52764420413631e-07,
      "loss": 0.1163,
      "step": 6059
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.199420704994289,
      "learning_rate": 8.515449860915426e-07,
      "loss": 0.2235,
      "step": 6060
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8715506688445197,
      "learning_rate": 8.503263431147102e-07,
      "loss": 0.1076,
      "step": 6061
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6935852907538813,
      "learning_rate": 8.491084917155973e-07,
      "loss": 0.1312,
      "step": 6062
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7555407117982189,
      "learning_rate": 8.478914321265169e-07,
      "loss": 0.1007,
      "step": 6063
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7038446619208261,
      "learning_rate": 8.466751645796306e-07,
      "loss": 0.0927,
      "step": 6064
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9356359035909217,
      "learning_rate": 8.454596893069517e-07,
      "loss": 0.1583,
      "step": 6065
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8853117175657235,
      "learning_rate": 8.442450065403385e-07,
      "loss": 0.1359,
      "step": 6066
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8808071848302962,
      "learning_rate": 8.430311165115001e-07,
      "loss": 0.1414,
      "step": 6067
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0730419107662135,
      "learning_rate": 8.418180194519954e-07,
      "loss": 0.1918,
      "step": 6068
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8580860712398342,
      "learning_rate": 8.406057155932279e-07,
      "loss": 0.1333,
      "step": 6069
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9505086751310611,
      "learning_rate": 8.393942051664538e-07,
      "loss": 0.1317,
      "step": 6070
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0779845594121606,
      "learning_rate": 8.381834884027773e-07,
      "loss": 0.2159,
      "step": 6071
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8998786541701671,
      "learning_rate": 8.369735655331507e-07,
      "loss": 0.1472,
      "step": 6072
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8632486980869826,
      "learning_rate": 8.357644367883739e-07,
      "loss": 0.1189,
      "step": 6073
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7678804756735101,
      "learning_rate": 8.345561023990966e-07,
      "loss": 0.142,
      "step": 6074
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8473279421571975,
      "learning_rate": 8.333485625958171e-07,
      "loss": 0.1106,
      "step": 6075
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8652640139008787,
      "learning_rate": 8.321418176088797e-07,
      "loss": 0.1338,
      "step": 6076
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6656726551143234,
      "learning_rate": 8.309358676684814e-07,
      "loss": 0.1039,
      "step": 6077
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2189564362254839,
      "learning_rate": 8.297307130046645e-07,
      "loss": 0.1975,
      "step": 6078
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.392876126457686,
      "learning_rate": 8.285263538473204e-07,
      "loss": 0.2822,
      "step": 6079
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1148474754891424,
      "learning_rate": 8.273227904261877e-07,
      "loss": 0.1798,
      "step": 6080
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0091922837294607,
      "learning_rate": 8.261200229708544e-07,
      "loss": 0.1655,
      "step": 6081
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9848769370174782,
      "learning_rate": 8.249180517107569e-07,
      "loss": 0.1746,
      "step": 6082
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0842552077356358,
      "learning_rate": 8.237168768751785e-07,
      "loss": 0.1919,
      "step": 6083
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.024480916926148,
      "learning_rate": 8.225164986932516e-07,
      "loss": 0.1486,
      "step": 6084
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.935971888305995,
      "learning_rate": 8.213169173939573e-07,
      "loss": 0.1847,
      "step": 6085
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9871876095384997,
      "learning_rate": 8.201181332061215e-07,
      "loss": 0.1432,
      "step": 6086
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.166746313936605,
      "learning_rate": 8.189201463584196e-07,
      "loss": 0.1591,
      "step": 6087
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8784183782724323,
      "learning_rate": 8.177229570793788e-07,
      "loss": 0.1488,
      "step": 6088
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0144954018961603,
      "learning_rate": 8.165265655973697e-07,
      "loss": 0.1371,
      "step": 6089
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9996278697781205,
      "learning_rate": 8.15330972140611e-07,
      "loss": 0.1857,
      "step": 6090
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9981021075062493,
      "learning_rate": 8.141361769371703e-07,
      "loss": 0.1326,
      "step": 6091
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9930793619480366,
      "learning_rate": 8.129421802149634e-07,
      "loss": 0.1597,
      "step": 6092
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3024659325195187,
      "learning_rate": 8.117489822017527e-07,
      "loss": 0.1929,
      "step": 6093
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1452659930659137,
      "learning_rate": 8.105565831251482e-07,
      "loss": 0.2121,
      "step": 6094
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8207914517503294,
      "learning_rate": 8.093649832126083e-07,
      "loss": 0.1186,
      "step": 6095
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0471114140576592,
      "learning_rate": 8.081741826914386e-07,
      "loss": 0.1858,
      "step": 6096
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7785130892873576,
      "learning_rate": 8.069841817887919e-07,
      "loss": 0.1183,
      "step": 6097
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8631191930257687,
      "learning_rate": 8.057949807316695e-07,
      "loss": 0.1654,
      "step": 6098
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9744288796597252,
      "learning_rate": 8.046065797469182e-07,
      "loss": 0.1897,
      "step": 6099
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9666763669119235,
      "learning_rate": 8.034189790612346e-07,
      "loss": 0.1834,
      "step": 6100
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9151940053568564,
      "learning_rate": 8.022321789011605e-07,
      "loss": 0.1208,
      "step": 6101
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7334484948256751,
      "learning_rate": 8.010461794930863e-07,
      "loss": 0.1104,
      "step": 6102
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9941851564493654,
      "learning_rate": 7.998609810632485e-07,
      "loss": 0.2032,
      "step": 6103
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8842185988292872,
      "learning_rate": 7.986765838377331e-07,
      "loss": 0.1784,
      "step": 6104
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1003159940142904,
      "learning_rate": 7.974929880424703e-07,
      "loss": 0.1909,
      "step": 6105
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2771901956935943,
      "learning_rate": 7.963101939032391e-07,
      "loss": 0.2117,
      "step": 6106
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7663633519301061,
      "learning_rate": 7.951282016456657e-07,
      "loss": 0.1441,
      "step": 6107
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9939968667576631,
      "learning_rate": 7.939470114952225e-07,
      "loss": 0.1883,
      "step": 6108
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8517936616019448,
      "learning_rate": 7.927666236772286e-07,
      "loss": 0.1528,
      "step": 6109
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2041061831995503,
      "learning_rate": 7.915870384168534e-07,
      "loss": 0.2086,
      "step": 6110
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8813208809662278,
      "learning_rate": 7.904082559391086e-07,
      "loss": 0.1318,
      "step": 6111
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9515325698442083,
      "learning_rate": 7.892302764688548e-07,
      "loss": 0.1601,
      "step": 6112
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8020032944557562,
      "learning_rate": 7.880531002308001e-07,
      "loss": 0.1633,
      "step": 6113
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9781906321162119,
      "learning_rate": 7.868767274494982e-07,
      "loss": 0.19,
      "step": 6114
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8019726629303815,
      "learning_rate": 7.857011583493518e-07,
      "loss": 0.1132,
      "step": 6115
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.97672221267092,
      "learning_rate": 7.845263931546049e-07,
      "loss": 0.1703,
      "step": 6116
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2286300500010932,
      "learning_rate": 7.833524320893537e-07,
      "loss": 0.192,
      "step": 6117
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.524333545120802,
      "learning_rate": 7.821792753775392e-07,
      "loss": 0.0914,
      "step": 6118
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1153487806857514,
      "learning_rate": 7.810069232429485e-07,
      "loss": 0.1893,
      "step": 6119
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0866797283985257,
      "learning_rate": 7.798353759092142e-07,
      "loss": 0.1719,
      "step": 6120
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9661404638720513,
      "learning_rate": 7.786646335998194e-07,
      "loss": 0.1663,
      "step": 6121
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9541699813682232,
      "learning_rate": 7.774946965380897e-07,
      "loss": 0.2175,
      "step": 6122
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.101808257978215,
      "learning_rate": 7.763255649471985e-07,
      "loss": 0.1846,
      "step": 6123
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0966982650458454,
      "learning_rate": 7.751572390501649e-07,
      "loss": 0.2146,
      "step": 6124
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0037108484619477,
      "learning_rate": 7.739897190698548e-07,
      "loss": 0.1446,
      "step": 6125
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8039800744948091,
      "learning_rate": 7.728230052289809e-07,
      "loss": 0.1198,
      "step": 6126
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9289607026068593,
      "learning_rate": 7.716570977501014e-07,
      "loss": 0.1515,
      "step": 6127
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.875895492993034,
      "learning_rate": 7.704919968556207e-07,
      "loss": 0.136,
      "step": 6128
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7788580175570488,
      "learning_rate": 7.693277027677898e-07,
      "loss": 0.1069,
      "step": 6129
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8790970746297723,
      "learning_rate": 7.68164215708705e-07,
      "loss": 0.142,
      "step": 6130
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8027184160123236,
      "learning_rate": 7.670015359003097e-07,
      "loss": 0.1402,
      "step": 6131
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1434865960537655,
      "learning_rate": 7.658396635643928e-07,
      "loss": 0.1783,
      "step": 6132
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0951267339067878,
      "learning_rate": 7.646785989225885e-07,
      "loss": 0.1855,
      "step": 6133
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0436976133548708,
      "learning_rate": 7.635183421963776e-07,
      "loss": 0.1734,
      "step": 6134
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.93164380967608,
      "learning_rate": 7.623588936070875e-07,
      "loss": 0.1382,
      "step": 6135
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.977981682286933,
      "learning_rate": 7.612002533758905e-07,
      "loss": 0.1555,
      "step": 6136
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9108394844083066,
      "learning_rate": 7.600424217238044e-07,
      "loss": 0.1349,
      "step": 6137
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7923254354344753,
      "learning_rate": 7.588853988716938e-07,
      "loss": 0.1191,
      "step": 6138
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7607966254961421,
      "learning_rate": 7.577291850402679e-07,
      "loss": 0.1068,
      "step": 6139
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1317271882017008,
      "learning_rate": 7.565737804500822e-07,
      "loss": 0.2225,
      "step": 6140
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.955854508839649,
      "learning_rate": 7.554191853215387e-07,
      "loss": 0.1721,
      "step": 6141
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9769691646902561,
      "learning_rate": 7.542653998748828e-07,
      "loss": 0.1724,
      "step": 6142
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.012325505636287,
      "learning_rate": 7.531124243302063e-07,
      "loss": 0.192,
      "step": 6143
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8752783373646479,
      "learning_rate": 7.519602589074493e-07,
      "loss": 0.1547,
      "step": 6144
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7807791017930841,
      "learning_rate": 7.508089038263943e-07,
      "loss": 0.1563,
      "step": 6145
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8970727027800284,
      "learning_rate": 7.496583593066686e-07,
      "loss": 0.1246,
      "step": 6146
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0242372005485993,
      "learning_rate": 7.485086255677466e-07,
      "loss": 0.1616,
      "step": 6147
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.18291011244177,
      "learning_rate": 7.473597028289475e-07,
      "loss": 0.2122,
      "step": 6148
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8516411687887598,
      "learning_rate": 7.462115913094364e-07,
      "loss": 0.1447,
      "step": 6149
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.889852145240436,
      "learning_rate": 7.45064291228223e-07,
      "loss": 0.1174,
      "step": 6150
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3418472995115502,
      "learning_rate": 7.439178028041621e-07,
      "loss": 0.2159,
      "step": 6151
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9582136359151241,
      "learning_rate": 7.42772126255955e-07,
      "loss": 0.1483,
      "step": 6152
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0356926047108834,
      "learning_rate": 7.416272618021458e-07,
      "loss": 0.2031,
      "step": 6153
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8634446375967592,
      "learning_rate": 7.404832096611242e-07,
      "loss": 0.1093,
      "step": 6154
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9902102049797653,
      "learning_rate": 7.393399700511284e-07,
      "loss": 0.1756,
      "step": 6155
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9045392286385434,
      "learning_rate": 7.381975431902372e-07,
      "loss": 0.1621,
      "step": 6156
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7515384207449244,
      "learning_rate": 7.37055929296377e-07,
      "loss": 0.1225,
      "step": 6157
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9286710664969411,
      "learning_rate": 7.359151285873172e-07,
      "loss": 0.1488,
      "step": 6158
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2092210606301028,
      "learning_rate": 7.347751412806737e-07,
      "loss": 0.2002,
      "step": 6159
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2041092978753376,
      "learning_rate": 7.336359675939064e-07,
      "loss": 0.2409,
      "step": 6160
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9695548839892117,
      "learning_rate": 7.324976077443202e-07,
      "loss": 0.164,
      "step": 6161
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0962321836794104,
      "learning_rate": 7.313600619490658e-07,
      "loss": 0.1938,
      "step": 6162
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9039609356037683,
      "learning_rate": 7.302233304251355e-07,
      "loss": 0.1425,
      "step": 6163
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9186361094846158,
      "learning_rate": 7.290874133893699e-07,
      "loss": 0.1445,
      "step": 6164
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0289703198061928,
      "learning_rate": 7.279523110584507e-07,
      "loss": 0.1513,
      "step": 6165
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0503353176811245,
      "learning_rate": 7.268180236489092e-07,
      "loss": 0.1652,
      "step": 6166
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8481553731394315,
      "learning_rate": 7.256845513771171e-07,
      "loss": 0.1416,
      "step": 6167
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9663251773634434,
      "learning_rate": 7.245518944592911e-07,
      "loss": 0.1581,
      "step": 6168
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9967230226747626,
      "learning_rate": 7.234200531114932e-07,
      "loss": 0.1622,
      "step": 6169
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.063421278436821,
      "learning_rate": 7.222890275496297e-07,
      "loss": 0.1656,
      "step": 6170
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6864006213515101,
      "learning_rate": 7.211588179894514e-07,
      "loss": 0.107,
      "step": 6171
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8462052004809886,
      "learning_rate": 7.200294246465534e-07,
      "loss": 0.1404,
      "step": 6172
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1924002487132512,
      "learning_rate": 7.189008477363746e-07,
      "loss": 0.1888,
      "step": 6173
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8906080676196526,
      "learning_rate": 7.177730874741984e-07,
      "loss": 0.1032,
      "step": 6174
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0183542846084526,
      "learning_rate": 7.166461440751526e-07,
      "loss": 0.1823,
      "step": 6175
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0056610724973303,
      "learning_rate": 7.155200177542098e-07,
      "loss": 0.1564,
      "step": 6176
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1340573619595236,
      "learning_rate": 7.143947087261854e-07,
      "loss": 0.1539,
      "step": 6177
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0043081449014173,
      "learning_rate": 7.132702172057393e-07,
      "loss": 0.2057,
      "step": 6178
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0782362356673272,
      "learning_rate": 7.121465434073766e-07,
      "loss": 0.1934,
      "step": 6179
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.239703825508989,
      "learning_rate": 7.110236875454446e-07,
      "loss": 0.1769,
      "step": 6180
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1267393632533662,
      "learning_rate": 7.09901649834136e-07,
      "loss": 0.1811,
      "step": 6181
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1339422132580366,
      "learning_rate": 7.087804304874863e-07,
      "loss": 0.1823,
      "step": 6182
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8173736291983926,
      "learning_rate": 7.076600297193764e-07,
      "loss": 0.1114,
      "step": 6183
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9485509978685372,
      "learning_rate": 7.065404477435301e-07,
      "loss": 0.1629,
      "step": 6184
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0898208480065668,
      "learning_rate": 7.054216847735146e-07,
      "loss": 0.186,
      "step": 6185
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.054525444735243,
      "learning_rate": 7.04303741022741e-07,
      "loss": 0.1767,
      "step": 6186
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9479647902711232,
      "learning_rate": 7.031866167044654e-07,
      "loss": 0.1395,
      "step": 6187
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8189701659694909,
      "learning_rate": 7.020703120317846e-07,
      "loss": 0.1154,
      "step": 6188
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0171741657423694,
      "learning_rate": 7.009548272176441e-07,
      "loss": 0.1377,
      "step": 6189
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8980623346236254,
      "learning_rate": 6.998401624748292e-07,
      "loss": 0.1325,
      "step": 6190
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9325814429803563,
      "learning_rate": 6.987263180159693e-07,
      "loss": 0.1106,
      "step": 6191
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0614430289371295,
      "learning_rate": 6.976132940535362e-07,
      "loss": 0.1902,
      "step": 6192
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9895215006617977,
      "learning_rate": 6.965010907998482e-07,
      "loss": 0.1477,
      "step": 6193
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9980111511448247,
      "learning_rate": 6.953897084670646e-07,
      "loss": 0.185,
      "step": 6194
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0103768602522145,
      "learning_rate": 6.942791472671895e-07,
      "loss": 0.1558,
      "step": 6195
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0260247126069824,
      "learning_rate": 6.931694074120699e-07,
      "loss": 0.1549,
      "step": 6196
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2750499296370776,
      "learning_rate": 6.920604891133948e-07,
      "loss": 0.2001,
      "step": 6197
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7716639438810377,
      "learning_rate": 6.909523925826994e-07,
      "loss": 0.1556,
      "step": 6198
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.986984242927735,
      "learning_rate": 6.898451180313581e-07,
      "loss": 0.189,
      "step": 6199
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.039981029488902,
      "learning_rate": 6.887386656705936e-07,
      "loss": 0.146,
      "step": 6200
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.995503247000664,
      "learning_rate": 6.876330357114674e-07,
      "loss": 0.1714,
      "step": 6201
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0476302590492692,
      "learning_rate": 6.865282283648867e-07,
      "loss": 0.1804,
      "step": 6202
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9254521656831755,
      "learning_rate": 6.854242438416003e-07,
      "loss": 0.1224,
      "step": 6203
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.117170914798823,
      "learning_rate": 6.843210823522001e-07,
      "loss": 0.1751,
      "step": 6204
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6020759506776878,
      "learning_rate": 6.832187441071226e-07,
      "loss": 0.1306,
      "step": 6205
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1204147055412714,
      "learning_rate": 6.821172293166445e-07,
      "loss": 0.1954,
      "step": 6206
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.884279666031796,
      "learning_rate": 6.810165381908884e-07,
      "loss": 0.1309,
      "step": 6207
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9468897272548826,
      "learning_rate": 6.799166709398175e-07,
      "loss": 0.144,
      "step": 6208
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9232994126549895,
      "learning_rate": 6.788176277732411e-07,
      "loss": 0.1381,
      "step": 6209
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9256460077272873,
      "learning_rate": 6.777194089008044e-07,
      "loss": 0.169,
      "step": 6210
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6835932232695976,
      "learning_rate": 6.766220145320035e-07,
      "loss": 0.1276,
      "step": 6211
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6643463968263651,
      "learning_rate": 6.755254448761728e-07,
      "loss": 0.0987,
      "step": 6212
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0260826350023649,
      "learning_rate": 6.744297001424904e-07,
      "loss": 0.1758,
      "step": 6213
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1132736379805734,
      "learning_rate": 6.733347805399764e-07,
      "loss": 0.1673,
      "step": 6214
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.846502914282829,
      "learning_rate": 6.722406862774944e-07,
      "loss": 0.117,
      "step": 6215
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7918961037359178,
      "learning_rate": 6.711474175637494e-07,
      "loss": 0.1103,
      "step": 6216
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8496341652328424,
      "learning_rate": 6.700549746072904e-07,
      "loss": 0.1581,
      "step": 6217
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0124500916911516,
      "learning_rate": 6.689633576165083e-07,
      "loss": 0.1751,
      "step": 6218
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1384272017395163,
      "learning_rate": 6.67872566799635e-07,
      "loss": 0.1819,
      "step": 6219
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.00706125606111,
      "learning_rate": 6.667826023647472e-07,
      "loss": 0.1687,
      "step": 6220
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.893069406294713,
      "learning_rate": 6.656934645197626e-07,
      "loss": 0.1356,
      "step": 6221
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9609383114905601,
      "learning_rate": 6.646051534724419e-07,
      "loss": 0.1847,
      "step": 6222
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8574887946637828,
      "learning_rate": 6.635176694303864e-07,
      "loss": 0.1283,
      "step": 6223
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8103146289707231,
      "learning_rate": 6.624310126010419e-07,
      "loss": 0.1464,
      "step": 6224
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9782819370412615,
      "learning_rate": 6.61345183191695e-07,
      "loss": 0.1663,
      "step": 6225
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9510302304816143,
      "learning_rate": 6.60260181409475e-07,
      "loss": 0.1836,
      "step": 6226
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9990670342413697,
      "learning_rate": 6.591760074613529e-07,
      "loss": 0.1793,
      "step": 6227
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9346110205167386,
      "learning_rate": 6.580926615541428e-07,
      "loss": 0.1393,
      "step": 6228
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0419359967959136,
      "learning_rate": 6.570101438944987e-07,
      "loss": 0.175,
      "step": 6229
