{
  "best_metric": 0.33440348505973816,
  "best_model_checkpoint": "final_models/lait_tha_mpt/checkpoint-10625",
  "epoch": 6.0,
  "eval_steps": 500,
  "global_step": 12750,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00047058823529411766,
      "grad_norm": 177.83850474140326,
      "learning_rate": 1.5e-07,
      "loss": 28.25,
      "step": 1
    },
    {
      "epoch": 0.0009411764705882353,
      "grad_norm": 63.61230097968654,
      "learning_rate": 3e-07,
      "loss": 12.0625,
      "step": 2
    },
    {
      "epoch": 0.001411764705882353,
      "grad_norm": 127.00871935299158,
      "learning_rate": 4.5e-07,
      "loss": 22.625,
      "step": 3
    },
    {
      "epoch": 0.0018823529411764706,
      "grad_norm": 61.40350935349764,
      "learning_rate": 6e-07,
      "loss": 11.75,
      "step": 4
    },
    {
      "epoch": 0.002352941176470588,
      "grad_norm": 124.34038471235046,
      "learning_rate": 7.499999999999999e-07,
      "loss": 22.25,
      "step": 5
    },
    {
      "epoch": 0.002823529411764706,
      "grad_norm": 133.25810511041888,
      "learning_rate": 9e-07,
      "loss": 21.875,
      "step": 6
    },
    {
      "epoch": 0.0032941176470588237,
      "grad_norm": 94.78102099919425,
      "learning_rate": 1.05e-06,
      "loss": 16.375,
      "step": 7
    },
    {
      "epoch": 0.0037647058823529413,
      "grad_norm": 141.8722944195454,
      "learning_rate": 1.2e-06,
      "loss": 23.375,
      "step": 8
    },
    {
      "epoch": 0.0042352941176470585,
      "grad_norm": 125.19277162321903,
      "learning_rate": 1.3499999999999998e-06,
      "loss": 20.625,
      "step": 9
    },
    {
      "epoch": 0.004705882352941176,
      "grad_norm": 161.13284067831705,
      "learning_rate": 1.4999999999999998e-06,
      "loss": 22.5,
      "step": 10
    },
    {
      "epoch": 0.0051764705882352945,
      "grad_norm": 172.7816158875274,
      "learning_rate": 1.6499999999999997e-06,
      "loss": 23.125,
      "step": 11
    },
    {
      "epoch": 0.005647058823529412,
      "grad_norm": 142.02111601787036,
      "learning_rate": 1.8e-06,
      "loss": 16.875,
      "step": 12
    },
    {
      "epoch": 0.00611764705882353,
      "grad_norm": 108.80893497878452,
      "learning_rate": 1.9499999999999995e-06,
      "loss": 12.0,
      "step": 13
    },
    {
      "epoch": 0.006588235294117647,
      "grad_norm": 116.9094700986542,
      "learning_rate": 2.1e-06,
      "loss": 9.8125,
      "step": 14
    },
    {
      "epoch": 0.007058823529411765,
      "grad_norm": 162.29362050230182,
      "learning_rate": 2.2499999999999996e-06,
      "loss": 13.125,
      "step": 15
    },
    {
      "epoch": 0.0075294117647058826,
      "grad_norm": 76.37176949853897,
      "learning_rate": 2.4e-06,
      "loss": 7.7812,
      "step": 16
    },
    {
      "epoch": 0.008,
      "grad_norm": 138.74574136900088,
      "learning_rate": 2.55e-06,
      "loss": 14.375,
      "step": 17
    },
    {
      "epoch": 0.008470588235294117,
      "grad_norm": 88.53350493248071,
      "learning_rate": 2.6999999999999996e-06,
      "loss": 10.8125,
      "step": 18
    },
    {
      "epoch": 0.008941176470588235,
      "grad_norm": 63.516977570185766,
      "learning_rate": 2.85e-06,
      "loss": 8.625,
      "step": 19
    },
    {
      "epoch": 0.009411764705882352,
      "grad_norm": 122.27683979620171,
      "learning_rate": 2.9999999999999997e-06,
      "loss": 13.5625,
      "step": 20
    },
    {
      "epoch": 0.00988235294117647,
      "grad_norm": 99.99873034096349,
      "learning_rate": 3.15e-06,
      "loss": 12.5,
      "step": 21
    },
    {
      "epoch": 0.010352941176470589,
      "grad_norm": 71.79387473739015,
      "learning_rate": 3.2999999999999993e-06,
      "loss": 9.8125,
      "step": 22
    },
    {
      "epoch": 0.010823529411764706,
      "grad_norm": 45.768164272053745,
      "learning_rate": 3.4499999999999996e-06,
      "loss": 7.2812,
      "step": 23
    },
    {
      "epoch": 0.011294117647058824,
      "grad_norm": 20.798485257611304,
      "learning_rate": 3.6e-06,
      "loss": 5.9062,
      "step": 24
    },
    {
      "epoch": 0.011764705882352941,
      "grad_norm": 58.10990597448891,
      "learning_rate": 3.7499999999999997e-06,
      "loss": 6.6875,
      "step": 25
    },
    {
      "epoch": 0.01223529411764706,
      "grad_norm": 95.11525900921276,
      "learning_rate": 3.899999999999999e-06,
      "loss": 6.25,
      "step": 26
    },
    {
      "epoch": 0.012705882352941176,
      "grad_norm": 23.642485301040338,
      "learning_rate": 4.049999999999999e-06,
      "loss": 5.0938,
      "step": 27
    },
    {
      "epoch": 0.013176470588235295,
      "grad_norm": 23.987306933945828,
      "learning_rate": 4.2e-06,
      "loss": 5.5625,
      "step": 28
    },
    {
      "epoch": 0.013647058823529411,
      "grad_norm": 64.23616910760174,
      "learning_rate": 4.35e-06,
      "loss": 6.0625,
      "step": 29
    },
    {
      "epoch": 0.01411764705882353,
      "grad_norm": 26.786107918019127,
      "learning_rate": 4.499999999999999e-06,
      "loss": 5.375,
      "step": 30
    },
    {
      "epoch": 0.014588235294117647,
      "grad_norm": 28.46017191854217,
      "learning_rate": 4.6499999999999995e-06,
      "loss": 5.625,
      "step": 31
    },
    {
      "epoch": 0.015058823529411765,
      "grad_norm": 26.261130166314462,
      "learning_rate": 4.8e-06,
      "loss": 5.4375,
      "step": 32
    },
    {
      "epoch": 0.015529411764705882,
      "grad_norm": 55.0488658197144,
      "learning_rate": 4.95e-06,
      "loss": 4.9062,
      "step": 33
    },
    {
      "epoch": 0.016,
      "grad_norm": 19.004376450512638,
      "learning_rate": 5.1e-06,
      "loss": 4.8438,
      "step": 34
    },
    {
      "epoch": 0.01647058823529412,
      "grad_norm": 28.976765997355795,
      "learning_rate": 5.25e-06,
      "loss": 4.6875,
      "step": 35
    },
    {
      "epoch": 0.016941176470588234,
      "grad_norm": 67.22051643767715,
      "learning_rate": 5.399999999999999e-06,
      "loss": 5.0312,
      "step": 36
    },
    {
      "epoch": 0.017411764705882352,
      "grad_norm": 23.59866526927036,
      "learning_rate": 5.549999999999999e-06,
      "loss": 5.0312,
      "step": 37
    },
    {
      "epoch": 0.01788235294117647,
      "grad_norm": 30.214714409958052,
      "learning_rate": 5.7e-06,
      "loss": 5.125,
      "step": 38
    },
    {
      "epoch": 0.01835294117647059,
      "grad_norm": 49.62466821308548,
      "learning_rate": 5.85e-06,
      "loss": 5.0625,
      "step": 39
    },
    {
      "epoch": 0.018823529411764704,
      "grad_norm": 31.515314340176015,
      "learning_rate": 5.999999999999999e-06,
      "loss": 5.0,
      "step": 40
    },
    {
      "epoch": 0.019294117647058823,
      "grad_norm": 28.26248354125833,
      "learning_rate": 6.1499999999999996e-06,
      "loss": 5.2188,
      "step": 41
    },
    {
      "epoch": 0.01976470588235294,
      "grad_norm": 12.88620813983396,
      "learning_rate": 6.3e-06,
      "loss": 3.375,
      "step": 42
    },
    {
      "epoch": 0.02023529411764706,
      "grad_norm": 27.81877381036393,
      "learning_rate": 6.449999999999999e-06,
      "loss": 5.125,
      "step": 43
    },
    {
      "epoch": 0.020705882352941178,
      "grad_norm": 28.884108141963793,
      "learning_rate": 6.599999999999999e-06,
      "loss": 4.3125,
      "step": 44
    },
    {
      "epoch": 0.021176470588235293,
      "grad_norm": 75.24375758493848,
      "learning_rate": 6.749999999999999e-06,
      "loss": 6.5625,
      "step": 45
    },
    {
      "epoch": 0.02164705882352941,
      "grad_norm": 23.00637647183345,
      "learning_rate": 6.899999999999999e-06,
      "loss": 4.4375,
      "step": 46
    },
    {
      "epoch": 0.02211764705882353,
      "grad_norm": 103.44774223037622,
      "learning_rate": 7.049999999999999e-06,
      "loss": 4.4688,
      "step": 47
    },
    {
      "epoch": 0.02258823529411765,
      "grad_norm": 23.93264524544019,
      "learning_rate": 7.2e-06,
      "loss": 4.6875,
      "step": 48
    },
    {
      "epoch": 0.023058823529411764,
      "grad_norm": 24.25652601372358,
      "learning_rate": 7.35e-06,
      "loss": 4.9062,
      "step": 49
    },
    {
      "epoch": 0.023529411764705882,
      "grad_norm": 48.384321473228006,
      "learning_rate": 7.499999999999999e-06,
      "loss": 6.5,
      "step": 50
    },
    {
      "epoch": 0.024,
      "grad_norm": 21.634637780834655,
      "learning_rate": 7.65e-06,
      "loss": 4.3438,
      "step": 51
    },
    {
      "epoch": 0.02447058823529412,
      "grad_norm": 22.22724769797757,
      "learning_rate": 7.799999999999998e-06,
      "loss": 4.2812,
      "step": 52
    },
    {
      "epoch": 0.024941176470588234,
      "grad_norm": 17.225667784335368,
      "learning_rate": 7.949999999999998e-06,
      "loss": 4.3125,
      "step": 53
    },
    {
      "epoch": 0.025411764705882352,
      "grad_norm": 23.19097947677121,
      "learning_rate": 8.099999999999999e-06,
      "loss": 4.5625,
      "step": 54
    },
    {
      "epoch": 0.02588235294117647,
      "grad_norm": 27.57658928691137,
      "learning_rate": 8.249999999999999e-06,
      "loss": 4.6562,
      "step": 55
    },
    {
      "epoch": 0.02635294117647059,
      "grad_norm": 17.23998717160848,
      "learning_rate": 8.4e-06,
      "loss": 3.75,
      "step": 56
    },
    {
      "epoch": 0.026823529411764704,
      "grad_norm": 19.29607504412151,
      "learning_rate": 8.55e-06,
      "loss": 4.3125,
      "step": 57
    },
    {
      "epoch": 0.027294117647058823,
      "grad_norm": 19.69664270825125,
      "learning_rate": 8.7e-06,
      "loss": 4.1875,
      "step": 58
    },
    {
      "epoch": 0.02776470588235294,
      "grad_norm": 21.440217963854334,
      "learning_rate": 8.849999999999998e-06,
      "loss": 3.9531,
      "step": 59
    },
    {
      "epoch": 0.02823529411764706,
      "grad_norm": 30.28531270458705,
      "learning_rate": 8.999999999999999e-06,
      "loss": 4.7812,
      "step": 60
    },
    {
      "epoch": 0.028705882352941175,
      "grad_norm": 18.390608693575082,
      "learning_rate": 9.149999999999999e-06,
      "loss": 3.9219,
      "step": 61
    },
    {
      "epoch": 0.029176470588235293,
      "grad_norm": 25.801955652460077,
      "learning_rate": 9.299999999999999e-06,
      "loss": 4.0625,
      "step": 62
    },
    {
      "epoch": 0.029647058823529412,
      "grad_norm": 21.714236973007992,
      "learning_rate": 9.45e-06,
      "loss": 3.8438,
      "step": 63
    },
    {
      "epoch": 0.03011764705882353,
      "grad_norm": 19.506996453240372,
      "learning_rate": 9.6e-06,
      "loss": 4.0938,
      "step": 64
    },
    {
      "epoch": 0.03058823529411765,
      "grad_norm": 19.6715916440055,
      "learning_rate": 9.75e-06,
      "loss": 3.9062,
      "step": 65
    },
    {
      "epoch": 0.031058823529411764,
      "grad_norm": 37.30781077981528,
      "learning_rate": 9.9e-06,
      "loss": 3.7656,
      "step": 66
    },
    {
      "epoch": 0.03152941176470588,
      "grad_norm": 29.6429224034447,
      "learning_rate": 1.005e-05,
      "loss": 4.125,
      "step": 67
    },
    {
      "epoch": 0.032,
      "grad_norm": 25.026089076914772,
      "learning_rate": 1.02e-05,
      "loss": 4.0312,
      "step": 68
    },
    {
      "epoch": 0.03247058823529412,
      "grad_norm": 14.112097501944222,
      "learning_rate": 1.035e-05,
      "loss": 2.7031,
      "step": 69
    },
    {
      "epoch": 0.03294117647058824,
      "grad_norm": 21.847054841901727,
      "learning_rate": 1.05e-05,
      "loss": 3.6094,
      "step": 70
    },
    {
      "epoch": 0.033411764705882356,
      "grad_norm": 22.372846099516433,
      "learning_rate": 1.0649999999999998e-05,
      "loss": 3.8438,
      "step": 71
    },
    {
      "epoch": 0.03388235294117647,
      "grad_norm": 11.656705409642372,
      "learning_rate": 1.0799999999999998e-05,
      "loss": 2.7656,
      "step": 72
    },
    {
      "epoch": 0.034352941176470586,
      "grad_norm": 23.92534785198683,
      "learning_rate": 1.0949999999999998e-05,
      "loss": 3.6875,
      "step": 73
    },
    {
      "epoch": 0.034823529411764705,
      "grad_norm": 22.030738511949703,
      "learning_rate": 1.1099999999999999e-05,
      "loss": 3.7969,
      "step": 74
    },
    {
      "epoch": 0.03529411764705882,
      "grad_norm": 20.62426189532418,
      "learning_rate": 1.1249999999999999e-05,
      "loss": 3.2969,
      "step": 75
    },
    {
      "epoch": 0.03576470588235294,
      "grad_norm": 22.492194977883702,
      "learning_rate": 1.14e-05,
      "loss": 3.4219,
      "step": 76
    },
    {
      "epoch": 0.03623529411764706,
      "grad_norm": 23.597416315087017,
      "learning_rate": 1.155e-05,
      "loss": 3.6562,
      "step": 77
    },
    {
      "epoch": 0.03670588235294118,
      "grad_norm": 24.481448808639758,
      "learning_rate": 1.17e-05,
      "loss": 3.5781,
      "step": 78
    },
    {
      "epoch": 0.0371764705882353,
      "grad_norm": 32.27151146012509,
      "learning_rate": 1.1849999999999998e-05,
      "loss": 3.9531,
      "step": 79
    },
    {
      "epoch": 0.03764705882352941,
      "grad_norm": 30.5386143044858,
      "learning_rate": 1.1999999999999999e-05,
      "loss": 3.8438,
      "step": 80
    },
    {
      "epoch": 0.03811764705882353,
      "grad_norm": 29.111044262215607,
      "learning_rate": 1.2149999999999999e-05,
      "loss": 3.875,
      "step": 81
    },
    {
      "epoch": 0.038588235294117645,
      "grad_norm": 18.093246880049787,
      "learning_rate": 1.2299999999999999e-05,
      "loss": 3.5156,
      "step": 82
    },
    {
      "epoch": 0.039058823529411764,
      "grad_norm": 21.540609924080854,
      "learning_rate": 1.245e-05,
      "loss": 3.3594,
      "step": 83
    },
    {
      "epoch": 0.03952941176470588,
      "grad_norm": 16.859076470295633,
      "learning_rate": 1.26e-05,
      "loss": 2.7344,
      "step": 84
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.076262430707676,
      "learning_rate": 1.275e-05,
      "loss": 3.2344,
      "step": 85
    },
    {
      "epoch": 0.04047058823529412,
      "grad_norm": 35.88813597989731,
      "learning_rate": 1.2899999999999998e-05,
      "loss": 3.2656,
      "step": 86
    },
    {
      "epoch": 0.04094117647058824,
      "grad_norm": 35.7157475323446,
      "learning_rate": 1.3049999999999999e-05,
      "loss": 3.3125,
      "step": 87
    },
    {
      "epoch": 0.041411764705882356,
      "grad_norm": 29.278512537125554,
      "learning_rate": 1.3199999999999997e-05,
      "loss": 3.0938,
      "step": 88
    },
    {
      "epoch": 0.04188235294117647,
      "grad_norm": 24.57280646335785,
      "learning_rate": 1.3349999999999998e-05,
      "loss": 3.0,
      "step": 89
    },
    {
      "epoch": 0.042352941176470586,
      "grad_norm": 28.577488092354354,
      "learning_rate": 1.3499999999999998e-05,
      "loss": 2.9844,
      "step": 90
    },
    {
      "epoch": 0.042823529411764705,
      "grad_norm": 23.90793800278827,
      "learning_rate": 1.3649999999999998e-05,
      "loss": 3.0469,
      "step": 91
    },
    {
      "epoch": 0.04329411764705882,
      "grad_norm": 20.79802197888147,
      "learning_rate": 1.3799999999999998e-05,
      "loss": 2.9062,
      "step": 92
    },
    {
      "epoch": 0.04376470588235294,
      "grad_norm": 25.00464987983649,
      "learning_rate": 1.3949999999999999e-05,
      "loss": 2.4375,
      "step": 93
    },
    {
      "epoch": 0.04423529411764706,
      "grad_norm": 35.76018621626229,
      "learning_rate": 1.4099999999999999e-05,
      "loss": 3.0781,
      "step": 94
    },
    {
      "epoch": 0.04470588235294118,
      "grad_norm": 39.57046081497139,
      "learning_rate": 1.4249999999999999e-05,
      "loss": 3.0938,
      "step": 95
    },
    {
      "epoch": 0.0451764705882353,
      "grad_norm": 28.581245245568024,
      "learning_rate": 1.44e-05,
      "loss": 2.4531,
      "step": 96
    },
    {
      "epoch": 0.04564705882352941,
      "grad_norm": 33.778133978498246,
      "learning_rate": 1.455e-05,
      "loss": 2.5469,
      "step": 97
    },
    {
      "epoch": 0.04611764705882353,
      "grad_norm": 27.209794076219865,
      "learning_rate": 1.47e-05,
      "loss": 2.5781,
      "step": 98
    },
    {
      "epoch": 0.046588235294117646,
      "grad_norm": 39.64276854545551,
      "learning_rate": 1.485e-05,
      "loss": 2.6719,
      "step": 99
    },
    {
      "epoch": 0.047058823529411764,
      "grad_norm": 28.158445979882075,
      "learning_rate": 1.4999999999999999e-05,
      "loss": 2.625,
      "step": 100
    },
    {
      "epoch": 0.04752941176470588,
      "grad_norm": 36.790679853400306,
      "learning_rate": 1.5149999999999999e-05,
      "loss": 2.5938,
      "step": 101
    },
    {
      "epoch": 0.048,
      "grad_norm": 33.289444250506094,
      "learning_rate": 1.53e-05,
      "loss": 2.4219,
      "step": 102
    },
    {
      "epoch": 0.04847058823529412,
      "grad_norm": 36.34248199650489,
      "learning_rate": 1.545e-05,
      "loss": 2.1406,
      "step": 103
    },
    {
      "epoch": 0.04894117647058824,
      "grad_norm": 27.65611176757927,
      "learning_rate": 1.5599999999999996e-05,
      "loss": 2.5156,
      "step": 104
    },
    {
      "epoch": 0.04941176470588235,
      "grad_norm": 26.62717215928147,
      "learning_rate": 1.5749999999999997e-05,
      "loss": 2.6406,
      "step": 105
    },
    {
      "epoch": 0.04988235294117647,
      "grad_norm": 22.51694728172711,
      "learning_rate": 1.5899999999999997e-05,
      "loss": 2.375,
      "step": 106
    },
    {
      "epoch": 0.050352941176470586,
      "grad_norm": 32.052153829755895,
      "learning_rate": 1.6049999999999997e-05,
      "loss": 2.4375,
      "step": 107
    },
    {
      "epoch": 0.050823529411764705,
      "grad_norm": 25.473604366325688,
      "learning_rate": 1.6199999999999997e-05,
      "loss": 1.7656,
      "step": 108
    },
    {
      "epoch": 0.05129411764705882,
      "grad_norm": 36.382977611627624,
      "learning_rate": 1.6349999999999998e-05,
      "loss": 2.0,
      "step": 109
    },
    {
      "epoch": 0.05176470588235294,
      "grad_norm": 23.13830452920154,
      "learning_rate": 1.6499999999999998e-05,
      "loss": 1.7891,
      "step": 110
    },
    {
      "epoch": 0.05223529411764706,
      "grad_norm": 23.257689055146397,
      "learning_rate": 1.6649999999999998e-05,
      "loss": 2.5,
      "step": 111
    },
    {
      "epoch": 0.05270588235294118,
      "grad_norm": 33.61175708099724,
      "learning_rate": 1.68e-05,
      "loss": 1.7344,
      "step": 112
    },
    {
      "epoch": 0.0531764705882353,
      "grad_norm": 33.791514331259286,
      "learning_rate": 1.695e-05,
      "loss": 1.6172,
      "step": 113
    },
    {
      "epoch": 0.05364705882352941,
      "grad_norm": 22.265727863755874,
      "learning_rate": 1.71e-05,
      "loss": 1.8281,
      "step": 114
    },
    {
      "epoch": 0.05411764705882353,
      "grad_norm": 24.940729406971407,
      "learning_rate": 1.725e-05,
      "loss": 1.7656,
      "step": 115
    },
    {
      "epoch": 0.054588235294117646,
      "grad_norm": 27.592405235498145,
      "learning_rate": 1.74e-05,
      "loss": 1.1484,
      "step": 116
    },
    {
      "epoch": 0.055058823529411764,
      "grad_norm": 27.92922319481223,
      "learning_rate": 1.755e-05,
      "loss": 1.3516,
      "step": 117
    },
    {
      "epoch": 0.05552941176470588,
      "grad_norm": 26.171259108289465,
      "learning_rate": 1.7699999999999997e-05,
      "loss": 1.1094,
      "step": 118
    },
    {
      "epoch": 0.056,
      "grad_norm": 19.76388368048904,
      "learning_rate": 1.7849999999999997e-05,
      "loss": 1.6172,
      "step": 119
    },
    {
      "epoch": 0.05647058823529412,
      "grad_norm": 19.94340747059346,
      "learning_rate": 1.7999999999999997e-05,
      "loss": 2.0312,
      "step": 120
    },
    {
      "epoch": 0.05694117647058824,
      "grad_norm": 27.039521056258618,
      "learning_rate": 1.8149999999999997e-05,
      "loss": 1.375,
      "step": 121
    },
    {
      "epoch": 0.05741176470588235,
      "grad_norm": 15.524415726767916,
      "learning_rate": 1.8299999999999998e-05,
      "loss": 1.3125,
      "step": 122
    },
    {
      "epoch": 0.05788235294117647,
      "grad_norm": 20.304554723228744,
      "learning_rate": 1.8449999999999998e-05,
      "loss": 1.5391,
      "step": 123
    },
    {
      "epoch": 0.05835294117647059,
      "grad_norm": 20.927766177171662,
      "learning_rate": 1.8599999999999998e-05,
      "loss": 1.2266,
      "step": 124
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 22.70489046716681,
      "learning_rate": 1.875e-05,
      "loss": 0.9375,
      "step": 125
    },
    {
      "epoch": 0.059294117647058824,
      "grad_norm": 13.128868788380322,
      "learning_rate": 1.89e-05,
      "loss": 1.1484,
      "step": 126
    },
    {
      "epoch": 0.05976470588235294,
      "grad_norm": 17.659830803893097,
      "learning_rate": 1.905e-05,
      "loss": 1.0703,
      "step": 127
    },
    {
      "epoch": 0.06023529411764706,
      "grad_norm": 22.21300236333354,
      "learning_rate": 1.92e-05,
      "loss": 1.5,
      "step": 128
    },
    {
      "epoch": 0.06070588235294118,
      "grad_norm": 20.75048784148114,
      "learning_rate": 1.935e-05,
      "loss": 1.5703,
      "step": 129
    },
    {
      "epoch": 0.0611764705882353,
      "grad_norm": 18.228185500251563,
      "learning_rate": 1.95e-05,
      "loss": 0.6797,
      "step": 130
    },
    {
      "epoch": 0.06164705882352941,
      "grad_norm": 17.57348216174964,
      "learning_rate": 1.965e-05,
      "loss": 0.9297,
      "step": 131
    },
    {
      "epoch": 0.06211764705882353,
      "grad_norm": 16.544499523704932,
      "learning_rate": 1.98e-05,
      "loss": 0.8672,
      "step": 132
    },
    {
      "epoch": 0.06258823529411765,
      "grad_norm": 15.793921984081052,
      "learning_rate": 1.995e-05,
      "loss": 0.6211,
      "step": 133
    },
    {
      "epoch": 0.06305882352941176,
      "grad_norm": 12.010240907380426,
      "learning_rate": 2.01e-05,
      "loss": 0.8047,
      "step": 134
    },
    {
      "epoch": 0.06352941176470588,
      "grad_norm": 14.558182002743258,
      "learning_rate": 2.025e-05,
      "loss": 0.707,
      "step": 135
    },
    {
      "epoch": 0.064,
      "grad_norm": 14.333503294488926,
      "learning_rate": 2.04e-05,
      "loss": 0.4922,
      "step": 136
    },
    {
      "epoch": 0.06447058823529411,
      "grad_norm": 14.277434520199478,
      "learning_rate": 2.055e-05,
      "loss": 0.8672,
      "step": 137
    },
    {
      "epoch": 0.06494117647058824,
      "grad_norm": 18.216092851265238,
      "learning_rate": 2.07e-05,
      "loss": 1.5078,
      "step": 138
    },
    {
      "epoch": 0.06541176470588235,
      "grad_norm": 18.03700701469668,
      "learning_rate": 2.085e-05,
      "loss": 1.6094,
      "step": 139
    },
    {
      "epoch": 0.06588235294117648,
      "grad_norm": 10.170428582418324,
      "learning_rate": 2.1e-05,
      "loss": 0.498,
      "step": 140
    },
    {
      "epoch": 0.06635294117647059,
      "grad_norm": 9.185160876639815,
      "learning_rate": 2.1149999999999996e-05,
      "loss": 0.4863,
      "step": 141
    },
    {
      "epoch": 0.06682352941176471,
      "grad_norm": 7.588825971236629,
      "learning_rate": 2.1299999999999996e-05,
      "loss": 0.4902,
      "step": 142
    },
    {
      "epoch": 0.06729411764705882,
      "grad_norm": 6.002617473664188,
      "learning_rate": 2.1449999999999996e-05,
      "loss": 1.5391,
      "step": 143
    },
    {
      "epoch": 0.06776470588235294,
      "grad_norm": 5.604171804734954,
      "learning_rate": 2.1599999999999996e-05,
      "loss": 1.0312,
      "step": 144
    },
    {
      "epoch": 0.06823529411764706,
      "grad_norm": 7.101340681853179,
      "learning_rate": 2.1749999999999997e-05,
      "loss": 0.6094,
      "step": 145
    },
    {
      "epoch": 0.06870588235294117,
      "grad_norm": 8.422214009404343,
      "learning_rate": 2.1899999999999997e-05,
      "loss": 0.5234,
      "step": 146
    },
    {
      "epoch": 0.0691764705882353,
      "grad_norm": 7.150234702023276,
      "learning_rate": 2.2049999999999997e-05,
      "loss": 0.4902,
      "step": 147
    },
    {
      "epoch": 0.06964705882352941,
      "grad_norm": 6.728459038577571,
      "learning_rate": 2.2199999999999998e-05,
      "loss": 0.2148,
      "step": 148
    },
    {
      "epoch": 0.07011764705882353,
      "grad_norm": 6.786701563351879,
      "learning_rate": 2.2349999999999998e-05,
      "loss": 0.9414,
      "step": 149
    },
    {
      "epoch": 0.07058823529411765,
      "grad_norm": 5.900796891852584,
      "learning_rate": 2.2499999999999998e-05,
      "loss": 0.2637,
      "step": 150
    },
    {
      "epoch": 0.07105882352941177,
      "grad_norm": 5.567841467150399,
      "learning_rate": 2.2649999999999998e-05,
      "loss": 0.6172,
      "step": 151
    },
    {
      "epoch": 0.07152941176470588,
      "grad_norm": 5.64060771709793,
      "learning_rate": 2.28e-05,
      "loss": 0.4102,
      "step": 152
    },
    {
      "epoch": 0.072,
      "grad_norm": 4.8207287667234455,
      "learning_rate": 2.295e-05,
      "loss": 1.1797,
      "step": 153
    },
    {
      "epoch": 0.07247058823529412,
      "grad_norm": 4.417657915948537,
      "learning_rate": 2.31e-05,
      "loss": 0.8281,
      "step": 154
    },
    {
      "epoch": 0.07294117647058823,
      "grad_norm": 7.734539403147178,
      "learning_rate": 2.325e-05,
      "loss": 0.3398,
      "step": 155
    },
    {
      "epoch": 0.07341176470588236,
      "grad_norm": 4.689729922284172,
      "learning_rate": 2.34e-05,
      "loss": 0.5586,
      "step": 156
    },
    {
      "epoch": 0.07388235294117647,
      "grad_norm": 4.416043122371213,
      "learning_rate": 2.3549999999999996e-05,
      "loss": 0.6641,
      "step": 157
    },
    {
      "epoch": 0.0743529411764706,
      "grad_norm": 4.134295200223024,
      "learning_rate": 2.3699999999999997e-05,
      "loss": 0.1807,
      "step": 158
    },
    {
      "epoch": 0.0748235294117647,
      "grad_norm": 4.5422428180453105,
      "learning_rate": 2.3849999999999997e-05,
      "loss": 0.668,
      "step": 159
    },
    {
      "epoch": 0.07529411764705882,
      "grad_norm": 3.5114997816246127,
      "learning_rate": 2.3999999999999997e-05,
      "loss": 0.2852,
      "step": 160
    },
    {
      "epoch": 0.07576470588235294,
      "grad_norm": 3.810598564174161,
      "learning_rate": 2.4149999999999997e-05,
      "loss": 0.7031,
      "step": 161
    },
    {
      "epoch": 0.07623529411764705,
      "grad_norm": 3.7732596955845428,
      "learning_rate": 2.4299999999999998e-05,
      "loss": 0.7734,
      "step": 162
    },
    {
      "epoch": 0.07670588235294118,
      "grad_norm": 3.3962647529565393,
      "learning_rate": 2.4449999999999998e-05,
      "loss": 0.7617,
      "step": 163
    },
    {
      "epoch": 0.07717647058823529,
      "grad_norm": 3.2134279995590345,
      "learning_rate": 2.4599999999999998e-05,
      "loss": 0.3262,
      "step": 164
    },
    {
      "epoch": 0.07764705882352942,
      "grad_norm": 3.100015190183976,
      "learning_rate": 2.475e-05,
      "loss": 0.1934,
      "step": 165
    },
    {
      "epoch": 0.07811764705882353,
      "grad_norm": 2.958354914233959,
      "learning_rate": 2.49e-05,
      "loss": 0.2461,
      "step": 166
    },
    {
      "epoch": 0.07858823529411765,
      "grad_norm": 2.9618484935077656,
      "learning_rate": 2.505e-05,
      "loss": 0.2969,
      "step": 167
    },
    {
      "epoch": 0.07905882352941176,
      "grad_norm": 2.5778270574586903,
      "learning_rate": 2.52e-05,
      "loss": 0.4082,
      "step": 168
    },
    {
      "epoch": 0.07952941176470588,
      "grad_norm": 2.593753520658777,
      "learning_rate": 2.535e-05,
      "loss": 0.459,
      "step": 169
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.9711860646750723,
      "learning_rate": 2.55e-05,
      "loss": 0.5078,
      "step": 170
    },
    {
      "epoch": 0.08047058823529411,
      "grad_norm": 3.066085913938986,
      "learning_rate": 2.565e-05,
      "loss": 0.4844,
      "step": 171
    },
    {
      "epoch": 0.08094117647058824,
      "grad_norm": 2.5596478952429194,
      "learning_rate": 2.5799999999999997e-05,
      "loss": 0.334,
      "step": 172
    },
    {
      "epoch": 0.08141176470588235,
      "grad_norm": 2.1402590365260274,
      "learning_rate": 2.5949999999999997e-05,
      "loss": 0.1885,
      "step": 173
    },
    {
      "epoch": 0.08188235294117648,
      "grad_norm": 3.9082646905555394,
      "learning_rate": 2.6099999999999997e-05,
      "loss": 0.5664,
      "step": 174
    },
    {
      "epoch": 0.08235294117647059,
      "grad_norm": 3.0555967324119306,
      "learning_rate": 2.6249999999999998e-05,
      "loss": 0.6172,
      "step": 175
    },
    {
      "epoch": 0.08282352941176471,
      "grad_norm": 4.545063899243026,
      "learning_rate": 2.6399999999999995e-05,
      "loss": 0.2217,
      "step": 176
    },
    {
      "epoch": 0.08329411764705882,
      "grad_norm": 4.1415275115438375,
      "learning_rate": 2.6549999999999995e-05,
      "loss": 0.5664,
      "step": 177
    },
    {
      "epoch": 0.08376470588235294,
      "grad_norm": 10.523443219657684,
      "learning_rate": 2.6699999999999995e-05,
      "loss": 0.6055,
      "step": 178
    },
    {
      "epoch": 0.08423529411764706,
      "grad_norm": 3.1451701005722046,
      "learning_rate": 2.6849999999999995e-05,
      "loss": 0.6719,
      "step": 179
    },
    {
      "epoch": 0.08470588235294117,
      "grad_norm": 98.23052122410776,
      "learning_rate": 2.6999999999999996e-05,
      "loss": 3.1094,
      "step": 180
    },
    {
      "epoch": 0.0851764705882353,
      "grad_norm": 4.600285799303798,
      "learning_rate": 2.7149999999999996e-05,
      "loss": 0.6562,
      "step": 181
    },
    {
      "epoch": 0.08564705882352941,
      "grad_norm": 101.43221646803615,
      "learning_rate": 2.7299999999999996e-05,
      "loss": 3.9531,
      "step": 182
    },
    {
      "epoch": 0.08611764705882353,
      "grad_norm": 5.699904888490198,
      "learning_rate": 2.7449999999999996e-05,
      "loss": 0.5156,
      "step": 183
    },
    {
      "epoch": 0.08658823529411765,
      "grad_norm": 12.168392543239367,
      "learning_rate": 2.7599999999999997e-05,
      "loss": 1.2969,
      "step": 184
    },
    {
      "epoch": 0.08705882352941176,
      "grad_norm": 3.3187017373086305,
      "learning_rate": 2.7749999999999997e-05,
      "loss": 0.4863,
      "step": 185
    },
    {
      "epoch": 0.08752941176470588,
      "grad_norm": 14.387877142770016,
      "learning_rate": 2.7899999999999997e-05,
      "loss": 1.4141,
      "step": 186
    },
    {
      "epoch": 0.088,
      "grad_norm": 4.536714304309607,
      "learning_rate": 2.8049999999999997e-05,
      "loss": 0.8008,
      "step": 187
    },
    {
      "epoch": 0.08847058823529412,
      "grad_norm": 46.429579114624715,
      "learning_rate": 2.8199999999999998e-05,
      "loss": 1.1562,
      "step": 188
    },
    {
      "epoch": 0.08894117647058823,
      "grad_norm": 6.720014119703345,
      "learning_rate": 2.8349999999999998e-05,
      "loss": 0.9844,
      "step": 189
    },
    {
      "epoch": 0.08941176470588236,
      "grad_norm": 6.033099809652258,
      "learning_rate": 2.8499999999999998e-05,
      "loss": 1.1172,
      "step": 190
    },
    {
      "epoch": 0.08988235294117647,
      "grad_norm": 5.490956748343037,
      "learning_rate": 2.865e-05,
      "loss": 0.4629,
      "step": 191
    },
    {
      "epoch": 0.0903529411764706,
      "grad_norm": 11.44671648402076,
      "learning_rate": 2.88e-05,
      "loss": 1.0938,
      "step": 192
    },
    {
      "epoch": 0.0908235294117647,
      "grad_norm": 4.593167912885566,
      "learning_rate": 2.895e-05,
      "loss": 0.5195,
      "step": 193
    },
    {
      "epoch": 0.09129411764705882,
      "grad_norm": 3.4751013870411307,
      "learning_rate": 2.91e-05,
      "loss": 0.668,
      "step": 194
    },
    {
      "epoch": 0.09176470588235294,
      "grad_norm": 3.1814186056554257,
      "learning_rate": 2.925e-05,
      "loss": 0.7891,
      "step": 195
    },
    {
      "epoch": 0.09223529411764705,
      "grad_norm": 3.390345173653914,
      "learning_rate": 2.94e-05,
      "loss": 0.6406,
      "step": 196
    },
    {
      "epoch": 0.09270588235294118,
      "grad_norm": 4.2484351523649595,
      "learning_rate": 2.955e-05,
      "loss": 0.6953,
      "step": 197
    },
    {
      "epoch": 0.09317647058823529,
      "grad_norm": 3.7788060432791535,
      "learning_rate": 2.97e-05,
      "loss": 0.7461,
      "step": 198
    },
    {
      "epoch": 0.09364705882352942,
      "grad_norm": 2.7526231057419195,
      "learning_rate": 2.985e-05,
      "loss": 0.3965,
      "step": 199
    },
    {
      "epoch": 0.09411764705882353,
      "grad_norm": 5.170275156473349,
      "learning_rate": 2.9999999999999997e-05,
      "loss": 0.5781,
      "step": 200
    },
    {
      "epoch": 0.09458823529411765,
      "grad_norm": 2.0174815283207157,
      "learning_rate": 3.0149999999999998e-05,
      "loss": 0.1416,
      "step": 201
    },
    {
      "epoch": 0.09505882352941177,
      "grad_norm": 3.8457132460177306,
      "learning_rate": 3.0299999999999998e-05,
      "loss": 0.5234,
      "step": 202
    },
    {
      "epoch": 0.09552941176470588,
      "grad_norm": 10.254970914516647,
      "learning_rate": 3.0449999999999998e-05,
      "loss": 1.4688,
      "step": 203
    },
    {
      "epoch": 0.096,
      "grad_norm": 2.798709424336467,
      "learning_rate": 3.06e-05,
      "loss": 0.6016,
      "step": 204
    },
    {
      "epoch": 0.09647058823529411,
      "grad_norm": 2.1011796250585277,
      "learning_rate": 3.0749999999999995e-05,
      "loss": 0.3965,
      "step": 205
    },
    {
      "epoch": 0.09694117647058824,
      "grad_norm": 10.782783917375218,
      "learning_rate": 3.09e-05,
      "loss": 1.1406,
      "step": 206
    },
    {
      "epoch": 0.09741176470588235,
      "grad_norm": 278.8354653836354,
      "learning_rate": 3.1049999999999996e-05,
      "loss": 2.5312,
      "step": 207
    },
    {
      "epoch": 0.09788235294117648,
      "grad_norm": 208.60750785180423,
      "learning_rate": 3.119999999999999e-05,
      "loss": 15.6875,
      "step": 208
    },
    {
      "epoch": 0.09835294117647059,
      "grad_norm": 26.754863655764105,
      "learning_rate": 3.1349999999999996e-05,
      "loss": 18.75,
      "step": 209
    },
    {
      "epoch": 0.0988235294117647,
      "grad_norm": 38.74930253997053,
      "learning_rate": 3.149999999999999e-05,
      "loss": 22.25,
      "step": 210
    },
    {
      "epoch": 0.09929411764705882,
      "grad_norm": 20.400037471547204,
      "learning_rate": 3.165e-05,
      "loss": 19.125,
      "step": 211
    },
    {
      "epoch": 0.09976470588235294,
      "grad_norm": 54.295338357228154,
      "learning_rate": 3.1799999999999994e-05,
      "loss": 14.5,
      "step": 212
    },
    {
      "epoch": 0.10023529411764706,
      "grad_norm": 62.44551675345745,
      "learning_rate": 3.195e-05,
      "loss": 14.3125,
      "step": 213
    },
    {
      "epoch": 0.10070588235294117,
      "grad_norm": 36.52812711744553,
      "learning_rate": 3.2099999999999994e-05,
      "loss": 11.25,
      "step": 214
    },
    {
      "epoch": 0.1011764705882353,
      "grad_norm": 28.326634139628613,
      "learning_rate": 3.225e-05,
      "loss": 10.25,
      "step": 215
    },
    {
      "epoch": 0.10164705882352941,
      "grad_norm": 26.651033358799722,
      "learning_rate": 3.2399999999999995e-05,
      "loss": 14.75,
      "step": 216
    },
    {
      "epoch": 0.10211764705882354,
      "grad_norm": 27.621683613801668,
      "learning_rate": 3.255e-05,
      "loss": 9.5,
      "step": 217
    },
    {
      "epoch": 0.10258823529411765,
      "grad_norm": 14.79244500662643,
      "learning_rate": 3.2699999999999995e-05,
      "loss": 9.5,
      "step": 218
    },
    {
      "epoch": 0.10305882352941176,
      "grad_norm": 23.638210237786797,
      "learning_rate": 3.285e-05,
      "loss": 12.375,
      "step": 219
    },
    {
      "epoch": 0.10352941176470588,
      "grad_norm": 17.215332743456226,
      "learning_rate": 3.2999999999999996e-05,
      "loss": 6.8438,
      "step": 220
    },
    {
      "epoch": 0.104,
      "grad_norm": 23.117587672583706,
      "learning_rate": 3.315e-05,
      "loss": 7.5938,
      "step": 221
    },
    {
      "epoch": 0.10447058823529412,
      "grad_norm": 23.504545244621372,
      "learning_rate": 3.3299999999999996e-05,
      "loss": 6.5938,
      "step": 222
    },
    {
      "epoch": 0.10494117647058823,
      "grad_norm": 18.44769078646809,
      "learning_rate": 3.345e-05,
      "loss": 12.8125,
      "step": 223
    },
    {
      "epoch": 0.10541176470588236,
      "grad_norm": 40.07937871415431,
      "learning_rate": 3.36e-05,
      "loss": 8.5625,
      "step": 224
    },
    {
      "epoch": 0.10588235294117647,
      "grad_norm": 34.197385052370606,
      "learning_rate": 3.375e-05,
      "loss": 9.1875,
      "step": 225
    },
    {
      "epoch": 0.1063529411764706,
      "grad_norm": 17.792131818569793,
      "learning_rate": 3.39e-05,
      "loss": 12.8125,
      "step": 226
    },
    {
      "epoch": 0.1068235294117647,
      "grad_norm": 16.64267606396611,
      "learning_rate": 3.405e-05,
      "loss": 6.75,
      "step": 227
    },
    {
      "epoch": 0.10729411764705882,
      "grad_norm": 45.72364772289574,
      "learning_rate": 3.42e-05,
      "loss": 9.3125,
      "step": 228
    },
    {
      "epoch": 0.10776470588235294,
      "grad_norm": 14.36307634699885,
      "learning_rate": 3.435e-05,
      "loss": 7.2188,
      "step": 229
    },
    {
      "epoch": 0.10823529411764705,
      "grad_norm": 19.107155476729808,
      "learning_rate": 3.45e-05,
      "loss": 6.9062,
      "step": 230
    },
    {
      "epoch": 0.10870588235294118,
      "grad_norm": 35.5611180159604,
      "learning_rate": 3.465e-05,
      "loss": 6.9688,
      "step": 231
    },
    {
      "epoch": 0.10917647058823529,
      "grad_norm": 39.43703312737912,
      "learning_rate": 3.48e-05,
      "loss": 7.6562,
      "step": 232
    },
    {
      "epoch": 0.10964705882352942,
      "grad_norm": 18.639508905887343,
      "learning_rate": 3.4949999999999996e-05,
      "loss": 6.5938,
      "step": 233
    },
    {
      "epoch": 0.11011764705882353,
      "grad_norm": 35.71551173003105,
      "learning_rate": 3.51e-05,
      "loss": 7.7188,
      "step": 234
    },
    {
      "epoch": 0.11058823529411765,
      "grad_norm": 18.715439078858715,
      "learning_rate": 3.5249999999999996e-05,
      "loss": 6.25,
      "step": 235
    },
    {
      "epoch": 0.11105882352941177,
      "grad_norm": 17.519137005772826,
      "learning_rate": 3.539999999999999e-05,
      "loss": 7.0625,
      "step": 236
    },
    {
      "epoch": 0.11152941176470588,
      "grad_norm": 20.982842695448984,
      "learning_rate": 3.555e-05,
      "loss": 5.0,
      "step": 237
    },
    {
      "epoch": 0.112,
      "grad_norm": 45.05761552650986,
      "learning_rate": 3.5699999999999994e-05,
      "loss": 6.5312,
      "step": 238
    },
    {
      "epoch": 0.11247058823529411,
      "grad_norm": 26.189781412198613,
      "learning_rate": 3.585e-05,
      "loss": 5.125,
      "step": 239
    },
    {
      "epoch": 0.11294117647058824,
      "grad_norm": 34.345315450036225,
      "learning_rate": 3.5999999999999994e-05,
      "loss": 6.0625,
      "step": 240
    },
    {
      "epoch": 0.11341176470588235,
      "grad_norm": 38.03194361355241,
      "learning_rate": 3.615e-05,
      "loss": 5.5625,
      "step": 241
    },
    {
      "epoch": 0.11388235294117648,
      "grad_norm": 39.231276611199675,
      "learning_rate": 3.6299999999999995e-05,
      "loss": 5.0,
      "step": 242
    },
    {
      "epoch": 0.11435294117647059,
      "grad_norm": 22.19358120813286,
      "learning_rate": 3.645e-05,
      "loss": 4.7812,
      "step": 243
    },
    {
      "epoch": 0.1148235294117647,
      "grad_norm": 31.736191838548613,
      "learning_rate": 3.6599999999999995e-05,
      "loss": 4.1562,
      "step": 244
    },
    {
      "epoch": 0.11529411764705882,
      "grad_norm": 38.629031250351744,
      "learning_rate": 3.675e-05,
      "loss": 3.5625,
      "step": 245
    },
    {
      "epoch": 0.11576470588235294,
      "grad_norm": 11.750256090128827,
      "learning_rate": 3.6899999999999996e-05,
      "loss": 3.0312,
      "step": 246
    },
    {
      "epoch": 0.11623529411764706,
      "grad_norm": 29.607289514837493,
      "learning_rate": 3.705e-05,
      "loss": 4.9062,
      "step": 247
    },
    {
      "epoch": 0.11670588235294117,
      "grad_norm": 17.697175107749153,
      "learning_rate": 3.7199999999999996e-05,
      "loss": 2.9844,
      "step": 248
    },
    {
      "epoch": 0.1171764705882353,
      "grad_norm": 14.33239490318074,
      "learning_rate": 3.735e-05,
      "loss": 2.4844,
      "step": 249
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 7.158670620241891,
      "learning_rate": 3.75e-05,
      "loss": 2.375,
      "step": 250
    },
    {
      "epoch": 0.11811764705882354,
      "grad_norm": 8.207956961509792,
      "learning_rate": 3.7649999999999994e-05,
      "loss": 2.0312,
      "step": 251
    },
    {
      "epoch": 0.11858823529411765,
      "grad_norm": 8.466928973545421,
      "learning_rate": 3.78e-05,
      "loss": 2.5312,
      "step": 252
    },
    {
      "epoch": 0.11905882352941176,
      "grad_norm": 9.947917028143486,
      "learning_rate": 3.7949999999999994e-05,
      "loss": 1.2422,
      "step": 253
    },
    {
      "epoch": 0.11952941176470588,
      "grad_norm": 9.637845703727594,
      "learning_rate": 3.81e-05,
      "loss": 0.5664,
      "step": 254
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.57675717745349,
      "learning_rate": 3.8249999999999995e-05,
      "loss": 1.0156,
      "step": 255
    },
    {
      "epoch": 0.12047058823529412,
      "grad_norm": 3.84424443526246,
      "learning_rate": 3.84e-05,
      "loss": 2.0781,
      "step": 256
    },
    {
      "epoch": 0.12094117647058823,
      "grad_norm": 3.106458322839173,
      "learning_rate": 3.8549999999999995e-05,
      "loss": 1.6797,
      "step": 257
    },
    {
      "epoch": 0.12141176470588236,
      "grad_norm": 9.553804602561966,
      "learning_rate": 3.87e-05,
      "loss": 0.457,
      "step": 258
    },
    {
      "epoch": 0.12188235294117647,
      "grad_norm": 11.259566718391254,
      "learning_rate": 3.8849999999999996e-05,
      "loss": 1.6719,
      "step": 259
    },
    {
      "epoch": 0.1223529411764706,
      "grad_norm": 3.6366701439547646,
      "learning_rate": 3.9e-05,
      "loss": 0.5195,
      "step": 260
    },
    {
      "epoch": 0.1228235294117647,
      "grad_norm": 25.72580523653768,
      "learning_rate": 3.9149999999999996e-05,
      "loss": 3.5781,
      "step": 261
    },
    {
      "epoch": 0.12329411764705882,
      "grad_norm": 19.69762804704518,
      "learning_rate": 3.93e-05,
      "loss": 2.2188,
      "step": 262
    },
    {
      "epoch": 0.12376470588235294,
      "grad_norm": 14.33300915217077,
      "learning_rate": 3.945e-05,
      "loss": 3.3281,
      "step": 263
    },
    {
      "epoch": 0.12423529411764705,
      "grad_norm": 52.143329578250956,
      "learning_rate": 3.96e-05,
      "loss": 5.0,
      "step": 264
    },
    {
      "epoch": 0.12470588235294118,
      "grad_norm": 2.9450621677384348,
      "learning_rate": 3.975e-05,
      "loss": 2.1094,
      "step": 265
    },
    {
      "epoch": 0.1251764705882353,
      "grad_norm": 7.058054380204918,
      "learning_rate": 3.99e-05,
      "loss": 1.5547,
      "step": 266
    },
    {
      "epoch": 0.12564705882352942,
      "grad_norm": 17.0559578843016,
      "learning_rate": 4.005e-05,
      "loss": 5.0938,
      "step": 267
    },
    {
      "epoch": 0.12611764705882353,
      "grad_norm": 7.102016121053011,
      "learning_rate": 4.02e-05,
      "loss": 1.9688,
      "step": 268
    },
    {
      "epoch": 0.12658823529411764,
      "grad_norm": 7.159421751574754,
      "learning_rate": 4.035e-05,
      "loss": 1.3672,
      "step": 269
    },
    {
      "epoch": 0.12705882352941175,
      "grad_norm": 19.222670782191862,
      "learning_rate": 4.05e-05,
      "loss": 3.2969,
      "step": 270
    },
    {
      "epoch": 0.1275294117647059,
      "grad_norm": 10.535115851289762,
      "learning_rate": 4.065e-05,
      "loss": 2.2188,
      "step": 271
    },
    {
      "epoch": 0.128,
      "grad_norm": 25.665798962379167,
      "learning_rate": 4.08e-05,
      "loss": 2.3281,
      "step": 272
    },
    {
      "epoch": 0.12847058823529411,
      "grad_norm": 7.699657226845019,
      "learning_rate": 4.095e-05,
      "loss": 1.9141,
      "step": 273
    },
    {
      "epoch": 0.12894117647058823,
      "grad_norm": 6.888152396240225,
      "learning_rate": 4.11e-05,
      "loss": 2.375,
      "step": 274
    },
    {
      "epoch": 0.12941176470588237,
      "grad_norm": 6.18015923802887,
      "learning_rate": 4.125e-05,
      "loss": 1.0703,
      "step": 275
    },
    {
      "epoch": 0.12988235294117648,
      "grad_norm": 5.10496475681139,
      "learning_rate": 4.14e-05,
      "loss": 1.1797,
      "step": 276
    },
    {
      "epoch": 0.1303529411764706,
      "grad_norm": 13.526194855791495,
      "learning_rate": 4.155e-05,
      "loss": 4.1562,
      "step": 277
    },
    {
      "epoch": 0.1308235294117647,
      "grad_norm": 5.033844064914956,
      "learning_rate": 4.17e-05,
      "loss": 0.8164,
      "step": 278
    },
    {
      "epoch": 0.1312941176470588,
      "grad_norm": 8.370075752240425,
      "learning_rate": 4.185e-05,
      "loss": 1.6641,
      "step": 279
    },
    {
      "epoch": 0.13176470588235295,
      "grad_norm": 1.9724706633962517,
      "learning_rate": 4.2e-05,
      "loss": 1.5781,
      "step": 280
    },
    {
      "epoch": 0.13223529411764706,
      "grad_norm": 1.705030903482689,
      "learning_rate": 4.215e-05,
      "loss": 2.1719,
      "step": 281
    },
    {
      "epoch": 0.13270588235294117,
      "grad_norm": 1.0367338077266852,
      "learning_rate": 4.229999999999999e-05,
      "loss": 0.8359,
      "step": 282
    },
    {
      "epoch": 0.13317647058823529,
      "grad_norm": 1.4931336129064048,
      "learning_rate": 4.2449999999999995e-05,
      "loss": 1.0234,
      "step": 283
    },
    {
      "epoch": 0.13364705882352942,
      "grad_norm": 2.283985610501559,
      "learning_rate": 4.259999999999999e-05,
      "loss": 2.6719,
      "step": 284
    },
    {
      "epoch": 0.13411764705882354,
      "grad_norm": 1.901368331521953,
      "learning_rate": 4.2749999999999996e-05,
      "loss": 0.7852,
      "step": 285
    },
    {
      "epoch": 0.13458823529411765,
      "grad_norm": 2.178093961622586,
      "learning_rate": 4.289999999999999e-05,
      "loss": 2.7656,
      "step": 286
    },
    {
      "epoch": 0.13505882352941176,
      "grad_norm": 1.818058257265634,
      "learning_rate": 4.3049999999999996e-05,
      "loss": 1.0312,
      "step": 287
    },
    {
      "epoch": 0.13552941176470587,
      "grad_norm": 2.040575150977686,
      "learning_rate": 4.319999999999999e-05,
      "loss": 1.5469,
      "step": 288
    },
    {
      "epoch": 0.136,
      "grad_norm": 2.2281630296280603,
      "learning_rate": 4.334999999999999e-05,
      "loss": 1.5391,
      "step": 289
    },
    {
      "epoch": 0.13647058823529412,
      "grad_norm": 1.8074805332261028,
      "learning_rate": 4.3499999999999993e-05,
      "loss": 1.2656,
      "step": 290
    },
    {
      "epoch": 0.13694117647058823,
      "grad_norm": 1.3146013537857597,
      "learning_rate": 4.364999999999999e-05,
      "loss": 1.7578,
      "step": 291
    },
    {
      "epoch": 0.13741176470588234,
      "grad_norm": 1.3814228645109183,
      "learning_rate": 4.3799999999999994e-05,
      "loss": 1.6094,
      "step": 292
    },
    {
      "epoch": 0.13788235294117648,
      "grad_norm": 1.5581586633892721,
      "learning_rate": 4.394999999999999e-05,
      "loss": 1.7344,
      "step": 293
    },
    {
      "epoch": 0.1383529411764706,
      "grad_norm": 3.7753912199456647,
      "learning_rate": 4.4099999999999995e-05,
      "loss": 1.4219,
      "step": 294
    },
    {
      "epoch": 0.1388235294117647,
      "grad_norm": 1.817069428946445,
      "learning_rate": 4.424999999999999e-05,
      "loss": 0.8789,
      "step": 295
    },
    {
      "epoch": 0.13929411764705882,
      "grad_norm": 2.2981355842629476,
      "learning_rate": 4.4399999999999995e-05,
      "loss": 1.4531,
      "step": 296
    },
    {
      "epoch": 0.13976470588235293,
      "grad_norm": 1.7082879495021697,
      "learning_rate": 4.454999999999999e-05,
      "loss": 1.5156,
      "step": 297
    },
    {
      "epoch": 0.14023529411764707,
      "grad_norm": 1.7948954973669538,
      "learning_rate": 4.4699999999999996e-05,
      "loss": 0.8828,
      "step": 298
    },
    {
      "epoch": 0.14070588235294118,
      "grad_norm": 2.03348913989381,
      "learning_rate": 4.484999999999999e-05,
      "loss": 1.4453,
      "step": 299
    },
    {
      "epoch": 0.1411764705882353,
      "grad_norm": 1.662911294547269,
      "learning_rate": 4.4999999999999996e-05,
      "loss": 0.9883,
      "step": 300
    },
    {
      "epoch": 0.1416470588235294,
      "grad_norm": 3.9611376027857412,
      "learning_rate": 4.514999999999999e-05,
      "loss": 2.6406,
      "step": 301
    },
    {
      "epoch": 0.14211764705882354,
      "grad_norm": 1.9465111655353384,
      "learning_rate": 4.5299999999999997e-05,
      "loss": 2.625,
      "step": 302
    },
    {
      "epoch": 0.14258823529411765,
      "grad_norm": 2.7392939453578418,
      "learning_rate": 4.5449999999999993e-05,
      "loss": 1.1484,
      "step": 303
    },
    {
      "epoch": 0.14305882352941177,
      "grad_norm": 1.9646301340251517,
      "learning_rate": 4.56e-05,
      "loss": 0.8164,
      "step": 304
    },
    {
      "epoch": 0.14352941176470588,
      "grad_norm": 2.4767883478850528,
      "learning_rate": 4.5749999999999994e-05,
      "loss": 1.3125,
      "step": 305
    },
    {
      "epoch": 0.144,
      "grad_norm": 2.054431719851156,
      "learning_rate": 4.59e-05,
      "loss": 1.5312,
      "step": 306
    },
    {
      "epoch": 0.14447058823529413,
      "grad_norm": 1.5172635366561873,
      "learning_rate": 4.6049999999999994e-05,
      "loss": 1.4531,
      "step": 307
    },
    {
      "epoch": 0.14494117647058824,
      "grad_norm": 1.1402233489302838,
      "learning_rate": 4.62e-05,
      "loss": 1.4922,
      "step": 308
    },
    {
      "epoch": 0.14541176470588235,
      "grad_norm": 1.5212716243577973,
      "learning_rate": 4.6349999999999995e-05,
      "loss": 1.4062,
      "step": 309
    },
    {
      "epoch": 0.14588235294117646,
      "grad_norm": 0.8797915210843892,
      "learning_rate": 4.65e-05,
      "loss": 0.3223,
      "step": 310
    },
    {
      "epoch": 0.14635294117647057,
      "grad_norm": 1.2513926253996344,
      "learning_rate": 4.6649999999999996e-05,
      "loss": 0.7266,
      "step": 311
    },
    {
      "epoch": 0.1468235294117647,
      "grad_norm": 1.9228799347324985,
      "learning_rate": 4.68e-05,
      "loss": 1.9922,
      "step": 312
    },
    {
      "epoch": 0.14729411764705883,
      "grad_norm": 1.8500993345457593,
      "learning_rate": 4.6949999999999996e-05,
      "loss": 1.3047,
      "step": 313
    },
    {
      "epoch": 0.14776470588235294,
      "grad_norm": 1.8107869953315365,
      "learning_rate": 4.709999999999999e-05,
      "loss": 0.9414,
      "step": 314
    },
    {
      "epoch": 0.14823529411764705,
      "grad_norm": 2.6690920569895784,
      "learning_rate": 4.7249999999999997e-05,
      "loss": 1.5078,
      "step": 315
    },
    {
      "epoch": 0.1487058823529412,
      "grad_norm": 0.8728929215061825,
      "learning_rate": 4.7399999999999993e-05,
      "loss": 0.4688,
      "step": 316
    },
    {
      "epoch": 0.1491764705882353,
      "grad_norm": 1.7906435936689404,
      "learning_rate": 4.755e-05,
      "loss": 1.1719,
      "step": 317
    },
    {
      "epoch": 0.1496470588235294,
      "grad_norm": 1.8291221820960322,
      "learning_rate": 4.7699999999999994e-05,
      "loss": 1.3203,
      "step": 318
    },
    {
      "epoch": 0.15011764705882352,
      "grad_norm": 2.065120216828793,
      "learning_rate": 4.785e-05,
      "loss": 1.6953,
      "step": 319
    },
    {
      "epoch": 0.15058823529411763,
      "grad_norm": 2.8988021158619293,
      "learning_rate": 4.7999999999999994e-05,
      "loss": 1.4375,
      "step": 320
    },
    {
      "epoch": 0.15105882352941177,
      "grad_norm": 2.8023225528466056,
      "learning_rate": 4.815e-05,
      "loss": 1.1562,
      "step": 321
    },
    {
      "epoch": 0.15152941176470588,
      "grad_norm": 1.9622721509367715,
      "learning_rate": 4.8299999999999995e-05,
      "loss": 0.7266,
      "step": 322
    },
    {
      "epoch": 0.152,
      "grad_norm": 14.094801015806713,
      "learning_rate": 4.845e-05,
      "loss": 1.8359,
      "step": 323
    },
    {
      "epoch": 0.1524705882352941,
      "grad_norm": 1.0996426473757637,
      "learning_rate": 4.8599999999999995e-05,
      "loss": 0.3965,
      "step": 324
    },
    {
      "epoch": 0.15294117647058825,
      "grad_norm": 1.2917326473811446,
      "learning_rate": 4.875e-05,
      "loss": 0.5195,
      "step": 325
    },
    {
      "epoch": 0.15341176470588236,
      "grad_norm": 1.606237678818691,
      "learning_rate": 4.8899999999999996e-05,
      "loss": 0.457,
      "step": 326
    },
    {
      "epoch": 0.15388235294117647,
      "grad_norm": 1.5102729299018767,
      "learning_rate": 4.905e-05,
      "loss": 0.5703,
      "step": 327
    },
    {
      "epoch": 0.15435294117647058,
      "grad_norm": 1.4168630137105194,
      "learning_rate": 4.9199999999999997e-05,
      "loss": 0.7773,
      "step": 328
    },
    {
      "epoch": 0.1548235294117647,
      "grad_norm": 8.064657284972142,
      "learning_rate": 4.935e-05,
      "loss": 1.1953,
      "step": 329
    },
    {
      "epoch": 0.15529411764705883,
      "grad_norm": 4.956546617701626,
      "learning_rate": 4.95e-05,
      "loss": 0.9102,
      "step": 330
    },
    {
      "epoch": 0.15576470588235294,
      "grad_norm": 3.0580470431867357,
      "learning_rate": 4.965e-05,
      "loss": 2.1562,
      "step": 331
    },
    {
      "epoch": 0.15623529411764706,
      "grad_norm": 1.2148237774103037,
      "learning_rate": 4.98e-05,
      "loss": 0.375,
      "step": 332
    },
    {
      "epoch": 0.15670588235294117,
      "grad_norm": 1.646667637027432,
      "learning_rate": 4.995e-05,
      "loss": 0.9492,
      "step": 333
    },
    {
      "epoch": 0.1571764705882353,
      "grad_norm": 3.8461176274706506,
      "learning_rate": 5.01e-05,
      "loss": 1.3828,
      "step": 334
    },
    {
      "epoch": 0.15764705882352942,
      "grad_norm": 0.9340180513669921,
      "learning_rate": 5.025e-05,
      "loss": 0.832,
      "step": 335
    },
    {
      "epoch": 0.15811764705882353,
      "grad_norm": 1.0463402037047196,
      "learning_rate": 5.04e-05,
      "loss": 0.8164,
      "step": 336
    },
    {
      "epoch": 0.15858823529411764,
      "grad_norm": 2.15575358195737,
      "learning_rate": 5.055e-05,
      "loss": 1.0859,
      "step": 337
    },
    {
      "epoch": 0.15905882352941175,
      "grad_norm": 3.6994193206189765,
      "learning_rate": 5.07e-05,
      "loss": 1.8359,
      "step": 338
    },
    {
      "epoch": 0.1595294117647059,
      "grad_norm": 1.7321201540603937,
      "learning_rate": 5.0849999999999996e-05,
      "loss": 1.25,
      "step": 339
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.957628361634743,
      "learning_rate": 5.1e-05,
      "loss": 0.9141,
      "step": 340
    },
    {
      "epoch": 0.16047058823529411,
      "grad_norm": 1.535064239542706,
      "learning_rate": 5.1149999999999996e-05,
      "loss": 1.3828,
      "step": 341
    },
    {
      "epoch": 0.16094117647058823,
      "grad_norm": 3.219049711781514,
      "learning_rate": 5.13e-05,
      "loss": 2.6562,
      "step": 342
    },
    {
      "epoch": 0.16141176470588237,
      "grad_norm": 2.1818150747717184,
      "learning_rate": 5.145e-05,
      "loss": 1.3828,
      "step": 343
    },
    {
      "epoch": 0.16188235294117648,
      "grad_norm": 1.7454068106846607,
      "learning_rate": 5.1599999999999994e-05,
      "loss": 0.6719,
      "step": 344
    },
    {
      "epoch": 0.1623529411764706,
      "grad_norm": 3.4500123446891466,
      "learning_rate": 5.174999999999999e-05,
      "loss": 2.1094,
      "step": 345
    },
    {
      "epoch": 0.1628235294117647,
      "grad_norm": 0.30928490211100745,
      "learning_rate": 5.1899999999999994e-05,
      "loss": 0.0593,
      "step": 346
    },
    {
      "epoch": 0.1632941176470588,
      "grad_norm": 3.0017100349806958,
      "learning_rate": 5.204999999999999e-05,
      "loss": 0.6953,
      "step": 347
    },
    {
      "epoch": 0.16376470588235295,
      "grad_norm": 1.1732076169602517,
      "learning_rate": 5.2199999999999995e-05,
      "loss": 0.4023,
      "step": 348
    },
    {
      "epoch": 0.16423529411764706,
      "grad_norm": 15.032873284852807,
      "learning_rate": 5.234999999999999e-05,
      "loss": 1.2734,
      "step": 349
    },
    {
      "epoch": 0.16470588235294117,
      "grad_norm": 3.3260225355897926,
      "learning_rate": 5.2499999999999995e-05,
      "loss": 1.7266,
      "step": 350
    },
    {
      "epoch": 0.16517647058823529,
      "grad_norm": 2.2968081345379683,
      "learning_rate": 5.264999999999999e-05,
      "loss": 0.7812,
      "step": 351
    },
    {
      "epoch": 0.16564705882352943,
      "grad_norm": 227.0576806840813,
      "learning_rate": 5.279999999999999e-05,
      "loss": 6.0938,
      "step": 352
    },
    {
      "epoch": 0.16611764705882354,
      "grad_norm": 1.6088062606014817,
      "learning_rate": 5.294999999999999e-05,
      "loss": 0.6133,
      "step": 353
    },
    {
      "epoch": 0.16658823529411765,
      "grad_norm": 18.86043159715291,
      "learning_rate": 5.309999999999999e-05,
      "loss": 4.9375,
      "step": 354
    },
    {
      "epoch": 0.16705882352941176,
      "grad_norm": 28.607432438048818,
      "learning_rate": 5.324999999999999e-05,