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7935254320390988,
      "learning_rate": 6.559284546889195e-07,
      "loss": 0.137,
      "step": 6230
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.042052888614404,
      "learning_rate": 6.548475941437437e-07,
      "loss": 0.1598,
      "step": 6231
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0181185555045325,
      "learning_rate": 6.53767562465153e-07,
      "loss": 0.1847,
      "step": 6232
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0246369637247872,
      "learning_rate": 6.526883598591694e-07,
      "loss": 0.1807,
      "step": 6233
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8926859957377062,
      "learning_rate": 6.5160998653166e-07,
      "loss": 0.1595,
      "step": 6234
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1756236492360923,
      "learning_rate": 6.505324426883303e-07,
      "loss": 0.1811,
      "step": 6235
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9200101837445687,
      "learning_rate": 6.494557285347297e-07,
      "loss": 0.1484,
      "step": 6236
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9308989064059969,
      "learning_rate": 6.483798442762479e-07,
      "loss": 0.1442,
      "step": 6237
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8566587174439826,
      "learning_rate": 6.473047901181185e-07,
      "loss": 0.1576,
      "step": 6238
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9387831528858295,
      "learning_rate": 6.462305662654122e-07,
      "loss": 0.1444,
      "step": 6239
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8118613276068842,
      "learning_rate": 6.451571729230466e-07,
      "loss": 0.1052,
      "step": 6240
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4421220022238005,
      "learning_rate": 6.440846102957776e-07,
      "loss": 0.1098,
      "step": 6241
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1267728512464357,
      "learning_rate": 6.43012878588204e-07,
      "loss": 0.1638,
      "step": 6242
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1094691275571125,
      "learning_rate": 6.419419780047659e-07,
      "loss": 0.1792,
      "step": 6243
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8874806944231751,
      "learning_rate": 6.408719087497428e-07,
      "loss": 0.1331,
      "step": 6244
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0030357743830205,
      "learning_rate": 6.398026710272609e-07,
      "loss": 0.1877,
      "step": 6245
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9597515717356001,
      "learning_rate": 6.387342650412826e-07,
      "loss": 0.1946,
      "step": 6246
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0583127357865145,
      "learning_rate": 6.37666690995613e-07,
      "loss": 0.1858,
      "step": 6247
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.889642739754243,
      "learning_rate": 6.365999490938995e-07,
      "loss": 0.1508,
      "step": 6248
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1653478679267737,
      "learning_rate": 6.355340395396303e-07,
      "loss": 0.1813,
      "step": 6249
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.895024954669452,
      "learning_rate": 6.344689625361339e-07,
      "loss": 0.1411,
      "step": 6250
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6960088724941282,
      "learning_rate": 6.334047182865815e-07,
      "loss": 0.1064,
      "step": 6251
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2241596519273352,
      "learning_rate": 6.323413069939849e-07,
      "loss": 0.2146,
      "step": 6252
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9023152636935614,
      "learning_rate": 6.312787288611965e-07,
      "loss": 0.1511,
      "step": 6253
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9221554338034569,
      "learning_rate": 6.3021698409091e-07,
      "loss": 0.1563,
      "step": 6254
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1332075820223064,
      "learning_rate": 6.291560728856599e-07,
      "loss": 0.1802,
      "step": 6255
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1005015940548575,
      "learning_rate": 6.280959954478233e-07,
      "loss": 0.1751,
      "step": 6256
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8862430180137111,
      "learning_rate": 6.270367519796155e-07,
      "loss": 0.1874,
      "step": 6257
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9403977681618588,
      "learning_rate": 6.259783426830957e-07,
      "loss": 0.1813,
      "step": 6258
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9010419636236114,
      "learning_rate": 6.24920767760161e-07,
      "loss": 0.1196,
      "step": 6259
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6743773773469434,
      "learning_rate": 6.238640274125518e-07,
      "loss": 0.109,
      "step": 6260
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7771613115728727,
      "learning_rate": 6.228081218418474e-07,
      "loss": 0.1655,
      "step": 6261
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.857793526083911,
      "learning_rate": 6.217530512494701e-07,
      "loss": 0.1289,
      "step": 6262
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.027098069717295,
      "learning_rate": 6.206988158366806e-07,
      "loss": 0.1521,
      "step": 6263
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8402333539093781,
      "learning_rate": 6.196454158045817e-07,
      "loss": 0.1219,
      "step": 6264
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9563946627141791,
      "learning_rate": 6.185928513541162e-07,
      "loss": 0.1928,
      "step": 6265
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7819378441519622,
      "learning_rate": 6.175411226860667e-07,
      "loss": 0.1372,
      "step": 6266
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.037719288637647,
      "learning_rate": 6.164902300010595e-07,
      "loss": 0.159,
      "step": 6267
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.100890765668324,
      "learning_rate": 6.154401734995596e-07,
      "loss": 0.2009,
      "step": 6268
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7844625497337301,
      "learning_rate": 6.143909533818704e-07,
      "loss": 0.1139,
      "step": 6269
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0762121409715362,
      "learning_rate": 6.133425698481377e-07,
      "loss": 0.1401,
      "step": 6270
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8591322116916041,
      "learning_rate": 6.122950230983476e-07,
      "loss": 0.1248,
      "step": 6271
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9848503867262209,
      "learning_rate": 6.112483133323277e-07,
      "loss": 0.1401,
      "step": 6272
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.885548882361572,
      "learning_rate": 6.102024407497442e-07,
      "loss": 0.1706,
      "step": 6273
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8009080176893196,
      "learning_rate": 6.091574055501043e-07,
      "loss": 0.124,
      "step": 6274
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8432006434097415,
      "learning_rate": 6.081132079327545e-07,
      "loss": 0.1424,
      "step": 6275
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7468176024959775,
      "learning_rate": 6.070698480968839e-07,
      "loss": 0.1247,
      "step": 6276
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7423936013498135,
      "learning_rate": 6.060273262415195e-07,
      "loss": 0.1114,
      "step": 6277
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9555201924686546,
      "learning_rate": 6.049856425655282e-07,
      "loss": 0.1567,
      "step": 6278
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6642573448764376,
      "learning_rate": 6.039447972676204e-07,
      "loss": 0.0983,
      "step": 6279
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0109414599524835,
      "learning_rate": 6.029047905463426e-07,
      "loss": 0.1692,
      "step": 6280
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7414731598576992,
      "learning_rate": 6.018656226000835e-07,
      "loss": 0.1432,
      "step": 6281
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9406293911704788,
      "learning_rate": 6.008272936270714e-07,
      "loss": 0.1514,
      "step": 6282
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.126949084429464,
      "learning_rate": 5.997898038253741e-07,
      "loss": 0.1992,
      "step": 6283
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.077464208791795,
      "learning_rate": 5.987531533928997e-07,
      "loss": 0.1912,
      "step": 6284
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9739589348780749,
      "learning_rate": 5.977173425273968e-07,
      "loss": 0.1683,
      "step": 6285
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8078965265186561,
      "learning_rate": 5.966823714264519e-07,
      "loss": 0.1129,
      "step": 6286
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6336835117876843,
      "learning_rate": 5.95648240287493e-07,
      "loss": 0.0992,
      "step": 6287
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0488789425810507,
      "learning_rate": 5.94614949307788e-07,
      "loss": 0.1621,
      "step": 6288
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9120279060567074,
      "learning_rate": 5.935824986844424e-07,
      "loss": 0.1528,
      "step": 6289
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9894668443124296,
      "learning_rate": 5.925508886144055e-07,
      "loss": 0.1321,
      "step": 6290
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6866931277850614,
      "learning_rate": 5.915201192944625e-07,
      "loss": 0.0822,
      "step": 6291
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8737681642287556,
      "learning_rate": 5.904901909212391e-07,
      "loss": 0.1625,
      "step": 6292
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2833847716477613,
      "learning_rate": 5.894611036912018e-07,
      "loss": 0.1826,
      "step": 6293
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9981923487265315,
      "learning_rate": 5.884328578006548e-07,
      "loss": 0.2155,
      "step": 6294
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.813583842154565,
      "learning_rate": 5.874054534457441e-07,
      "loss": 0.1288,
      "step": 6295
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3894503847509132,
      "learning_rate": 5.863788908224527e-07,
      "loss": 0.2117,
      "step": 6296
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9979169006250002,
      "learning_rate": 5.853531701266046e-07,
      "loss": 0.1903,
      "step": 6297
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0461993037947879,
      "learning_rate": 5.843282915538629e-07,
      "loss": 0.1592,
      "step": 6298
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7802901500769098,
      "learning_rate": 5.833042552997303e-07,
      "loss": 0.1313,
      "step": 6299
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8901947957229105,
      "learning_rate": 5.822810615595476e-07,
      "loss": 0.1246,
      "step": 6300
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0575206208146186,
      "learning_rate": 5.812587105284967e-07,
      "loss": 0.1592,
      "step": 6301
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9566914318955411,
      "learning_rate": 5.802372024015973e-07,
      "loss": 0.1546,
      "step": 6302
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9669475196316842,
      "learning_rate": 5.792165373737086e-07,
      "loss": 0.1846,
      "step": 6303
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9913844770937538,
      "learning_rate": 5.781967156395302e-07,
      "loss": 0.1473,
      "step": 6304
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9824181097664149,
      "learning_rate": 5.771777373935988e-07,
      "loss": 0.1675,
      "step": 6305
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2896950107835483,
      "learning_rate": 5.761596028302918e-07,
      "loss": 0.2143,
      "step": 6306
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9155980904289378,
      "learning_rate": 5.751423121438249e-07,
      "loss": 0.1303,
      "step": 6307
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2513061020020022,
      "learning_rate": 5.741258655282533e-07,
      "loss": 0.2045,
      "step": 6308
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9362728419103497,
      "learning_rate": 5.731102631774705e-07,
      "loss": 0.1472,
      "step": 6309
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.159023914631112,
      "learning_rate": 5.720955052852101e-07,
      "loss": 0.2032,
      "step": 6310
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.98608855880572,
      "learning_rate": 5.710815920450419e-07,
      "loss": 0.1263,
      "step": 6311
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9847878222889241,
      "learning_rate": 5.700685236503789e-07,
      "loss": 0.19,
      "step": 6312
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1085847897300618,
      "learning_rate": 5.690563002944704e-07,
      "loss": 0.1733,
      "step": 6313
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8370565128396594,
      "learning_rate": 5.680449221704038e-07,
      "loss": 0.1161,
      "step": 6314
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0038939228980803,
      "learning_rate": 5.670343894711072e-07,
      "loss": 0.2123,
      "step": 6315
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0704874804326,
      "learning_rate": 5.660247023893445e-07,
      "loss": 0.1554,
      "step": 6316
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7794792342410576,
      "learning_rate": 5.65015861117722e-07,
      "loss": 0.1476,
      "step": 6317
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0223832716264127,
      "learning_rate": 5.64007865848682e-07,
      "loss": 0.1684,
      "step": 6318
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.185889146564933,
      "learning_rate": 5.630007167745061e-07,
      "loss": 0.1957,
      "step": 6319
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8054054391485373,
      "learning_rate": 5.619944140873152e-07,
      "loss": 0.1384,
      "step": 6320
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2420299573179665,
      "learning_rate": 5.609889579790678e-07,
      "loss": 0.2054,
      "step": 6321
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7616327294359372,
      "learning_rate": 5.599843486415607e-07,
      "loss": 0.1395,
      "step": 6322
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9775296684308928,
      "learning_rate": 5.589805862664316e-07,
      "loss": 0.1542,
      "step": 6323
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9307535069843053,
      "learning_rate": 5.579776710451539e-07,
      "loss": 0.144,
      "step": 6324
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0050212838380548,
      "learning_rate": 5.569756031690399e-07,
      "loss": 0.1187,
      "step": 6325
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0623580751007327,
      "learning_rate": 5.559743828292413e-07,
      "loss": 0.1877,
      "step": 6326
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9902988796195,
      "learning_rate": 5.549740102167472e-07,
      "loss": 0.1637,
      "step": 6327
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8625682019438713,
      "learning_rate": 5.53974485522385e-07,
      "loss": 0.144,
      "step": 6328
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0647204828361037,
      "learning_rate": 5.529758089368215e-07,
      "loss": 0.1583,
      "step": 6329
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0281224170402394,
      "learning_rate": 5.519779806505599e-07,
      "loss": 0.1507,
      "step": 6330
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0346037656328568,
      "learning_rate": 5.509810008539435e-07,
      "loss": 0.1751,
      "step": 6331
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7424052746543686,
      "learning_rate": 5.49984869737153e-07,
      "loss": 0.139,
      "step": 6332
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1239726950026039,
      "learning_rate": 5.489895874902052e-07,
      "loss": 0.1971,
      "step": 6333
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1376911567211232,
      "learning_rate": 5.479951543029566e-07,
      "loss": 0.1964,
      "step": 6334
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8298414832175496,
      "learning_rate": 5.470015703651043e-07,
      "loss": 0.1663,
      "step": 6335
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.675882331440319,
      "learning_rate": 5.460088358661802e-07,
      "loss": 0.1289,
      "step": 6336
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0596667866550549,
      "learning_rate": 5.450169509955549e-07,
      "loss": 0.1487,
      "step": 6337
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8114730849049155,
      "learning_rate": 5.440259159424361e-07,
      "loss": 0.1076,
      "step": 6338
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8683828630136692,
      "learning_rate": 5.430357308958711e-07,
      "loss": 0.1709,
      "step": 6339
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1938311533404755,
      "learning_rate": 5.420463960447447e-07,
      "loss": 0.235,
      "step": 6340
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0498113883517028,
      "learning_rate": 5.410579115777781e-07,
      "loss": 0.2129,
      "step": 6341
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0582907525923018,
      "learning_rate": 5.400702776835314e-07,
      "loss": 0.1677,
      "step": 6342
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7590402506058977,
      "learning_rate": 5.390834945504031e-07,
      "loss": 0.1371,
      "step": 6343
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0112085087330025,
      "learning_rate": 5.380975623666279e-07,
      "loss": 0.149,
      "step": 6344
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9727409789707463,
      "learning_rate": 5.371124813202788e-07,
      "loss": 0.1346,
      "step": 6345
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.067419803079856,
      "learning_rate": 5.361282515992666e-07,
      "loss": 0.1959,
      "step": 6346
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.006667192035056,
      "learning_rate": 5.351448733913406e-07,
      "loss": 0.1743,
      "step": 6347
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6961740334951217,
      "learning_rate": 5.341623468840851e-07,
      "loss": 0.1226,
      "step": 6348
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9451164920063745,
      "learning_rate": 5.331806722649252e-07,
      "loss": 0.174,
      "step": 6349
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7914320523516993,
      "learning_rate": 5.321998497211206e-07,
      "loss": 0.1489,
      "step": 6350
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1142690808389233,
      "learning_rate": 5.3121987943977e-07,
      "loss": 0.1958,
      "step": 6351
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5625972881745678,
      "learning_rate": 5.302407616078092e-07,
      "loss": 0.0982,
      "step": 6352
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9118595306615143,
      "learning_rate": 5.292624964120113e-07,
      "loss": 0.1343,
      "step": 6353
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6161848332769246,
      "learning_rate": 5.282850840389875e-07,
      "loss": 0.1054,
      "step": 6354
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9315953265089535,
      "learning_rate": 5.273085246751852e-07,
      "loss": 0.158,
      "step": 6355
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9503710514920243,
      "learning_rate": 5.263328185068888e-07,
      "loss": 0.1484,
      "step": 6356
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.109034243759788,
      "learning_rate": 5.253579657202223e-07,
      "loss": 0.1828,
      "step": 6357
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9701613145284307,
      "learning_rate": 5.243839665011447e-07,
      "loss": 0.1689,
      "step": 6358
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0536977040942241,
      "learning_rate": 5.234108210354527e-07,
      "loss": 0.1615,
      "step": 6359
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9120923000888748,
      "learning_rate": 5.224385295087797e-07,
      "loss": 0.1329,
      "step": 6360
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.665783942298152,
      "learning_rate": 5.21467092106599e-07,
      "loss": 0.0889,
      "step": 6361
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8129534877449156,
      "learning_rate": 5.20496509014215e-07,
      "loss": 0.13,
      "step": 6362
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1365028813980407,
      "learning_rate": 5.195267804167753e-07,
      "loss": 0.1864,
      "step": 6363
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0929886090388992,
      "learning_rate": 5.185579064992618e-07,
      "loss": 0.1765,
      "step": 6364
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8343596349339035,
      "learning_rate": 5.17589887446493e-07,
      "loss": 0.144,
      "step": 6365
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8475793667492336,
      "learning_rate": 5.166227234431254e-07,
      "loss": 0.1388,
      "step": 6366
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0558267491181283,
      "learning_rate": 5.156564146736509e-07,
      "loss": 0.1676,
      "step": 6367
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8672959552998791,
      "learning_rate": 5.146909613224015e-07,
      "loss": 0.1237,
      "step": 6368
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9537162216136689,
      "learning_rate": 5.137263635735423e-07,
      "loss": 0.1666,
      "step": 6369
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8869111204772371,
      "learning_rate": 5.12762621611077e-07,
      "loss": 0.1326,
      "step": 6370
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8064632266417862,
      "learning_rate": 5.117997356188454e-07,
      "loss": 0.1266,
      "step": 6371
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0553248897841854,
      "learning_rate": 5.108377057805253e-07,
      "loss": 0.162,
      "step": 6372
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8249803394870762,
      "learning_rate": 5.09876532279629e-07,
      "loss": 0.1297,
      "step": 6373
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0610393567362673,
      "learning_rate": 5.089162152995075e-07,
      "loss": 0.172,
      "step": 6374
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7131092699894326,
      "learning_rate": 5.079567550233477e-07,
      "loss": 0.1355,
      "step": 6375
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0927078274854929,
      "learning_rate": 5.069981516341727e-07,
      "loss": 0.1858,
      "step": 6376
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8319940937292754,
      "learning_rate": 5.060404053148427e-07,
      "loss": 0.1653,
      "step": 6377
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9747227831021323,
      "learning_rate": 5.050835162480549e-07,
      "loss": 0.1636,
      "step": 6378
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7170252897512815,
      "learning_rate": 5.041274846163391e-07,
      "loss": 0.1364,
      "step": 6379
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2000447845135325,
      "learning_rate": 5.031723106020681e-07,
      "loss": 0.1822,
      "step": 6380
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0338266947945878,
      "learning_rate": 5.022179943874461e-07,
      "loss": 0.166,
      "step": 6381
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2488993678443545,
      "learning_rate": 5.012645361545159e-07,
      "loss": 0.2245,
      "step": 6382
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9060299133681259,
      "learning_rate": 5.003119360851555e-07,
      "loss": 0.1402,
      "step": 6383
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.992710997427097,
      "learning_rate": 4.993601943610798e-07,
      "loss": 0.1987,
      "step": 6384
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.137046372986187,
      "learning_rate": 4.9840931116384e-07,
      "loss": 0.189,
      "step": 6385
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8676078998304928,
      "learning_rate": 4.974592866748229e-07,
      "loss": 0.1674,
      "step": 6386
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8447930920612355,
      "learning_rate": 4.965101210752526e-07,
      "loss": 0.137,
      "step": 6387
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.772048849344421,
      "learning_rate": 4.955618145461883e-07,
      "loss": 0.1189,
      "step": 6388
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8303104590438161,
      "learning_rate": 4.946143672685256e-07,
      "loss": 0.1465,
      "step": 6389
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8641039053909849,
      "learning_rate": 4.936677794229955e-07,
      "loss": 0.1643,
      "step": 6390
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2135383284581949,
      "learning_rate": 4.927220511901693e-07,
      "loss": 0.1999,
      "step": 6391
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7884452403615588,
      "learning_rate": 4.917771827504475e-07,
      "loss": 0.1237,
      "step": 6392
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0230160452614838,
      "learning_rate": 4.908331742840711e-07,
      "loss": 0.154,
      "step": 6393
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9037995322282468,
      "learning_rate": 4.898900259711158e-07,
      "loss": 0.121,
      "step": 6394
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0407882777527573,
      "learning_rate": 4.889477379914936e-07,
      "loss": 0.1731,
      "step": 6395
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8675309108215828,
      "learning_rate": 4.880063105249522e-07,
      "loss": 0.1579,
      "step": 6396
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.18203651302903,
      "learning_rate": 4.870657437510751e-07,
      "loss": 0.1881,
      "step": 6397
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8848415487536554,
      "learning_rate": 4.861260378492816e-07,
      "loss": 0.1382,
      "step": 6398
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7470832430550791,
      "learning_rate": 4.851871929988267e-07,
      "loss": 0.1485,
      "step": 6399
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8672620485408103,