      "loss": 7.0938,
      "step": 355
    },
    {
      "epoch": 0.16752941176470587,
      "grad_norm": 50.14341850593775,
      "learning_rate": 5.339999999999999e-05,
      "loss": 5.5938,
      "step": 356
    },
    {
      "epoch": 0.168,
      "grad_norm": 5.675607469022968,
      "learning_rate": 5.3549999999999994e-05,
      "loss": 1.5625,
      "step": 357
    },
    {
      "epoch": 0.16847058823529412,
      "grad_norm": 6.32662784796487,
      "learning_rate": 5.369999999999999e-05,
      "loss": 1.1953,
      "step": 358
    },
    {
      "epoch": 0.16894117647058823,
      "grad_norm": 7.444180987953766,
      "learning_rate": 5.3849999999999994e-05,
      "loss": 1.9062,
      "step": 359
    },
    {
      "epoch": 0.16941176470588235,
      "grad_norm": 1.813931975897033,
      "learning_rate": 5.399999999999999e-05,
      "loss": 0.6367,
      "step": 360
    },
    {
      "epoch": 0.16988235294117648,
      "grad_norm": 4.667477702531829,
      "learning_rate": 5.4149999999999995e-05,
      "loss": 1.75,
      "step": 361
    },
    {
      "epoch": 0.1703529411764706,
      "grad_norm": 8.524803374082706,
      "learning_rate": 5.429999999999999e-05,
      "loss": 1.8672,
      "step": 362
    },
    {
      "epoch": 0.1708235294117647,
      "grad_norm": 11.318838256745765,
      "learning_rate": 5.4449999999999995e-05,
      "loss": 1.9922,
      "step": 363
    },
    {
      "epoch": 0.17129411764705882,
      "grad_norm": 5.7467313466073975,
      "learning_rate": 5.459999999999999e-05,
      "loss": 1.5938,
      "step": 364
    },
    {
      "epoch": 0.17176470588235293,
      "grad_norm": 1.5645128973229026,
      "learning_rate": 5.4749999999999996e-05,
      "loss": 0.4688,
      "step": 365
    },
    {
      "epoch": 0.17223529411764707,
      "grad_norm": 9.034446756297996,
      "learning_rate": 5.489999999999999e-05,
      "loss": 2.1719,
      "step": 366
    },
    {
      "epoch": 0.17270588235294118,
      "grad_norm": 3.969923986903702,
      "learning_rate": 5.5049999999999996e-05,
      "loss": 1.3984,
      "step": 367
    },
    {
      "epoch": 0.1731764705882353,
      "grad_norm": 5.709297915593139,
      "learning_rate": 5.519999999999999e-05,
      "loss": 3.2812,
      "step": 368
    },
    {
      "epoch": 0.1736470588235294,
      "grad_norm": 5.53242835427933,
      "learning_rate": 5.535e-05,
      "loss": 1.875,
      "step": 369
    },
    {
      "epoch": 0.17411764705882352,
      "grad_norm": 4.160527516715009,
      "learning_rate": 5.5499999999999994e-05,
      "loss": 1.2734,
      "step": 370
    },
    {
      "epoch": 0.17458823529411766,
      "grad_norm": 1.823671869588528,
      "learning_rate": 5.565e-05,
      "loss": 0.5234,
      "step": 371
    },
    {
      "epoch": 0.17505882352941177,
      "grad_norm": 7.210435403140748,
      "learning_rate": 5.5799999999999994e-05,
      "loss": 1.8672,
      "step": 372
    },
    {
      "epoch": 0.17552941176470588,
      "grad_norm": 2.33683764529406,
      "learning_rate": 5.595e-05,
      "loss": 1.3047,
      "step": 373
    },
    {
      "epoch": 0.176,
      "grad_norm": 2.3788919836887428,
      "learning_rate": 5.6099999999999995e-05,
      "loss": 1.5391,
      "step": 374
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 1.6807228344772962,
      "learning_rate": 5.625e-05,
      "loss": 1.0,
      "step": 375
    },
    {
      "epoch": 0.17694117647058824,
      "grad_norm": 2.8289833894801273,
      "learning_rate": 5.6399999999999995e-05,
      "loss": 1.875,
      "step": 376
    },
    {
      "epoch": 0.17741176470588235,
      "grad_norm": 1.8212519306421666,
      "learning_rate": 5.654999999999999e-05,
      "loss": 1.4062,
      "step": 377
    },
    {
      "epoch": 0.17788235294117646,
      "grad_norm": 1.0099654381909282,
      "learning_rate": 5.6699999999999996e-05,
      "loss": 0.4336,
      "step": 378
    },
    {
      "epoch": 0.17835294117647058,
      "grad_norm": 0.7450857286859774,
      "learning_rate": 5.684999999999999e-05,
      "loss": 0.1836,
      "step": 379
    },
    {
      "epoch": 0.17882352941176471,
      "grad_norm": 5.762379920436007,
      "learning_rate": 5.6999999999999996e-05,
      "loss": 1.8203,
      "step": 380
    },
    {
      "epoch": 0.17929411764705883,
      "grad_norm": 4.681459349129762,
      "learning_rate": 5.714999999999999e-05,
      "loss": 1.6562,
      "step": 381
    },
    {
      "epoch": 0.17976470588235294,
      "grad_norm": 2.620462965775208,
      "learning_rate": 5.73e-05,
      "loss": 2.875,
      "step": 382
    },
    {
      "epoch": 0.18023529411764705,
      "grad_norm": 1.5309789590401657,
      "learning_rate": 5.7449999999999994e-05,
      "loss": 1.3594,
      "step": 383
    },
    {
      "epoch": 0.1807058823529412,
      "grad_norm": 2.344574244934665,
      "learning_rate": 5.76e-05,
      "loss": 2.0,
      "step": 384
    },
    {
      "epoch": 0.1811764705882353,
      "grad_norm": 2.147065613895744,
      "learning_rate": 5.7749999999999994e-05,
      "loss": 1.8359,
      "step": 385
    },
    {
      "epoch": 0.1816470588235294,
      "grad_norm": 1.9172359260995238,
      "learning_rate": 5.79e-05,
      "loss": 1.6484,
      "step": 386
    },
    {
      "epoch": 0.18211764705882352,
      "grad_norm": 4.3909934972231515,
      "learning_rate": 5.8049999999999995e-05,
      "loss": 1.3359,
      "step": 387
    },
    {
      "epoch": 0.18258823529411763,
      "grad_norm": 1.9281730914193516,
      "learning_rate": 5.82e-05,
      "loss": 0.3984,
      "step": 388
    },
    {
      "epoch": 0.18305882352941177,
      "grad_norm": 0.8308637064809429,
      "learning_rate": 5.8349999999999995e-05,
      "loss": 0.459,
      "step": 389
    },
    {
      "epoch": 0.18352941176470589,
      "grad_norm": 2.021260906875496,
      "learning_rate": 5.85e-05,
      "loss": 1.6016,
      "step": 390
    },
    {
      "epoch": 0.184,
      "grad_norm": 2.457683447156207,
      "learning_rate": 5.8649999999999996e-05,
      "loss": 0.8242,
      "step": 391
    },
    {
      "epoch": 0.1844705882352941,
      "grad_norm": 1.2076734114660121,
      "learning_rate": 5.88e-05,
      "loss": 0.6953,
      "step": 392
    },
    {
      "epoch": 0.18494117647058825,
      "grad_norm": 1.3810297025466498,
      "learning_rate": 5.8949999999999996e-05,
      "loss": 1.1484,
      "step": 393
    },
    {
      "epoch": 0.18541176470588236,
      "grad_norm": 1.770623544019891,
      "learning_rate": 5.91e-05,
      "loss": 1.3984,
      "step": 394
    },
    {
      "epoch": 0.18588235294117647,
      "grad_norm": 1.110634205442013,
      "learning_rate": 5.925e-05,
      "loss": 0.9531,
      "step": 395
    },
    {
      "epoch": 0.18635294117647058,
      "grad_norm": 0.580751111505624,
      "learning_rate": 5.94e-05,
      "loss": 0.2734,
      "step": 396
    },
    {
      "epoch": 0.1868235294117647,
      "grad_norm": 1.336791980122874,
      "learning_rate": 5.955e-05,
      "loss": 1.3906,
      "step": 397
    },
    {
      "epoch": 0.18729411764705883,
      "grad_norm": 0.8586750867440825,
      "learning_rate": 5.97e-05,
      "loss": 0.8945,
      "step": 398
    },
    {
      "epoch": 0.18776470588235294,
      "grad_norm": 1.4631316222472441,
      "learning_rate": 5.985e-05,
      "loss": 1.3047,
      "step": 399
    },
    {
      "epoch": 0.18823529411764706,
      "grad_norm": 0.9549563002799903,
      "learning_rate": 5.9999999999999995e-05,
      "loss": 0.875,
      "step": 400
    },
    {
      "epoch": 0.18870588235294117,
      "grad_norm": 1.3449747930151619,
      "learning_rate": 6.015e-05,
      "loss": 1.5469,
      "step": 401
    },
    {
      "epoch": 0.1891764705882353,
      "grad_norm": 0.8832283836326429,
      "learning_rate": 6.0299999999999995e-05,
      "loss": 0.5859,
      "step": 402
    },
    {
      "epoch": 0.18964705882352942,
      "grad_norm": 1.0453705264120712,
      "learning_rate": 6.045e-05,
      "loss": 1.2969,
      "step": 403
    },
    {
      "epoch": 0.19011764705882353,
      "grad_norm": 1.2052440885020934,
      "learning_rate": 6.0599999999999996e-05,
      "loss": 0.9375,
      "step": 404
    },
    {
      "epoch": 0.19058823529411764,
      "grad_norm": 2.8845480032296225,
      "learning_rate": 6.075e-05,
      "loss": 2.2188,
      "step": 405
    },
    {
      "epoch": 0.19105882352941175,
      "grad_norm": 0.39054328593375176,
      "learning_rate": 6.0899999999999996e-05,
      "loss": 0.207,
      "step": 406
    },
    {
      "epoch": 0.1915294117647059,
      "grad_norm": 0.8926002962176048,
      "learning_rate": 6.104999999999999e-05,
      "loss": 1.1406,
      "step": 407
    },
    {
      "epoch": 0.192,
      "grad_norm": 1.3375240273876696,
      "learning_rate": 6.12e-05,
      "loss": 0.9844,
      "step": 408
    },
    {
      "epoch": 0.19247058823529412,
      "grad_norm": 0.5382706887871008,
      "learning_rate": 6.134999999999999e-05,
      "loss": 0.1797,
      "step": 409
    },
    {
      "epoch": 0.19294117647058823,
      "grad_norm": 1.684668123205898,
      "learning_rate": 6.149999999999999e-05,
      "loss": 0.75,
      "step": 410
    },
    {
      "epoch": 0.19341176470588237,
      "grad_norm": 2.1288687779474365,
      "learning_rate": 6.165e-05,
      "loss": 1.3047,
      "step": 411
    },
    {
      "epoch": 0.19388235294117648,
      "grad_norm": 1.2648508091854476,
      "learning_rate": 6.18e-05,
      "loss": 0.707,
      "step": 412
    },
    {
      "epoch": 0.1943529411764706,
      "grad_norm": 1.6178281576298543,
      "learning_rate": 6.194999999999999e-05,
      "loss": 1.375,
      "step": 413
    },
    {
      "epoch": 0.1948235294117647,
      "grad_norm": 1.2031179478986007,
      "learning_rate": 6.209999999999999e-05,
      "loss": 1.0,
      "step": 414
    },
    {
      "epoch": 0.1952941176470588,
      "grad_norm": 0.6058474620068227,
      "learning_rate": 6.225e-05,
      "loss": 0.4941,
      "step": 415
    },
    {
      "epoch": 0.19576470588235295,
      "grad_norm": 1.068439333326216,
      "learning_rate": 6.239999999999999e-05,
      "loss": 0.377,
      "step": 416
    },
    {
      "epoch": 0.19623529411764706,
      "grad_norm": 2.755269351996511,
      "learning_rate": 6.254999999999999e-05,
      "loss": 1.1719,
      "step": 417
    },
    {
      "epoch": 0.19670588235294117,
      "grad_norm": 2.23016319647157,
      "learning_rate": 6.269999999999999e-05,
      "loss": 1.6953,
      "step": 418
    },
    {
      "epoch": 0.1971764705882353,
      "grad_norm": 1.1520504817764754,
      "learning_rate": 6.285e-05,
      "loss": 1.2031,
      "step": 419
    },
    {
      "epoch": 0.1976470588235294,
      "grad_norm": 1.5974312138748932,
      "learning_rate": 6.299999999999999e-05,
      "loss": 1.2969,
      "step": 420
    },
    {
      "epoch": 0.19811764705882354,
      "grad_norm": 1.7906149242649918,
      "learning_rate": 6.314999999999999e-05,
      "loss": 1.1797,
      "step": 421
    },
    {
      "epoch": 0.19858823529411765,
      "grad_norm": 0.8862715476725402,
      "learning_rate": 6.33e-05,
      "loss": 0.4668,
      "step": 422
    },
    {
      "epoch": 0.19905882352941176,
      "grad_norm": 1.556021271756757,
      "learning_rate": 6.345e-05,
      "loss": 1.25,
      "step": 423
    },
    {
      "epoch": 0.19952941176470587,
      "grad_norm": 1.0201493603429734,
      "learning_rate": 6.359999999999999e-05,
      "loss": 1.1094,
      "step": 424
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3645488110091752,
      "learning_rate": 6.374999999999999e-05,
      "loss": 1.5469,
      "step": 425
    },
    {
      "epoch": 0.20047058823529412,
      "grad_norm": 1.0597772920368966,
      "learning_rate": 6.39e-05,
      "loss": 1.2266,
      "step": 426
    },
    {
      "epoch": 0.20094117647058823,
      "grad_norm": 0.7316444124090044,
      "learning_rate": 6.405e-05,
      "loss": 0.2656,
      "step": 427
    },
    {
      "epoch": 0.20141176470588235,
      "grad_norm": 2.05866573965444,
      "learning_rate": 6.419999999999999e-05,
      "loss": 1.5469,
      "step": 428
    },
    {
      "epoch": 0.20188235294117646,
      "grad_norm": 1.1262765221363606,
      "learning_rate": 6.434999999999999e-05,
      "loss": 0.8789,
      "step": 429
    },
    {
      "epoch": 0.2023529411764706,
      "grad_norm": 0.8520607903461879,
      "learning_rate": 6.45e-05,
      "loss": 0.7656,
      "step": 430
    },
    {
      "epoch": 0.2028235294117647,
      "grad_norm": 1.6105498736092418,
      "learning_rate": 6.465e-05,
      "loss": 1.5859,
      "step": 431
    },
    {
      "epoch": 0.20329411764705882,
      "grad_norm": 1.2537429951201353,
      "learning_rate": 6.479999999999999e-05,
      "loss": 0.6836,
      "step": 432
    },
    {
      "epoch": 0.20376470588235293,
      "grad_norm": 3.823727373377109,
      "learning_rate": 6.494999999999999e-05,
      "loss": 1.6172,
      "step": 433
    },
    {
      "epoch": 0.20423529411764707,
      "grad_norm": 1.1958520858512973,
      "learning_rate": 6.51e-05,
      "loss": 1.0078,
      "step": 434
    },
    {
      "epoch": 0.20470588235294118,
      "grad_norm": 1.1823461321978859,
      "learning_rate": 6.525e-05,
      "loss": 0.6914,
      "step": 435
    },
    {
      "epoch": 0.2051764705882353,
      "grad_norm": 1.2212456488953718,
      "learning_rate": 6.539999999999999e-05,
      "loss": 0.4883,
      "step": 436
    },
    {
      "epoch": 0.2056470588235294,
      "grad_norm": 1.0428448212408348,
      "learning_rate": 6.555e-05,
      "loss": 0.4551,
      "step": 437
    },
    {
      "epoch": 0.20611764705882352,
      "grad_norm": 1.709310550257011,
      "learning_rate": 6.57e-05,
      "loss": 1.5781,
      "step": 438
    },
    {
      "epoch": 0.20658823529411766,
      "grad_norm": 2.4531463640017606,
      "learning_rate": 6.584999999999999e-05,
      "loss": 1.5312,
      "step": 439
    },
    {
      "epoch": 0.20705882352941177,
      "grad_norm": 1.4267313646683242,
      "learning_rate": 6.599999999999999e-05,
      "loss": 1.0234,
      "step": 440
    },
    {
      "epoch": 0.20752941176470588,
      "grad_norm": 1.1472392412451382,
      "learning_rate": 6.615e-05,
      "loss": 1.2656,
      "step": 441
    },
    {
      "epoch": 0.208,
      "grad_norm": 0.6090622430154672,
      "learning_rate": 6.63e-05,
      "loss": 0.2988,
      "step": 442
    },
    {
      "epoch": 0.20847058823529413,
      "grad_norm": 2.254088184725621,
      "learning_rate": 6.644999999999999e-05,
      "loss": 1.6719,
      "step": 443
    },
    {
      "epoch": 0.20894117647058824,
      "grad_norm": 2.56238232378244,
      "learning_rate": 6.659999999999999e-05,
      "loss": 1.7734,
      "step": 444
    },
    {
      "epoch": 0.20941176470588235,
      "grad_norm": 1.52458702507323,
      "learning_rate": 6.675e-05,
      "loss": 1.1328,
      "step": 445
    },
    {
      "epoch": 0.20988235294117646,
      "grad_norm": 0.8063397484541528,
      "learning_rate": 6.69e-05,
      "loss": 0.3418,
      "step": 446
    },
    {
      "epoch": 0.21035294117647058,
      "grad_norm": 0.8685899615479538,
      "learning_rate": 6.704999999999999e-05,
      "loss": 0.4668,
      "step": 447
    },
    {
      "epoch": 0.21082352941176472,
      "grad_norm": 1.624994618929672,
      "learning_rate": 6.72e-05,
      "loss": 0.9141,
      "step": 448
    },
    {
      "epoch": 0.21129411764705883,
      "grad_norm": 0.7308538278506913,
      "learning_rate": 6.735e-05,
      "loss": 0.3477,
      "step": 449
    },
    {
      "epoch": 0.21176470588235294,
      "grad_norm": 0.6874887607889318,
      "learning_rate": 6.75e-05,
      "loss": 0.5664,
      "step": 450
    },
    {
      "epoch": 0.21223529411764705,
      "grad_norm": 1.370944681506273,
      "learning_rate": 6.764999999999999e-05,
      "loss": 1.7578,
      "step": 451
    },
    {
      "epoch": 0.2127058823529412,
      "grad_norm": 0.8030535189385252,
      "learning_rate": 6.78e-05,
      "loss": 0.7227,
      "step": 452
    },
    {
      "epoch": 0.2131764705882353,
      "grad_norm": 0.8845515923381506,
      "learning_rate": 6.795e-05,
      "loss": 0.5664,
      "step": 453
    },
    {
      "epoch": 0.2136470588235294,
      "grad_norm": 5.325323580591081,
      "learning_rate": 6.81e-05,
      "loss": 2.2812,
      "step": 454
    },
    {
      "epoch": 0.21411764705882352,
      "grad_norm": 1.5394333184890268,
      "learning_rate": 6.824999999999999e-05,
      "loss": 1.1953,
      "step": 455
    },
    {
      "epoch": 0.21458823529411764,
      "grad_norm": 1.1226342698201575,
      "learning_rate": 6.84e-05,
      "loss": 0.8867,
      "step": 456
    },
    {
      "epoch": 0.21505882352941177,
      "grad_norm": 0.8105094908952328,
      "learning_rate": 6.855e-05,
      "loss": 0.6836,
      "step": 457
    },
    {
      "epoch": 0.21552941176470589,
      "grad_norm": 1.2662280549639562,
      "learning_rate": 6.87e-05,
      "loss": 0.7734,
      "step": 458
    },
    {
      "epoch": 0.216,
      "grad_norm": 1.6663600005981147,
      "learning_rate": 6.884999999999999e-05,
      "loss": 1.4453,
      "step": 459
    },
    {
      "epoch": 0.2164705882352941,
      "grad_norm": 2.296558666486609,
      "learning_rate": 6.9e-05,
      "loss": 2.0938,
      "step": 460
    },
    {
      "epoch": 0.21694117647058825,
      "grad_norm": 1.1475521697621294,
      "learning_rate": 6.915e-05,
      "loss": 0.9727,
      "step": 461
    },
    {
      "epoch": 0.21741176470588236,
      "grad_norm": 2.5760216723453553,
      "learning_rate": 6.93e-05,
      "loss": 1.4297,
      "step": 462
    },
    {
      "epoch": 0.21788235294117647,
      "grad_norm": 1.0902344865063864,
      "learning_rate": 6.945e-05,
      "loss": 0.6328,
      "step": 463
    },
    {
      "epoch": 0.21835294117647058,
      "grad_norm": 1.387463441829491,
      "learning_rate": 6.96e-05,
      "loss": 1.5078,
      "step": 464
    },
    {
      "epoch": 0.2188235294117647,
      "grad_norm": 1.133552992761331,
      "learning_rate": 6.975e-05,
      "loss": 0.7539,
      "step": 465
    },
    {
      "epoch": 0.21929411764705883,
      "grad_norm": 1.8318016220419266,
      "learning_rate": 6.989999999999999e-05,
      "loss": 1.2578,
      "step": 466
    },
    {
      "epoch": 0.21976470588235295,
      "grad_norm": 0.7021703341549698,
      "learning_rate": 7.005e-05,
      "loss": 0.3945,
      "step": 467
    },
    {
      "epoch": 0.22023529411764706,
      "grad_norm": 1.5361210253799942,
      "learning_rate": 7.02e-05,
      "loss": 0.9844,
      "step": 468
    },
    {
      "epoch": 0.22070588235294117,
      "grad_norm": 1.728934692768208,
      "learning_rate": 7.034999999999999e-05,
      "loss": 1.2188,
      "step": 469
    },
    {
      "epoch": 0.2211764705882353,
      "grad_norm": 1.8693569572463504,
      "learning_rate": 7.049999999999999e-05,
      "loss": 2.0469,
      "step": 470
    },
    {
      "epoch": 0.22164705882352942,
      "grad_norm": 0.9294899280388638,
      "learning_rate": 7.065e-05,
      "loss": 0.9141,
      "step": 471
    },
    {
      "epoch": 0.22211764705882353,
      "grad_norm": 1.033885892838933,
      "learning_rate": 7.079999999999999e-05,
      "loss": 0.9688,
      "step": 472
    },
    {
      "epoch": 0.22258823529411764,
      "grad_norm": 1.0832304910310724,
      "learning_rate": 7.094999999999999e-05,
      "loss": 0.9414,
      "step": 473
    },
    {
      "epoch": 0.22305882352941175,
      "grad_norm": 0.6910928616427044,
      "learning_rate": 7.11e-05,
      "loss": 0.5469,
      "step": 474
    },
    {
      "epoch": 0.2235294117647059,
      "grad_norm": 1.131874785059174,
      "learning_rate": 7.125e-05,
      "loss": 0.8711,
      "step": 475
    },
    {
      "epoch": 0.224,
      "grad_norm": 0.893787686049386,
      "learning_rate": 7.139999999999999e-05,
      "loss": 0.4551,
      "step": 476
    },
    {
      "epoch": 0.22447058823529412,
      "grad_norm": 2.7595173442434042,
      "learning_rate": 7.154999999999999e-05,
      "loss": 1.3516,
      "step": 477
    },
    {
      "epoch": 0.22494117647058823,
      "grad_norm": 0.8701262006426771,
      "learning_rate": 7.17e-05,
      "loss": 0.6094,
      "step": 478
    },
    {
      "epoch": 0.22541176470588234,
      "grad_norm": 1.0198850146190348,
      "learning_rate": 7.184999999999998e-05,
      "loss": 0.4375,
      "step": 479
    },
    {
      "epoch": 0.22588235294117648,
      "grad_norm": 2.873893077611431,
      "learning_rate": 7.199999999999999e-05,
      "loss": 1.3281,
      "step": 480
    },
    {
      "epoch": 0.2263529411764706,
      "grad_norm": 1.4080120963508949,
      "learning_rate": 7.214999999999999e-05,
      "loss": 1.0547,
      "step": 481
    },
    {
      "epoch": 0.2268235294117647,
      "grad_norm": 3.202188066532375,
      "learning_rate": 7.23e-05,
      "loss": 1.6172,
      "step": 482
    },
    {
      "epoch": 0.2272941176470588,
      "grad_norm": 1.148309100860507,
      "learning_rate": 7.244999999999999e-05,
      "loss": 0.4902,
      "step": 483
    },
    {
      "epoch": 0.22776470588235295,
      "grad_norm": 2.7272125356624217,
      "learning_rate": 7.259999999999999e-05,
      "loss": 1.2188,
      "step": 484
    },
    {
      "epoch": 0.22823529411764706,
      "grad_norm": 1.0879564104876496,
      "learning_rate": 7.274999999999999e-05,
      "loss": 0.5898,
      "step": 485
    },
    {
      "epoch": 0.22870588235294118,
      "grad_norm": 2.4632528467139543,
      "learning_rate": 7.29e-05,
      "loss": 1.4219,
      "step": 486
    },
    {
      "epoch": 0.2291764705882353,
      "grad_norm": 2.0214944895698577,
      "learning_rate": 7.304999999999999e-05,
      "loss": 1.1641,
      "step": 487
    },
    {
      "epoch": 0.2296470588235294,
      "grad_norm": 1.8609058521290294,
      "learning_rate": 7.319999999999999e-05,
      "loss": 0.8906,
      "step": 488
    },
    {
      "epoch": 0.23011764705882354,
      "grad_norm": 2.072069697079097,
      "learning_rate": 7.335e-05,
      "loss": 1.2109,
      "step": 489
    },
    {
      "epoch": 0.23058823529411765,
      "grad_norm": 1.8753133886788198,
      "learning_rate": 7.35e-05,
      "loss": 1.7422,
      "step": 490
    },
    {
      "epoch": 0.23105882352941176,
      "grad_norm": 1.017206038592311,
      "learning_rate": 7.364999999999999e-05,
      "loss": 0.5898,
      "step": 491
    },
    {
      "epoch": 0.23152941176470587,
      "grad_norm": 1.5532042749321644,
      "learning_rate": 7.379999999999999e-05,
      "loss": 0.9883,
      "step": 492
    },
    {
      "epoch": 0.232,
      "grad_norm": 32.2266186121786,
      "learning_rate": 7.395e-05,
      "loss": 1.8359,
      "step": 493
    },
    {
      "epoch": 0.23247058823529412,
      "grad_norm": 3.9944943655372267,
      "learning_rate": 7.41e-05,
      "loss": 1.4766,
      "step": 494
    },
    {
      "epoch": 0.23294117647058823,
      "grad_norm": 3.5038127678602597,
      "learning_rate": 7.424999999999999e-05,
      "loss": 0.8594,
      "step": 495
    },
    {
      "epoch": 0.23341176470588235,
      "grad_norm": 2.2036770348300054,
      "learning_rate": 7.439999999999999e-05,
      "loss": 1.2578,
      "step": 496
    },
    {
      "epoch": 0.23388235294117646,
      "grad_norm": 1.7778694456503563,
      "learning_rate": 7.455e-05,
      "loss": 1.3281,
      "step": 497
    },
    {
      "epoch": 0.2343529411764706,
      "grad_norm": 1.2168293274805184,
      "learning_rate": 7.47e-05,
      "loss": 0.9102,
      "step": 498
    },
    {
      "epoch": 0.2348235294117647,
      "grad_norm": 6.4422822821395815,
      "learning_rate": 7.484999999999999e-05,
      "loss": 1.2891,
      "step": 499
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 5.563473602403891,
      "learning_rate": 7.5e-05,
      "loss": 1.2969,
      "step": 500
    },
    {
      "epoch": 0.23576470588235293,
      "grad_norm": 1.9882782632063172,
      "learning_rate": 7.515e-05,
      "loss": 1.2109,
      "step": 501
    },
    {
      "epoch": 0.23623529411764707,
      "grad_norm": 0.9868835327057199,
      "learning_rate": 7.529999999999999e-05,
      "loss": 0.5352,
      "step": 502
    },
    {
      "epoch": 0.23670588235294118,
      "grad_norm": 3.534441351554072,
      "learning_rate": 7.544999999999999e-05,
      "loss": 2.0781,
      "step": 503
    },
    {
      "epoch": 0.2371764705882353,
      "grad_norm": 1.8306222965744197,
      "learning_rate": 7.56e-05,
      "loss": 1.4922,
      "step": 504
    },
    {
      "epoch": 0.2376470588235294,
      "grad_norm": 3.748695508403138,
      "learning_rate": 7.575e-05,
      "loss": 1.1875,
      "step": 505
    },
    {
      "epoch": 0.23811764705882352,
      "grad_norm": 1.194769874667424,
      "learning_rate": 7.589999999999999e-05,
      "loss": 0.3555,
      "step": 506
    },
    {
      "epoch": 0.23858823529411766,
      "grad_norm": 1.829829252256181,
      "learning_rate": 7.604999999999999e-05,
      "loss": 0.5938,
      "step": 507
    },
    {
      "epoch": 0.23905882352941177,
      "grad_norm": 1.4999072097090542,
      "learning_rate": 7.62e-05,
      "loss": 1.125,
      "step": 508
    },
    {
      "epoch": 0.23952941176470588,
      "grad_norm": 1.774679377775179,
      "learning_rate": 7.635e-05,
      "loss": 1.0859,
      "step": 509
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.364568297524307,
      "learning_rate": 7.649999999999999e-05,
      "loss": 1.5234,
      "step": 510
    },
    {
      "epoch": 0.24047058823529413,
      "grad_norm": 1.6030488719489129,
      "learning_rate": 7.664999999999999e-05,
      "loss": 1.0625,
      "step": 511
    },
    {
      "epoch": 0.24094117647058824,
      "grad_norm": 0.7760661542753305,
      "learning_rate": 7.68e-05,
      "loss": 0.6328,
      "step": 512
    },
    {
      "epoch": 0.24141176470588235,
      "grad_norm": 0.644972772339755,
      "learning_rate": 7.695e-05,
      "loss": 0.2207,
      "step": 513
    },
    {
      "epoch": 0.24188235294117647,
      "grad_norm": 2.6005316578013065,
      "learning_rate": 7.709999999999999e-05,
      "loss": 1.5312,
      "step": 514
    },
    {
      "epoch": 0.24235294117647058,
      "grad_norm": 1.7921943841931898,
      "learning_rate": 7.725e-05,
      "loss": 0.9375,
      "step": 515
    },
    {
      "epoch": 0.24282352941176472,
      "grad_norm": 1.0354751453997126,
      "learning_rate": 7.74e-05,
      "loss": 1.1484,
      "step": 516
    },
    {
      "epoch": 0.24329411764705883,
      "grad_norm": 0.9669490046945501,
      "learning_rate": 7.755e-05,
      "loss": 0.6641,
      "step": 517
    },
    {
      "epoch": 0.24376470588235294,
      "grad_norm": 0.5840000217519992,
      "learning_rate": 7.769999999999999e-05,
      "loss": 0.2969,
      "step": 518
    },
    {
      "epoch": 0.24423529411764705,
      "grad_norm": 2.8616473373904516,
      "learning_rate": 7.785e-05,
      "loss": 1.2969,
      "step": 519
    },
    {
      "epoch": 0.2447058823529412,
      "grad_norm": 1.169345421577944,
      "learning_rate": 7.8e-05,
      "loss": 0.6484,
      "step": 520
    },
    {
      "epoch": 0.2451764705882353,
      "grad_norm": 0.9306041572466784,
      "learning_rate": 7.815e-05,
      "loss": 0.6445,
      "step": 521
    },
    {
      "epoch": 0.2456470588235294,
      "grad_norm": 1.0821881314156205,
      "learning_rate": 7.829999999999999e-05,
      "loss": 0.7344,
      "step": 522
    },
    {
      "epoch": 0.24611764705882352,
      "grad_norm": 1.1319113128334441,
      "learning_rate": 7.845e-05,
      "loss": 0.7852,
      "step": 523
    },
    {
      "epoch": 0.24658823529411764,
      "grad_norm": 2.323791443307664,
      "learning_rate": 7.86e-05,
      "loss": 1.6406,
      "step": 524
    },
    {
      "epoch": 0.24705882352941178,
      "grad_norm": 1.492837981577801,
      "learning_rate": 7.874999999999999e-05,
      "loss": 1.1719,
      "step": 525
    },
    {
      "epoch": 0.2475294117647059,
      "grad_norm": 3.3816080306978975,
      "learning_rate": 7.89e-05,
      "loss": 2.0469,
      "step": 526
    },
    {
      "epoch": 0.248,
      "grad_norm": 1.4711717566970781,
      "learning_rate": 7.905e-05,
      "loss": 1.4766,
      "step": 527
    },
    {
      "epoch": 0.2484705882352941,
      "grad_norm": 1.022191563837131,
      "learning_rate": 7.92e-05,
      "loss": 0.459,
      "step": 528
    },
    {
      "epoch": 0.24894117647058825,
      "grad_norm": 1.3275332593996132,
      "learning_rate": 7.934999999999999e-05,
      "loss": 0.6445,
      "step": 529
    },
    {
      "epoch": 0.24941176470588236,
      "grad_norm": 1.3691267367644373,
      "learning_rate": 7.95e-05,
      "loss": 0.6953,
      "step": 530
    },
    {
      "epoch": 0.24988235294117647,
      "grad_norm": 1.2667900788729953,
      "learning_rate": 7.965e-05,
      "loss": 0.7812,
      "step": 531
    },
    {
      "epoch": 0.2503529411764706,
      "grad_norm": 1.0682280484006395,
      "learning_rate": 7.98e-05,
      "loss": 0.7539,
      "step": 532
    },
    {
      "epoch": 0.2508235294117647,
      "grad_norm": 1.9762501924871554,
      "learning_rate": 7.994999999999999e-05,
      "loss": 1.7422,
      "step": 533
    },
    {
      "epoch": 0.25129411764705883,
      "grad_norm": 2.744533073823166,
      "learning_rate": 8.01e-05,
      "loss": 1.0781,
      "step": 534
    },
    {
      "epoch": 0.25176470588235295,
      "grad_norm": 0.6110382627871339,
      "learning_rate": 8.025e-05,
      "loss": 0.2715,
      "step": 535
    },
    {
      "epoch": 0.25223529411764706,
      "grad_norm": 2.355078551352758,
      "learning_rate": 8.04e-05,
      "loss": 1.8125,
      "step": 536
    },
    {
      "epoch": 0.25270588235294117,
      "grad_norm": 1.1473388635863346,
      "learning_rate": 8.054999999999999e-05,
      "loss": 0.7539,
      "step": 537
    },
    {
      "epoch": 0.2531764705882353,
      "grad_norm": 2.59313650349663,
      "learning_rate": 8.07e-05,
      "loss": 1.2109,
      "step": 538
    },
    {
      "epoch": 0.2536470588235294,
      "grad_norm": 0.7827325398642256,
      "learning_rate": 8.085e-05,
      "loss": 0.3379,
      "step": 539
    },
    {
      "epoch": 0.2541176470588235,
      "grad_norm": 1.4008073477294432,
      "learning_rate": 8.1e-05,
      "loss": 1.2812,
      "step": 540
    },
    {
      "epoch": 0.25458823529411767,
      "grad_norm": 0.5532203862437474,
      "learning_rate": 8.115e-05,
      "loss": 0.2314,
      "step": 541
    },
    {
      "epoch": 0.2550588235294118,
      "grad_norm": 0.7522246964121714,
      "learning_rate": 8.13e-05,
      "loss": 0.3613,
      "step": 542
    },
    {
      "epoch": 0.2555294117647059,
      "grad_norm": 2.0620058323608785,
      "learning_rate": 8.145e-05,
      "loss": 1.0312,
      "step": 543
    },
    {
      "epoch": 0.256,
      "grad_norm": 0.7923272273935483,
      "learning_rate": 8.16e-05,
      "loss": 0.3203,
      "step": 544
    },
    {
      "epoch": 0.2564705882352941,
      "grad_norm": 2.7906849044096593,
      "learning_rate": 8.175e-05,
      "loss": 2.1719,
      "step": 545
    },
    {
      "epoch": 0.25694117647058823,
      "grad_norm": 2.3766326723376787,
      "learning_rate": 8.19e-05,
      "loss": 1.8672,
      "step": 546
    },
    {
      "epoch": 0.25741176470588234,
      "grad_norm": 1.1874429524732637,
      "learning_rate": 8.205e-05,
      "loss": 0.7109,
      "step": 547
    },
    {
      "epoch": 0.25788235294117645,
      "grad_norm": 1.87341072161763,
      "learning_rate": 8.22e-05,
      "loss": 1.5781,
      "step": 548
    },
    {
      "epoch": 0.25835294117647056,
      "grad_norm": 1.0360805712098646,
      "learning_rate": 8.235e-05,
      "loss": 0.4648,
      "step": 549
    },
    {
      "epoch": 0.25882352941176473,
      "grad_norm": 1.7610703989628884,
      "learning_rate": 8.25e-05,
      "loss": 0.8711,
      "step": 550
    },
    {
      "epoch": 0.25929411764705884,
      "grad_norm": 1.901735185467671,
      "learning_rate": 8.265e-05,
      "loss": 1.1953,
      "step": 551
    },
    {
      "epoch": 0.25976470588235295,
      "grad_norm": 1.1599540998313695,
      "learning_rate": 8.28e-05,
      "loss": 0.6445,
      "step": 552
    },
    {
      "epoch": 0.26023529411764706,
      "grad_norm": 1.8966701897686007,
      "learning_rate": 8.295e-05,
      "loss": 1.1719,
      "step": 553
    },
    {
      "epoch": 0.2607058823529412,
      "grad_norm": 0.9347278255367935,
      "learning_rate": 8.31e-05,
      "loss": 0.3359,
      "step": 554
    },
    {
      "epoch": 0.2611764705882353,
      "grad_norm": 1.0127196702467012,
      "learning_rate": 8.325e-05,
      "loss": 0.3965,
      "step": 555
    },
    {
      "epoch": 0.2616470588235294,
      "grad_norm": 0.8148895714721599,
      "learning_rate": 8.34e-05,
      "loss": 0.5898,
      "step": 556
    },
    {
      "epoch": 0.2621176470588235,
      "grad_norm": 0.6522153841159589,
      "learning_rate": 8.355e-05,
      "loss": 0.2637,
      "step": 557
    },
    {
      "epoch": 0.2625882352941176,
      "grad_norm": 0.9108284838290271,
      "learning_rate": 8.37e-05,
      "loss": 0.2988,
      "step": 558
    },
    {
      "epoch": 0.2630588235294118,
      "grad_norm": 1.4191696569173449,
      "learning_rate": 8.385e-05,
      "loss": 0.7852,
      "step": 559
    },
    {
      "epoch": 0.2635294117647059,
      "grad_norm": 1.7434749079267127,
      "learning_rate": 8.4e-05,
      "loss": 1.2969,
      "step": 560
    },
    {
      "epoch": 0.264,
      "grad_norm": 1.5130082175800181,
      "learning_rate": 8.415e-05,
      "loss": 0.8711,
      "step": 561
    },
    {
      "epoch": 0.2644705882352941,
      "grad_norm": 0.9462211145365491,
      "learning_rate": 8.43e-05,
      "loss": 0.4531,
      "step": 562
    },
    {
      "epoch": 0.26494117647058824,
      "grad_norm": 1.6016481678762378,
      "learning_rate": 8.444999999999998e-05,
      "loss": 1.6172,
      "step": 563
    },
    {
      "epoch": 0.26541176470588235,
      "grad_norm": 1.6039830922718643,
      "learning_rate": 8.459999999999998e-05,
      "loss": 0.7227,
      "step": 564
    },
    {
      "epoch": 0.26588235294117646,
      "grad_norm": 1.3063558500721058,
      "learning_rate": 8.474999999999999e-05,
      "loss": 1.1016,
      "step": 565
    },
    {
      "epoch": 0.26635294117647057,
      "grad_norm": 0.7885738874361724,
      "learning_rate": 8.489999999999999e-05,
      "loss": 0.457,
      "step": 566
    },
    {
      "epoch": 0.2668235294117647,
      "grad_norm": 2.0816618873459194,
      "learning_rate": 8.504999999999998e-05,
      "loss": 1.3906,
      "step": 567
    },
    {
      "epoch": 0.26729411764705885,
      "grad_norm": 1.1965043748020052,
      "learning_rate": 8.519999999999998e-05,
      "loss": 0.8867,
      "step": 568
    },
    {
      "epoch": 0.26776470588235296,
      "grad_norm": 0.7815907408664307,
      "learning_rate": 8.534999999999999e-05,
      "loss": 0.4219,
      "step": 569
    },
    {
      "epoch": 0.26823529411764707,
      "grad_norm": 0.7830118727287328,
      "learning_rate": 8.549999999999999e-05,
      "loss": 0.3652,
      "step": 570
    },
    {
      "epoch": 0.2687058823529412,
      "grad_norm": 4.8974662387242045,
      "learning_rate": 8.564999999999998e-05,
      "loss": 1.3359,
      "step": 571
    },
    {
      "epoch": 0.2691764705882353,
      "grad_norm": 22.77407678164597,
      "learning_rate": 8.579999999999998e-05,
      "loss": 1.3438,
      "step": 572
    },
    {
      "epoch": 0.2696470588235294,
      "grad_norm": 2.1622537081740303,
      "learning_rate": 8.594999999999999e-05,
      "loss": 1.7422,
      "step": 573
    },
    {
      "epoch": 0.2701176470588235,
      "grad_norm": 0.9520028632332723,
      "learning_rate": 8.609999999999999e-05,
      "loss": 0.5781,
      "step": 574
    },
    {
      "epoch": 0.27058823529411763,
      "grad_norm": 1.5551399574485705,
      "learning_rate": 8.624999999999998e-05,
      "loss": 1.1016,
      "step": 575
    },
    {
      "epoch": 0.27105882352941174,
      "grad_norm": 0.8835490614865124,
      "learning_rate": 8.639999999999999e-05,
      "loss": 0.4043,
      "step": 576
    },
    {
      "epoch": 0.2715294117647059,
      "grad_norm": 1.0572219443634914,
      "learning_rate": 8.654999999999999e-05,
      "loss": 0.9023,
      "step": 577
    },
    {
      "epoch": 0.272,
      "grad_norm": 0.7917148532212626,
      "learning_rate": 8.669999999999998e-05,
      "loss": 0.416,
      "step": 578
    },
    {
      "epoch": 0.27247058823529413,
      "grad_norm": 1.9057561750390335,
      "learning_rate": 8.684999999999998e-05,
      "loss": 1.3594,
      "step": 579
    },
    {
      "epoch": 0.27294117647058824,
      "grad_norm": 0.8559299714996063,
      "learning_rate": 8.699999999999999e-05,
      "loss": 0.3809,
      "step": 580
    },
    {
      "epoch": 0.27341176470588235,
      "grad_norm": 1.341876588084525,
      "learning_rate": 8.714999999999999e-05,
      "loss": 1.0391,
      "step": 581
    },
    {
      "epoch": 0.27388235294117647,
      "grad_norm": 1.1776589626685519,
      "learning_rate": 8.729999999999998e-05,
      "loss": 0.5039,
      "step": 582
    },
    {
      "epoch": 0.2743529411764706,
      "grad_norm": 1.3056315174804332,
      "learning_rate": 8.744999999999998e-05,
      "loss": 0.5039,
      "step": 583
    },
    {
      "epoch": 0.2748235294117647,
      "grad_norm": 1.4290801033953682,
      "learning_rate": 8.759999999999999e-05,
      "loss": 0.832,
      "step": 584
    },
    {
      "epoch": 0.2752941176470588,
      "grad_norm": 1.2608740833041303,
      "learning_rate": 8.774999999999999e-05,
      "loss": 0.5547,
      "step": 585
    },
    {
      "epoch": 0.27576470588235297,
      "grad_norm": 1.9032945779733537,
      "learning_rate": 8.789999999999998e-05,
      "loss": 1.2344,
      "step": 586
    },
    {
      "epoch": 0.2762352941176471,
      "grad_norm": 2.224497870921282,
      "learning_rate": 8.804999999999999e-05,
      "loss": 1.5938,
      "step": 587
    },
    {
      "epoch": 0.2767058823529412,
      "grad_norm": 0.8728628480808134,
      "learning_rate": 8.819999999999999e-05,
      "loss": 0.4727,
      "step": 588
    },
    {
      "epoch": 0.2771764705882353,
      "grad_norm": 1.9121672692212723,
      "learning_rate": 8.834999999999999e-05,
      "loss": 1.2656,
      "step": 589
    },
    {
      "epoch": 0.2776470588235294,
      "grad_norm": 1.5889612527959494,
      "learning_rate": 8.849999999999998e-05,
      "loss": 0.7266,
      "step": 590
    },
    {
      "epoch": 0.2781176470588235,
      "grad_norm": 1.5511974780599285,
      "learning_rate": 8.864999999999999e-05,
      "loss": 1.3203,
      "step": 591
    },
    {
      "epoch": 0.27858823529411764,
      "grad_norm": 1.8222877245108924,
      "learning_rate": 8.879999999999999e-05,
      "loss": 0.8086,
      "step": 592
    },
    {
      "epoch": 0.27905882352941175,
      "grad_norm": 1.8867371881998702,
      "learning_rate": 8.895e-05,
      "loss": 1.1016,
      "step": 593
    },
    {
      "epoch": 0.27952941176470586,
      "grad_norm": 0.5870921030666038,
      "learning_rate": 8.909999999999998e-05,
      "loss": 0.2002,
      "step": 594
    },
    {
      "epoch": 0.28,
      "grad_norm": 5.96982175434989,
      "learning_rate": 8.924999999999999e-05,
      "loss": 1.3828,
      "step": 595
    },
    {
      "epoch": 0.28047058823529414,
      "grad_norm": 2.006878322787248,
      "learning_rate": 8.939999999999999e-05,
      "loss": 1.3281,
      "step": 596
    },
    {
      "epoch": 0.28094117647058825,
      "grad_norm": 1.5350316992621356,
      "learning_rate": 8.955e-05,
      "loss": 0.9922,
      "step": 597
    },
    {
      "epoch": 0.28141176470588236,
      "grad_norm": 1.3846863906229117,
      "learning_rate": 8.969999999999998e-05,
      "loss": 0.8828,
      "step": 598
    },
    {
      "epoch": 0.2818823529411765,
      "grad_norm": 1.0808045004674323,
      "learning_rate": 8.984999999999999e-05,
      "loss": 0.9688,
      "step": 599
    },
    {
      "epoch": 0.2823529411764706,
      "grad_norm": 1.3598551406657415,
      "learning_rate": 8.999999999999999e-05,
      "loss": 0.7539,
      "step": 600
    },
    {
      "epoch": 0.2828235294117647,
      "grad_norm": 2.4845424170791874,
      "learning_rate": 9.014999999999998e-05,
      "loss": 1.5312,
      "step": 601
    },
    {
      "epoch": 0.2832941176470588,
      "grad_norm": 1.6375224235369896,
      "learning_rate": 9.029999999999999e-05,
      "loss": 0.5391,
      "step": 602
    },
    {
      "epoch": 0.2837647058823529,
      "grad_norm": 21.649881795472268,
      "learning_rate": 9.044999999999999e-05,
      "loss": 2.4375,
      "step": 603
    },
    {
      "epoch": 0.2842352941176471,
      "grad_norm": 1.719543682359555,
      "learning_rate": 9.059999999999999e-05,
      "loss": 1.2031,
      "step": 604
    },
    {
      "epoch": 0.2847058823529412,
      "grad_norm": 3.7318817108675297,
      "learning_rate": 9.074999999999998e-05,
      "loss": 2.5781,
      "step": 605
    },
    {
      "epoch": 0.2851764705882353,
      "grad_norm": 1.646908736780664,
      "learning_rate": 9.089999999999999e-05,
      "loss": 1.1562,
      "step": 606
    },
    {
      "epoch": 0.2856470588235294,
      "grad_norm": 9.964757984464082,
      "learning_rate": 9.104999999999999e-05,
      "loss": 0.75,
      "step": 607
    },
    {
      "epoch": 0.28611764705882353,
      "grad_norm": 3.322089556964142,
      "learning_rate": 9.12e-05,
      "loss": 0.4297,
      "step": 608
    },
    {
      "epoch": 0.28658823529411764,
      "grad_norm": 6.5497380996404875,
      "learning_rate": 9.134999999999998e-05,
      "loss": 1.0078,
      "step": 609
    },
    {
      "epoch": 0.28705882352941176,
      "grad_norm": 1.013140207874275,
      "learning_rate": 9.149999999999999e-05,
      "loss": 0.6094,
      "step": 610
    },
    {
      "epoch": 0.28752941176470587,
      "grad_norm": 2.3691521400700317,
      "learning_rate": 9.164999999999999e-05,
      "loss": 1.5547,
      "step": 611
    },
    {
      "epoch": 0.288,
      "grad_norm": 0.9910069619575543,
      "learning_rate": 9.18e-05,
      "loss": 0.5664,
      "step": 612
    },
    {
      "epoch": 0.28847058823529415,
      "grad_norm": 2.1805520007673276,
      "learning_rate": 9.194999999999999e-05,
      "loss": 0.6133,
      "step": 613
    },
    {
      "epoch": 0.28894117647058826,
      "grad_norm": 2.8870480469356323,
      "learning_rate": 9.209999999999999e-05,
      "loss": 1.1094,
      "step": 614
    },
    {
      "epoch": 0.28941176470588237,
      "grad_norm": 1.241932240692055,
      "learning_rate": 9.224999999999999e-05,
      "loss": 0.9492,
      "step": 615
    },
    {
      "epoch": 0.2898823529411765,
      "grad_norm": 2.533324932724292,
      "learning_rate": 9.24e-05,
      "loss": 1.6797,
      "step": 616
    },
    {
      "epoch": 0.2903529411764706,
      "grad_norm": 1.3184507594335,
      "learning_rate": 9.254999999999999e-05,
      "loss": 0.9414,
      "step": 617
    },
    {
      "epoch": 0.2908235294117647,
      "grad_norm": 1.196290133610639,
      "learning_rate": 9.269999999999999e-05,
      "loss": 0.707,
      "step": 618
    },
    {
      "epoch": 0.2912941176470588,
      "grad_norm": 1.963111220967039,
      "learning_rate": 9.285e-05,
      "loss": 1.3594,
      "step": 619
    },
    {
      "epoch": 0.2917647058823529,
      "grad_norm": 2.5900014851568187,
      "learning_rate": 9.3e-05,
      "loss": 1.4688,
      "step": 620
    },
    {
      "epoch": 0.29223529411764704,
      "grad_norm": 4.028230513711322,
      "learning_rate": 9.314999999999999e-05,
      "loss": 1.5156,
      "step": 621
    },
    {
      "epoch": 0.29270588235294115,
      "grad_norm": 2.2419943466006806,
      "learning_rate": 9.329999999999999e-05,
      "loss": 1.4922,
      "step": 622
    },
    {
      "epoch": 0.2931764705882353,
      "grad_norm": 4.092137094146537,
      "learning_rate": 9.345e-05,
      "loss": 2.9844,
      "step": 623
    },
    {
      "epoch": 0.2936470588235294,
      "grad_norm": 1.1963697679607372,
      "learning_rate": 9.36e-05,
      "loss": 0.9219,
      "step": 624
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 3.0133200745776505,
      "learning_rate": 9.374999999999999e-05,
      "loss": 1.2422,
      "step": 625
    },
    {
      "epoch": 0.29458823529411765,
      "grad_norm": 1.61059044837171,
      "learning_rate": 9.389999999999999e-05,
      "loss": 0.9609,
      "step": 626
    },
    {
      "epoch": 0.29505882352941176,
      "grad_norm": 1.2754179016741933,
      "learning_rate": 9.405e-05,
      "loss": 0.7773,
      "step": 627
    },
    {
      "epoch": 0.2955294117647059,
      "grad_norm": 2.3630717215457717,
      "learning_rate": 9.419999999999999e-05,
      "loss": 1.5469,
      "step": 628
    },
    {
      "epoch": 0.296,
      "grad_norm": 1.9261327855460038,
      "learning_rate": 9.434999999999999e-05,
      "loss": 1.3438,
      "step": 629
    },
    {
      "epoch": 0.2964705882352941,
      "grad_norm": 1.8309642209141557,
      "learning_rate": 9.449999999999999e-05,
      "loss": 1.7344,
      "step": 630
    },
    {
      "epoch": 0.2969411764705882,
      "grad_norm": 1.336644005877765,
      "learning_rate": 9.465e-05,
      "loss": 1.0859,
      "step": 631
    },
    {
      "epoch": 0.2974117647058824,
      "grad_norm": 109.44951956742975,
      "learning_rate": 9.479999999999999e-05,
      "loss": 2.5469,
      "step": 632
    },
    {
      "epoch": 0.2978823529411765,
      "grad_norm": 8.037028824651495,
      "learning_rate": 9.494999999999999e-05,
      "loss": 1.6641,
      "step": 633
    },
    {
      "epoch": 0.2983529411764706,
      "grad_norm": 13.052022638377522,
      "learning_rate": 9.51e-05,
      "loss": 3.4062,
      "step": 634
    },
    {
      "epoch": 0.2988235294117647,
      "grad_norm": 65.4282950708209,
      "learning_rate": 9.525e-05,
      "loss": 4.0312,
      "step": 635
    },
    {
      "epoch": 0.2992941176470588,
      "grad_norm": 21.79644915842182,
      "learning_rate": 9.539999999999999e-05,
      "loss": 5.1562,
      "step": 636
    },
    {
      "epoch": 0.29976470588235293,
      "grad_norm": 22.302322599204206,
      "learning_rate": 9.554999999999999e-05,
      "loss": 1.5,
      "step": 637
    },
    {
      "epoch": 0.30023529411764704,
      "grad_norm": 14.988408373541008,
      "learning_rate": 9.57e-05,
      "loss": 3.4062,
      "step": 638
    },
    {
      "epoch": 0.30070588235294116,
      "grad_norm": 24.974186414055218,
      "learning_rate": 9.585e-05,
      "loss": 5.9375,
      "step": 639
    },
    {
      "epoch": 0.30117647058823527,
      "grad_norm": 4.029779774300509,
      "learning_rate": 9.599999999999999e-05,
      "loss": 1.0547,
      "step": 640
    },
    {
      "epoch": 0.30164705882352943,
      "grad_norm": 10.294430481790688,
      "learning_rate": 9.614999999999999e-05,
      "loss": 2.0,
      "step": 641
    },
    {
      "epoch": 0.30211764705882355,
      "grad_norm": 32.13278216659781,
      "learning_rate": 9.63e-05,
      "loss": 1.9922,
      "step": 642
    },
    {
      "epoch": 0.30258823529411766,
      "grad_norm": 5.849179588435844,
      "learning_rate": 9.645e-05,
      "loss": 0.3379,
      "step": 643
    },
    {
      "epoch": 0.30305882352941177,
      "grad_norm": 11.110880220555188,
      "learning_rate": 9.659999999999999e-05,
      "loss": 2.5,
      "step": 644
    },
    {
      "epoch": 0.3035294117647059,
      "grad_norm": 16.971854308233343,
      "learning_rate": 9.675e-05,
      "loss": 3.625,
      "step": 645
    },
    {
      "epoch": 0.304,
      "grad_norm": 5.578266497301158,
      "learning_rate": 9.69e-05,
      "loss": 1.1484,
      "step": 646
    },
    {
      "epoch": 0.3044705882352941,
      "grad_norm": 6.655891442111778,
      "learning_rate": 9.705e-05,
      "loss": 1.875,
      "step": 647
    },
    {
      "epoch": 0.3049411764705882,
      "grad_norm": 18.99033733088897,
      "learning_rate": 9.719999999999999e-05,
      "loss": 3.1875,
      "step": 648
    },
    {
      "epoch": 0.3054117647058823,
      "grad_norm": 26.16093438205782,
      "learning_rate": 9.735e-05,
      "loss": 2.9062,
      "step": 649
    },
    {
      "epoch": 0.3058823529411765,
      "grad_norm": 3.9139223053352006,
      "learning_rate": 9.75e-05,
      "loss": 1.0234,
      "step": 650
    },
    {
      "epoch": 0.3063529411764706,
      "grad_norm": 9.031505139513163,
      "learning_rate": 9.764999999999999e-05,
      "loss": 1.7891,
      "step": 651
    },
    {
      "epoch": 0.3068235294117647,
      "grad_norm": 10.049098809225697,
      "learning_rate": 9.779999999999999e-05,
      "loss": 2.0781,
      "step": 652
    },
    {
      "epoch": 0.30729411764705883,
      "grad_norm": 11.0669374291548,
      "learning_rate": 9.795e-05,
      "loss": 2.5312,
      "step": 653
    },
    {
      "epoch": 0.30776470588235294,
      "grad_norm": 3.029364159361763,
      "learning_rate": 9.81e-05,
      "loss": 0.6172,
      "step": 654
    },
    {
      "epoch": 0.30823529411764705,
      "grad_norm": 5.218362057296631,
      "learning_rate": 9.824999999999999e-05,
      "loss": 1.2969,
      "step": 655
    },
    {
      "epoch": 0.30870588235294116,
      "grad_norm": 13.678859670165886,
      "learning_rate": 9.839999999999999e-05,
      "loss": 1.1016,
      "step": 656
    },
    {
      "epoch": 0.3091764705882353,
      "grad_norm": 2.4213397624947604,
      "learning_rate": 9.855e-05,
      "loss": 0.9453,
      "step": 657
    },
    {
      "epoch": 0.3096470588235294,
      "grad_norm": 4.103146933417112,
      "learning_rate": 9.87e-05,
      "loss": 2.0781,
      "step": 658
    },
    {
      "epoch": 0.31011764705882355,
      "grad_norm": 2.6688784372151773,
      "learning_rate": 9.884999999999999e-05,
      "loss": 2.1406,
      "step": 659
    },
    {
      "epoch": 0.31058823529411766,
      "grad_norm": 3.9148011605020407,
      "learning_rate": 9.9e-05,
      "loss": 1.7891,
      "step": 660
    },
    {
      "epoch": 0.3110588235294118,
      "grad_norm": 4.616970188415253,
      "learning_rate": 9.915e-05,
      "loss": 2.125,
      "step": 661
    },
    {
      "epoch": 0.3115294117647059,
      "grad_norm": 8.386318344388048,
      "learning_rate": 9.93e-05,
      "loss": 0.9414,
      "step": 662
    },
    {
      "epoch": 0.312,
      "grad_norm": 2.618972323134557,
      "learning_rate": 9.944999999999999e-05,
      "loss": 1.3594,
      "step": 663
    },
    {
      "epoch": 0.3124705882352941,
      "grad_norm": 7.085034133202365,
      "learning_rate": 9.96e-05,
      "loss": 1.5547,
      "step": 664
    },
    {
      "epoch": 0.3129411764705882,
      "grad_norm": 13.865452829192384,
      "learning_rate": 9.975e-05,
      "loss": 2.3594,
      "step": 665
    },
    {
      "epoch": 0.31341176470588233,
      "grad_norm": 7.423292734655408,
      "learning_rate": 9.99e-05,
      "loss": 1.1016,
      "step": 666
    },
    {
      "epoch": 0.31388235294117645,
      "grad_norm": 3.074832814442863,
      "learning_rate": 0.00010004999999999999,
      "loss": 1.5703,
      "step": 667
    },
    {
      "epoch": 0.3143529411764706,
      "grad_norm": 6.20157418407865,
      "learning_rate": 0.0001002,
      "loss": 1.3047,
      "step": 668
    },
    {
      "epoch": 0.3148235294117647,
      "grad_norm": 3.7453552231236182,
      "learning_rate": 0.00010035,
      "loss": 1.5,
      "step": 669
    },
    {
      "epoch": 0.31529411764705884,
      "grad_norm": 6.3278914702579785,
      "learning_rate": 0.0001005,
      "loss": 1.5469,
      "step": 670
    },
    {
      "epoch": 0.31576470588235295,
      "grad_norm": 7.9544217743363514,
      "learning_rate": 0.00010065,
      "loss": 1.8828,
      "step": 671
    },
    {
      "epoch": 0.31623529411764706,
      "grad_norm": 10.029669375092784,
      "learning_rate": 0.0001008,
      "loss": 2.6406,
      "step": 672
    },
    {
      "epoch": 0.31670588235294117,
      "grad_norm": 2.8701640375482933,
      "learning_rate": 0.00010095,
      "loss": 1.0625,
      "step": 673
    },
    {
      "epoch": 0.3171764705882353,
      "grad_norm": 9.786806227521831,
      "learning_rate": 0.0001011,
      "loss": 1.6719,
      "step": 674
    },
    {
      "epoch": 0.3176470588235294,
      "grad_norm": 9.019785767023842,
      "learning_rate": 0.00010125,
      "loss": 1.4688,
      "step": 675
    },
    {
      "epoch": 0.3181176470588235,
      "grad_norm": 2.3577980733782833,
      "learning_rate": 0.0001014,
      "loss": 0.918,
      "step": 676
    },
    {
      "epoch": 0.31858823529411767,
      "grad_norm": 11.370018136844497,
      "learning_rate": 0.00010155,
      "loss": 1.7578,
      "step": 677
    },
    {
      "epoch": 0.3190588235294118,
      "grad_norm": 4.809413310637468,
      "learning_rate": 0.00010169999999999999,
      "loss": 0.8906,
      "step": 678
    },
    {
      "epoch": 0.3195294117647059,
      "grad_norm": 3.1543534147634724,
      "learning_rate": 0.00010185,
      "loss": 0.8047,
      "step": 679
    },
    {
      "epoch": 0.32,
      "grad_norm": 3.4335008374752714,
      "learning_rate": 0.000102,
      "loss": 0.9727,
      "step": 680
    },
    {
      "epoch": 0.3204705882352941,
      "grad_norm": 11.2558519695157,
      "learning_rate": 0.00010215,
      "loss": 1.0938,
      "step": 681
    },
    {
      "epoch": 0.32094117647058823,
      "grad_norm": 5.195315533311945,
      "learning_rate": 0.00010229999999999999,
      "loss": 1.8438,
      "step": 682
    },
    {
      "epoch": 0.32141176470588234,
      "grad_norm": 9.356135744496699,
      "learning_rate": 0.00010245,
      "loss": 1.7188,
      "step": 683
    },
    {
      "epoch": 0.32188235294117645,
      "grad_norm": 4.568701156460155,
      "learning_rate": 0.0001026,
      "loss": 1.3906,
      "step": 684
    },
    {
      "epoch": 0.32235294117647056,
      "grad_norm": 10.778516283189708,
      "learning_rate": 0.00010275,
      "loss": 2.0781,
      "step": 685
    },
    {
      "epoch": 0.32282352941176473,
      "grad_norm": 10.333847017951387,
      "learning_rate": 0.0001029,
      "loss": 1.4688,
      "step": 686
    },
    {
      "epoch": 0.32329411764705884,
      "grad_norm": 4.172235502231599,
      "learning_rate": 0.00010305,
      "loss": 1.1953,
      "step": 687
    },
    {
      "epoch": 0.32376470588235295,
      "grad_norm": 2.7565805165055783,
      "learning_rate": 0.00010319999999999999,
      "loss": 1.4219,
      "step": 688
    },
    {
      "epoch": 0.32423529411764707,
      "grad_norm": 9.640771704649982,
      "learning_rate": 0.00010334999999999998,
      "loss": 2.2969,
      "step": 689
    },
    {
      "epoch": 0.3247058823529412,
      "grad_norm": 2.7640615078785227,
      "learning_rate": 0.00010349999999999998,
      "loss": 0.8086,
      "step": 690
    },
    {
      "epoch": 0.3251764705882353,
      "grad_norm": 4.099901373744279,
      "learning_rate": 0.00010364999999999999,
      "loss": 1.4922,
      "step": 691
    },
    {
      "epoch": 0.3256470588235294,
      "grad_norm": 4.952082162076946,
      "learning_rate": 0.00010379999999999999,
      "loss": 1.2031,
      "step": 692
    },
    {
      "epoch": 0.3261176470588235,
      "grad_norm": 11.81272949025244,
      "learning_rate": 0.00010394999999999998,
      "loss": 1.2578,
      "step": 693
    },
    {
      "epoch": 0.3265882352941176,
      "grad_norm": 15.938932997693493,
      "learning_rate": 0.00010409999999999998,
      "loss": 2.5625,
      "step": 694
    },
    {
      "epoch": 0.3270588235294118,
      "grad_norm": 0.8603286505738479,
      "learning_rate": 0.00010424999999999999,
      "loss": 0.375,
      "step": 695
    },
    {
      "epoch": 0.3275294117647059,
      "grad_norm": 3.1574411048343936,
      "learning_rate": 0.00010439999999999999,
      "loss": 1.5312,
      "step": 696
    },
    {
      "epoch": 0.328,
      "grad_norm": 6.023138397375894,
      "learning_rate": 0.00010454999999999998,
      "loss": 1.6328,
      "step": 697
    },
    {
      "epoch": 0.3284705882352941,
      "grad_norm": 2.9988373185444823,
      "learning_rate": 0.00010469999999999998,
      "loss": 1.3906,
      "step": 698
    },
    {
      "epoch": 0.32894117647058824,
      "grad_norm": 2.266615688239139,
      "learning_rate": 0.00010484999999999999,
      "loss": 0.8867,
      "step": 699
    },
    {
      "epoch": 0.32941176470588235,
      "grad_norm": 8.796051949970456,
      "learning_rate": 0.00010499999999999999,
      "loss": 1.5391,
      "step": 700
    },
    {
      "epoch": 0.32988235294117646,
      "grad_norm": 5.456288382952397,
      "learning_rate": 0.00010514999999999998,
      "loss": 1.6797,
      "step": 701
    },
    {
      "epoch": 0.33035294117647057,
      "grad_norm": 6.220980280268066,
      "learning_rate": 0.00010529999999999998,
      "loss": 0.8867,
      "step": 702
    },
    {
      "epoch": 0.3308235294117647,
      "grad_norm": 2.5643792896744633,
      "learning_rate": 0.00010544999999999999,
      "loss": 0.7812,
      "step": 703
    },
    {
      "epoch": 0.33129411764705885,
      "grad_norm": 2.071936180472948,
      "learning_rate": 0.00010559999999999998,
      "loss": 1.2578,
      "step": 704
    },
    {
      "epoch": 0.33176470588235296,
      "grad_norm": 8.489600718442029,
      "learning_rate": 0.00010574999999999998,
      "loss": 1.2891,
      "step": 705
    },
    {
      "epoch": 0.3322352941176471,
      "grad_norm": 5.8073446139427425,
      "learning_rate": 0.00010589999999999999,
      "loss": 1.0391,
      "step": 706
    },
    {
      "epoch": 0.3327058823529412,
      "grad_norm": 7.674809038687456,
      "learning_rate": 0.00010604999999999999,
      "loss": 1.5469,
      "step": 707
    },
    {
      "epoch": 0.3331764705882353,
      "grad_norm": 1.44916257937421,
      "learning_rate": 0.00010619999999999998,
      "loss": 1.0391,
      "step": 708
    },
    {
      "epoch": 0.3336470588235294,
      "grad_norm": 3.468340715532432,
      "learning_rate": 0.00010634999999999998,
      "loss": 1.7031,
      "step": 709
    },
    {
      "epoch": 0.3341176470588235,
      "grad_norm": 1.0418206386171942,
      "learning_rate": 0.00010649999999999999,
      "loss": 0.9805,
      "step": 710
    },
    {
      "epoch": 0.33458823529411763,
      "grad_norm": 3.253348402024685,
      "learning_rate": 0.00010664999999999999,