      "learning_rate": 4.842492093788014e-07,
      "loss": 0.1005,
      "step": 6400
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0693028326845089,
      "learning_rate": 4.833120871681313e-07,
      "loss": 0.1952,
      "step": 6401
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.89817158696219,
      "learning_rate": 4.823758265455803e-07,
      "loss": 0.1629,
      "step": 6402
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9389234499997499,
      "learning_rate": 4.814404276897461e-07,
      "loss": 0.1847,
      "step": 6403
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3805380072437428,
      "learning_rate": 4.805058907790611e-07,
      "loss": 0.2236,
      "step": 6404
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9185964155534543,
      "learning_rate": 4.795722159917959e-07,
      "loss": 0.1532,
      "step": 6405
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8995469898040998,
      "learning_rate": 4.786394035060538e-07,
      "loss": 0.1344,
      "step": 6406
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0231218896030658,
      "learning_rate": 4.777074534997755e-07,
      "loss": 0.1614,
      "step": 6407
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8868462163412947,
      "learning_rate": 4.767763661507374e-07,
      "loss": 0.1249,
      "step": 6408
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.034421449688889,
      "learning_rate": 4.7584614163654896e-07,
      "loss": 0.1621,
      "step": 6409
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0452204289544866,
      "learning_rate": 4.749167801346577e-07,
      "loss": 0.1933,
      "step": 6410
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8128865763691816,
      "learning_rate": 4.7398828182234457e-07,
      "loss": 0.138,
      "step": 6411
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8791060882141107,
      "learning_rate": 4.7306064687672737e-07,
      "loss": 0.1312,
      "step": 6412
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7557405145665133,
      "learning_rate": 4.7213387547475896e-07,
      "loss": 0.1303,
      "step": 6413
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.907209228617094,
      "learning_rate": 4.7120796779322744e-07,
      "loss": 0.1514,
      "step": 6414
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9135308649235993,
      "learning_rate": 4.7028292400875474e-07,
      "loss": 0.1052,
      "step": 6415
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1628995042620305,
      "learning_rate": 4.693587442977998e-07,
      "loss": 0.1855,
      "step": 6416
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1751885371104345,
      "learning_rate": 4.684354288366555e-07,
      "loss": 0.1895,
      "step": 6417
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9383083686692925,
      "learning_rate": 4.675129778014509e-07,
      "loss": 0.1673,
      "step": 6418
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0513667540498797,
      "learning_rate": 4.665913913681497e-07,
      "loss": 0.1596,
      "step": 6419
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1893424639207533,
      "learning_rate": 4.656706697125496e-07,
      "loss": 0.1868,
      "step": 6420
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0345221289420548,
      "learning_rate": 4.647508130102857e-07,
      "loss": 0.1765,
      "step": 6421
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.119065406141127,
      "learning_rate": 4.6383182143682594e-07,
      "loss": 0.1715,
      "step": 6422
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0770133808447624,
      "learning_rate": 4.629136951674745e-07,
      "loss": 0.1876,
      "step": 6423
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9513060236042091,
      "learning_rate": 4.619964343773692e-07,
      "loss": 0.1753,
      "step": 6424
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8666653119867324,
      "learning_rate": 4.6108003924148494e-07,
      "loss": 0.1093,
      "step": 6425
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.85487304570555,
      "learning_rate": 4.6016450993462855e-07,
      "loss": 0.146,
      "step": 6426
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1136577620509371,
      "learning_rate": 4.592498466314449e-07,
      "loss": 0.2144,
      "step": 6427
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7011679828914752,
      "learning_rate": 4.5833604950641096e-07,
      "loss": 0.1248,
      "step": 6428
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8470328694884893,
      "learning_rate": 4.5742311873384013e-07,
      "loss": 0.0955,
      "step": 6429
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6837451875300236,
      "learning_rate": 4.565110544878798e-07,
      "loss": 0.104,
      "step": 6430
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8954120052792939,
      "learning_rate": 4.5559985694251186e-07,
      "loss": 0.1405,
      "step": 6431
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1019082372884303,
      "learning_rate": 4.546895262715539e-07,
      "loss": 0.1863,
      "step": 6432
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0381878063235304,
      "learning_rate": 4.537800626486577e-07,
      "loss": 0.1591,
      "step": 6433
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1127247709874724,
      "learning_rate": 4.528714662473088e-07,
      "loss": 0.1533,
      "step": 6434
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.646117499795589,
      "learning_rate": 4.5196373724082754e-07,
      "loss": 0.0855,
      "step": 6435
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0335400954746639,
      "learning_rate": 4.510568758023709e-07,
      "loss": 0.141,
      "step": 6436
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9958253356718576,
      "learning_rate": 4.5015088210492717e-07,
      "loss": 0.1774,
      "step": 6437
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8166083199526809,
      "learning_rate": 4.492457563213226e-07,
      "loss": 0.1083,
      "step": 6438
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9716060690372261,
      "learning_rate": 4.4834149862421403e-07,
      "loss": 0.1636,
      "step": 6439
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0405519442068278,
      "learning_rate": 4.4743810918609463e-07,
      "loss": 0.1545,
      "step": 6440
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8819017154034535,
      "learning_rate": 4.4653558817929264e-07,
      "loss": 0.1561,
      "step": 6441
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.13205047983376,
      "learning_rate": 4.4563393577596925e-07,
      "loss": 0.1848,
      "step": 6442
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.903405096997521,
      "learning_rate": 4.4473315214812127e-07,
      "loss": 0.1521,
      "step": 6443
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.171827189825303,
      "learning_rate": 4.438332374675791e-07,
      "loss": 0.2009,
      "step": 6444
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.869932608833549,
      "learning_rate": 4.4293419190600764e-07,
      "loss": 0.1351,
      "step": 6445
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7694561381004996,
      "learning_rate": 4.420360156349041e-07,
      "loss": 0.1486,
      "step": 6446
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0374554250753323,
      "learning_rate": 4.411387088256036e-07,
      "loss": 0.1521,
      "step": 6447
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.074006989542153,
      "learning_rate": 4.402422716492727e-07,
      "loss": 0.1948,
      "step": 6448
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8598074327594086,
      "learning_rate": 4.3934670427691284e-07,
      "loss": 0.1209,
      "step": 6449
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0412107908330104,
      "learning_rate": 4.384520068793591e-07,
      "loss": 0.152,
      "step": 6450
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8705221625334096,
      "learning_rate": 4.3755817962728096e-07,
      "loss": 0.1351,
      "step": 6451
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9274617552430099,
      "learning_rate": 4.3666522269118215e-07,
      "loss": 0.1902,
      "step": 6452
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9976743129947836,
      "learning_rate": 4.357731362413997e-07,
      "loss": 0.1725,
      "step": 6453
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9340032688502382,
      "learning_rate": 4.3488192044810586e-07,
      "loss": 0.1457,
      "step": 6454
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0530111800883135,
      "learning_rate": 4.3399157548130565e-07,
      "loss": 0.1705,
      "step": 6455
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8170778766830011,
      "learning_rate": 4.331021015108372e-07,
      "loss": 0.1388,
      "step": 6456
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7546400708734039,
      "learning_rate": 4.322134987063731e-07,
      "loss": 0.1376,
      "step": 6457
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8107502280700065,
      "learning_rate": 4.313257672374227e-07,
      "loss": 0.1048,
      "step": 6458
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0726992280511949,
      "learning_rate": 4.30438907273325e-07,
      "loss": 0.1539,
      "step": 6459
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.055569913458303,
      "learning_rate": 4.2955291898325536e-07,
      "loss": 0.1523,
      "step": 6460
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9616161707263087,
      "learning_rate": 4.286678025362212e-07,
      "loss": 0.1468,
      "step": 6461
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.953575077812341,
      "learning_rate": 4.277835581010642e-07,
      "loss": 0.1271,
      "step": 6462
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0355726775644365,
      "learning_rate": 4.2690018584645996e-07,
      "loss": 0.1959,
      "step": 6463
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.213549376469964,
      "learning_rate": 4.260176859409182e-07,
      "loss": 0.2098,
      "step": 6464
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0115768362932047,
      "learning_rate": 4.2513605855278085e-07,
      "loss": 0.172,
      "step": 6465
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8909129033414762,
      "learning_rate": 4.242553038502245e-07,
      "loss": 0.1407,
      "step": 6466
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.743871526709058,
      "learning_rate": 4.2337542200125926e-07,
      "loss": 0.1299,
      "step": 6467
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6488685374319328,
      "learning_rate": 4.2249641317372804e-07,
      "loss": 0.1156,
      "step": 6468
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8448071152873876,
      "learning_rate": 4.216182775353073e-07,
      "loss": 0.1654,
      "step": 6469
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7087261492290345,
      "learning_rate": 4.20741015253508e-07,
      "loss": 0.1179,
      "step": 6470
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8868306110551291,
      "learning_rate": 4.1986462649567294e-07,
      "loss": 0.1435,
      "step": 6471
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8161106589166939,
      "learning_rate": 4.18989111428979e-07,
      "loss": 0.1262,
      "step": 6472
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.917232189949663,
      "learning_rate": 4.181144702204376e-07,
      "loss": 0.1515,
      "step": 6473
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.581779098374162,
      "learning_rate": 4.1724070303689136e-07,
      "loss": 0.1265,
      "step": 6474
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.107703673919751,
      "learning_rate": 4.163678100450175e-07,
      "loss": 0.1833,
      "step": 6475
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.896800873808154,
      "learning_rate": 4.1549579141132615e-07,
      "loss": 0.1635,
      "step": 6476
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9908288057270572,
      "learning_rate": 4.1462464730216033e-07,
      "loss": 0.1581,
      "step": 6477
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7440298303116876,
      "learning_rate": 4.1375437788369667e-07,
      "loss": 0.1335,
      "step": 6478
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.282407051469742,
      "learning_rate": 4.128849833219456e-07,
      "loss": 0.1912,
      "step": 6479
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.081437284401157,
      "learning_rate": 4.1201646378274785e-07,
      "loss": 0.197,
      "step": 6480
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1757527714252658,
      "learning_rate": 4.1114881943178143e-07,
      "loss": 0.2061,
      "step": 6481
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.155212154795072,
      "learning_rate": 4.102820504345545e-07,
      "loss": 0.1693,
      "step": 6482
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9712598078953576,
      "learning_rate": 4.0941615695640867e-07,
      "loss": 0.1895,
      "step": 6483
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7471690672061314,
      "learning_rate": 4.085511391625208e-07,
      "loss": 0.1246,
      "step": 6484
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1826021928746024,
      "learning_rate": 4.076869972178954e-07,
      "loss": 0.208,
      "step": 6485
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9552185486071433,
      "learning_rate": 4.0682373128737517e-07,
      "loss": 0.1465,
      "step": 6486
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8788915218191641,
      "learning_rate": 4.0596134153563395e-07,
      "loss": 0.1418,
      "step": 6487
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9850177161247985,
      "learning_rate": 4.050998281271773e-07,
      "loss": 0.1878,
      "step": 6488
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0057863075292472,
      "learning_rate": 4.0423919122634547e-07,
      "loss": 0.1648,
      "step": 6489
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9703155084764146,
      "learning_rate": 4.0337943099731046e-07,
      "loss": 0.1614,
      "step": 6490
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2780864770273908,
      "learning_rate": 4.025205476040761e-07,
      "loss": 0.2051,
      "step": 6491
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8821344164709313,
      "learning_rate": 4.016625412104824e-07,
      "loss": 0.1357,
      "step": 6492
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9302304373117717,
      "learning_rate": 4.008054119801985e-07,
      "loss": 0.1476,
      "step": 6493
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8109677424672546,
      "learning_rate": 3.999491600767269e-07,
      "loss": 0.1444,
      "step": 6494
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2689648179918647,
      "learning_rate": 3.9909378566340417e-07,
      "loss": 0.192,
      "step": 6495
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.317319543632297,
      "learning_rate": 3.982392889033987e-07,
      "loss": 0.1801,
      "step": 6496
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.097641840689414,
      "learning_rate": 3.973856699597106e-07,
      "loss": 0.1926,
      "step": 6497
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.865846921185871,
      "learning_rate": 3.9653292899517414e-07,
      "loss": 0.1461,
      "step": 6498
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.989683764736022,
      "learning_rate": 3.956810661724547e-07,
      "loss": 0.1696,
      "step": 6499
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9756648782807744,
      "learning_rate": 3.9483008165405123e-07,
      "loss": 0.1817,
      "step": 6500
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5613915559447358,
      "learning_rate": 3.9397997560229437e-07,
      "loss": 0.1131,
      "step": 6501
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.008566720028377,
      "learning_rate": 3.9313074817934783e-07,
      "loss": 0.1527,
      "step": 6502
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1722708014827223,
      "learning_rate": 3.9228239954720693e-07,
      "loss": 0.179,
      "step": 6503
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.096787406809332,
      "learning_rate": 3.9143492986770007e-07,
      "loss": 0.1673,
      "step": 6504
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0682586899980069,
      "learning_rate": 3.905883393024873e-07,
      "loss": 0.2069,
      "step": 6505
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.264826663646371,
      "learning_rate": 3.897426280130617e-07,
      "loss": 0.21,
      "step": 6506
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5669285699780408,
      "learning_rate": 3.888977961607482e-07,
      "loss": 0.1055,
      "step": 6507
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0185287585168588,
      "learning_rate": 3.880538439067039e-07,
      "loss": 0.1617,
      "step": 6508
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0171604367433305,
      "learning_rate": 3.872107714119189e-07,
      "loss": 0.1687,
      "step": 6509
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0556227379999779,
      "learning_rate": 3.863685788372146e-07,
      "loss": 0.1848,
      "step": 6510
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6028610893912756,
      "learning_rate": 3.8552726634324413e-07,
      "loss": 0.109,
      "step": 6511
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7670237576698306,
      "learning_rate": 3.8468683409049466e-07,
      "loss": 0.1694,
      "step": 6512
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9934111821423252,
      "learning_rate": 3.838472822392819e-07,
      "loss": 0.1359,
      "step": 6513
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.003227345760076,
      "learning_rate": 3.8300861094975997e-07,
      "loss": 0.1791,
      "step": 6514
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.064978338384462,
      "learning_rate": 3.821708203819069e-07,
      "loss": 0.1894,
      "step": 6515
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0130215871248245,
      "learning_rate": 3.813339106955388e-07,
      "loss": 0.1855,
      "step": 6516
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0222025365445833,
      "learning_rate": 3.8049788205030136e-07,
      "loss": 0.1604,
      "step": 6517
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9056031836229449,
      "learning_rate": 3.796627346056725e-07,
      "loss": 0.1416,
      "step": 6518
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.151253380161323,
      "learning_rate": 3.78828468520962e-07,
      "loss": 0.1908,
      "step": 6519
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8883815463969833,
      "learning_rate": 3.7799508395531206e-07,
      "loss": 0.133,
      "step": 6520
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9875291166161327,
      "learning_rate": 3.77162581067696e-07,
      "loss": 0.1619,
      "step": 6521
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.309152225975833,
      "learning_rate": 3.763309600169196e-07,
      "loss": 0.2586,
      "step": 6522
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9295171634152507,
      "learning_rate": 3.755002209616193e-07,
      "loss": 0.1317,
      "step": 6523
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9364247972245481,
      "learning_rate": 3.746703640602639e-07,
      "loss": 0.1477,
      "step": 6524
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0878118754137895,
      "learning_rate": 3.7384138947115565e-07,
      "loss": 0.1489,
      "step": 6525
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8301701944822767,
      "learning_rate": 3.730132973524264e-07,
      "loss": 0.1715,
      "step": 6526
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7107326026389474,
      "learning_rate": 3.721860878620398e-07,
      "loss": 0.1302,
      "step": 6527
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0219976931003387,
      "learning_rate": 3.7135976115779127e-07,
      "loss": 0.1654,
      "step": 6528
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8994930083728846,
      "learning_rate": 3.705343173973086e-07,
      "loss": 0.1677,
      "step": 6529
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7993534409902192,
      "learning_rate": 3.697097567380503e-07,
      "loss": 0.1211,
      "step": 6530
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8433669191744421,
      "learning_rate": 3.688860793373078e-07,
      "loss": 0.1475,
      "step": 6531
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.642017097339313,
      "learning_rate": 3.6806328535220147e-07,
      "loss": 0.0871,
      "step": 6532
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8956533472298287,
      "learning_rate": 3.672413749396853e-07,
      "loss": 0.1557,
      "step": 6533
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8549453310518719,
      "learning_rate": 3.6642034825654427e-07,
      "loss": 0.124,
      "step": 6534
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0403827706534647,
      "learning_rate": 3.656002054593949e-07,
      "loss": 0.1674,
      "step": 6535
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8813547320067389,
      "learning_rate": 3.6478094670468367e-07,
      "loss": 0.1211,
      "step": 6536
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7598998566142101,
      "learning_rate": 3.639625721486911e-07,
      "loss": 0.104,
      "step": 6537
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9000704563939006,
      "learning_rate": 3.631450819475274e-07,
      "loss": 0.1417,
      "step": 6538
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0877764961658094,
      "learning_rate": 3.6232847625713386e-07,
      "loss": 0.1819,
      "step": 6539
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8388659602356552,
      "learning_rate": 3.615127552332831e-07,
      "loss": 0.1522,
      "step": 6540
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9662056558674348,
      "learning_rate": 3.606979190315796e-07,
      "loss": 0.1654,
      "step": 6541
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9706670629200765,
      "learning_rate": 3.5988396780745836e-07,
      "loss": 0.1646,
      "step": 6542
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0814522148929162,
      "learning_rate": 3.59070901716187e-07,
      "loss": 0.1963,
      "step": 6543
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7558270547264629,
      "learning_rate": 3.582587209128624e-07,
      "loss": 0.1101,
      "step": 6544
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8978317037987603,
      "learning_rate": 3.5744742555241354e-07,
      "loss": 0.1413,
      "step": 6545
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9096836905331253,
      "learning_rate": 3.5663701578960096e-07,
      "loss": 0.121,
      "step": 6546
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.00582611429529,
      "learning_rate": 3.5582749177901497e-07,
      "loss": 0.1853,
      "step": 6547
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1640381679664376,
      "learning_rate": 3.550188536750776e-07,
      "loss": 0.1898,
      "step": 6548
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6902101338254265,
      "learning_rate": 3.542111016320421e-07,
      "loss": 0.1301,
      "step": 6549
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.955197945022915,
      "learning_rate": 3.5340423580399296e-07,
      "loss": 0.1647,
      "step": 6550
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.140311955598939,
      "learning_rate": 3.525982563448449e-07,
      "loss": 0.1986,
      "step": 6551
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.95068368921538,
      "learning_rate": 3.5179316340834377e-07,
      "loss": 0.1726,
      "step": 6552
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0281036899925298,
      "learning_rate": 3.5098895714806625e-07,
      "loss": 0.1951,
      "step": 6553
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7488986129334169,
      "learning_rate": 3.501856377174201e-07,
      "loss": 0.1466,
      "step": 6554
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0974802154508676,
      "learning_rate": 3.493832052696439e-07,
      "loss": 0.1687,
      "step": 6555
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3699203070728392,
      "learning_rate": 3.485816599578068e-07,
      "loss": 0.2349,
      "step": 6556
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.814473312491895,
      "learning_rate": 3.477810019348088e-07,
      "loss": 0.1212,
      "step": 6557
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8310675606196032,
      "learning_rate": 3.4698123135338047e-07,
      "loss": 0.1131,
      "step": 6558
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0879481998263407,
      "learning_rate": 3.4618234836608424e-07,
      "loss": 0.1687,
      "step": 6559
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1889208860488423,
      "learning_rate": 3.4538435312531207e-07,
      "loss": 0.2106,
      "step": 6560
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1244545905105425,
      "learning_rate": 3.4458724578328774e-07,
      "loss": 0.2091,
      "step": 6561
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1082101657036865,
      "learning_rate": 3.437910264920624e-07,
      "loss": 0.2006,
      "step": 6562
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0402475983565507,
      "learning_rate": 3.429956954035213e-07,
      "loss": 0.1414,
      "step": 6563
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2047701986373314,
      "learning_rate": 3.4220125266937967e-07,
      "loss": 0.2273,
      "step": 6564
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8768276410422384,
      "learning_rate": 3.414076984411818e-07,
      "loss": 0.154,
      "step": 6565
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0809589769565038,
      "learning_rate": 3.4061503287030393e-07,
      "loss": 0.164,
      "step": 6566
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.136133764892567,
      "learning_rate": 3.398232561079523e-07,
      "loss": 0.1359,
      "step": 6567
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8034942994577804,
      "learning_rate": 3.3903236830516384e-07,
      "loss": 0.1336,
      "step": 6568
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9982189516096012,
      "learning_rate": 3.38242369612804e-07,
      "loss": 0.1985,
      "step": 6569
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.02777525926754,