      "loss": 0.9141,
      "step": 711
    },
    {
      "epoch": 0.33505882352941174,
      "grad_norm": 2.48448009135229,
      "learning_rate": 0.00010679999999999998,
      "loss": 0.3809,
      "step": 712
    },
    {
      "epoch": 0.3355294117647059,
      "grad_norm": 8.34321363160101,
      "learning_rate": 0.00010694999999999998,
      "loss": 0.9727,
      "step": 713
    },
    {
      "epoch": 0.336,
      "grad_norm": 42.31868252201944,
      "learning_rate": 0.00010709999999999999,
      "loss": 2.0938,
      "step": 714
    },
    {
      "epoch": 0.33647058823529413,
      "grad_norm": 2.0301649932290853,
      "learning_rate": 0.00010724999999999999,
      "loss": 0.7695,
      "step": 715
    },
    {
      "epoch": 0.33694117647058824,
      "grad_norm": 3.2562821874712533,
      "learning_rate": 0.00010739999999999998,
      "loss": 2.0469,
      "step": 716
    },
    {
      "epoch": 0.33741176470588236,
      "grad_norm": 2.343485044984807,
      "learning_rate": 0.00010754999999999999,
      "loss": 1.0938,
      "step": 717
    },
    {
      "epoch": 0.33788235294117647,
      "grad_norm": 6.337773869032026,
      "learning_rate": 0.00010769999999999999,
      "loss": 1.5625,
      "step": 718
    },
    {
      "epoch": 0.3383529411764706,
      "grad_norm": 4.030955869794676,
      "learning_rate": 0.00010784999999999999,
      "loss": 0.1826,
      "step": 719
    },
    {
      "epoch": 0.3388235294117647,
      "grad_norm": 1.0653899286305355,
      "learning_rate": 0.00010799999999999998,
      "loss": 1.0078,
      "step": 720
    },
    {
      "epoch": 0.3392941176470588,
      "grad_norm": 5.114186077383256,
      "learning_rate": 0.00010814999999999999,
      "loss": 1.2422,
      "step": 721
    },
    {
      "epoch": 0.33976470588235297,
      "grad_norm": 7.3273835898151445,
      "learning_rate": 0.00010829999999999999,
      "loss": 0.9102,
      "step": 722
    },
    {
      "epoch": 0.3402352941176471,
      "grad_norm": 0.73864604423886,
      "learning_rate": 0.00010845,
      "loss": 0.3418,
      "step": 723
    },
    {
      "epoch": 0.3407058823529412,
      "grad_norm": 4.418443398007158,
      "learning_rate": 0.00010859999999999998,
      "loss": 0.7383,
      "step": 724
    },
    {
      "epoch": 0.3411764705882353,
      "grad_norm": 52.96933895657383,
      "learning_rate": 0.00010874999999999999,
      "loss": 2.6562,
      "step": 725
    },
    {
      "epoch": 0.3416470588235294,
      "grad_norm": 1.0694948791099197,
      "learning_rate": 0.00010889999999999999,
      "loss": 0.3789,
      "step": 726
    },
    {
      "epoch": 0.3421176470588235,
      "grad_norm": 10.569140915020917,
      "learning_rate": 0.00010904999999999998,
      "loss": 1.4766,
      "step": 727
    },
    {
      "epoch": 0.34258823529411764,
      "grad_norm": 7.425494443307618,
      "learning_rate": 0.00010919999999999998,
      "loss": 1.3516,
      "step": 728
    },
    {
      "epoch": 0.34305882352941175,
      "grad_norm": 3.1231012781928467,
      "learning_rate": 0.00010934999999999999,
      "loss": 1.375,
      "step": 729
    },
    {
      "epoch": 0.34352941176470586,
      "grad_norm": 1.59125730646577,
      "learning_rate": 0.00010949999999999999,
      "loss": 0.6055,
      "step": 730
    },
    {
      "epoch": 0.344,
      "grad_norm": 3.999043833516949,
      "learning_rate": 0.00010964999999999998,
      "loss": 0.543,
      "step": 731
    },
    {
      "epoch": 0.34447058823529414,
      "grad_norm": 9.515937470018997,
      "learning_rate": 0.00010979999999999999,
      "loss": 0.8164,
      "step": 732
    },
    {
      "epoch": 0.34494117647058825,
      "grad_norm": 2.7200776991453393,
      "learning_rate": 0.00010994999999999999,
      "loss": 1.1953,
      "step": 733
    },
    {
      "epoch": 0.34541176470588236,
      "grad_norm": 1.6089445631897341,
      "learning_rate": 0.00011009999999999999,
      "loss": 0.8477,
      "step": 734
    },
    {
      "epoch": 0.3458823529411765,
      "grad_norm": 2.45422679017346,
      "learning_rate": 0.00011024999999999998,
      "loss": 1.2422,
      "step": 735
    },
    {
      "epoch": 0.3463529411764706,
      "grad_norm": 3.600072067233595,
      "learning_rate": 0.00011039999999999999,
      "loss": 1.1406,
      "step": 736
    },
    {
      "epoch": 0.3468235294117647,
      "grad_norm": 15.230812691726223,
      "learning_rate": 0.00011054999999999999,
      "loss": 0.9766,
      "step": 737
    },
    {
      "epoch": 0.3472941176470588,
      "grad_norm": 1.0120835436279372,
      "learning_rate": 0.0001107,
      "loss": 0.6445,
      "step": 738
    },
    {
      "epoch": 0.3477647058823529,
      "grad_norm": 1.4237014439455198,
      "learning_rate": 0.00011084999999999998,
      "loss": 0.8242,
      "step": 739
    },
    {
      "epoch": 0.34823529411764703,
      "grad_norm": 0.9508331561574108,
      "learning_rate": 0.00011099999999999999,
      "loss": 0.4551,
      "step": 740
    },
    {
      "epoch": 0.3487058823529412,
      "grad_norm": 2.2416673359528,
      "learning_rate": 0.00011114999999999999,
      "loss": 0.6641,
      "step": 741
    },
    {
      "epoch": 0.3491764705882353,
      "grad_norm": 12.467609040903312,
      "learning_rate": 0.0001113,
      "loss": 1.4609,
      "step": 742
    },
    {
      "epoch": 0.3496470588235294,
      "grad_norm": 1.8835041913721196,
      "learning_rate": 0.00011144999999999998,
      "loss": 1.3438,
      "step": 743
    },
    {
      "epoch": 0.35011764705882353,
      "grad_norm": 1.3919790270729808,
      "learning_rate": 0.00011159999999999999,
      "loss": 0.4121,
      "step": 744
    },
    {
      "epoch": 0.35058823529411764,
      "grad_norm": 2.2125064022115075,
      "learning_rate": 0.00011174999999999999,
      "loss": 0.8555,
      "step": 745
    },
    {
      "epoch": 0.35105882352941176,
      "grad_norm": 2.5860257235932798,
      "learning_rate": 0.0001119,
      "loss": 1.1875,
      "step": 746
    },
    {
      "epoch": 0.35152941176470587,
      "grad_norm": 1.3256920521389264,
      "learning_rate": 0.00011204999999999999,
      "loss": 0.793,
      "step": 747
    },
    {
      "epoch": 0.352,
      "grad_norm": 2.3465720805184995,
      "learning_rate": 0.00011219999999999999,
      "loss": 1.4375,
      "step": 748
    },
    {
      "epoch": 0.3524705882352941,
      "grad_norm": 1.0263219169919384,
      "learning_rate": 0.00011235,
      "loss": 0.5469,
      "step": 749
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.6106423729991617,
      "learning_rate": 0.0001125,
      "loss": 0.3418,
      "step": 750
    },
    {
      "epoch": 0.35341176470588237,
      "grad_norm": 1.3433736416662487,
      "learning_rate": 0.00011264999999999999,
      "loss": 0.832,
      "step": 751
    },
    {
      "epoch": 0.3538823529411765,
      "grad_norm": 1.4559456442996241,
      "learning_rate": 0.00011279999999999999,
      "loss": 0.8359,
      "step": 752
    },
    {
      "epoch": 0.3543529411764706,
      "grad_norm": 1.0625148488495102,
      "learning_rate": 0.00011295,
      "loss": 0.5312,
      "step": 753
    },
    {
      "epoch": 0.3548235294117647,
      "grad_norm": 1.4165298951313297,
      "learning_rate": 0.00011309999999999998,
      "loss": 0.8906,
      "step": 754
    },
    {
      "epoch": 0.3552941176470588,
      "grad_norm": 1.1096406228047415,
      "learning_rate": 0.00011324999999999999,
      "loss": 0.5273,
      "step": 755
    },
    {
      "epoch": 0.3557647058823529,
      "grad_norm": 1.7774692002470496,
      "learning_rate": 0.00011339999999999999,
      "loss": 1.5703,
      "step": 756
    },
    {
      "epoch": 0.35623529411764704,
      "grad_norm": 1.0761944615083878,
      "learning_rate": 0.00011355,
      "loss": 0.6367,
      "step": 757
    },
    {
      "epoch": 0.35670588235294115,
      "grad_norm": 2.444944201856324,
      "learning_rate": 0.00011369999999999999,
      "loss": 1.4688,
      "step": 758
    },
    {
      "epoch": 0.3571764705882353,
      "grad_norm": 1.6891685515157406,
      "learning_rate": 0.00011384999999999999,
      "loss": 1.1328,
      "step": 759
    },
    {
      "epoch": 0.35764705882352943,
      "grad_norm": 1.957549759444073,
      "learning_rate": 0.00011399999999999999,
      "loss": 1.4141,
      "step": 760
    },
    {
      "epoch": 0.35811764705882354,
      "grad_norm": 0.6934747289897923,
      "learning_rate": 0.00011415,
      "loss": 0.457,
      "step": 761
    },
    {
      "epoch": 0.35858823529411765,
      "grad_norm": 1.7190479403531875,
      "learning_rate": 0.00011429999999999999,
      "loss": 1.5156,
      "step": 762
    },
    {
      "epoch": 0.35905882352941176,
      "grad_norm": 1.6369729046779817,
      "learning_rate": 0.00011444999999999999,
      "loss": 0.9609,
      "step": 763
    },
    {
      "epoch": 0.3595294117647059,
      "grad_norm": 1.4921113581428254,
      "learning_rate": 0.0001146,
      "loss": 0.9492,
      "step": 764
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.469776670742512,
      "learning_rate": 0.00011475,
      "loss": 0.6641,
      "step": 765
    },
    {
      "epoch": 0.3604705882352941,
      "grad_norm": 1.288690034507791,
      "learning_rate": 0.00011489999999999999,
      "loss": 0.5664,
      "step": 766
    },
    {
      "epoch": 0.3609411764705882,
      "grad_norm": 2.016285410580494,
      "learning_rate": 0.00011504999999999999,
      "loss": 0.9883,
      "step": 767
    },
    {
      "epoch": 0.3614117647058824,
      "grad_norm": 1.378692016775361,
      "learning_rate": 0.0001152,
      "loss": 0.8984,
      "step": 768
    },
    {
      "epoch": 0.3618823529411765,
      "grad_norm": 0.8031325935126996,
      "learning_rate": 0.00011535,
      "loss": 0.4355,
      "step": 769
    },
    {
      "epoch": 0.3623529411764706,
      "grad_norm": 1.0846803188934933,
      "learning_rate": 0.00011549999999999999,
      "loss": 0.6992,
      "step": 770
    },
    {
      "epoch": 0.3628235294117647,
      "grad_norm": 1.6887260054581759,
      "learning_rate": 0.00011564999999999999,
      "loss": 1.0,
      "step": 771
    },
    {
      "epoch": 0.3632941176470588,
      "grad_norm": 0.8880698434106633,
      "learning_rate": 0.0001158,
      "loss": 0.5039,
      "step": 772
    },
    {
      "epoch": 0.36376470588235293,
      "grad_norm": 1.1854568972844048,
      "learning_rate": 0.00011595,
      "loss": 0.4297,
      "step": 773
    },
    {
      "epoch": 0.36423529411764705,
      "grad_norm": 0.9323231243879929,
      "learning_rate": 0.00011609999999999999,
      "loss": 0.6172,
      "step": 774
    },
    {
      "epoch": 0.36470588235294116,
      "grad_norm": 0.4801146226090603,
      "learning_rate": 0.00011624999999999999,
      "loss": 0.1855,
      "step": 775
    },
    {
      "epoch": 0.36517647058823527,
      "grad_norm": 1.1243990529490224,
      "learning_rate": 0.0001164,
      "loss": 0.8203,
      "step": 776
    },
    {
      "epoch": 0.36564705882352944,
      "grad_norm": 1.7979913881796368,
      "learning_rate": 0.00011654999999999999,
      "loss": 1.2734,
      "step": 777
    },
    {
      "epoch": 0.36611764705882355,
      "grad_norm": 1.772119656677502,
      "learning_rate": 0.00011669999999999999,
      "loss": 1.3516,
      "step": 778
    },
    {
      "epoch": 0.36658823529411766,
      "grad_norm": 1.294833226544372,
      "learning_rate": 0.00011685,
      "loss": 1.0078,
      "step": 779
    },
    {
      "epoch": 0.36705882352941177,
      "grad_norm": 1.4845801952638495,
      "learning_rate": 0.000117,
      "loss": 0.8789,
      "step": 780
    },
    {
      "epoch": 0.3675294117647059,
      "grad_norm": 0.9902621868914657,
      "learning_rate": 0.00011714999999999999,
      "loss": 0.5625,
      "step": 781
    },
    {
      "epoch": 0.368,
      "grad_norm": 1.4661006149265114,
      "learning_rate": 0.00011729999999999999,
      "loss": 1.0469,
      "step": 782
    },
    {
      "epoch": 0.3684705882352941,
      "grad_norm": 2.344334806998753,
      "learning_rate": 0.00011745,
      "loss": 1.5938,
      "step": 783
    },
    {
      "epoch": 0.3689411764705882,
      "grad_norm": 1.38569382303774,
      "learning_rate": 0.0001176,
      "loss": 1.1641,
      "step": 784
    },
    {
      "epoch": 0.36941176470588233,
      "grad_norm": 0.687631299825845,
      "learning_rate": 0.00011774999999999999,
      "loss": 0.3574,
      "step": 785
    },
    {
      "epoch": 0.3698823529411765,
      "grad_norm": 1.9747697938632167,
      "learning_rate": 0.00011789999999999999,
      "loss": 1.2344,
      "step": 786
    },
    {
      "epoch": 0.3703529411764706,
      "grad_norm": 2.2448214719818202,
      "learning_rate": 0.00011805,
      "loss": 1.2109,
      "step": 787
    },
    {
      "epoch": 0.3708235294117647,
      "grad_norm": 1.0155929114601483,
      "learning_rate": 0.0001182,
      "loss": 0.8281,
      "step": 788
    },
    {
      "epoch": 0.37129411764705883,
      "grad_norm": 1.3825613388146971,
      "learning_rate": 0.00011834999999999999,
      "loss": 0.875,
      "step": 789
    },
    {
      "epoch": 0.37176470588235294,
      "grad_norm": 1.0014102548335637,
      "learning_rate": 0.0001185,
      "loss": 0.6836,
      "step": 790
    },
    {
      "epoch": 0.37223529411764705,
      "grad_norm": 1.7629194879099255,
      "learning_rate": 0.00011865,
      "loss": 1.5,
      "step": 791
    },
    {
      "epoch": 0.37270588235294116,
      "grad_norm": 1.0891454017516056,
      "learning_rate": 0.0001188,
      "loss": 0.6641,
      "step": 792
    },
    {
      "epoch": 0.3731764705882353,
      "grad_norm": 1.4194643430506897,
      "learning_rate": 0.00011894999999999999,
      "loss": 0.8867,
      "step": 793
    },
    {
      "epoch": 0.3736470588235294,
      "grad_norm": 1.9604478113893522,
      "learning_rate": 0.0001191,
      "loss": 1.1328,
      "step": 794
    },
    {
      "epoch": 0.37411764705882355,
      "grad_norm": 1.2193393738528813,
      "learning_rate": 0.00011925,
      "loss": 1.1172,
      "step": 795
    },
    {
      "epoch": 0.37458823529411767,
      "grad_norm": 0.7922766124950656,
      "learning_rate": 0.0001194,
      "loss": 0.5898,
      "step": 796
    },
    {
      "epoch": 0.3750588235294118,
      "grad_norm": 0.7215347075631988,
      "learning_rate": 0.00011954999999999999,
      "loss": 0.2734,
      "step": 797
    },
    {
      "epoch": 0.3755294117647059,
      "grad_norm": 0.6079282674167896,
      "learning_rate": 0.0001197,
      "loss": 0.2559,
      "step": 798
    },
    {
      "epoch": 0.376,
      "grad_norm": 1.3742307671330496,
      "learning_rate": 0.00011985,
      "loss": 0.6133,
      "step": 799
    },
    {
      "epoch": 0.3764705882352941,
      "grad_norm": 1.3882822383898843,
      "learning_rate": 0.00011999999999999999,
      "loss": 0.5742,
      "step": 800
    },
    {
      "epoch": 0.3769411764705882,
      "grad_norm": 1.6262654575336717,
      "learning_rate": 0.00012014999999999999,
      "loss": 1.0078,
      "step": 801
    },
    {
      "epoch": 0.37741176470588234,
      "grad_norm": 0.7610713861730333,
      "learning_rate": 0.0001203,
      "loss": 0.5117,
      "step": 802
    },
    {
      "epoch": 0.37788235294117645,
      "grad_norm": 1.2305200699127934,
      "learning_rate": 0.00012045,
      "loss": 0.8555,
      "step": 803
    },
    {
      "epoch": 0.3783529411764706,
      "grad_norm": 1.1510540832317053,
      "learning_rate": 0.00012059999999999999,
      "loss": 0.5898,
      "step": 804
    },
    {
      "epoch": 0.3788235294117647,
      "grad_norm": 1.809097980033926,
      "learning_rate": 0.00012075,
      "loss": 1.2812,
      "step": 805
    },
    {
      "epoch": 0.37929411764705884,
      "grad_norm": 1.5420762400590327,
      "learning_rate": 0.0001209,
      "loss": 1.4141,
      "step": 806
    },
    {
      "epoch": 0.37976470588235295,
      "grad_norm": 1.1472754695530651,
      "learning_rate": 0.00012105,
      "loss": 0.8711,
      "step": 807
    },
    {
      "epoch": 0.38023529411764706,
      "grad_norm": 1.8264552304810129,
      "learning_rate": 0.00012119999999999999,
      "loss": 1.1641,
      "step": 808
    },
    {
      "epoch": 0.38070588235294117,
      "grad_norm": 1.1643814919938769,
      "learning_rate": 0.00012135,
      "loss": 0.4648,
      "step": 809
    },
    {
      "epoch": 0.3811764705882353,
      "grad_norm": 1.0468980857972636,
      "learning_rate": 0.0001215,
      "loss": 0.6211,
      "step": 810
    },
    {
      "epoch": 0.3816470588235294,
      "grad_norm": 1.0153830226586948,
      "learning_rate": 0.00012165,
      "loss": 0.5195,
      "step": 811
    },
    {
      "epoch": 0.3821176470588235,
      "grad_norm": 1.127251427609683,
      "learning_rate": 0.00012179999999999999,
      "loss": 0.6992,
      "step": 812
    },
    {
      "epoch": 0.3825882352941177,
      "grad_norm": 1.5838991545321426,
      "learning_rate": 0.00012194999999999998,
      "loss": 1.0,
      "step": 813
    },
    {
      "epoch": 0.3830588235294118,
      "grad_norm": 1.4577187507451153,
      "learning_rate": 0.00012209999999999999,
      "loss": 1.1406,
      "step": 814
    },
    {
      "epoch": 0.3835294117647059,
      "grad_norm": 0.5612167223622818,
      "learning_rate": 0.00012225,
      "loss": 0.2012,
      "step": 815
    },
    {
      "epoch": 0.384,
      "grad_norm": 1.3065253280338713,
      "learning_rate": 0.0001224,
      "loss": 1.1641,
      "step": 816
    },
    {
      "epoch": 0.3844705882352941,
      "grad_norm": 0.35797939803795276,
      "learning_rate": 0.00012254999999999997,
      "loss": 0.1934,
      "step": 817
    },
    {
      "epoch": 0.38494117647058823,
      "grad_norm": 1.9396305365963797,
      "learning_rate": 0.00012269999999999997,
      "loss": 1.6016,
      "step": 818
    },
    {
      "epoch": 0.38541176470588234,
      "grad_norm": 1.937087544521548,
      "learning_rate": 0.00012284999999999998,
      "loss": 1.7266,
      "step": 819
    },
    {
      "epoch": 0.38588235294117645,
      "grad_norm": 1.6492059730395436,
      "learning_rate": 0.00012299999999999998,
      "loss": 1.3047,
      "step": 820
    },
    {
      "epoch": 0.38635294117647057,
      "grad_norm": 0.8206411233009858,
      "learning_rate": 0.00012314999999999998,
      "loss": 0.625,
      "step": 821
    },
    {
      "epoch": 0.38682352941176473,
      "grad_norm": 1.7846976284648464,
      "learning_rate": 0.0001233,
      "loss": 1.1562,
      "step": 822
    },
    {
      "epoch": 0.38729411764705884,
      "grad_norm": 0.2671214617883406,
      "learning_rate": 0.00012345,
      "loss": 0.106,
      "step": 823
    },
    {
      "epoch": 0.38776470588235296,
      "grad_norm": 25.94345855172889,
      "learning_rate": 0.0001236,
      "loss": 1.4922,
      "step": 824
    },
    {
      "epoch": 0.38823529411764707,
      "grad_norm": 2.0346168829395928,
      "learning_rate": 0.00012374999999999997,
      "loss": 0.9727,
      "step": 825
    },
    {
      "epoch": 0.3887058823529412,
      "grad_norm": 30.922210818002554,
      "learning_rate": 0.00012389999999999998,
      "loss": 3.2344,
      "step": 826
    },
    {
      "epoch": 0.3891764705882353,
      "grad_norm": 12.672771472972101,
      "learning_rate": 0.00012404999999999998,
      "loss": 0.9219,
      "step": 827
    },
    {
      "epoch": 0.3896470588235294,
      "grad_norm": 2.2851518798660337,
      "learning_rate": 0.00012419999999999998,
      "loss": 1.125,
      "step": 828
    },
    {
      "epoch": 0.3901176470588235,
      "grad_norm": 2.0349901445891594,
      "learning_rate": 0.00012435,
      "loss": 1.0938,
      "step": 829
    },
    {
      "epoch": 0.3905882352941176,
      "grad_norm": 4.6600966085351265,
      "learning_rate": 0.0001245,
      "loss": 1.9141,
      "step": 830
    },
    {
      "epoch": 0.3910588235294118,
      "grad_norm": 17.788808379743955,
      "learning_rate": 0.00012465,
      "loss": 1.1562,
      "step": 831
    },
    {
      "epoch": 0.3915294117647059,
      "grad_norm": 1.279953042860862,
      "learning_rate": 0.00012479999999999997,
      "loss": 0.6016,
      "step": 832
    },
    {
      "epoch": 0.392,
      "grad_norm": 6.824219840463681,
      "learning_rate": 0.00012494999999999997,
      "loss": 2.0625,
      "step": 833
    },
    {
      "epoch": 0.3924705882352941,
      "grad_norm": 5.157714911210055,
      "learning_rate": 0.00012509999999999998,
      "loss": 1.3359,
      "step": 834
    },
    {
      "epoch": 0.39294117647058824,
      "grad_norm": 0.8599288542472394,
      "learning_rate": 0.00012524999999999998,
      "loss": 0.2754,
      "step": 835
    },
    {
      "epoch": 0.39341176470588235,
      "grad_norm": 3.106808824741971,
      "learning_rate": 0.00012539999999999999,
      "loss": 1.7656,
      "step": 836
    },
    {
      "epoch": 0.39388235294117646,
      "grad_norm": 2.335395827493474,
      "learning_rate": 0.00012555,
      "loss": 0.2754,
      "step": 837
    },
    {
      "epoch": 0.3943529411764706,
      "grad_norm": 1.498424343332123,
      "learning_rate": 0.0001257,
      "loss": 0.7695,
      "step": 838
    },
    {
      "epoch": 0.3948235294117647,
      "grad_norm": 1.5319966154475004,
      "learning_rate": 0.00012585,
      "loss": 0.7812,
      "step": 839
    },
    {
      "epoch": 0.3952941176470588,
      "grad_norm": 1.6112238795286444,
      "learning_rate": 0.00012599999999999997,
      "loss": 0.8906,
      "step": 840
    },
    {
      "epoch": 0.39576470588235296,
      "grad_norm": 6.884043768983477,
      "learning_rate": 0.00012614999999999998,
      "loss": 1.1719,
      "step": 841
    },
    {
      "epoch": 0.3962352941176471,
      "grad_norm": 0.5222967903330523,
      "learning_rate": 0.00012629999999999998,
      "loss": 0.2178,
      "step": 842
    },
    {
      "epoch": 0.3967058823529412,
      "grad_norm": 4.725681341744438,
      "learning_rate": 0.00012644999999999998,
      "loss": 0.5742,
      "step": 843
    },
    {
      "epoch": 0.3971764705882353,
      "grad_norm": 1.2779432278271647,
      "learning_rate": 0.0001266,
      "loss": 0.5117,
      "step": 844
    },
    {
      "epoch": 0.3976470588235294,
      "grad_norm": 0.4996136241718225,
      "learning_rate": 0.00012675,
      "loss": 0.0752,
      "step": 845
    },
    {
      "epoch": 0.3981176470588235,
      "grad_norm": 35.92109625616015,
      "learning_rate": 0.0001269,
      "loss": 4.1875,
      "step": 846
    },
    {
      "epoch": 0.39858823529411763,
      "grad_norm": 3.5883534024588855,
      "learning_rate": 0.00012705,
      "loss": 1.9453,
      "step": 847
    },
    {
      "epoch": 0.39905882352941174,
      "grad_norm": 1.940945924756021,
      "learning_rate": 0.00012719999999999997,
      "loss": 1.3672,
      "step": 848
    },
    {
      "epoch": 0.39952941176470586,
      "grad_norm": 4.914306037134052,
      "learning_rate": 0.00012734999999999998,
      "loss": 1.4453,
      "step": 849
    },
    {
      "epoch": 0.4,
      "grad_norm": 3.1772729795171033,
      "learning_rate": 0.00012749999999999998,
      "loss": 0.3711,
      "step": 850
    },
    {
      "epoch": 0.40047058823529413,
      "grad_norm": 2.8279556436877034,
      "learning_rate": 0.00012764999999999999,
      "loss": 1.7969,
      "step": 851
    },
    {
      "epoch": 0.40094117647058825,
      "grad_norm": 1.347611583821308,
      "learning_rate": 0.0001278,
      "loss": 1.1953,
      "step": 852
    },
    {
      "epoch": 0.40141176470588236,
      "grad_norm": 1.318793683536631,
      "learning_rate": 0.00012795,
      "loss": 0.7695,
      "step": 853
    },
    {
      "epoch": 0.40188235294117647,
      "grad_norm": 1.6025483323910241,
      "learning_rate": 0.0001281,
      "loss": 1.0078,
      "step": 854
    },
    {
      "epoch": 0.4023529411764706,
      "grad_norm": 0.7090083133700753,
      "learning_rate": 0.00012824999999999997,
      "loss": 0.4688,
      "step": 855
    },
    {
      "epoch": 0.4028235294117647,
      "grad_norm": 1.3920482384879924,
      "learning_rate": 0.00012839999999999998,
      "loss": 0.9219,
      "step": 856
    },
    {
      "epoch": 0.4032941176470588,
      "grad_norm": 1.8194004206244012,
      "learning_rate": 0.00012854999999999998,
      "loss": 1.6562,
      "step": 857
    },
    {
      "epoch": 0.4037647058823529,
      "grad_norm": 1.7503376468053455,
      "learning_rate": 0.00012869999999999998,
      "loss": 1.1641,
      "step": 858
    },
    {
      "epoch": 0.4042352941176471,
      "grad_norm": 1.5744284694511936,
      "learning_rate": 0.00012885,
      "loss": 1.1094,
      "step": 859
    },
    {
      "epoch": 0.4047058823529412,
      "grad_norm": 1.339629786087434,
      "learning_rate": 0.000129,
      "loss": 0.7266,
      "step": 860
    },
    {
      "epoch": 0.4051764705882353,
      "grad_norm": 0.937932540591346,
      "learning_rate": 0.00012915,
      "loss": 0.793,
      "step": 861
    },
    {
      "epoch": 0.4056470588235294,
      "grad_norm": 1.5676461450323653,
      "learning_rate": 0.0001293,
      "loss": 0.9219,
      "step": 862
    },
    {
      "epoch": 0.4061176470588235,
      "grad_norm": 0.8585405007440612,
      "learning_rate": 0.00012944999999999998,
      "loss": 0.4668,
      "step": 863
    },
    {
      "epoch": 0.40658823529411764,
      "grad_norm": 1.062117088391807,
      "learning_rate": 0.00012959999999999998,
      "loss": 0.707,
      "step": 864
    },
    {
      "epoch": 0.40705882352941175,
      "grad_norm": 1.9372668410048637,
      "learning_rate": 0.00012974999999999998,
      "loss": 1.2812,
      "step": 865
    },
    {
      "epoch": 0.40752941176470586,
      "grad_norm": 3.5632151065173687,
      "learning_rate": 0.00012989999999999999,
      "loss": 0.6797,
      "step": 866
    },
    {
      "epoch": 0.408,
      "grad_norm": 1.5234189841148775,
      "learning_rate": 0.00013005,
      "loss": 1.2422,
      "step": 867
    },
    {
      "epoch": 0.40847058823529414,
      "grad_norm": 1.325036666942065,
      "learning_rate": 0.0001302,
      "loss": 1.1719,
      "step": 868
    },
    {
      "epoch": 0.40894117647058825,
      "grad_norm": 1.49239823853274,
      "learning_rate": 0.00013035,
      "loss": 0.9609,
      "step": 869
    },
    {
      "epoch": 0.40941176470588236,
      "grad_norm": 1.4571162949081258,
      "learning_rate": 0.0001305,
      "loss": 0.8672,
      "step": 870
    },
    {
      "epoch": 0.4098823529411765,
      "grad_norm": 1.3134493540698988,
      "learning_rate": 0.00013064999999999998,
      "loss": 0.9805,
      "step": 871
    },
    {
      "epoch": 0.4103529411764706,
      "grad_norm": 2.9514625182157923,
      "learning_rate": 0.00013079999999999998,
      "loss": 1.6328,
      "step": 872
    },
    {
      "epoch": 0.4108235294117647,
      "grad_norm": 1.675722468146762,
      "learning_rate": 0.00013094999999999998,
      "loss": 1.1406,
      "step": 873
    },
    {
      "epoch": 0.4112941176470588,
      "grad_norm": 2.310590795846875,
      "learning_rate": 0.0001311,
      "loss": 1.8125,
      "step": 874
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 1.6458032888654865,
      "learning_rate": 0.00013125,
      "loss": 0.7812,
      "step": 875
    },
    {
      "epoch": 0.41223529411764703,
      "grad_norm": 1.1930959913864478,
      "learning_rate": 0.0001314,
      "loss": 0.5195,
      "step": 876
    },
    {
      "epoch": 0.4127058823529412,
      "grad_norm": 1.0826016118436108,
      "learning_rate": 0.00013155,
      "loss": 0.4902,
      "step": 877
    },
    {
      "epoch": 0.4131764705882353,
      "grad_norm": 2.709729145866106,
      "learning_rate": 0.00013169999999999998,
      "loss": 1.1719,
      "step": 878
    },
    {
      "epoch": 0.4136470588235294,
      "grad_norm": 1.4809647198168887,
      "learning_rate": 0.00013184999999999998,
      "loss": 0.957,
      "step": 879
    },
    {
      "epoch": 0.41411764705882353,
      "grad_norm": 1.57566725636465,
      "learning_rate": 0.00013199999999999998,
      "loss": 1.0234,
      "step": 880
    },
    {
      "epoch": 0.41458823529411765,
      "grad_norm": 0.8100087035269697,
      "learning_rate": 0.00013215,
      "loss": 0.5195,
      "step": 881
    },
    {
      "epoch": 0.41505882352941176,
      "grad_norm": 1.0611349780385937,
      "learning_rate": 0.0001323,
      "loss": 0.5508,
      "step": 882
    },
    {
      "epoch": 0.41552941176470587,
      "grad_norm": 2.5648240348289506,
      "learning_rate": 0.00013245,
      "loss": 1.3516,
      "step": 883
    },
    {
      "epoch": 0.416,
      "grad_norm": 1.5553873345619467,
      "learning_rate": 0.0001326,
      "loss": 0.8711,
      "step": 884
    },
    {
      "epoch": 0.4164705882352941,
      "grad_norm": 1.8920408506143584,
      "learning_rate": 0.00013275,
      "loss": 1.2266,
      "step": 885
    },
    {
      "epoch": 0.41694117647058826,
      "grad_norm": 0.2903796678102538,
      "learning_rate": 0.00013289999999999998,
      "loss": 0.0791,
      "step": 886
    },
    {
      "epoch": 0.41741176470588237,
      "grad_norm": 2.453782036033775,
      "learning_rate": 0.00013304999999999998,
      "loss": 1.4688,
      "step": 887
    },
    {
      "epoch": 0.4178823529411765,
      "grad_norm": 0.6357885331473118,
      "learning_rate": 0.00013319999999999999,
      "loss": 0.2754,
      "step": 888
    },
    {
      "epoch": 0.4183529411764706,
      "grad_norm": 1.7338689461605852,
      "learning_rate": 0.00013335,
      "loss": 0.8945,
      "step": 889
    },
    {
      "epoch": 0.4188235294117647,
      "grad_norm": 1.0453321322246938,
      "learning_rate": 0.0001335,
      "loss": 0.5859,
      "step": 890
    },
    {
      "epoch": 0.4192941176470588,
      "grad_norm": 1.0807857957437992,
      "learning_rate": 0.00013365,
      "loss": 0.7891,
      "step": 891
    },
    {
      "epoch": 0.41976470588235293,
      "grad_norm": 1.1285305521988642,
      "learning_rate": 0.0001338,
      "loss": 0.6484,
      "step": 892
    },
    {
      "epoch": 0.42023529411764704,
      "grad_norm": 1.3208415614848275,
      "learning_rate": 0.00013395,
      "loss": 0.8828,
      "step": 893
    },
    {
      "epoch": 0.42070588235294115,
      "grad_norm": 0.8515465826198987,
      "learning_rate": 0.00013409999999999998,
      "loss": 0.5977,
      "step": 894
    },
    {
      "epoch": 0.4211764705882353,
      "grad_norm": 1.6260559028172805,
      "learning_rate": 0.00013424999999999998,
      "loss": 1.1953,
      "step": 895
    },
    {
      "epoch": 0.42164705882352943,
      "grad_norm": 1.1918509280087473,
      "learning_rate": 0.0001344,
      "loss": 0.6719,
      "step": 896
    },
    {
      "epoch": 0.42211764705882354,
      "grad_norm": 1.4203714833939243,
      "learning_rate": 0.00013455,
      "loss": 1.1562,
      "step": 897
    },
    {
      "epoch": 0.42258823529411765,
      "grad_norm": 1.1088569493340263,
      "learning_rate": 0.0001347,
      "loss": 0.5664,
      "step": 898
    },
    {
      "epoch": 0.42305882352941176,
      "grad_norm": 1.2896987729180573,
      "learning_rate": 0.00013485,
      "loss": 0.9453,
      "step": 899
    },
    {
      "epoch": 0.4235294117647059,
      "grad_norm": 1.041735770611394,
      "learning_rate": 0.000135,
      "loss": 0.7969,
      "step": 900
    },
    {
      "epoch": 0.424,
      "grad_norm": 0.9833934724925755,
      "learning_rate": 0.00013514999999999998,
      "loss": 0.625,
      "step": 901
    },
    {
      "epoch": 0.4244705882352941,
      "grad_norm": 73.50624390714178,
      "learning_rate": 0.00013529999999999998,
      "loss": 0.8516,
      "step": 902
    },
    {
      "epoch": 0.4249411764705882,
      "grad_norm": 1.162406043958616,
      "learning_rate": 0.00013544999999999999,
      "loss": 0.6523,
      "step": 903
    },
    {
      "epoch": 0.4254117647058824,
      "grad_norm": 2.006489467897351,
      "learning_rate": 0.0001356,
      "loss": 1.8438,
      "step": 904
    },
    {
      "epoch": 0.4258823529411765,
      "grad_norm": 1.1358527133394678,
      "learning_rate": 0.00013575,
      "loss": 0.5742,
      "step": 905
    },
    {
      "epoch": 0.4263529411764706,
      "grad_norm": 1.4389615871310715,
      "learning_rate": 0.0001359,
      "loss": 0.8984,
      "step": 906
    },
    {
      "epoch": 0.4268235294117647,
      "grad_norm": 10.26546054546065,
      "learning_rate": 0.00013605,
      "loss": 1.3984,
      "step": 907
    },
    {
      "epoch": 0.4272941176470588,
      "grad_norm": 2.653930014701672,
      "learning_rate": 0.0001362,
      "loss": 1.5547,
      "step": 908
    },
    {
      "epoch": 0.42776470588235294,
      "grad_norm": 0.8038945270035074,
      "learning_rate": 0.00013634999999999998,
      "loss": 0.2754,
      "step": 909
    },
    {
      "epoch": 0.42823529411764705,
      "grad_norm": 1.8505478372588873,
      "learning_rate": 0.00013649999999999998,
      "loss": 1.1094,
      "step": 910
    },
    {
      "epoch": 0.42870588235294116,
      "grad_norm": 2.981571102178724,
      "learning_rate": 0.00013665,
      "loss": 1.6016,
      "step": 911
    },
    {
      "epoch": 0.42917647058823527,
      "grad_norm": 2.6606063099099493,
      "learning_rate": 0.0001368,
      "loss": 1.3281,
      "step": 912
    },
    {
      "epoch": 0.42964705882352944,
      "grad_norm": 1.869835916411865,
      "learning_rate": 0.00013695,
      "loss": 0.5547,
      "step": 913
    },
    {
      "epoch": 0.43011764705882355,
      "grad_norm": 1.6007425771982116,
      "learning_rate": 0.0001371,
      "loss": 0.6641,
      "step": 914
    },
    {
      "epoch": 0.43058823529411766,
      "grad_norm": 1.3448326893378677,
      "learning_rate": 0.00013725,
      "loss": 0.5352,
      "step": 915
    },
    {
      "epoch": 0.43105882352941177,
      "grad_norm": 2.3007908374329413,
      "learning_rate": 0.0001374,
      "loss": 1.2031,
      "step": 916
    },
    {
      "epoch": 0.4315294117647059,
      "grad_norm": 2.1407425262833537,
      "learning_rate": 0.00013754999999999998,
      "loss": 1.1953,
      "step": 917
    },
    {
      "epoch": 0.432,
      "grad_norm": 1.1894672492205574,
      "learning_rate": 0.00013769999999999999,
      "loss": 0.6328,
      "step": 918
    },
    {
      "epoch": 0.4324705882352941,
      "grad_norm": 1.7582628229921256,
      "learning_rate": 0.00013785,
      "loss": 1.0938,
      "step": 919
    },
    {
      "epoch": 0.4329411764705882,
      "grad_norm": 1.4961483707227492,
      "learning_rate": 0.000138,
      "loss": 0.8281,
      "step": 920
    },
    {
      "epoch": 0.43341176470588233,
      "grad_norm": 2.208041927266448,
      "learning_rate": 0.00013815,
      "loss": 1.3281,
      "step": 921
    },
    {
      "epoch": 0.4338823529411765,
      "grad_norm": 1.1978187833914957,
      "learning_rate": 0.0001383,
      "loss": 0.9766,
      "step": 922
    },
    {
      "epoch": 0.4343529411764706,
      "grad_norm": 1.732359647738536,
      "learning_rate": 0.00013845,
      "loss": 1.125,
      "step": 923
    },
    {
      "epoch": 0.4348235294117647,
      "grad_norm": 0.9289790717360139,
      "learning_rate": 0.0001386,
      "loss": 0.3887,
      "step": 924
    },
    {
      "epoch": 0.43529411764705883,
      "grad_norm": 0.7115753112234828,
      "learning_rate": 0.00013874999999999998,
      "loss": 0.2559,
      "step": 925
    },
    {
      "epoch": 0.43576470588235294,
      "grad_norm": 1.5193102325962613,
      "learning_rate": 0.0001389,
      "loss": 1.1562,
      "step": 926
    },
    {
      "epoch": 0.43623529411764705,
      "grad_norm": 2.281221326340915,
      "learning_rate": 0.00013905,
      "loss": 2.0625,
      "step": 927
    },
    {
      "epoch": 0.43670588235294117,
      "grad_norm": 1.5503436754730344,
      "learning_rate": 0.0001392,
      "loss": 0.9766,
      "step": 928
    },
    {
      "epoch": 0.4371764705882353,
      "grad_norm": 1.512662893816738,
      "learning_rate": 0.00013935,
      "loss": 1.5938,
      "step": 929
    },
    {
      "epoch": 0.4376470588235294,
      "grad_norm": 0.747286023966291,
      "learning_rate": 0.0001395,
      "loss": 0.2988,
      "step": 930
    },
    {
      "epoch": 0.43811764705882356,
      "grad_norm": 2.4366262828962126,
      "learning_rate": 0.00013965,
      "loss": 1.6172,
      "step": 931
    },
    {
      "epoch": 0.43858823529411767,
      "grad_norm": 1.275003608002007,
      "learning_rate": 0.00013979999999999998,
      "loss": 0.6328,
      "step": 932
    },
    {
      "epoch": 0.4390588235294118,
      "grad_norm": 2.7030972923378007,
      "learning_rate": 0.00013995,
      "loss": 1.4453,
      "step": 933
    },
    {
      "epoch": 0.4395294117647059,
      "grad_norm": 1.2449181300781689,
      "learning_rate": 0.0001401,
      "loss": 0.6836,
      "step": 934
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7175886874856192,
      "learning_rate": 0.00014025,
      "loss": 0.957,
      "step": 935
    },
    {
      "epoch": 0.4404705882352941,
      "grad_norm": 0.6918782912507632,
      "learning_rate": 0.0001404,
      "loss": 0.2832,
      "step": 936
    },
    {
      "epoch": 0.4409411764705882,
      "grad_norm": 1.622129141399119,
      "learning_rate": 0.00014055,
      "loss": 1.0234,
      "step": 937
    },
    {
      "epoch": 0.44141176470588234,
      "grad_norm": 1.580510952342206,
      "learning_rate": 0.00014069999999999998,
      "loss": 0.9531,
      "step": 938
    },
    {
      "epoch": 0.44188235294117645,
      "grad_norm": 2.2955086110906655,
      "learning_rate": 0.00014084999999999998,
      "loss": 1.2969,
      "step": 939
    },
    {
      "epoch": 0.4423529411764706,
      "grad_norm": 0.6356702781024233,
      "learning_rate": 0.00014099999999999998,
      "loss": 0.3555,
      "step": 940
    },
    {
      "epoch": 0.4428235294117647,
      "grad_norm": 1.4277685540141571,
      "learning_rate": 0.00014115,
      "loss": 1.0859,
      "step": 941
    },
    {
      "epoch": 0.44329411764705884,
      "grad_norm": 2.2325490515313557,
      "learning_rate": 0.0001413,
      "loss": 1.5312,
      "step": 942
    },
    {
      "epoch": 0.44376470588235295,
      "grad_norm": 0.8788329243845371,
      "learning_rate": 0.00014144999999999997,
      "loss": 0.5547,
      "step": 943
    },
    {
      "epoch": 0.44423529411764706,
      "grad_norm": 1.4349651014106486,
      "learning_rate": 0.00014159999999999997,
      "loss": 0.6875,
      "step": 944
    },
    {
      "epoch": 0.4447058823529412,
      "grad_norm": 0.48242744369577767,
      "learning_rate": 0.00014174999999999998,
      "loss": 0.1543,
      "step": 945
    },
    {
      "epoch": 0.4451764705882353,
      "grad_norm": 0.7465819523742282,
      "learning_rate": 0.00014189999999999998,
      "loss": 0.334,
      "step": 946
    },
    {
      "epoch": 0.4456470588235294,
      "grad_norm": 2.5579729687598034,
      "learning_rate": 0.00014204999999999998,
      "loss": 1.1484,
      "step": 947
    },
    {
      "epoch": 0.4461176470588235,
      "grad_norm": 2.0631813244816732,
      "learning_rate": 0.0001422,
      "loss": 1.1641,
      "step": 948
    },
    {
      "epoch": 0.4465882352941177,
      "grad_norm": 1.4597763163139041,
      "learning_rate": 0.00014235,
      "loss": 1.2188,
      "step": 949
    },
    {
      "epoch": 0.4470588235294118,
      "grad_norm": 1.1600017087105188,
      "learning_rate": 0.0001425,
      "loss": 1.0547,
      "step": 950
    },
    {
      "epoch": 0.4475294117647059,
      "grad_norm": 1.4296316826508746,
      "learning_rate": 0.00014264999999999997,
      "loss": 1.25,
      "step": 951
    },
    {
      "epoch": 0.448,
      "grad_norm": 0.8708787526707037,
      "learning_rate": 0.00014279999999999997,
      "loss": 0.5117,
      "step": 952
    },
    {
      "epoch": 0.4484705882352941,
      "grad_norm": 1.12134311042452,
      "learning_rate": 0.00014294999999999998,
      "loss": 0.7266,
      "step": 953
    },
    {
      "epoch": 0.44894117647058823,
      "grad_norm": 0.8045197925207725,
      "learning_rate": 0.00014309999999999998,
      "loss": 0.3418,
      "step": 954
    },
    {
      "epoch": 0.44941176470588234,
      "grad_norm": 0.9857034273173599,
      "learning_rate": 0.00014324999999999999,
      "loss": 0.6055,
      "step": 955
    },
    {
      "epoch": 0.44988235294117646,
      "grad_norm": 0.603541899495135,
      "learning_rate": 0.0001434,
      "loss": 0.2812,
      "step": 956
    },
    {
      "epoch": 0.45035294117647057,
      "grad_norm": 1.3584917321590366,
      "learning_rate": 0.00014355,
      "loss": 0.7969,
      "step": 957
    },
    {
      "epoch": 0.4508235294117647,
      "grad_norm": 2.374915410200759,
      "learning_rate": 0.00014369999999999997,
      "loss": 0.8438,
      "step": 958
    },
    {
      "epoch": 0.45129411764705885,
      "grad_norm": 1.3826032431224626,
      "learning_rate": 0.00014384999999999997,
      "loss": 0.8906,
      "step": 959
    },
    {
      "epoch": 0.45176470588235296,
      "grad_norm": 0.9008507790243037,
      "learning_rate": 0.00014399999999999998,
      "loss": 0.582,
      "step": 960
    },
    {
      "epoch": 0.45223529411764707,
      "grad_norm": 1.2253728399329158,
      "learning_rate": 0.00014414999999999998,
      "loss": 0.9375,
      "step": 961
    },
    {
      "epoch": 0.4527058823529412,
      "grad_norm": 1.2976543569971168,
      "learning_rate": 0.00014429999999999998,
      "loss": 1.0547,
      "step": 962
    },
    {
      "epoch": 0.4531764705882353,
      "grad_norm": 1.23961830313092,
      "learning_rate": 0.00014445,
      "loss": 0.5781,
      "step": 963
    },
    {
      "epoch": 0.4536470588235294,
      "grad_norm": 1.4381219405883696,
      "learning_rate": 0.0001446,
      "loss": 1.1875,
      "step": 964
    },
    {
      "epoch": 0.4541176470588235,
      "grad_norm": 0.9694778368966258,
      "learning_rate": 0.00014475,
      "loss": 0.6172,
      "step": 965
    },
    {
      "epoch": 0.4545882352941176,
      "grad_norm": 1.0194429988230311,
      "learning_rate": 0.00014489999999999997,
      "loss": 0.3887,
      "step": 966
    },
    {
      "epoch": 0.45505882352941174,
      "grad_norm": 1.9900407889232414,
      "learning_rate": 0.00014504999999999997,
      "loss": 1.1406,
      "step": 967
    },
    {
      "epoch": 0.4555294117647059,
      "grad_norm": 0.7547299761343831,
      "learning_rate": 0.00014519999999999998,
      "loss": 0.2715,
      "step": 968
    },
    {
      "epoch": 0.456,
      "grad_norm": 1.2522294948023196,
      "learning_rate": 0.00014534999999999998,
      "loss": 0.8633,
      "step": 969
    },
    {
      "epoch": 0.45647058823529413,
      "grad_norm": 0.6827829308347648,
      "learning_rate": 0.00014549999999999999,
      "loss": 0.3086,
      "step": 970
    },
    {
      "epoch": 0.45694117647058824,
      "grad_norm": 1.110301241672015,
      "learning_rate": 0.00014565,
      "loss": 0.793,
      "step": 971
    },
    {
      "epoch": 0.45741176470588235,
      "grad_norm": 0.404282575072616,
      "learning_rate": 0.0001458,
      "loss": 0.1553,
      "step": 972
    },
    {
      "epoch": 0.45788235294117646,
      "grad_norm": 0.8348230803187805,
      "learning_rate": 0.00014595,
      "loss": 0.2617,
      "step": 973
    },
    {
      "epoch": 0.4583529411764706,
      "grad_norm": 2.4105900636911337,
      "learning_rate": 0.00014609999999999997,
      "loss": 1.2422,
      "step": 974
    },
    {
      "epoch": 0.4588235294117647,
      "grad_norm": 1.4380366161448488,
      "learning_rate": 0.00014624999999999998,
      "loss": 1.1094,
      "step": 975
    },
    {
      "epoch": 0.4592941176470588,
      "grad_norm": 0.82883052700294,
      "learning_rate": 0.00014639999999999998,
      "loss": 0.5781,
      "step": 976
    },
    {
      "epoch": 0.45976470588235296,
      "grad_norm": 0.4740817463968322,
      "learning_rate": 0.00014654999999999998,
      "loss": 0.1699,
      "step": 977
    },
    {
      "epoch": 0.4602352941176471,
      "grad_norm": 1.7320236083830354,
      "learning_rate": 0.0001467,
      "loss": 0.7734,
      "step": 978
    },
    {
      "epoch": 0.4607058823529412,
      "grad_norm": 0.6766322503038612,
      "learning_rate": 0.00014685,
      "loss": 0.334,
      "step": 979
    },
    {
      "epoch": 0.4611764705882353,
      "grad_norm": 1.1630848428086236,
      "learning_rate": 0.000147,
      "loss": 0.416,
      "step": 980
    },
    {
      "epoch": 0.4616470588235294,
      "grad_norm": 1.5340177337073058,
      "learning_rate": 0.00014714999999999997,
      "loss": 0.8984,
      "step": 981
    },
    {
      "epoch": 0.4621176470588235,
      "grad_norm": 2.9218429151244867,
      "learning_rate": 0.00014729999999999998,
      "loss": 1.7656,
      "step": 982
    },
    {
      "epoch": 0.46258823529411763,
      "grad_norm": 0.6910538842793117,
      "learning_rate": 0.00014744999999999998,
      "loss": 0.373,
      "step": 983
    },
    {
      "epoch": 0.46305882352941174,
      "grad_norm": 1.495065474445182,
      "learning_rate": 0.00014759999999999998,
      "loss": 1.0391,
      "step": 984
    },
    {
      "epoch": 0.46352941176470586,
      "grad_norm": 1.8325559478284112,
      "learning_rate": 0.00014774999999999999,
      "loss": 1.4453,
      "step": 985
    },
    {
      "epoch": 0.464,
      "grad_norm": 1.2347162916351442,
      "learning_rate": 0.0001479,
      "loss": 0.5664,
      "step": 986
    },
    {
      "epoch": 0.46447058823529414,
      "grad_norm": 1.7569208163789103,
      "learning_rate": 0.00014805,
      "loss": 1.2969,
      "step": 987
    },
    {
      "epoch": 0.46494117647058825,
      "grad_norm": 1.2692283757177132,
      "learning_rate": 0.0001482,
      "loss": 0.7109,
      "step": 988
    },
    {
      "epoch": 0.46541176470588236,
      "grad_norm": 0.48546665938419176,
      "learning_rate": 0.00014834999999999997,
      "loss": 0.1885,
      "step": 989
    },
    {
      "epoch": 0.46588235294117647,
      "grad_norm": 1.627316716078771,
      "learning_rate": 0.00014849999999999998,
      "loss": 0.8398,
      "step": 990
    },
    {
      "epoch": 0.4663529411764706,
      "grad_norm": 0.7350315813171754,
      "learning_rate": 0.00014864999999999998,
      "loss": 0.252,
      "step": 991
    },
    {
      "epoch": 0.4668235294117647,
      "grad_norm": 2.4871103571305193,
      "learning_rate": 0.00014879999999999998,
      "loss": 1.6562,
      "step": 992
    },
    {
      "epoch": 0.4672941176470588,
      "grad_norm": 0.6103906232528058,
      "learning_rate": 0.00014895,
      "loss": 0.2188,
      "step": 993
    },
    {
      "epoch": 0.4677647058823529,
      "grad_norm": 0.9831963954749142,
      "learning_rate": 0.0001491,
      "loss": 0.6172,
      "step": 994
    },
    {
      "epoch": 0.4682352941176471,
      "grad_norm": 1.7371086725435103,
      "learning_rate": 0.00014925,
      "loss": 1.2422,
      "step": 995
    },
    {
      "epoch": 0.4687058823529412,
      "grad_norm": 0.4428458074473174,
      "learning_rate": 0.0001494,
      "loss": 0.1206,
      "step": 996
    },
    {
      "epoch": 0.4691764705882353,
      "grad_norm": 1.5674444952738547,
      "learning_rate": 0.00014954999999999998,
      "loss": 1.1641,
      "step": 997
    },
    {
      "epoch": 0.4696470588235294,
      "grad_norm": 0.8810140157037369,
      "learning_rate": 0.00014969999999999998,
      "loss": 0.5352,
      "step": 998
    },
    {
      "epoch": 0.47011764705882353,
      "grad_norm": 1.039437981305701,
      "learning_rate": 0.00014984999999999998,
      "loss": 0.3301,
      "step": 999
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 1.6466200570949614,
      "learning_rate": 0.00015,
      "loss": 1.2891,
      "step": 1000
    },
    {
      "epoch": 0.47105882352941175,
      "grad_norm": 0.9004887118396746,
      "learning_rate": 0.00015014999999999996,
      "loss": 0.543,
      "step": 1001
    },
    {
      "epoch": 0.47152941176470586,
      "grad_norm": 0.9753467547865163,
      "learning_rate": 0.0001503,
      "loss": 0.5547,
      "step": 1002
    },
    {
      "epoch": 0.472,
      "grad_norm": 1.6884449658766847,
      "learning_rate": 0.00015044999999999997,
      "loss": 0.9414,
      "step": 1003
    },
    {
      "epoch": 0.47247058823529414,
      "grad_norm": 0.8490252866760727,
      "learning_rate": 0.00015059999999999997,
      "loss": 0.6641,
      "step": 1004
    },
    {
      "epoch": 0.47294117647058825,
      "grad_norm": 2.0419738835668038,
      "learning_rate": 0.00015074999999999998,
      "loss": 1.6953,
      "step": 1005
    },
    {
      "epoch": 0.47341176470588237,
      "grad_norm": 0.7440834982632337,
      "learning_rate": 0.00015089999999999998,
      "loss": 0.6641,
      "step": 1006
    },
    {
      "epoch": 0.4738823529411765,
      "grad_norm": 1.8582831637863124,
      "learning_rate": 0.00015104999999999996,
      "loss": 1.2422,
      "step": 1007
    },
    {
      "epoch": 0.4743529411764706,
      "grad_norm": 1.2782882500114336,
      "learning_rate": 0.0001512,
      "loss": 0.7461,
      "step": 1008
    },
    {
      "epoch": 0.4748235294117647,
      "grad_norm": 0.9683745304524223,
      "learning_rate": 0.00015134999999999997,
      "loss": 0.6719,
      "step": 1009
    },
    {
      "epoch": 0.4752941176470588,
      "grad_norm": 1.3954393112795098,
      "learning_rate": 0.0001515,
      "loss": 0.8242,
      "step": 1010
    },
    {
      "epoch": 0.4757647058823529,
      "grad_norm": 1.1118533817893865,
      "learning_rate": 0.00015164999999999997,
      "loss": 0.6289,
      "step": 1011
    },
    {
      "epoch": 0.47623529411764703,
      "grad_norm": 0.9158747287882046,
      "learning_rate": 0.00015179999999999998,
      "loss": 0.5273,
      "step": 1012
    },
    {
      "epoch": 0.4767058823529412,
      "grad_norm": 1.218599720782386,
      "learning_rate": 0.00015194999999999998,
      "loss": 0.6797,
      "step": 1013
    },
    {
      "epoch": 0.4771764705882353,
      "grad_norm": 1.9142894020848618,
      "learning_rate": 0.00015209999999999998,
      "loss": 1.1641,
      "step": 1014
    },
    {
      "epoch": 0.4776470588235294,
      "grad_norm": 1.3141583348105084,
      "learning_rate": 0.00015224999999999996,
      "loss": 1.1641,
      "step": 1015
    },
    {
      "epoch": 0.47811764705882354,
      "grad_norm": 1.2741606930272071,
      "learning_rate": 0.0001524,
      "loss": 0.9141,
      "step": 1016
    },
    {
      "epoch": 0.47858823529411765,
      "grad_norm": 1.2937806576957915,
      "learning_rate": 0.00015254999999999997,
      "loss": 0.5938,
      "step": 1017
    },
    {
      "epoch": 0.47905882352941176,
      "grad_norm": 1.330621974223714,
      "learning_rate": 0.0001527,
      "loss": 0.8867,
      "step": 1018
    },
    {
      "epoch": 0.47952941176470587,
      "grad_norm": 0.877811788677112,
      "learning_rate": 0.00015284999999999997,
      "loss": 0.3926,
      "step": 1019
    },
    {
      "epoch": 0.48,
      "grad_norm": 2.263137472820019,
      "learning_rate": 0.00015299999999999998,
      "loss": 1.2812,
      "step": 1020
    },
    {
      "epoch": 0.4804705882352941,
      "grad_norm": 1.3503449387232747,
      "learning_rate": 0.00015314999999999998,
      "loss": 0.9102,
      "step": 1021
    },
    {
      "epoch": 0.48094117647058826,
      "grad_norm": 1.412636639139652,
      "learning_rate": 0.00015329999999999999,
      "loss": 0.9336,
      "step": 1022
    },
    {
      "epoch": 0.4814117647058824,
      "grad_norm": 1.0188032247542889,
      "learning_rate": 0.00015344999999999996,
      "loss": 0.9766,
      "step": 1023
    },
    {
      "epoch": 0.4818823529411765,
      "grad_norm": 0.9246998763871551,
      "learning_rate": 0.0001536,
      "loss": 0.4004,
      "step": 1024
    },
    {
      "epoch": 0.4823529411764706,
      "grad_norm": 2.12339234205907,
      "learning_rate": 0.00015374999999999997,
      "loss": 1.4141,
      "step": 1025
    },
    {
      "epoch": 0.4828235294117647,
      "grad_norm": 1.493731779666277,
      "learning_rate": 0.0001539,
      "loss": 0.9883,
      "step": 1026
    },
    {
      "epoch": 0.4832941176470588,
      "grad_norm": 1.387252981718701,
      "learning_rate": 0.00015404999999999998,
      "loss": 1.1172,
      "step": 1027
    },
    {
      "epoch": 0.48376470588235293,
      "grad_norm": 0.9567684794118091,
      "learning_rate": 0.00015419999999999998,
      "loss": 0.668,
      "step": 1028
    },
    {
      "epoch": 0.48423529411764704,
      "grad_norm": 0.8322782350316934,
      "learning_rate": 0.00015434999999999998,
      "loss": 0.4727,
      "step": 1029
    },
    {
      "epoch": 0.48470588235294115,
      "grad_norm": 0.7578610953526673,
      "learning_rate": 0.0001545,
      "loss": 0.375,
      "step": 1030
    },
    {
      "epoch": 0.4851764705882353,
      "grad_norm": 0.529202199921313,
      "learning_rate": 0.00015464999999999996,
      "loss": 0.2188,
      "step": 1031
    },
    {
      "epoch": 0.48564705882352943,
      "grad_norm": 2.36773447117524,
      "learning_rate": 0.0001548,
      "loss": 0.9922,
      "step": 1032
    },
    {
      "epoch": 0.48611764705882354,
      "grad_norm": 1.0083441347035587,
      "learning_rate": 0.00015494999999999997,
      "loss": 0.4785,
      "step": 1033
    },
    {
      "epoch": 0.48658823529411765,
      "grad_norm": 1.3109052405930162,
      "learning_rate": 0.0001551,
      "loss": 0.8438,
      "step": 1034
    },
    {
      "epoch": 0.48705882352941177,
      "grad_norm": 1.3141193535614857,
      "learning_rate": 0.00015524999999999998,
      "loss": 0.4512,
      "step": 1035
    },
    {
      "epoch": 0.4875294117647059,
      "grad_norm": 1.461320886723835,
      "learning_rate": 0.00015539999999999998,
      "loss": 1.0156,
      "step": 1036
    },
    {
      "epoch": 0.488,
      "grad_norm": 0.604281525085278,
      "learning_rate": 0.00015554999999999999,
      "loss": 0.3223,
      "step": 1037
    },
    {
      "epoch": 0.4884705882352941,
      "grad_norm": 0.8803539440479061,
      "learning_rate": 0.0001557,
      "loss": 0.3789,
      "step": 1038
    },
    {
      "epoch": 0.4889411764705882,
      "grad_norm": 1.0309893581142788,
      "learning_rate": 0.00015584999999999997,
      "loss": 0.5547,
      "step": 1039
    },
    {
      "epoch": 0.4894117647058824,
      "grad_norm": 1.8562616994609795,
      "learning_rate": 0.000156,
      "loss": 0.9844,
      "step": 1040
    },
    {
      "epoch": 0.4898823529411765,
      "grad_norm": 1.2182347405058103,
      "learning_rate": 0.00015614999999999997,
      "loss": 0.5898,
      "step": 1041
    },
    {
      "epoch": 0.4903529411764706,
      "grad_norm": 0.9243667420586493,
      "learning_rate": 0.0001563,
      "loss": 0.5117,
      "step": 1042
    },
    {
      "epoch": 0.4908235294117647,
      "grad_norm": 1.8512149122924906,
      "learning_rate": 0.00015644999999999998,
      "loss": 1.0156,
      "step": 1043
    },
    {
      "epoch": 0.4912941176470588,
      "grad_norm": 1.708182504687622,
      "learning_rate": 0.00015659999999999998,
      "loss": 1.25,
      "step": 1044
    },
    {
      "epoch": 0.49176470588235294,
      "grad_norm": 0.9162307007066678,
      "learning_rate": 0.00015675,
      "loss": 0.6562,
      "step": 1045
    },
    {
      "epoch": 0.49223529411764705,
      "grad_norm": 1.7654572736635947,
      "learning_rate": 0.0001569,
      "loss": 1.3359,
      "step": 1046
    },
    {
      "epoch": 0.49270588235294116,
      "grad_norm": 1.1979813388667677,
      "learning_rate": 0.00015704999999999997,
      "loss": 0.6602,
      "step": 1047
    },
    {
      "epoch": 0.49317647058823527,
      "grad_norm": 1.2020579610310531,
      "learning_rate": 0.0001572,
      "loss": 0.8086,
      "step": 1048
    },
    {
      "epoch": 0.49364705882352944,
      "grad_norm": 1.7876445311909572,
      "learning_rate": 0.00015734999999999998,
      "loss": 1.4062,
      "step": 1049
    },
    {
      "epoch": 0.49411764705882355,
      "grad_norm": 1.222855954616941,
      "learning_rate": 0.00015749999999999998,
      "loss": 0.7148,
      "step": 1050
    },
    {
      "epoch": 0.49458823529411766,
      "grad_norm": 0.7473814603239506,
      "learning_rate": 0.00015764999999999998,
      "loss": 0.3926,
      "step": 1051
    },
    {
      "epoch": 0.4950588235294118,
      "grad_norm": 0.7376229166515713,
      "learning_rate": 0.0001578,
      "loss": 0.3789,
      "step": 1052
    },
    {
      "epoch": 0.4955294117647059,
      "grad_norm": 0.2516967680187441,
      "learning_rate": 0.00015794999999999996,
      "loss": 0.1396,
      "step": 1053
    },
    {
      "epoch": 0.496,
      "grad_norm": 2.1824005677621514,
      "learning_rate": 0.0001581,
      "loss": 1.2266,
      "step": 1054
    },
    {
      "epoch": 0.4964705882352941,
      "grad_norm": 1.8285989090255907,
      "learning_rate": 0.00015824999999999997,
      "loss": 1.0391,
      "step": 1055
    },
    {
      "epoch": 0.4969411764705882,
      "grad_norm": 0.27777906233711785,
      "learning_rate": 0.0001584,
      "loss": 0.063,
      "step": 1056
    },
    {
      "epoch": 0.49741176470588233,
      "grad_norm": 1.406040022270699,
      "learning_rate": 0.00015854999999999998,
      "loss": 0.9062,
      "step": 1057
    },
    {
      "epoch": 0.4978823529411765,
      "grad_norm": 1.4053304435826124,
      "learning_rate": 0.00015869999999999998,
      "loss": 0.957,
      "step": 1058
    },
    {
      "epoch": 0.4983529411764706,
      "grad_norm": 0.6620263522266773,
      "learning_rate": 0.00015884999999999999,
      "loss": 0.4258,
      "step": 1059
    },
    {
      "epoch": 0.4988235294117647,
      "grad_norm": 1.292547024623744,
      "learning_rate": 0.000159,
      "loss": 0.8789,
      "step": 1060
    },
    {
      "epoch": 0.49929411764705883,
      "grad_norm": 1.1664406533717913,
      "learning_rate": 0.00015914999999999997,
      "loss": 0.8945,
      "step": 1061
    },
    {
      "epoch": 0.49976470588235294,
      "grad_norm": 1.5414427590094273,
      "learning_rate": 0.0001593,
      "loss": 0.9414,
      "step": 1062
    },
    {
      "epoch": 0.5002352941176471,
      "grad_norm": 1.1083481384202392,
      "learning_rate": 0.00015944999999999997,
      "loss": 0.8203,
      "step": 1063
    },
    {
      "epoch": 0.5007058823529412,
      "grad_norm": 0.8215954047977583,
      "learning_rate": 0.0001596,
      "loss": 0.3691,
      "step": 1064
    },
    {
      "epoch": 0.5011764705882353,
      "grad_norm": 0.8289479330290973,
      "learning_rate": 0.00015974999999999998,
      "loss": 0.6719,
      "step": 1065
    },
    {
      "epoch": 0.5016470588235294,
      "grad_norm": 1.8267993563028226,
      "learning_rate": 0.00015989999999999998,
      "loss": 1.25,
      "step": 1066
    },
    {
      "epoch": 0.5021176470588236,
      "grad_norm": 1.4753533853844625,
      "learning_rate": 0.00016005,
      "loss": 1.1719,
      "step": 1067
    },
    {
      "epoch": 0.5025882352941177,
      "grad_norm": 1.335220916430335,
      "learning_rate": 0.0001602,
      "loss": 0.7539,
      "step": 1068
    },
    {
      "epoch": 0.5030588235294118,
      "grad_norm": 1.1306682421537804,
      "learning_rate": 0.00016034999999999997,
      "loss": 0.8086,
      "step": 1069
    },
    {
      "epoch": 0.5035294117647059,
      "grad_norm": 1.6613267978086899,
      "learning_rate": 0.0001605,
      "loss": 1.1172,
      "step": 1070
    },
    {
      "epoch": 0.504,
      "grad_norm": 0.6296625360771693,
      "learning_rate": 0.00016064999999999997,
      "loss": 0.2969,
      "step": 1071
    },
    {
      "epoch": 0.5044705882352941,
      "grad_norm": 1.0077891642607175,
      "learning_rate": 0.0001608,
      "loss": 0.8945,
      "step": 1072
    },
    {
      "epoch": 0.5049411764705882,
      "grad_norm": 1.8088372515084334,
      "learning_rate": 0.00016094999999999998,
      "loss": 0.9453,
      "step": 1073
    },
    {
      "epoch": 0.5054117647058823,
      "grad_norm": 1.6313832019154326,
      "learning_rate": 0.00016109999999999999,
      "loss": 0.9102,
      "step": 1074
    },
    {
      "epoch": 0.5058823529411764,
      "grad_norm": 0.8524056008046521,
      "learning_rate": 0.00016125,
      "loss": 0.5938,
      "step": 1075
    },
    {
      "epoch": 0.5063529411764706,
      "grad_norm": 1.3347987558715746,
      "learning_rate": 0.0001614,
      "loss": 1.2578,
      "step": 1076
    },
    {
      "epoch": 0.5068235294117647,
      "grad_norm": 0.8079324587028714,
      "learning_rate": 0.00016154999999999997,
      "loss": 0.5352,
      "step": 1077
    },
    {
      "epoch": 0.5072941176470588,
      "grad_norm": 1.475882446616363,
      "learning_rate": 0.0001617,
      "loss": 1.0859,
      "step": 1078
    },
    {
      "epoch": 0.5077647058823529,
      "grad_norm": 1.3560435081753137,
      "learning_rate": 0.00016184999999999998,
      "loss": 0.9297,
      "step": 1079
    },
    {
      "epoch": 0.508235294117647,
      "grad_norm": 1.2769614077957114,
      "learning_rate": 0.000162,
      "loss": 1.0781,
      "step": 1080
    },
    {
      "epoch": 0.5087058823529412,
      "grad_norm": 1.4515361673454326,
      "learning_rate": 0.00016214999999999998,
      "loss": 0.832,
      "step": 1081
    },
    {
      "epoch": 0.5091764705882353,
      "grad_norm": 0.456939936939439,
      "learning_rate": 0.0001623,
      "loss": 0.2432,
      "step": 1082
    },
    {
      "epoch": 0.5096470588235295,
      "grad_norm": 1.1578458966698668,
      "learning_rate": 0.00016245,
      "loss": 0.7305,
      "step": 1083
    },
    {
      "epoch": 0.5101176470588236,
      "grad_norm": 0.6625897392581014,
      "learning_rate": 0.0001626,
      "loss": 0.2832,
      "step": 1084
    },
    {
      "epoch": 0.5105882352941177,
      "grad_norm": 1.315669213724012,
      "learning_rate": 0.00016274999999999997,
      "loss": 0.7773,
      "step": 1085
    },
    {
      "epoch": 0.5110588235294118,
      "grad_norm": 1.443009813184858,
      "learning_rate": 0.0001629,
      "loss": 0.8203,
      "step": 1086
    },
    {
      "epoch": 0.5115294117647059,
      "grad_norm": 1.804014363556358,
      "learning_rate": 0.00016304999999999998,
      "loss": 1.1719,
      "step": 1087
    },
    {
      "epoch": 0.512,
      "grad_norm": 1.1317169764922765,
      "learning_rate": 0.0001632,
      "loss": 1.0469,
      "step": 1088
    },
    {
      "epoch": 0.5124705882352941,
      "grad_norm": 1.2111365942073617,
      "learning_rate": 0.00016334999999999999,
      "loss": 0.7852,
      "step": 1089
    },
    {
      "epoch": 0.5129411764705882,
      "grad_norm": 1.8040476581617308,
      "learning_rate": 0.0001635,
      "loss": 1.2266,
      "step": 1090
    },
    {
      "epoch": 0.5134117647058823,
      "grad_norm": 1.1757407142481149,
      "learning_rate": 0.00016365,
      "loss": 0.918,
      "step": 1091
    },
    {
      "epoch": 0.5138823529411765,
      "grad_norm": 0.9125690409835434,
      "learning_rate": 0.0001638,
      "loss": 0.5039,
      "step": 1092
    },
    {
      "epoch": 0.5143529411764706,
      "grad_norm": 0.8091048774040145,
      "learning_rate": 0.00016394999999999997,
      "loss": 0.416,
      "step": 1093
    },
    {
      "epoch": 0.5148235294117647,
      "grad_norm": 1.2905126289538913,
      "learning_rate": 0.0001641,
      "loss": 0.6562,
      "step": 1094
    },
    {
      "epoch": 0.5152941176470588,
      "grad_norm": 0.7197248436846455,
      "learning_rate": 0.00016424999999999998,
      "loss": 0.2773,
      "step": 1095
    },
    {
      "epoch": 0.5157647058823529,
      "grad_norm": 2.1727993117041247,
      "learning_rate": 0.0001644,
      "loss": 1.375,
      "step": 1096
    },
    {
      "epoch": 0.516235294117647,
      "grad_norm": 1.073749602735268,
      "learning_rate": 0.00016455,
      "loss": 0.7461,
      "step": 1097
    },
    {
      "epoch": 0.5167058823529411,
      "grad_norm": 0.8471514958998734,
      "learning_rate": 0.0001647,
      "loss": 0.6641,
      "step": 1098
    },
    {
      "epoch": 0.5171764705882353,
      "grad_norm": 1.3247178176418153,
      "learning_rate": 0.00016485,
      "loss": 0.5781,
      "step": 1099
    },
    {
      "epoch": 0.5176470588235295,
      "grad_norm": 1.3564252774855048,
      "learning_rate": 0.000165,
      "loss": 1.1641,
      "step": 1100
    },
    {
      "epoch": 0.5181176470588236,
      "grad_norm": 1.058648977685216,
      "learning_rate": 0.00016514999999999998,
      "loss": 0.8047,
      "step": 1101
    },
    {
      "epoch": 0.5185882352941177,
      "grad_norm": 0.4701587247063496,
      "learning_rate": 0.0001653,
      "loss": 0.3105,
      "step": 1102
    },
    {
      "epoch": 0.5190588235294118,
      "grad_norm": 1.227655728864705,
      "learning_rate": 0.00016544999999999998,
      "loss": 0.7188,
      "step": 1103
    },
    {
      "epoch": 0.5195294117647059,
      "grad_norm": 1.857706623130839,
      "learning_rate": 0.0001656,
      "loss": 1.2422,
      "step": 1104
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4468804915733127,
      "learning_rate": 0.00016575,
      "loss": 0.7891,
      "step": 1105
    },
    {
      "epoch": 0.5204705882352941,
      "grad_norm": 1.1122199197731708,
      "learning_rate": 0.0001659,
      "loss": 0.5312,
      "step": 1106
    },
    {
      "epoch": 0.5209411764705882,
      "grad_norm": 1.5246748242151182,
      "learning_rate": 0.00016604999999999997,
      "loss": 0.9141,
      "step": 1107
    },
    {
      "epoch": 0.5214117647058824,
      "grad_norm": 1.9466535535201808,
      "learning_rate": 0.0001662,
      "loss": 1.1484,
      "step": 1108
    },
    {
      "epoch": 0.5218823529411765,
      "grad_norm": 0.9366652605124595,
      "learning_rate": 0.00016634999999999998,
      "loss": 0.5742,
      "step": 1109
    },
    {
      "epoch": 0.5223529411764706,
      "grad_norm": 1.061792920287728,
      "learning_rate": 0.0001665,
      "loss": 0.8555,
      "step": 1110
    },
    {
      "epoch": 0.5228235294117647,
      "grad_norm": 1.2840399527194455,
      "learning_rate": 0.00016664999999999998,
      "loss": 0.9609,
      "step": 1111
    },
    {
      "epoch": 0.5232941176470588,
      "grad_norm": 1.1853703258338264,
      "learning_rate": 0.0001668,
      "loss": 0.8789,
      "step": 1112
    },
    {
      "epoch": 0.5237647058823529,
      "grad_norm": 1.887352251740274,
      "learning_rate": 0.00016695,
      "loss": 0.957,
      "step": 1113
    },
    {
      "epoch": 0.524235294117647,
      "grad_norm": 0.4110325394985463,
      "learning_rate": 0.0001671,
      "loss": 0.0981,
      "step": 1114
    },
    {
      "epoch": 0.5247058823529411,
      "grad_norm": 1.2279174502867107,
      "learning_rate": 0.00016724999999999997,
      "loss": 1.0234,
      "step": 1115
    },
    {
      "epoch": 0.5251764705882352,
      "grad_norm": 1.4740413709958409,
      "learning_rate": 0.0001674,
      "loss": 0.9922,
      "step": 1116
    },
    {
      "epoch": 0.5256470588235295,
      "grad_norm": 1.4390643923146589,
      "learning_rate": 0.00016754999999999998,
      "loss": 1.0703,
      "step": 1117
    },
    {
      "epoch": 0.5261176470588236,
      "grad_norm": 0.6287396520811959,
      "learning_rate": 0.0001677,
      "loss": 0.3789,
      "step": 1118
    },
    {
      "epoch": 0.5265882352941177,
      "grad_norm": 1.1614154495688518,
      "learning_rate": 0.00016785,
      "loss": 0.6133,
      "step": 1119
    },
    {
      "epoch": 0.5270588235294118,
      "grad_norm": 0.8732733844347376,
      "learning_rate": 0.000168,
      "loss": 0.4434,
      "step": 1120
    },
    {
      "epoch": 0.5275294117647059,
      "grad_norm": 2.5496371596387912,
      "learning_rate": 0.00016815,
      "loss": 1.6328,
      "step": 1121
    },
    {
      "epoch": 0.528,
      "grad_norm": 1.0211236760783986,
      "learning_rate": 0.0001683,
      "loss": 0.5703,
      "step": 1122
    },
    {
      "epoch": 0.5284705882352941,
      "grad_norm": 1.197771504934843,
      "learning_rate": 0.00016844999999999997,
      "loss": 0.6797,
      "step": 1123
    },
    {
      "epoch": 0.5289411764705882,
      "grad_norm": 0.6015301508151119,
      "learning_rate": 0.0001686,
      "loss": 0.2227,
      "step": 1124
    },
    {
      "epoch": 0.5294117647058824,
      "grad_norm": 1.6041005469547278,
      "learning_rate": 0.00016874999999999998,
      "loss": 1.1484,
      "step": 1125
    },
    {
      "epoch": 0.5298823529411765,
      "grad_norm": 0.8830403961317976,
      "learning_rate": 0.00016889999999999996,
      "loss": 0.4727,
      "step": 1126
    },
    {
      "epoch": 0.5303529411764706,
      "grad_norm": 1.9801810211157855,
      "learning_rate": 0.00016905,
      "loss": 0.9258,
      "step": 1127
    },
    {
      "epoch": 0.5308235294117647,
      "grad_norm": 2.634866001773711,
      "learning_rate": 0.00016919999999999997,
      "loss": 1.5156,
      "step": 1128
    },
    {
      "epoch": 0.5312941176470588,
      "grad_norm": 0.556295338733792,
      "learning_rate": 0.00016935,
      "loss": 0.1699,
      "step": 1129
    },
    {
      "epoch": 0.5317647058823529,
      "grad_norm": 1.5709502375470332,
      "learning_rate": 0.00016949999999999997,
      "loss": 0.9297,
      "step": 1130
    },
    {
      "epoch": 0.532235294117647,
      "grad_norm": 0.890376178419782,
      "learning_rate": 0.00016964999999999998,
      "loss": 0.6367,
      "step": 1131
    },
    {
      "epoch": 0.5327058823529411,
      "grad_norm": 1.680800082719743,
      "learning_rate": 0.00016979999999999998,
      "loss": 1.0781,
      "step": 1132
    },
    {
      "epoch": 0.5331764705882353,
      "grad_norm": 1.1418213043230723,
      "learning_rate": 0.00016994999999999998,
      "loss": 0.4785,
      "step": 1133
    },
    {
      "epoch": 0.5336470588235294,
      "grad_norm": 1.4865665795972447,
      "learning_rate": 0.00017009999999999996,
      "loss": 1.0156,
      "step": 1134
    },
    {
      "epoch": 0.5341176470588235,
      "grad_norm": 1.871059858419479,
      "learning_rate": 0.00017025,
      "loss": 1.1953,
      "step": 1135
    },
    {
      "epoch": 0.5345882352941177,
      "grad_norm": 1.9936047559209922,
      "learning_rate": 0.00017039999999999997,
      "loss": 0.9805,
      "step": 1136
    },
    {
      "epoch": 0.5350588235294118,
      "grad_norm": 1.4207779195684145,
      "learning_rate": 0.00017055,
      "loss": 0.8203,
      "step": 1137
    },
    {
      "epoch": 0.5355294117647059,
      "grad_norm": 0.8822147288113762,
      "learning_rate": 0.00017069999999999998,
      "loss": 0.4062,
      "step": 1138
    },
    {
      "epoch": 0.536,
      "grad_norm": 0.8386176339046963,
      "learning_rate": 0.00017084999999999998,
      "loss": 0.4004,
      "step": 1139
    },
    {
      "epoch": 0.5364705882352941,
      "grad_norm": 0.7870151453240151,
      "learning_rate": 0.00017099999999999998,
      "loss": 0.3164,
      "step": 1140
    },
    {
      "epoch": 0.5369411764705883,
      "grad_norm": 2.277846145246399,
      "learning_rate": 0.00017114999999999999,
      "loss": 1.4609,
      "step": 1141
    },
    {
      "epoch": 0.5374117647058824,
      "grad_norm": 0.4086143829780317,
      "learning_rate": 0.00017129999999999996,
      "loss": 0.1406,
      "step": 1142
    },
    {
      "epoch": 0.5378823529411765,
      "grad_norm": 1.7899696815751414,
      "learning_rate": 0.00017145,
      "loss": 0.7695,
      "step": 1143
    },
    {
      "epoch": 0.5383529411764706,
      "grad_norm": 2.2350111246783038,
      "learning_rate": 0.00017159999999999997,
      "loss": 1.3672,
      "step": 1144
    },
    {
      "epoch": 0.5388235294117647,
      "grad_norm": 1.427476292834741,
      "learning_rate": 0.00017175,
      "loss": 0.875,
      "step": 1145
    },
    {
      "epoch": 0.5392941176470588,
      "grad_norm": 0.8732041020629203,
      "learning_rate": 0.00017189999999999998,
      "loss": 0.6758,
      "step": 1146
    },
    {
      "epoch": 0.5397647058823529,
      "grad_norm": 1.4898154554267955,
      "learning_rate": 0.00017204999999999998,
      "loss": 0.6289,
      "step": 1147
    },
    {
      "epoch": 0.540235294117647,
      "grad_norm": 0.5385731629504852,
      "learning_rate": 0.00017219999999999998,
      "loss": 0.2139,
      "step": 1148
    },
    {
      "epoch": 0.5407058823529411,
      "grad_norm": 2.3544841236078353,
      "learning_rate": 0.00017235,
      "loss": 0.9141,
      "step": 1149
    },
    {
      "epoch": 0.5411764705882353,
      "grad_norm": 1.4983110252119127,
      "learning_rate": 0.00017249999999999996,
      "loss": 0.9219,
      "step": 1150
    },
    {
      "epoch": 0.5416470588235294,
      "grad_norm": 0.7236848914704032,
      "learning_rate": 0.00017265,
      "loss": 0.2949,
      "step": 1151
    },
    {
      "epoch": 0.5421176470588235,
      "grad_norm": 1.0127384283624805,
      "learning_rate": 0.00017279999999999997,
      "loss": 0.3672,
      "step": 1152
    },
    {
      "epoch": 0.5425882352941176,
      "grad_norm": 1.5915380813712445,
      "learning_rate": 0.00017294999999999998,
      "loss": 1.125,
      "step": 1153
    },
    {
      "epoch": 0.5430588235294118,
      "grad_norm": 1.2162519185869682,
      "learning_rate": 0.00017309999999999998,
      "loss": 0.9336,
      "step": 1154
    },
    {
      "epoch": 0.5435294117647059,
      "grad_norm": 0.7121766143023992,
      "learning_rate": 0.00017324999999999998,
      "loss": 0.3301,
      "step": 1155
    },
    {
      "epoch": 0.544,
      "grad_norm": 0.7567991984193552,
      "learning_rate": 0.00017339999999999996,
      "loss": 0.4414,
      "step": 1156
    },
    {
      "epoch": 0.5444705882352942,
      "grad_norm": 1.4628114574109572,
      "learning_rate": 0.00017355,
      "loss": 0.9297,
      "step": 1157
    },
    {
      "epoch": 0.5449411764705883,
      "grad_norm": 1.8402121387139088,
      "learning_rate": 0.00017369999999999997,
      "loss": 1.1719,
      "step": 1158
    },
    {
      "epoch": 0.5454117647058824,
      "grad_norm": 1.2167614853382531,
      "learning_rate": 0.00017385,
      "loss": 0.6914,
      "step": 1159
    },
    {
      "epoch": 0.5458823529411765,
      "grad_norm": 1.6225289402482834,
      "learning_rate": 0.00017399999999999997,
      "loss": 1.0391,
      "step": 1160
    },
    {
      "epoch": 0.5463529411764706,
      "grad_norm": 1.638953504098315,
      "learning_rate": 0.00017414999999999998,
      "loss": 1.1875,
      "step": 1161
    },
    {
      "epoch": 0.5468235294117647,
      "grad_norm": 0.7930920941635764,
      "learning_rate": 0.00017429999999999998,
      "loss": 0.4629,
      "step": 1162
    },
    {
      "epoch": 0.5472941176470588,
      "grad_norm": 0.8025690943227214,
      "learning_rate": 0.00017444999999999998,
      "loss": 0.5664,
      "step": 1163
    },
    {
      "epoch": 0.5477647058823529,
      "grad_norm": 10.026274164735305,
      "learning_rate": 0.00017459999999999996,
      "loss": 1.3203,
      "step": 1164
    },
    {
      "epoch": 0.548235294117647,
      "grad_norm": 0.7910071274072534,
      "learning_rate": 0.00017475,
      "loss": 0.498,
      "step": 1165
    },
    {
      "epoch": 0.5487058823529412,
      "grad_norm": 4.910079760776811,
      "learning_rate": 0.00017489999999999997,
      "loss": 1.5938,
      "step": 1166
    },
    {
      "epoch": 0.5491764705882353,
      "grad_norm": 12.765307372358125,
      "learning_rate": 0.00017505,
      "loss": 2.7344,
      "step": 1167
    },
    {
      "epoch": 0.5496470588235294,
      "grad_norm": 18.25027024194492,
      "learning_rate": 0.00017519999999999998,
      "loss": 1.4141,
      "step": 1168
    },
    {
      "epoch": 0.5501176470588235,
      "grad_norm": 1.6974246347216935,
      "learning_rate": 0.00017534999999999998,
      "loss": 0.8125,
      "step": 1169
    },
    {
      "epoch": 0.5505882352941176,
      "grad_norm": 1.8397354645100925,
      "learning_rate": 0.00017549999999999998,
      "loss": 1.1875,
      "step": 1170
    },
    {
      "epoch": 0.5510588235294117,
      "grad_norm": 0.9169491745342703,
      "learning_rate": 0.00017565,
      "loss": 0.4805,
      "step": 1171
    },
    {
      "epoch": 0.5515294117647059,
      "grad_norm": 1.2857642481797198,
      "learning_rate": 0.00017579999999999996,
      "loss": 0.6875,
      "step": 1172
    },
    {
      "epoch": 0.552,
      "grad_norm": 0.9938644709001511,
      "learning_rate": 0.00017595,
      "loss": 0.5625,
      "step": 1173
    },
    {
      "epoch": 0.5524705882352942,
      "grad_norm": 0.7091796605449777,
      "learning_rate": 0.00017609999999999997,
      "loss": 0.3105,
      "step": 1174
    },
    {
      "epoch": 0.5529411764705883,
      "grad_norm": 2.858247420043814,
      "learning_rate": 0.00017625,
      "loss": 1.5781,
      "step": 1175
    },
    {
      "epoch": 0.5534117647058824,
      "grad_norm": 1.000383763547897,
      "learning_rate": 0.00017639999999999998,
      "loss": 0.6602,
      "step": 1176
    },
    {
      "epoch": 0.5538823529411765,
      "grad_norm": 1.0725063608751357,
      "learning_rate": 0.00017654999999999998,
      "loss": 1.0078,
      "step": 1177
    },
    {
      "epoch": 0.5543529411764706,
      "grad_norm": 1.00525216736857,
      "learning_rate": 0.00017669999999999999,
      "loss": 0.7695,
      "step": 1178
    },
    {
      "epoch": 0.5548235294117647,
      "grad_norm": 0.5045419308136362,
      "learning_rate": 0.00017685,
      "loss": 0.1797,
      "step": 1179
    },
    {
      "epoch": 0.5552941176470588,
      "grad_norm": 1.7306911633409587,
      "learning_rate": 0.00017699999999999997,
      "loss": 1.1406,
      "step": 1180
    },
    {
      "epoch": 0.5557647058823529,
      "grad_norm": 1.4226976065416208,
      "learning_rate": 0.00017715,
      "loss": 0.9297,
      "step": 1181
    },
    {
      "epoch": 0.556235294117647,
      "grad_norm": 0.9198145934806441,
      "learning_rate": 0.00017729999999999997,
      "loss": 0.4316,
      "step": 1182
    },
    {
      "epoch": 0.5567058823529412,
      "grad_norm": 0.8511519439922713,
      "learning_rate": 0.00017745,
      "loss": 0.4004,
      "step": 1183
    },
    {
      "epoch": 0.5571764705882353,
      "grad_norm": 1.512096165671481,
      "learning_rate": 0.00017759999999999998,
      "loss": 0.9883,
      "step": 1184
    },
    {
      "epoch": 0.5576470588235294,
      "grad_norm": 0.7903114550983494,
      "learning_rate": 0.00017774999999999998,
      "loss": 0.3008,
      "step": 1185
    },
    {
      "epoch": 0.5581176470588235,
      "grad_norm": 1.1373890331874323,
      "learning_rate": 0.0001779,
      "loss": 0.7422,
      "step": 1186
    },
    {
      "epoch": 0.5585882352941176,
      "grad_norm": 1.9147992863065377,
      "learning_rate": 0.00017805,
      "loss": 1.3047,
      "step": 1187
    },
    {
      "epoch": 0.5590588235294117,
      "grad_norm": 1.0105282292429576,
      "learning_rate": 0.00017819999999999997,
      "loss": 0.8359,
      "step": 1188
    },
    {
      "epoch": 0.5595294117647058,
      "grad_norm": 0.5754408585236246,
      "learning_rate": 0.00017835,
      "loss": 0.2656,
      "step": 1189
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5347384017658915,
      "learning_rate": 0.00017849999999999997,
      "loss": 0.1738,
      "step": 1190
    },
    {
      "epoch": 0.5604705882352942,
      "grad_norm": 1.67366707201809,
      "learning_rate": 0.00017865,
      "loss": 1.1484,
      "step": 1191
    },
    {
      "epoch": 0.5609411764705883,
      "grad_norm": 1.5360500737267109,
      "learning_rate": 0.00017879999999999998,
      "loss": 1.0078,
      "step": 1192
    },
    {
      "epoch": 0.5614117647058824,
      "grad_norm": 2.146167629903488,
      "learning_rate": 0.00017894999999999999,
      "loss": 1.0859,
      "step": 1193
    },
    {
      "epoch": 0.5618823529411765,
      "grad_norm": 1.0412866784792898,
      "learning_rate": 0.0001791,
      "loss": 0.6133,
      "step": 1194
    },
    {
      "epoch": 0.5623529411764706,
      "grad_norm": 0.9635367918687597,
      "learning_rate": 0.00017925,
      "loss": 0.4434,
      "step": 1195
    },
    {
      "epoch": 0.5628235294117647,
      "grad_norm": 0.8303663982939894,
      "learning_rate": 0.00017939999999999997,
      "loss": 0.3379,
      "step": 1196
    },
    {
      "epoch": 0.5632941176470588,
      "grad_norm": 1.1380176278821994,
      "learning_rate": 0.00017955,
      "loss": 0.5977,
      "step": 1197
    },
    {
      "epoch": 0.563764705882353,
      "grad_norm": 1.5677490408621966,
      "learning_rate": 0.00017969999999999998,
      "loss": 0.8984,
      "step": 1198
    },
    {
      "epoch": 0.5642352941176471,
      "grad_norm": 0.3210660369285074,
      "learning_rate": 0.00017984999999999998,
      "loss": 0.051,
      "step": 1199
    },
    {
      "epoch": 0.5647058823529412,
      "grad_norm": 0.791865101956121,
      "learning_rate": 0.00017999999999999998,
      "loss": 0.459,
      "step": 1200
    },
    {
      "epoch": 0.5651764705882353,
      "grad_norm": 1.8045730801310897,
      "learning_rate": 0.00018015,
      "loss": 1.0234,
      "step": 1201
    },
    {
      "epoch": 0.5656470588235294,
      "grad_norm": 0.9642333557879157,
      "learning_rate": 0.00018029999999999996,
      "loss": 0.2715,
      "step": 1202
    },
    {
      "epoch": 0.5661176470588235,
      "grad_norm": 1.1627935240252978,
      "learning_rate": 0.00018045,
      "loss": 0.8203,
      "step": 1203
    },
    {
      "epoch": 0.5665882352941176,
      "grad_norm": 0.6922921735654562,
      "learning_rate": 0.00018059999999999997,
      "loss": 0.3965,
      "step": 1204
    },
    {
      "epoch": 0.5670588235294117,
      "grad_norm": 0.9697453503544513,
      "learning_rate": 0.00018075,
      "loss": 0.5664,
      "step": 1205
    },
    {
      "epoch": 0.5675294117647058,
      "grad_norm": 0.8019830715681444,
      "learning_rate": 0.00018089999999999998,
      "loss": 0.4355,
      "step": 1206
    },
    {
      "epoch": 0.568,
      "grad_norm": 0.8697818776165842,
      "learning_rate": 0.00018104999999999998,
      "loss": 0.4492,
      "step": 1207
    },
    {
      "epoch": 0.5684705882352942,
      "grad_norm": 0.520837230322498,
      "learning_rate": 0.00018119999999999999,
      "loss": 0.1631,
      "step": 1208
    },
    {
      "epoch": 0.5689411764705883,
      "grad_norm": 0.57657443805488,
      "learning_rate": 0.00018135,
      "loss": 0.1943,
      "step": 1209
    },
    {
      "epoch": 0.5694117647058824,
      "grad_norm": 1.1774606987765557,
      "learning_rate": 0.00018149999999999997,
      "loss": 0.5898,
      "step": 1210
    },
    {
      "epoch": 0.5698823529411765,
      "grad_norm": 0.9847505918143542,
      "learning_rate": 0.00018165,
      "loss": 0.5898,
      "step": 1211
    },
    {
      "epoch": 0.5703529411764706,
      "grad_norm": 1.6407223999952725,
      "learning_rate": 0.00018179999999999997,
      "loss": 0.8398,
      "step": 1212
    },
    {
      "epoch": 0.5708235294117647,
      "grad_norm": 0.5526795737269398,
      "learning_rate": 0.00018195,
      "loss": 0.3516,
      "step": 1213
    },
    {
      "epoch": 0.5712941176470588,
      "grad_norm": 0.6824209972023055,
      "learning_rate": 0.00018209999999999998,
      "loss": 0.3516,
      "step": 1214
    },
    {
      "epoch": 0.571764705882353,
      "grad_norm": 0.5451048570029042,
      "learning_rate": 0.00018224999999999998,
      "loss": 0.2598,
      "step": 1215
    },
    {
      "epoch": 0.5722352941176471,
      "grad_norm": 2.1064550740246677,
      "learning_rate": 0.0001824,
      "loss": 1.2734,
      "step": 1216
    },
    {
      "epoch": 0.5727058823529412,
      "grad_norm": 1.4599804727534271,
      "learning_rate": 0.00018255,
      "loss": 1.0078,
      "step": 1217
    },
    {
      "epoch": 0.5731764705882353,
      "grad_norm": 0.9930714232913739,
      "learning_rate": 0.00018269999999999997,
      "loss": 0.7109,
      "step": 1218
    },
    {
      "epoch": 0.5736470588235294,
      "grad_norm": 1.2568459254291517,
      "learning_rate": 0.00018285,
      "loss": 1.1484,
      "step": 1219
    },
    {
      "epoch": 0.5741176470588235,
      "grad_norm": 1.000359047221528,
      "learning_rate": 0.00018299999999999998,
      "loss": 0.7344,
      "step": 1220
    },
    {
      "epoch": 0.5745882352941176,
      "grad_norm": 1.324893196410496,
      "learning_rate": 0.00018315,
      "loss": 0.8711,
      "step": 1221
    },
    {
      "epoch": 0.5750588235294117,
      "grad_norm": 1.722788153446002,
      "learning_rate": 0.00018329999999999998,
      "loss": 0.9727,
      "step": 1222
    },
    {
      "epoch": 0.5755294117647058,
      "grad_norm": 0.6485615276780248,
      "learning_rate": 0.00018345,
      "loss": 0.2949,
      "step": 1223
    },
    {
      "epoch": 0.576,
      "grad_norm": 2.297549892285664,
      "learning_rate": 0.0001836,
      "loss": 1.7812,
      "step": 1224
    },
    {
      "epoch": 0.5764705882352941,
      "grad_norm": 0.7268937720223517,
      "learning_rate": 0.00018375,
      "loss": 0.3027,
      "step": 1225
    },
    {
      "epoch": 0.5769411764705883,
      "grad_norm": 1.4612950226937573,
      "learning_rate": 0.00018389999999999997,
      "loss": 0.7617,
      "step": 1226
    },
    {
      "epoch": 0.5774117647058824,
      "grad_norm": 0.8379732144935045,
      "learning_rate": 0.00018405,
      "loss": 0.4727,
      "step": 1227
    },
    {
      "epoch": 0.5778823529411765,
      "grad_norm": 298.6776847862006,
      "learning_rate": 0.00018419999999999998,
      "loss": 1.4453,
      "step": 1228
    },
    {
      "epoch": 0.5783529411764706,
      "grad_norm": 0.7164101381916744,
      "learning_rate": 0.00018435,
      "loss": 0.2773,
      "step": 1229
    },
    {
      "epoch": 0.5788235294117647,
      "grad_norm": 16.356908891239293,
      "learning_rate": 0.00018449999999999999,
      "loss": 2.1875,
      "step": 1230
    },
    {
      "epoch": 0.5792941176470588,
      "grad_norm": 1.6813050229311672,
      "learning_rate": 0.00018465,
      "loss": 1.0469,
      "step": 1231
    },
    {
      "epoch": 0.579764705882353,
      "grad_norm": 1.4844230020481302,
      "learning_rate": 0.0001848,
      "loss": 1.2109,
      "step": 1232
    },
    {
      "epoch": 0.5802352941176471,
      "grad_norm": 1.5840387537453249,
      "learning_rate": 0.00018495,
      "loss": 1.0391,
      "step": 1233
    },
    {
      "epoch": 0.5807058823529412,
      "grad_norm": 1.294432406115366,
      "learning_rate": 0.00018509999999999997,
      "loss": 0.793,
      "step": 1234
    },
    {
      "epoch": 0.5811764705882353,
      "grad_norm": 1.412649813848568,
      "learning_rate": 0.00018525,
      "loss": 1.2656,
      "step": 1235
    },
    {
      "epoch": 0.5816470588235294,
      "grad_norm": 1.7262739098613424,
      "learning_rate": 0.00018539999999999998,
      "loss": 1.4609,
      "step": 1236
    },
    {
      "epoch": 0.5821176470588235,
      "grad_norm": 1.1304196767846901,
      "learning_rate": 0.00018555,
      "loss": 0.3789,
      "step": 1237
    },
    {
      "epoch": 0.5825882352941176,
      "grad_norm": 2.901345544206578,
      "learning_rate": 0.0001857,
      "loss": 0.1807,
      "step": 1238
    },
    {
      "epoch": 0.5830588235294117,
      "grad_norm": 1.5306032643693674,
      "learning_rate": 0.00018585,
      "loss": 1.0078,
      "step": 1239
    },
    {
      "epoch": 0.5835294117647059,
      "grad_norm": 10.17887044437438,
      "learning_rate": 0.000186,
      "loss": 0.4004,
      "step": 1240
    },
    {
      "epoch": 0.584,
      "grad_norm": 3.89951406519595,
      "learning_rate": 0.00018615,
      "loss": 1.25,
      "step": 1241
    },
    {
      "epoch": 0.5844705882352941,
      "grad_norm": 0.9627998083988218,
      "learning_rate": 0.00018629999999999997,
      "loss": 0.5273,
      "step": 1242
    },
    {
      "epoch": 0.5849411764705882,
      "grad_norm": 0.5398691997567118,
      "learning_rate": 0.00018645,
      "loss": 0.1504,
      "step": 1243
    },
    {
      "epoch": 0.5854117647058823,
      "grad_norm": 1.131616485719048,
      "learning_rate": 0.00018659999999999998,
      "loss": 0.6836,
      "step": 1244
    },
    {
      "epoch": 0.5858823529411765,
      "grad_norm": 1.1407024829527177,
      "learning_rate": 0.00018675,
      "loss": 0.5469,
      "step": 1245
    },
    {
      "epoch": 0.5863529411764706,
      "grad_norm": 2.0407284256570897,
      "learning_rate": 0.0001869,
      "loss": 1.0938,
      "step": 1246
    },
    {
      "epoch": 0.5868235294117647,
      "grad_norm": 1.75322074547588,
      "learning_rate": 0.00018705,
      "loss": 1.4766,
      "step": 1247
    },
    {
      "epoch": 0.5872941176470589,
      "grad_norm": 0.8881259004800878,
      "learning_rate": 0.0001872,
      "loss": 0.3574,
      "step": 1248
    },
    {
      "epoch": 0.587764705882353,
      "grad_norm": 0.7162420448684806,
      "learning_rate": 0.00018735,
      "loss": 0.3477,
      "step": 1249
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 1.3500856152928014,
      "learning_rate": 0.00018749999999999998,
      "loss": 0.6016,
      "step": 1250
    },
    {
      "epoch": 0.5887058823529412,
      "grad_norm": 1.2374117550513308,
      "learning_rate": 0.00018764999999999998,
      "loss": 0.6172,
      "step": 1251
    },
    {
      "epoch": 0.5891764705882353,
      "grad_norm": 0.6370807904569663,
      "learning_rate": 0.00018779999999999998,
      "loss": 0.2754,
      "step": 1252
    },
    {
      "epoch": 0.5896470588235294,
      "grad_norm": 1.4528311014404536,
      "learning_rate": 0.00018794999999999996,
      "loss": 0.6328,
      "step": 1253
    },
    {
      "epoch": 0.5901176470588235,
      "grad_norm": 1.6516667929555184,
      "learning_rate": 0.0001881,
      "loss": 1.0391,
      "step": 1254
    },
    {
      "epoch": 0.5905882352941176,
      "grad_norm": 0.6085083647214331,
      "learning_rate": 0.00018824999999999997,
      "loss": 0.4707,
      "step": 1255
    },
    {
      "epoch": 0.5910588235294117,
      "grad_norm": 1.6677235224158895,
      "learning_rate": 0.00018839999999999997,
      "loss": 0.9961,
      "step": 1256
    },
    {
      "epoch": 0.5915294117647059,
      "grad_norm": 0.6415258348175239,
      "learning_rate": 0.00018854999999999998,
      "loss": 0.3145,
      "step": 1257
    },
    {
      "epoch": 0.592,
      "grad_norm": 1.2880371871691203,
      "learning_rate": 0.00018869999999999998,
      "loss": 0.8984,
      "step": 1258
    },
    {
      "epoch": 0.5924705882352941,
      "grad_norm": 1.104912386337902,
      "learning_rate": 0.00018884999999999996,
      "loss": 0.6055,
      "step": 1259
    },
    {
      "epoch": 0.5929411764705882,
      "grad_norm": 0.7857324291764884,
      "learning_rate": 0.00018899999999999999,
      "loss": 0.2578,
      "step": 1260
    },
    {
      "epoch": 0.5934117647058823,
      "grad_norm": 1.1097215361622663,
      "learning_rate": 0.00018914999999999996,
      "loss": 0.5508,
      "step": 1261
    },
    {
      "epoch": 0.5938823529411764,
      "grad_norm": 1.2860949484165327,
      "learning_rate": 0.0001893,
      "loss": 0.7305,
      "step": 1262
    },
    {
      "epoch": 0.5943529411764706,
      "grad_norm": 3.6186847588902245,
      "learning_rate": 0.00018944999999999997,
      "loss": 2.0781,
      "step": 1263
    },
    {
      "epoch": 0.5948235294117648,
      "grad_norm": 1.9191104579689233,
      "learning_rate": 0.00018959999999999997,
      "loss": 0.9922,
      "step": 1264
    },
    {
      "epoch": 0.5952941176470589,
      "grad_norm": 1.4865272954858515,
      "learning_rate": 0.00018974999999999998,
      "loss": 0.8398,
      "step": 1265
    },
    {
      "epoch": 0.595764705882353,
      "grad_norm": 1.5927181219368085,
      "learning_rate": 0.00018989999999999998,
      "loss": 1.3438,
      "step": 1266
    },
    {
      "epoch": 0.5962352941176471,
      "grad_norm": 1.4573697315824998,
      "learning_rate": 0.00019004999999999996,
      "loss": 0.7539,
      "step": 1267
    },
    {
      "epoch": 0.5967058823529412,
      "grad_norm": 0.6129113867230573,
      "learning_rate": 0.0001902,
      "loss": 0.3047,
      "step": 1268
    },
    {
      "epoch": 0.5971764705882353,
      "grad_norm": 0.6139399903448344,
      "learning_rate": 0.00019034999999999996,
      "loss": 0.2949,
      "step": 1269
    },
    {
      "epoch": 0.5976470588235294,
      "grad_norm": 1.735891697074178,
      "learning_rate": 0.0001905,
      "loss": 1.3594,
      "step": 1270
    },
    {
      "epoch": 0.5981176470588235,
      "grad_norm": 2.3928544141152077,
      "learning_rate": 0.00019064999999999997,
      "loss": 1.8047,
      "step": 1271
    },
    {
      "epoch": 0.5985882352941176,
      "grad_norm": 1.3957056514489012,
      "learning_rate": 0.00019079999999999998,
      "loss": 0.9766,
      "step": 1272
    },
    {
      "epoch": 0.5990588235294118,
      "grad_norm": 0.9769557216646707,
      "learning_rate": 0.00019094999999999998,
      "loss": 0.4902,
      "step": 1273
    },
    {
      "epoch": 0.5995294117647059,
      "grad_norm": 2.251264486731168,
      "learning_rate": 0.00019109999999999998,
      "loss": 1.4453,
      "step": 1274
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.26886381233417,
      "learning_rate": 0.00019124999999999996,
      "loss": 0.8516,
      "step": 1275
    },
    {
      "epoch": 0.6004705882352941,
      "grad_norm": 1.2506274936112824,
      "learning_rate": 0.0001914,
      "loss": 1.1406,
      "step": 1276
    },
    {
      "epoch": 0.6009411764705882,
      "grad_norm": 1.0928374193896104,
      "learning_rate": 0.00019154999999999997,
      "loss": 0.8125,
      "step": 1277
    },
    {
      "epoch": 0.6014117647058823,
      "grad_norm": 1.3842953991095475,
      "learning_rate": 0.0001917,
      "loss": 0.7969,
      "step": 1278
    },
    {
      "epoch": 0.6018823529411764,
      "grad_norm": 0.8400983082740633,
      "learning_rate": 0.00019184999999999997,
      "loss": 0.2832,
      "step": 1279
    },
    {
      "epoch": 0.6023529411764705,
      "grad_norm": 0.717572879348489,
      "learning_rate": 0.00019199999999999998,
      "loss": 0.332,
      "step": 1280
    },
    {
      "epoch": 0.6028235294117648,
      "grad_norm": 0.6727178673205456,
      "learning_rate": 0.00019214999999999998,
      "loss": 0.1895,
      "step": 1281
    },
    {
      "epoch": 0.6032941176470589,
      "grad_norm": 0.8349938488246564,
      "learning_rate": 0.00019229999999999999,
      "loss": 0.291,
      "step": 1282
    },
    {
      "epoch": 0.603764705882353,
      "grad_norm": 0.9086296285023809,
      "learning_rate": 0.00019244999999999996,
      "loss": 0.4883,
      "step": 1283
    },
    {
      "epoch": 0.6042352941176471,
      "grad_norm": 1.1207185635683128,
      "learning_rate": 0.0001926,
      "loss": 0.4766,
      "step": 1284
    },
    {
      "epoch": 0.6047058823529412,
      "grad_norm": 1.5133676127334388,
      "learning_rate": 0.00019274999999999997,
      "loss": 0.7539,
      "step": 1285
    },
    {
      "epoch": 0.6051764705882353,
      "grad_norm": 1.183404857022851,
      "learning_rate": 0.0001929,
      "loss": 0.6562,
      "step": 1286
    },
    {
      "epoch": 0.6056470588235294,
      "grad_norm": 1.1572732518951094,
      "learning_rate": 0.00019304999999999998,
      "loss": 0.3711,
      "step": 1287
    },
    {
      "epoch": 0.6061176470588235,
      "grad_norm": 1.5974241609996869,
      "learning_rate": 0.00019319999999999998,
      "loss": 0.8477,
      "step": 1288
    },
    {
      "epoch": 0.6065882352941177,
      "grad_norm": 1.0595238765535118,
      "learning_rate": 0.00019334999999999998,
      "loss": 0.6094,
      "step": 1289
    },
    {
      "epoch": 0.6070588235294118,
      "grad_norm": 0.3767458746157722,
      "learning_rate": 0.0001935,
      "loss": 0.1357,
      "step": 1290
    },
    {
      "epoch": 0.6075294117647059,
      "grad_norm": 0.6121421281710041,
      "learning_rate": 0.00019364999999999996,
      "loss": 0.2051,
      "step": 1291
    },
    {
      "epoch": 0.608,
      "grad_norm": 1.7420897008218732,
      "learning_rate": 0.0001938,
      "loss": 0.9297,
      "step": 1292
    },
    {
      "epoch": 0.6084705882352941,
      "grad_norm": 0.602769292947047,
      "learning_rate": 0.00019394999999999997,
      "loss": 0.2012,
      "step": 1293
    },
    {
      "epoch": 0.6089411764705882,
      "grad_norm": 2.513254072990886,
      "learning_rate": 0.0001941,
      "loss": 1.7344,
      "step": 1294
    },
    {
      "epoch": 0.6094117647058823,
      "grad_norm": 0.9788814113854332,
      "learning_rate": 0.00019424999999999998,
      "loss": 0.4941,
      "step": 1295
    },
    {
      "epoch": 0.6098823529411764,
      "grad_norm": 1.1701658955526788,
      "learning_rate": 0.00019439999999999998,
      "loss": 0.7891,
      "step": 1296
    },
    {
      "epoch": 0.6103529411764705,
      "grad_norm": 0.6292386263703216,
      "learning_rate": 0.00019454999999999999,
      "loss": 0.2598,
      "step": 1297
    },
    {
      "epoch": 0.6108235294117647,
      "grad_norm": 1.854125919885355,
      "learning_rate": 0.0001947,
      "loss": 1.0469,
      "step": 1298
    },
    {
      "epoch": 0.6112941176470589,
      "grad_norm": 1.15111407023886,
      "learning_rate": 0.00019484999999999997,
      "loss": 0.6445,
      "step": 1299
    },
    {
      "epoch": 0.611764705882353,
      "grad_norm": 1.8517008485075666,
      "learning_rate": 0.000195,
      "loss": 1.2266,
      "step": 1300
    },
    {
      "epoch": 0.6122352941176471,
      "grad_norm": 1.3752947068295516,
      "learning_rate": 0.00019514999999999997,
      "loss": 0.9102,
      "step": 1301
    },
    {
      "epoch": 0.6127058823529412,
      "grad_norm": 0.6163123994317461,
      "learning_rate": 0.00019529999999999998,
      "loss": 0.25,
      "step": 1302
    },
    {
      "epoch": 0.6131764705882353,
      "grad_norm": 1.1670266069480755,
      "learning_rate": 0.00019544999999999998,
      "loss": 0.7422,
      "step": 1303
    },
    {
      "epoch": 0.6136470588235294,
      "grad_norm": 1.435177602301921,
      "learning_rate": 0.00019559999999999998,
      "loss": 0.9102,
      "step": 1304
    },
    {
      "epoch": 0.6141176470588235,
      "grad_norm": 1.1391773820927853,
      "learning_rate": 0.00019574999999999996,
      "loss": 0.7383,
      "step": 1305
    },
    {
      "epoch": 0.6145882352941177,
      "grad_norm": 1.73455944284283,
      "learning_rate": 0.0001959,
      "loss": 1.1328,
      "step": 1306
    },
    {
      "epoch": 0.6150588235294118,
      "grad_norm": 1.035226412637068,
      "learning_rate": 0.00019604999999999997,
      "loss": 0.5977,
      "step": 1307
    },
    {
      "epoch": 0.6155294117647059,
      "grad_norm": 0.9250747779565208,
      "learning_rate": 0.0001962,
      "loss": 0.4824,
      "step": 1308
    },
    {
      "epoch": 0.616,
      "grad_norm": 0.5837617040454154,
      "learning_rate": 0.00019634999999999998,
      "loss": 0.1426,
      "step": 1309
    },
    {
      "epoch": 0.6164705882352941,
      "grad_norm": 2.4842664696487486,
      "learning_rate": 0.00019649999999999998,
      "loss": 1.4219,
      "step": 1310
    },
    {
      "epoch": 0.6169411764705882,
      "grad_norm": 0.1464528659477329,
      "learning_rate": 0.00019664999999999998,
      "loss": 0.0645,
      "step": 1311
    },
    {
      "epoch": 0.6174117647058823,
      "grad_norm": 1.79147963615439,
      "learning_rate": 0.00019679999999999999,
      "loss": 1.0469,
      "step": 1312
    },
    {
      "epoch": 0.6178823529411764,
      "grad_norm": 1.7409593231719447,
      "learning_rate": 0.00019694999999999996,
      "loss": 1.1406,
      "step": 1313
    },
    {
      "epoch": 0.6183529411764705,
      "grad_norm": 1.3292401387764206,
      "learning_rate": 0.0001971,
      "loss": 0.7266,
      "step": 1314
    },
    {
      "epoch": 0.6188235294117647,
      "grad_norm": 1.2120342066055196,
      "learning_rate": 0.00019724999999999997,
      "loss": 0.9219,
      "step": 1315
    },
    {
      "epoch": 0.6192941176470588,
      "grad_norm": 1.3481157799731145,
      "learning_rate": 0.0001974,
      "loss": 1.0391,
      "step": 1316
    },
    {
      "epoch": 0.619764705882353,
      "grad_norm": 1.209523234008638,
      "learning_rate": 0.00019754999999999998,
      "loss": 0.8125,
      "step": 1317
    },
    {
      "epoch": 0.6202352941176471,
      "grad_norm": 1.3333805312015201,
      "learning_rate": 0.00019769999999999998,
      "loss": 0.9258,
      "step": 1318
    },
    {
      "epoch": 0.6207058823529412,
      "grad_norm": 0.6345687439468422,
      "learning_rate": 0.00019784999999999998,
      "loss": 0.2441,
      "step": 1319
    },
    {
      "epoch": 0.6211764705882353,
      "grad_norm": 1.1760799979990053,
      "learning_rate": 0.000198,
      "loss": 0.5039,
      "step": 1320
    },
    {
      "epoch": 0.6216470588235294,
      "grad_norm": 0.6935479414742585,
      "learning_rate": 0.00019814999999999996,
      "loss": 0.1973,
      "step": 1321
    },
    {
      "epoch": 0.6221176470588236,
      "grad_norm": 1.4073576631794584,
      "learning_rate": 0.0001983,
      "loss": 0.9961,
      "step": 1322
    },
    {
      "epoch": 0.6225882352941177,
      "grad_norm": 0.5837026970316742,
      "learning_rate": 0.00019844999999999997,
      "loss": 0.3203,
      "step": 1323
    },
    {
      "epoch": 0.6230588235294118,
      "grad_norm": 0.9935607463663886,
      "learning_rate": 0.0001986,
      "loss": 0.6445,
      "step": 1324
    },
    {
      "epoch": 0.6235294117647059,
      "grad_norm": 0.5133896485487621,
      "learning_rate": 0.00019874999999999998,
      "loss": 0.1885,
      "step": 1325
    },
    {
      "epoch": 0.624,
      "grad_norm": 0.8676499037739085,
      "learning_rate": 0.00019889999999999998,
      "loss": 0.4297,
      "step": 1326
    },
    {
      "epoch": 0.6244705882352941,
      "grad_norm": 1.1992781095598257,
      "learning_rate": 0.00019905,
      "loss": 1.1484,
      "step": 1327
    },
    {
      "epoch": 0.6249411764705882,
      "grad_norm": 0.45765603958731094,
      "learning_rate": 0.0001992,
      "loss": 0.1826,
      "step": 1328
    },
    {
      "epoch": 0.6254117647058823,
      "grad_norm": 1.3711569207486707,
      "learning_rate": 0.00019934999999999997,
      "loss": 0.6133,
      "step": 1329
    },
    {
      "epoch": 0.6258823529411764,
      "grad_norm": 1.009803831431976,
      "learning_rate": 0.0001995,
      "loss": 0.9961,
      "step": 1330
    },
    {
      "epoch": 0.6263529411764706,
      "grad_norm": 1.212818623928074,
      "learning_rate": 0.00019964999999999997,
      "loss": 1.2031,
      "step": 1331
    },
    {
      "epoch": 0.6268235294117647,
      "grad_norm": 1.13612414089325,
      "learning_rate": 0.0001998,
      "loss": 0.5312,
      "step": 1332
    },
    {
      "epoch": 0.6272941176470588,
      "grad_norm": 0.8853950766268208,
      "learning_rate": 0.00019994999999999998,
      "loss": 0.9258,
      "step": 1333
    },
    {
      "epoch": 0.6277647058823529,
      "grad_norm": 1.2921609356089265,
      "learning_rate": 0.00020009999999999998,
      "loss": 0.6406,
      "step": 1334
    },
    {
      "epoch": 0.6282352941176471,
      "grad_norm": 0.6795183425092829,
      "learning_rate": 0.00020025,
      "loss": 0.3418,
      "step": 1335
    },
    {
      "epoch": 0.6287058823529412,
      "grad_norm": 0.8046294567680323,
      "learning_rate": 0.0002004,
      "loss": 0.4863,
      "step": 1336
    },
    {
      "epoch": 0.6291764705882353,
      "grad_norm": 1.162764072923724,
      "learning_rate": 0.00020054999999999997,
      "loss": 0.6133,
      "step": 1337
    },
    {
      "epoch": 0.6296470588235294,
      "grad_norm": 0.979762919532506,
      "learning_rate": 0.0002007,
      "loss": 0.5703,
      "step": 1338
    },
    {
      "epoch": 0.6301176470588236,
      "grad_norm": 1.8916621463592598,
      "learning_rate": 0.00020084999999999998,
      "loss": 1.7422,
      "step": 1339
    },
    {
      "epoch": 0.6305882352941177,
      "grad_norm": 1.230307412035277,
      "learning_rate": 0.000201,
      "loss": 0.8359,
      "step": 1340
    },
    {
      "epoch": 0.6310588235294118,
      "grad_norm": 1.332874164928239,
      "learning_rate": 0.00020114999999999998,
      "loss": 0.9258,
      "step": 1341
    },
    {
      "epoch": 0.6315294117647059,
      "grad_norm": 1.0247982208247426,
      "learning_rate": 0.0002013,
      "loss": 0.8789,
      "step": 1342
    },
    {
      "epoch": 0.632,
      "grad_norm": 1.041454781368992,
      "learning_rate": 0.00020145,
      "loss": 0.4961,
      "step": 1343
    },
    {
      "epoch": 0.6324705882352941,
      "grad_norm": 1.9636497607064172,
      "learning_rate": 0.0002016,
      "loss": 1.2656,
      "step": 1344
    },
    {
      "epoch": 0.6329411764705882,
      "grad_norm": 1.1919451951926465,
      "learning_rate": 0.00020174999999999997,
      "loss": 0.6523,
      "step": 1345
    },
    {
      "epoch": 0.6334117647058823,
      "grad_norm": 1.229094903091439,
      "learning_rate": 0.0002019,
      "loss": 0.8008,
      "step": 1346
    },
    {
      "epoch": 0.6338823529411765,
      "grad_norm": 1.089458132866111,
      "learning_rate": 0.00020204999999999998,
      "loss": 0.6758,
      "step": 1347
    },
    {
      "epoch": 0.6343529411764706,
      "grad_norm": 1.7753172229589687,
      "learning_rate": 0.0002022,
      "loss": 1.0078,
      "step": 1348
    },
    {
      "epoch": 0.6348235294117647,
      "grad_norm": 2.017466582737705,
      "learning_rate": 0.00020234999999999999,
      "loss": 1.4375,
      "step": 1349
    },
    {
      "epoch": 0.6352941176470588,
      "grad_norm": 1.8787147682314524,
      "learning_rate": 0.0002025,
      "loss": 1.375,
      "step": 1350
    },
    {
      "epoch": 0.6357647058823529,
      "grad_norm": 1.268839041716915,
      "learning_rate": 0.00020264999999999997,
      "loss": 0.8438,
      "step": 1351
    },
    {
      "epoch": 0.636235294117647,
      "grad_norm": 1.3504450495207254,
      "learning_rate": 0.0002028,
      "loss": 0.9531,
      "step": 1352
    },
    {
      "epoch": 0.6367058823529411,
      "grad_norm": 0.8105387857644983,
      "learning_rate": 0.00020294999999999997,
      "loss": 0.2324,
      "step": 1353
    },
    {
      "epoch": 0.6371764705882353,
      "grad_norm": 1.5018450163505719,
      "learning_rate": 0.0002031,
      "loss": 0.6133,
      "step": 1354
    },
    {
      "epoch": 0.6376470588235295,
      "grad_norm": 2.3560806714493454,
      "learning_rate": 0.00020324999999999998,
      "loss": 1.0625,
      "step": 1355
    },
    {
      "epoch": 0.6381176470588236,
      "grad_norm": 0.7620483325732562,
      "learning_rate": 0.00020339999999999998,
      "loss": 0.3789,
      "step": 1356
    },
    {
      "epoch": 0.6385882352941177,
      "grad_norm": 0.6168966588495473,
      "learning_rate": 0.00020355,
      "loss": 0.168,
      "step": 1357
    },
    {
      "epoch": 0.6390588235294118,
      "grad_norm": 0.9495091930347848,
      "learning_rate": 0.0002037,
      "loss": 0.5195,
      "step": 1358
    },
    {
      "epoch": 0.6395294117647059,
      "grad_norm": 1.5482146175607425,
      "learning_rate": 0.00020384999999999997,
      "loss": 0.9219,
      "step": 1359
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4359367545640267,
      "learning_rate": 0.000204,
      "loss": 0.4785,
      "step": 1360
    },
    {
      "epoch": 0.6404705882352941,
      "grad_norm": 0.8305370009520975,
      "learning_rate": 0.00020414999999999997,
      "loss": 0.3926,
      "step": 1361
    },
    {
      "epoch": 0.6409411764705882,
      "grad_norm": 1.1144511871454879,
      "learning_rate": 0.0002043,
      "loss": 0.7188,
      "step": 1362
    },
    {
      "epoch": 0.6414117647058823,
      "grad_norm": 2.415501537127222,
      "learning_rate": 0.00020444999999999998,
      "loss": 1.6406,
      "step": 1363
    },
    {
      "epoch": 0.6418823529411765,
      "grad_norm": 1.291178013109687,
      "learning_rate": 0.00020459999999999999,
      "loss": 1.1953,
      "step": 1364
    },
    {
      "epoch": 0.6423529411764706,
      "grad_norm": 1.2274866051578615,
      "learning_rate": 0.00020475,
      "loss": 0.7383,
      "step": 1365
    },
    {
      "epoch": 0.6428235294117647,
      "grad_norm": 1.7636623761165011,
      "learning_rate": 0.0002049,
      "loss": 1.3438,
      "step": 1366
    },
    {
      "epoch": 0.6432941176470588,
      "grad_norm": 1.3526698714284309,
      "learning_rate": 0.00020504999999999997,
      "loss": 0.8945,
      "step": 1367
    },
    {
      "epoch": 0.6437647058823529,
      "grad_norm": 0.7317039280836795,
      "learning_rate": 0.0002052,
      "loss": 0.3066,
      "step": 1368
    },
    {
      "epoch": 0.644235294117647,
      "grad_norm": 0.7982314447457662,
      "learning_rate": 0.00020534999999999998,
      "loss": 0.4453,
      "step": 1369
    },
    {
      "epoch": 0.6447058823529411,
      "grad_norm": 0.9178853301339972,
      "learning_rate": 0.0002055,
      "loss": 0.3848,
      "step": 1370
    },
    {
      "epoch": 0.6451764705882352,
      "grad_norm": 0.8330614091471168,
      "learning_rate": 0.00020564999999999998,
      "loss": 0.4258,
      "step": 1371
    },
    {
      "epoch": 0.6456470588235295,
      "grad_norm": 1.371212527208834,
      "learning_rate": 0.0002058,
      "loss": 0.7773,
      "step": 1372
    },
    {
      "epoch": 0.6461176470588236,
      "grad_norm": 0.6099586578950865,
      "learning_rate": 0.00020595,
      "loss": 0.1855,
      "step": 1373
    },
    {
      "epoch": 0.6465882352941177,
      "grad_norm": 1.1309077453987417,
      "learning_rate": 0.0002061,
      "loss": 0.5625,
      "step": 1374
    },
    {
      "epoch": 0.6470588235294118,
      "grad_norm": 1.6356803869932524,
      "learning_rate": 0.00020624999999999997,
      "loss": 0.9102,
      "step": 1375
    },
    {
      "epoch": 0.6475294117647059,
      "grad_norm": 1.0523188413595317,
      "learning_rate": 0.00020639999999999998,
      "loss": 0.7852,
      "step": 1376
    },
    {
      "epoch": 0.648,
      "grad_norm": 0.7160747158012357,
      "learning_rate": 0.00020654999999999998,
      "loss": 0.2832,
      "step": 1377
    },
    {
      "epoch": 0.6484705882352941,
      "grad_norm": 1.4226413943229066,
      "learning_rate": 0.00020669999999999996,
      "loss": 1.0,
      "step": 1378
    },
    {
      "epoch": 0.6489411764705882,
      "grad_norm": 1.6241692436752613,
      "learning_rate": 0.00020684999999999999,
      "loss": 1.0,
      "step": 1379
    },
    {
      "epoch": 0.6494117647058824,
      "grad_norm": 1.5168948389416375,
      "learning_rate": 0.00020699999999999996,
      "loss": 1.4922,
      "step": 1380
    },
    {
      "epoch": 0.6498823529411765,
      "grad_norm": 0.5390644785379506,
      "learning_rate": 0.00020715,
      "loss": 0.1758,
      "step": 1381
    },
    {
      "epoch": 0.6503529411764706,
      "grad_norm": 1.2286184464172734,
      "learning_rate": 0.00020729999999999997,
      "loss": 0.7617,
      "step": 1382
    },
    {
      "epoch": 0.6508235294117647,
      "grad_norm": 0.8931543265372629,
      "learning_rate": 0.00020744999999999997,
      "loss": 0.5469,
      "step": 1383
    },
    {
      "epoch": 0.6512941176470588,
      "grad_norm": 1.5593855317607894,
      "learning_rate": 0.00020759999999999998,
      "loss": 0.6289,
      "step": 1384
    },
    {
      "epoch": 0.6517647058823529,
      "grad_norm": 0.5984815721800818,
      "learning_rate": 0.00020774999999999998,
      "loss": 0.208,
      "step": 1385
    },
    {
      "epoch": 0.652235294117647,
      "grad_norm": 1.198908308460601,
      "learning_rate": 0.00020789999999999996,
      "loss": 0.4785,
      "step": 1386
    },
    {
      "epoch": 0.6527058823529411,
      "grad_norm": 0.9796054248428431,
      "learning_rate": 0.00020805,
      "loss": 0.459,
      "step": 1387
    },
    {
      "epoch": 0.6531764705882352,
      "grad_norm": 2.1611241199306455,
      "learning_rate": 0.00020819999999999996,
      "loss": 0.8711,
      "step": 1388
    },
    {
      "epoch": 0.6536470588235294,
      "grad_norm": 0.39960761295787717,
      "learning_rate": 0.00020835,
      "loss": 0.1699,
      "step": 1389
    },
    {
      "epoch": 0.6541176470588236,
      "grad_norm": 0.6014928703093253,
      "learning_rate": 0.00020849999999999997,
      "loss": 0.3164,
      "step": 1390
    },
    {
      "epoch": 0.6545882352941177,
      "grad_norm": 1.7722608183681634,
      "learning_rate": 0.00020864999999999998,
      "loss": 1.2578,
      "step": 1391
    },
    {
      "epoch": 0.6550588235294118,
      "grad_norm": 1.6318129240627175,
      "learning_rate": 0.00020879999999999998,
      "loss": 0.9102,
      "step": 1392
    },
    {
      "epoch": 0.6555294117647059,
      "grad_norm": 1.3069961313788405,
      "learning_rate": 0.00020894999999999998,
      "loss": 0.8281,
      "step": 1393
    },
    {
      "epoch": 0.656,
      "grad_norm": 1.1317019988751722,
      "learning_rate": 0.00020909999999999996,
      "loss": 0.5664,
      "step": 1394
    },
    {
      "epoch": 0.6564705882352941,
      "grad_norm": 0.44410109967622635,
      "learning_rate": 0.00020925,
      "loss": 0.1582,
      "step": 1395
    },
    {
      "epoch": 0.6569411764705883,
      "grad_norm": 0.7438014891124466,
      "learning_rate": 0.00020939999999999997,
      "loss": 0.3418,
      "step": 1396
    },
    {
      "epoch": 0.6574117647058824,
      "grad_norm": 1.253390299994102,
      "learning_rate": 0.00020955,
      "loss": 0.7227,
      "step": 1397
    },
    {
      "epoch": 0.6578823529411765,
      "grad_norm": 2.2143350907580817,
      "learning_rate": 0.00020969999999999997,
      "loss": 1.1406,
      "step": 1398
    },
    {
      "epoch": 0.6583529411764706,
      "grad_norm": 0.9878000611505937,
      "learning_rate": 0.00020984999999999998,
      "loss": 0.7734,
      "step": 1399
    },
    {
      "epoch": 0.6588235294117647,
      "grad_norm": 1.1588655832598052,
      "learning_rate": 0.00020999999999999998,
      "loss": 0.5469,
      "step": 1400
    },
    {
      "epoch": 0.6592941176470588,
      "grad_norm": 0.7428897420361568,
      "learning_rate": 0.00021014999999999999,
      "loss": 0.3281,
      "step": 1401
    },
    {
      "epoch": 0.6597647058823529,
      "grad_norm": 0.3500956591339803,
      "learning_rate": 0.00021029999999999996,
      "loss": 0.0894,
      "step": 1402
    },
    {
      "epoch": 0.660235294117647,
      "grad_norm": 1.795714062485441,
      "learning_rate": 0.00021045,
      "loss": 1.0234,
      "step": 1403
    },
    {
      "epoch": 0.6607058823529411,
      "grad_norm": 1.2431108327162474,
      "learning_rate": 0.00021059999999999997,
      "loss": 0.9258,
      "step": 1404
    },
    {
      "epoch": 0.6611764705882353,
      "grad_norm": 1.0753441632482819,
      "learning_rate": 0.00021074999999999997,
      "loss": 0.7344,
      "step": 1405
    },
    {
      "epoch": 0.6616470588235294,
      "grad_norm": 1.0861994107876571,
      "learning_rate": 0.00021089999999999998,
      "loss": 0.5508,
      "step": 1406
    },
    {
      "epoch": 0.6621176470588235,
      "grad_norm": 0.7684818895970308,
      "learning_rate": 0.00021104999999999998,
      "loss": 0.4434,
      "step": 1407
    },
    {
      "epoch": 0.6625882352941177,
      "grad_norm": 1.1488408007307913,
      "learning_rate": 0.00021119999999999996,
      "loss": 0.8125,
      "step": 1408
    },
    {
      "epoch": 0.6630588235294118,
      "grad_norm": 1.137810830659117,
      "learning_rate": 0.00021135,
      "loss": 0.7383,
      "step": 1409
    },
    {
      "epoch": 0.6635294117647059,
      "grad_norm": 1.0981043294177013,
      "learning_rate": 0.00021149999999999996,
      "loss": 0.8281,
      "step": 1410
    },
    {
      "epoch": 0.664,
      "grad_norm": 1.5789484064650592,
      "learning_rate": 0.00021165,
      "loss": 0.7461,
      "step": 1411
    },
    {
      "epoch": 0.6644705882352941,
      "grad_norm": 0.45146048991235876,
      "learning_rate": 0.00021179999999999997,
      "loss": 0.2314,
      "step": 1412
    },
    {
      "epoch": 0.6649411764705883,
      "grad_norm": 1.3759144499071674,
      "learning_rate": 0.00021194999999999997,
      "loss": 0.9414,
      "step": 1413
    },
    {
      "epoch": 0.6654117647058824,
      "grad_norm": 0.586973982765501,
      "learning_rate": 0.00021209999999999998,
      "loss": 0.1738,
      "step": 1414
    },
    {
      "epoch": 0.6658823529411765,
      "grad_norm": 0.6315522389739558,
      "learning_rate": 0.00021224999999999998,
      "loss": 0.2246,
      "step": 1415
    },
    {
      "epoch": 0.6663529411764706,
      "grad_norm": 0.7783956685248593,
      "learning_rate": 0.00021239999999999996,
      "loss": 0.3438,
      "step": 1416
    },