      "learning_rate": 3.374532601815722e-07,
      "loss": 0.1591,
      "step": 6570
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0826231575601886,
      "learning_rate": 3.3666504016199586e-07,
      "loss": 0.1693,
      "step": 6571
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9733790090647892,
      "learning_rate": 3.35877709704433e-07,
      "loss": 0.1225,
      "step": 6572
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8063944014097106,
      "learning_rate": 3.3509126895907186e-07,
      "loss": 0.1497,
      "step": 6573
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7984358656295524,
      "learning_rate": 3.343057180759313e-07,
      "loss": 0.1191,
      "step": 6574
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8268560964927839,
      "learning_rate": 3.3352105720486027e-07,
      "loss": 0.1332,
      "step": 6575
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.015142484497892,
      "learning_rate": 3.3273728649553863e-07,
      "loss": 0.1403,
      "step": 6576
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8574767530874033,
      "learning_rate": 3.3195440609747496e-07,
      "loss": 0.1568,
      "step": 6577
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8976287344952174,
      "learning_rate": 3.311724161600105e-07,
      "loss": 0.1775,
      "step": 6578
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7736575301619936,
      "learning_rate": 3.303913168323131e-07,
      "loss": 0.1046,
      "step": 6579
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7322249549765311,
      "learning_rate": 3.296111082633824e-07,
      "loss": 0.1372,
      "step": 6580
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9280715034259762,
      "learning_rate": 3.2883179060204994e-07,
      "loss": 0.1449,
      "step": 6581
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1052150353396726,
      "learning_rate": 3.280533639969752e-07,
      "loss": 0.1721,
      "step": 6582
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9922173758037507,
      "learning_rate": 3.272758285966482e-07,
      "loss": 0.1826,
      "step": 6583
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7875461382178235,
      "learning_rate": 3.2649918454938866e-07,
      "loss": 0.1576,
      "step": 6584
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8779228830794216,
      "learning_rate": 3.2572343200334646e-07,
      "loss": 0.1762,
      "step": 6585
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7658826541318579,
      "learning_rate": 3.24948571106502e-07,
      "loss": 0.1635,
      "step": 6586
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8411251851074788,
      "learning_rate": 3.241746020066655e-07,
      "loss": 0.1447,
      "step": 6587
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1134114911987871,
      "learning_rate": 3.2340152485147544e-07,
      "loss": 0.1691,
      "step": 6588
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.856687338466489,
      "learning_rate": 3.226293397884028e-07,
      "loss": 0.1446,
      "step": 6589
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9504562480092675,
      "learning_rate": 3.2185804696474586e-07,
      "loss": 0.15,
      "step": 6590
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2006452986744334,
      "learning_rate": 3.2108764652763467e-07,
      "loss": 0.183,
      "step": 6591
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1917619237533197,
      "learning_rate": 3.203181386240273e-07,
      "loss": 0.2188,
      "step": 6592
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9488561665167081,
      "learning_rate": 3.1954952340071344e-07,
      "loss": 0.1496,
      "step": 6593
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0120851279874687,
      "learning_rate": 3.187818010043109e-07,
      "loss": 0.1271,
      "step": 6594
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7736068664262942,
      "learning_rate": 3.1801497158126806e-07,
      "loss": 0.1423,
      "step": 6595
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8292479212760469,
      "learning_rate": 3.1724903527786233e-07,
      "loss": 0.1636,
      "step": 6596
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7028020536128271,
      "learning_rate": 3.164839922402019e-07,
      "loss": 0.0989,
      "step": 6597
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0252751921226475,
      "learning_rate": 3.1571984261422275e-07,
      "loss": 0.1485,
      "step": 6598
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0309274672242745,
      "learning_rate": 3.1495658654569273e-07,
      "loss": 0.1847,
      "step": 6599
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9886273596206779,
      "learning_rate": 3.1419422418020706e-07,
      "loss": 0.1439,
      "step": 6600
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1299984692462481,
      "learning_rate": 3.1343275566319166e-07,
      "loss": 0.1864,
      "step": 6601
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0895693647275881,
      "learning_rate": 3.126721811399019e-07,
      "loss": 0.1819,
      "step": 6602
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9584299114018205,
      "learning_rate": 3.1191250075542137e-07,
      "loss": 0.1597,
      "step": 6603
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.013272709964075,
      "learning_rate": 3.111537146546656e-07,
      "loss": 0.1507,
      "step": 6604
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6960865398307213,
      "learning_rate": 3.1039582298237723e-07,
      "loss": 0.1158,
      "step": 6605
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9649240524826111,
      "learning_rate": 3.0963882588312956e-07,
      "loss": 0.1537,
      "step": 6606
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9512347960010288,
      "learning_rate": 3.0888272350132474e-07,
      "loss": 0.1695,
      "step": 6607
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8354475531704867,
      "learning_rate": 3.0812751598119474e-07,
      "loss": 0.1287,
      "step": 6608
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3614181477512404,
      "learning_rate": 3.073732034667992e-07,
      "loss": 0.1722,
      "step": 6609
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.128437758210328,
      "learning_rate": 3.066197861020287e-07,
      "loss": 0.1817,
      "step": 6610
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.278323403117601,
      "learning_rate": 3.058672640306021e-07,
      "loss": 0.2203,
      "step": 6611
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.927759220749876,
      "learning_rate": 3.051156373960695e-07,
      "loss": 0.1461,
      "step": 6612
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7852338785709027,
      "learning_rate": 3.0436490634180693e-07,
      "loss": 0.1497,
      "step": 6613
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9150770539355506,
      "learning_rate": 3.0361507101102194e-07,
      "loss": 0.1629,
      "step": 6614
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0524752225839542,
      "learning_rate": 3.0286613154675125e-07,
      "loss": 0.1965,
      "step": 6615
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1172043250432258,
      "learning_rate": 3.021180880918595e-07,
      "loss": 0.1681,
      "step": 6616
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9713451782661265,
      "learning_rate": 3.013709407890408e-07,
      "loss": 0.1956,
      "step": 6617
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0129546656229351,
      "learning_rate": 3.006246897808185e-07,
      "loss": 0.1558,
      "step": 6618
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1158046260658976,
      "learning_rate": 2.998793352095453e-07,
      "loss": 0.2053,
      "step": 6619
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7012829817839722,
      "learning_rate": 2.991348772174024e-07,
      "loss": 0.1635,
      "step": 6620
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7621891692892512,
      "learning_rate": 2.9839131594639967e-07,
      "loss": 0.1286,
      "step": 6621
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7692795225306365,
      "learning_rate": 2.976486515383764e-07,
      "loss": 0.1412,
      "step": 6622
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8686282571969726,
      "learning_rate": 2.969068841350009e-07,
      "loss": 0.1287,
      "step": 6623
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7328120633589157,
      "learning_rate": 2.9616601387777053e-07,
      "loss": 0.1339,
      "step": 6624
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2295634185224427,
      "learning_rate": 2.954260409080106e-07,
      "loss": 0.2054,
      "step": 6625
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.975385218802498,
      "learning_rate": 2.946869653668766e-07,
      "loss": 0.1724,
      "step": 6626
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9944115560919254,
      "learning_rate": 2.9394878739535115e-07,
      "loss": 0.1763,
      "step": 6627
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5561884290660416,
      "learning_rate": 2.932115071342473e-07,
      "loss": 0.0937,
      "step": 6628
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8211377836272143,
      "learning_rate": 2.9247512472420637e-07,
      "loss": 0.143,
      "step": 6629
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6816106400405629,
      "learning_rate": 2.917396403056971e-07,
      "loss": 0.1272,
      "step": 6630
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7874203318456808,
      "learning_rate": 2.9100505401901903e-07,
      "loss": 0.1197,
      "step": 6631
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9541211153683593,
      "learning_rate": 2.9027136600429827e-07,
      "loss": 0.1474,
      "step": 6632
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0419436612363782,
      "learning_rate": 2.895385764014919e-07,
      "loss": 0.1675,
      "step": 6633
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7313657012592987,
      "learning_rate": 2.88806685350384e-07,
      "loss": 0.0897,
      "step": 6634
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1751544951110067,
      "learning_rate": 2.8807569299058704e-07,
      "loss": 0.1832,
      "step": 6635
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.87394548536581,
      "learning_rate": 2.873455994615437e-07,
      "loss": 0.1567,
      "step": 6636
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.848540032767753,
      "learning_rate": 2.8661640490252275e-07,
      "loss": 0.1624,
      "step": 6637
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1583998791422128,
      "learning_rate": 2.8588810945262444e-07,
      "loss": 0.1626,
      "step": 6638
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0481943384045274,
      "learning_rate": 2.851607132507744e-07,
      "loss": 0.1344,
      "step": 6639
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7557330697740667,
      "learning_rate": 2.8443421643573e-07,
      "loss": 0.1443,
      "step": 6640
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.079365388849619,
      "learning_rate": 2.837086191460736e-07,
      "loss": 0.1537,
      "step": 6641
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8645301192705651,
      "learning_rate": 2.8298392152021846e-07,
      "loss": 0.1257,
      "step": 6642
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9452217071725098,
      "learning_rate": 2.8226012369640567e-07,
      "loss": 0.1708,
      "step": 6643
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6575937846407753,
      "learning_rate": 2.8153722581270414e-07,
      "loss": 0.0842,
      "step": 6644
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.85116681734894,
      "learning_rate": 2.80815228007012e-07,
      "loss": 0.1304,
      "step": 6645
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8196352757830315,
      "learning_rate": 2.80094130417054e-07,
      "loss": 0.14,
      "step": 6646
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0628025874867095,
      "learning_rate": 2.793739331803852e-07,
      "loss": 0.1947,
      "step": 6647
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1177429552312959,
      "learning_rate": 2.786546364343867e-07,
      "loss": 0.1757,
      "step": 6648
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7312960174397884,
      "learning_rate": 2.7793624031627107e-07,
      "loss": 0.127,
      "step": 6649
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7359838654132836,
      "learning_rate": 2.7721874496307645e-07,
      "loss": 0.1435,
      "step": 6650
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9826294479012873,
      "learning_rate": 2.765021505116694e-07,
      "loss": 0.1449,
      "step": 6651
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8377987551786372,
      "learning_rate": 2.7578645709874453e-07,
      "loss": 0.1547,
      "step": 6652
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.004328620497427,
      "learning_rate": 2.750716648608265e-07,
      "loss": 0.1522,
      "step": 6653
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7000974968985629,
      "learning_rate": 2.743577739342662e-07,
      "loss": 0.1293,
      "step": 6654
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.026127375034486,
      "learning_rate": 2.7364478445524257e-07,
      "loss": 0.2004,
      "step": 6655
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6239615986397823,
      "learning_rate": 2.7293269655976287e-07,
      "loss": 0.0863,
      "step": 6656
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.993312021105059,
      "learning_rate": 2.722215103836634e-07,
      "loss": 0.1849,
      "step": 6657
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0139377405710521,
      "learning_rate": 2.7151122606260683e-07,
      "loss": 0.188,
      "step": 6658
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.190791376272607,
      "learning_rate": 2.708018437320842e-07,
      "loss": 0.2356,
      "step": 6659
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9823940295867932,
      "learning_rate": 2.700933635274167e-07,
      "loss": 0.1693,
      "step": 6660
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9885278584573925,
      "learning_rate": 2.693857855837501e-07,
      "loss": 0.1654,
      "step": 6661
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.157938190274593,
      "learning_rate": 2.6867911003606083e-07,
      "loss": 0.1793,
      "step": 6662
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8425218021527714,
      "learning_rate": 2.6797333701915064e-07,
      "loss": 0.1553,
      "step": 6663
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9835979923039619,
      "learning_rate": 2.6726846666765115e-07,
      "loss": 0.1561,
      "step": 6664
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0083558905833165,
      "learning_rate": 2.6656449911602047e-07,
      "loss": 0.1678,
      "step": 6665
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8986191501543311,
      "learning_rate": 2.6586143449854553e-07,
      "loss": 0.1278,
      "step": 6666
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8114877995375229,
      "learning_rate": 2.6515927294934074e-07,
      "loss": 0.1196,
      "step": 6667
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8275420902361819,
      "learning_rate": 2.6445801460234786e-07,
      "loss": 0.1251,
      "step": 6668
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8567256928654707,
      "learning_rate": 2.6375765959133605e-07,
      "loss": 0.1206,
      "step": 6669
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8239166944593588,
      "learning_rate": 2.6305820804990336e-07,
      "loss": 0.1132,
      "step": 6670
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8002047464781497,
      "learning_rate": 2.6235966011147474e-07,
      "loss": 0.113,
      "step": 6671
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7788489659890118,
      "learning_rate": 2.616620159093025e-07,
      "loss": 0.1105,
      "step": 6672
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8506461770175697,
      "learning_rate": 2.609652755764669e-07,
      "loss": 0.1237,
      "step": 6673
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8267672891438624,
      "learning_rate": 2.602694392458754e-07,
      "loss": 0.1292,
      "step": 6674
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8155835992315059,
      "learning_rate": 2.5957450705026477e-07,
      "loss": 0.1587,
      "step": 6675
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1255525061762994,
      "learning_rate": 2.588804791221966e-07,
      "loss": 0.1696,
      "step": 6676
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9200730104289826,
      "learning_rate": 2.5818735559406174e-07,
      "loss": 0.1438,
      "step": 6677
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1151723915832161,
      "learning_rate": 2.574951365980782e-07,
      "loss": 0.1631,
      "step": 6678
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9706705628537926,
      "learning_rate": 2.568038222662911e-07,
      "loss": 0.1611,
      "step": 6679
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8375354864081039,
      "learning_rate": 2.5611341273057367e-07,
      "loss": 0.1625,
      "step": 6680
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0352167944023354,
      "learning_rate": 2.554239081226245e-07,
      "loss": 0.1623,
      "step": 6681
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1577350788844432,
      "learning_rate": 2.5473530857397397e-07,
      "loss": 0.1872,
      "step": 6682
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0765132097359427,
      "learning_rate": 2.5404761421597477e-07,
      "loss": 0.1533,
      "step": 6683
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8086476054473513,
      "learning_rate": 2.5336082517981087e-07,
      "loss": 0.1275,
      "step": 6684
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0969198175037491,
      "learning_rate": 2.526749415964902e-07,
      "loss": 0.143,
      "step": 6685
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6274480009715614,
      "learning_rate": 2.519899635968498e-07,
      "loss": 0.1025,
      "step": 6686
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0289831294528313,
      "learning_rate": 2.5130589131155456e-07,
      "loss": 0.1627,
      "step": 6687
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.141361479274327,
      "learning_rate": 2.506227248710952e-07,
      "loss": 0.2035,
      "step": 6688
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0732214347186955,
      "learning_rate": 2.499404644057907e-07,
      "loss": 0.1711,
      "step": 6689
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.078564032124749,
      "learning_rate": 2.492591100457864e-07,
      "loss": 0.1805,
      "step": 6690
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0026996718355696,
      "learning_rate": 2.485786619210551e-07,
      "loss": 0.171,
      "step": 6691
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1276088585751578,
      "learning_rate": 2.478991201613973e-07,
      "loss": 0.1927,
      "step": 6692
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8259801642402856,
      "learning_rate": 2.472204848964388e-07,
      "loss": 0.1221,
      "step": 6693
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7321050038793474,
      "learning_rate": 2.4654275625563494e-07,
      "loss": 0.1165,
      "step": 6694
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9595477855416754,
      "learning_rate": 2.4586593436826736e-07,
      "loss": 0.155,
      "step": 6695
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.012408104952246,
      "learning_rate": 2.451900193634432e-07,
      "loss": 0.1637,
      "step": 6696
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0478786614380016,
      "learning_rate": 2.4451501137009837e-07,
      "loss": 0.1868,
      "step": 6697
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9831590381526357,
      "learning_rate": 2.438409105169953e-07,
      "loss": 0.1471,
      "step": 6698
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9200254236748565,
      "learning_rate": 2.4316771693272235e-07,
      "loss": 0.1833,
      "step": 6699
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1306890587841099,
      "learning_rate": 2.4249543074569615e-07,
      "loss": 0.2137,
      "step": 6700
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0327272427140246,
      "learning_rate": 2.4182405208416084e-07,
      "loss": 0.1434,
      "step": 6701
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8563372905697705,
      "learning_rate": 2.411535810761839e-07,
      "loss": 0.1254,
      "step": 6702
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0486879698177756,
      "learning_rate": 2.404840178496637e-07,
      "loss": 0.1251,
      "step": 6703
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0175661968996863,
      "learning_rate": 2.3981536253232293e-07,
      "loss": 0.1708,
      "step": 6704
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0037161600615598,
      "learning_rate": 2.3914761525171357e-07,
      "loss": 0.1751,
      "step": 6705
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9068768901769396,
      "learning_rate": 2.3848077613521146e-07,
      "loss": 0.1313,
      "step": 6706
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.08900783272371,
      "learning_rate": 2.3781484531002153e-07,
      "loss": 0.1756,
      "step": 6707
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9861045997301896,
      "learning_rate": 2.3714982290317378e-07,
      "loss": 0.1724,
      "step": 6708
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8600836028666228,
      "learning_rate": 2.3648570904152623e-07,
      "loss": 0.1395,
      "step": 6709
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1264065551759803,
      "learning_rate": 2.3582250385176196e-07,
      "loss": 0.1772,
      "step": 6710
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0170562906502958,
      "learning_rate": 2.351602074603926e-07,
      "loss": 0.1784,
      "step": 6711
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7077727724972602,
      "learning_rate": 2.344988199937559e-07,
      "loss": 0.1201,
      "step": 6712
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9078826414633829,
      "learning_rate": 2.3383834157801488e-07,
      "loss": 0.1194,
      "step": 6713
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6366206323870837,
      "learning_rate": 2.3317877233916099e-07,
      "loss": 0.1329,
      "step": 6714
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.90466204577714,
      "learning_rate": 2.3252011240301076e-07,
      "loss": 0.1427,
      "step": 6715
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7319442519345528,
      "learning_rate": 2.3186236189520871e-07,
      "loss": 0.1112,
      "step": 6716
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1310513202235326,
      "learning_rate": 2.312055209412245e-07,
      "loss": 0.17,
      "step": 6717
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0884141335528246,
      "learning_rate": 2.305495896663551e-07,
      "loss": 0.2074,
      "step": 6718
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0318554833307692,
      "learning_rate": 2.2989456819572375e-07,
      "loss": 0.1963,
      "step": 6719
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0828734760347496,
      "learning_rate": 2.2924045665427995e-07,
      "loss": 0.1327,
      "step": 6720
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.014443844098307,
      "learning_rate": 2.2858725516680004e-07,
      "loss": 0.146,
      "step": 6721
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.824328367063193,
      "learning_rate": 2.279349638578865e-07,
      "loss": 0.1371,
      "step": 6722
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8035756610096375,
      "learning_rate": 2.2728358285196816e-07,
      "loss": 0.1649,
      "step": 6723
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0172173420235422,
      "learning_rate": 2.266331122733001e-07,
      "loss": 0.1786,
      "step": 6724
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9526066231190604,
      "learning_rate": 2.259835522459647e-07,
      "loss": 0.1467,
      "step": 6725
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8436062421839398,
      "learning_rate": 2.2533490289386785e-07,
      "loss": 0.1428,
      "step": 6726
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9434540414125806,
      "learning_rate": 2.246871643407461e-07,
      "loss": 0.1219,
      "step": 6727
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9563641041477071,
      "learning_rate": 2.2404033671015845e-07,
      "loss": 0.1558,
      "step": 6728
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.795439573271076,
      "learning_rate": 2.2339442012549173e-07,
      "loss": 0.117,
      "step": 6729
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0729858226882558,
      "learning_rate": 2.227494147099596e-07,
      "loss": 0.1962,
      "step": 6730
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9940246777118744,
      "learning_rate": 2.2210532058660085e-07,
      "loss": 0.153,
      "step": 6731
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8956076044655401,
      "learning_rate": 2.2146213787827996e-07,
      "loss": 0.1771,
      "step": 6732
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8824056545303289,
      "learning_rate": 2.208198667076883e-07,
      "loss": 0.1076,
      "step": 6733
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.012451149655774,
      "learning_rate": 2.201785071973439e-07,
      "loss": 0.1836,
      "step": 6734
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.031169732380788,
      "learning_rate": 2.1953805946959006e-07,
      "loss": 0.1623,
      "step": 6735
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8322791958333444,
      "learning_rate": 2.1889852364659626e-07,
      "loss": 0.1054,
      "step": 6736
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5280690386015066,
      "learning_rate": 2.1825989985035877e-07,
      "loss": 0.2885,
      "step": 6737
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9772694647092872,
      "learning_rate": 2.1762218820269854e-07,
      "loss": 0.2014,
      "step": 6738
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9963411976620876,
      "learning_rate": 2.1698538882526432e-07,
      "loss": 0.1658,
      "step": 6739