    {
      "epoch": 0.6668235294117647,
      "grad_norm": 1.1934423515574242,
      "learning_rate": 0.00021255,
      "loss": 0.5195,
      "step": 1417
    },
    {
      "epoch": 0.6672941176470588,
      "grad_norm": 1.1709019956481284,
      "learning_rate": 0.00021269999999999997,
      "loss": 0.8125,
      "step": 1418
    },
    {
      "epoch": 0.6677647058823529,
      "grad_norm": 0.8563228516955469,
      "learning_rate": 0.00021285,
      "loss": 0.4219,
      "step": 1419
    },
    {
      "epoch": 0.668235294117647,
      "grad_norm": 1.5174189675185974,
      "learning_rate": 0.00021299999999999997,
      "loss": 1.1016,
      "step": 1420
    },
    {
      "epoch": 0.6687058823529411,
      "grad_norm": 1.039206084916156,
      "learning_rate": 0.00021314999999999998,
      "loss": 1.0312,
      "step": 1421
    },
    {
      "epoch": 0.6691764705882353,
      "grad_norm": 0.8954637577543642,
      "learning_rate": 0.00021329999999999998,
      "loss": 0.6914,
      "step": 1422
    },
    {
      "epoch": 0.6696470588235294,
      "grad_norm": 1.0216076481489915,
      "learning_rate": 0.00021344999999999998,
      "loss": 0.8867,
      "step": 1423
    },
    {
      "epoch": 0.6701176470588235,
      "grad_norm": 0.9751494299134359,
      "learning_rate": 0.00021359999999999996,
      "loss": 0.8438,
      "step": 1424
    },
    {
      "epoch": 0.6705882352941176,
      "grad_norm": 1.5341142852217546,
      "learning_rate": 0.00021375,
      "loss": 1.25,
      "step": 1425
    },
    {
      "epoch": 0.6710588235294118,
      "grad_norm": 1.1954591749611607,
      "learning_rate": 0.00021389999999999997,
      "loss": 1.1016,
      "step": 1426
    },
    {
      "epoch": 0.6715294117647059,
      "grad_norm": 0.8931230756841549,
      "learning_rate": 0.00021405,
      "loss": 0.8398,
      "step": 1427
    },
    {
      "epoch": 0.672,
      "grad_norm": 0.8559801324100814,
      "learning_rate": 0.00021419999999999998,
      "loss": 0.7656,
      "step": 1428
    },
    {
      "epoch": 0.6724705882352942,
      "grad_norm": 1.593589373258132,
      "learning_rate": 0.00021434999999999998,
      "loss": 1.0391,
      "step": 1429
    },
    {
      "epoch": 0.6729411764705883,
      "grad_norm": 1.4383129280266675,
      "learning_rate": 0.00021449999999999998,
      "loss": 1.3828,
      "step": 1430
    },
    {
      "epoch": 0.6734117647058824,
      "grad_norm": 0.8373754562093485,
      "learning_rate": 0.00021464999999999999,
      "loss": 0.2793,
      "step": 1431
    },
    {
      "epoch": 0.6738823529411765,
      "grad_norm": 0.9659201321770517,
      "learning_rate": 0.00021479999999999996,
      "loss": 0.4648,
      "step": 1432
    },
    {
      "epoch": 0.6743529411764706,
      "grad_norm": 1.4727088811937832,
      "learning_rate": 0.00021495,
      "loss": 0.8984,
      "step": 1433
    },
    {
      "epoch": 0.6748235294117647,
      "grad_norm": 0.6618889564250904,
      "learning_rate": 0.00021509999999999997,
      "loss": 0.4766,
      "step": 1434
    },
    {
      "epoch": 0.6752941176470588,
      "grad_norm": 0.9570201725665136,
      "learning_rate": 0.00021525,
      "loss": 0.6016,
      "step": 1435
    },
    {
      "epoch": 0.6757647058823529,
      "grad_norm": 1.5107160715504357,
      "learning_rate": 0.00021539999999999998,
      "loss": 0.8906,
      "step": 1436
    },
    {
      "epoch": 0.676235294117647,
      "grad_norm": 1.0397505557088782,
      "learning_rate": 0.00021554999999999998,
      "loss": 0.7695,
      "step": 1437
    },
    {
      "epoch": 0.6767058823529412,
      "grad_norm": 0.7734482982531135,
      "learning_rate": 0.00021569999999999998,
      "loss": 0.3438,
      "step": 1438
    },
    {
      "epoch": 0.6771764705882353,
      "grad_norm": 1.811856263652381,
      "learning_rate": 0.00021585,
      "loss": 1.0781,
      "step": 1439
    },
    {
      "epoch": 0.6776470588235294,
      "grad_norm": 0.6421079828260646,
      "learning_rate": 0.00021599999999999996,
      "loss": 0.3965,
      "step": 1440
    },
    {
      "epoch": 0.6781176470588235,
      "grad_norm": 1.4426256058098976,
      "learning_rate": 0.00021615,
      "loss": 0.8398,
      "step": 1441
    },
    {
      "epoch": 0.6785882352941176,
      "grad_norm": 1.1140888382896892,
      "learning_rate": 0.00021629999999999997,
      "loss": 0.8125,
      "step": 1442
    },
    {
      "epoch": 0.6790588235294117,
      "grad_norm": 1.0755026338778906,
      "learning_rate": 0.00021645,
      "loss": 0.9531,
      "step": 1443
    },
    {
      "epoch": 0.6795294117647059,
      "grad_norm": 0.7187932893371463,
      "learning_rate": 0.00021659999999999998,
      "loss": 0.4688,
      "step": 1444
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6935055184911392,
      "learning_rate": 0.00021674999999999998,
      "loss": 0.2578,
      "step": 1445
    },
    {
      "epoch": 0.6804705882352942,
      "grad_norm": 1.6350115522301474,
      "learning_rate": 0.0002169,
      "loss": 1.3281,
      "step": 1446
    },
    {
      "epoch": 0.6809411764705883,
      "grad_norm": 1.2742851629353895,
      "learning_rate": 0.00021705,
      "loss": 1.2656,
      "step": 1447
    },
    {
      "epoch": 0.6814117647058824,
      "grad_norm": 0.8056864107065848,
      "learning_rate": 0.00021719999999999997,
      "loss": 0.3242,
      "step": 1448
    },
    {
      "epoch": 0.6818823529411765,
      "grad_norm": 1.67607115456727,
      "learning_rate": 0.00021735,
      "loss": 1.1016,
      "step": 1449
    },
    {
      "epoch": 0.6823529411764706,
      "grad_norm": 1.5973810407045257,
      "learning_rate": 0.00021749999999999997,
      "loss": 0.9844,
      "step": 1450
    },
    {
      "epoch": 0.6828235294117647,
      "grad_norm": 1.2698682328547948,
      "learning_rate": 0.00021764999999999998,
      "loss": 1.25,
      "step": 1451
    },
    {
      "epoch": 0.6832941176470588,
      "grad_norm": 1.1421757367735843,
      "learning_rate": 0.00021779999999999998,
      "loss": 1.0,
      "step": 1452
    },
    {
      "epoch": 0.6837647058823529,
      "grad_norm": 1.3461753442417932,
      "learning_rate": 0.00021794999999999999,
      "loss": 1.0703,
      "step": 1453
    },
    {
      "epoch": 0.684235294117647,
      "grad_norm": 0.608719864174986,
      "learning_rate": 0.00021809999999999996,
      "loss": 0.2969,
      "step": 1454
    },
    {
      "epoch": 0.6847058823529412,
      "grad_norm": 1.565890462591123,
      "learning_rate": 0.00021825,
      "loss": 1.0156,
      "step": 1455
    },
    {
      "epoch": 0.6851764705882353,
      "grad_norm": 0.7665977837134493,
      "learning_rate": 0.00021839999999999997,
      "loss": 0.2324,
      "step": 1456
    },
    {
      "epoch": 0.6856470588235294,
      "grad_norm": 1.689487638051788,
      "learning_rate": 0.00021855,
      "loss": 1.1719,
      "step": 1457
    },
    {
      "epoch": 0.6861176470588235,
      "grad_norm": 0.29921454277197845,
      "learning_rate": 0.00021869999999999998,
      "loss": 0.1104,
      "step": 1458
    },
    {
      "epoch": 0.6865882352941176,
      "grad_norm": 1.3700129526082756,
      "learning_rate": 0.00021884999999999998,
      "loss": 0.8672,
      "step": 1459
    },
    {
      "epoch": 0.6870588235294117,
      "grad_norm": 1.7465764394107992,
      "learning_rate": 0.00021899999999999998,
      "loss": 1.5312,
      "step": 1460
    },
    {
      "epoch": 0.6875294117647058,
      "grad_norm": 0.608785854996387,
      "learning_rate": 0.00021915,
      "loss": 0.2617,
      "step": 1461
    },
    {
      "epoch": 0.688,
      "grad_norm": 1.4715898165566739,
      "learning_rate": 0.00021929999999999996,
      "loss": 0.9297,
      "step": 1462
    },
    {
      "epoch": 0.6884705882352942,
      "grad_norm": 0.7785506617332789,
      "learning_rate": 0.00021945,
      "loss": 0.3926,
      "step": 1463
    },
    {
      "epoch": 0.6889411764705883,
      "grad_norm": 1.6819486878853196,
      "learning_rate": 0.00021959999999999997,
      "loss": 1.1953,
      "step": 1464
    },
    {
      "epoch": 0.6894117647058824,
      "grad_norm": 0.6230854657787543,
      "learning_rate": 0.00021975,
      "loss": 0.2637,
      "step": 1465
    },
    {
      "epoch": 0.6898823529411765,
      "grad_norm": 0.9557044730513905,
      "learning_rate": 0.00021989999999999998,
      "loss": 0.5469,
      "step": 1466
    },
    {
      "epoch": 0.6903529411764706,
      "grad_norm": 1.2200777909744294,
      "learning_rate": 0.00022004999999999998,
      "loss": 0.4453,
      "step": 1467
    },
    {
      "epoch": 0.6908235294117647,
      "grad_norm": 2.1135103881131325,
      "learning_rate": 0.00022019999999999999,
      "loss": 1.2969,
      "step": 1468
    },
    {
      "epoch": 0.6912941176470588,
      "grad_norm": 2.2341676805761406,
      "learning_rate": 0.00022035,
      "loss": 1.375,
      "step": 1469
    },
    {
      "epoch": 0.691764705882353,
      "grad_norm": 1.7959944297091397,
      "learning_rate": 0.00022049999999999997,
      "loss": 1.7891,
      "step": 1470
    },
    {
      "epoch": 0.6922352941176471,
      "grad_norm": 0.6577613909715155,
      "learning_rate": 0.00022065,
      "loss": 0.5469,
      "step": 1471
    },
    {
      "epoch": 0.6927058823529412,
      "grad_norm": 0.8354817226881451,
      "learning_rate": 0.00022079999999999997,
      "loss": 0.6172,
      "step": 1472
    },
    {
      "epoch": 0.6931764705882353,
      "grad_norm": 1.5874813380254265,
      "learning_rate": 0.00022095,
      "loss": 0.7812,
      "step": 1473
    },
    {
      "epoch": 0.6936470588235294,
      "grad_norm": 1.33209703584775,
      "learning_rate": 0.00022109999999999998,
      "loss": 0.8516,
      "step": 1474
    },
    {
      "epoch": 0.6941176470588235,
      "grad_norm": 1.5273079721251595,
      "learning_rate": 0.00022124999999999998,
      "loss": 0.8984,
      "step": 1475
    },
    {
      "epoch": 0.6945882352941176,
      "grad_norm": 1.2849300739554343,
      "learning_rate": 0.0002214,
      "loss": 0.9805,
      "step": 1476
    },
    {
      "epoch": 0.6950588235294117,
      "grad_norm": 0.4910795080024348,
      "learning_rate": 0.00022155,
      "loss": 0.1602,
      "step": 1477
    },
    {
      "epoch": 0.6955294117647058,
      "grad_norm": 1.0076647764933908,
      "learning_rate": 0.00022169999999999997,
      "loss": 0.7148,
      "step": 1478
    },
    {
      "epoch": 0.696,
      "grad_norm": 0.5174207120854862,
      "learning_rate": 0.00022185,
      "loss": 0.1592,
      "step": 1479
    },
    {
      "epoch": 0.6964705882352941,
      "grad_norm": 1.0234940228884009,
      "learning_rate": 0.00022199999999999998,
      "loss": 0.3555,
      "step": 1480
    },
    {
      "epoch": 0.6969411764705883,
      "grad_norm": 0.6942184415564192,
      "learning_rate": 0.00022215,
      "loss": 0.2578,
      "step": 1481
    },
    {
      "epoch": 0.6974117647058824,
      "grad_norm": 1.9326566209189513,
      "learning_rate": 0.00022229999999999998,
      "loss": 1.1562,
      "step": 1482
    },
    {
      "epoch": 0.6978823529411765,
      "grad_norm": 1.0961534093950012,
      "learning_rate": 0.00022244999999999999,
      "loss": 0.6172,
      "step": 1483
    },
    {
      "epoch": 0.6983529411764706,
      "grad_norm": 1.593146644161451,
      "learning_rate": 0.0002226,
      "loss": 0.8867,
      "step": 1484
    },
    {
      "epoch": 0.6988235294117647,
      "grad_norm": 1.2551547171867492,
      "learning_rate": 0.00022275,
      "loss": 0.9023,
      "step": 1485
    },
    {
      "epoch": 0.6992941176470588,
      "grad_norm": 0.6442608705313921,
      "learning_rate": 0.00022289999999999997,
      "loss": 0.249,
      "step": 1486
    },
    {
      "epoch": 0.699764705882353,
      "grad_norm": 0.910647913975328,
      "learning_rate": 0.00022305,
      "loss": 0.7422,
      "step": 1487
    },
    {
      "epoch": 0.7002352941176471,
      "grad_norm": 0.3806072023577959,
      "learning_rate": 0.00022319999999999998,
      "loss": 0.1279,
      "step": 1488
    },
    {
      "epoch": 0.7007058823529412,
      "grad_norm": 1.3879620416111997,
      "learning_rate": 0.00022335,
      "loss": 0.7695,
      "step": 1489
    },
    {
      "epoch": 0.7011764705882353,
      "grad_norm": 1.4342944107245845,
      "learning_rate": 0.00022349999999999998,
      "loss": 0.7852,
      "step": 1490
    },
    {
      "epoch": 0.7016470588235294,
      "grad_norm": 3.810257476556067,
      "learning_rate": 0.00022365,
      "loss": 0.2676,
      "step": 1491
    },
    {
      "epoch": 0.7021176470588235,
      "grad_norm": 0.36249320695578013,
      "learning_rate": 0.0002238,
      "loss": 0.0552,
      "step": 1492
    },
    {
      "epoch": 0.7025882352941176,
      "grad_norm": 1.1182622763146928,
      "learning_rate": 0.00022395,
      "loss": 0.5078,
      "step": 1493
    },
    {
      "epoch": 0.7030588235294117,
      "grad_norm": 1.6336763316600593,
      "learning_rate": 0.00022409999999999997,
      "loss": 1.0859,
      "step": 1494
    },
    {
      "epoch": 0.7035294117647058,
      "grad_norm": 1.3260618318213675,
      "learning_rate": 0.00022425,
      "loss": 0.7031,
      "step": 1495
    },
    {
      "epoch": 0.704,
      "grad_norm": 1.5456221489780415,
      "learning_rate": 0.00022439999999999998,
      "loss": 0.9727,
      "step": 1496
    },
    {
      "epoch": 0.7044705882352941,
      "grad_norm": 2.460370652259074,
      "learning_rate": 0.00022455,
      "loss": 1.1406,
      "step": 1497
    },
    {
      "epoch": 0.7049411764705882,
      "grad_norm": 2.2479690732057973,
      "learning_rate": 0.0002247,
      "loss": 1.6719,
      "step": 1498
    },
    {
      "epoch": 0.7054117647058824,
      "grad_norm": 3.285223862541638,
      "learning_rate": 0.00022485,
      "loss": 0.4609,
      "step": 1499
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 1.7236346457961902,
      "learning_rate": 0.000225,
      "loss": 1.0234,
      "step": 1500
    },
    {
      "epoch": 0.7063529411764706,
      "grad_norm": 19.945537398434716,
      "learning_rate": 0.00022514999999999997,
      "loss": 1.6875,
      "step": 1501
    },
    {
      "epoch": 0.7068235294117647,
      "grad_norm": 1.336127302144649,
      "learning_rate": 0.00022529999999999997,
      "loss": 0.6094,
      "step": 1502
    },
    {
      "epoch": 0.7072941176470589,
      "grad_norm": 1.5050030072754634,
      "learning_rate": 0.00022544999999999995,
      "loss": 0.7461,
      "step": 1503
    },
    {
      "epoch": 0.707764705882353,
      "grad_norm": 13.141912340678331,
      "learning_rate": 0.00022559999999999998,
      "loss": 2.3281,
      "step": 1504
    },
    {
      "epoch": 0.7082352941176471,
      "grad_norm": 6.109662587862222,
      "learning_rate": 0.00022574999999999996,
      "loss": 2.1562,
      "step": 1505
    },
    {
      "epoch": 0.7087058823529412,
      "grad_norm": 0.572884077779372,
      "learning_rate": 0.0002259,
      "loss": 0.2275,
      "step": 1506
    },
    {
      "epoch": 0.7091764705882353,
      "grad_norm": 1.169543392214809,
      "learning_rate": 0.00022604999999999997,
      "loss": 0.543,
      "step": 1507
    },
    {
      "epoch": 0.7096470588235294,
      "grad_norm": 1.3806464139165362,
      "learning_rate": 0.00022619999999999997,
      "loss": 0.4043,
      "step": 1508
    },
    {
      "epoch": 0.7101176470588235,
      "grad_norm": 2.604354318165391,
      "learning_rate": 0.00022634999999999997,
      "loss": 1.2031,
      "step": 1509
    },
    {
      "epoch": 0.7105882352941176,
      "grad_norm": 4.5858821968754535,
      "learning_rate": 0.00022649999999999998,
      "loss": 0.9688,
      "step": 1510
    },
    {
      "epoch": 0.7110588235294117,
      "grad_norm": 0.8533533646606941,
      "learning_rate": 0.00022664999999999995,
      "loss": 0.4062,
      "step": 1511
    },
    {
      "epoch": 0.7115294117647059,
      "grad_norm": 2.3102249363926886,
      "learning_rate": 0.00022679999999999998,
      "loss": 2.0312,
      "step": 1512
    },
    {
      "epoch": 0.712,
      "grad_norm": 1.6165646009832915,
      "learning_rate": 0.00022694999999999996,
      "loss": 1.625,
      "step": 1513
    },
    {
      "epoch": 0.7124705882352941,
      "grad_norm": 1.5261990633576499,
      "learning_rate": 0.0002271,
      "loss": 1.0547,
      "step": 1514
    },
    {
      "epoch": 0.7129411764705882,
      "grad_norm": 0.9335196663650129,
      "learning_rate": 0.00022724999999999997,
      "loss": 0.543,
      "step": 1515
    },
    {
      "epoch": 0.7134117647058823,
      "grad_norm": 93.05505682547556,
      "learning_rate": 0.00022739999999999997,
      "loss": 9.75,
      "step": 1516
    },
    {
      "epoch": 0.7138823529411765,
      "grad_norm": 6.316455980384331,
      "learning_rate": 0.00022754999999999997,
      "loss": 1.0938,
      "step": 1517
    },
    {
      "epoch": 0.7143529411764706,
      "grad_norm": 10.22822602571248,
      "learning_rate": 0.00022769999999999998,
      "loss": 0.7891,
      "step": 1518
    },
    {
      "epoch": 0.7148235294117647,
      "grad_norm": 16.473402346151495,
      "learning_rate": 0.00022784999999999995,
      "loss": 0.9414,
      "step": 1519
    },
    {
      "epoch": 0.7152941176470589,
      "grad_norm": 1.8121281647080472,
      "learning_rate": 0.00022799999999999999,
      "loss": 0.5859,
      "step": 1520
    },
    {
      "epoch": 0.715764705882353,
      "grad_norm": 0.9394053864877588,
      "learning_rate": 0.00022814999999999996,
      "loss": 0.375,
      "step": 1521
    },
    {
      "epoch": 0.7162352941176471,
      "grad_norm": 0.6382595363083942,
      "learning_rate": 0.0002283,
      "loss": 0.2295,
      "step": 1522
    },
    {
      "epoch": 0.7167058823529412,
      "grad_norm": 1.4430840099304567,
      "learning_rate": 0.00022844999999999997,
      "loss": 1.2188,
      "step": 1523
    },
    {
      "epoch": 0.7171764705882353,
      "grad_norm": 1.1657618366611648,
      "learning_rate": 0.00022859999999999997,
      "loss": 0.6953,
      "step": 1524
    },
    {
      "epoch": 0.7176470588235294,
      "grad_norm": 2.024851337416224,
      "learning_rate": 0.00022874999999999998,
      "loss": 1.0859,
      "step": 1525
    },
    {
      "epoch": 0.7181176470588235,
      "grad_norm": 1.0273915113681076,
      "learning_rate": 0.00022889999999999998,
      "loss": 0.5938,
      "step": 1526
    },
    {
      "epoch": 0.7185882352941176,
      "grad_norm": 2.935698883352516,
      "learning_rate": 0.00022904999999999996,
      "loss": 1.4219,
      "step": 1527
    },
    {
      "epoch": 0.7190588235294118,
      "grad_norm": 27.404578084539484,
      "learning_rate": 0.0002292,
      "loss": 1.8594,
      "step": 1528
    },
    {
      "epoch": 0.7195294117647059,
      "grad_norm": 1.1112449606625723,
      "learning_rate": 0.00022934999999999996,
      "loss": 0.6641,
      "step": 1529
    },
    {
      "epoch": 0.72,
      "grad_norm": 452.5002827124615,
      "learning_rate": 0.0002295,
      "loss": 4.7812,
      "step": 1530
    },
    {
      "epoch": 0.7204705882352941,
      "grad_norm": 8.783559809681494,
      "learning_rate": 0.00022964999999999997,
      "loss": 1.5391,
      "step": 1531
    },
    {
      "epoch": 0.7209411764705882,
      "grad_norm": 17.720069771143677,
      "learning_rate": 0.00022979999999999997,
      "loss": 2.375,
      "step": 1532
    },
    {
      "epoch": 0.7214117647058823,
      "grad_norm": 5.033120345293809,
      "learning_rate": 0.00022994999999999998,
      "loss": 1.1406,
      "step": 1533
    },
    {
      "epoch": 0.7218823529411764,
      "grad_norm": 7.804842260821223,
      "learning_rate": 0.00023009999999999998,
      "loss": 0.957,
      "step": 1534
    },
    {
      "epoch": 0.7223529411764706,
      "grad_norm": 1.227893016849942,
      "learning_rate": 0.00023024999999999996,
      "loss": 0.8164,
      "step": 1535
    },
    {
      "epoch": 0.7228235294117648,
      "grad_norm": 0.3581297831489376,
      "learning_rate": 0.0002304,
      "loss": 0.1079,
      "step": 1536
    },
    {
      "epoch": 0.7232941176470589,
      "grad_norm": 1.6010111123447708,
      "learning_rate": 0.00023054999999999997,
      "loss": 0.793,
      "step": 1537
    },
    {
      "epoch": 0.723764705882353,
      "grad_norm": 1.5092000058843666,
      "learning_rate": 0.0002307,
      "loss": 1.0312,
      "step": 1538
    },
    {
      "epoch": 0.7242352941176471,
      "grad_norm": 1.868585534053655,
      "learning_rate": 0.00023084999999999997,
      "loss": 1.5859,
      "step": 1539
    },
    {
      "epoch": 0.7247058823529412,
      "grad_norm": 2.001413749025945,
      "learning_rate": 0.00023099999999999998,
      "loss": 1.5781,
      "step": 1540
    },
    {
      "epoch": 0.7251764705882353,
      "grad_norm": 11.74311131866442,
      "learning_rate": 0.00023114999999999998,
      "loss": 0.6133,
      "step": 1541
    },
    {
      "epoch": 0.7256470588235294,
      "grad_norm": 2.130116632541914,
      "learning_rate": 0.00023129999999999998,
      "loss": 1.3203,
      "step": 1542
    },
    {
      "epoch": 0.7261176470588235,
      "grad_norm": 0.4047229753793201,
      "learning_rate": 0.00023144999999999996,
      "loss": 0.0967,
      "step": 1543
    },
    {
      "epoch": 0.7265882352941176,
      "grad_norm": 1.0748316903511383,
      "learning_rate": 0.0002316,
      "loss": 0.5234,
      "step": 1544
    },
    {
      "epoch": 0.7270588235294118,
      "grad_norm": 1.844923744139989,
      "learning_rate": 0.00023174999999999997,
      "loss": 1.0547,
      "step": 1545
    },
    {
      "epoch": 0.7275294117647059,
      "grad_norm": 1.2596599516610292,
      "learning_rate": 0.0002319,
      "loss": 0.6992,
      "step": 1546
    },
    {
      "epoch": 0.728,
      "grad_norm": 0.9447155441404004,
      "learning_rate": 0.00023204999999999998,
      "loss": 0.5664,
      "step": 1547
    },
    {
      "epoch": 0.7284705882352941,
      "grad_norm": 1.4852351122903986,
      "learning_rate": 0.00023219999999999998,
      "loss": 0.6484,
      "step": 1548
    },
    {
      "epoch": 0.7289411764705882,
      "grad_norm": 1.701699131248738,
      "learning_rate": 0.00023234999999999998,
      "loss": 1.0469,
      "step": 1549
    },
    {
      "epoch": 0.7294117647058823,
      "grad_norm": 0.7623463422270648,
      "learning_rate": 0.00023249999999999999,
      "loss": 0.3652,
      "step": 1550
    },
    {
      "epoch": 0.7298823529411764,
      "grad_norm": 1.5488204173613396,
      "learning_rate": 0.00023264999999999996,
      "loss": 1.0391,
      "step": 1551
    },
    {
      "epoch": 0.7303529411764705,
      "grad_norm": 0.688293947831101,
      "learning_rate": 0.0002328,
      "loss": 0.3691,
      "step": 1552
    },
    {
      "epoch": 0.7308235294117648,
      "grad_norm": 1.2502053672417708,
      "learning_rate": 0.00023294999999999997,
      "loss": 0.7773,
      "step": 1553
    },
    {
      "epoch": 0.7312941176470589,
      "grad_norm": 0.8882286145063663,
      "learning_rate": 0.00023309999999999997,
      "loss": 0.6719,
      "step": 1554
    },
    {
      "epoch": 0.731764705882353,
      "grad_norm": 1.0753092479130166,
      "learning_rate": 0.00023324999999999998,
      "loss": 0.7891,
      "step": 1555
    },
    {
      "epoch": 0.7322352941176471,
      "grad_norm": 1.5541978571629078,
      "learning_rate": 0.00023339999999999998,
      "loss": 1.1016,
      "step": 1556
    },
    {
      "epoch": 0.7327058823529412,
      "grad_norm": 1.0165277951589438,
      "learning_rate": 0.00023354999999999996,
      "loss": 0.6758,
      "step": 1557
    },
    {
      "epoch": 0.7331764705882353,
      "grad_norm": 0.6219860720061929,
      "learning_rate": 0.0002337,
      "loss": 0.3887,
      "step": 1558
    },
    {
      "epoch": 0.7336470588235294,
      "grad_norm": 1.0760101403509479,
      "learning_rate": 0.00023384999999999997,
      "loss": 0.6523,
      "step": 1559
    },
    {
      "epoch": 0.7341176470588235,
      "grad_norm": 0.7887523815770784,
      "learning_rate": 0.000234,
      "loss": 0.4277,
      "step": 1560
    },
    {
      "epoch": 0.7345882352941177,
      "grad_norm": 0.861503383315111,
      "learning_rate": 0.00023414999999999997,
      "loss": 0.3516,
      "step": 1561
    },
    {
      "epoch": 0.7350588235294118,
      "grad_norm": 1.0714970556313912,
      "learning_rate": 0.00023429999999999998,
      "loss": 0.9102,
      "step": 1562
    },
    {
      "epoch": 0.7355294117647059,
      "grad_norm": 1.3426915486776978,
      "learning_rate": 0.00023444999999999998,
      "loss": 1.0547,
      "step": 1563
    },
    {
      "epoch": 0.736,
      "grad_norm": 0.49694721452030927,
      "learning_rate": 0.00023459999999999998,
      "loss": 0.1387,
      "step": 1564
    },
    {
      "epoch": 0.7364705882352941,
      "grad_norm": 0.8804046031966105,
      "learning_rate": 0.00023474999999999996,
      "loss": 0.5391,
      "step": 1565
    },
    {
      "epoch": 0.7369411764705882,
      "grad_norm": 0.9858220351830503,
      "learning_rate": 0.0002349,
      "loss": 0.4727,
      "step": 1566
    },
    {
      "epoch": 0.7374117647058823,
      "grad_norm": 3.761542820619597,
      "learning_rate": 0.00023504999999999997,
      "loss": 1.7578,
      "step": 1567
    },
    {
      "epoch": 0.7378823529411764,
      "grad_norm": 2.677465131271867,
      "learning_rate": 0.0002352,
      "loss": 1.1406,
      "step": 1568
    },
    {
      "epoch": 0.7383529411764705,
      "grad_norm": 0.8348873578323757,
      "learning_rate": 0.00023534999999999997,
      "loss": 0.373,
      "step": 1569
    },
    {
      "epoch": 0.7388235294117647,
      "grad_norm": 2.594029676903348,
      "learning_rate": 0.00023549999999999998,
      "loss": 0.6914,
      "step": 1570
    },
    {
      "epoch": 0.7392941176470588,
      "grad_norm": 0.9483318085438224,
      "learning_rate": 0.00023564999999999998,
      "loss": 0.4531,
      "step": 1571
    },
    {
      "epoch": 0.739764705882353,
      "grad_norm": 2.7715473902747805,
      "learning_rate": 0.00023579999999999999,
      "loss": 2.1094,
      "step": 1572
    },
    {
      "epoch": 0.7402352941176471,
      "grad_norm": 1.1443007148175781,
      "learning_rate": 0.00023594999999999996,
      "loss": 0.7969,
      "step": 1573
    },
    {
      "epoch": 0.7407058823529412,
      "grad_norm": 1.2965788466425072,
      "learning_rate": 0.0002361,
      "loss": 0.7227,
      "step": 1574
    },
    {
      "epoch": 0.7411764705882353,
      "grad_norm": 1.7063630465028128,
      "learning_rate": 0.00023624999999999997,
      "loss": 1.0469,
      "step": 1575
    },
    {
      "epoch": 0.7416470588235294,
      "grad_norm": 0.8972202386290664,
      "learning_rate": 0.0002364,
      "loss": 0.668,
      "step": 1576
    },
    {
      "epoch": 0.7421176470588235,
      "grad_norm": 1.7710807489262805,
      "learning_rate": 0.00023654999999999998,
      "loss": 1.0,
      "step": 1577
    },
    {
      "epoch": 0.7425882352941177,
      "grad_norm": 1.0939409867607393,
      "learning_rate": 0.00023669999999999998,
      "loss": 0.5234,
      "step": 1578
    },
    {
      "epoch": 0.7430588235294118,
      "grad_norm": 1.473998443431307,
      "learning_rate": 0.00023684999999999998,
      "loss": 0.9297,
      "step": 1579
    },
    {
      "epoch": 0.7435294117647059,
      "grad_norm": 1.2955988592861902,
      "learning_rate": 0.000237,
      "loss": 0.8438,
      "step": 1580
    },
    {
      "epoch": 0.744,
      "grad_norm": 1.1066820044011532,
      "learning_rate": 0.00023714999999999996,
      "loss": 0.4727,
      "step": 1581
    },
    {
      "epoch": 0.7444705882352941,
      "grad_norm": 0.4775008963563021,
      "learning_rate": 0.0002373,
      "loss": 0.1289,
      "step": 1582
    },
    {
      "epoch": 0.7449411764705882,
      "grad_norm": 2.1798673698196267,
      "learning_rate": 0.00023744999999999997,
      "loss": 1.8203,
      "step": 1583
    },
    {
      "epoch": 0.7454117647058823,
      "grad_norm": 0.6630443583502378,
      "learning_rate": 0.0002376,
      "loss": 0.3164,
      "step": 1584
    },
    {
      "epoch": 0.7458823529411764,
      "grad_norm": 1.1780526408396992,
      "learning_rate": 0.00023774999999999998,
      "loss": 0.8672,
      "step": 1585
    },
    {
      "epoch": 0.7463529411764706,
      "grad_norm": 1.1520126306515808,
      "learning_rate": 0.00023789999999999998,
      "loss": 0.9453,
      "step": 1586
    },
    {
      "epoch": 0.7468235294117647,
      "grad_norm": 1.655855870060096,
      "learning_rate": 0.00023804999999999999,
      "loss": 1.1016,
      "step": 1587
    },
    {
      "epoch": 0.7472941176470588,
      "grad_norm": 0.8745348773835908,
      "learning_rate": 0.0002382,
      "loss": 0.6641,
      "step": 1588
    },
    {
      "epoch": 0.7477647058823529,
      "grad_norm": 1.2251978754141106,
      "learning_rate": 0.00023834999999999997,
      "loss": 0.6641,
      "step": 1589
    },
    {
      "epoch": 0.7482352941176471,
      "grad_norm": 1.0591960504145277,
      "learning_rate": 0.0002385,
      "loss": 0.4004,
      "step": 1590
    },
    {
      "epoch": 0.7487058823529412,
      "grad_norm": 1.484620691254011,
      "learning_rate": 0.00023864999999999997,
      "loss": 0.6875,
      "step": 1591
    },
    {
      "epoch": 0.7491764705882353,
      "grad_norm": 1.0558392654835502,
      "learning_rate": 0.0002388,
      "loss": 0.6328,
      "step": 1592
    },
    {
      "epoch": 0.7496470588235294,
      "grad_norm": 0.6205285415139906,
      "learning_rate": 0.00023894999999999998,
      "loss": 0.3223,
      "step": 1593
    },
    {
      "epoch": 0.7501176470588236,
      "grad_norm": 0.5923446643632677,
      "learning_rate": 0.00023909999999999998,
      "loss": 0.3379,
      "step": 1594
    },
    {
      "epoch": 0.7505882352941177,
      "grad_norm": 0.7318891526625996,
      "learning_rate": 0.00023925,
      "loss": 0.3555,
      "step": 1595
    },
    {
      "epoch": 0.7510588235294118,
      "grad_norm": 1.9387829119799445,
      "learning_rate": 0.0002394,
      "loss": 1.1562,
      "step": 1596
    },
    {
      "epoch": 0.7515294117647059,
      "grad_norm": 0.7716323299641085,
      "learning_rate": 0.00023954999999999997,
      "loss": 0.2812,
      "step": 1597
    },
    {
      "epoch": 0.752,
      "grad_norm": 1.1610797376778357,
      "learning_rate": 0.0002397,
      "loss": 0.6094,
      "step": 1598
    },
    {
      "epoch": 0.7524705882352941,
      "grad_norm": 1.083181964243726,
      "learning_rate": 0.00023984999999999998,
      "loss": 0.5117,
      "step": 1599
    },
    {
      "epoch": 0.7529411764705882,
      "grad_norm": 1.2137311928280465,
      "learning_rate": 0.00023999999999999998,
      "loss": 0.6953,
      "step": 1600
    },
    {
      "epoch": 0.7534117647058823,
      "grad_norm": 0.6773925426110222,
      "learning_rate": 0.00024014999999999998,
      "loss": 0.252,
      "step": 1601
    },
    {
      "epoch": 0.7538823529411764,
      "grad_norm": 1.5811277579768634,
      "learning_rate": 0.00024029999999999999,
      "loss": 0.8906,
      "step": 1602
    },
    {
      "epoch": 0.7543529411764706,
      "grad_norm": 0.8743853077990432,
      "learning_rate": 0.00024044999999999996,
      "loss": 0.5664,
      "step": 1603
    },
    {
      "epoch": 0.7548235294117647,
      "grad_norm": 0.3654028180514291,
      "learning_rate": 0.0002406,
      "loss": 0.0981,
      "step": 1604
    },
    {
      "epoch": 0.7552941176470588,
      "grad_norm": 1.1156426907670638,
      "learning_rate": 0.00024074999999999997,
      "loss": 0.7539,
      "step": 1605
    },
    {
      "epoch": 0.7557647058823529,
      "grad_norm": 0.9121313791773898,
      "learning_rate": 0.0002409,
      "loss": 0.5586,
      "step": 1606
    },
    {
      "epoch": 0.756235294117647,
      "grad_norm": 0.9415206305800796,
      "learning_rate": 0.00024104999999999998,
      "loss": 0.4746,
      "step": 1607
    },
    {
      "epoch": 0.7567058823529412,
      "grad_norm": 0.9257899402213038,
      "learning_rate": 0.00024119999999999998,
      "loss": 0.6719,
      "step": 1608
    },
    {
      "epoch": 0.7571764705882353,
      "grad_norm": 1.0370767164015153,
      "learning_rate": 0.00024134999999999998,
      "loss": 0.543,
      "step": 1609
    },
    {
      "epoch": 0.7576470588235295,
      "grad_norm": 0.8632714106993751,
      "learning_rate": 0.0002415,
      "loss": 0.3613,
      "step": 1610
    },
    {
      "epoch": 0.7581176470588236,
      "grad_norm": 1.1371025245305306,
      "learning_rate": 0.00024164999999999996,
      "loss": 0.6836,
      "step": 1611
    },
    {
      "epoch": 0.7585882352941177,
      "grad_norm": 1.1334876698671188,
      "learning_rate": 0.0002418,
      "loss": 0.8359,
      "step": 1612
    },
    {
      "epoch": 0.7590588235294118,
      "grad_norm": 1.2826739726820924,
      "learning_rate": 0.00024194999999999997,
      "loss": 0.7891,
      "step": 1613
    },
    {
      "epoch": 0.7595294117647059,
      "grad_norm": 0.8755018029138764,
      "learning_rate": 0.0002421,
      "loss": 0.291,
      "step": 1614
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2076516994666038,
      "learning_rate": 0.00024224999999999998,
      "loss": 0.5664,
      "step": 1615
    },
    {
      "epoch": 0.7604705882352941,
      "grad_norm": 1.3514529001568618,
      "learning_rate": 0.00024239999999999998,
      "loss": 0.6133,
      "step": 1616
    },
    {
      "epoch": 0.7609411764705882,
      "grad_norm": 0.7017106296259661,
      "learning_rate": 0.00024255,
      "loss": 0.3066,
      "step": 1617
    },
    {
      "epoch": 0.7614117647058823,
      "grad_norm": 1.67758344590153,
      "learning_rate": 0.0002427,
      "loss": 0.875,
      "step": 1618
    },
    {
      "epoch": 0.7618823529411765,
      "grad_norm": 1.739350447497331,
      "learning_rate": 0.00024284999999999997,
      "loss": 1.0156,
      "step": 1619
    },
    {
      "epoch": 0.7623529411764706,
      "grad_norm": 0.8998545808308535,
      "learning_rate": 0.000243,
      "loss": 0.4141,
      "step": 1620
    },
    {
      "epoch": 0.7628235294117647,
      "grad_norm": 1.585646756696069,
      "learning_rate": 0.00024314999999999997,
      "loss": 0.9844,
      "step": 1621
    },
    {
      "epoch": 0.7632941176470588,
      "grad_norm": 0.4960563648243108,
      "learning_rate": 0.0002433,
      "loss": 0.1777,
      "step": 1622
    },
    {
      "epoch": 0.7637647058823529,
      "grad_norm": 2.149868770009893,
      "learning_rate": 0.00024344999999999998,
      "loss": 2.0938,
      "step": 1623
    },
    {
      "epoch": 0.764235294117647,
      "grad_norm": 1.2797467488778236,
      "learning_rate": 0.00024359999999999999,
      "loss": 0.7734,
      "step": 1624
    },
    {
      "epoch": 0.7647058823529411,
      "grad_norm": 1.2550727894369482,
      "learning_rate": 0.00024375,
      "loss": 0.9609,
      "step": 1625
    },
    {
      "epoch": 0.7651764705882353,
      "grad_norm": 1.7775564939499608,
      "learning_rate": 0.00024389999999999997,
      "loss": 1.3906,
      "step": 1626
    },
    {
      "epoch": 0.7656470588235295,
      "grad_norm": 1.2166526163874785,
      "learning_rate": 0.00024404999999999997,
      "loss": 0.7383,
      "step": 1627
    },
    {
      "epoch": 0.7661176470588236,
      "grad_norm": 1.52507591090041,
      "learning_rate": 0.00024419999999999997,
      "loss": 1.3516,
      "step": 1628
    },
    {
      "epoch": 0.7665882352941177,
      "grad_norm": 1.2537203743574583,
      "learning_rate": 0.00024435,
      "loss": 0.3926,
      "step": 1629
    },
    {
      "epoch": 0.7670588235294118,
      "grad_norm": 1.2259603163757138,
      "learning_rate": 0.0002445,
      "loss": 0.6055,
      "step": 1630
    },
    {
      "epoch": 0.7675294117647059,
      "grad_norm": 1.051088712499284,
      "learning_rate": 0.00024464999999999996,
      "loss": 0.8203,
      "step": 1631
    },
    {
      "epoch": 0.768,
      "grad_norm": 1.140110421003342,
      "learning_rate": 0.0002448,
      "loss": 0.8281,
      "step": 1632
    },
    {
      "epoch": 0.7684705882352941,
      "grad_norm": 0.5644362713282332,
      "learning_rate": 0.00024494999999999996,
      "loss": 0.1992,
      "step": 1633
    },
    {
      "epoch": 0.7689411764705882,
      "grad_norm": 0.8863494904998397,
      "learning_rate": 0.00024509999999999994,
      "loss": 0.5625,
      "step": 1634
    },
    {
      "epoch": 0.7694117647058824,
      "grad_norm": 3.2937463235905686,
      "learning_rate": 0.00024524999999999997,
      "loss": 1.4609,
      "step": 1635
    },
    {
      "epoch": 0.7698823529411765,
      "grad_norm": 0.7955436721863322,
      "learning_rate": 0.00024539999999999995,
      "loss": 0.4102,
      "step": 1636
    },
    {
      "epoch": 0.7703529411764706,
      "grad_norm": 1.2402849145210888,
      "learning_rate": 0.00024555,
      "loss": 0.7539,
      "step": 1637
    },
    {
      "epoch": 0.7708235294117647,
      "grad_norm": 1.239121668576133,
      "learning_rate": 0.00024569999999999995,
      "loss": 0.9453,
      "step": 1638
    },
    {
      "epoch": 0.7712941176470588,
      "grad_norm": 1.055617721611829,
      "learning_rate": 0.00024585,
      "loss": 0.8711,
      "step": 1639
    },
    {
      "epoch": 0.7717647058823529,
      "grad_norm": 0.5030055951003735,
      "learning_rate": 0.00024599999999999996,
      "loss": 0.1592,
      "step": 1640
    },
    {
      "epoch": 0.772235294117647,
      "grad_norm": 1.299768349351133,
      "learning_rate": 0.00024615,
      "loss": 0.5352,
      "step": 1641
    },
    {
      "epoch": 0.7727058823529411,
      "grad_norm": 1.6758039067478003,
      "learning_rate": 0.00024629999999999997,
      "loss": 0.9609,
      "step": 1642
    },
    {
      "epoch": 0.7731764705882352,
      "grad_norm": 1.6432963290664884,
      "learning_rate": 0.00024645,
      "loss": 1.2656,
      "step": 1643
    },
    {
      "epoch": 0.7736470588235295,
      "grad_norm": 3.3603203658047747,
      "learning_rate": 0.0002466,
      "loss": 0.6133,
      "step": 1644
    },
    {
      "epoch": 0.7741176470588236,
      "grad_norm": 1.6622451207421172,
      "learning_rate": 0.00024675,
      "loss": 1.0,
      "step": 1645
    },
    {
      "epoch": 0.7745882352941177,
      "grad_norm": 3.4201340547732673,
      "learning_rate": 0.0002469,
      "loss": 1.3281,
      "step": 1646
    },
    {
      "epoch": 0.7750588235294118,
      "grad_norm": 1.029487434014366,
      "learning_rate": 0.00024704999999999996,
      "loss": 0.5977,
      "step": 1647
    },
    {
      "epoch": 0.7755294117647059,
      "grad_norm": 0.9019781865859627,
      "learning_rate": 0.0002472,
      "loss": 0.5117,
      "step": 1648
    },
    {
      "epoch": 0.776,
      "grad_norm": 1.1311803452847429,
      "learning_rate": 0.00024734999999999997,
      "loss": 1.0078,
      "step": 1649
    },
    {
      "epoch": 0.7764705882352941,
      "grad_norm": 2.2886533628475183,
      "learning_rate": 0.00024749999999999994,
      "loss": 1.5469,
      "step": 1650
    },
    {
      "epoch": 0.7769411764705882,
      "grad_norm": 1.3357924033247717,
      "learning_rate": 0.00024765,
      "loss": 1.0625,
      "step": 1651
    },
    {
      "epoch": 0.7774117647058824,
      "grad_norm": 1.8624549775978467,
      "learning_rate": 0.00024779999999999995,
      "loss": 1.1562,
      "step": 1652
    },
    {
      "epoch": 0.7778823529411765,
      "grad_norm": 1.5988700584153244,
      "learning_rate": 0.00024795,
      "loss": 0.9766,
      "step": 1653
    },
    {
      "epoch": 0.7783529411764706,
      "grad_norm": 1.6867064867891843,
      "learning_rate": 0.00024809999999999996,
      "loss": 1.2266,
      "step": 1654
    },
    {
      "epoch": 0.7788235294117647,
      "grad_norm": 1.4350575577552223,
      "learning_rate": 0.00024825,
      "loss": 0.8086,
      "step": 1655
    },
    {
      "epoch": 0.7792941176470588,
      "grad_norm": 1.5646555682440006,
      "learning_rate": 0.00024839999999999997,
      "loss": 0.8633,
      "step": 1656
    },
    {
      "epoch": 0.7797647058823529,
      "grad_norm": 1.7470924937978962,
      "learning_rate": 0.00024855,
      "loss": 0.9961,
      "step": 1657
    },
    {
      "epoch": 0.780235294117647,
      "grad_norm": 0.9038415047456198,
      "learning_rate": 0.0002487,
      "loss": 0.373,
      "step": 1658
    },
    {
      "epoch": 0.7807058823529411,
      "grad_norm": 1.3290602374331502,
      "learning_rate": 0.00024885,
      "loss": 1.0391,
      "step": 1659
    },
    {
      "epoch": 0.7811764705882352,
      "grad_norm": 1.3562342693742762,
      "learning_rate": 0.000249,
      "loss": 0.9375,
      "step": 1660
    },
    {
      "epoch": 0.7816470588235294,
      "grad_norm": 0.8031266676071844,
      "learning_rate": 0.00024914999999999996,
      "loss": 0.582,
      "step": 1661
    },
    {
      "epoch": 0.7821176470588236,
      "grad_norm": 1.0718736444415935,
      "learning_rate": 0.0002493,
      "loss": 0.7422,
      "step": 1662
    },
    {
      "epoch": 0.7825882352941177,
      "grad_norm": 1.1195163841258076,
      "learning_rate": 0.00024944999999999996,
      "loss": 0.7344,
      "step": 1663
    },
    {
      "epoch": 0.7830588235294118,
      "grad_norm": 0.9658541843319733,
      "learning_rate": 0.00024959999999999994,
      "loss": 0.4141,
      "step": 1664
    },
    {
      "epoch": 0.7835294117647059,
      "grad_norm": 1.225609369303925,
      "learning_rate": 0.00024974999999999997,
      "loss": 0.9258,
      "step": 1665
    },
    {
      "epoch": 0.784,
      "grad_norm": 0.9072968842111326,
      "learning_rate": 0.00024989999999999995,
      "loss": 0.6328,
      "step": 1666
    },
    {
      "epoch": 0.7844705882352941,
      "grad_norm": 1.1130568091295856,
      "learning_rate": 0.00025005,
      "loss": 1.125,
      "step": 1667
    },
    {
      "epoch": 0.7849411764705883,
      "grad_norm": 2.1297600472309757,
      "learning_rate": 0.00025019999999999996,
      "loss": 0.7266,
      "step": 1668
    },
    {
      "epoch": 0.7854117647058824,
      "grad_norm": 0.941545224577073,
      "learning_rate": 0.00025035,
      "loss": 0.6094,
      "step": 1669
    },
    {
      "epoch": 0.7858823529411765,
      "grad_norm": 1.8791460400288058,
      "learning_rate": 0.00025049999999999996,
      "loss": 1.4844,
      "step": 1670
    },
    {
      "epoch": 0.7863529411764706,
      "grad_norm": 1.6453505316554982,
      "learning_rate": 0.00025065,
      "loss": 0.7266,
      "step": 1671
    },
    {
      "epoch": 0.7868235294117647,
      "grad_norm": 1.6509259822076123,
      "learning_rate": 0.00025079999999999997,
      "loss": 0.9766,
      "step": 1672
    },
    {
      "epoch": 0.7872941176470588,
      "grad_norm": 1.1190178872029746,
      "learning_rate": 0.00025095,
      "loss": 0.7969,
      "step": 1673
    },
    {
      "epoch": 0.7877647058823529,
      "grad_norm": 0.26148783727165487,
      "learning_rate": 0.0002511,
      "loss": 0.063,
      "step": 1674
    },
    {
      "epoch": 0.788235294117647,
      "grad_norm": 0.6999842783126248,
      "learning_rate": 0.00025125,
      "loss": 0.5039,
      "step": 1675
    },
    {
      "epoch": 0.7887058823529411,
      "grad_norm": 1.2503337495414917,
      "learning_rate": 0.0002514,
      "loss": 0.9062,
      "step": 1676
    },
    {
      "epoch": 0.7891764705882353,
      "grad_norm": 1.3828884311309062,
      "learning_rate": 0.00025154999999999996,
      "loss": 1.0078,
      "step": 1677
    },
    {
      "epoch": 0.7896470588235294,
      "grad_norm": 1.2007783305983006,
      "learning_rate": 0.0002517,
      "loss": 0.9336,
      "step": 1678
    },
    {
      "epoch": 0.7901176470588235,
      "grad_norm": 0.6922575946926929,
      "learning_rate": 0.00025184999999999997,
      "loss": 0.3242,
      "step": 1679
    },
    {
      "epoch": 0.7905882352941176,
      "grad_norm": 1.4223263384112657,
      "learning_rate": 0.00025199999999999995,
      "loss": 0.9219,
      "step": 1680
    },
    {
      "epoch": 0.7910588235294118,
      "grad_norm": 1.4837694087291267,
      "learning_rate": 0.00025215,
      "loss": 0.9531,
      "step": 1681
    },
    {
      "epoch": 0.7915294117647059,
      "grad_norm": 1.5094149647964972,
      "learning_rate": 0.00025229999999999995,
      "loss": 1.1172,
      "step": 1682
    },
    {
      "epoch": 0.792,
      "grad_norm": 0.7776860027568693,
      "learning_rate": 0.00025245,
      "loss": 0.582,
      "step": 1683
    },
    {
      "epoch": 0.7924705882352941,
      "grad_norm": 1.5366528457357136,
      "learning_rate": 0.00025259999999999996,
      "loss": 0.7227,
      "step": 1684
    },
    {
      "epoch": 0.7929411764705883,
      "grad_norm": 1.2663974432602125,
      "learning_rate": 0.00025275,
      "loss": 0.8008,
      "step": 1685
    },
    {
      "epoch": 0.7934117647058824,
      "grad_norm": 1.7825831954938272,
      "learning_rate": 0.00025289999999999997,
      "loss": 1.2656,
      "step": 1686
    },
    {
      "epoch": 0.7938823529411765,
      "grad_norm": 0.9326025274051397,
      "learning_rate": 0.00025305,
      "loss": 0.7539,
      "step": 1687
    },
    {
      "epoch": 0.7943529411764706,
      "grad_norm": 0.6834060703314604,
      "learning_rate": 0.0002532,
      "loss": 0.3145,
      "step": 1688
    },
    {
      "epoch": 0.7948235294117647,
      "grad_norm": 1.7047720835305307,
      "learning_rate": 0.00025335,
      "loss": 1.4141,
      "step": 1689
    },
    {
      "epoch": 0.7952941176470588,
      "grad_norm": 1.6019369505120515,
      "learning_rate": 0.0002535,
      "loss": 0.9297,
      "step": 1690
    },
    {
      "epoch": 0.7957647058823529,
      "grad_norm": 0.8815439708375525,
      "learning_rate": 0.00025365,
      "loss": 0.5742,
      "step": 1691
    },
    {
      "epoch": 0.796235294117647,
      "grad_norm": 1.450897198143577,
      "learning_rate": 0.0002538,
      "loss": 0.8516,
      "step": 1692
    },
    {
      "epoch": 0.7967058823529412,
      "grad_norm": 0.6610888171161894,
      "learning_rate": 0.00025394999999999997,
      "loss": 0.3418,
      "step": 1693
    },
    {
      "epoch": 0.7971764705882353,
      "grad_norm": 1.147492996589895,
      "learning_rate": 0.0002541,
      "loss": 1.0312,
      "step": 1694
    },
    {
      "epoch": 0.7976470588235294,
      "grad_norm": 0.671984655072403,
      "learning_rate": 0.00025425,
      "loss": 0.3223,
      "step": 1695
    },
    {
      "epoch": 0.7981176470588235,
      "grad_norm": 1.771127274393749,
      "learning_rate": 0.00025439999999999995,
      "loss": 1.0547,
      "step": 1696
    },
    {
      "epoch": 0.7985882352941176,
      "grad_norm": 1.9815815322251944,
      "learning_rate": 0.00025455,
      "loss": 1.3984,
      "step": 1697
    },
    {
      "epoch": 0.7990588235294117,
      "grad_norm": 0.9398219259048619,
      "learning_rate": 0.00025469999999999996,
      "loss": 0.9102,
      "step": 1698
    },
    {
      "epoch": 0.7995294117647059,
      "grad_norm": 1.1024781407085034,
      "learning_rate": 0.00025485,
      "loss": 0.4922,
      "step": 1699
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5513957850735743,
      "learning_rate": 0.00025499999999999996,
      "loss": 0.1963,
      "step": 1700
    },
    {
      "epoch": 0.8004705882352942,
      "grad_norm": 0.8895838169385062,
      "learning_rate": 0.00025515,
      "loss": 0.4551,
      "step": 1701
    },
    {
      "epoch": 0.8009411764705883,
      "grad_norm": 1.9973479136083152,
      "learning_rate": 0.00025529999999999997,
      "loss": 0.8867,
      "step": 1702
    },
    {
      "epoch": 0.8014117647058824,
      "grad_norm": 1.2143312403145807,
      "learning_rate": 0.00025545,
      "loss": 0.6289,
      "step": 1703
    },
    {
      "epoch": 0.8018823529411765,
      "grad_norm": 1.4760628877355664,
      "learning_rate": 0.0002556,
      "loss": 0.9727,
      "step": 1704
    },
    {
      "epoch": 0.8023529411764706,
      "grad_norm": 1.695696585669628,
      "learning_rate": 0.00025575,
      "loss": 1.2031,
      "step": 1705
    },
    {
      "epoch": 0.8028235294117647,
      "grad_norm": 1.4475872277802804,
      "learning_rate": 0.0002559,
      "loss": 0.7031,
      "step": 1706
    },
    {
      "epoch": 0.8032941176470588,
      "grad_norm": 1.4969248538709072,
      "learning_rate": 0.00025604999999999996,
      "loss": 1.125,
      "step": 1707
    },
    {
      "epoch": 0.8037647058823529,
      "grad_norm": 0.9148502088536773,
      "learning_rate": 0.0002562,
      "loss": 0.4082,
      "step": 1708
    },
    {
      "epoch": 0.804235294117647,
      "grad_norm": 1.4281612583858783,
      "learning_rate": 0.00025634999999999997,
      "loss": 1.2188,
      "step": 1709
    },
    {
      "epoch": 0.8047058823529412,
      "grad_norm": 0.6540347621372996,
      "learning_rate": 0.00025649999999999995,
      "loss": 0.1748,
      "step": 1710
    },
    {
      "epoch": 0.8051764705882353,
      "grad_norm": 0.8373788578450159,
      "learning_rate": 0.00025665,
      "loss": 0.4375,
      "step": 1711
    },
    {
      "epoch": 0.8056470588235294,
      "grad_norm": 1.354346994296625,
      "learning_rate": 0.00025679999999999995,
      "loss": 0.8242,
      "step": 1712
    },
    {
      "epoch": 0.8061176470588235,
      "grad_norm": 1.649501242642325,
      "learning_rate": 0.00025695,
      "loss": 0.6758,
      "step": 1713
    },
    {
      "epoch": 0.8065882352941176,
      "grad_norm": 1.5331783861323443,
      "learning_rate": 0.00025709999999999996,
      "loss": 0.8281,
      "step": 1714
    },
    {
      "epoch": 0.8070588235294117,
      "grad_norm": 1.0631146263034408,
      "learning_rate": 0.00025725,
      "loss": 0.7383,
      "step": 1715
    },
    {
      "epoch": 0.8075294117647058,
      "grad_norm": 0.7320088642803494,
      "learning_rate": 0.00025739999999999997,
      "loss": 0.3438,
      "step": 1716
    },
    {
      "epoch": 0.808,
      "grad_norm": 0.8279080986412882,
      "learning_rate": 0.00025755,
      "loss": 0.4043,
      "step": 1717
    },
    {
      "epoch": 0.8084705882352942,
      "grad_norm": 1.08441779957978,
      "learning_rate": 0.0002577,
      "loss": 0.6133,
      "step": 1718
    },
    {
      "epoch": 0.8089411764705883,
      "grad_norm": 1.8034778085305896,
      "learning_rate": 0.00025785,
      "loss": 1.0938,
      "step": 1719
    },
    {
      "epoch": 0.8094117647058824,
      "grad_norm": 1.2040256574110022,
      "learning_rate": 0.000258,
      "loss": 0.8516,
      "step": 1720
    },
    {
      "epoch": 0.8098823529411765,
      "grad_norm": 1.4121053222297473,
      "learning_rate": 0.00025815,
      "loss": 0.8008,
      "step": 1721
    },
    {
      "epoch": 0.8103529411764706,
      "grad_norm": 1.590584981069569,
      "learning_rate": 0.0002583,
      "loss": 1.0078,
      "step": 1722
    },
    {
      "epoch": 0.8108235294117647,
      "grad_norm": 1.0867137645100218,
      "learning_rate": 0.00025844999999999997,
      "loss": 0.4102,
      "step": 1723
    },
    {
      "epoch": 0.8112941176470588,
      "grad_norm": 1.0218606657308211,
      "learning_rate": 0.0002586,
      "loss": 0.4668,
      "step": 1724
    },
    {
      "epoch": 0.8117647058823529,
      "grad_norm": 1.3645823453887282,
      "learning_rate": 0.00025875,
      "loss": 0.9961,
      "step": 1725
    },
    {
      "epoch": 0.812235294117647,
      "grad_norm": 1.3402836872407184,
      "learning_rate": 0.00025889999999999995,
      "loss": 0.8281,
      "step": 1726
    },
    {
      "epoch": 0.8127058823529412,
      "grad_norm": 1.9217689286874748,
      "learning_rate": 0.00025905,
      "loss": 1.1562,
      "step": 1727
    },
    {
      "epoch": 0.8131764705882353,
      "grad_norm": 0.8978455515430421,
      "learning_rate": 0.00025919999999999996,
      "loss": 0.332,
      "step": 1728
    },
    {
      "epoch": 0.8136470588235294,
      "grad_norm": 1.3195576134165319,
      "learning_rate": 0.00025935,
      "loss": 0.9062,
      "step": 1729
    },
    {
      "epoch": 0.8141176470588235,
      "grad_norm": 0.9279128176764816,
      "learning_rate": 0.00025949999999999997,
      "loss": 0.7188,
      "step": 1730
    },
    {
      "epoch": 0.8145882352941176,
      "grad_norm": 0.7405553539659622,
      "learning_rate": 0.00025965,
      "loss": 0.2617,
      "step": 1731
    },
    {
      "epoch": 0.8150588235294117,
      "grad_norm": 0.8684768207107986,
      "learning_rate": 0.00025979999999999997,
      "loss": 0.4199,
      "step": 1732
    },
    {
      "epoch": 0.8155294117647058,
      "grad_norm": 1.4271762140010986,
      "learning_rate": 0.00025995,
      "loss": 0.6562,
      "step": 1733
    },
    {
      "epoch": 0.816,
      "grad_norm": 1.514989924612964,
      "learning_rate": 0.0002601,
      "loss": 0.7383,
      "step": 1734
    },
    {
      "epoch": 0.8164705882352942,
      "grad_norm": 1.263099419306706,
      "learning_rate": 0.00026025,
      "loss": 0.6562,
      "step": 1735
    },
    {
      "epoch": 0.8169411764705883,
      "grad_norm": 0.8438346983249485,
      "learning_rate": 0.0002604,
      "loss": 0.3438,
      "step": 1736
    },
    {
      "epoch": 0.8174117647058824,
      "grad_norm": 0.9486661250856315,
      "learning_rate": 0.00026055,
      "loss": 0.4082,
      "step": 1737
    },
    {
      "epoch": 0.8178823529411765,
      "grad_norm": 0.8862100155253968,
      "learning_rate": 0.0002607,
      "loss": 0.3418,
      "step": 1738
    },
    {
      "epoch": 0.8183529411764706,
      "grad_norm": 0.770820015480115,
      "learning_rate": 0.00026084999999999997,
      "loss": 0.4004,
      "step": 1739
    },
    {
      "epoch": 0.8188235294117647,
      "grad_norm": 1.9963039578781372,
      "learning_rate": 0.000261,
      "loss": 1.0781,
      "step": 1740
    },
    {
      "epoch": 0.8192941176470588,
      "grad_norm": 0.9076960318211708,
      "learning_rate": 0.00026115,
      "loss": 0.5742,
      "step": 1741
    },
    {
      "epoch": 0.819764705882353,
      "grad_norm": 1.214683799814928,
      "learning_rate": 0.00026129999999999995,
      "loss": 0.7656,
      "step": 1742
    },
    {
      "epoch": 0.8202352941176471,
      "grad_norm": 1.4353098917986962,
      "learning_rate": 0.00026145,
      "loss": 0.8203,
      "step": 1743
    },
    {
      "epoch": 0.8207058823529412,
      "grad_norm": 0.5220351754211525,
      "learning_rate": 0.00026159999999999996,
      "loss": 0.2061,
      "step": 1744
    },
    {
      "epoch": 0.8211764705882353,
      "grad_norm": 0.9815344552245308,
      "learning_rate": 0.00026175,
      "loss": 0.2656,
      "step": 1745
    },
    {
      "epoch": 0.8216470588235294,
      "grad_norm": 2.1311762222625488,
      "learning_rate": 0.00026189999999999997,
      "loss": 0.9844,
      "step": 1746
    },
    {
      "epoch": 0.8221176470588235,
      "grad_norm": 0.6658510148707273,
      "learning_rate": 0.00026205,
      "loss": 0.3477,
      "step": 1747
    },
    {
      "epoch": 0.8225882352941176,
      "grad_norm": 0.9589850166385508,
      "learning_rate": 0.0002622,
      "loss": 0.5625,
      "step": 1748
    },
    {
      "epoch": 0.8230588235294117,
      "grad_norm": 0.9210842574916869,
      "learning_rate": 0.00026235,
      "loss": 0.457,
      "step": 1749
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.6069023271641225,
      "learning_rate": 0.0002625,
      "loss": 0.2295,
      "step": 1750
    },
    {
      "epoch": 0.824,
      "grad_norm": 1.6872455469190344,
      "learning_rate": 0.00026264999999999996,
      "loss": 1.0156,
      "step": 1751
    },
    {
      "epoch": 0.8244705882352941,
      "grad_norm": 1.4842706188532244,
      "learning_rate": 0.0002628,
      "loss": 0.8281,
      "step": 1752
    },
    {
      "epoch": 0.8249411764705883,
      "grad_norm": 1.142770319774701,
      "learning_rate": 0.00026294999999999997,
      "loss": 0.4941,
      "step": 1753
    },
    {
      "epoch": 0.8254117647058824,
      "grad_norm": 0.9762256383182953,
      "learning_rate": 0.0002631,
      "loss": 0.5352,
      "step": 1754
    },
    {
      "epoch": 0.8258823529411765,
      "grad_norm": 1.4410949944398597,
      "learning_rate": 0.00026325,
      "loss": 0.7695,
      "step": 1755
    },
    {
      "epoch": 0.8263529411764706,
      "grad_norm": 0.7483280040945792,
      "learning_rate": 0.00026339999999999995,
      "loss": 0.3633,
      "step": 1756
    },
    {
      "epoch": 0.8268235294117647,
      "grad_norm": 1.3725258077729119,
      "learning_rate": 0.00026355,
      "loss": 0.8711,
      "step": 1757
    },
    {
      "epoch": 0.8272941176470588,
      "grad_norm": 1.0305891967291965,
      "learning_rate": 0.00026369999999999996,
      "loss": 0.5391,
      "step": 1758
    },
    {
      "epoch": 0.827764705882353,
      "grad_norm": 1.1506289602252395,
      "learning_rate": 0.00026384999999999994,
      "loss": 0.5039,
      "step": 1759
    },
    {
      "epoch": 0.8282352941176471,
      "grad_norm": 3.405923111392192,
      "learning_rate": 0.00026399999999999997,
      "loss": 2.6562,
      "step": 1760
    },
    {
      "epoch": 0.8287058823529412,
      "grad_norm": 0.9217447452823468,
      "learning_rate": 0.00026414999999999994,
      "loss": 0.6055,
      "step": 1761
    },
    {
      "epoch": 0.8291764705882353,
      "grad_norm": 1.525528993984812,
      "learning_rate": 0.0002643,
      "loss": 0.9453,
      "step": 1762
    },
    {
      "epoch": 0.8296470588235294,
      "grad_norm": 1.524347598550406,
      "learning_rate": 0.00026444999999999995,
      "loss": 0.9727,
      "step": 1763
    },
    {
      "epoch": 0.8301176470588235,
      "grad_norm": 0.710553110903477,
      "learning_rate": 0.0002646,
      "loss": 0.2236,
      "step": 1764
    },
    {
      "epoch": 0.8305882352941176,
      "grad_norm": 1.976111854245375,
      "learning_rate": 0.00026474999999999996,
      "loss": 1.3203,
      "step": 1765
    },
    {
      "epoch": 0.8310588235294117,
      "grad_norm": 1.341993315406632,
      "learning_rate": 0.0002649,
      "loss": 0.543,
      "step": 1766
    },
    {
      "epoch": 0.8315294117647059,
      "grad_norm": 0.989834890870922,
      "learning_rate": 0.00026504999999999996,
      "loss": 0.8867,
      "step": 1767
    },
    {
      "epoch": 0.832,
      "grad_norm": 1.70738089751935,
      "learning_rate": 0.0002652,
      "loss": 1.125,
      "step": 1768
    },
    {
      "epoch": 0.8324705882352941,
      "grad_norm": 0.5745103230351019,
      "learning_rate": 0.00026534999999999997,
      "loss": 0.25,
      "step": 1769
    },
    {
      "epoch": 0.8329411764705882,
      "grad_norm": 0.9241944203819533,
      "learning_rate": 0.0002655,
      "loss": 0.6562,
      "step": 1770
    },
    {
      "epoch": 0.8334117647058824,
      "grad_norm": 0.7179290330361564,
      "learning_rate": 0.00026565,
      "loss": 0.2051,
      "step": 1771
    },
    {
      "epoch": 0.8338823529411765,
      "grad_norm": 1.142351389824931,
      "learning_rate": 0.00026579999999999996,
      "loss": 0.7422,
      "step": 1772
    },
    {
      "epoch": 0.8343529411764706,
      "grad_norm": 2.7309417891662204,
      "learning_rate": 0.00026595,
      "loss": 1.8828,
      "step": 1773
    },
    {
      "epoch": 0.8348235294117647,
      "grad_norm": 1.841900448356742,
      "learning_rate": 0.00026609999999999996,
      "loss": 0.4297,
      "step": 1774
    },
    {
      "epoch": 0.8352941176470589,
      "grad_norm": 1.243550682748326,
      "learning_rate": 0.00026624999999999994,
      "loss": 0.3418,
      "step": 1775
    },
    {
      "epoch": 0.835764705882353,
      "grad_norm": 2.8905681272503676,
      "learning_rate": 0.00026639999999999997,
      "loss": 0.1924,
      "step": 1776
    },
    {
      "epoch": 0.8362352941176471,
      "grad_norm": 3.4656334351050906,
      "learning_rate": 0.00026654999999999995,
      "loss": 1.6016,
      "step": 1777
    },
    {
      "epoch": 0.8367058823529412,
      "grad_norm": 13.66192676700596,
      "learning_rate": 0.0002667,
      "loss": 1.375,
      "step": 1778
    },
    {
      "epoch": 0.8371764705882353,
      "grad_norm": 34.412817357647555,
      "learning_rate": 0.00026684999999999995,
      "loss": 2.2188,
      "step": 1779
    },
    {
      "epoch": 0.8376470588235294,
      "grad_norm": 22.608677247110844,
      "learning_rate": 0.000267,
      "loss": 4.5938,
      "step": 1780
    },
    {
      "epoch": 0.8381176470588235,
      "grad_norm": 19.8922184933866,
      "learning_rate": 0.00026714999999999996,
      "loss": 4.0312,
      "step": 1781
    },
    {
      "epoch": 0.8385882352941176,
      "grad_norm": 14.762253929777383,
      "learning_rate": 0.0002673,
      "loss": 3.0156,
      "step": 1782
    },
    {
      "epoch": 0.8390588235294117,
      "grad_norm": 27.108162867392025,
      "learning_rate": 0.00026744999999999997,
      "loss": 4.2812,
      "step": 1783
    },
    {
      "epoch": 0.8395294117647059,
      "grad_norm": 16.19438340539569,
      "learning_rate": 0.0002676,
      "loss": 1.4531,
      "step": 1784
    },
    {
      "epoch": 0.84,
      "grad_norm": 11.330601881095102,
      "learning_rate": 0.00026775,
      "loss": 2.0781,
      "step": 1785
    },
    {
      "epoch": 0.8404705882352941,
      "grad_norm": 11.933762385916218,
      "learning_rate": 0.0002679,
      "loss": 2.0938,
      "step": 1786
    },
    {
      "epoch": 0.8409411764705882,
      "grad_norm": 11.509134000887732,
      "learning_rate": 0.00026805,
      "loss": 2.0781,
      "step": 1787
    },
    {
      "epoch": 0.8414117647058823,
      "grad_norm": 9.562846994911599,
      "learning_rate": 0.00026819999999999996,
      "loss": 1.1562,
      "step": 1788
    },
    {
      "epoch": 0.8418823529411765,
      "grad_norm": 7.272747574343054,
      "learning_rate": 0.00026835,
      "loss": 2.2188,
      "step": 1789
    },
    {
      "epoch": 0.8423529411764706,
      "grad_norm": 21.05815977909142,
      "learning_rate": 0.00026849999999999997,
      "loss": 1.5391,
      "step": 1790
    },
    {
      "epoch": 0.8428235294117647,
      "grad_norm": 3.188093933709084,
      "learning_rate": 0.00026864999999999994,
      "loss": 0.4551,
      "step": 1791
    },
    {
      "epoch": 0.8432941176470589,
      "grad_norm": 1.172784043347137,
      "learning_rate": 0.0002688,
      "loss": 0.3125,
      "step": 1792
    },
    {
      "epoch": 0.843764705882353,
      "grad_norm": 4.487621596249018,
      "learning_rate": 0.00026894999999999995,
      "loss": 2.0,
      "step": 1793
    },
    {
      "epoch": 0.8442352941176471,
      "grad_norm": 5.7594916669736955,
      "learning_rate": 0.0002691,
      "loss": 1.9375,
      "step": 1794
    },
    {
      "epoch": 0.8447058823529412,
      "grad_norm": 1.323655022527766,
      "learning_rate": 0.00026924999999999996,
      "loss": 0.6992,
      "step": 1795
    },
    {
      "epoch": 0.8451764705882353,
      "grad_norm": 1.4486036206267667,
      "learning_rate": 0.0002694,
      "loss": 0.9844,
      "step": 1796
    },
    {
      "epoch": 0.8456470588235294,
      "grad_norm": 1.5007218281713404,
      "learning_rate": 0.00026954999999999997,
      "loss": 1.0156,
      "step": 1797
    },
    {
      "epoch": 0.8461176470588235,
      "grad_norm": 1.2212863091935224,
      "learning_rate": 0.0002697,
      "loss": 0.6875,
      "step": 1798
    },
    {
      "epoch": 0.8465882352941176,
      "grad_norm": 2.1745893263313074,
      "learning_rate": 0.00026984999999999997,
      "loss": 1.3906,
      "step": 1799
    },
    {
      "epoch": 0.8470588235294118,
      "grad_norm": 1.57980453761785,
      "learning_rate": 0.00027,
      "loss": 0.9766,
      "step": 1800
    },
    {
      "epoch": 0.8475294117647059,
      "grad_norm": 1.5287649140367972,
      "learning_rate": 0.00027015,
      "loss": 0.9766,
      "step": 1801
    },
    {
      "epoch": 0.848,
      "grad_norm": 1.65729182229328,
      "learning_rate": 0.00027029999999999996,
      "loss": 1.3672,
      "step": 1802
    },
    {
      "epoch": 0.8484705882352941,
      "grad_norm": 0.7976661968327552,
      "learning_rate": 0.00027045,
      "loss": 0.5078,
      "step": 1803
    },
    {
      "epoch": 0.8489411764705882,
      "grad_norm": 0.9021556741862409,
      "learning_rate": 0.00027059999999999996,
      "loss": 0.4922,
      "step": 1804
    },
    {
      "epoch": 0.8494117647058823,
      "grad_norm": 2.2580899373588346,
      "learning_rate": 0.00027074999999999994,
      "loss": 0.5664,
      "step": 1805
    },
    {
      "epoch": 0.8498823529411764,
      "grad_norm": 1.5862233816738172,
      "learning_rate": 0.00027089999999999997,
      "loss": 0.625,
      "step": 1806
    },
    {
      "epoch": 0.8503529411764705,
      "grad_norm": 1.766026007336605,
      "learning_rate": 0.00027104999999999995,
      "loss": 0.7148,
      "step": 1807
    },
    {
      "epoch": 0.8508235294117648,
      "grad_norm": 1.449432557762882,
      "learning_rate": 0.0002712,
      "loss": 0.6875,
      "step": 1808
    },
    {
      "epoch": 0.8512941176470589,
      "grad_norm": 1.8446943423542899,
      "learning_rate": 0.00027134999999999995,
      "loss": 1.0391,
      "step": 1809
    },
    {
      "epoch": 0.851764705882353,
      "grad_norm": 1.4288440001040255,
      "learning_rate": 0.0002715,
      "loss": 1.2031,
      "step": 1810
    },
    {
      "epoch": 0.8522352941176471,
      "grad_norm": 1.3934792610583615,
      "learning_rate": 0.00027164999999999996,
      "loss": 1.2031,
      "step": 1811
    },
    {
      "epoch": 0.8527058823529412,
      "grad_norm": 0.9977337232484981,
      "learning_rate": 0.0002718,
      "loss": 0.6602,
      "step": 1812
    },
    {
      "epoch": 0.8531764705882353,
      "grad_norm": 1.866127160783351,
      "learning_rate": 0.00027194999999999997,
      "loss": 1.5547,
      "step": 1813
    },
    {
      "epoch": 0.8536470588235294,
      "grad_norm": 3.5415952582254726,
      "learning_rate": 0.0002721,
      "loss": 0.6406,
      "step": 1814
    },
    {
      "epoch": 0.8541176470588235,
      "grad_norm": 1.0693464565642967,
      "learning_rate": 0.00027225,
      "loss": 0.7461,
      "step": 1815
    },
    {
      "epoch": 0.8545882352941176,
      "grad_norm": 0.8709406545005657,
      "learning_rate": 0.0002724,
      "loss": 0.8281,
      "step": 1816
    },
    {
      "epoch": 0.8550588235294118,
      "grad_norm": 0.8306402838575642,
      "learning_rate": 0.00027255,
      "loss": 0.5586,
      "step": 1817
    },
    {
      "epoch": 0.8555294117647059,
      "grad_norm": 2.057435338818722,
      "learning_rate": 0.00027269999999999996,
      "loss": 1.8828,
      "step": 1818
    },
    {
      "epoch": 0.856,
      "grad_norm": 1.0430250264159855,
      "learning_rate": 0.00027285,
      "loss": 0.7109,
      "step": 1819
    },
    {
      "epoch": 0.8564705882352941,
      "grad_norm": 0.9796288125868112,
      "learning_rate": 0.00027299999999999997,
      "loss": 0.4941,
      "step": 1820
    },
    {
      "epoch": 0.8569411764705882,
      "grad_norm": 1.2210563286198313,
      "learning_rate": 0.00027314999999999994,
      "loss": 1.2422,
      "step": 1821
    },
    {
      "epoch": 0.8574117647058823,
      "grad_norm": 0.9206343694040683,
      "learning_rate": 0.0002733,
      "loss": 0.5273,
      "step": 1822
    },
    {
      "epoch": 0.8578823529411764,
      "grad_norm": 1.2275492158827401,
      "learning_rate": 0.00027344999999999995,
      "loss": 1.1484,
      "step": 1823
    },
    {
      "epoch": 0.8583529411764705,
      "grad_norm": 0.8286677212826333,
      "learning_rate": 0.0002736,
      "loss": 0.707,
      "step": 1824
    },
    {
      "epoch": 0.8588235294117647,
      "grad_norm": 1.1738800533701974,
      "learning_rate": 0.00027374999999999996,
      "loss": 0.793,
      "step": 1825
    },
    {
      "epoch": 0.8592941176470589,
      "grad_norm": 1.188802508925923,
      "learning_rate": 0.0002739,
      "loss": 0.7422,
      "step": 1826
    },
    {
      "epoch": 0.859764705882353,
      "grad_norm": 0.9371762639045739,
      "learning_rate": 0.00027404999999999997,
      "loss": 0.7148,
      "step": 1827
    },
    {
      "epoch": 0.8602352941176471,
      "grad_norm": 0.704071203636433,
      "learning_rate": 0.0002742,
      "loss": 0.5352,
      "step": 1828
    },
    {
      "epoch": 0.8607058823529412,
      "grad_norm": 0.8922336784099114,
      "learning_rate": 0.00027435,
      "loss": 0.5664,
      "step": 1829
    },
    {
      "epoch": 0.8611764705882353,
      "grad_norm": 1.4480370283379758,
      "learning_rate": 0.0002745,
      "loss": 1.1328,
      "step": 1830
    },
    {
      "epoch": 0.8616470588235294,
      "grad_norm": 1.2639526689138283,
      "learning_rate": 0.00027465,
      "loss": 0.9883,
      "step": 1831
    },
    {
      "epoch": 0.8621176470588235,
      "grad_norm": 0.9246292206435893,
      "learning_rate": 0.0002748,
      "loss": 0.4102,
      "step": 1832
    },
    {
      "epoch": 0.8625882352941177,
      "grad_norm": 1.237988796177922,
      "learning_rate": 0.00027495,
      "loss": 0.7656,
      "step": 1833
    },
    {
      "epoch": 0.8630588235294118,
      "grad_norm": 0.8679989071390133,
      "learning_rate": 0.00027509999999999996,
      "loss": 0.6445,
      "step": 1834
    },
    {
      "epoch": 0.8635294117647059,
      "grad_norm": 0.5088629957835269,
      "learning_rate": 0.00027525,
      "loss": 0.1533,
      "step": 1835
    },
    {
      "epoch": 0.864,
      "grad_norm": 0.5363177301247363,
      "learning_rate": 0.00027539999999999997,
      "loss": 0.2422,
      "step": 1836
    },
    {
      "epoch": 0.8644705882352941,
      "grad_norm": 1.3949316541602428,
      "learning_rate": 0.00027554999999999995,
      "loss": 0.8164,
      "step": 1837
    },
    {
      "epoch": 0.8649411764705882,
      "grad_norm": 0.4219627130513268,
      "learning_rate": 0.0002757,
      "loss": 0.1157,
      "step": 1838
    },
    {
      "epoch": 0.8654117647058823,
      "grad_norm": 0.6884840877133763,
      "learning_rate": 0.00027584999999999996,
      "loss": 0.2598,
      "step": 1839
    },
    {
      "epoch": 0.8658823529411764,
      "grad_norm": 0.6795522138852924,
      "learning_rate": 0.000276,
      "loss": 0.2617,
      "step": 1840
    },
    {
      "epoch": 0.8663529411764705,
      "grad_norm": 1.4166872455645563,
      "learning_rate": 0.00027614999999999996,
      "loss": 1.0703,
      "step": 1841
    },
    {
      "epoch": 0.8668235294117647,
      "grad_norm": 0.5423514748291302,
      "learning_rate": 0.0002763,
      "loss": 0.2617,
      "step": 1842
    },
    {
      "epoch": 0.8672941176470588,
      "grad_norm": 0.7884834357842325,
      "learning_rate": 0.00027644999999999997,
      "loss": 0.3125,
      "step": 1843
    },
    {
      "epoch": 0.867764705882353,
      "grad_norm": 1.020123110119524,
      "learning_rate": 0.0002766,
      "loss": 0.8242,
      "step": 1844
    },
    {
      "epoch": 0.8682352941176471,
      "grad_norm": 1.8844005070342003,
      "learning_rate": 0.00027675,
      "loss": 1.4844,
      "step": 1845
    },
    {
      "epoch": 0.8687058823529412,
      "grad_norm": 1.4536301077220644,
      "learning_rate": 0.0002769,
      "loss": 1.0781,
      "step": 1846
    },
    {
      "epoch": 0.8691764705882353,
      "grad_norm": 0.4002945762923648,
      "learning_rate": 0.00027705,
      "loss": 0.2812,
      "step": 1847
    },
    {
      "epoch": 0.8696470588235294,
      "grad_norm": 0.8622988014373819,
      "learning_rate": 0.0002772,
      "loss": 0.5117,
      "step": 1848
    },
    {
      "epoch": 0.8701176470588236,
      "grad_norm": 1.1211146637476932,
      "learning_rate": 0.00027735,
      "loss": 0.668,
      "step": 1849
    },
    {
      "epoch": 0.8705882352941177,
      "grad_norm": 0.44330365102554203,
      "learning_rate": 0.00027749999999999997,
      "loss": 0.1865,
      "step": 1850
    },
    {
      "epoch": 0.8710588235294118,
      "grad_norm": 1.5325915775486025,
      "learning_rate": 0.00027764999999999995,
      "loss": 1.1797,
      "step": 1851
    },
    {
      "epoch": 0.8715294117647059,
      "grad_norm": 0.5474659878757517,
      "learning_rate": 0.0002778,
      "loss": 0.2129,
      "step": 1852
    },
    {
      "epoch": 0.872,
      "grad_norm": 1.287916554298411,
      "learning_rate": 0.00027794999999999995,
      "loss": 1.0547,
      "step": 1853
    },
    {
      "epoch": 0.8724705882352941,
      "grad_norm": 1.594188017704641,
      "learning_rate": 0.0002781,
      "loss": 1.625,
      "step": 1854
    },
    {
      "epoch": 0.8729411764705882,
      "grad_norm": 0.9409156901047616,
      "learning_rate": 0.00027824999999999996,
      "loss": 0.7656,
      "step": 1855
    },
    {
      "epoch": 0.8734117647058823,
      "grad_norm": 0.8347349462104322,
      "learning_rate": 0.0002784,
      "loss": 0.6758,
      "step": 1856
    },
    {
      "epoch": 0.8738823529411764,
      "grad_norm": 1.7642682957585583,
      "learning_rate": 0.00027854999999999997,
      "loss": 1.0312,
      "step": 1857
    },
    {
      "epoch": 0.8743529411764706,
      "grad_norm": 1.553408798351185,
      "learning_rate": 0.0002787,
      "loss": 0.8281,
      "step": 1858
    },
    {
      "epoch": 0.8748235294117647,
      "grad_norm": 0.4848617364776964,
      "learning_rate": 0.00027885,
      "loss": 0.1602,
      "step": 1859
    },
    {
      "epoch": 0.8752941176470588,
      "grad_norm": 0.8655537692463416,
      "learning_rate": 0.000279,
      "loss": 0.416,
      "step": 1860
    },
    {
      "epoch": 0.8757647058823529,
      "grad_norm": 0.9526365143913945,
      "learning_rate": 0.00027915,
      "loss": 0.4395,
      "step": 1861
    },
    {
      "epoch": 0.8762352941176471,
      "grad_norm": 1.5369932309508116,
      "learning_rate": 0.0002793,
      "loss": 0.875,
      "step": 1862
    },
    {
      "epoch": 0.8767058823529412,
      "grad_norm": 0.4183858156699978,
      "learning_rate": 0.00027945,
      "loss": 0.2852,
      "step": 1863
    },
    {
      "epoch": 0.8771764705882353,
      "grad_norm": 1.173729018065345,
      "learning_rate": 0.00027959999999999997,
      "loss": 0.5586,
      "step": 1864
    },
    {
      "epoch": 0.8776470588235294,
      "grad_norm": 1.5250573571952821,
      "learning_rate": 0.00027975,
      "loss": 0.6211,
      "step": 1865
    },
    {
      "epoch": 0.8781176470588236,
      "grad_norm": 1.4584519345329678,
      "learning_rate": 0.0002799,
      "loss": 1.1172,
      "step": 1866
    },
    {
      "epoch": 0.8785882352941177,
      "grad_norm": 1.3192421117490185,
      "learning_rate": 0.00028004999999999995,
      "loss": 0.9219,
      "step": 1867
    },
    {
      "epoch": 0.8790588235294118,
      "grad_norm": 1.2854338166067432,
      "learning_rate": 0.0002802,
      "loss": 1.1797,
      "step": 1868
    },
    {
      "epoch": 0.8795294117647059,
      "grad_norm": 0.9381960679656568,
      "learning_rate": 0.00028034999999999996,
      "loss": 0.8984,
      "step": 1869
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2366572128772733,
      "learning_rate": 0.0002805,
      "loss": 0.9922,
      "step": 1870
    },
    {
      "epoch": 0.8804705882352941,
      "grad_norm": 1.715621090760781,
      "learning_rate": 0.00028064999999999996,
      "loss": 0.8633,
      "step": 1871
    },
    {
      "epoch": 0.8809411764705882,
      "grad_norm": 1.0790159967440378,
      "learning_rate": 0.0002808,
      "loss": 0.7617,
      "step": 1872
    },
    {
      "epoch": 0.8814117647058823,
      "grad_norm": 1.1381840614675276,
      "learning_rate": 0.00028094999999999997,
      "loss": 0.7617,
      "step": 1873
    },
    {
      "epoch": 0.8818823529411765,
      "grad_norm": 0.7310223881363398,
      "learning_rate": 0.0002811,
      "loss": 0.668,
      "step": 1874
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 1.662141025266446,
      "learning_rate": 0.00028125,
      "loss": 0.5742,
      "step": 1875
    },
    {
      "epoch": 0.8828235294117647,
      "grad_norm": 2.0540610011383804,
      "learning_rate": 0.00028139999999999996,
      "loss": 1.1016,
      "step": 1876
    },
    {
      "epoch": 0.8832941176470588,
      "grad_norm": 1.1699516541810293,
      "learning_rate": 0.00028155,
      "loss": 0.5703,
      "step": 1877
    },
    {
      "epoch": 0.8837647058823529,
      "grad_norm": 0.6661699055407206,
      "learning_rate": 0.00028169999999999996,
      "loss": 0.4355,
      "step": 1878
    },
    {
      "epoch": 0.884235294117647,
      "grad_norm": 2.0254924098038685,
      "learning_rate": 0.00028185,
      "loss": 1.1562,
      "step": 1879
    },
    {
      "epoch": 0.8847058823529412,
      "grad_norm": 0.8176685195781662,
      "learning_rate": 0.00028199999999999997,
      "loss": 0.3066,
      "step": 1880
    },
    {
      "epoch": 0.8851764705882353,
      "grad_norm": 0.7873711617402352,
      "learning_rate": 0.00028215,
      "loss": 0.459,
      "step": 1881
    },
    {
      "epoch": 0.8856470588235295,
      "grad_norm": 0.7149483629861934,
      "learning_rate": 0.0002823,
      "loss": 0.3652,
      "step": 1882
    },
    {
      "epoch": 0.8861176470588236,
      "grad_norm": 1.1650520470232684,
      "learning_rate": 0.00028244999999999995,
      "loss": 0.6016,
      "step": 1883
    },
    {
      "epoch": 0.8865882352941177,
      "grad_norm": 1.1398857292590279,
      "learning_rate": 0.0002826,
      "loss": 0.7266,
      "step": 1884
    },
    {
      "epoch": 0.8870588235294118,
      "grad_norm": 1.432118383137582,
      "learning_rate": 0.00028274999999999996,
      "loss": 1.0156,
      "step": 1885
    },
    {
      "epoch": 0.8875294117647059,
      "grad_norm": 0.9941932083935625,
      "learning_rate": 0.00028289999999999994,
      "loss": 0.6758,
      "step": 1886
    },
    {
      "epoch": 0.888,
      "grad_norm": 1.2854279365464654,
      "learning_rate": 0.00028304999999999997,
      "loss": 0.7383,
      "step": 1887
    },
    {
      "epoch": 0.8884705882352941,
      "grad_norm": 1.1991393275687168,
      "learning_rate": 0.00028319999999999994,
      "loss": 0.4668,
      "step": 1888
    },
    {
      "epoch": 0.8889411764705882,
      "grad_norm": 1.992396038338569,
      "learning_rate": 0.00028335,
      "loss": 1.0547,
      "step": 1889
    },
    {
      "epoch": 0.8894117647058823,
      "grad_norm": 0.990472874894226,
      "learning_rate": 0.00028349999999999995,
      "loss": 0.4785,
      "step": 1890
    },
    {
      "epoch": 0.8898823529411765,
      "grad_norm": 0.9836686884249948,
      "learning_rate": 0.00028365,
      "loss": 0.6289,
      "step": 1891
    },
    {
      "epoch": 0.8903529411764706,
      "grad_norm": 1.101829348742268,
      "learning_rate": 0.00028379999999999996,
      "loss": 0.8633,
      "step": 1892
    },
    {
      "epoch": 0.8908235294117647,
      "grad_norm": 1.0117913289224825,
      "learning_rate": 0.00028395,
      "loss": 0.6602,
      "step": 1893
    },
    {
      "epoch": 0.8912941176470588,
      "grad_norm": 0.4392112398692379,
      "learning_rate": 0.00028409999999999997,
      "loss": 0.1768,
      "step": 1894
    },
    {
      "epoch": 0.8917647058823529,
      "grad_norm": 1.656982991154685,
      "learning_rate": 0.00028425,
      "loss": 0.9609,
      "step": 1895
    },
    {
      "epoch": 0.892235294117647,
      "grad_norm": 0.621882879271925,
      "learning_rate": 0.0002844,
      "loss": 0.3594,
      "step": 1896
    },
    {
      "epoch": 0.8927058823529411,
      "grad_norm": 1.2520719289155748,
      "learning_rate": 0.00028455,
      "loss": 0.6719,
      "step": 1897
    },
    {
      "epoch": 0.8931764705882353,
      "grad_norm": 1.552463294741675,
      "learning_rate": 0.0002847,
      "loss": 1.1641,
      "step": 1898
    },
    {
      "epoch": 0.8936470588235295,
      "grad_norm": 0.6526803262812187,
      "learning_rate": 0.00028484999999999996,
      "loss": 0.4453,
      "step": 1899
    },
    {
      "epoch": 0.8941176470588236,
      "grad_norm": 0.7584751889138172,
      "learning_rate": 0.000285,
      "loss": 0.375,
      "step": 1900
    },
    {
      "epoch": 0.8945882352941177,
      "grad_norm": 0.7075685766546136,
      "learning_rate": 0.00028514999999999997,
      "loss": 0.291,
      "step": 1901
    },
    {
      "epoch": 0.8950588235294118,
      "grad_norm": 2.8738168217713613,
      "learning_rate": 0.00028529999999999994,
      "loss": 0.8242,
      "step": 1902
    },
    {
      "epoch": 0.8955294117647059,
      "grad_norm": 1.9564664082119287,
      "learning_rate": 0.00028544999999999997,
      "loss": 0.875,
      "step": 1903
    },
    {
      "epoch": 0.896,
      "grad_norm": 0.7125732850382733,
      "learning_rate": 0.00028559999999999995,
      "loss": 0.3164,
      "step": 1904
    },
    {
      "epoch": 0.8964705882352941,
      "grad_norm": 0.9062815942029959,
      "learning_rate": 0.00028575,
      "loss": 0.3008,
      "step": 1905
    },
    {
      "epoch": 0.8969411764705882,
      "grad_norm": 0.961710957637222,
      "learning_rate": 0.00028589999999999996,
      "loss": 0.5312,
      "step": 1906
    },
    {
      "epoch": 0.8974117647058824,
      "grad_norm": 1.7377821367532496,
      "learning_rate": 0.00028605,
      "loss": 1.2188,
      "step": 1907
    },
    {
      "epoch": 0.8978823529411765,
      "grad_norm": 2.795508233506692,
      "learning_rate": 0.00028619999999999996,
      "loss": 1.4219,
      "step": 1908
    },
    {
      "epoch": 0.8983529411764706,
      "grad_norm": 1.7257105778709947,
      "learning_rate": 0.00028635,
      "loss": 1.1328,
      "step": 1909
    },
    {
      "epoch": 0.8988235294117647,
      "grad_norm": 0.8761387804855634,
      "learning_rate": 0.00028649999999999997,
      "loss": 0.5195,
      "step": 1910
    },
    {
      "epoch": 0.8992941176470588,
      "grad_norm": 0.9773584628420604,
      "learning_rate": 0.00028665,
      "loss": 0.4492,
      "step": 1911
    },
    {
      "epoch": 0.8997647058823529,
      "grad_norm": 1.366188879183032,
      "learning_rate": 0.0002868,
      "loss": 0.625,
      "step": 1912
    },
    {
      "epoch": 0.900235294117647,
      "grad_norm": 1.0532339476693329,
      "learning_rate": 0.00028694999999999995,
      "loss": 0.3457,
      "step": 1913
    },
    {
      "epoch": 0.9007058823529411,
      "grad_norm": 0.9927561314003303,
      "learning_rate": 0.0002871,
      "loss": 0.2637,
      "step": 1914
    },
    {
      "epoch": 0.9011764705882352,
      "grad_norm": 3.16292909206933,
      "learning_rate": 0.00028724999999999996,
      "loss": 0.8594,
      "step": 1915
    },
    {
      "epoch": 0.9016470588235294,
      "grad_norm": 1.9129531601716385,
      "learning_rate": 0.00028739999999999994,
      "loss": 1.0703,
      "step": 1916
    },
    {
      "epoch": 0.9021176470588236,
      "grad_norm": 0.9490171581580067,
      "learning_rate": 0.00028754999999999997,
      "loss": 0.707,
      "step": 1917
    },
    {
      "epoch": 0.9025882352941177,
      "grad_norm": 0.4572241851535307,
      "learning_rate": 0.00028769999999999995,
      "loss": 0.1602,
      "step": 1918
    },
    {
      "epoch": 0.9030588235294118,
      "grad_norm": 1.8017394506938038,
      "learning_rate": 0.00028785,
      "loss": 1.0781,
      "step": 1919
    },
    {
      "epoch": 0.9035294117647059,
      "grad_norm": 2.068071047921743,
      "learning_rate": 0.00028799999999999995,
      "loss": 0.8711,
      "step": 1920
    },
    {
      "epoch": 0.904,
      "grad_norm": 0.6488814702660117,
      "learning_rate": 0.00028815,
      "loss": 0.2354,
      "step": 1921
    },
    {
      "epoch": 0.9044705882352941,
      "grad_norm": 0.7811015221821748,
      "learning_rate": 0.00028829999999999996,
      "loss": 0.3516,
      "step": 1922
    },
    {
      "epoch": 0.9049411764705882,
      "grad_norm": 0.5912618703700292,
      "learning_rate": 0.00028845,
      "loss": 0.2051,
      "step": 1923
    },
    {
      "epoch": 0.9054117647058824,
      "grad_norm": 1.0285073006201677,
      "learning_rate": 0.00028859999999999997,
      "loss": 0.4863,
      "step": 1924
    },
    {
      "epoch": 0.9058823529411765,
      "grad_norm": 1.7507421780127566,
      "learning_rate": 0.00028875,
      "loss": 0.707,
      "step": 1925
    },
    {
      "epoch": 0.9063529411764706,
      "grad_norm": 0.9160303199023127,
      "learning_rate": 0.0002889,
      "loss": 0.252,
      "step": 1926
    },
    {
      "epoch": 0.9068235294117647,
      "grad_norm": 1.9100172245115525,
      "learning_rate": 0.00028905,
      "loss": 1.1328,
      "step": 1927
    },
    {
      "epoch": 0.9072941176470588,
      "grad_norm": 1.6167882809154688,
      "learning_rate": 0.0002892,
      "loss": 1.0,
      "step": 1928
    },
    {
      "epoch": 0.9077647058823529,
      "grad_norm": 0.5697311281885786,
      "learning_rate": 0.00028934999999999996,
      "loss": 0.3477,
      "step": 1929
    },
    {
      "epoch": 0.908235294117647,
      "grad_norm": 0.9186984190141684,
      "learning_rate": 0.0002895,
      "loss": 0.7383,
      "step": 1930
    },
    {
      "epoch": 0.9087058823529411,
      "grad_norm": 1.376664174998991,
      "learning_rate": 0.00028964999999999997,
      "loss": 0.9258,
      "step": 1931
    },
    {
      "epoch": 0.9091764705882353,
      "grad_norm": 0.9407722143046978,
      "learning_rate": 0.00028979999999999994,
      "loss": 0.5938,
      "step": 1932
    },
    {
      "epoch": 0.9096470588235294,
      "grad_norm": 1.915631096308762,
      "learning_rate": 0.00028995,
      "loss": 1.0781,
      "step": 1933
    },
    {
      "epoch": 0.9101176470588235,
      "grad_norm": 1.7210820038012198,
      "learning_rate": 0.00029009999999999995,
      "loss": 0.918,
      "step": 1934
    },
    {
      "epoch": 0.9105882352941177,
      "grad_norm": 0.7968090568216538,
      "learning_rate": 0.00029025,
      "loss": 0.4668,
      "step": 1935
    },
    {
      "epoch": 0.9110588235294118,
      "grad_norm": 5.678703862711699,
      "learning_rate": 0.00029039999999999996,
      "loss": 0.8555,
      "step": 1936
    },
    {
      "epoch": 0.9115294117647059,
      "grad_norm": 0.5622379725412932,
      "learning_rate": 0.00029055,
      "loss": 0.2715,
      "step": 1937
    },
    {
      "epoch": 0.912,
      "grad_norm": 1.9414640791005187,
      "learning_rate": 0.00029069999999999996,
      "loss": 1.0625,
      "step": 1938
    },
    {
      "epoch": 0.9124705882352941,
      "grad_norm": 0.5394902264553274,
      "learning_rate": 0.00029085,
      "loss": 0.1982,
      "step": 1939
    },
    {
      "epoch": 0.9129411764705883,
      "grad_norm": 1.5626602245950378,
      "learning_rate": 0.00029099999999999997,
      "loss": 0.7266,
      "step": 1940
    },
    {
      "epoch": 0.9134117647058824,
      "grad_norm": 1.3199400411964135,
      "learning_rate": 0.00029115,
      "loss": 0.6367,
      "step": 1941
    },
    {
      "epoch": 0.9138823529411765,
      "grad_norm": 0.9964529931007995,
      "learning_rate": 0.0002913,
      "loss": 0.9102,
      "step": 1942
    },
    {
      "epoch": 0.9143529411764706,
      "grad_norm": 1.1100603284213246,
      "learning_rate": 0.00029145,
      "loss": 0.8984,
      "step": 1943
    },
    {
      "epoch": 0.9148235294117647,
      "grad_norm": 1.4818733599362761,
      "learning_rate": 0.0002916,
      "loss": 0.9453,
      "step": 1944
    },
    {
      "epoch": 0.9152941176470588,
      "grad_norm": 0.6595317537709391,
      "learning_rate": 0.00029174999999999996,
      "loss": 0.2432,
      "step": 1945
    },
    {
      "epoch": 0.9157647058823529,
      "grad_norm": 1.3199647120866622,
      "learning_rate": 0.0002919,
      "loss": 0.707,
      "step": 1946
    },
    {
      "epoch": 0.916235294117647,
      "grad_norm": 1.8210335771395092,
      "learning_rate": 0.00029204999999999997,
      "loss": 0.8906,
      "step": 1947
    },
    {
      "epoch": 0.9167058823529411,
      "grad_norm": 1.076666803960329,
      "learning_rate": 0.00029219999999999995,
      "loss": 0.6172,
      "step": 1948
    },
    {
      "epoch": 0.9171764705882353,
      "grad_norm": 0.4595320311366626,
      "learning_rate": 0.00029235,
      "loss": 0.1416,
      "step": 1949
    },
    {
      "epoch": 0.9176470588235294,
      "grad_norm": 1.4064109885140075,
      "learning_rate": 0.00029249999999999995,
      "loss": 0.7344,
      "step": 1950
    },
    {
      "epoch": 0.9181176470588235,
      "grad_norm": 1.2357870801549375,
      "learning_rate": 0.00029265,
      "loss": 0.6094,
      "step": 1951
    },
    {
      "epoch": 0.9185882352941176,
      "grad_norm": 1.4898178828586697,
      "learning_rate": 0.00029279999999999996,
      "loss": 0.9453,
      "step": 1952
    },
    {
      "epoch": 0.9190588235294118,
      "grad_norm": 1.4849614850895052,
      "learning_rate": 0.00029295,
      "loss": 0.9883,
      "step": 1953
    },
    {
      "epoch": 0.9195294117647059,
      "grad_norm": 0.7974555896633891,
      "learning_rate": 0.00029309999999999997,
      "loss": 0.4121,
      "step": 1954
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.304004353198665,
      "learning_rate": 0.00029325,
      "loss": 0.9258,
      "step": 1955
    },
    {
      "epoch": 0.9204705882352942,
      "grad_norm": 5.34199740746897,
      "learning_rate": 0.0002934,
      "loss": 0.8164,
      "step": 1956
    },
    {
      "epoch": 0.9209411764705883,
      "grad_norm": 1.4176272123341078,
      "learning_rate": 0.00029355,
      "loss": 0.5312,
      "step": 1957
    },
    {
      "epoch": 0.9214117647058824,
      "grad_norm": 1.5291215302928278,
      "learning_rate": 0.0002937,
      "loss": 0.6133,
      "step": 1958
    },
    {
      "epoch": 0.9218823529411765,
      "grad_norm": 0.772372778411312,
      "learning_rate": 0.00029384999999999996,
      "loss": 0.3047,
      "step": 1959
    },
    {
      "epoch": 0.9223529411764706,
      "grad_norm": 1.8203681482122263,
      "learning_rate": 0.000294,
      "loss": 1.0781,
      "step": 1960
    },
    {
      "epoch": 0.9228235294117647,
      "grad_norm": 1.2025589950397089,
      "learning_rate": 0.00029414999999999997,
      "loss": 0.7188,
      "step": 1961
    },
    {
      "epoch": 0.9232941176470588,
      "grad_norm": 1.128320262344535,
      "learning_rate": 0.00029429999999999994,
      "loss": 0.6562,
      "step": 1962
    },
    {
      "epoch": 0.9237647058823529,
      "grad_norm": 0.5593577348629377,
      "learning_rate": 0.00029445,
      "loss": 0.2832,
      "step": 1963
    },
    {
      "epoch": 0.924235294117647,
      "grad_norm": 2.02467019242733,
      "learning_rate": 0.00029459999999999995,
      "loss": 1.0859,
      "step": 1964
    },
    {
      "epoch": 0.9247058823529412,
      "grad_norm": 2.0867648279975866,
      "learning_rate": 0.00029475,
      "loss": 1.5703,
      "step": 1965
    },
    {
      "epoch": 0.9251764705882353,
      "grad_norm": 1.4140919142524802,
      "learning_rate": 0.00029489999999999996,
      "loss": 0.8828,
      "step": 1966
    },
    {
      "epoch": 0.9256470588235294,
      "grad_norm": 1.4330477179399765,
      "learning_rate": 0.00029505,
      "loss": 0.8867,
      "step": 1967
    },
    {
      "epoch": 0.9261176470588235,
      "grad_norm": 1.6617875337592063,
      "learning_rate": 0.00029519999999999997,
      "loss": 1.25,
      "step": 1968
    },
    {
      "epoch": 0.9265882352941176,
      "grad_norm": 1.678221427687536,
      "learning_rate": 0.00029535,
      "loss": 1.1016,
      "step": 1969
    },
    {
      "epoch": 0.9270588235294117,
      "grad_norm": 1.4395412463328425,
      "learning_rate": 0.00029549999999999997,
      "loss": 0.7773,
      "step": 1970
    },
    {
      "epoch": 0.9275294117647059,
      "grad_norm": 0.8452632192811274,
      "learning_rate": 0.00029565,
      "loss": 0.3242,
      "step": 1971
    },
    {
      "epoch": 0.928,
      "grad_norm": 0.5264998247527982,
      "learning_rate": 0.0002958,
      "loss": 0.1797,
      "step": 1972
    },
    {
      "epoch": 0.9284705882352942,
      "grad_norm": 0.8654508955871403,
      "learning_rate": 0.00029595,
      "loss": 0.4355,
      "step": 1973
    },
    {
      "epoch": 0.9289411764705883,
      "grad_norm": 0.9905504131784794,
      "learning_rate": 0.0002961,
      "loss": 0.7148,
      "step": 1974
    },
    {
      "epoch": 0.9294117647058824,
      "grad_norm": 1.494110217872643,
      "learning_rate": 0.00029624999999999996,
      "loss": 1.2812,
      "step": 1975
    },
    {
      "epoch": 0.9298823529411765,
      "grad_norm": 0.49554763807892344,
      "learning_rate": 0.0002964,
      "loss": 0.1602,
      "step": 1976
    },
    {
      "epoch": 0.9303529411764706,
      "grad_norm": 0.7906964622755289,
      "learning_rate": 0.00029654999999999997,
      "loss": 0.4863,
      "step": 1977
    },
    {
      "epoch": 0.9308235294117647,
      "grad_norm": 2.3389257650065236,
      "learning_rate": 0.00029669999999999995,
      "loss": 1.2734,
      "step": 1978
    },
    {
      "epoch": 0.9312941176470588,
      "grad_norm": 1.5547203788940742,
      "learning_rate": 0.00029685,
      "loss": 0.9805,
      "step": 1979
    },
    {
      "epoch": 0.9317647058823529,
      "grad_norm": 1.4918990605392133,
      "learning_rate": 0.00029699999999999996,
      "loss": 1.1719,
      "step": 1980
    },
    {
      "epoch": 0.932235294117647,
      "grad_norm": 0.9750173808385745,
      "learning_rate": 0.00029715,
      "loss": 0.4395,
      "step": 1981
    },
    {
      "epoch": 0.9327058823529412,
      "grad_norm": 0.9171492783946407,
      "learning_rate": 0.00029729999999999996,
      "loss": 0.3203,
      "step": 1982
    },
    {
      "epoch": 0.9331764705882353,
      "grad_norm": 1.2097662070369444,
      "learning_rate": 0.00029745,
      "loss": 0.5508,
      "step": 1983
    },
    {
      "epoch": 0.9336470588235294,
      "grad_norm": 0.992200527087095,
      "learning_rate": 0.00029759999999999997,
      "loss": 0.3809,
      "step": 1984
    },
    {
      "epoch": 0.9341176470588235,
      "grad_norm": 1.830476039730717,
      "learning_rate": 0.00029775,
      "loss": 1.2969,
      "step": 1985
    },
    {
      "epoch": 0.9345882352941176,
      "grad_norm": 1.375310227191483,
      "learning_rate": 0.0002979,
      "loss": 0.7383,
      "step": 1986
    },
    {
      "epoch": 0.9350588235294117,
      "grad_norm": 1.2521586720776514,
      "learning_rate": 0.00029805,
      "loss": 0.9883,
      "step": 1987
    },
    {
      "epoch": 0.9355294117647058,
      "grad_norm": 1.2341465990059863,
      "learning_rate": 0.0002982,
      "loss": 1.1328,
      "step": 1988
    },
    {
      "epoch": 0.936,
      "grad_norm": 1.1454799161525113,
      "learning_rate": 0.00029835,
      "loss": 0.6641,
      "step": 1989
    },
    {
      "epoch": 0.9364705882352942,
      "grad_norm": 2.684480514981279,
      "learning_rate": 0.0002985,
      "loss": 1.4453,
      "step": 1990
    },
    {
      "epoch": 0.9369411764705883,
      "grad_norm": 2.470833274389761,
      "learning_rate": 0.00029864999999999997,
      "loss": 1.1484,
      "step": 1991
    },
    {
      "epoch": 0.9374117647058824,
      "grad_norm": 0.6616781152111081,
      "learning_rate": 0.0002988,
      "loss": 0.2227,
      "step": 1992
    },
    {
      "epoch": 0.9378823529411765,
      "grad_norm": 1.4847415990247577,
      "learning_rate": 0.00029895,
      "loss": 1.0312,
      "step": 1993
    },
    {
      "epoch": 0.9383529411764706,
      "grad_norm": 0.9070446221799453,
      "learning_rate": 0.00029909999999999995,
      "loss": 0.4316,
      "step": 1994
    },
    {
      "epoch": 0.9388235294117647,
      "grad_norm": 2.3817981662702414,
      "learning_rate": 0.00029925,
      "loss": 1.5859,
      "step": 1995
    },
    {
      "epoch": 0.9392941176470588,
      "grad_norm": 1.2719061846424005,
      "learning_rate": 0.00029939999999999996,
      "loss": 0.7578,
      "step": 1996
    },
    {
      "epoch": 0.939764705882353,
      "grad_norm": 0.9676611093358695,
      "learning_rate": 0.00029955,
      "loss": 0.6562,
      "step": 1997
    },
    {
      "epoch": 0.9402352941176471,
      "grad_norm": 1.3947566326082035,
      "learning_rate": 0.00029969999999999997,
      "loss": 0.8633,
      "step": 1998
    },
    {
      "epoch": 0.9407058823529412,
      "grad_norm": 1.0802628996408743,
      "learning_rate": 0.00029985,
      "loss": 0.8203,
      "step": 1999
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.8095952523201979,
      "learning_rate": 0.0003,
      "loss": 0.5312,
      "step": 2000
    },
    {
      "epoch": 0.9416470588235294,
      "grad_norm": 1.4866343844293843,
      "learning_rate": 0.0002999999935946321,
      "loss": 0.6406,
      "step": 2001
    },
    {
      "epoch": 0.9421176470588235,
      "grad_norm": 1.5776865130416733,
      "learning_rate": 0.00029999997437852905,
      "loss": 1.5703,
      "step": 2002
    },
    {
      "epoch": 0.9425882352941176,
      "grad_norm": 0.6475151526331178,
      "learning_rate": 0.0002999999423516924,
      "loss": 0.4219,
      "step": 2003
    },
    {
      "epoch": 0.9430588235294117,
      "grad_norm": 1.1509971255874059,
      "learning_rate": 0.00029999989751412494,
      "loss": 0.5898,
      "step": 2004
    },
    {
      "epoch": 0.9435294117647058,
      "grad_norm": 1.930749030195737,
      "learning_rate": 0.00029999983986583054,
      "loss": 1.5234,
      "step": 2005
    },
    {
      "epoch": 0.944,
      "grad_norm": 1.063293988249853,
      "learning_rate": 0.000299999769406814,
      "loss": 0.7578,
      "step": 2006
    },
    {
      "epoch": 0.9444705882352942,
      "grad_norm": 0.902606800787882,
      "learning_rate": 0.0002999996861370814,
      "loss": 1.0859,
      "step": 2007
    },
    {
      "epoch": 0.9449411764705883,
      "grad_norm": 0.7799414717316769,
      "learning_rate": 0.0002999995900566399,
      "loss": 0.4512,
      "step": 2008
    },
    {
      "epoch": 0.9454117647058824,
      "grad_norm": 0.8233525441635793,
      "learning_rate": 0.0002999994811654977,
      "loss": 0.4434,
      "step": 2009
    },
    {
      "epoch": 0.9458823529411765,
      "grad_norm": 0.9730573148168729,
      "learning_rate": 0.000299999359463664,
      "loss": 0.4043,
      "step": 2010
    },
    {
      "epoch": 0.9463529411764706,
      "grad_norm": 1.718614407728104,
      "learning_rate": 0.0002999992249511493,
      "loss": 0.9141,
      "step": 2011
    },
    {
      "epoch": 0.9468235294117647,
      "grad_norm": 0.7893585000670594,
      "learning_rate": 0.00029999907762796505,
      "loss": 0.4238,
      "step": 2012
    },
    {
      "epoch": 0.9472941176470588,
      "grad_norm": 1.3170603945265091,
      "learning_rate": 0.00029999891749412375,
      "loss": 0.9023,
      "step": 2013
    },
    {
      "epoch": 0.947764705882353,
      "grad_norm": 1.016284414207036,
      "learning_rate": 0.0002999987445496392,
      "loss": 0.8125,
      "step": 2014
    },
    {
      "epoch": 0.9482352941176471,
      "grad_norm": 0.9177212159751115,
      "learning_rate": 0.00029999855879452616,
      "loss": 0.3945,
      "step": 2015
    },
    {
      "epoch": 0.9487058823529412,
      "grad_norm": 1.7559624639495517,
      "learning_rate": 0.00029999836022880045,
      "loss": 0.8633,
      "step": 2016
    },
    {
      "epoch": 0.9491764705882353,
      "grad_norm": 1.798023491694529,
      "learning_rate": 0.00029999814885247906,
      "loss": 0.793,
      "step": 2017
    },
    {
      "epoch": 0.9496470588235294,
      "grad_norm": 0.9343120976056981,
      "learning_rate": 0.00029999792466558,
      "loss": 0.3594,
      "step": 2018
    },
    {
      "epoch": 0.9501176470588235,
      "grad_norm": 1.685911148673008,
      "learning_rate": 0.0002999976876681224,
      "loss": 1.0,
      "step": 2019
    },
    {
      "epoch": 0.9505882352941176,
      "grad_norm": 0.5652113401151938,
      "learning_rate": 0.00029999743786012656,
      "loss": 0.3477,
      "step": 2020
    },
    {
      "epoch": 0.9510588235294117,
      "grad_norm": 0.8581258038409852,
      "learning_rate": 0.0002999971752416138,
      "loss": 0.4922,
      "step": 2021
    },
    {
      "epoch": 0.9515294117647058,
      "grad_norm": 4.562995420713493,
      "learning_rate": 0.00029999689981260655,
      "loss": 0.5352,
      "step": 2022
    },
    {
      "epoch": 0.952,
      "grad_norm": 1.0808370522083683,
      "learning_rate": 0.0002999966115731283,
      "loss": 0.4121,
      "step": 2023
    },
    {
      "epoch": 0.9524705882352941,
      "grad_norm": 1.7812042814585733,
      "learning_rate": 0.00029999631052320374,
      "loss": 1.3203,
      "step": 2024
    },
    {
      "epoch": 0.9529411764705882,
      "grad_norm": 1.2944965477788124,
      "learning_rate": 0.0002999959966628585,
      "loss": 0.7266,
      "step": 2025
    },
    {
      "epoch": 0.9534117647058824,
      "grad_norm": 0.5069568511338856,
      "learning_rate": 0.0002999956699921194,
      "loss": 0.1426,
      "step": 2026
    },
    {
      "epoch": 0.9538823529411765,
      "grad_norm": 1.277243311601461,
      "learning_rate": 0.0002999953305110144,
      "loss": 0.5742,
      "step": 2027
    },
    {
      "epoch": 0.9543529411764706,
      "grad_norm": 1.2359650569495533,
      "learning_rate": 0.0002999949782195724,
      "loss": 0.7031,
      "step": 2028
    },
    {
      "epoch": 0.9548235294117647,
      "grad_norm": 0.7619933136042559,
      "learning_rate": 0.0002999946131178236,
      "loss": 0.418,
      "step": 2029
    },
    {
      "epoch": 0.9552941176470588,
      "grad_norm": 1.0428207196597705,
      "learning_rate": 0.0002999942352057991,
      "loss": 0.4199,
      "step": 2030
    },
    {
      "epoch": 0.955764705882353,
      "grad_norm": 0.4581199309245707,
      "learning_rate": 0.00029999384448353114,
      "loss": 0.1172,
      "step": 2031
    },
    {
      "epoch": 0.9562352941176471,
      "grad_norm": 1.0012363067500023,
      "learning_rate": 0.00029999344095105326,
      "loss": 0.6797,
      "step": 2032
    },
    {
      "epoch": 0.9567058823529412,
      "grad_norm": 0.7733467257264347,
      "learning_rate": 0.0002999930246083997,
      "loss": 0.3945,
      "step": 2033
    },
    {
      "epoch": 0.9571764705882353,
      "grad_norm": 1.243885583674872,
      "learning_rate": 0.0002999925954556062,
      "loss": 0.4785,
      "step": 2034
    },
    {
      "epoch": 0.9576470588235294,
      "grad_norm": 1.284347004216204,
      "learning_rate": 0.0002999921534927093,
      "loss": 0.9453,
      "step": 2035
    },
    {
      "epoch": 0.9581176470588235,
      "grad_norm": 0.8641550414505339,
      "learning_rate": 0.00029999169871974677,
      "loss": 0.4512,
      "step": 2036
    },
    {
      "epoch": 0.9585882352941176,
      "grad_norm": 0.8708679449458168,
      "learning_rate": 0.0002999912311367575,
      "loss": 0.3945,
      "step": 2037
    },
    {
      "epoch": 0.9590588235294117,
      "grad_norm": 1.7871122586779258,
      "learning_rate": 0.00029999075074378136,
      "loss": 1.2344,
      "step": 2038
    },
    {
      "epoch": 0.9595294117647059,
      "grad_norm": 0.5685735340005911,
      "learning_rate": 0.00029999025754085944,
      "loss": 0.2793,
      "step": 2039
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6228913590244544,
      "learning_rate": 0.00029998975152803383,
      "loss": 0.4121,
      "step": 2040
    },
    {
      "epoch": 0.9604705882352941,
      "grad_norm": 1.7315900011465917,
      "learning_rate": 0.0002999892327053477,
      "loss": 0.8008,
      "step": 2041
    },
    {
      "epoch": 0.9609411764705882,
      "grad_norm": 1.272599357532502,
      "learning_rate": 0.0002999887010728454,
      "loss": 1.25,
      "step": 2042
    },
    {
      "epoch": 0.9614117647058823,
      "grad_norm": 0.2121659037685555,
      "learning_rate": 0.0002999881566305724,
      "loss": 0.0223,
      "step": 2043
    },
    {
      "epoch": 0.9618823529411765,
      "grad_norm": 0.7123440403100898,
      "learning_rate": 0.0002999875993785751,
      "loss": 0.3926,
      "step": 2044
    },
    {
      "epoch": 0.9623529411764706,
      "grad_norm": 0.5350404570614108,
      "learning_rate": 0.0002999870293169011,
      "loss": 0.1934,
      "step": 2045
    },
    {
      "epoch": 0.9628235294117647,
      "grad_norm": 2.554032680901925,
      "learning_rate": 0.00029998644644559916,
      "loss": 1.4531,
      "step": 2046
    },
    {
      "epoch": 0.9632941176470589,
      "grad_norm": 1.174419267357845,
      "learning_rate": 0.000299985850764719,
      "loss": 0.5312,
      "step": 2047
    },
    {
      "epoch": 0.963764705882353,
      "grad_norm": 1.3160667626853617,
      "learning_rate": 0.00029998524227431146,
      "loss": 1.0312,
      "step": 2048
    },
    {
      "epoch": 0.9642352941176471,
      "grad_norm": 1.05008245530684,
      "learning_rate": 0.00029998462097442866,
      "loss": 1.0469,
      "step": 2049
    },
    {
      "epoch": 0.9647058823529412,
      "grad_norm": 1.2596601295696526,
      "learning_rate": 0.0002999839868651235,
      "loss": 0.707,
      "step": 2050
    },
    {
      "epoch": 0.9651764705882353,
      "grad_norm": 0.9429955601264314,
      "learning_rate": 0.0002999833399464502,
      "loss": 0.5078,
      "step": 2051
    },
    {
      "epoch": 0.9656470588235294,
      "grad_norm": 1.2192542116932252,
      "learning_rate": 0.000299982680218464,
      "loss": 0.9727,
      "step": 2052
    },
    {
      "epoch": 0.9661176470588235,
      "grad_norm": 0.5510327220312426,
      "learning_rate": 0.00029998200768122123,
      "loss": 0.3496,
      "step": 2053
    },
    {
      "epoch": 0.9665882352941176,
      "grad_norm": 1.2245635012252838,
      "learning_rate": 0.00029998132233477936,
      "loss": 1.0938,
      "step": 2054
    },
    {
      "epoch": 0.9670588235294117,
      "grad_norm": 1.1522496040839285,
      "learning_rate": 0.0002999806241791969,
      "loss": 0.9492,
      "step": 2055
    },
    {
      "epoch": 0.9675294117647059,
      "grad_norm": 0.8766319739879043,
      "learning_rate": 0.0002999799132145335,
      "loss": 0.8086,
      "step": 2056
    },
    {
      "epoch": 0.968,
      "grad_norm": 1.5371648001428553,
      "learning_rate": 0.00029997918944084986,
      "loss": 1.1562,
      "step": 2057
    },
    {
      "epoch": 0.9684705882352941,
      "grad_norm": 0.8501477298232,
      "learning_rate": 0.0002999784528582078,
      "loss": 0.7109,
      "step": 2058
    },
    {
      "epoch": 0.9689411764705882,
      "grad_norm": 1.1889299267900277,
      "learning_rate": 0.0002999777034666702,
      "loss": 0.793,
      "step": 2059
    },
    {
      "epoch": 0.9694117647058823,
      "grad_norm": 0.780414129673238,
      "learning_rate": 0.00029997694126630115,
      "loss": 0.4961,
      "step": 2060
    },
    {
      "epoch": 0.9698823529411764,
      "grad_norm": 1.1972211019543935,
      "learning_rate": 0.00029997616625716567,
      "loss": 1.0469,
      "step": 2061
    },
    {
      "epoch": 0.9703529411764706,
      "grad_norm": 1.0262975283139677,
      "learning_rate": 0.0002999753784393299,
      "loss": 0.6602,
      "step": 2062
    },
    {
      "epoch": 0.9708235294117648,
      "grad_norm": 1.4377054521643249,
      "learning_rate": 0.00029997457781286125,
      "loss": 0.9805,
      "step": 2063
    },
    {
      "epoch": 0.9712941176470589,
      "grad_norm": 1.0482574781045133,
      "learning_rate": 0.000299973764377828,
      "loss": 0.6484,
      "step": 2064
    },
    {
      "epoch": 0.971764705882353,
      "grad_norm": 0.657951895298403,
      "learning_rate": 0.00029997293813429965,
      "loss": 0.3613,
      "step": 2065
    },
    {
      "epoch": 0.9722352941176471,
      "grad_norm": 2.220141111672204,
      "learning_rate": 0.0002999720990823468,
      "loss": 1.625,
      "step": 2066
    },
    {
      "epoch": 0.9727058823529412,
      "grad_norm": 0.8937260203839222,
      "learning_rate": 0.0002999712472220411,
      "loss": 0.7266,
      "step": 2067
    },
    {
      "epoch": 0.9731764705882353,
      "grad_norm": 1.4376486745741275,
      "learning_rate": 0.0002999703825534552,
      "loss": 1.3438,
      "step": 2068
    },
    {
      "epoch": 0.9736470588235294,
      "grad_norm": 0.9200831162735563,
      "learning_rate": 0.0002999695050766631,
      "loss": 0.2773,
      "step": 2069
    },
    {
      "epoch": 0.9741176470588235,
      "grad_norm": 0.29044480297045167,
      "learning_rate": 0.0002999686147917396,
      "loss": 0.084,
      "step": 2070
    },
    {
      "epoch": 0.9745882352941176,
      "grad_norm": 1.8780842660203856,
      "learning_rate": 0.0002999677116987609,
      "loss": 0.9609,
      "step": 2071
    },
    {
      "epoch": 0.9750588235294118,
      "grad_norm": 1.3672551575498584,
      "learning_rate": 0.00029996679579780394,
      "loss": 0.9531,
      "step": 2072
    },
    {
      "epoch": 0.9755294117647059,
      "grad_norm": 1.3407425597849365,
      "learning_rate": 0.0002999658670889471,
      "loss": 0.793,
      "step": 2073
    },
    {
      "epoch": 0.976,
      "grad_norm": 1.6086641831526967,
      "learning_rate": 0.0002999649255722696,
      "loss": 1.1797,
      "step": 2074
    },
    {
      "epoch": 0.9764705882352941,
      "grad_norm": 0.7065282878393042,
      "learning_rate": 0.0002999639712478519,
      "loss": 0.4629,
      "step": 2075
    },
    {
      "epoch": 0.9769411764705882,
      "grad_norm": 0.5607024893246083,
      "learning_rate": 0.00029996300411577547,
      "loss": 0.4531,
      "step": 2076
    },
    {
      "epoch": 0.9774117647058823,
      "grad_norm": 0.8305291160267193,
      "learning_rate": 0.00029996202417612294,
      "loss": 0.4629,
      "step": 2077
    },
    {
      "epoch": 0.9778823529411764,
      "grad_norm": 1.1782132232315587,
      "learning_rate": 0.000299961031428978,
      "loss": 0.5781,
      "step": 2078
    },
    {
      "epoch": 0.9783529411764705,
      "grad_norm": 0.7372745612595075,
      "learning_rate": 0.00029996002587442537,
      "loss": 0.3242,
      "step": 2079
    },
    {
      "epoch": 0.9788235294117648,
      "grad_norm": 1.577957551135549,
      "learning_rate": 0.000299959007512551,
      "loss": 0.7266,
      "step": 2080
    },
    {
      "epoch": 0.9792941176470589,
      "grad_norm": 1.4372630255780108,
      "learning_rate": 0.00029995797634344186,
      "loss": 0.6367,
      "step": 2081
    },
    {
      "epoch": 0.979764705882353,
      "grad_norm": 1.1907820964876763,
      "learning_rate": 0.000299956932367186,
      "loss": 0.707,
      "step": 2082
    },
    {
      "epoch": 0.9802352941176471,
      "grad_norm": 0.5223965157136968,
      "learning_rate": 0.0002999558755838726,
      "loss": 0.1885,
      "step": 2083
    },
    {
      "epoch": 0.9807058823529412,
      "grad_norm": 1.846720308665745,
      "learning_rate": 0.00029995480599359186,
      "loss": 1.2344,
      "step": 2084
    },
    {
      "epoch": 0.9811764705882353,
      "grad_norm": 1.6632953172092189,
      "learning_rate": 0.00029995372359643515,
      "loss": 0.8398,
      "step": 2085
    },
    {
      "epoch": 0.9816470588235294,
      "grad_norm": 0.8523695912947089,
      "learning_rate": 0.00029995262839249497,
      "loss": 0.4512,
      "step": 2086
    },
    {
      "epoch": 0.9821176470588235,
      "grad_norm": 1.0751493330436055,
      "learning_rate": 0.0002999515203818648,
      "loss": 0.6289,
      "step": 2087
    },
    {
      "epoch": 0.9825882352941177,
      "grad_norm": 1.110252102742101,
      "learning_rate": 0.0002999503995646393,
      "loss": 0.5234,
      "step": 2088
    },
    {
      "epoch": 0.9830588235294118,
      "grad_norm": 0.7012666101699335,
      "learning_rate": 0.00029994926594091413,
      "loss": 0.4668,
      "step": 2089
    },
    {
      "epoch": 0.9835294117647059,
      "grad_norm": 0.806461150272455,
      "learning_rate": 0.00029994811951078616,
      "loss": 0.3398,
      "step": 2090
    },
    {
      "epoch": 0.984,
      "grad_norm": 1.53409154726614,
      "learning_rate": 0.0002999469602743533,
      "loss": 0.7969,
      "step": 2091
    },
    {
      "epoch": 0.9844705882352941,
      "grad_norm": 1.0635795721046937,
      "learning_rate": 0.0002999457882317146,
      "loss": 0.5859,
      "step": 2092
    },
    {
      "epoch": 0.9849411764705882,
      "grad_norm": 0.7361930140954956,
      "learning_rate": 0.00029994460338297,
      "loss": 0.4043,
      "step": 2093
    },
    {
      "epoch": 0.9854117647058823,
      "grad_norm": 0.6017395968135901,
      "learning_rate": 0.0002999434057282209,
      "loss": 0.2324,
      "step": 2094
    },
    {
      "epoch": 0.9858823529411764,
      "grad_norm": 1.3847634147284307,
      "learning_rate": 0.0002999421952675694,
      "loss": 0.9375,
      "step": 2095
    },
    {
      "epoch": 0.9863529411764705,
      "grad_norm": 0.7315194486528899,
      "learning_rate": 0.00029994097200111903,
      "loss": 0.498,
      "step": 2096
    },
    {
      "epoch": 0.9868235294117647,
      "grad_norm": 1.0177208626559229,
      "learning_rate": 0.0002999397359289741,
      "loss": 0.7891,
      "step": 2097
    },
    {
      "epoch": 0.9872941176470589,
      "grad_norm": 1.159775970318152,
      "learning_rate": 0.00029993848705124035,
      "loss": 0.5273,
      "step": 2098
    },
    {
      "epoch": 0.987764705882353,
      "grad_norm": 1.2339489305641373,
      "learning_rate": 0.0002999372253680244,
      "loss": 0.7344,
      "step": 2099
    },
    {
      "epoch": 0.9882352941176471,
      "grad_norm": 0.5527486787372567,
      "learning_rate": 0.0002999359508794339,
      "loss": 0.2949,
      "step": 2100
    },
    {
      "epoch": 0.9887058823529412,
      "grad_norm": 0.6547199020258242,
      "learning_rate": 0.0002999346635855778,
      "loss": 0.543,
      "step": 2101
    },
    {
      "epoch": 0.9891764705882353,
      "grad_norm": 0.7048620059341282,
      "learning_rate": 0.00029993336348656594,
      "loss": 0.2969,
      "step": 2102
    },
    {
      "epoch": 0.9896470588235294,
      "grad_norm": 0.9395908495613434,
      "learning_rate": 0.00029993205058250947,
      "loss": 0.4004,
      "step": 2103
    },
    {
      "epoch": 0.9901176470588235,
      "grad_norm": 1.8737827068969433,
      "learning_rate": 0.00029993072487352043,
      "loss": 1.2734,
      "step": 2104
    },
    {
      "epoch": 0.9905882352941177,
      "grad_norm": 1.0610518651443457,
      "learning_rate": 0.0002999293863597121,
      "loss": 0.4707,
      "step": 2105
    },
    {
      "epoch": 0.9910588235294118,
      "grad_norm": 0.5030323252171235,
      "learning_rate": 0.00029992803504119876,
      "loss": 0.291,
      "step": 2106
    },
    {
      "epoch": 0.9915294117647059,
      "grad_norm": 0.8671922250215823,
      "learning_rate": 0.0002999266709180959,
      "loss": 0.2949,
      "step": 2107
    },
    {
      "epoch": 0.992,
      "grad_norm": 0.9930876323316005,
      "learning_rate": 0.0002999252939905199,
      "loss": 0.5547,
      "step": 2108
    },
    {
      "epoch": 0.9924705882352941,
      "grad_norm": 1.3830066689260054,
      "learning_rate": 0.0002999239042585884,
      "loss": 0.8125,
      "step": 2109
    },
    {
      "epoch": 0.9929411764705882,
      "grad_norm": 2.115595291209749,
      "learning_rate": 0.0002999225017224201,
      "loss": 1.1797,
      "step": 2110
    },
    {
      "epoch": 0.9934117647058823,
      "grad_norm": 1.2643235663438561,
      "learning_rate": 0.0002999210863821349,
      "loss": 0.9453,
      "step": 2111
    },
    {
      "epoch": 0.9938823529411764,
      "grad_norm": 1.1046689062725477,
      "learning_rate": 0.0002999196582378535,
      "loss": 0.8125,
      "step": 2112
    },
    {
      "epoch": 0.9943529411764706,
      "grad_norm": 1.0239847728262053,
      "learning_rate": 0.0002999182172896979,
      "loss": 0.6367,
      "step": 2113
    },
    {
      "epoch": 0.9948235294117647,
      "grad_norm": 0.6361813626884262,
      "learning_rate": 0.0002999167635377912,
      "loss": 0.3164,
      "step": 2114
    },
    {
      "epoch": 0.9952941176470588,
      "grad_norm": 1.44824293137948,
      "learning_rate": 0.00029991529698225757,
      "loss": 0.8242,
      "step": 2115
    },
    {
      "epoch": 0.995764705882353,
      "grad_norm": 0.6207633232870053,
      "learning_rate": 0.0002999138176232223,
      "loss": 0.208,
      "step": 2116
    },
    {
      "epoch": 0.9962352941176471,
      "grad_norm": 1.2847952040196813,
      "learning_rate": 0.00029991232546081156,
      "loss": 0.7578,
      "step": 2117
    },
    {
      "epoch": 0.9967058823529412,
      "grad_norm": 0.27646490380342215,
      "learning_rate": 0.000299910820495153,
      "loss": 0.1455,
      "step": 2118
    },
    {
      "epoch": 0.9971764705882353,
      "grad_norm": 0.8084593209568853,
      "learning_rate": 0.00029990930272637506,
      "loss": 0.2578,
      "step": 2119
    },
    {
      "epoch": 0.9976470588235294,