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9543734217586894,
      "learning_rate": 2.1634950183952897e-07,
      "loss": 0.1877,
      "step": 6740
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.133498945206632,
      "learning_rate": 2.1571452736679266e-07,
      "loss": 0.1777,
      "step": 6741
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.252649209739906,
      "learning_rate": 2.150804655281802e-07,
      "loss": 0.218,
      "step": 6742
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9260883962658802,
      "learning_rate": 2.1444731644464422e-07,
      "loss": 0.1935,
      "step": 6743
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7756229770376946,
      "learning_rate": 2.1381508023696095e-07,
      "loss": 0.128,
      "step": 6744
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0744534020311167,
      "learning_rate": 2.1318375702573445e-07,
      "loss": 0.1506,
      "step": 6745
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9790213493799531,
      "learning_rate": 2.1255334693139396e-07,
      "loss": 0.1789,
      "step": 6746
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.014456228385367,
      "learning_rate": 2.1192385007419325e-07,
      "loss": 0.1583,
      "step": 6747
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9726128222690806,
      "learning_rate": 2.1129526657421462e-07,
      "loss": 0.1395,
      "step": 6748
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0480702252575425,
      "learning_rate": 2.1066759655136216e-07,
      "loss": 0.1949,
      "step": 6749
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9251505464067075,
      "learning_rate": 2.1004084012537008e-07,
      "loss": 0.1476,
      "step": 6750
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0444927302495286,
      "learning_rate": 2.0941499741579607e-07,
      "loss": 0.1816,
      "step": 6751
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6937258058917575,
      "learning_rate": 2.08790068542023e-07,
      "loss": 0.116,
      "step": 6752
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.665808402027934,
      "learning_rate": 2.0816605362326102e-07,
      "loss": 0.1114,
      "step": 6753
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0593196446942101,
      "learning_rate": 2.075429527785444e-07,
      "loss": 0.1633,
      "step": 6754
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0440140460702834,
      "learning_rate": 2.0692076612673474e-07,
      "loss": 0.1509,
      "step": 6755
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8143268402190774,
      "learning_rate": 2.0629949378651702e-07,
      "loss": 0.1219,
      "step": 6756
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.956364049779853,
      "learning_rate": 2.0567913587640375e-07,
      "loss": 0.1409,
      "step": 6757
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1029134722279517,
      "learning_rate": 2.0505969251473245e-07,
      "loss": 0.2,
      "step": 6758
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.093290526690002,
      "learning_rate": 2.0444116381966527e-07,
      "loss": 0.1643,
      "step": 6759
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0073043726104889,
      "learning_rate": 2.0382354990919174e-07,
      "loss": 0.1733,
      "step": 6760
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2034475784381755,
      "learning_rate": 2.032068509011248e-07,
      "loss": 0.2171,
      "step": 6761
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7845397303968557,
      "learning_rate": 2.0259106691310482e-07,
      "loss": 0.1339,
      "step": 6762
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0280729390011725,
      "learning_rate": 2.0197619806259561e-07,
      "loss": 0.1522,
      "step": 6763
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.028247938623212,
      "learning_rate": 2.0136224446688836e-07,
      "loss": 0.1843,
      "step": 6764
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.94611881732135,
      "learning_rate": 2.0074920624309824e-07,
      "loss": 0.1705,
      "step": 6765
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9174364743102539,
      "learning_rate": 2.0013708350816675e-07,
      "loss": 0.1408,
      "step": 6766
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9312328786789269,
      "learning_rate": 1.995258763788599e-07,
      "loss": 0.1387,
      "step": 6767
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.103724422168606,
      "learning_rate": 1.9891558497177e-07,
      "loss": 0.143,
      "step": 6768
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0269077014687187,
      "learning_rate": 1.983062094033139e-07,
      "loss": 0.1709,
      "step": 6769
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3199742643808787,
      "learning_rate": 1.9769774978973423e-07,
      "loss": 0.2248,
      "step": 6770
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.114703220683054,
      "learning_rate": 1.9709020624709752e-07,
      "loss": 0.1832,
      "step": 6771
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0106402372272287,
      "learning_rate": 1.964835788912983e-07,
      "loss": 0.1352,
      "step": 6772
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9436781271395638,
      "learning_rate": 1.9587786783805408e-07,
      "loss": 0.1606,
      "step": 6773
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9080278328034828,
      "learning_rate": 1.95273073202909e-07,
      "loss": 0.1301,
      "step": 6774
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9945442752322082,
      "learning_rate": 1.9466919510123028e-07,
      "loss": 0.1399,
      "step": 6775
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8769700505060033,
      "learning_rate": 1.940662336482124e-07,
      "loss": 0.1298,
      "step": 6776
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9746005730966486,
      "learning_rate": 1.9346418895887508e-07,
      "loss": 0.1611,
      "step": 6777
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9146394840734171,
      "learning_rate": 1.9286306114806086e-07,
      "loss": 0.1366,
      "step": 6778
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2354333036826854,
      "learning_rate": 1.9226285033043967e-07,
      "loss": 0.1961,
      "step": 6779
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.979747714937446,
      "learning_rate": 1.9166355662050496e-07,
      "loss": 0.1445,
      "step": 6780
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8447773918798533,
      "learning_rate": 1.9106518013257692e-07,
      "loss": 0.1282,
      "step": 6781
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.99086148726328,
      "learning_rate": 1.904677209807987e-07,
      "loss": 0.1282,
      "step": 6782
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0160127955853104,
      "learning_rate": 1.8987117927914077e-07,
      "loss": 0.1328,
      "step": 6783
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1023543365938702,
      "learning_rate": 1.8927555514139662e-07,
      "loss": 0.1727,
      "step": 6784
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1766364740573778,
      "learning_rate": 1.886808486811864e-07,
      "loss": 0.2475,
      "step": 6785
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9423938932289952,
      "learning_rate": 1.8808706001195332e-07,
      "loss": 0.1521,
      "step": 6786
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9729259324778075,
      "learning_rate": 1.8749418924696727e-07,
      "loss": 0.1305,
      "step": 6787
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9248349212095932,
      "learning_rate": 1.869022364993217e-07,
      "loss": 0.1707,
      "step": 6788
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1245511083907447,
      "learning_rate": 1.863112018819352e-07,
      "loss": 0.1721,
      "step": 6789
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.852660704903649,
      "learning_rate": 1.8572108550755253e-07,
      "loss": 0.1072,
      "step": 6790
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7774192655014818,
      "learning_rate": 1.8513188748874201e-07,
      "loss": 0.0947,
      "step": 6791
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.112694410758652,
      "learning_rate": 1.845436079378965e-07,
      "loss": 0.1878,
      "step": 6792
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.354375617679019,
      "learning_rate": 1.8395624696723457e-07,
      "loss": 0.2098,
      "step": 6793
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1071921577629877,
      "learning_rate": 1.833698046887994e-07,
      "loss": 0.1794,
      "step": 6794
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1382392362398797,
      "learning_rate": 1.8278428121445812e-07,
      "loss": 0.2032,
      "step": 6795
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9712796762694941,
      "learning_rate": 1.8219967665590422e-07,
      "loss": 0.118,
      "step": 6796
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0066596934531054,
      "learning_rate": 1.81615991124654e-07,
      "loss": 0.1439,
      "step": 6797
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9870830836275584,
      "learning_rate": 1.8103322473204954e-07,
      "loss": 0.175,
      "step": 6798
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0096027253721307,
      "learning_rate": 1.80451377589258e-07,
      "loss": 0.1838,
      "step": 6799
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0142456734755338,
      "learning_rate": 1.7987044980726954e-07,
      "loss": 0.1848,
      "step": 6800
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8570320286690623,
      "learning_rate": 1.7929044149690044e-07,
      "loss": 0.1599,
      "step": 6801
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7547474577260419,
      "learning_rate": 1.787113527687917e-07,
      "loss": 0.1133,
      "step": 6802
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.974620879857114,
      "learning_rate": 1.781331837334077e-07,
      "loss": 0.1524,
      "step": 6803
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1398265511814931,
      "learning_rate": 1.7755593450103802e-07,
      "loss": 0.1793,
      "step": 6804
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1207655921191881,
      "learning_rate": 1.7697960518179626e-07,
      "loss": 0.1974,
      "step": 6805
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8862655907905792,
      "learning_rate": 1.764041958856222e-07,
      "loss": 0.1619,
      "step": 6806
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1513751816395996,
      "learning_rate": 1.7582970672227918e-07,
      "loss": 0.178,
      "step": 6807
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9378278663179058,
      "learning_rate": 1.7525613780135343e-07,
      "loss": 0.1558,
      "step": 6808
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1056748457585173,
      "learning_rate": 1.7468348923225797e-07,
      "loss": 0.1528,
      "step": 6809
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0222818524322523,
      "learning_rate": 1.7411176112422823e-07,
      "loss": 0.136,
      "step": 6810
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8079899017361339,
      "learning_rate": 1.735409535863264e-07,
      "loss": 0.1207,
      "step": 6811
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7099181268290162,
      "learning_rate": 1.7297106672743758e-07,
      "loss": 0.1266,
      "step": 6812
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0435795623900979,
      "learning_rate": 1.7240210065627094e-07,
      "loss": 0.1374,
      "step": 6813
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9459127143358255,
      "learning_rate": 1.718340554813608e-07,
      "loss": 0.1861,
      "step": 6814
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9578330210788875,
      "learning_rate": 1.7126693131106542e-07,
      "loss": 0.1667,
      "step": 6815
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0156030790565675,
      "learning_rate": 1.707007282535672e-07,
      "loss": 0.1577,
      "step": 6816
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9079643661929773,
      "learning_rate": 1.7013544641687363e-07,
      "loss": 0.1417,
      "step": 6817
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7561887728279548,
      "learning_rate": 1.695710859088162e-07,
      "loss": 0.1533,
      "step": 6818
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3462879323708707,
      "learning_rate": 1.6900764683704996e-07,
      "loss": 0.2081,
      "step": 6819
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.99541328036293,
      "learning_rate": 1.6844512930905444e-07,
      "loss": 0.1362,
      "step": 6820
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1495349320148396,
      "learning_rate": 1.6788353343213437e-07,
      "loss": 0.2165,
      "step": 6821
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.017767985687681,
      "learning_rate": 1.673228593134174e-07,
      "loss": 0.166,
      "step": 6822
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9868431142243442,
      "learning_rate": 1.667631070598552e-07,
      "loss": 0.147,
      "step": 6823
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0126302945884185,
      "learning_rate": 1.662042767782257e-07,
      "loss": 0.1869,
      "step": 6824
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8684027348539511,
      "learning_rate": 1.6564636857512805e-07,
      "loss": 0.1427,
      "step": 6825
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.095211437168177,
      "learning_rate": 1.650893825569877e-07,
      "loss": 0.1766,
      "step": 6826
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0844122036317763,
      "learning_rate": 1.6453331883005242e-07,
      "loss": 0.1798,
      "step": 6827
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7792485030088863,
      "learning_rate": 1.639781775003968e-07,
      "loss": 0.1458,
      "step": 6828
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8481350409316776,
      "learning_rate": 1.6342395867391614e-07,
      "loss": 0.1494,
      "step": 6829
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0624312129570461,
      "learning_rate": 1.6287066245633254e-07,
      "loss": 0.1648,
      "step": 6830
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.154335516833946,
      "learning_rate": 1.6231828895318992e-07,
      "loss": 0.1424,
      "step": 6831
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0101539612005428,
      "learning_rate": 1.6176683826985783e-07,
      "loss": 0.1534,
      "step": 6832
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7825455891168713,
      "learning_rate": 1.612163105115283e-07,
      "loss": 0.1337,
      "step": 6833
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9814518092291986,
      "learning_rate": 1.6066670578321953e-07,
      "loss": 0.1546,
      "step": 6834
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9846516738820059,
      "learning_rate": 1.6011802418977097e-07,
      "loss": 0.1707,
      "step": 6835
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2561405259857576,
      "learning_rate": 1.5957026583584833e-07,
      "loss": 0.2176,
      "step": 6836
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0616547584092364,
      "learning_rate": 1.5902343082593919e-07,
      "loss": 0.1923,
      "step": 6837
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.21433307855318,
      "learning_rate": 1.5847751926435618e-07,
      "loss": 0.2004,
      "step": 6838
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9732220053610612,
      "learning_rate": 1.57932531255236e-07,
      "loss": 0.1532,
      "step": 6839
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0721191161610808,
      "learning_rate": 1.5738846690253883e-07,
      "loss": 0.1844,
      "step": 6840
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7367615112818118,
      "learning_rate": 1.568453263100478e-07,
      "loss": 0.1192,
      "step": 6841
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0285972660886171,
      "learning_rate": 1.563031095813705e-07,
      "loss": 0.1714,
      "step": 6842
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1037997840668385,
      "learning_rate": 1.5576181681993928e-07,
      "loss": 0.159,
      "step": 6843
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9266977943529282,
      "learning_rate": 1.5522144812900875e-07,
      "loss": 0.1607,
      "step": 6844
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0291541873348247,
      "learning_rate": 1.5468200361165752e-07,
      "loss": 0.1627,
      "step": 6845
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9830241068235963,
      "learning_rate": 1.541434833707889e-07,
      "loss": 0.1986,
      "step": 6846
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7975243805304287,
      "learning_rate": 1.5360588750912897e-07,
      "loss": 0.135,
      "step": 6847
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9284062720392169,
      "learning_rate": 1.5306921612922742e-07,
      "loss": 0.1733,
      "step": 6848
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7675289180221971,
      "learning_rate": 1.5253346933345846e-07,
      "loss": 0.1192,
      "step": 6849
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0012495692163907,
      "learning_rate": 1.5199864722401758e-07,
      "loss": 0.1985,
      "step": 6850
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0767112626210948,
      "learning_rate": 1.5146474990292815e-07,
      "loss": 0.1599,
      "step": 6851
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8361832294703531,
      "learning_rate": 1.5093177747203315e-07,
      "loss": 0.114,
      "step": 6852
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1691105339445367,
      "learning_rate": 1.5039973003300125e-07,
      "loss": 0.1959,
      "step": 6853
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.015301717942936,
      "learning_rate": 1.4986860768732402e-07,
      "loss": 0.1855,
      "step": 6854
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.001258828550003,
      "learning_rate": 1.4933841053631547e-07,
      "loss": 0.1494,
      "step": 6855
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9314623884934662,
      "learning_rate": 1.4880913868111525e-07,
      "loss": 0.1304,
      "step": 6856
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0575387703299641,
      "learning_rate": 1.4828079222268477e-07,
      "loss": 0.1358,
      "step": 6857
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.967803336688854,
      "learning_rate": 1.477533712618101e-07,
      "loss": 0.1647,
      "step": 6858
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7792933362088337,
      "learning_rate": 1.4722687589910024e-07,
      "loss": 0.1105,
      "step": 6859
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0322848378259848,
      "learning_rate": 1.4670130623498812e-07,
      "loss": 0.1695,
      "step": 6860
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7792738474141399,
      "learning_rate": 1.4617666236972805e-07,
      "loss": 0.1222,
      "step": 6861
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0177908163604303,
      "learning_rate": 1.4565294440340105e-07,
      "loss": 0.1663,
      "step": 6862
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8263887906331342,
      "learning_rate": 1.4513015243590944e-07,
      "loss": 0.1155,
      "step": 6863
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0075184391409626,
      "learning_rate": 1.4460828656697844e-07,
      "loss": 0.1679,
      "step": 6864
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1410329235693637,
      "learning_rate": 1.4408734689615845e-07,
      "loss": 0.1601,
      "step": 6865
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0434385779581652,
      "learning_rate": 1.4356733352282105e-07,
      "loss": 0.1471,
      "step": 6866
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.212335567364262,
      "learning_rate": 1.4304824654616355e-07,
      "loss": 0.2087,
      "step": 6867
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8955785094588053,
      "learning_rate": 1.4253008606520403e-07,
      "loss": 0.128,
      "step": 6868
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1990913870746325,
      "learning_rate": 1.4201285217878558e-07,
      "loss": 0.1787,
      "step": 6869
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8501299172632159,
      "learning_rate": 1.4149654498557374e-07,
      "loss": 0.1411,
      "step": 6870
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0648243759354779,
      "learning_rate": 1.4098116458405865e-07,
      "loss": 0.1788,
      "step": 6871
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1117321612656634,
      "learning_rate": 1.404667110725505e-07,
      "loss": 0.1919,
      "step": 6872
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0051994081301492,
      "learning_rate": 1.3995318454918582e-07,
      "loss": 0.1727,
      "step": 6873
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8733244448772443,
      "learning_rate": 1.3944058511192348e-07,
      "loss": 0.1653,
      "step": 6874
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8248642055521022,
      "learning_rate": 1.3892891285854472e-07,
      "loss": 0.1678,
      "step": 6875
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.040102610725063,
      "learning_rate": 1.3841816788665475e-07,
      "loss": 0.1674,
      "step": 6876
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0465346620653737,
      "learning_rate": 1.3790835029368177e-07,
      "loss": 0.2106,
      "step": 6877
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9718102152972311,
      "learning_rate": 1.3739946017687578e-07,
      "loss": 0.1658,
      "step": 6878
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9880900417321249,
      "learning_rate": 1.3689149763331188e-07,
      "loss": 0.1937,
      "step": 6879
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0688475872829633,
      "learning_rate": 1.3638446275988704e-07,
      "loss": 0.1845,
      "step": 6880
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7375971134015602,
      "learning_rate": 1.3587835565332163e-07,
      "loss": 0.0847,
      "step": 6881
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9916687036891483,
      "learning_rate": 1.35373176410159e-07,
      "loss": 0.1705,
      "step": 6882
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7535632217084743,
      "learning_rate": 1.3486892512676487e-07,
      "loss": 0.1397,
      "step": 6883
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8426302529136925,
      "learning_rate": 1.3436560189932946e-07,
      "loss": 0.1642,
      "step": 6884
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.126881376383651,
      "learning_rate": 1.3386320682386378e-07,
      "loss": 0.2253,
      "step": 6885
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1001898073756695,
      "learning_rate": 1.3336173999620395e-07,
      "loss": 0.1747,
      "step": 6886
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8640578017306897,
      "learning_rate": 1.328612015120079e-07,
      "loss": 0.1401,
      "step": 6887
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8951414773510545,
      "learning_rate": 1.3236159146675642e-07,
      "loss": 0.135,
      "step": 6888
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.220633420885509,
      "learning_rate": 1.318629099557539e-07,
      "loss": 0.2181,
      "step": 6889
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9883186681332496,
      "learning_rate": 1.3136515707412645e-07,
      "loss": 0.188,
      "step": 6890
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.877711501819609,
      "learning_rate": 1.3086833291682422e-07,
      "loss": 0.1414,
      "step": 6891
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9343673676747426,
      "learning_rate": 1.3037243757861972e-07,
      "loss": 0.1216,
      "step": 6892
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.807103721283885,
      "learning_rate": 1.2987747115410786e-07,
      "loss": 0.1272,
      "step": 6893
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2335430650237793,
      "learning_rate": 1.2938343373770757e-07,
      "loss": 0.1911,
      "step": 6894
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7659543403398328,
      "learning_rate": 1.2889032542365843e-07,
      "loss": 0.1131,
      "step": 6895
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7711357798791394,
      "learning_rate": 1.283981463060252e-07,
      "loss": 0.1138,
      "step": 6896
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.062924210897339,
      "learning_rate": 1.2790689647869448e-07,
      "loss": 0.1852,
      "step": 6897
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7583324833634865,
      "learning_rate": 1.2741657603537462e-07,
      "loss": 0.1464,
      "step": 6898
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7306173758575644,
      "learning_rate": 1.26927185069598e-07,
      "loss": 0.123,
      "step": 6899
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9021564794402062,
      "learning_rate": 1.2643872367471888e-07,
      "loss": 0.1257,
      "step": 6900
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7719130946016068,
      "learning_rate": 1.2595119194391548e-07,
      "loss": 0.1401,
      "step": 6901
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.039889120537121,
      "learning_rate": 1.2546458997018617e-07,
      "loss": 0.1796,
      "step": 6902
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8245906176885905,
      "learning_rate": 1.2497891784635396e-07,
      "loss": 0.1143,
      "step": 6903
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7596167999096254,
      "learning_rate": 1.2449417566506473e-07,
      "loss": 0.1621,
      "step": 6904
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7801814233950001,
      "learning_rate": 1.2401036351878558e-07,
      "loss": 0.1161,
      "step": 6905
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6441709767186422,
      "learning_rate": 1.2352748149980666e-07,
      "loss": 0.0864,
      "step": 6906
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0410202981008367,
      "learning_rate": 1.230455297002414e-07,
      "loss": 0.1772,
      "step": 6907
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2021837901248675,
      "learning_rate": 1.2256450821202582e-07,
      "loss": 0.2268,
      "step": 6908
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6626419927594319,
      "learning_rate": 1.2208441712691698e-07,