      "grad_norm": 0.8754503347079469,
      "learning_rate": 0.00029990777215460737,
      "loss": 0.6094,
      "step": 2120
    },
    {
      "epoch": 0.9981176470588236,
      "grad_norm": 1.220733315731403,
      "learning_rate": 0.0002999062287799806,
      "loss": 0.8516,
      "step": 2121
    },
    {
      "epoch": 0.9985882352941177,
      "grad_norm": 1.2846051005625578,
      "learning_rate": 0.0002999046726026266,
      "loss": 1.0312,
      "step": 2122
    },
    {
      "epoch": 0.9990588235294118,
      "grad_norm": 0.8379202726414325,
      "learning_rate": 0.00029990310362267833,
      "loss": 0.7812,
      "step": 2123
    },
    {
      "epoch": 0.9995294117647059,
      "grad_norm": 0.9136682415624522,
      "learning_rate": 0.0002999015218402697,
      "loss": 0.832,
      "step": 2124
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.041172297210159,
      "learning_rate": 0.0002998999272555359,
      "loss": 0.9414,
      "step": 2125
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.6415559649467468,
      "eval_runtime": 44.1002,
      "eval_samples_per_second": 17.007,
      "eval_steps_per_second": 8.503,
      "step": 2125
    },
    {
      "epoch": 1.0004705882352942,
      "grad_norm": 1.1451693840772939,
      "learning_rate": 0.000299898319868613,
      "loss": 0.7305,
      "step": 2126
    },
    {
      "epoch": 1.0009411764705882,
      "grad_norm": 0.8508119525818632,
      "learning_rate": 0.00029989669967963835,
      "loss": 0.5977,
      "step": 2127
    },
    {
      "epoch": 1.0014117647058824,
      "grad_norm": 0.48424627177716295,
      "learning_rate": 0.00029989506668875033,
      "loss": 0.3379,
      "step": 2128
    },
    {
      "epoch": 1.0018823529411764,
      "grad_norm": 0.41539817084349756,
      "learning_rate": 0.00029989342089608835,
      "loss": 0.2021,
      "step": 2129
    },
    {
      "epoch": 1.0023529411764707,
      "grad_norm": 1.040098805790498,
      "learning_rate": 0.000299891762301793,
      "loss": 0.3887,
      "step": 2130
    },
    {
      "epoch": 1.0028235294117647,
      "grad_norm": 1.7416624917071288,
      "learning_rate": 0.00029989009090600595,
      "loss": 0.9062,
      "step": 2131
    },
    {
      "epoch": 1.003294117647059,
      "grad_norm": 0.8694520299045322,
      "learning_rate": 0.00029988840670887,
      "loss": 0.4258,
      "step": 2132
    },
    {
      "epoch": 1.003764705882353,
      "grad_norm": 1.2304885140961912,
      "learning_rate": 0.0002998867097105288,
      "loss": 0.9258,
      "step": 2133
    },
    {
      "epoch": 1.0042352941176471,
      "grad_norm": 1.307887250842866,
      "learning_rate": 0.00029988499991112747,
      "loss": 1.3047,
      "step": 2134
    },
    {
      "epoch": 1.0047058823529411,
      "grad_norm": 0.6712499087710271,
      "learning_rate": 0.000299883277310812,
      "loss": 0.3184,
      "step": 2135
    },
    {
      "epoch": 1.0051764705882353,
      "grad_norm": 1.2033283123276541,
      "learning_rate": 0.0002998815419097294,
      "loss": 0.6289,
      "step": 2136
    },
    {
      "epoch": 1.0056470588235293,
      "grad_norm": 1.579114319926738,
      "learning_rate": 0.0002998797937080279,
      "loss": 0.7344,
      "step": 2137
    },
    {
      "epoch": 1.0061176470588236,
      "grad_norm": 1.1704491845255676,
      "learning_rate": 0.000299878032705857,
      "loss": 0.9766,
      "step": 2138
    },
    {
      "epoch": 1.0065882352941176,
      "grad_norm": 1.2599869195626794,
      "learning_rate": 0.0002998762589033669,
      "loss": 0.832,
      "step": 2139
    },
    {
      "epoch": 1.0070588235294118,
      "grad_norm": 0.7254267820629493,
      "learning_rate": 0.00029987447230070917,
      "loss": 0.4102,
      "step": 2140
    },
    {
      "epoch": 1.0075294117647058,
      "grad_norm": 0.8065501976245463,
      "learning_rate": 0.00029987267289803635,
      "loss": 0.3223,
      "step": 2141
    },
    {
      "epoch": 1.008,
      "grad_norm": 0.965268521940052,
      "learning_rate": 0.00029987086069550216,
      "loss": 0.3105,
      "step": 2142
    },
    {
      "epoch": 1.008470588235294,
      "grad_norm": 1.492221767643024,
      "learning_rate": 0.0002998690356932613,
      "loss": 0.5781,
      "step": 2143
    },
    {
      "epoch": 1.0089411764705882,
      "grad_norm": 0.680204870789465,
      "learning_rate": 0.0002998671978914698,
      "loss": 0.2637,
      "step": 2144
    },
    {
      "epoch": 1.0094117647058825,
      "grad_norm": 1.430899449207437,
      "learning_rate": 0.00029986534729028444,
      "loss": 1.0156,
      "step": 2145
    },
    {
      "epoch": 1.0098823529411765,
      "grad_norm": 0.939578937264321,
      "learning_rate": 0.00029986348388986333,
      "loss": 0.6602,
      "step": 2146
    },
    {
      "epoch": 1.0103529411764707,
      "grad_norm": 0.435893517593319,
      "learning_rate": 0.00029986160769036565,
      "loss": 0.1318,
      "step": 2147
    },
    {
      "epoch": 1.0108235294117647,
      "grad_norm": 1.03030034827643,
      "learning_rate": 0.0002998597186919515,
      "loss": 0.5039,
      "step": 2148
    },
    {
      "epoch": 1.011294117647059,
      "grad_norm": 1.3095566097292781,
      "learning_rate": 0.0002998578168947824,
      "loss": 0.7578,
      "step": 2149
    },
    {
      "epoch": 1.011764705882353,
      "grad_norm": 1.2396269005255616,
      "learning_rate": 0.00029985590229902073,
      "loss": 0.6875,
      "step": 2150
    },
    {
      "epoch": 1.0122352941176471,
      "grad_norm": 1.1059803070034269,
      "learning_rate": 0.0002998539749048299,
      "loss": 0.8516,
      "step": 2151
    },
    {
      "epoch": 1.0127058823529411,
      "grad_norm": 0.5236451529005193,
      "learning_rate": 0.0002998520347123746,
      "loss": 0.1504,
      "step": 2152
    },
    {
      "epoch": 1.0131764705882353,
      "grad_norm": 1.4386550729105594,
      "learning_rate": 0.00029985008172182055,
      "loss": 0.8164,
      "step": 2153
    },
    {
      "epoch": 1.0136470588235293,
      "grad_norm": 1.1201261005743761,
      "learning_rate": 0.00029984811593333446,
      "loss": 0.9609,
      "step": 2154
    },
    {
      "epoch": 1.0141176470588236,
      "grad_norm": 0.8273486643878427,
      "learning_rate": 0.00029984613734708427,
      "loss": 0.4238,
      "step": 2155
    },
    {
      "epoch": 1.0145882352941176,
      "grad_norm": 0.6046841072044955,
      "learning_rate": 0.00029984414596323896,
      "loss": 0.2324,
      "step": 2156
    },
    {
      "epoch": 1.0150588235294118,
      "grad_norm": 1.3513960206689946,
      "learning_rate": 0.00029984214178196863,
      "loss": 0.8125,
      "step": 2157
    },
    {
      "epoch": 1.0155294117647058,
      "grad_norm": 1.328235005612072,
      "learning_rate": 0.00029984012480344444,
      "loss": 0.8828,
      "step": 2158
    },
    {
      "epoch": 1.016,
      "grad_norm": 0.40180274969113056,
      "learning_rate": 0.0002998380950278386,
      "loss": 0.0957,
      "step": 2159
    },
    {
      "epoch": 1.016470588235294,
      "grad_norm": 1.3038544880285137,
      "learning_rate": 0.00029983605245532453,
      "loss": 0.6406,
      "step": 2160
    },
    {
      "epoch": 1.0169411764705882,
      "grad_norm": 1.3701058985843138,
      "learning_rate": 0.0002998339970860766,
      "loss": 0.6992,
      "step": 2161
    },
    {
      "epoch": 1.0174117647058825,
      "grad_norm": 1.8288874000134483,
      "learning_rate": 0.00029983192892027037,
      "loss": 1.1719,
      "step": 2162
    },
    {
      "epoch": 1.0178823529411765,
      "grad_norm": 0.5562872871064832,
      "learning_rate": 0.0002998298479580825,
      "loss": 0.3926,
      "step": 2163
    },
    {
      "epoch": 1.0183529411764707,
      "grad_norm": 1.3474424379571226,
      "learning_rate": 0.0002998277541996907,
      "loss": 1.2188,
      "step": 2164
    },
    {
      "epoch": 1.0188235294117647,
      "grad_norm": 0.5880083136003766,
      "learning_rate": 0.0002998256476452738,
      "loss": 0.4043,
      "step": 2165
    },
    {
      "epoch": 1.019294117647059,
      "grad_norm": 0.3664462707881798,
      "learning_rate": 0.0002998235282950117,
      "loss": 0.1177,
      "step": 2166
    },
    {
      "epoch": 1.019764705882353,
      "grad_norm": 1.0185702181835454,
      "learning_rate": 0.0002998213961490854,
      "loss": 0.4531,
      "step": 2167
    },
    {
      "epoch": 1.0202352941176471,
      "grad_norm": 1.6076821036303723,
      "learning_rate": 0.00029981925120767705,
      "loss": 0.875,
      "step": 2168
    },
    {
      "epoch": 1.0207058823529411,
      "grad_norm": 1.0447334985337289,
      "learning_rate": 0.00029981709347096975,
      "loss": 0.7148,
      "step": 2169
    },
    {
      "epoch": 1.0211764705882354,
      "grad_norm": 0.6781826627636653,
      "learning_rate": 0.00029981492293914777,
      "loss": 0.377,
      "step": 2170
    },
    {
      "epoch": 1.0216470588235294,
      "grad_norm": 1.0000425081542645,
      "learning_rate": 0.0002998127396123966,
      "loss": 0.8047,
      "step": 2171
    },
    {
      "epoch": 1.0221176470588236,
      "grad_norm": 0.5368584676368385,
      "learning_rate": 0.00029981054349090264,
      "loss": 0.2676,
      "step": 2172
    },
    {
      "epoch": 1.0225882352941176,
      "grad_norm": 1.5357579060586002,
      "learning_rate": 0.0002998083345748534,
      "loss": 1.0703,
      "step": 2173
    },
    {
      "epoch": 1.0230588235294118,
      "grad_norm": 0.9742619330490886,
      "learning_rate": 0.0002998061128644376,
      "loss": 0.4805,
      "step": 2174
    },
    {
      "epoch": 1.0235294117647058,
      "grad_norm": 0.7144806019809049,
      "learning_rate": 0.00029980387835984494,
      "loss": 0.4043,
      "step": 2175
    },
    {
      "epoch": 1.024,
      "grad_norm": 1.6790268998699058,
      "learning_rate": 0.0002998016310612663,
      "loss": 1.0938,
      "step": 2176
    },
    {
      "epoch": 1.024470588235294,
      "grad_norm": 1.0840751318214608,
      "learning_rate": 0.0002997993709688937,
      "loss": 0.8242,
      "step": 2177
    },
    {
      "epoch": 1.0249411764705882,
      "grad_norm": 0.8499775715929478,
      "learning_rate": 0.0002997970980829199,
      "loss": 0.5898,
      "step": 2178
    },
    {
      "epoch": 1.0254117647058822,
      "grad_norm": 0.8584861926411164,
      "learning_rate": 0.0002997948124035393,
      "loss": 0.3711,
      "step": 2179
    },
    {
      "epoch": 1.0258823529411765,
      "grad_norm": 0.5658419330847851,
      "learning_rate": 0.00029979251393094685,
      "loss": 0.2471,
      "step": 2180
    },
    {
      "epoch": 1.0263529411764707,
      "grad_norm": 1.425557686349109,
      "learning_rate": 0.00029979020266533913,
      "loss": 0.7578,
      "step": 2181
    },
    {
      "epoch": 1.0268235294117647,
      "grad_norm": 1.510077763315445,
      "learning_rate": 0.0002997878786069133,
      "loss": 0.7578,
      "step": 2182
    },
    {
      "epoch": 1.027294117647059,
      "grad_norm": 1.029609280878026,
      "learning_rate": 0.0002997855417558679,
      "loss": 0.7383,
      "step": 2183
    },
    {
      "epoch": 1.027764705882353,
      "grad_norm": 2.7496794773165,
      "learning_rate": 0.0002997831921124026,
      "loss": 1.1953,
      "step": 2184
    },
    {
      "epoch": 1.0282352941176471,
      "grad_norm": 0.5845977630511099,
      "learning_rate": 0.000299780829676718,
      "loss": 0.25,
      "step": 2185
    },
    {
      "epoch": 1.0287058823529411,
      "grad_norm": 1.4700733033952584,
      "learning_rate": 0.00029977845444901586,
      "loss": 0.8086,
      "step": 2186
    },
    {
      "epoch": 1.0291764705882354,
      "grad_norm": 0.7320262496705034,
      "learning_rate": 0.0002997760664294991,
      "loss": 0.3223,
      "step": 2187
    },
    {
      "epoch": 1.0296470588235294,
      "grad_norm": 1.5068972608980429,
      "learning_rate": 0.00029977366561837154,
      "loss": 0.7812,
      "step": 2188
    },
    {
      "epoch": 1.0301176470588236,
      "grad_norm": 0.8983824163157588,
      "learning_rate": 0.0002997712520158383,
      "loss": 0.6055,
      "step": 2189
    },
    {
      "epoch": 1.0305882352941176,
      "grad_norm": 0.6712819407631212,
      "learning_rate": 0.00029976882562210563,
      "loss": 0.3828,
      "step": 2190
    },
    {
      "epoch": 1.0310588235294118,
      "grad_norm": 0.5756117090980619,
      "learning_rate": 0.00029976638643738055,
      "loss": 0.291,
      "step": 2191
    },
    {
      "epoch": 1.0315294117647058,
      "grad_norm": 1.057423114291906,
      "learning_rate": 0.0002997639344618714,
      "loss": 0.7891,
      "step": 2192
    },
    {
      "epoch": 1.032,
      "grad_norm": 1.2676638763697854,
      "learning_rate": 0.0002997614696957878,
      "loss": 0.7422,
      "step": 2193
    },
    {
      "epoch": 1.032470588235294,
      "grad_norm": 0.990118612243761,
      "learning_rate": 0.00029975899213934004,
      "loss": 0.4883,
      "step": 2194
    },
    {
      "epoch": 1.0329411764705883,
      "grad_norm": 0.6099166657609626,
      "learning_rate": 0.0002997565017927398,
      "loss": 0.252,
      "step": 2195
    },
    {
      "epoch": 1.0334117647058823,
      "grad_norm": 1.1491635966042701,
      "learning_rate": 0.00029975399865619977,
      "loss": 0.5625,
      "step": 2196
    },
    {
      "epoch": 1.0338823529411765,
      "grad_norm": 2.326573690940198,
      "learning_rate": 0.00029975148272993373,
      "loss": 1.0312,
      "step": 2197
    },
    {
      "epoch": 1.0343529411764707,
      "grad_norm": 1.0840478038187096,
      "learning_rate": 0.0002997489540141565,
      "loss": 0.4473,
      "step": 2198
    },
    {
      "epoch": 1.0348235294117647,
      "grad_norm": 0.7018478142055723,
      "learning_rate": 0.00029974641250908406,
      "loss": 0.3438,
      "step": 2199
    },
    {
      "epoch": 1.035294117647059,
      "grad_norm": 0.6476743472353678,
      "learning_rate": 0.0002997438582149335,
      "loss": 0.3008,
      "step": 2200
    },
    {
      "epoch": 1.035764705882353,
      "grad_norm": 1.0013953066528447,
      "learning_rate": 0.000299741291131923,
      "loss": 0.4297,
      "step": 2201
    },
    {
      "epoch": 1.0362352941176471,
      "grad_norm": 0.9053808827863238,
      "learning_rate": 0.00029973871126027173,
      "loss": 0.418,
      "step": 2202
    },
    {
      "epoch": 1.0367058823529411,
      "grad_norm": 1.5509377808761755,
      "learning_rate": 0.00029973611860020013,
      "loss": 0.7578,
      "step": 2203
    },
    {
      "epoch": 1.0371764705882354,
      "grad_norm": 1.070731377465039,
      "learning_rate": 0.0002997335131519294,
      "loss": 0.75,
      "step": 2204
    },
    {
      "epoch": 1.0376470588235294,
      "grad_norm": 0.8515784421102937,
      "learning_rate": 0.0002997308949156823,
      "loss": 0.6172,
      "step": 2205
    },
    {
      "epoch": 1.0381176470588236,
      "grad_norm": 1.0871140905643966,
      "learning_rate": 0.00029972826389168235,
      "loss": 0.7305,
      "step": 2206
    },
    {
      "epoch": 1.0385882352941176,
      "grad_norm": 1.6467296245350584,
      "learning_rate": 0.00029972562008015425,
      "loss": 1.1328,
      "step": 2207
    },
    {
      "epoch": 1.0390588235294118,
      "grad_norm": 0.7260418801304411,
      "learning_rate": 0.0002997229634813238,
      "loss": 0.4316,
      "step": 2208
    },
    {
      "epoch": 1.0395294117647058,
      "grad_norm": 1.0367604996698219,
      "learning_rate": 0.0002997202940954179,
      "loss": 0.4746,
      "step": 2209
    },
    {
      "epoch": 1.04,
      "grad_norm": 0.5734246796590837,
      "learning_rate": 0.00029971761192266447,
      "loss": 0.1738,
      "step": 2210
    },
    {
      "epoch": 1.040470588235294,
      "grad_norm": 1.128245973860106,
      "learning_rate": 0.0002997149169632926,
      "loss": 0.5859,
      "step": 2211
    },
    {
      "epoch": 1.0409411764705883,
      "grad_norm": 1.1123323583955311,
      "learning_rate": 0.0002997122092175325,
      "loss": 0.6719,
      "step": 2212
    },
    {
      "epoch": 1.0414117647058823,
      "grad_norm": 0.8391945997197743,
      "learning_rate": 0.00029970948868561547,
      "loss": 0.377,
      "step": 2213
    },
    {
      "epoch": 1.0418823529411765,
      "grad_norm": 1.0319638945551473,
      "learning_rate": 0.00029970675536777365,
      "loss": 0.8828,
      "step": 2214
    },
    {
      "epoch": 1.0423529411764705,
      "grad_norm": 1.2295271164298582,
      "learning_rate": 0.0002997040092642407,
      "loss": 0.7812,
      "step": 2215
    },
    {
      "epoch": 1.0428235294117647,
      "grad_norm": 0.5469874016962275,
      "learning_rate": 0.00029970125037525106,
      "loss": 0.2031,
      "step": 2216
    },
    {
      "epoch": 1.043294117647059,
      "grad_norm": 0.9071956183977703,
      "learning_rate": 0.00029969847870104033,
      "loss": 0.7969,
      "step": 2217
    },
    {
      "epoch": 1.043764705882353,
      "grad_norm": 0.880489444745452,
      "learning_rate": 0.00029969569424184527,
      "loss": 0.4395,
      "step": 2218
    },
    {
      "epoch": 1.0442352941176472,
      "grad_norm": 0.9182294513193785,
      "learning_rate": 0.0002996928969979037,
      "loss": 0.8281,
      "step": 2219
    },
    {
      "epoch": 1.0447058823529412,
      "grad_norm": 0.38864573994758034,
      "learning_rate": 0.00029969008696945445,
      "loss": 0.1426,
      "step": 2220
    },
    {
      "epoch": 1.0451764705882354,
      "grad_norm": 1.0592688782576474,
      "learning_rate": 0.00029968726415673753,
      "loss": 0.7891,
      "step": 2221
    },
    {
      "epoch": 1.0456470588235294,
      "grad_norm": 0.5527316607259622,
      "learning_rate": 0.00029968442855999406,
      "loss": 0.3086,
      "step": 2222
    },
    {
      "epoch": 1.0461176470588236,
      "grad_norm": 1.2506835659949578,
      "learning_rate": 0.0002996815801794662,
      "loss": 0.9648,
      "step": 2223
    },
    {
      "epoch": 1.0465882352941176,
      "grad_norm": 1.016303579814898,
      "learning_rate": 0.0002996787190153972,
      "loss": 1.0312,
      "step": 2224
    },
    {
      "epoch": 1.0470588235294118,
      "grad_norm": 1.0535084764386926,
      "learning_rate": 0.0002996758450680314,
      "loss": 0.5039,
      "step": 2225
    },
    {
      "epoch": 1.0475294117647058,
      "grad_norm": 0.5340419365107648,
      "learning_rate": 0.0002996729583376143,
      "loss": 0.3418,
      "step": 2226
    },
    {
      "epoch": 1.048,
      "grad_norm": 0.6121273320216362,
      "learning_rate": 0.00029967005882439244,
      "loss": 0.2852,
      "step": 2227
    },
    {
      "epoch": 1.048470588235294,
      "grad_norm": 0.8979612774138352,
      "learning_rate": 0.0002996671465286134,
      "loss": 0.4551,
      "step": 2228
    },
    {
      "epoch": 1.0489411764705883,
      "grad_norm": 0.6797762739040429,
      "learning_rate": 0.00029966422145052594,
      "loss": 0.4512,
      "step": 2229
    },
    {
      "epoch": 1.0494117647058823,
      "grad_norm": 0.558236148574722,
      "learning_rate": 0.00029966128359037986,
      "loss": 0.3223,
      "step": 2230
    },
    {
      "epoch": 1.0498823529411765,
      "grad_norm": 1.1816128357964923,
      "learning_rate": 0.0002996583329484261,
      "loss": 0.6289,
      "step": 2231
    },
    {
      "epoch": 1.0503529411764705,
      "grad_norm": 1.199336320064049,
      "learning_rate": 0.00029965536952491665,
      "loss": 0.7969,
      "step": 2232
    },
    {
      "epoch": 1.0508235294117647,
      "grad_norm": 1.0179179853824458,
      "learning_rate": 0.00029965239332010454,
      "loss": 0.7812,
      "step": 2233
    },
    {
      "epoch": 1.0512941176470587,
      "grad_norm": 1.0817261146879151,
      "learning_rate": 0.000299649404334244,
      "loss": 0.8906,
      "step": 2234
    },
    {
      "epoch": 1.051764705882353,
      "grad_norm": 0.7731157441104521,
      "learning_rate": 0.0002996464025675903,
      "loss": 0.4375,
      "step": 2235
    },
    {
      "epoch": 1.0522352941176472,
      "grad_norm": 1.0501937978740916,
      "learning_rate": 0.00029964338802039987,
      "loss": 0.7812,
      "step": 2236
    },
    {
      "epoch": 1.0527058823529412,
      "grad_norm": 0.7208505340543534,
      "learning_rate": 0.00029964036069293006,
      "loss": 0.5586,
      "step": 2237
    },
    {
      "epoch": 1.0531764705882354,
      "grad_norm": 0.6418269663841951,
      "learning_rate": 0.0002996373205854395,
      "loss": 0.3574,
      "step": 2238
    },
    {
      "epoch": 1.0536470588235294,
      "grad_norm": 1.4552852983791527,
      "learning_rate": 0.00029963426769818774,
      "loss": 0.9727,
      "step": 2239
    },
    {
      "epoch": 1.0541176470588236,
      "grad_norm": 1.0882721002119902,
      "learning_rate": 0.0002996312020314356,
      "loss": 0.4668,
      "step": 2240
    },
    {
      "epoch": 1.0545882352941176,
      "grad_norm": 1.0280913374801506,
      "learning_rate": 0.00029962812358544485,
      "loss": 0.5703,
      "step": 2241
    },
    {
      "epoch": 1.0550588235294118,
      "grad_norm": 0.861625769938223,
      "learning_rate": 0.0002996250323604785,
      "loss": 0.3906,
      "step": 2242
    },
    {
      "epoch": 1.0555294117647058,
      "grad_norm": 0.6034192564277264,
      "learning_rate": 0.00029962192835680037,
      "loss": 0.2598,
      "step": 2243
    },
    {
      "epoch": 1.056,
      "grad_norm": 0.822569352860452,
      "learning_rate": 0.0002996188115746757,
      "loss": 0.5586,
      "step": 2244
    },
    {
      "epoch": 1.056470588235294,
      "grad_norm": 0.7731881187154274,
      "learning_rate": 0.0002996156820143707,
      "loss": 0.3086,
      "step": 2245
    },
    {
      "epoch": 1.0569411764705883,
      "grad_norm": 1.4324106028544454,
      "learning_rate": 0.00029961253967615255,
      "loss": 1.2422,
      "step": 2246
    },
    {
      "epoch": 1.0574117647058823,
      "grad_norm": 1.1928035059080122,
      "learning_rate": 0.0002996093845602897,
      "loss": 0.8945,
      "step": 2247
    },
    {
      "epoch": 1.0578823529411765,
      "grad_norm": 0.7386555588755482,
      "learning_rate": 0.00029960621666705155,
      "loss": 0.5234,
      "step": 2248
    },
    {
      "epoch": 1.0583529411764705,
      "grad_norm": 1.3309455453434278,
      "learning_rate": 0.00029960303599670866,
      "loss": 1.1328,
      "step": 2249
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 0.9750792534791624,
      "learning_rate": 0.0002995998425495327,
      "loss": 0.7266,
      "step": 2250
    },
    {
      "epoch": 1.0592941176470587,
      "grad_norm": 1.3748459918828038,
      "learning_rate": 0.0002995966363257964,
      "loss": 0.9453,
      "step": 2251
    },
    {
      "epoch": 1.059764705882353,
      "grad_norm": 0.9254466901043957,
      "learning_rate": 0.0002995934173257736,
      "loss": 0.5312,
      "step": 2252
    },
    {
      "epoch": 1.0602352941176472,
      "grad_norm": 0.9058024280333888,
      "learning_rate": 0.0002995901855497393,
      "loss": 0.5078,
      "step": 2253
    },
    {
      "epoch": 1.0607058823529412,
      "grad_norm": 1.071970866271475,
      "learning_rate": 0.00029958694099796934,
      "loss": 0.7383,
      "step": 2254
    },
    {
      "epoch": 1.0611764705882354,
      "grad_norm": 0.9546286952040328,
      "learning_rate": 0.0002995836836707409,
      "loss": 0.6367,
      "step": 2255
    },
    {
      "epoch": 1.0616470588235294,
      "grad_norm": 1.1467898224436208,
      "learning_rate": 0.00029958041356833217,
      "loss": 0.9883,
      "step": 2256
    },
    {
      "epoch": 1.0621176470588236,
      "grad_norm": 1.3929252307788254,
      "learning_rate": 0.00029957713069102244,
      "loss": 0.8438,
      "step": 2257
    },
    {
      "epoch": 1.0625882352941176,
      "grad_norm": 0.6781728644050676,
      "learning_rate": 0.00029957383503909204,
      "loss": 0.2793,
      "step": 2258
    },
    {
      "epoch": 1.0630588235294118,
      "grad_norm": 0.3528833902373964,
      "learning_rate": 0.00029957052661282256,
      "loss": 0.1191,
      "step": 2259
    },
    {
      "epoch": 1.0635294117647058,
      "grad_norm": 1.1415582546258027,
      "learning_rate": 0.0002995672054124964,
      "loss": 0.9297,
      "step": 2260
    },
    {
      "epoch": 1.064,
      "grad_norm": 2.028896016489529,
      "learning_rate": 0.0002995638714383973,
      "loss": 1.0312,
      "step": 2261
    },
    {
      "epoch": 1.064470588235294,
      "grad_norm": 0.857631335016029,
      "learning_rate": 0.00029956052469080994,
      "loss": 0.8047,
      "step": 2262
    },
    {
      "epoch": 1.0649411764705883,
      "grad_norm": 0.6877402993841945,
      "learning_rate": 0.0002995571651700202,
      "loss": 0.3809,
      "step": 2263
    },
    {
      "epoch": 1.0654117647058823,
      "grad_norm": 0.7790192744533485,
      "learning_rate": 0.000299553792876315,
      "loss": 0.5156,
      "step": 2264
    },
    {
      "epoch": 1.0658823529411765,
      "grad_norm": 0.7243809716153041,
      "learning_rate": 0.0002995504078099823,
      "loss": 0.3438,
      "step": 2265
    },
    {
      "epoch": 1.0663529411764705,
      "grad_norm": 0.9882555519817071,
      "learning_rate": 0.00029954700997131126,
      "loss": 0.5156,
      "step": 2266
    },
    {
      "epoch": 1.0668235294117647,
      "grad_norm": 0.6235412784461537,
      "learning_rate": 0.000299543599360592,
      "loss": 0.2598,
      "step": 2267
    },
    {
      "epoch": 1.0672941176470587,
      "grad_norm": 0.8996618058125393,
      "learning_rate": 0.0002995401759781159,
      "loss": 0.3887,
      "step": 2268
    },
    {
      "epoch": 1.067764705882353,
      "grad_norm": 0.9106573379777336,
      "learning_rate": 0.00029953673982417527,
      "loss": 0.8281,
      "step": 2269
    },
    {
      "epoch": 1.0682352941176472,
      "grad_norm": 0.6069511734979836,
      "learning_rate": 0.00029953329089906357,
      "loss": 0.3262,
      "step": 2270
    },
    {
      "epoch": 1.0687058823529412,
      "grad_norm": 0.6918101454242442,
      "learning_rate": 0.0002995298292030754,
      "loss": 0.3809,
      "step": 2271
    },
    {
      "epoch": 1.0691764705882354,
      "grad_norm": 0.8953693794123693,
      "learning_rate": 0.00029952635473650633,
      "loss": 0.8594,
      "step": 2272
    },
    {
      "epoch": 1.0696470588235294,
      "grad_norm": 1.6246283317405492,
      "learning_rate": 0.00029952286749965315,
      "loss": 1.0547,
      "step": 2273
    },
    {
      "epoch": 1.0701176470588236,
      "grad_norm": 1.1809375503981956,
      "learning_rate": 0.0002995193674928137,
      "loss": 0.8125,
      "step": 2274
    },
    {
      "epoch": 1.0705882352941176,
      "grad_norm": 0.8706224907229168,
      "learning_rate": 0.0002995158547162869,
      "loss": 0.5703,
      "step": 2275
    },
    {
      "epoch": 1.0710588235294118,
      "grad_norm": 1.4725377888090274,
      "learning_rate": 0.00029951232917037264,
      "loss": 0.9258,
      "step": 2276
    },
    {
      "epoch": 1.0715294117647058,
      "grad_norm": 1.3985529396153071,
      "learning_rate": 0.0002995087908553722,
      "loss": 0.7031,
      "step": 2277
    },
    {
      "epoch": 1.072,
      "grad_norm": 0.661098903339752,
      "learning_rate": 0.00029950523977158766,
      "loss": 0.3867,
      "step": 2278
    },
    {
      "epoch": 1.072470588235294,
      "grad_norm": 2.122200697800015,
      "learning_rate": 0.00029950167591932233,
      "loss": 1.8203,
      "step": 2279
    },
    {
      "epoch": 1.0729411764705883,
      "grad_norm": 0.9835709557235969,
      "learning_rate": 0.0002994980992988806,
      "loss": 0.9453,
      "step": 2280
    },
    {
      "epoch": 1.0734117647058823,
      "grad_norm": 0.6308382593790706,
      "learning_rate": 0.0002994945099105679,
      "loss": 0.5703,
      "step": 2281
    },
    {
      "epoch": 1.0738823529411765,
      "grad_norm": 0.47279704711706555,
      "learning_rate": 0.00029949090775469077,
      "loss": 0.1445,
      "step": 2282
    },
    {
      "epoch": 1.0743529411764705,
      "grad_norm": 1.2966374292348417,
      "learning_rate": 0.00029948729283155683,
      "loss": 0.9023,
      "step": 2283
    },
    {
      "epoch": 1.0748235294117647,
      "grad_norm": 0.9493398860508837,
      "learning_rate": 0.0002994836651414749,
      "loss": 0.5664,
      "step": 2284
    },
    {
      "epoch": 1.0752941176470587,
      "grad_norm": 0.9339868969641096,
      "learning_rate": 0.00029948002468475475,
      "loss": 0.6797,
      "step": 2285
    },
    {
      "epoch": 1.075764705882353,
      "grad_norm": 0.6135388391101992,
      "learning_rate": 0.0002994763714617073,
      "loss": 0.4922,
      "step": 2286
    },
    {
      "epoch": 1.076235294117647,
      "grad_norm": 1.250774435655936,
      "learning_rate": 0.00029947270547264456,
      "loss": 1.0234,
      "step": 2287
    },
    {
      "epoch": 1.0767058823529412,
      "grad_norm": 1.2760194728493695,
      "learning_rate": 0.00029946902671787955,
      "loss": 1.1484,
      "step": 2288
    },
    {
      "epoch": 1.0771764705882352,
      "grad_norm": 0.5649037250439675,
      "learning_rate": 0.00029946533519772657,
      "loss": 0.1973,
      "step": 2289
    },
    {
      "epoch": 1.0776470588235294,
      "grad_norm": 1.031526885094723,
      "learning_rate": 0.0002994616309125008,
      "loss": 0.6836,
      "step": 2290
    },
    {
      "epoch": 1.0781176470588236,
      "grad_norm": 0.37140166789240364,
      "learning_rate": 0.0002994579138625187,
      "loss": 0.1035,
      "step": 2291
    },
    {
      "epoch": 1.0785882352941176,
      "grad_norm": 1.084102362192583,
      "learning_rate": 0.00029945418404809767,
      "loss": 0.6797,
      "step": 2292
    },
    {
      "epoch": 1.0790588235294118,
      "grad_norm": 0.5364236907950947,
      "learning_rate": 0.0002994504414695562,
      "loss": 0.2207,
      "step": 2293
    },
    {
      "epoch": 1.0795294117647058,
      "grad_norm": 1.03060631199911,
      "learning_rate": 0.000299446686127214,
      "loss": 0.7109,
      "step": 2294
    },
    {
      "epoch": 1.08,
      "grad_norm": 1.1861039184176696,
      "learning_rate": 0.00029944291802139184,
      "loss": 0.6562,
      "step": 2295
    },
    {
      "epoch": 1.080470588235294,
      "grad_norm": 0.6345297167211073,
      "learning_rate": 0.00029943913715241137,
      "loss": 0.4043,
      "step": 2296
    },
    {
      "epoch": 1.0809411764705883,
      "grad_norm": 0.9671243059916581,
      "learning_rate": 0.0002994353435205956,
      "loss": 0.7227,
      "step": 2297
    },
    {
      "epoch": 1.0814117647058823,
      "grad_norm": 0.8481455691911628,
      "learning_rate": 0.00029943153712626855,
      "loss": 0.4902,
      "step": 2298
    },
    {
      "epoch": 1.0818823529411765,
      "grad_norm": 0.6561861497451956,
      "learning_rate": 0.0002994277179697553,
      "loss": 0.373,
      "step": 2299
    },
    {
      "epoch": 1.0823529411764705,
      "grad_norm": 1.2136560980999807,
      "learning_rate": 0.000299423886051382,
      "loss": 1.0391,
      "step": 2300
    },
    {
      "epoch": 1.0828235294117647,
      "grad_norm": 0.8632152602260921,
      "learning_rate": 0.00029942004137147586,
      "loss": 0.5469,
      "step": 2301
    },
    {
      "epoch": 1.0832941176470587,
      "grad_norm": 0.9441936250471642,
      "learning_rate": 0.0002994161839303653,
      "loss": 0.6133,
      "step": 2302
    },
    {
      "epoch": 1.083764705882353,
      "grad_norm": 0.5764959841301478,
      "learning_rate": 0.00029941231372837974,
      "loss": 0.3105,
      "step": 2303
    },
    {
      "epoch": 1.084235294117647,
      "grad_norm": 0.7143413773483748,
      "learning_rate": 0.0002994084307658497,
      "loss": 0.4062,
      "step": 2304
    },
    {
      "epoch": 1.0847058823529412,
      "grad_norm": 1.4498656561712016,
      "learning_rate": 0.0002994045350431069,
      "loss": 0.8281,
      "step": 2305
    },
    {
      "epoch": 1.0851764705882352,
      "grad_norm": 0.6218928959024943,
      "learning_rate": 0.00029940062656048396,
      "loss": 0.2852,
      "step": 2306
    },
    {
      "epoch": 1.0856470588235294,
      "grad_norm": 0.868126577889834,
      "learning_rate": 0.0002993967053183147,
      "loss": 0.707,
      "step": 2307
    },
    {
      "epoch": 1.0861176470588236,
      "grad_norm": 0.8825751807321968,
      "learning_rate": 0.000299392771316934,
      "loss": 0.5391,
      "step": 2308
    },
    {
      "epoch": 1.0865882352941176,
      "grad_norm": 1.3459694077438162,
      "learning_rate": 0.00029938882455667786,
      "loss": 1.0469,
      "step": 2309
    },
    {
      "epoch": 1.0870588235294119,
      "grad_norm": 1.0087119279208623,
      "learning_rate": 0.00029938486503788337,
      "loss": 0.4434,
      "step": 2310
    },
    {
      "epoch": 1.0875294117647059,
      "grad_norm": 1.2961885866695033,
      "learning_rate": 0.0002993808927608887,
      "loss": 0.8945,
      "step": 2311
    },
    {
      "epoch": 1.088,
      "grad_norm": 1.8577006612441749,
      "learning_rate": 0.000299376907726033,
      "loss": 1.4688,
      "step": 2312
    },
    {
      "epoch": 1.088470588235294,
      "grad_norm": 0.823691821779601,
      "learning_rate": 0.0002993729099336568,
      "loss": 0.4297,
      "step": 2313
    },
    {
      "epoch": 1.0889411764705883,
      "grad_norm": 1.099417990257321,
      "learning_rate": 0.00029936889938410137,
      "loss": 1.3203,
      "step": 2314
    },
    {
      "epoch": 1.0894117647058823,
      "grad_norm": 1.979163944750431,
      "learning_rate": 0.0002993648760777092,
      "loss": 1.2422,
      "step": 2315
    },
    {
      "epoch": 1.0898823529411765,
      "grad_norm": 2.048180815559026,
      "learning_rate": 0.0002993608400148241,
      "loss": 1.4297,
      "step": 2316
    },
    {
      "epoch": 1.0903529411764705,
      "grad_norm": 0.563040867069171,
      "learning_rate": 0.00029935679119579066,
      "loss": 0.418,
      "step": 2317
    },
    {
      "epoch": 1.0908235294117647,
      "grad_norm": 1.3743947894199342,
      "learning_rate": 0.0002993527296209546,
      "loss": 1.3359,
      "step": 2318
    },
    {
      "epoch": 1.0912941176470587,
      "grad_norm": 1.0804537371890333,
      "learning_rate": 0.0002993486552906629,
      "loss": 0.4844,
      "step": 2319
    },
    {
      "epoch": 1.091764705882353,
      "grad_norm": 0.5622084903107715,
      "learning_rate": 0.00029934456820526344,
      "loss": 0.293,
      "step": 2320
    },
    {
      "epoch": 1.092235294117647,
      "grad_norm": 0.9045441535489116,
      "learning_rate": 0.00029934046836510537,
      "loss": 0.4121,
      "step": 2321
    },
    {
      "epoch": 1.0927058823529412,
      "grad_norm": 1.025966867494636,
      "learning_rate": 0.00029933635577053877,
      "loss": 0.457,
      "step": 2322
    },
    {
      "epoch": 1.0931764705882352,
      "grad_norm": 0.6510259890961,
      "learning_rate": 0.00029933223042191487,
      "loss": 0.2041,
      "step": 2323
    },
    {
      "epoch": 1.0936470588235294,
      "grad_norm": 0.7529218091901518,
      "learning_rate": 0.000299328092319586,
      "loss": 0.4043,
      "step": 2324
    },
    {
      "epoch": 1.0941176470588236,
      "grad_norm": 1.212005362564546,
      "learning_rate": 0.0002993239414639057,
      "loss": 0.8125,
      "step": 2325
    },
    {
      "epoch": 1.0945882352941176,
      "grad_norm": 1.3668956802468668,
      "learning_rate": 0.0002993197778552283,
      "loss": 1.0547,
      "step": 2326
    },
    {
      "epoch": 1.0950588235294119,
      "grad_norm": 0.5512009598110116,
      "learning_rate": 0.0002993156014939095,
      "loss": 0.375,
      "step": 2327
    },
    {
      "epoch": 1.0955294117647059,
      "grad_norm": 1.2318515507981265,
      "learning_rate": 0.00029931141238030587,
      "loss": 0.6641,
      "step": 2328
    },
    {
      "epoch": 1.096,
      "grad_norm": 0.7313737107912864,
      "learning_rate": 0.00029930721051477533,
      "loss": 0.248,
      "step": 2329
    },
    {
      "epoch": 1.096470588235294,
      "grad_norm": 1.0593685041550867,
      "learning_rate": 0.0002993029958976766,
      "loss": 0.7266,
      "step": 2330
    },
    {
      "epoch": 1.0969411764705883,
      "grad_norm": 0.6720259708084707,
      "learning_rate": 0.00029929876852936973,
      "loss": 0.3887,
      "step": 2331
    },
    {
      "epoch": 1.0974117647058823,
      "grad_norm": 0.9571138963000427,
      "learning_rate": 0.00029929452841021573,
      "loss": 0.7969,
      "step": 2332
    },
    {
      "epoch": 1.0978823529411765,
      "grad_norm": 0.5811538401645405,
      "learning_rate": 0.00029929027554057675,
      "loss": 0.375,
      "step": 2333
    },
    {
      "epoch": 1.0983529411764705,
      "grad_norm": 0.7694981348186709,
      "learning_rate": 0.0002992860099208159,
      "loss": 0.459,
      "step": 2334
    },
    {
      "epoch": 1.0988235294117648,
      "grad_norm": 0.6322817033644857,
      "learning_rate": 0.0002992817315512976,
      "loss": 0.2324,
      "step": 2335
    },
    {
      "epoch": 1.0992941176470588,
      "grad_norm": 0.9770182514414005,
      "learning_rate": 0.0002992774404323872,
      "loss": 0.5352,
      "step": 2336
    },
    {
      "epoch": 1.099764705882353,
      "grad_norm": 0.7874771542157177,
      "learning_rate": 0.0002992731365644512,
      "loss": 0.3613,
      "step": 2337
    },
    {
      "epoch": 1.100235294117647,
      "grad_norm": 0.8146629009258072,
      "learning_rate": 0.0002992688199478571,
      "loss": 0.3398,
      "step": 2338
    },
    {
      "epoch": 1.1007058823529412,
      "grad_norm": 0.6540310662846377,
      "learning_rate": 0.00029926449058297364,
      "loss": 0.4629,
      "step": 2339
    },
    {
      "epoch": 1.1011764705882352,
      "grad_norm": 0.7269102229512869,
      "learning_rate": 0.0002992601484701706,
      "loss": 0.377,
      "step": 2340
    },
    {
      "epoch": 1.1016470588235294,
      "grad_norm": 1.1148325246630915,
      "learning_rate": 0.0002992557936098187,
      "loss": 0.6367,
      "step": 2341
    },
    {
      "epoch": 1.1021176470588236,
      "grad_norm": 0.7663627296757505,
      "learning_rate": 0.00029925142600229,
      "loss": 0.4961,
      "step": 2342
    },
    {
      "epoch": 1.1025882352941176,
      "grad_norm": 0.6885343074509696,
      "learning_rate": 0.00029924704564795735,
      "loss": 0.2773,
      "step": 2343
    },
    {
      "epoch": 1.1030588235294116,
      "grad_norm": 0.8684136947690744,
      "learning_rate": 0.000299242652547195,
      "loss": 0.3867,
      "step": 2344
    },
    {
      "epoch": 1.1035294117647059,
      "grad_norm": 0.8532770976164762,
      "learning_rate": 0.0002992382467003781,
      "loss": 0.6367,
      "step": 2345
    },
    {
      "epoch": 1.104,
      "grad_norm": 0.5595951349804645,
      "learning_rate": 0.00029923382810788287,
      "loss": 0.2734,
      "step": 2346
    },
    {
      "epoch": 1.104470588235294,
      "grad_norm": 0.6642054402998699,
      "learning_rate": 0.0002992293967700868,
      "loss": 0.3828,
      "step": 2347
    },
    {
      "epoch": 1.1049411764705883,
      "grad_norm": 0.9803471673386224,
      "learning_rate": 0.00029922495268736824,
      "loss": 0.7461,
      "step": 2348
    },
    {
      "epoch": 1.1054117647058823,
      "grad_norm": 0.7843592577995755,
      "learning_rate": 0.0002992204958601068,
      "loss": 0.793,
      "step": 2349
    },
    {
      "epoch": 1.1058823529411765,
      "grad_norm": 0.9636660420097797,
      "learning_rate": 0.0002992160262886831,
      "loss": 0.8008,
      "step": 2350
    },
    {
      "epoch": 1.1063529411764705,
      "grad_norm": 1.1157655070544479,
      "learning_rate": 0.0002992115439734788,
      "loss": 0.8867,
      "step": 2351
    },
    {
      "epoch": 1.1068235294117648,
      "grad_norm": 0.5367570391066809,
      "learning_rate": 0.0002992070489148768,
      "loss": 0.207,
      "step": 2352
    },
    {
      "epoch": 1.1072941176470588,
      "grad_norm": 1.4086312607026517,
      "learning_rate": 0.000299202541113261,
      "loss": 0.9648,
      "step": 2353
    },
    {
      "epoch": 1.107764705882353,
      "grad_norm": 0.8833378017391281,
      "learning_rate": 0.0002991980205690163,
      "loss": 0.7734,
      "step": 2354
    },
    {
      "epoch": 1.108235294117647,
      "grad_norm": 0.5779984642697173,
      "learning_rate": 0.0002991934872825289,
      "loss": 0.3125,
      "step": 2355
    },
    {
      "epoch": 1.1087058823529412,
      "grad_norm": 0.8249179326207874,
      "learning_rate": 0.0002991889412541858,
      "loss": 0.5273,
      "step": 2356
    },
    {
      "epoch": 1.1091764705882352,
      "grad_norm": 1.04191422086701,
      "learning_rate": 0.0002991843824843754,
      "loss": 0.7617,
      "step": 2357
    },
    {
      "epoch": 1.1096470588235294,
      "grad_norm": 1.0729580575630286,
      "learning_rate": 0.000299179810973487,
      "loss": 0.7695,
      "step": 2358
    },
    {
      "epoch": 1.1101176470588234,
      "grad_norm": 1.2427786930285618,
      "learning_rate": 0.000299175226721911,
      "loss": 0.6875,
      "step": 2359
    },
    {
      "epoch": 1.1105882352941177,
      "grad_norm": 1.582126917334028,
      "learning_rate": 0.00029917062973003897,
      "loss": 0.9336,
      "step": 2360
    },
    {
      "epoch": 1.1110588235294117,
      "grad_norm": 1.2436380000608451,
      "learning_rate": 0.0002991660199982634,
      "loss": 0.582,
      "step": 2361
    },
    {
      "epoch": 1.1115294117647059,
      "grad_norm": 1.7260733471486451,
      "learning_rate": 0.0002991613975269782,
      "loss": 1.125,
      "step": 2362
    },
    {
      "epoch": 1.112,
      "grad_norm": 1.801217964963359,
      "learning_rate": 0.0002991567623165779,
      "loss": 1.6328,
      "step": 2363
    },
    {
      "epoch": 1.112470588235294,
      "grad_norm": 1.1758261919487616,
      "learning_rate": 0.00029915211436745855,
      "loss": 0.7891,
      "step": 2364
    },
    {
      "epoch": 1.1129411764705883,
      "grad_norm": 0.9352754460180425,
      "learning_rate": 0.00029914745368001704,
      "loss": 0.4922,
      "step": 2365
    },
    {
      "epoch": 1.1134117647058823,
      "grad_norm": 1.126940636404244,
      "learning_rate": 0.0002991427802546514,
      "loss": 0.75,
      "step": 2366
    },
    {
      "epoch": 1.1138823529411765,
      "grad_norm": 1.0083088591559157,
      "learning_rate": 0.0002991380940917608,
      "loss": 0.8555,
      "step": 2367
    },
    {
      "epoch": 1.1143529411764705,
      "grad_norm": 1.2522787462805147,
      "learning_rate": 0.0002991333951917454,
      "loss": 1.0,
      "step": 2368
    },
    {
      "epoch": 1.1148235294117648,
      "grad_norm": 0.8439743029105201,
      "learning_rate": 0.00029912868355500666,
      "loss": 0.5703,
      "step": 2369
    },
    {
      "epoch": 1.1152941176470588,
      "grad_norm": 1.4145134392276069,
      "learning_rate": 0.0002991239591819468,
      "loss": 0.5625,
      "step": 2370
    },
    {
      "epoch": 1.115764705882353,
      "grad_norm": 0.6915977665988902,
      "learning_rate": 0.00029911922207296937,
      "loss": 0.2891,
      "step": 2371
    },
    {
      "epoch": 1.116235294117647,
      "grad_norm": 2.292615732542327,
      "learning_rate": 0.00029911447222847893,
      "loss": 1.2812,
      "step": 2372
    },
    {
      "epoch": 1.1167058823529412,
      "grad_norm": 1.141438193546743,
      "learning_rate": 0.00029910970964888126,
      "loss": 0.7305,
      "step": 2373
    },
    {
      "epoch": 1.1171764705882352,
      "grad_norm": 1.0243533449243076,
      "learning_rate": 0.0002991049343345829,
      "loss": 0.6875,
      "step": 2374
    },
    {
      "epoch": 1.1176470588235294,
      "grad_norm": 0.6125154030672609,
      "learning_rate": 0.00029910014628599184,
      "loss": 0.4219,
      "step": 2375
    },
    {
      "epoch": 1.1181176470588234,
      "grad_norm": 0.7592200267383443,
      "learning_rate": 0.00029909534550351695,
      "loss": 0.3281,
      "step": 2376
    },
    {
      "epoch": 1.1185882352941177,
      "grad_norm": 1.2205986112163214,
      "learning_rate": 0.00029909053198756827,
      "loss": 0.7695,
      "step": 2377
    },
    {
      "epoch": 1.1190588235294117,
      "grad_norm": 1.102148608802305,
      "learning_rate": 0.00029908570573855683,
      "loss": 0.8242,
      "step": 2378
    },
    {
      "epoch": 1.1195294117647059,
      "grad_norm": 1.6576635113664404,
      "learning_rate": 0.0002990808667568949,
      "loss": 1.1328,
      "step": 2379
    },
    {
      "epoch": 1.12,
      "grad_norm": 1.0845264491075324,
      "learning_rate": 0.0002990760150429957,
      "loss": 0.6328,
      "step": 2380
    },
    {
      "epoch": 1.120470588235294,
      "grad_norm": 1.280734100055955,
      "learning_rate": 0.0002990711505972736,
      "loss": 0.9844,
      "step": 2381
    },
    {
      "epoch": 1.1209411764705883,
      "grad_norm": 0.7446659007038725,
      "learning_rate": 0.00029906627342014407,
      "loss": 0.3594,
      "step": 2382
    },
    {
      "epoch": 1.1214117647058823,
      "grad_norm": 0.8794506617248612,
      "learning_rate": 0.00029906138351202356,
      "loss": 0.7266,
      "step": 2383
    },
    {
      "epoch": 1.1218823529411766,
      "grad_norm": 3.026414151816613,
      "learning_rate": 0.0002990564808733298,
      "loss": 1.4375,
      "step": 2384
    },
    {
      "epoch": 1.1223529411764706,
      "grad_norm": 1.2484686066992494,
      "learning_rate": 0.00029905156550448145,
      "loss": 0.7695,
      "step": 2385
    },
    {
      "epoch": 1.1228235294117648,
      "grad_norm": 0.9636202894001088,
      "learning_rate": 0.00029904663740589827,
      "loss": 0.9609,
      "step": 2386
    },
    {
      "epoch": 1.1232941176470588,
      "grad_norm": 0.5955441926816564,
      "learning_rate": 0.0002990416965780012,
      "loss": 0.4668,
      "step": 2387
    },
    {
      "epoch": 1.123764705882353,
      "grad_norm": 0.07362604413812945,
      "learning_rate": 0.0002990367430212123,
      "loss": 0.0173,
      "step": 2388
    },
    {
      "epoch": 1.124235294117647,
      "grad_norm": 0.4645997612820491,
      "learning_rate": 0.0002990317767359544,
      "loss": 0.0864,
      "step": 2389
    },
    {
      "epoch": 1.1247058823529412,
      "grad_norm": 1.7516959974765804,
      "learning_rate": 0.0002990267977226518,
      "loss": 0.7773,
      "step": 2390
    },
    {
      "epoch": 1.1251764705882352,
      "grad_norm": 1.4281944915504579,
      "learning_rate": 0.00029902180598172977,
      "loss": 0.6055,
      "step": 2391
    },
    {
      "epoch": 1.1256470588235294,
      "grad_norm": 1.063316612304659,
      "learning_rate": 0.00029901680151361447,
      "loss": 0.5352,
      "step": 2392
    },
    {
      "epoch": 1.1261176470588234,
      "grad_norm": 0.2521646884693583,
      "learning_rate": 0.00029901178431873346,
      "loss": 0.0486,
      "step": 2393
    },
    {
      "epoch": 1.1265882352941177,
      "grad_norm": 0.8128112194280062,
      "learning_rate": 0.0002990067543975151,
      "loss": 0.2695,
      "step": 2394
    },
    {
      "epoch": 1.1270588235294117,
      "grad_norm": 0.806186376451435,
      "learning_rate": 0.0002990017117503891,
      "loss": 0.457,
      "step": 2395
    },
    {
      "epoch": 1.127529411764706,
      "grad_norm": 0.8017060636779506,
      "learning_rate": 0.0002989966563777861,
      "loss": 0.6523,
      "step": 2396
    },
    {
      "epoch": 1.1280000000000001,
      "grad_norm": 1.0416231686582649,
      "learning_rate": 0.00029899158828013774,
      "loss": 0.6094,
      "step": 2397
    },
    {
      "epoch": 1.1284705882352941,
      "grad_norm": 0.44843221171365605,
      "learning_rate": 0.000298986507457877,
      "loss": 0.1641,
      "step": 2398
    },
    {
      "epoch": 1.1289411764705881,
      "grad_norm": 0.8496859799536318,
      "learning_rate": 0.0002989814139114377,
      "loss": 0.3652,
      "step": 2399
    },
    {
      "epoch": 1.1294117647058823,
      "grad_norm": 0.5581281384611202,
      "learning_rate": 0.0002989763076412549,
      "loss": 0.252,
      "step": 2400
    },
    {
      "epoch": 1.1298823529411766,
      "grad_norm": 1.858852270882668,
      "learning_rate": 0.00029897118864776476,
      "loss": 1.1562,
      "step": 2401
    },
    {
      "epoch": 1.1303529411764706,
      "grad_norm": 0.7811621303757983,
      "learning_rate": 0.00029896605693140437,
      "loss": 0.293,
      "step": 2402
    },
    {
      "epoch": 1.1308235294117648,
      "grad_norm": 0.8991112872084441,
      "learning_rate": 0.0002989609124926121,
      "loss": 0.7305,
      "step": 2403
    },
    {
      "epoch": 1.1312941176470588,
      "grad_norm": 1.1598497398910432,
      "learning_rate": 0.0002989557553318272,
      "loss": 0.8281,
      "step": 2404
    },
    {
      "epoch": 1.131764705882353,
      "grad_norm": 0.6312268161726515,
      "learning_rate": 0.0002989505854494902,
      "loss": 0.3203,
      "step": 2405
    },
    {
      "epoch": 1.132235294117647,
      "grad_norm": 1.0793859587583012,
      "learning_rate": 0.00029894540284604255,
      "loss": 0.7891,
      "step": 2406
    },
    {
      "epoch": 1.1327058823529412,
      "grad_norm": 0.9043164373063319,
      "learning_rate": 0.000298940207521927,
      "loss": 0.4668,
      "step": 2407
    },
    {
      "epoch": 1.1331764705882352,
      "grad_norm": 0.4733864397194864,
      "learning_rate": 0.00029893499947758706,
      "loss": 0.2041,
      "step": 2408
    },
    {
      "epoch": 1.1336470588235295,
      "grad_norm": 0.7848795601498042,
      "learning_rate": 0.00029892977871346776,
      "loss": 0.5547,
      "step": 2409
    },
    {
      "epoch": 1.1341176470588235,
      "grad_norm": 1.0012102152654672,
      "learning_rate": 0.00029892454523001484,
      "loss": 0.5703,
      "step": 2410
    },
    {
      "epoch": 1.1345882352941177,
      "grad_norm": 0.6551346941469213,
      "learning_rate": 0.00029891929902767526,
      "loss": 0.2422,
      "step": 2411
    },
    {
      "epoch": 1.1350588235294117,
      "grad_norm": 1.0310869054782423,
      "learning_rate": 0.0002989140401068971,
      "loss": 0.5781,
      "step": 2412
    },
    {
      "epoch": 1.135529411764706,
      "grad_norm": 0.5627491180589538,
      "learning_rate": 0.0002989087684681295,
      "loss": 0.2012,
      "step": 2413
    },
    {
      "epoch": 1.1360000000000001,
      "grad_norm": 1.0964886606539903,
      "learning_rate": 0.00029890348411182267,
      "loss": 0.6914,
      "step": 2414
    },
    {
      "epoch": 1.1364705882352941,
      "grad_norm": 1.4104171762567854,
      "learning_rate": 0.00029889818703842796,
      "loss": 0.9922,
      "step": 2415
    },
    {
      "epoch": 1.1369411764705881,
      "grad_norm": 1.7541607348902346,
      "learning_rate": 0.0002988928772483977,
      "loss": 1.3281,
      "step": 2416
    },
    {
      "epoch": 1.1374117647058823,
      "grad_norm": 0.6599832514285109,
      "learning_rate": 0.0002988875547421854,
      "loss": 0.4844,
      "step": 2417
    },
    {
      "epoch": 1.1378823529411766,
      "grad_norm": 0.7223483031017279,
      "learning_rate": 0.00029888221952024567,
      "loss": 0.6055,
      "step": 2418
    },
    {
      "epoch": 1.1383529411764706,
      "grad_norm": 0.7286405375483104,
      "learning_rate": 0.00029887687158303416,
      "loss": 0.5234,
      "step": 2419
    },
    {
      "epoch": 1.1388235294117648,
      "grad_norm": 1.3766147738484527,
      "learning_rate": 0.00029887151093100755,
      "loss": 0.7695,
      "step": 2420
    },
    {
      "epoch": 1.1392941176470588,
      "grad_norm": 1.0729961570875162,
      "learning_rate": 0.00029886613756462365,
      "loss": 0.7969,
      "step": 2421
    },
    {
      "epoch": 1.139764705882353,
      "grad_norm": 0.8637677998722649,
      "learning_rate": 0.00029886075148434146,
      "loss": 0.7617,
      "step": 2422
    },
    {
      "epoch": 1.140235294117647,
      "grad_norm": 1.1088417116266456,
      "learning_rate": 0.0002988553526906209,
      "loss": 0.6641,
      "step": 2423
    },
    {
      "epoch": 1.1407058823529412,
      "grad_norm": 0.8637727059494472,
      "learning_rate": 0.0002988499411839232,
      "loss": 0.8359,
      "step": 2424
    },
    {
      "epoch": 1.1411764705882352,
      "grad_norm": 1.5372895418138126,
      "learning_rate": 0.0002988445169647103,
      "loss": 1.1875,
      "step": 2425
    },
    {
      "epoch": 1.1416470588235295,
      "grad_norm": 0.3640471210176801,
      "learning_rate": 0.00029883908003344554,
      "loss": 0.1338,
      "step": 2426
    },
    {
      "epoch": 1.1421176470588235,
      "grad_norm": 0.9183862700705618,
      "learning_rate": 0.00029883363039059343,
      "loss": 0.4199,
      "step": 2427
    },
    {
      "epoch": 1.1425882352941177,
      "grad_norm": 0.6744778301386125,
      "learning_rate": 0.0002988281680366192,
      "loss": 0.4473,
      "step": 2428
    },
    {
      "epoch": 1.1430588235294117,
      "grad_norm": 1.417938494434292,
      "learning_rate": 0.0002988226929719894,
      "loss": 1.4141,
      "step": 2429
    },
    {
      "epoch": 1.143529411764706,
      "grad_norm": 0.5788700733493756,
      "learning_rate": 0.0002988172051971717,
      "loss": 0.2539,
      "step": 2430
    },
    {
      "epoch": 1.144,
      "grad_norm": 0.3459256835738663,
      "learning_rate": 0.0002988117047126346,
      "loss": 0.1069,
      "step": 2431
    },
    {
      "epoch": 1.1444705882352941,
      "grad_norm": 0.6337276531176129,
      "learning_rate": 0.00029880619151884813,
      "loss": 0.2148,
      "step": 2432
    },
    {
      "epoch": 1.1449411764705881,
      "grad_norm": 1.3580794553693993,
      "learning_rate": 0.000298800665616283,
      "loss": 0.8438,
      "step": 2433
    },
    {
      "epoch": 1.1454117647058824,
      "grad_norm": 0.96295890610476,
      "learning_rate": 0.00029879512700541115,
      "loss": 0.5781,
      "step": 2434
    },
    {
      "epoch": 1.1458823529411766,
      "grad_norm": 0.28563492403489704,
      "learning_rate": 0.0002987895756867056,
      "loss": 0.0903,
      "step": 2435
    },
    {
      "epoch": 1.1463529411764706,
      "grad_norm": 1.2014318140857148,
      "learning_rate": 0.00029878401166064046,
      "loss": 0.8867,
      "step": 2436
    },
    {
      "epoch": 1.1468235294117648,
      "grad_norm": 0.9928590470285922,
      "learning_rate": 0.000298778434927691,
      "loss": 0.7266,
      "step": 2437
    },
    {
      "epoch": 1.1472941176470588,
      "grad_norm": 0.708234647821124,
      "learning_rate": 0.0002987728454883334,
      "loss": 0.3496,
      "step": 2438
    },
    {
      "epoch": 1.147764705882353,
      "grad_norm": 0.8822803997262942,
      "learning_rate": 0.00029876724334304506,
      "loss": 0.6211,
      "step": 2439
    },
    {
      "epoch": 1.148235294117647,
      "grad_norm": 0.5705555103169772,
      "learning_rate": 0.00029876162849230446,
      "loss": 0.249,
      "step": 2440
    },
    {
      "epoch": 1.1487058823529412,
      "grad_norm": 1.1149873380755195,
      "learning_rate": 0.00029875600093659115,
      "loss": 0.7891,
      "step": 2441
    },
    {
      "epoch": 1.1491764705882352,
      "grad_norm": 1.4379648944944412,
      "learning_rate": 0.00029875036067638565,
      "loss": 1.1172,
      "step": 2442
    },
    {
      "epoch": 1.1496470588235295,
      "grad_norm": 1.4347302251940168,
      "learning_rate": 0.00029874470771216983,
      "loss": 1.1094,
      "step": 2443
    },
    {
      "epoch": 1.1501176470588235,
      "grad_norm": 0.5670877380993351,
      "learning_rate": 0.0002987390420444262,
      "loss": 0.4375,
      "step": 2444
    },
    {
      "epoch": 1.1505882352941177,
      "grad_norm": 0.6619056080340112,
      "learning_rate": 0.00029873336367363903,
      "loss": 0.4688,
      "step": 2445
    },
    {
      "epoch": 1.1510588235294117,
      "grad_norm": 0.9167895826204294,
      "learning_rate": 0.00029872767260029294,
      "loss": 0.6055,
      "step": 2446
    },
    {
      "epoch": 1.151529411764706,
      "grad_norm": 0.9150851828839354,
      "learning_rate": 0.00029872196882487417,
      "loss": 0.6875,
      "step": 2447
    },
    {
      "epoch": 1.152,
      "grad_norm": 1.7682874469051595,
      "learning_rate": 0.00029871625234786974,
      "loss": 1.3516,
      "step": 2448
    },
    {
      "epoch": 1.1524705882352941,
      "grad_norm": 1.4508835360921593,
      "learning_rate": 0.0002987105231697679,
      "loss": 1.0469,
      "step": 2449
    },
    {
      "epoch": 1.1529411764705881,
      "grad_norm": 0.8272898232466562,
      "learning_rate": 0.000298704781291058,
      "loss": 0.4297,
      "step": 2450
    },
    {
      "epoch": 1.1534117647058824,
      "grad_norm": 0.9221723008456759,
      "learning_rate": 0.0002986990267122304,
      "loss": 0.3574,
      "step": 2451
    },
    {
      "epoch": 1.1538823529411766,
      "grad_norm": 1.2629875191006574,
      "learning_rate": 0.0002986932594337765,
      "loss": 1.1328,
      "step": 2452
    },
    {
      "epoch": 1.1543529411764706,
      "grad_norm": 1.05243383923291,
      "learning_rate": 0.0002986874794561889,
      "loss": 0.7148,
      "step": 2453
    },
    {
      "epoch": 1.1548235294117646,
      "grad_norm": 1.3506844204464075,
      "learning_rate": 0.0002986816867799613,
      "loss": 1.2734,
      "step": 2454
    },
    {
      "epoch": 1.1552941176470588,
      "grad_norm": 0.48497313454089425,
      "learning_rate": 0.0002986758814055883,
      "loss": 0.2578,
      "step": 2455
    },
    {
      "epoch": 1.155764705882353,
      "grad_norm": 0.998648126282261,
      "learning_rate": 0.0002986700633335659,
      "loss": 0.8906,
      "step": 2456
    },
    {
      "epoch": 1.156235294117647,
      "grad_norm": 0.9778757633211168,
      "learning_rate": 0.0002986642325643908,
      "loss": 0.6602,
      "step": 2457
    },
    {
      "epoch": 1.1567058823529413,
      "grad_norm": 0.8867500687853507,
      "learning_rate": 0.00029865838909856103,
      "loss": 0.7344,
      "step": 2458
    },
    {
      "epoch": 1.1571764705882353,
      "grad_norm": 1.156468517165448,
      "learning_rate": 0.0002986525329365757,
      "loss": 1.1953,
      "step": 2459
    },
    {
      "epoch": 1.1576470588235295,
      "grad_norm": 0.8454510448483039,
      "learning_rate": 0.00029864666407893487,
      "loss": 0.5195,
      "step": 2460
    },
    {
      "epoch": 1.1581176470588235,