      "loss": 0.1172,
      "step": 6909
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9762276305307642,
      "learning_rate": 1.2160525653649612e-07,
      "loss": 0.1275,
      "step": 6910
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.058085175121899,
      "learning_rate": 1.211270265321657e-07,
      "loss": 0.182,
      "step": 6911
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0497647619294852,
      "learning_rate": 1.2064972720515212e-07,
      "loss": 0.1759,
      "step": 6912
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0813809603471911,
      "learning_rate": 1.2017335864650203e-07,
      "loss": 0.1868,
      "step": 6913
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7989126268666471,
      "learning_rate": 1.196979209470872e-07,
      "loss": 0.1693,
      "step": 6914
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.978452244598369,
      "learning_rate": 1.1922341419760008e-07,
      "loss": 0.1904,
      "step": 6915
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9177666824822925,
      "learning_rate": 1.1874983848855604e-07,
      "loss": 0.1292,
      "step": 6916
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0826419225178698,
      "learning_rate": 1.1827719391029225e-07,
      "loss": 0.1811,
      "step": 6917
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7690066591901719,
      "learning_rate": 1.1780548055296936e-07,
      "loss": 0.122,
      "step": 6918
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2246651204166494,
      "learning_rate": 1.1733469850656931e-07,
      "loss": 0.1905,
      "step": 6919
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1238667593114104,
      "learning_rate": 1.1686484786089747e-07,
      "loss": 0.207,
      "step": 6920
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9747297005547378,
      "learning_rate": 1.1639592870558048e-07,
      "loss": 0.1361,
      "step": 6921
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8187022322931965,
      "learning_rate": 1.1592794113006845e-07,
      "loss": 0.1537,
      "step": 6922
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1102674764004108,
      "learning_rate": 1.1546088522363219e-07,
      "loss": 0.1899,
      "step": 6923
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7075808873327298,
      "learning_rate": 1.1499476107536655e-07,
      "loss": 0.0996,
      "step": 6924
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0803615244105516,
      "learning_rate": 1.1452956877418764e-07,
      "loss": 0.1752,
      "step": 6925
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.237496649926138,
      "learning_rate": 1.1406530840883334e-07,
      "loss": 0.2013,
      "step": 6926
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7954974255427564,
      "learning_rate": 1.136019800678645e-07,
      "loss": 0.1219,
      "step": 6927
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9967805143938459,
      "learning_rate": 1.1313958383966428e-07,
      "loss": 0.1649,
      "step": 6928
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7159837535860237,
      "learning_rate": 1.1267811981243882e-07,
      "loss": 0.1081,
      "step": 6929
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8359313315181557,
      "learning_rate": 1.1221758807421435e-07,
      "loss": 0.1367,
      "step": 6930
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8070837742125326,
      "learning_rate": 1.1175798871284006e-07,
      "loss": 0.1318,
      "step": 6931
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.794275802656754,
      "learning_rate": 1.1129932181598856e-07,
      "loss": 0.1235,
      "step": 6932
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1966159876135878,
      "learning_rate": 1.108415874711527e-07,
      "loss": 0.1987,
      "step": 6933
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8261832940836285,
      "learning_rate": 1.103847857656487e-07,
      "loss": 0.1099,
      "step": 6934
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0266613638903914,
      "learning_rate": 1.0992891678661466e-07,
      "loss": 0.1513,
      "step": 6935
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9331578311293341,
      "learning_rate": 1.0947398062101045e-07,
      "loss": 0.162,
      "step": 6936
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9484773346406137,
      "learning_rate": 1.0901997735561776e-07,
      "loss": 0.1392,
      "step": 6937
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.877577674585599,
      "learning_rate": 1.0856690707704176e-07,
      "loss": 0.1373,
      "step": 6938
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7359981688196937,
      "learning_rate": 1.0811476987170777e-07,
      "loss": 0.1162,
      "step": 6939
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8747269360441735,
      "learning_rate": 1.0766356582586346e-07,
      "loss": 0.1236,
      "step": 6940
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9785448245176511,
      "learning_rate": 1.0721329502557998e-07,
      "loss": 0.1566,
      "step": 6941
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1579438074836157,
      "learning_rate": 1.0676395755674973e-07,
      "loss": 0.1692,
      "step": 6942
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9503997292611295,
      "learning_rate": 1.0631555350508582e-07,
      "loss": 0.1471,
      "step": 6943
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0033255358415225,
      "learning_rate": 1.0586808295612539e-07,
      "loss": 0.1926,
      "step": 6944
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9649499364955014,
      "learning_rate": 1.0542154599522514e-07,
      "loss": 0.1539,
      "step": 6945
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8265417903652713,
      "learning_rate": 1.0497594270756528e-07,
      "loss": 0.1088,
      "step": 6946
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1065140627433878,
      "learning_rate": 1.0453127317814893e-07,
      "loss": 0.1692,
      "step": 6947
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0319409985190968,
      "learning_rate": 1.0408753749179823e-07,
      "loss": 0.1396,
      "step": 6948
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0404655209606903,
      "learning_rate": 1.0364473573315881e-07,
      "loss": 0.1673,
      "step": 6949
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1247742042901339,
      "learning_rate": 1.0320286798669866e-07,
      "loss": 0.1969,
      "step": 6950
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7124769309542707,
      "learning_rate": 1.027619343367059e-07,
      "loss": 0.0965,
      "step": 6951
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9031262664229643,
      "learning_rate": 1.023219348672927e-07,
      "loss": 0.1521,
      "step": 6952
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8381989219143028,
      "learning_rate": 1.0188286966239192e-07,
      "loss": 0.1454,
      "step": 6953
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9227675066641171,
      "learning_rate": 1.0144473880575768e-07,
      "loss": 0.1627,
      "step": 6954
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.146477928778864,
      "learning_rate": 1.0100754238096589e-07,
      "loss": 0.1946,
      "step": 6955
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0154914022686132,
      "learning_rate": 1.0057128047141595e-07,
      "loss": 0.1613,
      "step": 6956
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0913733916168618,
      "learning_rate": 1.0013595316032631e-07,
      "loss": 0.1885,
      "step": 6957
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8640093656692655,
      "learning_rate": 9.970156053073943e-08,
      "loss": 0.1272,
      "step": 6958
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0680869611371815,
      "learning_rate": 9.926810266551846e-08,
      "loss": 0.181,
      "step": 6959
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2424681281862482,
      "learning_rate": 9.88355796473478e-08,
      "loss": 0.1763,
      "step": 6960
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8896922965322368,
      "learning_rate": 9.840399155873481e-08,
      "loss": 0.1501,
      "step": 6961
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9925847711156711,
      "learning_rate": 9.797333848200697e-08,
      "loss": 0.182,
      "step": 6962
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.031894349965629,
      "learning_rate": 9.75436204993152e-08,
      "loss": 0.1521,
      "step": 6963
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9180696057251551,
      "learning_rate": 9.711483769263009e-08,
      "loss": 0.1808,
      "step": 6964
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1383570756316745,
      "learning_rate": 9.668699014374561e-08,
      "loss": 0.2309,
      "step": 6965
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0305743378514032,
      "learning_rate": 9.626007793427594e-08,
      "loss": 0.1606,
      "step": 6966
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8381107163402245,
      "learning_rate": 9.58341011456576e-08,
      "loss": 0.1691,
      "step": 6967
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6027096068304391,
      "learning_rate": 9.540905985914894e-08,
      "loss": 0.0824,
      "step": 6968
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9200510133713593,
      "learning_rate": 9.498495415582842e-08,
      "loss": 0.1458,
      "step": 6969
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.055291907111611,
      "learning_rate": 9.456178411659744e-08,
      "loss": 0.1689,
      "step": 6970
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1751543791541668,
      "learning_rate": 9.413954982217921e-08,
      "loss": 0.2077,
      "step": 6971
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2126832459883061,
      "learning_rate": 9.371825135311652e-08,
      "loss": 0.2243,
      "step": 6972
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8385530446159899,
      "learning_rate": 9.329788878977508e-08,
      "loss": 0.1447,
      "step": 6973
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0532505459681425,
      "learning_rate": 9.287846221234298e-08,
      "loss": 0.1633,
      "step": 6974
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.974995952349562,
      "learning_rate": 9.245997170082733e-08,
      "loss": 0.1498,
      "step": 6975
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.100508714808763,
      "learning_rate": 9.20424173350587e-08,
      "loss": 0.2052,
      "step": 6976
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.148564326871821,
      "learning_rate": 9.16257991946884e-08,
      "loss": 0.1967,
      "step": 6977
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.161186775073153,
      "learning_rate": 9.121011735918895e-08,
      "loss": 0.1725,
      "step": 6978
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0719089407269116,
      "learning_rate": 9.079537190785359e-08,
      "loss": 0.1634,
      "step": 6979
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0515948872860046,
      "learning_rate": 9.03815629197985e-08,
      "loss": 0.1393,
      "step": 6980
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.073805003163807,
      "learning_rate": 8.99686904739605e-08,
      "loss": 0.173,
      "step": 6981
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2063120131641136,
      "learning_rate": 8.955675464909774e-08,
      "loss": 0.1754,
      "step": 6982
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7927009938293855,
      "learning_rate": 8.914575552378901e-08,
      "loss": 0.1135,
      "step": 6983
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9169515131823162,
      "learning_rate": 8.873569317643604e-08,
      "loss": 0.127,
      "step": 6984
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8224274799941372,
      "learning_rate": 8.832656768526072e-08,
      "loss": 0.117,
      "step": 6985
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9200720761655052,
      "learning_rate": 8.791837912830615e-08,
      "loss": 0.13,
      "step": 6986
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0177390240974866,
      "learning_rate": 8.751112758343782e-08,
      "loss": 0.1513,
      "step": 6987
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8040040487009378,
      "learning_rate": 8.71048131283403e-08,
      "loss": 0.1644,
      "step": 6988
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8101141854816508,
      "learning_rate": 8.66994358405221e-08,
      "loss": 0.1416,
      "step": 6989
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.917024736915328,
      "learning_rate": 8.629499579731026e-08,
      "loss": 0.1756,
      "step": 6990
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1623350934465588,
      "learning_rate": 8.589149307585586e-08,
      "loss": 0.2068,
      "step": 6991
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9852066247858529,
      "learning_rate": 8.548892775312844e-08,
      "loss": 0.1622,
      "step": 6992
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9540801564633401,
      "learning_rate": 8.508729990592102e-08,
      "loss": 0.1387,
      "step": 6993
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9861813244108999,
      "learning_rate": 8.468660961084674e-08,
      "loss": 0.1619,
      "step": 6994
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1166845453035672,
      "learning_rate": 8.428685694433891e-08,
      "loss": 0.2069,
      "step": 6995
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2924889724761786,
      "learning_rate": 8.388804198265321e-08,
      "loss": 0.1976,
      "step": 6996
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9526324579270022,
      "learning_rate": 8.349016480186656e-08,
      "loss": 0.138,
      "step": 6997
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0747713239233667,
      "learning_rate": 8.309322547787713e-08,
      "loss": 0.148,
      "step": 6998
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9660227252042999,
      "learning_rate": 8.269722408640324e-08,
      "loss": 0.1511,
      "step": 6999
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0087833132393098,
      "learning_rate": 8.230216070298502e-08,
      "loss": 0.1598,
      "step": 7000
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7152347525241377,
      "learning_rate": 8.190803540298331e-08,
      "loss": 0.11,
      "step": 7001
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.054730053694937,
      "learning_rate": 8.151484826157963e-08,
      "loss": 0.1605,
      "step": 7002
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9315182783968157,
      "learning_rate": 8.112259935377786e-08,
      "loss": 0.189,
      "step": 7003
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1291506339491824,
      "learning_rate": 8.07312887544015e-08,
      "loss": 0.1903,
      "step": 7004
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8260030822275065,
      "learning_rate": 8.034091653809584e-08,
      "loss": 0.1344,
      "step": 7005
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.212945458502857,
      "learning_rate": 7.995148277932685e-08,
      "loss": 0.1714,
      "step": 7006
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6702863924064277,
      "learning_rate": 7.956298755238179e-08,
      "loss": 0.0768,
      "step": 7007
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9002958350063365,
      "learning_rate": 7.91754309313686e-08,
      "loss": 0.1365,
      "step": 7008
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2152535770728965,
      "learning_rate": 7.878881299021645e-08,
      "loss": 0.1917,
      "step": 7009
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6844110970902733,
      "learning_rate": 7.840313380267473e-08,
      "loss": 0.1434,
      "step": 7010
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9978101672618794,
      "learning_rate": 7.801839344231454e-08,
      "loss": 0.1488,
      "step": 7011
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0447438043261512,
      "learning_rate": 7.763459198252831e-08,
      "loss": 0.1723,
      "step": 7012
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1202286791690361,
      "learning_rate": 7.725172949652749e-08,
      "loss": 0.1759,
      "step": 7013
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0791106556770236,
      "learning_rate": 7.686980605734695e-08,
      "loss": 0.2037,
      "step": 7014
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7887193245136266,
      "learning_rate": 7.648882173784011e-08,
      "loss": 0.1042,
      "step": 7015
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9201503694891774,
      "learning_rate": 7.610877661068273e-08,
      "loss": 0.1709,
      "step": 7016
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7511217234758081,
      "learning_rate": 7.572967074837068e-08,
      "loss": 0.1261,
      "step": 7017
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9753919936055008,
      "learning_rate": 7.535150422322058e-08,
      "loss": 0.1683,
      "step": 7018
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1904585194941226,
      "learning_rate": 7.497427710737137e-08,
      "loss": 0.2544,
      "step": 7019
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8916280664734003,
      "learning_rate": 7.459798947278052e-08,
      "loss": 0.121,
      "step": 7020
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0580553714593297,
      "learning_rate": 7.422264139122837e-08,
      "loss": 0.1646,
      "step": 7021
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1215091846467247,
      "learning_rate": 7.384823293431376e-08,
      "loss": 0.1692,
      "step": 7022
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0711031375615445,
      "learning_rate": 7.347476417345844e-08,
      "loss": 0.1473,
      "step": 7023
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9523342540473632,
      "learning_rate": 7.31022351799049e-08,
      "loss": 0.1668,
      "step": 7024
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.106292044854448,
      "learning_rate": 7.273064602471347e-08,
      "loss": 0.1881,
      "step": 7025
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8545987396965311,
      "learning_rate": 7.23599967787686e-08,
      "loss": 0.1268,
      "step": 7026
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0527027312857609,
      "learning_rate": 7.19902875127737e-08,
      "loss": 0.1855,
      "step": 7027
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1455493082986261,
      "learning_rate": 7.162151829725294e-08,
      "loss": 0.1996,
      "step": 7028
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9139721571737562,
      "learning_rate": 7.125368920255171e-08,
      "loss": 0.1136,
      "step": 7029
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8426379912520281,
      "learning_rate": 7.088680029883666e-08,
      "loss": 0.1606,
      "step": 7030
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.946100134389037,
      "learning_rate": 7.052085165609346e-08,
      "loss": 0.1707,
      "step": 7031
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8119875457454561,
      "learning_rate": 7.015584334412962e-08,
      "loss": 0.1203,
      "step": 7032
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7069691820667958,
      "learning_rate": 6.979177543257221e-08,
      "loss": 0.0931,
      "step": 7033
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.04679186233318,
      "learning_rate": 6.942864799087012e-08,
      "loss": 0.19,
      "step": 7034
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1796330774852635,
      "learning_rate": 6.906646108829241e-08,
      "loss": 0.2154,
      "step": 7035
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0087850062045989,
      "learning_rate": 6.870521479392877e-08,
      "loss": 0.1623,
      "step": 7036
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1124501498450161,
      "learning_rate": 6.834490917668857e-08,
      "loss": 0.1719,
      "step": 7037
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2013818197090123,
      "learning_rate": 6.798554430530291e-08,
      "loss": 0.1917,
      "step": 7038
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1379517922246167,
      "learning_rate": 6.762712024832307e-08,
      "loss": 0.1775,
      "step": 7039
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8633570667462709,
      "learning_rate": 6.726963707412105e-08,
      "loss": 0.1271,
      "step": 7040
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7673023260235989,
      "learning_rate": 6.691309485088893e-08,
      "loss": 0.143,
      "step": 7041
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9301210642044782,
      "learning_rate": 6.655749364663899e-08,
      "loss": 0.1565,
      "step": 7042
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8745504846071491,
      "learning_rate": 6.620283352920532e-08,
      "loss": 0.1409,
      "step": 7043
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9953145554382922,
      "learning_rate": 6.584911456624155e-08,
      "loss": 0.1722,
      "step": 7044
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8967535253780745,
      "learning_rate": 6.54963368252215e-08,
      "loss": 0.1488,
      "step": 7045
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6057901333070779,
      "learning_rate": 6.514450037343967e-08,
      "loss": 0.1263,
      "step": 7046
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1072689922349281,
      "learning_rate": 6.479360527801182e-08,
      "loss": 0.1728,
      "step": 7047
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9055335826003975,
      "learning_rate": 6.444365160587329e-08,
      "loss": 0.1467,
      "step": 7048
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0323953404476154,
      "learning_rate": 6.409463942378013e-08,
      "loss": 0.1927,
      "step": 7049
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9973299311542495,
      "learning_rate": 6.374656879830854e-08,
      "loss": 0.1492,
      "step": 7050
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2156904163525593,
      "learning_rate": 6.339943979585539e-08,
      "loss": 0.2485,
      "step": 7051
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.083884964488857,
      "learning_rate": 6.305325248263772e-08,
      "loss": 0.1563,
      "step": 7052
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9798685167991897,
      "learning_rate": 6.270800692469326e-08,
      "loss": 0.1885,
      "step": 7053
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0310496609265638,
      "learning_rate": 6.236370318787987e-08,
      "loss": 0.1717,
      "step": 7054
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0047648088706729,
      "learning_rate": 6.20203413378756e-08,
      "loss": 0.1657,
      "step": 7055
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9563596982256766,
      "learning_rate": 6.167792144017914e-08,
      "loss": 0.1797,
      "step": 7056
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0990766121585174,
      "learning_rate": 6.133644356010937e-08,
      "loss": 0.174,
      "step": 7057
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5602632717713026,
      "learning_rate": 6.099590776280528e-08,
      "loss": 0.0646,
      "step": 7058
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.05487130623426,
      "learning_rate": 6.065631411322603e-08,
      "loss": 0.1769,
      "step": 7059
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9480356050453882,
      "learning_rate": 6.031766267615258e-08,
      "loss": 0.177,
      "step": 7060
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1392833493943963,
      "learning_rate": 5.997995351618324e-08,
      "loss": 0.2103,
      "step": 7061
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1229122868914487,
      "learning_rate": 5.964318669773983e-08,
      "loss": 0.2067,
      "step": 7062
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8403392117757154,
      "learning_rate": 5.9307362285061506e-08,
      "loss": 0.1282,
      "step": 7063
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9586021300076366,
      "learning_rate": 5.8972480342209815e-08,
      "loss": 0.1608,
      "step": 7064
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0175765619891144,
      "learning_rate": 5.863854093306587e-08,
      "loss": 0.1955,
      "step": 7065
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9411246167175679,
      "learning_rate": 5.830554412133038e-08,
      "loss": 0.15,
      "step": 7066
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.138469488854617,
      "learning_rate": 5.7973489970524765e-08,
      "loss": 0.1751,
      "step": 7067
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9099681757254275,
      "learning_rate": 5.7642378543990574e-08,
      "loss": 0.151,
      "step": 7068
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9639249396763565,
      "learning_rate": 5.731220990488951e-08,
      "loss": 0.1414,
      "step": 7069
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1143911186013638,
      "learning_rate": 5.6982984116203957e-08,
      "loss": 0.1958,
      "step": 7070
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9863776859870137,
      "learning_rate": 5.66547012407348e-08,
      "loss": 0.1818,
      "step": 7071
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0352749073467844,
      "learning_rate": 5.632736134110528e-08,
      "loss": 0.1534,
      "step": 7072
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8887867656221394,
      "learning_rate": 5.600096447975656e-08,
      "loss": 0.1433,
      "step": 7073
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8658589670185685,
      "learning_rate": 5.567551071895161e-08,
      "loss": 0.1621,
      "step": 7074
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8960107262050628,
      "learning_rate": 5.535100012077299e-08,
      "loss": 0.1147,
      "step": 7075
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.917790257757191,
      "learning_rate": 5.5027432747122856e-08,
      "loss": 0.1316,
      "step": 7076
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8314344666687191,
      "learning_rate": 5.4704808659724605e-08,
      "loss": 0.1594,
      "step": 7077
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0605990609014848,
      "learning_rate": 5.438312792012013e-08,
      "loss": 0.1715,
      "step": 7078
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0792548561402124,
      "learning_rate": 5.4062390589671995e-08,
      "loss": 0.1647,
      "step": 7079
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1977355579808646,
      "learning_rate": 5.374259672956405e-08,
      "loss": 0.1674,
      "step": 7080
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1260653423575646,
      "learning_rate": 5.342374640079806e-08,
      "loss": 0.1837,
      "step": 7081
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1443169958670973,
      "learning_rate": 5.310583966419758e-08,
      "loss": 0.1665,
      "step": 7082
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3508639457021028,
      "learning_rate": 5.2788876580404655e-08,
      "loss": 0.1945,
      "step": 7083
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8851511866204479,
      "learning_rate": 5.24728572098826e-08,
      "loss": 0.1132,
      "step": 7084
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6638674643897843,
      "learning_rate": 5.2157781612914294e-08,
      "loss": 0.1386,
      "step": 7085
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9774552805776522,
      "learning_rate": 5.184364984960222e-08,
      "loss": 0.2021,
      "step": 7086
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.577528746012248,
      "learning_rate": 5.1530461979869e-08,
      "loss": 0.1001,
      "step": 7087
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0316230000814415,
      "learning_rate": 5.121821806345739e-08,
      "loss": 0.1916,
      "step": 7088
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0083787115060228,
      "learning_rate": 5.0906918159930317e-08,
      "loss": 0.1547,
      "step": 7089
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1536402655496538,
      "learning_rate": 5.0596562328670276e-08,
      "loss": 0.1826,
      "step": 7090
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.230927145391705,
      "learning_rate": 5.028715062887879e-08,
      "loss": 0.2176,
      "step": 7091
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2918328757811322,
      "learning_rate": 4.997868311957921e-08,
      "loss": 0.2439,
      "step": 7092
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8290716527260996,
      "learning_rate": 4.967115985961335e-08,
      "loss": 0.1386,
      "step": 7093
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8768183180387156,
      "learning_rate": 4.936458090764373e-08,
      "loss": 0.1652,
      "step": 7094
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7566050134981219,
      "learning_rate": 4.905894632215136e-08,
      "loss": 0.126,
      "step": 7095
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9282169703255335,
      "learning_rate": 4.875425616143903e-08,
      "loss": 0.1553,
      "step": 7096
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8427655628596161,
      "learning_rate": 4.8450510483627475e-08,
      "loss": 0.1537,
      "step": 7097
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.137470986325778,
      "learning_rate": 4.8147709346659244e-08,
      "loss": 0.1945,
      "step": 7098
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1352126481513605,
      "learning_rate": 4.784585280829479e-08,
      "loss": 0.1565,
      "step": 7099
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1504103441553881,
      "learning_rate": 4.754494092611639e-08,
      "loss": 0.2085,
      "step": 7100
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9968633346442629,
      "learning_rate": 4.7244973757523685e-08,
      "loss": 0.2032,
      "step": 7101
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0165172992525902,
      "learning_rate": 4.694595135973756e-08,
      "loss": 0.1724,
      "step": 7102
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0122556234174085,
      "learning_rate": 4.664787378979907e-08,
      "loss": 0.1634,
      "step": 7103
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9293130591260527,
      "learning_rate": 4.635074110456883e-08,
      "loss": 0.1819,
      "step": 7104
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5769741320825271,
      "learning_rate": 4.6054553360725396e-08,
      "loss": 0.1143,
      "step": 7105
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1041168200322484,
      "learning_rate": 4.575931061477024e-08,
      "loss": 0.1956,
      "step": 7106
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.781405133150644,
      "learning_rate": 4.546501292302164e-08,
      "loss": 0.1242,
      "step": 7107
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9367363618602934,
      "learning_rate": 4.517166034161913e-08,
      "loss": 0.1819,
      "step": 7108
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9798406071983519,
      "learning_rate": 4.487925292652184e-08,
      "loss": 0.1696,
      "step": 7109
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.22317732364511,
      "learning_rate": 4.458779073350905e-08,
      "loss": 0.2078,
      "step": 7110
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1818405245834898,
      "learning_rate": 4.429727381817794e-08,
      "loss": 0.1939,
      "step": 7111
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7829602075323107,
      "learning_rate": 4.400770223594697e-08,
      "loss": 0.14,
      "step": 7112
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0664728644433223,
      "learning_rate": 4.371907604205417e-08,
      "loss": 0.1581,
      "step": 7113
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9623471058950345,
      "learning_rate": 4.34313952915566e-08,
      "loss": 0.162,
      "step": 7114
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8707344576490329,
      "learning_rate": 4.3144660039330934e-08,
      "loss": 0.1626,
      "step": 7115
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6640611329853282,
      "learning_rate": 4.285887034007452e-08,
      "loss": 0.1411,
      "step": 7116
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9995482486343169,
      "learning_rate": 4.2574026248303754e-08,
      "loss": 0.1438,
      "step": 7117
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.764746758357648,
      "learning_rate": 4.229012781835351e-08,
      "loss": 0.135,
      "step": 7118
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.042184059637096,
      "learning_rate": 4.200717510437935e-08,
      "loss": 0.1845,
      "step": 7119
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8231663939172652,
      "learning_rate": 4.172516816035754e-08,
      "loss": 0.1189,
      "step": 7120
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1482020077152615,
      "learning_rate": 4.1444107040081726e-08,
      "loss": 0.2023,
      "step": 7121
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5814978856094835,
      "learning_rate": 4.116399179716679e-08,
      "loss": 0.0988,
      "step": 7122
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0806382615287595,
      "learning_rate": 4.088482248504666e-08,
      "loss": 0.1924,
      "step": 7123
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6798039409914879,
      "learning_rate": 4.0606599156973736e-08,
      "loss": 0.1137,
      "step": 7124
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.071348682999224,
      "learning_rate": 4.032932186602223e-08,
      "loss": 0.1698,
      "step": 7125
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0809547600928047,
      "learning_rate": 4.0052990665083724e-08,
      "loss": 0.1758,
      "step": 7126
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8891227682595524,
      "learning_rate": 3.9777605606870497e-08,
      "loss": 0.1429,
      "step": 7127
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8231024386783966,
      "learning_rate": 3.9503166743914415e-08,
      "loss": 0.1662,
      "step": 7128
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9594891247269897,
      "learning_rate": 3.922967412856582e-08,
      "loss": 0.1647,
      "step": 7129
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9131795656641276,
      "learning_rate": 3.895712781299576e-08,
      "loss": 0.1652,
      "step": 7130
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7228939317846501,
      "learning_rate": 3.868552784919433e-08,
      "loss": 0.1246,
      "step": 7131
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0129629393894297,
      "learning_rate": 3.8414874288970613e-08,
      "loss": 0.1874,
      "step": 7132
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.067232098632618,
      "learning_rate": 3.814516718395389e-08,
      "loss": 0.1718,
      "step": 7133
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.816623799691802,
      "learning_rate": 3.787640658559244e-08,
      "loss": 0.1104,
      "step": 7134
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7774751482357395,
      "learning_rate": 3.7608592545153586e-08,
      "loss": 0.12,
      "step": 7135
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0292186791493732,
      "learning_rate": 3.734172511372591e-08,
      "loss": 0.1623,
      "step": 7136
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.797518342786685,
      "learning_rate": 3.707580434221481e-08,
      "loss": 0.125,
      "step": 7137
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.052172413316302,
      "learning_rate": 3.681083028134746e-08,
      "loss": 0.1918,
      "step": 7138
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9519166433316231,
      "learning_rate": 3.654680298166846e-08,
      "loss": 0.1474,
      "step": 7139
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8345894715268443,
      "learning_rate": 3.628372249354362e-08,
      "loss": 0.1558,
      "step": 7140
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8075613870443407,
      "learning_rate": 3.6021588867156674e-08,
      "loss": 0.1315,
      "step": 7141
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1038562707070174,
      "learning_rate": 3.576040215251209e-08,
      "loss": 0.2047,
      "step": 7142
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9076058316018562,
      "learning_rate": 3.5500162399432216e-08,
      "loss": 0.116,
      "step": 7143
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0130202700218056,
      "learning_rate": 3.524086965755957e-08,
      "loss": 0.1608,
      "step": 7144
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1517224146659804,
      "learning_rate": 3.498252397635626e-08,
      "loss": 0.1573,
      "step": 7145
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.04800406541103,
      "learning_rate": 3.472512540510342e-08,
      "loss": 0.1801,
      "step": 7146
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1235299369201044,
      "learning_rate": 3.446867399290177e-08,
      "loss": 0.1915,
      "step": 7147
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.82576423000679,
      "learning_rate": 3.421316978867106e-08,
      "loss": 0.1177,
      "step": 7148
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2336444681280472,
      "learning_rate": 3.395861284114954e-08,
      "loss": 0.178,
      "step": 7149
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0511690062893986,
      "learning_rate": 3.3705003198896135e-08,
      "loss": 0.1353,
      "step": 7150
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1623202608562717,
      "learning_rate": 3.3452340910288815e-08,
      "loss": 0.1662,
      "step": 7151
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8899130861015885,
      "learning_rate": 3.320062602352459e-08,
      "loss": 0.1623,
      "step": 7152
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.880947129280693,
      "learning_rate": 3.294985858661948e-08,
      "loss": 0.1659,
      "step": 7153
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.111357308523082,
      "learning_rate": 3.270003864740911e-08,
      "loss": 0.1985,
      "step": 7154
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2365417875639373,
      "learning_rate": 3.245116625354816e-08,
      "loss": 0.1787,
      "step": 7155
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9860269263970403,
      "learning_rate": 3.220324145251141e-08,
      "loss": 0.1624,
      "step": 7156
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1226807968652617,
      "learning_rate": 3.195626429159104e-08,
      "loss": 0.1942,
      "step": 7157
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9293532972832927,
      "learning_rate": 3.171023481790048e-08,
      "loss": 0.1474,
      "step": 7158
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.13725811102622,
      "learning_rate": 3.146515307837106e-08,
      "loss": 0.1948,
      "step": 7159
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7634710690128641,
      "learning_rate": 3.1221019119753725e-08,
      "loss": 0.1375,
      "step": 7160
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7103459842648718,
      "learning_rate": 3.097783298861845e-08,
      "loss": 0.1235,
      "step": 7161
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8916320531948476,
      "learning_rate": 3.0735594731355346e-08,
      "loss": 0.133,
      "step": 7162
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0495667457310545,
      "learning_rate": 3.0494304394172446e-08,
      "loss": 0.1681,
      "step": 7163
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9538875134378775,
      "learning_rate": 3.0253962023097386e-08,
      "loss": 0.1546,
      "step": 7164
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7676624392891073,
      "learning_rate": 3.001456766397737e-08,
      "loss": 0.1301,
      "step": 7165
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8575544276570009,
      "learning_rate": 2.9776121362478095e-08,
      "loss": 0.1255,
      "step": 7166
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2077879297113376,
      "learning_rate": 2.95386231640854e-08,
      "loss": 0.1803,
      "step": 7167
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1869999781766283,
      "learning_rate": 2.9302073114103047e-08,
      "loss": 0.2136,
      "step": 7168
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0599898584056127,
      "learning_rate": 2.9066471257654384e-08,
      "loss": 0.1879,
      "step": 7169
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0912505396712935,
      "learning_rate": 2.8831817639682347e-08,
      "loss": 0.1818,
      "step": 7170
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2673451256923198,
      "learning_rate": 2.859811230494891e-08,
      "loss": 0.2063,
      "step": 7171
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7425769934436572,
      "learning_rate": 2.8365355298034525e-08,
      "loss": 0.1129,
      "step": 7172
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5497071382974267,
      "learning_rate": 2.8133546663339228e-08,
      "loss": 0.1447,
      "step": 7173
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7809700254729391,
      "learning_rate": 2.7902686445082648e-08,
      "loss": 0.1332,
      "step": 7174
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7839604332547094,
      "learning_rate": 2.767277468730123e-08,
      "loss": 0.1244,
      "step": 7175
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9290002981960046,
      "learning_rate": 2.744381143385433e-08,
      "loss": 0.1276,
      "step": 7176
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6519756782079191,
      "learning_rate": 2.7215796728416453e-08,
      "loss": 0.1087,
      "step": 7177
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8148680928790976,
      "learning_rate": 2.6988730614483927e-08,
      "loss": 0.1581,
      "step": 7178
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.01240355340718,
      "learning_rate": 2.6762613135370984e-08,
      "loss": 0.1533,
      "step": 7179
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9768972303493653,
      "learning_rate": 2.653744433421035e-08,
      "loss": 0.1506,
      "step": 7180
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0130511256957622,
      "learning_rate": 2.6313224253955437e-08,
      "loss": 0.2188,
      "step": 7181
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0026290211568003,
      "learning_rate": 2.608995293737704e-08,
      "loss": 0.171,
      "step": 7182
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9479050653312758,
      "learning_rate": 2.5867630427065527e-08,
      "loss": 0.136,
      "step": 7183
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1745330928350917,
      "learning_rate": 2.564625676543142e-08,
      "loss": 0.2282,
      "step": 7184
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.750369407768691,
      "learning_rate": 2.5425831994702055e-08,
      "loss": 0.0973,
      "step": 7185
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1555046253244834,
      "learning_rate": 2.5206356156924906e-08,
      "loss": 0.1633,
      "step": 7186
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9774818080683998,
      "learning_rate": 2.4987829293967036e-08,
      "loss": 0.1797,
      "step": 7187
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.677692783631185,
      "learning_rate": 2.4770251447513992e-08,
      "loss": 0.1178,
      "step": 7188
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7593551966305777,
      "learning_rate": 2.4553622659070352e-08,
      "loss": 0.1131,
      "step": 7189
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9900820148602745,
      "learning_rate": 2.4337942969958618e-08,
      "loss": 0.1621,
      "step": 7190
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9186930840477887,
      "learning_rate": 2.412321242132143e-08,
      "loss": 0.1347,
      "step": 7191
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8633800034696428,
      "learning_rate": 2.390943105412047e-08,
      "loss": 0.1396,
      "step": 7192
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8770752937983549,
      "learning_rate": 2.3696598909135336e-08,
      "loss": 0.1334,
      "step": 7193
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7448222391558278,
      "learning_rate": 2.348471602696578e-08,
      "loss": 0.1211,
      "step": 7194
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8979615622741122,
      "learning_rate": 2.3273782448030024e-08,
      "loss": 0.1711,
      "step": 7195
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2793402219024927,
      "learning_rate": 2.3063798212564214e-08,
      "loss": 0.2351,
      "step": 7196
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0648073295227254,
      "learning_rate": 2.2854763360624088e-08,
      "loss": 0.19,
      "step": 7197
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.114090649327106,
      "learning_rate": 2.2646677932085524e-08,
      "loss": 0.1891,
      "step": 7198
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1266808776654769,
      "learning_rate": 2.2439541966641775e-08,
      "loss": 0.152,
      "step": 7199
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7972374345818846,
      "learning_rate": 2.223335550380512e-08,
      "loss": 0.1495,
      "step": 7200
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.010171096651144,
      "learning_rate": 2.2028118582906877e-08,
      "loss": 0.1682,
      "step": 7201
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.842067508137843,
      "learning_rate": 2.182383124309795e-08,
      "loss": 0.1332,
      "step": 7202
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9704381048106576,
      "learning_rate": 2.1620493523347164e-08,
      "loss": 0.1488,
      "step": 7203
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1933360995515734,
      "learning_rate": 2.1418105462442383e-08,
      "loss": 0.2011,
      "step": 7204
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.089040431988871,
      "learning_rate": 2.1216667098990508e-08,
      "loss": 0.1362,
      "step": 7205
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0373171487015365,
      "learning_rate": 2.101617847141746e-08,
      "loss": 0.1914,
      "step": 7206
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7544403257449053,
      "learning_rate": 2.0816639617967648e-08,
      "loss": 0.1319,
      "step": 7207
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0375810487948913,
      "learning_rate": 2.061805057670452e-08,
      "loss": 0.1614,
      "step": 7208
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.027484278318831,
      "learning_rate": 2.042041138551054e-08,
      "loss": 0.1617,
      "step": 7209
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9209026914053957,
      "learning_rate": 2.0223722082085563e-08,
      "loss": 0.1543,
      "step": 7210
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0514316699325272,
      "learning_rate": 2.0027982703950676e-08,
      "loss": 0.1488,
      "step": 7211
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0098928158279667,
      "learning_rate": 1.9833193288443797e-08,
      "loss": 0.1685,
      "step": 7212
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8569993196651691,
      "learning_rate": 1.9639353872722422e-08,
      "loss": 0.1228,
      "step": 7213
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9464963849491995,
      "learning_rate": 1.944646449376253e-08,
      "loss": 0.1454,
      "step": 7214
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6481379278355931,
      "learning_rate": 1.9254525188359686e-08,
      "loss": 0.1375,
      "step": 7215
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9326711465592867,
      "learning_rate": 1.9063535993126826e-08,
      "loss": 0.1711,
      "step": 7216
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7846424844874648,
      "learning_rate": 1.887349694449647e-08,
      "loss": 0.1455,
      "step": 7217
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.040236365964663,
      "learning_rate": 1.868440807872074e-08,
      "loss": 0.2096,
      "step": 7218
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2277799145822967,
      "learning_rate": 1.8496269431867996e-08,
      "loss": 0.1917,
      "step": 7219
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.087620456197298,
      "learning_rate": 1.8309081039828424e-08,
      "loss": 0.1813,
      "step": 7220
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.988527968024858,
      "learning_rate": 1.812284293830846e-08,
      "loss": 0.1693,
      "step": 7221
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.174760467243767,
      "learning_rate": 1.793755516283524e-08,
      "loss": 0.1696,
      "step": 7222
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.856082902665888,
      "learning_rate": 1.7753217748752717e-08,
      "loss": 0.1444,
      "step": 7223
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2434046270594465,
      "learning_rate": 1.7569830731224978e-08,
      "loss": 0.2321,
      "step": 7224
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9672506749297931,
      "learning_rate": 1.7387394145234048e-08,
      "loss": 0.1582,
      "step": 7225
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9261964630001277,
      "learning_rate": 1.7205908025580976e-08,
      "loss": 0.1554,
      "step": 7226
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7729255813328016,
      "learning_rate": 1.7025372406885287e-08,
      "loss": 0.1316,
      "step": 7227
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0059223979963525,
      "learning_rate": 1.684578732358555e-08,
      "loss": 0.1569,
      "step": 7228
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0446198285826305,
      "learning_rate": 1.6667152809938802e-08,
      "loss": 0.1595,
      "step": 7229
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0095543271374934,
      "learning_rate": 1.648946890002112e-08,
      "loss": 0.1918,
      "step": 7230
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3723155739933133,
      "learning_rate": 1.631273562772595e-08,
      "loss": 0.2075,
      "step": 7231
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9762424147694152,
      "learning_rate": 1.613695302676688e-08,
      "loss": 0.1497,
      "step": 7232
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.029560168958185,
      "learning_rate": 1.5962121130675968e-08,
      "loss": 0.1285,
      "step": 7233
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5832255259110695,
      "learning_rate": 1.578823997280321e-08,
      "loss": 0.0971,
      "step": 7234
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7837626497995986,
      "learning_rate": 1.561530958631763e-08,
      "loss": 0.1148,
      "step": 7235
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9782952044692776,
      "learning_rate": 1.5443330004206724e-08,
      "loss": 0.1179,
      "step": 7236
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9750351367830732,
      "learning_rate": 1.527230125927648e-08,
      "loss": 0.1588,
      "step": 7237
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.289382533541146,
      "learning_rate": 1.510222338415246e-08,
      "loss": 0.1741,
      "step": 7238
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9982328209354667,
      "learning_rate": 1.4933096411277605e-08,
      "loss": 0.1695,
      "step": 7239
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.04918327919573,
      "learning_rate": 1.4764920372914437e-08,
      "loss": 0.2032,
      "step": 7240
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8385409122443841,
      "learning_rate": 1.459769530114341e-08,
      "loss": 0.1213,
      "step": 7241
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9173073402864916,
      "learning_rate": 1.4431421227863452e-08,
      "loss": 0.1477,
      "step": 7242
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9260723412031067,
      "learning_rate": 1.4266098184792521e-08,
      "loss": 0.1259,
      "step": 7243
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1246784388623132,
      "learning_rate": 1.410172620346817e-08,
      "loss": 0.1927,
      "step": 7244
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7363178390574152,
      "learning_rate": 1.3938305315244204e-08,
      "loss": 0.1188,
      "step": 7245
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.187290787046644,
      "learning_rate": 1.377583555129458e-08,
      "loss": 0.1643,
      "step": 7246
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8644567440517933,
      "learning_rate": 1.3614316942611172e-08,
      "loss": 0.1745,
      "step": 7247
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1389402791725147,
      "learning_rate": 1.3453749520005999e-08,
      "loss": 0.1849,
      "step": 7248
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5376808021453415,
      "learning_rate": 1.3294133314106784e-08,
      "loss": 0.1154,
      "step": 7249
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.964977313300608,
      "learning_rate": 1.3135468355361946e-08,
      "loss": 0.1339,
      "step": 7250
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8092263798219078,
      "learning_rate": 1.2977754674038389e-08,
      "loss": 0.1589,
      "step": 7251
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9753080140822372,
      "learning_rate": 1.2820992300220381e-08,
      "loss": 0.1768,
      "step": 7252
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.157298772867146,
      "learning_rate": 1.2665181263812332e-08,
      "loss": 0.151,
      "step": 7253
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8593569325985204,
      "learning_rate": 1.2510321594534913e-08,
      "loss": 0.1592,
      "step": 7254
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8763379770282465,
      "learning_rate": 1.235641332192894e-08,
      "loss": 0.1391,
      "step": 7255
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1127012964508203,
      "learning_rate": 1.2203456475354258e-08,
      "loss": 0.2151,
      "step": 7256
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9515050685915679,
      "learning_rate": 1.2051451083988086e-08,
      "loss": 0.146,
      "step": 7257
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6718886105895752,
      "learning_rate": 1.1900397176825562e-08,
      "loss": 0.1052,
      "step": 7258
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0012577090661603,
      "learning_rate": 1.1750294782682525e-08,
      "loss": 0.1584,
      "step": 7259
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.849355626647822,
      "learning_rate": 1.1601143930191072e-08,
      "loss": 0.1121,
      "step": 7260
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9037331862099334,
      "learning_rate": 1.1452944647802889e-08,
      "loss": 0.1462,
      "step": 7261
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1097003924124773,
      "learning_rate": 1.1305696963788693e-08,
      "loss": 0.1887,
      "step": 7262
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.124875701096774,
      "learning_rate": 1.1159400906236018e-08,
      "loss": 0.1615,
      "step": 7263
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9528369971918839,
      "learning_rate": 1.1014056503051984e-08,
      "loss": 0.1464,
      "step": 7264
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8427228653486243,
      "learning_rate": 1.0869663781962191e-08,
      "loss": 0.1261,
      "step": 7265
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9144785428667102,
      "learning_rate": 1.072622277051072e-08,
      "loss": 0.1345,
      "step": 7266
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7500716329958492,
      "learning_rate": 1.0583733496059566e-08,
      "loss": 0.1432,
      "step": 7267
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0088395873869167,
      "learning_rate": 1.0442195985789771e-08,
      "loss": 0.1641,
      "step": 7268
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1239988962832834,
      "learning_rate": 1.0301610266700291e-08,
      "loss": 0.18,
      "step": 7269
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8319645276727875,
      "learning_rate": 1.0161976365609117e-08,
      "loss": 0.1256,
      "step": 7270
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0874139074927234,
      "learning_rate": 1.002329430915161e-08,
      "loss": 0.1777,
      "step": 7271
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8915205529451817,
      "learning_rate": 9.885564123783276e-09,
      "loss": 0.1141,
      "step": 7272
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.092484900392333,
      "learning_rate": 9.748785835776431e-09,
      "loss": 0.1911,
      "step": 7273
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0762850586966235,
      "learning_rate": 9.612959471222428e-09,
      "loss": 0.16,
      "step": 7274
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2123503611139166,
      "learning_rate": 9.478085056031095e-09,
      "loss": 0.1792,
      "step": 7275
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8380386613981863,
      "learning_rate": 9.34416261593074e-09,
      "loss": 0.1535,
      "step": 7276
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9871674725544453,
      "learning_rate": 9.211192176468153e-09,
      "loss": 0.1628,
      "step": 7277
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8228464023622025,
      "learning_rate": 9.079173763007488e-09,
      "loss": 0.1461,
      "step": 7278
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8482191488217737,
      "learning_rate": 8.948107400733596e-09,
      "loss": 0.1664,
      "step": 7279
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8540544718796427,
      "learning_rate": 8.817993114647039e-09,
      "loss": 0.1423,
      "step": 7280
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3316424281721553,
      "learning_rate": 8.68883092956796e-09,
      "loss": 0.2056,
      "step": 7281
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7698594825504441,
      "learning_rate": 8.560620870136094e-09,
      "loss": 0.1205,
      "step": 7282
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9917504534722746,
      "learning_rate": 8.433362960806879e-09,
      "loss": 0.1349,
      "step": 7283
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1031158520198243,
      "learning_rate": 8.307057225856451e-09,
      "loss": 0.1633,
      "step": 7284
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0593188248946113,
      "learning_rate": 8.181703689378318e-09,
      "loss": 0.1819,
      "step": 7285
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8318870856893527,
      "learning_rate": 8.05730237528446e-09,
      "loss": 0.1257,
      "step": 7286
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9516334213539437,
      "learning_rate": 7.93385330730534e-09,
      "loss": 0.1761,
      "step": 7287
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.024260309863008,
      "learning_rate": 7.811356508989343e-09,
      "loss": 0.1852,
      "step": 7288
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3534508501554439,
      "learning_rate": 7.689812003703889e-09,
      "loss": 0.2047,
      "step": 7289
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6709242278965117,
      "learning_rate": 7.569219814634877e-09,
      "loss": 0.1064,
      "step": 7290
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9889618753104136,
      "learning_rate": 7.4495799647850145e-09,
      "loss": 0.1622,
      "step": 7291
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7780964501208475,
      "learning_rate": 7.330892476976603e-09,
      "loss": 0.1603,
      "step": 7292
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9836105095780547,
      "learning_rate": 7.2131573738504215e-09,
      "loss": 0.1263,
      "step": 7293
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9207374576315601,
      "learning_rate": 7.096374677865725e-09,
      "loss": 0.1255,
      "step": 7294
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6140018417570645,
      "learning_rate": 6.980544411298584e-09,
      "loss": 0.2622,
      "step": 7295
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9766592591255876,
      "learning_rate": 6.865666596244658e-09,
      "loss": 0.1588,
      "step": 7296
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9876291005244302,
      "learning_rate": 6.7517412546180824e-09,
      "loss": 0.1427,
      "step": 7297
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9901286785854136,
      "learning_rate": 6.638768408150364e-09,
      "loss": 0.1608,
      "step": 7298
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2148659920416132,
      "learning_rate": 6.5267480783920425e-09,
      "loss": 0.125,
      "step": 7299
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9811744867323801,
      "learning_rate": 6.4156802867121334e-09,
      "loss": 0.1228,
      "step": 7300
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9399186308217071,
      "learning_rate": 6.305565054296469e-09,
      "loss": 0.1731,
      "step": 7301
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.735161724371514,
      "learning_rate": 6.19640240215158e-09,
      "loss": 0.1387,
      "step": 7302
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0750703866323073,
      "learning_rate": 6.0881923511002525e-09,
      "loss": 0.1477,
      "step": 7303
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8462896118346415,
      "learning_rate": 5.980934921783754e-09,
      "loss": 0.1392,
      "step": 7304
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8748958631299202,
      "learning_rate": 5.874630134663495e-09,
      "loss": 0.1692,
      "step": 7305
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0011727154256957,
      "learning_rate": 5.769278010016588e-09,
      "loss": 0.1758,
      "step": 7306
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8701555751928727,
      "learning_rate": 5.664878567940291e-09,
      "loss": 0.1355,
      "step": 7307
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.847642738538414,
      "learning_rate": 5.561431828349229e-09,
      "loss": 0.1187,
      "step": 7308
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0211249685906114,
      "learning_rate": 5.458937810977061e-09,
      "loss": 0.1819,
      "step": 7309
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.095285407817043,
      "learning_rate": 5.35739653537426e-09,
      "loss": 0.1901,
      "step": 7310
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7631258917254583,
      "learning_rate": 5.256808020911441e-09,
      "loss": 0.1099,
      "step": 7311
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9709794182574086,
      "learning_rate": 5.15717228677548e-09,
      "loss": 0.1713,
      "step": 7312
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7240899504050898,
      "learning_rate": 5.05848935197395e-09,
      "loss": 0.0958,
      "step": 7313
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9478565351997758,
      "learning_rate": 4.960759235330126e-09,
      "loss": 0.1723,
      "step": 7314
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0383124331202234,
      "learning_rate": 4.863981955486874e-09,
      "loss": 0.181,
      "step": 7315
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9119123475860408,
      "learning_rate": 4.768157530905537e-09,
      "loss": 0.1336,
      "step": 7316
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1138665966261432,
      "learning_rate": 4.6732859798653825e-09,
      "loss": 0.1664,
      "step": 7317
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8862830376148857,
      "learning_rate": 4.579367320462491e-09,
      "loss": 0.1634,
      "step": 7318
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8772252008345854,
      "learning_rate": 4.486401570614196e-09,
      "loss": 0.1598,
      "step": 7319
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8481296613885574,
      "learning_rate": 4.39438874805298e-09,
      "loss": 0.114,
      "step": 7320
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0606002998023984,
      "learning_rate": 4.3033288703320245e-09,
      "loss": 0.1448,
      "step": 7321
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1362527032118839,
      "learning_rate": 4.213221954820212e-09,
      "loss": 0.187,
      "step": 7322
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0233662832087336,
      "learning_rate": 4.124068018707683e-09,
      "loss": 0.1689,
      "step": 7323
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7274149033223958,
      "learning_rate": 4.035867078999723e-09,
      "loss": 0.1442,
      "step": 7324
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0521711151182591,
      "learning_rate": 3.94861915252176e-09,
      "loss": 0.1665,
      "step": 7325
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9383749304079361,
      "learning_rate": 3.862324255917704e-09,
      "loss": 0.1755,
      "step": 7326
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1633300745439148,
      "learning_rate": 3.776982405647167e-09,
      "loss": 0.1629,
      "step": 7327
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1981072672178041,
      "learning_rate": 3.692593617991569e-09,
      "loss": 0.2145,
      "step": 7328
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9794479382122182,
      "learning_rate": 3.6091579090469233e-09,
      "loss": 0.1816,
      "step": 7329
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9570975839741919,
      "learning_rate": 3.526675294730497e-09,
      "loss": 0.1642,
      "step": 7330
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8883787274567082,
      "learning_rate": 3.4451457907758167e-09,
      "loss": 0.1359,
      "step": 7331
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0110702097677413,
      "learning_rate": 3.364569412734886e-09,
      "loss": 0.1682,
      "step": 7332
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5581022894592724,
      "learning_rate": 3.2849461759781877e-09,
      "loss": 0.0757,
      "step": 7333
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9247373091213249,
      "learning_rate": 3.2062760956952377e-09,
      "loss": 0.1656,
      "step": 7334
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.050303215645153,
      "learning_rate": 3.128559186892366e-09,
      "loss": 0.1912,
      "step": 7335
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.833293330729349,
      "learning_rate": 3.0517954643943802e-09,
      "loss": 0.1576,
      "step": 7336
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8419055023895928,
      "learning_rate": 2.9759849428445674e-09,
      "loss": 0.1448,
      "step": 7337
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9032495486336977,
      "learning_rate": 2.9011276367041374e-09,
      "loss": 0.1832,
      "step": 7338
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6673626163816087,
      "learning_rate": 2.827223560253334e-09,
      "loss": 0.0769,
      "step": 7339
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0815135840921177,
      "learning_rate": 2.7542727275886583e-09,
      "loss": 0.1765,
      "step": 7340
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8561218748411754,
      "learning_rate": 2.682275152627312e-09,
      "loss": 0.1647,
      "step": 7341
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1610374096440863,
      "learning_rate": 2.6112308491021978e-09,
      "loss": 0.1961,
      "step": 7342
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6974386501106745,
      "learning_rate": 2.5411398305663636e-09,
      "loss": 0.0788,
      "step": 7343
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2176600254128493,
      "learning_rate": 2.4720021103891155e-09,
      "loss": 0.198,
      "step": 7344
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9570140820143032,
      "learning_rate": 2.403817701759903e-09,
      "loss": 0.1701,
      "step": 7345
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7857036663351461,
      "learning_rate": 2.336586617684988e-09,
      "loss": 0.1254,
      "step": 7346
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6775817002517139,
      "learning_rate": 2.270308870989113e-09,
      "loss": 0.1115,
      "step": 7347
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0876478113727341,
      "learning_rate": 2.2049844743149416e-09,
      "loss": 0.143,
      "step": 7348
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9473478118799338,
      "learning_rate": 2.1406134401241728e-09,
      "loss": 0.1605,
      "step": 7349
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9114193696716774,
      "learning_rate": 2.077195780695318e-09,
      "loss": 0.1744,
      "step": 7350
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9412166446615565,
      "learning_rate": 2.0147315081259223e-09,
      "loss": 0.143,
      "step": 7351
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9853472283261606,
      "learning_rate": 1.953220634332009e-09,
      "loss": 0.1777,
      "step": 7352
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7457698400318497,
      "learning_rate": 1.892663171046416e-09,
      "loss": 0.132,
      "step": 7353
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7697231425471306,
      "learning_rate": 1.8330591298215684e-09,
      "loss": 0.0865,
      "step": 7354
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9304385715754818,
      "learning_rate": 1.7744085220267049e-09,
      "loss": 0.171,
      "step": 7355
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9120212585183252,
      "learning_rate": 1.7167113588500984e-09,
      "loss": 0.1492,
      "step": 7356
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.196727409562461,
      "learning_rate": 1.6599676512979445e-09,
      "loss": 0.2217,
      "step": 7357
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.82212991324107,
      "learning_rate": 1.604177410194363e-09,
      "loss": 0.1141,
      "step": 7358
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0225809684560312,
      "learning_rate": 1.5493406461813965e-09,
      "loss": 0.1905,
      "step": 7359
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9313745254500596,
      "learning_rate": 1.4954573697206764e-09,
      "loss": 0.1504,
      "step": 7360
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9540168990297494,
      "learning_rate": 1.4425275910889824e-09,
      "loss": 0.1413,
      "step": 7361
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0895854662339497,
      "learning_rate": 1.390551320384903e-09,
      "loss": 0.1442,
      "step": 7362
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.740663540148301,
      "learning_rate": 1.3395285675216197e-09,
      "loss": 0.1118,
      "step": 7363
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0301102290498831,
      "learning_rate": 1.2894593422335677e-09,
      "loss": 0.1653,
      "step": 7364
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5927225765594528,
      "learning_rate": 1.2403436540703306e-09,
      "loss": 0.1281,
      "step": 7365
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9344783013356373,
      "learning_rate": 1.1921815124021906e-09,
      "loss": 0.1511,
      "step": 7366
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.272488954068171,
      "learning_rate": 1.1449729264156883e-09,
      "loss": 0.143,
      "step": 7367
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8732665830035861,
      "learning_rate": 1.0987179051169528e-09,
      "loss": 0.1412,
      "step": 7368
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0200599366518175,
      "learning_rate": 1.0534164573283712e-09,
      "loss": 0.1671,
      "step": 7369
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8726796458450906,
      "learning_rate": 1.0090685916924746e-09,
      "loss": 0.155,
      "step": 7370
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2194908087329297,
      "learning_rate": 9.656743166680526e-10,
      "loss": 0.2373,
      "step": 7371
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1611830769482292,
      "learning_rate": 9.23233640533483e-10,
      "loss": 0.1392,
      "step": 7372
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8643668038490419,
      "learning_rate": 8.817465713839568e-10,
      "loss": 0.1358,
      "step": 7373
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9479602972180314,
      "learning_rate": 8.412131171348093e-10,
      "loss": 0.1563,
      "step": 7374
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0099703459035925,
      "learning_rate": 8.016332855165232e-10,
      "loss": 0.1442,
      "step": 7375
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8511616543262075,
      "learning_rate": 7.630070840797255e-10,
      "loss": 0.1417,
      "step": 7376
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1072196153609812,
      "learning_rate": 7.253345201924111e-10,
      "loss": 0.1713,
      "step": 7377
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9585863667482212,
      "learning_rate": 6.88615601041609e-10,
      "loss": 0.1781,
      "step": 7378
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8103151370175125,
      "learning_rate": 6.528503336311609e-10,
      "loss": 0.1547,
      "step": 7379
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.265027966466253,
      "learning_rate": 6.180387247839426e-10,
      "loss": 0.196,
      "step": 7380
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2484218772950444,
      "learning_rate": 5.841807811396427e-10,
      "loss": 0.1844,
      "step": 7381
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1088687512433664,
      "learning_rate": 5.512765091575389e-10,
      "loss": 0.221,
      "step": 7382
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0203734573582837,
      "learning_rate": 5.193259151142771e-10,
      "loss": 0.1575,
      "step": 7383
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6966041603371325,
      "learning_rate": 4.883290051049816e-10,
      "loss": 0.0961,
      "step": 7384
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9273647952756037,
      "learning_rate": 4.5828578504159004e-10,
      "loss": 0.1899,
      "step": 7385
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0937148129790664,
      "learning_rate": 4.291962606556288e-10,
      "loss": 0.2017,
      "step": 7386
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2504495608465191,
      "learning_rate": 4.0106043749599256e-10,
      "loss": 0.2026,
      "step": 7387
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9078404868242123,
      "learning_rate": 3.738783209300545e-10,
      "loss": 0.1901,
      "step": 7388
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8615352085141567,
      "learning_rate": 3.4764991614255616e-10,
      "loss": 0.1256,
      "step": 7389
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9503880497530746,
      "learning_rate": 3.2237522813727273e-10,
      "loss": 0.1345,
      "step": 7390
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9190127242673465,
      "learning_rate": 2.9805426173479257e-10,
      "loss": 0.0995,
      "step": 7391
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0856927359291453,
      "learning_rate": 2.746870215752928e-10,
      "loss": 0.1671,
      "step": 7392
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8797208430439248,
      "learning_rate": 2.522735121157638e-10,
      "loss": 0.1502,
      "step": 7393
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1766230155997932,
      "learning_rate": 2.3081373763167436e-10,
      "loss": 0.2112,
      "step": 7394
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0034046694287782,
      "learning_rate": 2.1030770221641684e-10,
      "loss": 0.1717,
      "step": 7395
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0200840543863872,
      "learning_rate": 1.9075540978241714e-10,
      "loss": 0.1884,
      "step": 7396
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.14572582015939,
      "learning_rate": 1.7215686405891442e-10,
      "loss": 0.1946,
      "step": 7397
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0505690416300588,
      "learning_rate": 1.5451206859418143e-10,
      "loss": 0.1694,
      "step": 7398
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8229424702458777,
      "learning_rate": 1.378210267527491e-10,
      "loss": 0.186,
      "step": 7399
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7758891303816176,
      "learning_rate": 1.220837417204024e-10,
      "loss": 0.1561,
      "step": 7400
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.111874036608757,
      "learning_rate": 1.0730021649751899e-10,
      "loss": 0.1701,
      "step": 7401
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8714765681728576,
      "learning_rate": 9.347045390517561e-11,
      "loss": 0.1461,
      "step": 7402
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8452264516684526,
      "learning_rate": 8.059445658070708e-11,
      "loss": 0.1542,
      "step": 7403
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0527982172143069,
      "learning_rate": 6.867222698103693e-11,
      "loss": 0.1689,
      "step": 7404
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0765845134661567,
      "learning_rate": 5.770376738045702e-11,
      "loss": 0.1669,
      "step": 7405
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0996113919522528,
      "learning_rate": 4.768907987062754e-11,
      "loss": 0.1719,
      "step": 7406
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7948748653045761,
      "learning_rate": 3.8628166362242315e-11,
      "loss": 0.1764,
      "step": 7407
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1164557534378285,
      "learning_rate": 3.052102858336348e-11,
      "loss": 0.1277,
      "step": 7408
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0693356460774184,
      "learning_rate": 2.3367668081086813e-11,
      "loss": 0.1723,
      "step": 7409
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6104702938633813,
      "learning_rate": 1.7168086219876424e-11,
      "loss": 0.1016,
      "step": 7410
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8963425757551352,
      "learning_rate": 1.1922284181564714e-11,
      "loss": 0.1399,
      "step": 7411
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0374687282553519,
      "learning_rate": 7.630262968127966e-12,
      "loss": 0.1671,
      "step": 7412
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1719537044381103,
      "learning_rate": 4.292023397245437e-12,
      "loss": 0.2004,
      "step": 7413
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0656576361137509,
      "learning_rate": 1.9075661056300323e-12,
      "loss": 0.1787,
      "step": 7414
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3256955823670467,
      "learning_rate": 4.768915490283022e-13,
      "loss": 0.2237,
      "step": 7415
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.82913469848174,
      "learning_rate": 0.0,
      "loss": 0.1521,
      "step": 7416
    },
    {
      "epoch": 1.0,
      "step": 7416,
      "total_flos": 1350402630340608.0,
      "train_loss": 0.18874426638986816,
      "train_runtime": 19625.7622,
      "train_samples_per_second": 3.023,
      "train_steps_per_second": 0.378
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 7416,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "total_flos": 1350402630340608.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}