      "grad_norm": 0.7750743056228809,
      "learning_rate": 0.00029864078252613983,
      "loss": 0.7148,
      "step": 2461
    },
    {
      "epoch": 1.1585882352941177,
      "grad_norm": 0.9323672184925377,
      "learning_rate": 0.0002986348882786929,
      "loss": 0.7539,
      "step": 2462
    },
    {
      "epoch": 1.1590588235294117,
      "grad_norm": 1.170023189045642,
      "learning_rate": 0.00029862898133709747,
      "loss": 0.5898,
      "step": 2463
    },
    {
      "epoch": 1.159529411764706,
      "grad_norm": 0.8416549045545307,
      "learning_rate": 0.000298623061701858,
      "loss": 0.6484,
      "step": 2464
    },
    {
      "epoch": 1.16,
      "grad_norm": 0.8960355196000919,
      "learning_rate": 0.00029861712937348007,
      "loss": 0.7031,
      "step": 2465
    },
    {
      "epoch": 1.1604705882352941,
      "grad_norm": 0.6329404159834963,
      "learning_rate": 0.0002986111843524703,
      "loss": 0.5039,
      "step": 2466
    },
    {
      "epoch": 1.1609411764705881,
      "grad_norm": 0.6737939293028594,
      "learning_rate": 0.00029860522663933655,
      "loss": 0.3926,
      "step": 2467
    },
    {
      "epoch": 1.1614117647058824,
      "grad_norm": 0.9757667554027326,
      "learning_rate": 0.00029859925623458745,
      "loss": 0.6289,
      "step": 2468
    },
    {
      "epoch": 1.1618823529411766,
      "grad_norm": 0.9743877468625676,
      "learning_rate": 0.000298593273138733,
      "loss": 0.5195,
      "step": 2469
    },
    {
      "epoch": 1.1623529411764706,
      "grad_norm": 0.6393813014295363,
      "learning_rate": 0.00029858727735228416,
      "loss": 0.416,
      "step": 2470
    },
    {
      "epoch": 1.1628235294117646,
      "grad_norm": 1.1643380871291809,
      "learning_rate": 0.00029858126887575306,
      "loss": 0.9023,
      "step": 2471
    },
    {
      "epoch": 1.1632941176470588,
      "grad_norm": 0.8922688322343245,
      "learning_rate": 0.00029857524770965285,
      "loss": 0.7461,
      "step": 2472
    },
    {
      "epoch": 1.163764705882353,
      "grad_norm": 0.7979797682236333,
      "learning_rate": 0.00029856921385449763,
      "loss": 0.4863,
      "step": 2473
    },
    {
      "epoch": 1.164235294117647,
      "grad_norm": 1.1207779452920417,
      "learning_rate": 0.0002985631673108029,
      "loss": 1.1719,
      "step": 2474
    },
    {
      "epoch": 1.1647058823529413,
      "grad_norm": 0.4370597235839614,
      "learning_rate": 0.00029855710807908493,
      "loss": 0.2002,
      "step": 2475
    },
    {
      "epoch": 1.1651764705882353,
      "grad_norm": 0.6086522291508367,
      "learning_rate": 0.00029855103615986133,
      "loss": 0.3652,
      "step": 2476
    },
    {
      "epoch": 1.1656470588235295,
      "grad_norm": 0.8404311138651958,
      "learning_rate": 0.0002985449515536505,
      "loss": 0.4746,
      "step": 2477
    },
    {
      "epoch": 1.1661176470588235,
      "grad_norm": 0.7415578431839855,
      "learning_rate": 0.0002985388542609723,
      "loss": 0.832,
      "step": 2478
    },
    {
      "epoch": 1.1665882352941177,
      "grad_norm": 1.116726409014953,
      "learning_rate": 0.0002985327442823473,
      "loss": 1.0781,
      "step": 2479
    },
    {
      "epoch": 1.1670588235294117,
      "grad_norm": 0.5546628165694906,
      "learning_rate": 0.00029852662161829744,
      "loss": 0.4316,
      "step": 2480
    },
    {
      "epoch": 1.167529411764706,
      "grad_norm": 0.9163388177187942,
      "learning_rate": 0.0002985204862693455,
      "loss": 0.7031,
      "step": 2481
    },
    {
      "epoch": 1.168,
      "grad_norm": 0.41079324525412136,
      "learning_rate": 0.00029851433823601563,
      "loss": 0.3457,
      "step": 2482
    },
    {
      "epoch": 1.1684705882352941,
      "grad_norm": 0.954361662801579,
      "learning_rate": 0.00029850817751883276,
      "loss": 0.5156,
      "step": 2483
    },
    {
      "epoch": 1.1689411764705882,
      "grad_norm": 0.8786559944266982,
      "learning_rate": 0.0002985020041183231,
      "loss": 0.4492,
      "step": 2484
    },
    {
      "epoch": 1.1694117647058824,
      "grad_norm": 1.2250486686881799,
      "learning_rate": 0.00029849581803501393,
      "loss": 0.6602,
      "step": 2485
    },
    {
      "epoch": 1.1698823529411766,
      "grad_norm": 1.0225788942141885,
      "learning_rate": 0.0002984896192694335,
      "loss": 0.6797,
      "step": 2486
    },
    {
      "epoch": 1.1703529411764706,
      "grad_norm": 1.1453493775852406,
      "learning_rate": 0.00029848340782211125,
      "loss": 0.8047,
      "step": 2487
    },
    {
      "epoch": 1.1708235294117646,
      "grad_norm": 0.8612011983074787,
      "learning_rate": 0.0002984771836935777,
      "loss": 0.6211,
      "step": 2488
    },
    {
      "epoch": 1.1712941176470588,
      "grad_norm": 1.176783326779863,
      "learning_rate": 0.0002984709468843644,
      "loss": 0.707,
      "step": 2489
    },
    {
      "epoch": 1.171764705882353,
      "grad_norm": 1.0992256862194811,
      "learning_rate": 0.0002984646973950039,
      "loss": 0.7383,
      "step": 2490
    },
    {
      "epoch": 1.172235294117647,
      "grad_norm": 0.39655070328847186,
      "learning_rate": 0.00029845843522603006,
      "loss": 0.2578,
      "step": 2491
    },
    {
      "epoch": 1.1727058823529413,
      "grad_norm": 0.8901711731784201,
      "learning_rate": 0.0002984521603779777,
      "loss": 0.7617,
      "step": 2492
    },
    {
      "epoch": 1.1731764705882353,
      "grad_norm": 1.8445202158800758,
      "learning_rate": 0.0002984458728513827,
      "loss": 1.2344,
      "step": 2493
    },
    {
      "epoch": 1.1736470588235295,
      "grad_norm": 0.9983946922518832,
      "learning_rate": 0.000298439572646782,
      "loss": 0.9727,
      "step": 2494
    },
    {
      "epoch": 1.1741176470588235,
      "grad_norm": 0.3250979228316444,
      "learning_rate": 0.0002984332597647137,
      "loss": 0.1011,
      "step": 2495
    },
    {
      "epoch": 1.1745882352941177,
      "grad_norm": 0.7830877026093879,
      "learning_rate": 0.000298426934205717,
      "loss": 0.4785,
      "step": 2496
    },
    {
      "epoch": 1.1750588235294117,
      "grad_norm": 1.0652739052751399,
      "learning_rate": 0.00029842059597033204,
      "loss": 0.5,
      "step": 2497
    },
    {
      "epoch": 1.175529411764706,
      "grad_norm": 0.8038657421474975,
      "learning_rate": 0.0002984142450591002,
      "loss": 0.4863,
      "step": 2498
    },
    {
      "epoch": 1.176,
      "grad_norm": 0.7857065364193357,
      "learning_rate": 0.0002984078814725639,
      "loss": 0.4395,
      "step": 2499
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.7781455228617316,
      "learning_rate": 0.0002984015052112665,
      "loss": 0.5,
      "step": 2500
    },
    {
      "epoch": 1.1769411764705882,
      "grad_norm": 0.3684267897976682,
      "learning_rate": 0.0002983951162757527,
      "loss": 0.1807,
      "step": 2501
    },
    {
      "epoch": 1.1774117647058824,
      "grad_norm": 1.0931616013782306,
      "learning_rate": 0.0002983887146665681,
      "loss": 0.7812,
      "step": 2502
    },
    {
      "epoch": 1.1778823529411764,
      "grad_norm": 0.8862285572539447,
      "learning_rate": 0.00029838230038425944,
      "loss": 0.8125,
      "step": 2503
    },
    {
      "epoch": 1.1783529411764706,
      "grad_norm": 0.8999708216042188,
      "learning_rate": 0.0002983758734293745,
      "loss": 0.5547,
      "step": 2504
    },
    {
      "epoch": 1.1788235294117646,
      "grad_norm": 1.5483618792208205,
      "learning_rate": 0.00029836943380246215,
      "loss": 0.9844,
      "step": 2505
    },
    {
      "epoch": 1.1792941176470588,
      "grad_norm": 0.5784255213705254,
      "learning_rate": 0.00029836298150407245,
      "loss": 0.3066,
      "step": 2506
    },
    {
      "epoch": 1.179764705882353,
      "grad_norm": 1.0550759554676812,
      "learning_rate": 0.00029835651653475643,
      "loss": 0.6484,
      "step": 2507
    },
    {
      "epoch": 1.180235294117647,
      "grad_norm": 0.17469268234825935,
      "learning_rate": 0.00029835003889506616,
      "loss": 0.0791,
      "step": 2508
    },
    {
      "epoch": 1.1807058823529413,
      "grad_norm": 0.5924126609129914,
      "learning_rate": 0.000298343548585555,
      "loss": 0.2402,
      "step": 2509
    },
    {
      "epoch": 1.1811764705882353,
      "grad_norm": 1.8686639895460386,
      "learning_rate": 0.0002983370456067771,
      "loss": 1.5703,
      "step": 2510
    },
    {
      "epoch": 1.1816470588235295,
      "grad_norm": 1.1109526880362501,
      "learning_rate": 0.0002983305299592879,
      "loss": 0.8828,
      "step": 2511
    },
    {
      "epoch": 1.1821176470588235,
      "grad_norm": 0.799915225952583,
      "learning_rate": 0.0002983240016436439,
      "loss": 0.4785,
      "step": 2512
    },
    {
      "epoch": 1.1825882352941177,
      "grad_norm": 0.46831184703821815,
      "learning_rate": 0.0002983174606604027,
      "loss": 0.2754,
      "step": 2513
    },
    {
      "epoch": 1.1830588235294117,
      "grad_norm": 0.8367766446097563,
      "learning_rate": 0.0002983109070101228,
      "loss": 0.3086,
      "step": 2514
    },
    {
      "epoch": 1.183529411764706,
      "grad_norm": 1.2676401660428163,
      "learning_rate": 0.00029830434069336403,
      "loss": 0.875,
      "step": 2515
    },
    {
      "epoch": 1.184,
      "grad_norm": 1.7820486123768355,
      "learning_rate": 0.0002982977617106871,
      "loss": 1.0938,
      "step": 2516
    },
    {
      "epoch": 1.1844705882352942,
      "grad_norm": 0.9205771009306458,
      "learning_rate": 0.0002982911700626539,
      "loss": 0.5469,
      "step": 2517
    },
    {
      "epoch": 1.1849411764705882,
      "grad_norm": 0.821094441115192,
      "learning_rate": 0.0002982845657498274,
      "loss": 0.7344,
      "step": 2518
    },
    {
      "epoch": 1.1854117647058824,
      "grad_norm": 0.9172746336282726,
      "learning_rate": 0.0002982779487727718,
      "loss": 0.7617,
      "step": 2519
    },
    {
      "epoch": 1.1858823529411764,
      "grad_norm": 0.8458254732687475,
      "learning_rate": 0.0002982713191320519,
      "loss": 0.6133,
      "step": 2520
    },
    {
      "epoch": 1.1863529411764706,
      "grad_norm": 0.7651756105865903,
      "learning_rate": 0.00029826467682823415,
      "loss": 0.5508,
      "step": 2521
    },
    {
      "epoch": 1.1868235294117646,
      "grad_norm": 1.620550956072359,
      "learning_rate": 0.0002982580218618858,
      "loss": 0.7969,
      "step": 2522
    },
    {
      "epoch": 1.1872941176470588,
      "grad_norm": 0.7005918607008528,
      "learning_rate": 0.0002982513542335751,
      "loss": 0.3301,
      "step": 2523
    },
    {
      "epoch": 1.187764705882353,
      "grad_norm": 0.9986061189850286,
      "learning_rate": 0.00029824467394387166,
      "loss": 0.4863,
      "step": 2524
    },
    {
      "epoch": 1.188235294117647,
      "grad_norm": 1.1659306238285057,
      "learning_rate": 0.0002982379809933459,
      "loss": 1.0156,
      "step": 2525
    },
    {
      "epoch": 1.188705882352941,
      "grad_norm": 0.5936314798006616,
      "learning_rate": 0.00029823127538256943,
      "loss": 0.3262,
      "step": 2526
    },
    {
      "epoch": 1.1891764705882353,
      "grad_norm": 0.5717094105192694,
      "learning_rate": 0.000298224557112115,
      "loss": 0.2451,
      "step": 2527
    },
    {
      "epoch": 1.1896470588235295,
      "grad_norm": 0.8836337235258725,
      "learning_rate": 0.0002982178261825563,
      "loss": 0.3867,
      "step": 2528
    },
    {
      "epoch": 1.1901176470588235,
      "grad_norm": 1.5858928587581247,
      "learning_rate": 0.0002982110825944683,
      "loss": 0.9336,
      "step": 2529
    },
    {
      "epoch": 1.1905882352941177,
      "grad_norm": 1.5273270828406806,
      "learning_rate": 0.0002982043263484269,
      "loss": 0.9375,
      "step": 2530
    },
    {
      "epoch": 1.1910588235294117,
      "grad_norm": 0.8194177783729901,
      "learning_rate": 0.00029819755744500904,
      "loss": 0.8477,
      "step": 2531
    },
    {
      "epoch": 1.191529411764706,
      "grad_norm": 1.0810264356839803,
      "learning_rate": 0.0002981907758847929,
      "loss": 0.793,
      "step": 2532
    },
    {
      "epoch": 1.192,
      "grad_norm": 0.7836024057590668,
      "learning_rate": 0.0002981839816683576,
      "loss": 0.6133,
      "step": 2533
    },
    {
      "epoch": 1.1924705882352942,
      "grad_norm": 0.5431244755424938,
      "learning_rate": 0.00029817717479628346,
      "loss": 0.1758,
      "step": 2534
    },
    {
      "epoch": 1.1929411764705882,
      "grad_norm": 1.0311648999390295,
      "learning_rate": 0.00029817035526915177,
      "loss": 0.7656,
      "step": 2535
    },
    {
      "epoch": 1.1934117647058824,
      "grad_norm": 1.56128671276236,
      "learning_rate": 0.00029816352308754496,
      "loss": 1.0625,
      "step": 2536
    },
    {
      "epoch": 1.1938823529411764,
      "grad_norm": 0.700514454844549,
      "learning_rate": 0.00029815667825204654,
      "loss": 0.3418,
      "step": 2537
    },
    {
      "epoch": 1.1943529411764706,
      "grad_norm": 0.6955922377491287,
      "learning_rate": 0.0002981498207632411,
      "loss": 0.2373,
      "step": 2538
    },
    {
      "epoch": 1.1948235294117646,
      "grad_norm": 1.2307859594140138,
      "learning_rate": 0.0002981429506217143,
      "loss": 0.9609,
      "step": 2539
    },
    {
      "epoch": 1.1952941176470588,
      "grad_norm": 0.8026955863463063,
      "learning_rate": 0.0002981360678280529,
      "loss": 0.3672,
      "step": 2540
    },
    {
      "epoch": 1.195764705882353,
      "grad_norm": 0.9464741909572604,
      "learning_rate": 0.00029812917238284466,
      "loss": 0.6406,
      "step": 2541
    },
    {
      "epoch": 1.196235294117647,
      "grad_norm": 0.579395970659104,
      "learning_rate": 0.00029812226428667854,
      "loss": 0.3086,
      "step": 2542
    },
    {
      "epoch": 1.196705882352941,
      "grad_norm": 0.9268359765627162,
      "learning_rate": 0.0002981153435401445,
      "loss": 0.7695,
      "step": 2543
    },
    {
      "epoch": 1.1971764705882353,
      "grad_norm": 1.1635579034161427,
      "learning_rate": 0.00029810841014383366,
      "loss": 0.9531,
      "step": 2544
    },
    {
      "epoch": 1.1976470588235295,
      "grad_norm": 1.2798793712120171,
      "learning_rate": 0.00029810146409833815,
      "loss": 0.8945,
      "step": 2545
    },
    {
      "epoch": 1.1981176470588235,
      "grad_norm": 1.5311614410545795,
      "learning_rate": 0.0002980945054042511,
      "loss": 1.4141,
      "step": 2546
    },
    {
      "epoch": 1.1985882352941177,
      "grad_norm": 0.5015911815962762,
      "learning_rate": 0.00029808753406216693,
      "loss": 0.4258,
      "step": 2547
    },
    {
      "epoch": 1.1990588235294117,
      "grad_norm": 1.2090559119973476,
      "learning_rate": 0.000298080550072681,
      "loss": 0.8398,
      "step": 2548
    },
    {
      "epoch": 1.199529411764706,
      "grad_norm": 0.5385834780039815,
      "learning_rate": 0.0002980735534363897,
      "loss": 0.123,
      "step": 2549
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.0798947221640907,
      "learning_rate": 0.0002980665441538907,
      "loss": 1.1641,
      "step": 2550
    },
    {
      "epoch": 1.2004705882352942,
      "grad_norm": 1.204204320972639,
      "learning_rate": 0.00029805952222578257,
      "loss": 1.0703,
      "step": 2551
    },
    {
      "epoch": 1.2009411764705882,
      "grad_norm": 0.9007545767206081,
      "learning_rate": 0.000298052487652665,
      "loss": 0.7031,
      "step": 2552
    },
    {
      "epoch": 1.2014117647058824,
      "grad_norm": 1.160176848559561,
      "learning_rate": 0.0002980454404351388,
      "loss": 0.9258,
      "step": 2553
    },
    {
      "epoch": 1.2018823529411764,
      "grad_norm": 1.3577525731779458,
      "learning_rate": 0.0002980383805738058,
      "loss": 0.7695,
      "step": 2554
    },
    {
      "epoch": 1.2023529411764706,
      "grad_norm": 0.8450068493361357,
      "learning_rate": 0.00029803130806926897,
      "loss": 0.5352,
      "step": 2555
    },
    {
      "epoch": 1.2028235294117646,
      "grad_norm": 1.1834848154336273,
      "learning_rate": 0.0002980242229221324,
      "loss": 0.5352,
      "step": 2556
    },
    {
      "epoch": 1.2032941176470588,
      "grad_norm": 1.3943548799384466,
      "learning_rate": 0.00029801712513300105,
      "loss": 0.8164,
      "step": 2557
    },
    {
      "epoch": 1.2037647058823528,
      "grad_norm": 0.815970703794272,
      "learning_rate": 0.00029801001470248123,
      "loss": 0.5352,
      "step": 2558
    },
    {
      "epoch": 1.204235294117647,
      "grad_norm": 1.0351142989994637,
      "learning_rate": 0.00029800289163118014,
      "loss": 0.7852,
      "step": 2559
    },
    {
      "epoch": 1.204705882352941,
      "grad_norm": 0.9651227233954907,
      "learning_rate": 0.00029799575591970623,
      "loss": 0.8086,
      "step": 2560
    },
    {
      "epoch": 1.2051764705882353,
      "grad_norm": 1.1067050787559205,
      "learning_rate": 0.00029798860756866883,
      "loss": 0.9453,
      "step": 2561
    },
    {
      "epoch": 1.2056470588235295,
      "grad_norm": 0.531324885509593,
      "learning_rate": 0.00029798144657867845,
      "loss": 0.1904,
      "step": 2562
    },
    {
      "epoch": 1.2061176470588235,
      "grad_norm": 1.2469915508920941,
      "learning_rate": 0.0002979742729503467,
      "loss": 0.6133,
      "step": 2563
    },
    {
      "epoch": 1.2065882352941177,
      "grad_norm": 0.5305751324324456,
      "learning_rate": 0.0002979670866842862,
      "loss": 0.2217,
      "step": 2564
    },
    {
      "epoch": 1.2070588235294117,
      "grad_norm": 1.264194668695696,
      "learning_rate": 0.0002979598877811107,
      "loss": 1.0156,
      "step": 2565
    },
    {
      "epoch": 1.207529411764706,
      "grad_norm": 0.6146984642636657,
      "learning_rate": 0.0002979526762414351,
      "loss": 0.3574,
      "step": 2566
    },
    {
      "epoch": 1.208,
      "grad_norm": 1.2667682682491088,
      "learning_rate": 0.0002979454520658752,
      "loss": 0.8672,
      "step": 2567
    },
    {
      "epoch": 1.2084705882352942,
      "grad_norm": 1.4655150502197183,
      "learning_rate": 0.0002979382152550481,
      "loss": 0.8086,
      "step": 2568
    },
    {
      "epoch": 1.2089411764705882,
      "grad_norm": 2.019563515764998,
      "learning_rate": 0.0002979309658095717,
      "loss": 0.6094,
      "step": 2569
    },
    {
      "epoch": 1.2094117647058824,
      "grad_norm": 0.8504951601878544,
      "learning_rate": 0.0002979237037300653,
      "loss": 0.5859,
      "step": 2570
    },
    {
      "epoch": 1.2098823529411764,
      "grad_norm": 0.4280110971641485,
      "learning_rate": 0.00029791642901714897,
      "loss": 0.2139,
      "step": 2571
    },
    {
      "epoch": 1.2103529411764706,
      "grad_norm": 0.6068788727079037,
      "learning_rate": 0.0002979091416714441,
      "loss": 0.3262,
      "step": 2572
    },
    {
      "epoch": 1.2108235294117646,
      "grad_norm": 0.5726836242232237,
      "learning_rate": 0.00029790184169357307,
      "loss": 0.3867,
      "step": 2573
    },
    {
      "epoch": 1.2112941176470589,
      "grad_norm": 0.9601520928757906,
      "learning_rate": 0.00029789452908415923,
      "loss": 0.4629,
      "step": 2574
    },
    {
      "epoch": 1.2117647058823529,
      "grad_norm": 1.3494868238446003,
      "learning_rate": 0.0002978872038438273,
      "loss": 0.7305,
      "step": 2575
    },
    {
      "epoch": 1.212235294117647,
      "grad_norm": 0.6002506221025014,
      "learning_rate": 0.0002978798659732027,
      "loss": 0.2598,
      "step": 2576
    },
    {
      "epoch": 1.212705882352941,
      "grad_norm": 0.9949404559015643,
      "learning_rate": 0.00029787251547291215,
      "loss": 0.7148,
      "step": 2577
    },
    {
      "epoch": 1.2131764705882353,
      "grad_norm": 0.6054813802225998,
      "learning_rate": 0.00029786515234358357,
      "loss": 0.3066,
      "step": 2578
    },
    {
      "epoch": 1.2136470588235295,
      "grad_norm": 1.2382586751540345,
      "learning_rate": 0.0002978577765858457,
      "loss": 0.9414,
      "step": 2579
    },
    {
      "epoch": 1.2141176470588235,
      "grad_norm": 0.8720548032261931,
      "learning_rate": 0.00029785038820032844,
      "loss": 0.4902,
      "step": 2580
    },
    {
      "epoch": 1.2145882352941175,
      "grad_norm": 0.6989358436371561,
      "learning_rate": 0.0002978429871876628,
      "loss": 0.3203,
      "step": 2581
    },
    {
      "epoch": 1.2150588235294117,
      "grad_norm": 0.7459123165400682,
      "learning_rate": 0.00029783557354848094,
      "loss": 0.2852,
      "step": 2582
    },
    {
      "epoch": 1.215529411764706,
      "grad_norm": 1.6416117021761543,
      "learning_rate": 0.0002978281472834159,
      "loss": 1.0078,
      "step": 2583
    },
    {
      "epoch": 1.216,
      "grad_norm": 0.5689427485736205,
      "learning_rate": 0.00029782070839310206,
      "loss": 0.3457,
      "step": 2584
    },
    {
      "epoch": 1.2164705882352942,
      "grad_norm": 1.0548161761629014,
      "learning_rate": 0.0002978132568781746,
      "loss": 0.8633,
      "step": 2585
    },
    {
      "epoch": 1.2169411764705882,
      "grad_norm": 1.1289285797866437,
      "learning_rate": 0.00029780579273927006,
      "loss": 0.7578,
      "step": 2586
    },
    {
      "epoch": 1.2174117647058824,
      "grad_norm": 0.9225155805420421,
      "learning_rate": 0.0002977983159770258,
      "loss": 0.4961,
      "step": 2587
    },
    {
      "epoch": 1.2178823529411764,
      "grad_norm": 1.0698450706058933,
      "learning_rate": 0.0002977908265920804,
      "loss": 0.9883,
      "step": 2588
    },
    {
      "epoch": 1.2183529411764706,
      "grad_norm": 1.1026406742704293,
      "learning_rate": 0.00029778332458507353,
      "loss": 0.7305,
      "step": 2589
    },
    {
      "epoch": 1.2188235294117646,
      "grad_norm": 0.7525061789854488,
      "learning_rate": 0.0002977758099566458,
      "loss": 0.5117,
      "step": 2590
    },
    {
      "epoch": 1.2192941176470589,
      "grad_norm": 0.5301392398332454,
      "learning_rate": 0.0002977682827074391,
      "loss": 0.3555,
      "step": 2591
    },
    {
      "epoch": 1.2197647058823529,
      "grad_norm": 0.47144356095888335,
      "learning_rate": 0.0002977607428380963,
      "loss": 0.4395,
      "step": 2592
    },
    {
      "epoch": 1.220235294117647,
      "grad_norm": 0.8324616873882688,
      "learning_rate": 0.00029775319034926125,
      "loss": 0.6602,
      "step": 2593
    },
    {
      "epoch": 1.220705882352941,
      "grad_norm": 2.2862248175596838,
      "learning_rate": 0.000297745625241579,
      "loss": 1.0078,
      "step": 2594
    },
    {
      "epoch": 1.2211764705882353,
      "grad_norm": 0.9215681203250575,
      "learning_rate": 0.0002977380475156958,
      "loss": 0.5469,
      "step": 2595
    },
    {
      "epoch": 1.2216470588235295,
      "grad_norm": 0.4564897910979463,
      "learning_rate": 0.00029773045717225856,
      "loss": 0.1367,
      "step": 2596
    },
    {
      "epoch": 1.2221176470588235,
      "grad_norm": 0.8730840263611676,
      "learning_rate": 0.0002977228542119157,
      "loss": 0.9805,
      "step": 2597
    },
    {
      "epoch": 1.2225882352941175,
      "grad_norm": 0.7038217644118696,
      "learning_rate": 0.0002977152386353165,
      "loss": 0.3867,
      "step": 2598
    },
    {
      "epoch": 1.2230588235294118,
      "grad_norm": 0.5698203267394444,
      "learning_rate": 0.0002977076104431114,
      "loss": 0.2266,
      "step": 2599
    },
    {
      "epoch": 1.223529411764706,
      "grad_norm": 0.8533918575065508,
      "learning_rate": 0.00029769996963595184,
      "loss": 0.5625,
      "step": 2600
    },
    {
      "epoch": 1.224,
      "grad_norm": 0.750462204628515,
      "learning_rate": 0.0002976923162144904,
      "loss": 0.5781,
      "step": 2601
    },
    {
      "epoch": 1.2244705882352942,
      "grad_norm": 0.7606439089634469,
      "learning_rate": 0.0002976846501793808,
      "loss": 0.6094,
      "step": 2602
    },
    {
      "epoch": 1.2249411764705882,
      "grad_norm": 1.4549242057038438,
      "learning_rate": 0.00029767697153127766,
      "loss": 0.8555,
      "step": 2603
    },
    {
      "epoch": 1.2254117647058824,
      "grad_norm": 1.3436305361738485,
      "learning_rate": 0.0002976692802708368,
      "loss": 1.4453,
      "step": 2604
    },
    {
      "epoch": 1.2258823529411764,
      "grad_norm": 0.8645961537962426,
      "learning_rate": 0.000297661576398715,
      "loss": 0.7734,
      "step": 2605
    },
    {
      "epoch": 1.2263529411764706,
      "grad_norm": 1.9583632786052798,
      "learning_rate": 0.0002976538599155704,
      "loss": 1.2578,
      "step": 2606
    },
    {
      "epoch": 1.2268235294117646,
      "grad_norm": 0.6075552449460492,
      "learning_rate": 0.00029764613082206183,
      "loss": 0.3945,
      "step": 2607
    },
    {
      "epoch": 1.2272941176470589,
      "grad_norm": 0.6327621689043752,
      "learning_rate": 0.0002976383891188496,
      "loss": 0.6211,
      "step": 2608
    },
    {
      "epoch": 1.2277647058823529,
      "grad_norm": 0.5179328197088245,
      "learning_rate": 0.0002976306348065947,
      "loss": 0.1924,
      "step": 2609
    },
    {
      "epoch": 1.228235294117647,
      "grad_norm": 1.1475710517246638,
      "learning_rate": 0.00029762286788595954,
      "loss": 0.7188,
      "step": 2610
    },
    {
      "epoch": 1.228705882352941,
      "grad_norm": 1.1781382347295124,
      "learning_rate": 0.0002976150883576073,
      "loss": 0.9648,
      "step": 2611
    },
    {
      "epoch": 1.2291764705882353,
      "grad_norm": 1.0600823990244983,
      "learning_rate": 0.0002976072962222025,
      "loss": 0.6836,
      "step": 2612
    },
    {
      "epoch": 1.2296470588235293,
      "grad_norm": 0.6211591494573159,
      "learning_rate": 0.0002975994914804106,
      "loss": 0.3711,
      "step": 2613
    },
    {
      "epoch": 1.2301176470588235,
      "grad_norm": 1.1801165610764488,
      "learning_rate": 0.00029759167413289815,
      "loss": 1.4219,
      "step": 2614
    },
    {
      "epoch": 1.2305882352941175,
      "grad_norm": 0.9537366663483676,
      "learning_rate": 0.00029758384418033276,
      "loss": 0.8086,
      "step": 2615
    },
    {
      "epoch": 1.2310588235294118,
      "grad_norm": 0.5919655253681699,
      "learning_rate": 0.00029757600162338326,
      "loss": 0.3027,
      "step": 2616
    },
    {
      "epoch": 1.231529411764706,
      "grad_norm": 1.0546857814247588,
      "learning_rate": 0.0002975681464627193,
      "loss": 0.7461,
      "step": 2617
    },
    {
      "epoch": 1.232,
      "grad_norm": 1.211764850054186,
      "learning_rate": 0.00029756027869901183,
      "loss": 0.6172,
      "step": 2618
    },
    {
      "epoch": 1.2324705882352942,
      "grad_norm": 0.9007468195858214,
      "learning_rate": 0.0002975523983329328,
      "loss": 0.5664,
      "step": 2619
    },
    {
      "epoch": 1.2329411764705882,
      "grad_norm": 0.9937759925433481,
      "learning_rate": 0.00029754450536515517,
      "loss": 0.5234,
      "step": 2620
    },
    {
      "epoch": 1.2334117647058824,
      "grad_norm": 0.8551073872925369,
      "learning_rate": 0.00029753659979635315,
      "loss": 0.5234,
      "step": 2621
    },
    {
      "epoch": 1.2338823529411764,
      "grad_norm": 0.6934653711445434,
      "learning_rate": 0.00029752868162720176,
      "loss": 0.3516,
      "step": 2622
    },
    {
      "epoch": 1.2343529411764707,
      "grad_norm": 1.1989425413173556,
      "learning_rate": 0.0002975207508583774,
      "loss": 0.8164,
      "step": 2623
    },
    {
      "epoch": 1.2348235294117647,
      "grad_norm": 1.1230272655789175,
      "learning_rate": 0.00029751280749055727,
      "loss": 0.6875,
      "step": 2624
    },
    {
      "epoch": 1.2352941176470589,
      "grad_norm": 0.825619327501281,
      "learning_rate": 0.0002975048515244199,
      "loss": 0.5703,
      "step": 2625
    },
    {
      "epoch": 1.2357647058823529,
      "grad_norm": 0.9645225031012338,
      "learning_rate": 0.0002974968829606447,
      "loss": 0.7812,
      "step": 2626
    },
    {
      "epoch": 1.236235294117647,
      "grad_norm": 0.9548599330258208,
      "learning_rate": 0.0002974889017999122,
      "loss": 0.7188,
      "step": 2627
    },
    {
      "epoch": 1.236705882352941,
      "grad_norm": 0.570390667176483,
      "learning_rate": 0.00029748090804290403,
      "loss": 0.3926,
      "step": 2628
    },
    {
      "epoch": 1.2371764705882353,
      "grad_norm": 1.018565922738356,
      "learning_rate": 0.00029747290169030295,
      "loss": 0.7617,
      "step": 2629
    },
    {
      "epoch": 1.2376470588235293,
      "grad_norm": 0.645774313621377,
      "learning_rate": 0.0002974648827427927,
      "loss": 0.248,
      "step": 2630
    },
    {
      "epoch": 1.2381176470588235,
      "grad_norm": 0.8684516579615612,
      "learning_rate": 0.0002974568512010582,
      "loss": 0.7031,
      "step": 2631
    },
    {
      "epoch": 1.2385882352941175,
      "grad_norm": 1.2150286850570968,
      "learning_rate": 0.00029744880706578526,
      "loss": 1.1484,
      "step": 2632
    },
    {
      "epoch": 1.2390588235294118,
      "grad_norm": 0.7267827169202095,
      "learning_rate": 0.00029744075033766103,
      "loss": 0.6133,
      "step": 2633
    },
    {
      "epoch": 1.239529411764706,
      "grad_norm": 0.6448311337995338,
      "learning_rate": 0.0002974326810173735,
      "loss": 0.5156,
      "step": 2634
    },
    {
      "epoch": 1.24,
      "grad_norm": 0.47460357748588233,
      "learning_rate": 0.00029742459910561187,
      "loss": 0.25,
      "step": 2635
    },
    {
      "epoch": 1.2404705882352942,
      "grad_norm": 1.1252537611441251,
      "learning_rate": 0.00029741650460306636,
      "loss": 1.0078,
      "step": 2636
    },
    {
      "epoch": 1.2409411764705882,
      "grad_norm": 0.8535157362772592,
      "learning_rate": 0.0002974083975104283,
      "loss": 0.7148,
      "step": 2637
    },
    {
      "epoch": 1.2414117647058824,
      "grad_norm": 0.5900924741169118,
      "learning_rate": 0.00029740027782839003,
      "loss": 0.4941,
      "step": 2638
    },
    {
      "epoch": 1.2418823529411764,
      "grad_norm": 1.198193650698389,
      "learning_rate": 0.00029739214555764504,
      "loss": 1.2656,
      "step": 2639
    },
    {
      "epoch": 1.2423529411764707,
      "grad_norm": 0.6812497561360744,
      "learning_rate": 0.00029738400069888787,
      "loss": 0.6484,
      "step": 2640
    },
    {
      "epoch": 1.2428235294117647,
      "grad_norm": 0.6078888445051119,
      "learning_rate": 0.00029737584325281415,
      "loss": 0.3281,
      "step": 2641
    },
    {
      "epoch": 1.2432941176470589,
      "grad_norm": 0.7658282238339923,
      "learning_rate": 0.0002973676732201205,
      "loss": 0.5938,
      "step": 2642
    },
    {
      "epoch": 1.2437647058823529,
      "grad_norm": 0.8978349741603746,
      "learning_rate": 0.00029735949060150476,
      "loss": 0.7812,
      "step": 2643
    },
    {
      "epoch": 1.244235294117647,
      "grad_norm": 0.8858743821196974,
      "learning_rate": 0.0002973512953976657,
      "loss": 0.6406,
      "step": 2644
    },
    {
      "epoch": 1.244705882352941,
      "grad_norm": 0.48771746622573664,
      "learning_rate": 0.0002973430876093033,
      "loss": 0.1641,
      "step": 2645
    },
    {
      "epoch": 1.2451764705882353,
      "grad_norm": 1.097360260404042,
      "learning_rate": 0.00029733486723711845,
      "loss": 0.9219,
      "step": 2646
    },
    {
      "epoch": 1.2456470588235293,
      "grad_norm": 1.048583975544324,
      "learning_rate": 0.0002973266342818133,
      "loss": 1.1953,
      "step": 2647
    },
    {
      "epoch": 1.2461176470588236,
      "grad_norm": 0.9282833049035631,
      "learning_rate": 0.00029731838874409093,
      "loss": 0.9453,
      "step": 2648
    },
    {
      "epoch": 1.2465882352941176,
      "grad_norm": 1.2735599180001815,
      "learning_rate": 0.0002973101306246556,
      "loss": 0.7305,
      "step": 2649
    },
    {
      "epoch": 1.2470588235294118,
      "grad_norm": 0.6588503792921433,
      "learning_rate": 0.0002973018599242125,
      "loss": 0.3008,
      "step": 2650
    },
    {
      "epoch": 1.247529411764706,
      "grad_norm": 0.9340731420067002,
      "learning_rate": 0.0002972935766434681,
      "loss": 0.8047,
      "step": 2651
    },
    {
      "epoch": 1.248,
      "grad_norm": 0.28652657215182287,
      "learning_rate": 0.00029728528078312976,
      "loss": 0.0938,
      "step": 2652
    },
    {
      "epoch": 1.248470588235294,
      "grad_norm": 0.7141958816703997,
      "learning_rate": 0.00029727697234390604,
      "loss": 0.2969,
      "step": 2653
    },
    {
      "epoch": 1.2489411764705882,
      "grad_norm": 0.706962273007187,
      "learning_rate": 0.0002972686513265065,
      "loss": 0.3301,
      "step": 2654
    },
    {
      "epoch": 1.2494117647058824,
      "grad_norm": 0.5119710692767419,
      "learning_rate": 0.00029726031773164176,
      "loss": 0.4004,
      "step": 2655
    },
    {
      "epoch": 1.2498823529411764,
      "grad_norm": 1.2850965160333712,
      "learning_rate": 0.00029725197156002357,
      "loss": 0.7422,
      "step": 2656
    },
    {
      "epoch": 1.2503529411764707,
      "grad_norm": 0.9926014876619053,
      "learning_rate": 0.00029724361281236475,
      "loss": 0.6289,
      "step": 2657
    },
    {
      "epoch": 1.2508235294117647,
      "grad_norm": 0.7749949513451011,
      "learning_rate": 0.0002972352414893792,
      "loss": 0.4336,
      "step": 2658
    },
    {
      "epoch": 1.251294117647059,
      "grad_norm": 0.8990289827589055,
      "learning_rate": 0.00029722685759178187,
      "loss": 0.8125,
      "step": 2659
    },
    {
      "epoch": 1.251764705882353,
      "grad_norm": 4.864434532493067,
      "learning_rate": 0.0002972184611202886,
      "loss": 0.9922,
      "step": 2660
    },
    {
      "epoch": 1.2522352941176471,
      "grad_norm": 0.8491349653478893,
      "learning_rate": 0.00029721005207561683,
      "loss": 0.5586,
      "step": 2661
    },
    {
      "epoch": 1.2527058823529411,
      "grad_norm": 0.9657277626784888,
      "learning_rate": 0.00029720163045848444,
      "loss": 0.5,
      "step": 2662
    },
    {
      "epoch": 1.2531764705882353,
      "grad_norm": 0.9337918385149163,
      "learning_rate": 0.0002971931962696109,
      "loss": 0.6289,
      "step": 2663
    },
    {
      "epoch": 1.2536470588235293,
      "grad_norm": 0.7560817643749147,
      "learning_rate": 0.00029718474950971626,
      "loss": 0.064,
      "step": 2664
    },
    {
      "epoch": 1.2541176470588236,
      "grad_norm": 5.536652094751073,
      "learning_rate": 0.0002971762901795222,
      "loss": 1.1797,
      "step": 2665
    },
    {
      "epoch": 1.2545882352941176,
      "grad_norm": 0.8929352352070555,
      "learning_rate": 0.000297167818279751,
      "loss": 0.6875,
      "step": 2666
    },
    {
      "epoch": 1.2550588235294118,
      "grad_norm": 0.5076528536228933,
      "learning_rate": 0.0002971593338111263,
      "loss": 0.2217,
      "step": 2667
    },
    {
      "epoch": 1.255529411764706,
      "grad_norm": 0.9986204803388653,
      "learning_rate": 0.0002971508367743727,
      "loss": 0.6172,
      "step": 2668
    },
    {
      "epoch": 1.256,
      "grad_norm": 1.1387202944607056,
      "learning_rate": 0.0002971423271702158,
      "loss": 0.625,
      "step": 2669
    },
    {
      "epoch": 1.256470588235294,
      "grad_norm": 0.9665190055490297,
      "learning_rate": 0.00029713380499938246,
      "loss": 0.7461,
      "step": 2670
    },
    {
      "epoch": 1.2569411764705882,
      "grad_norm": 1.627485079560235,
      "learning_rate": 0.00029712527026260047,
      "loss": 1.2734,
      "step": 2671
    },
    {
      "epoch": 1.2574117647058825,
      "grad_norm": 0.9148301254032601,
      "learning_rate": 0.0002971167229605988,
      "loss": 1.1328,
      "step": 2672
    },
    {
      "epoch": 1.2578823529411765,
      "grad_norm": 0.7554235624879997,
      "learning_rate": 0.00029710816309410737,
      "loss": 0.6484,
      "step": 2673
    },
    {
      "epoch": 1.2583529411764705,
      "grad_norm": 1.0806072340137773,
      "learning_rate": 0.00029709959066385723,
      "loss": 0.6445,
      "step": 2674
    },
    {
      "epoch": 1.2588235294117647,
      "grad_norm": 1.6547852364661284,
      "learning_rate": 0.0002970910056705806,
      "loss": 0.8867,
      "step": 2675
    },
    {
      "epoch": 1.259294117647059,
      "grad_norm": 1.3294114965135027,
      "learning_rate": 0.0002970824081150105,
      "loss": 0.9844,
      "step": 2676
    },
    {
      "epoch": 1.259764705882353,
      "grad_norm": 0.43356826101810003,
      "learning_rate": 0.00029707379799788134,
      "loss": 0.2041,
      "step": 2677
    },
    {
      "epoch": 1.2602352941176471,
      "grad_norm": 0.5887983159862414,
      "learning_rate": 0.0002970651753199285,
      "loss": 0.2695,
      "step": 2678
    },
    {
      "epoch": 1.2607058823529411,
      "grad_norm": 0.9091219822821306,
      "learning_rate": 0.0002970565400818883,
      "loss": 0.6523,
      "step": 2679
    },
    {
      "epoch": 1.2611764705882353,
      "grad_norm": 0.6907707753434427,
      "learning_rate": 0.0002970478922844983,
      "loss": 0.2393,
      "step": 2680
    },
    {
      "epoch": 1.2616470588235293,
      "grad_norm": 1.3754442091224768,
      "learning_rate": 0.00029703923192849694,
      "loss": 0.6055,
      "step": 2681
    },
    {
      "epoch": 1.2621176470588236,
      "grad_norm": 1.5276656052970197,
      "learning_rate": 0.000297030559014624,
      "loss": 0.9297,
      "step": 2682
    },
    {
      "epoch": 1.2625882352941176,
      "grad_norm": 0.964010762496715,
      "learning_rate": 0.00029702187354362016,
      "loss": 0.6797,
      "step": 2683
    },
    {
      "epoch": 1.2630588235294118,
      "grad_norm": 0.8443567129213929,
      "learning_rate": 0.0002970131755162272,
      "loss": 0.9258,
      "step": 2684
    },
    {
      "epoch": 1.263529411764706,
      "grad_norm": 0.8928413065895424,
      "learning_rate": 0.0002970044649331879,
      "loss": 0.4277,
      "step": 2685
    },
    {
      "epoch": 1.264,
      "grad_norm": 0.6921804895003827,
      "learning_rate": 0.0002969957417952463,
      "loss": 0.4141,
      "step": 2686
    },
    {
      "epoch": 1.264470588235294,
      "grad_norm": 0.9228895218854314,
      "learning_rate": 0.00029698700610314725,
      "loss": 0.4297,
      "step": 2687
    },
    {
      "epoch": 1.2649411764705882,
      "grad_norm": 0.9624582715842527,
      "learning_rate": 0.000296978257857637,
      "loss": 0.4453,
      "step": 2688
    },
    {
      "epoch": 1.2654117647058825,
      "grad_norm": 1.7673442570750575,
      "learning_rate": 0.00029696949705946255,
      "loss": 0.8203,
      "step": 2689
    },
    {
      "epoch": 1.2658823529411765,
      "grad_norm": 1.0626169960802625,
      "learning_rate": 0.0002969607237093722,
      "loss": 0.75,
      "step": 2690
    },
    {
      "epoch": 1.2663529411764705,
      "grad_norm": 0.6442936628310021,
      "learning_rate": 0.00029695193780811523,
      "loss": 0.3574,
      "step": 2691
    },
    {
      "epoch": 1.2668235294117647,
      "grad_norm": 0.9155095807526917,
      "learning_rate": 0.0002969431393564419,
      "loss": 0.6797,
      "step": 2692
    },
    {
      "epoch": 1.267294117647059,
      "grad_norm": 0.7116445387354416,
      "learning_rate": 0.0002969343283551038,
      "loss": 0.373,
      "step": 2693
    },
    {
      "epoch": 1.267764705882353,
      "grad_norm": 0.8281587673217304,
      "learning_rate": 0.0002969255048048533,
      "loss": 0.3086,
      "step": 2694
    },
    {
      "epoch": 1.2682352941176471,
      "grad_norm": 1.862341492982116,
      "learning_rate": 0.00029691666870644405,
      "loss": 1.0469,
      "step": 2695
    },
    {
      "epoch": 1.2687058823529411,
      "grad_norm": 0.7012911195090643,
      "learning_rate": 0.0002969078200606307,
      "loss": 0.3281,
      "step": 2696
    },
    {
      "epoch": 1.2691764705882354,
      "grad_norm": 0.3685416866805656,
      "learning_rate": 0.0002968989588681688,
      "loss": 0.2578,
      "step": 2697
    },
    {
      "epoch": 1.2696470588235294,
      "grad_norm": 0.7689817888123366,
      "learning_rate": 0.0002968900851298154,
      "loss": 0.5039,
      "step": 2698
    },
    {
      "epoch": 1.2701176470588236,
      "grad_norm": 1.0135941980405456,
      "learning_rate": 0.00029688119884632827,
      "loss": 0.7695,
      "step": 2699
    },
    {
      "epoch": 1.2705882352941176,
      "grad_norm": 0.5121456194489702,
      "learning_rate": 0.0002968723000184662,
      "loss": 0.2617,
      "step": 2700
    },
    {
      "epoch": 1.2710588235294118,
      "grad_norm": 1.1044209359641681,
      "learning_rate": 0.0002968633886469894,
      "loss": 0.8828,
      "step": 2701
    },
    {
      "epoch": 1.271529411764706,
      "grad_norm": 1.0971941413393942,
      "learning_rate": 0.00029685446473265876,
      "loss": 1.1016,
      "step": 2702
    },
    {
      "epoch": 1.272,
      "grad_norm": 0.5449357884761746,
      "learning_rate": 0.00029684552827623657,
      "loss": 0.2275,
      "step": 2703
    },
    {
      "epoch": 1.272470588235294,
      "grad_norm": 0.546097056811807,
      "learning_rate": 0.000296836579278486,
      "loss": 0.4824,
      "step": 2704
    },
    {
      "epoch": 1.2729411764705882,
      "grad_norm": 0.6117549961560325,
      "learning_rate": 0.0002968276177401713,
      "loss": 0.3496,
      "step": 2705
    },
    {
      "epoch": 1.2734117647058825,
      "grad_norm": 1.3121778079105282,
      "learning_rate": 0.00029681864366205786,
      "loss": 0.9766,
      "step": 2706
    },
    {
      "epoch": 1.2738823529411765,
      "grad_norm": 0.8048176203376181,
      "learning_rate": 0.0002968096570449121,
      "loss": 0.5391,
      "step": 2707
    },
    {
      "epoch": 1.2743529411764705,
      "grad_norm": 0.6587161289203706,
      "learning_rate": 0.0002968006578895016,
      "loss": 0.4277,
      "step": 2708
    },
    {
      "epoch": 1.2748235294117647,
      "grad_norm": 0.6453668369960551,
      "learning_rate": 0.0002967916461965948,
      "loss": 0.2432,
      "step": 2709
    },
    {
      "epoch": 1.275294117647059,
      "grad_norm": 0.8127677008902098,
      "learning_rate": 0.00029678262196696143,
      "loss": 0.7383,
      "step": 2710
    },
    {
      "epoch": 1.275764705882353,
      "grad_norm": 0.7474351881716204,
      "learning_rate": 0.0002967735852013722,
      "loss": 0.3633,
      "step": 2711
    },
    {
      "epoch": 1.2762352941176471,
      "grad_norm": 0.7673217574712129,
      "learning_rate": 0.0002967645359005988,
      "loss": 0.5391,
      "step": 2712
    },
    {
      "epoch": 1.2767058823529411,
      "grad_norm": 0.690502804228809,
      "learning_rate": 0.0002967554740654142,
      "loss": 0.3496,
      "step": 2713
    },
    {
      "epoch": 1.2771764705882354,
      "grad_norm": 0.584336850143643,
      "learning_rate": 0.0002967463996965923,
      "loss": 0.2793,
      "step": 2714
    },
    {
      "epoch": 1.2776470588235294,
      "grad_norm": 1.2964091009804084,
      "learning_rate": 0.0002967373127949081,
      "loss": 1.0781,
      "step": 2715
    },
    {
      "epoch": 1.2781176470588236,
      "grad_norm": 0.8685553311567316,
      "learning_rate": 0.0002967282133611375,
      "loss": 0.707,
      "step": 2716
    },
    {
      "epoch": 1.2785882352941176,
      "grad_norm": 1.1153419520610142,
      "learning_rate": 0.0002967191013960579,
      "loss": 0.9453,
      "step": 2717
    },
    {
      "epoch": 1.2790588235294118,
      "grad_norm": 0.9403667463972659,
      "learning_rate": 0.00029670997690044745,
      "loss": 0.8164,
      "step": 2718
    },
    {
      "epoch": 1.2795294117647058,
      "grad_norm": 0.628752674105734,
      "learning_rate": 0.0002967008398750853,
      "loss": 0.3379,
      "step": 2719
    },
    {
      "epoch": 1.28,
      "grad_norm": 0.8070324091942569,
      "learning_rate": 0.00029669169032075186,
      "loss": 0.3574,
      "step": 2720
    },
    {
      "epoch": 1.280470588235294,
      "grad_norm": 0.8069938067396034,
      "learning_rate": 0.0002966825282382285,
      "loss": 0.4316,
      "step": 2721
    },
    {
      "epoch": 1.2809411764705882,
      "grad_norm": 0.6720321139082609,
      "learning_rate": 0.0002966733536282978,
      "loss": 0.3457,
      "step": 2722
    },
    {
      "epoch": 1.2814117647058825,
      "grad_norm": 0.5674544129278745,
      "learning_rate": 0.0002966641664917433,
      "loss": 0.2305,
      "step": 2723
    },
    {
      "epoch": 1.2818823529411765,
      "grad_norm": 0.4467490367918732,
      "learning_rate": 0.0002966549668293495,
      "loss": 0.1836,
      "step": 2724
    },
    {
      "epoch": 1.2823529411764705,
      "grad_norm": 1.1545300678311918,
      "learning_rate": 0.0002966457546419023,
      "loss": 0.6367,
      "step": 2725
    },
    {
      "epoch": 1.2828235294117647,
      "grad_norm": 0.6463415722224337,
      "learning_rate": 0.0002966365299301883,
      "loss": 0.3203,
      "step": 2726
    },
    {
      "epoch": 1.283294117647059,
      "grad_norm": 1.0997624054293136,
      "learning_rate": 0.00029662729269499546,
      "loss": 0.8984,
      "step": 2727
    },
    {
      "epoch": 1.283764705882353,
      "grad_norm": 0.5525284028735346,
      "learning_rate": 0.00029661804293711254,
      "loss": 0.2598,
      "step": 2728
    },
    {
      "epoch": 1.2842352941176471,
      "grad_norm": 1.7198509840204008,
      "learning_rate": 0.00029660878065732965,
      "loss": 1.125,
      "step": 2729
    },
    {
      "epoch": 1.2847058823529411,
      "grad_norm": 0.519035623092611,
      "learning_rate": 0.0002965995058564377,
      "loss": 0.1641,
      "step": 2730
    },
    {
      "epoch": 1.2851764705882354,
      "grad_norm": 1.017401741832296,
      "learning_rate": 0.000296590218535229,
      "loss": 0.625,
      "step": 2731
    },
    {
      "epoch": 1.2856470588235294,
      "grad_norm": 0.5863931530812708,
      "learning_rate": 0.00029658091869449657,
      "loss": 0.2539,
      "step": 2732
    },
    {
      "epoch": 1.2861176470588236,
      "grad_norm": 0.530041607156701,
      "learning_rate": 0.0002965716063350347,
      "loss": 0.209,
      "step": 2733
    },
    {
      "epoch": 1.2865882352941176,
      "grad_norm": 1.302175010475994,
      "learning_rate": 0.0002965622814576387,
      "loss": 1.0156,
      "step": 2734
    },
    {
      "epoch": 1.2870588235294118,
      "grad_norm": 1.2837707735975774,
      "learning_rate": 0.000296552944063105,
      "loss": 1.0625,
      "step": 2735
    },
    {
      "epoch": 1.2875294117647058,
      "grad_norm": 0.7662972082944669,
      "learning_rate": 0.0002965435941522311,
      "loss": 0.291,
      "step": 2736
    },
    {
      "epoch": 1.288,
      "grad_norm": 0.6333534105949379,
      "learning_rate": 0.00029653423172581534,
      "loss": 0.375,
      "step": 2737
    },
    {
      "epoch": 1.288470588235294,
      "grad_norm": 1.0728486068145056,
      "learning_rate": 0.0002965248567846575,
      "loss": 0.875,
      "step": 2738
    },
    {
      "epoch": 1.2889411764705883,
      "grad_norm": 0.7517865779389166,
      "learning_rate": 0.0002965154693295582,
      "loss": 0.7109,
      "step": 2739
    },
    {
      "epoch": 1.2894117647058825,
      "grad_norm": 0.41318595435019834,
      "learning_rate": 0.00029650606936131913,
      "loss": 0.332,
      "step": 2740
    },
    {
      "epoch": 1.2898823529411765,
      "grad_norm": 0.9966632096713004,
      "learning_rate": 0.0002964966568807432,
      "loss": 0.9375,
      "step": 2741
    },
    {
      "epoch": 1.2903529411764705,
      "grad_norm": 0.9353243227130541,
      "learning_rate": 0.00029648723188863417,
      "loss": 0.5781,
      "step": 2742
    },
    {
      "epoch": 1.2908235294117647,
      "grad_norm": 1.1667261982210368,
      "learning_rate": 0.000296477794385797,
      "loss": 0.6406,
      "step": 2743
    },
    {
      "epoch": 1.291294117647059,
      "grad_norm": 0.9000241375744734,
      "learning_rate": 0.00029646834437303777,
      "loss": 0.625,
      "step": 2744
    },
    {
      "epoch": 1.291764705882353,
      "grad_norm": 0.7977374966271644,
      "learning_rate": 0.00029645888185116353,
      "loss": 0.4219,
      "step": 2745
    },
    {
      "epoch": 1.292235294117647,
      "grad_norm": 0.9276109212784219,
      "learning_rate": 0.00029644940682098234,
      "loss": 0.6367,
      "step": 2746
    },
    {
      "epoch": 1.2927058823529411,
      "grad_norm": 1.6898546160050891,
      "learning_rate": 0.00029643991928330346,
      "loss": 0.8203,
      "step": 2747
    },
    {
      "epoch": 1.2931764705882354,
      "grad_norm": 0.9862721734771597,
      "learning_rate": 0.0002964304192389372,
      "loss": 0.7148,
      "step": 2748
    },
    {
      "epoch": 1.2936470588235294,
      "grad_norm": 1.1824252525845156,
      "learning_rate": 0.000296420906688695,
      "loss": 0.7188,
      "step": 2749
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 1.038265731742481,
      "learning_rate": 0.00029641138163338907,
      "loss": 0.7305,
      "step": 2750
    },
    {
      "epoch": 1.2945882352941176,
      "grad_norm": 0.6272565415053639,
      "learning_rate": 0.00029640184407383303,
      "loss": 0.2871,
      "step": 2751
    },
    {
      "epoch": 1.2950588235294118,
      "grad_norm": 0.8613505181273008,
      "learning_rate": 0.00029639229401084137,
      "loss": 0.5664,
      "step": 2752
    },
    {
      "epoch": 1.2955294117647058,
      "grad_norm": 1.0346138205132709,
      "learning_rate": 0.00029638273144522983,
      "loss": 0.4766,
      "step": 2753
    },
    {
      "epoch": 1.296,
      "grad_norm": 0.9509896172969328,
      "learning_rate": 0.00029637315637781494,
      "loss": 0.7656,
      "step": 2754
    },
    {
      "epoch": 1.296470588235294,
      "grad_norm": 0.6073502883376412,
      "learning_rate": 0.0002963635688094146,
      "loss": 0.3496,
      "step": 2755
    },
    {
      "epoch": 1.2969411764705883,
      "grad_norm": 0.45042422850428476,
      "learning_rate": 0.0002963539687408476,
      "loss": 0.1699,
      "step": 2756
    },
    {
      "epoch": 1.2974117647058825,
      "grad_norm": 0.6340699237059022,
      "learning_rate": 0.00029634435617293374,
      "loss": 0.5117,
      "step": 2757
    },
    {
      "epoch": 1.2978823529411765,
      "grad_norm": 1.3700803872442526,
      "learning_rate": 0.00029633473110649407,
      "loss": 1.2266,
      "step": 2758
    },
    {
      "epoch": 1.2983529411764705,
      "grad_norm": 0.679715521867732,
      "learning_rate": 0.00029632509354235063,
      "loss": 0.5625,
      "step": 2759
    },
    {
      "epoch": 1.2988235294117647,
      "grad_norm": 0.6711083579421047,
      "learning_rate": 0.00029631544348132643,
      "loss": 0.5859,
      "step": 2760
    },
    {
      "epoch": 1.299294117647059,
      "grad_norm": 0.6468861267598254,
      "learning_rate": 0.0002963057809242457,
      "loss": 0.2451,
      "step": 2761
    },
    {
      "epoch": 1.299764705882353,
      "grad_norm": 1.5686564749317908,
      "learning_rate": 0.00029629610587193365,
      "loss": 1.2031,
      "step": 2762
    },
    {
      "epoch": 1.300235294117647,
      "grad_norm": 1.3259829399866185,
      "learning_rate": 0.00029628641832521663,
      "loss": 1.3203,
      "step": 2763
    },
    {
      "epoch": 1.3007058823529412,
      "grad_norm": 0.5624237449340347,
      "learning_rate": 0.0002962767182849219,
      "loss": 0.2969,
      "step": 2764
    },
    {
      "epoch": 1.3011764705882354,
      "grad_norm": 0.9357857105082366,
      "learning_rate": 0.00029626700575187794,
      "loss": 0.3887,
      "step": 2765
    },
    {
      "epoch": 1.3016470588235294,
      "grad_norm": 1.0244537514939236,
      "learning_rate": 0.0002962572807269143,
      "loss": 0.6836,
      "step": 2766
    },
    {
      "epoch": 1.3021176470588236,
      "grad_norm": 0.8067140835106326,
      "learning_rate": 0.00029624754321086146,
      "loss": 0.4805,
      "step": 2767
    },
    {
      "epoch": 1.3025882352941176,
      "grad_norm": 1.168230630268414,
      "learning_rate": 0.00029623779320455106,
      "loss": 0.8945,
      "step": 2768
    },
    {
      "epoch": 1.3030588235294118,
      "grad_norm": 0.975931220777782,
      "learning_rate": 0.00029622803070881586,
      "loss": 0.7773,
      "step": 2769
    },
    {
      "epoch": 1.3035294117647058,
      "grad_norm": 0.41258945228037763,
      "learning_rate": 0.00029621825572448965,
      "loss": 0.1416,
      "step": 2770
    },
    {
      "epoch": 1.304,
      "grad_norm": 0.7689526465326538,
      "learning_rate": 0.0002962084682524071,
      "loss": 0.6133,
      "step": 2771
    },
    {
      "epoch": 1.304470588235294,
      "grad_norm": 0.5938351717732433,
      "learning_rate": 0.0002961986682934043,
      "loss": 0.2559,
      "step": 2772
    },
    {
      "epoch": 1.3049411764705883,
      "grad_norm": 1.2459026143128908,
      "learning_rate": 0.0002961888558483181,
      "loss": 0.8828,
      "step": 2773
    },
    {
      "epoch": 1.3054117647058823,
      "grad_norm": 0.7904967396539176,
      "learning_rate": 0.0002961790309179866,
      "loss": 0.5352,
      "step": 2774
    },
    {
      "epoch": 1.3058823529411765,
      "grad_norm": 0.8898754760063249,
      "learning_rate": 0.00029616919350324873,
      "loss": 0.8203,
      "step": 2775
    },
    {
      "epoch": 1.3063529411764705,
      "grad_norm": 0.6153429350358564,
      "learning_rate": 0.0002961593436049449,
      "loss": 0.3906,
      "step": 2776
    },
    {
      "epoch": 1.3068235294117647,
      "grad_norm": 1.2830111152950718,
      "learning_rate": 0.0002961494812239162,
      "loss": 1.1172,
      "step": 2777
    },
    {
      "epoch": 1.307294117647059,
      "grad_norm": 0.6510649622917385,
      "learning_rate": 0.00029613960636100487,
      "loss": 0.373,
      "step": 2778
    },
    {
      "epoch": 1.307764705882353,
      "grad_norm": 1.1891767254952077,
      "learning_rate": 0.0002961297190170544,
      "loss": 0.5938,
      "step": 2779
    },
    {
      "epoch": 1.308235294117647,
      "grad_norm": 0.9889594343611078,
      "learning_rate": 0.00029611981919290916,
      "loss": 0.5859,
      "step": 2780
    },
    {
      "epoch": 1.3087058823529412,
      "grad_norm": 0.6545808557708029,
      "learning_rate": 0.0002961099068894146,
      "loss": 0.2832,
      "step": 2781
    },
    {
      "epoch": 1.3091764705882354,
      "grad_norm": 0.7060075282182445,
      "learning_rate": 0.00029609998210741744,
      "loss": 0.3066,
      "step": 2782
    },
    {
      "epoch": 1.3096470588235294,
      "grad_norm": 1.1292319152262216,
      "learning_rate": 0.0002960900448477651,
      "loss": 0.7812,
      "step": 2783
    },
    {
      "epoch": 1.3101176470588236,
      "grad_norm": 0.14298920797519862,
      "learning_rate": 0.0002960800951113064,
      "loss": 0.0552,
      "step": 2784
    },
    {
      "epoch": 1.3105882352941176,
      "grad_norm": 0.5512650977629816,
      "learning_rate": 0.00029607013289889106,
      "loss": 0.3633,
      "step": 2785
    },
    {
      "epoch": 1.3110588235294118,
      "grad_norm": 0.9323856969461578,
      "learning_rate": 0.0002960601582113699,
      "loss": 0.7578,
      "step": 2786
    },
    {
      "epoch": 1.3115294117647058,
      "grad_norm": 1.143124220236149,
      "learning_rate": 0.00029605017104959475,
      "loss": 0.6094,
      "step": 2787
    },
    {
      "epoch": 1.312,
      "grad_norm": 0.44458231079298866,
      "learning_rate": 0.00029604017141441865,
      "loss": 0.2656,
      "step": 2788
    },
    {
      "epoch": 1.312470588235294,
      "grad_norm": 0.9434955889342105,
      "learning_rate": 0.0002960301593066956,
      "loss": 0.6328,
      "step": 2789
    },
    {
      "epoch": 1.3129411764705883,
      "grad_norm": 0.4186710334338029,
      "learning_rate": 0.0002960201347272807,
      "loss": 0.2363,
      "step": 2790
    },
    {
      "epoch": 1.3134117647058823,
      "grad_norm": 0.6419504353522757,
      "learning_rate": 0.00029601009767703007,
      "loss": 0.2598,
      "step": 2791
    },
    {
      "epoch": 1.3138823529411765,
      "grad_norm": 0.9903456162804762,
      "learning_rate": 0.0002960000481568009,
      "loss": 0.9336,
      "step": 2792
    },
    {
      "epoch": 1.3143529411764705,
      "grad_norm": 1.2675823847398842,
      "learning_rate": 0.0002959899861674515,
      "loss": 0.9531,
      "step": 2793
    },
    {
      "epoch": 1.3148235294117647,
      "grad_norm": 0.49431416105844456,
      "learning_rate": 0.0002959799117098412,
      "loss": 0.25,
      "step": 2794
    },
    {
      "epoch": 1.315294117647059,
      "grad_norm": 1.061255450882752,
      "learning_rate": 0.00029596982478483037,
      "loss": 0.8477,
      "step": 2795
    },
    {
      "epoch": 1.315764705882353,
      "grad_norm": 0.428458855468648,
      "learning_rate": 0.00029595972539328053,
      "loss": 0.3184,
      "step": 2796
    },
    {
      "epoch": 1.316235294117647,
      "grad_norm": 0.8060977230741155,
      "learning_rate": 0.00029594961353605426,
      "loss": 0.6016,
      "step": 2797
    },
    {
      "epoch": 1.3167058823529412,
      "grad_norm": 0.8228540233005409,
      "learning_rate": 0.0002959394892140151,
      "loss": 0.4961,
      "step": 2798
    },
    {
      "epoch": 1.3171764705882354,
      "grad_norm": 0.33833972539250434,
      "learning_rate": 0.00029592935242802774,
      "loss": 0.1963,
      "step": 2799
    },
    {
      "epoch": 1.3176470588235294,
      "grad_norm": 1.023498404533829,
      "learning_rate": 0.0002959192031789579,
      "loss": 0.6016,
      "step": 2800
    },
    {
      "epoch": 1.3181176470588234,
      "grad_norm": 0.8650788444056418,
      "learning_rate": 0.00029590904146767236,
      "loss": 0.7578,
      "step": 2801
    },
    {
      "epoch": 1.3185882352941176,
      "grad_norm": 0.9596498633070935,
      "learning_rate": 0.000295898867295039,
      "loss": 0.6094,
      "step": 2802
    },
    {
      "epoch": 1.3190588235294118,
      "grad_norm": 0.6065837923239519,
      "learning_rate": 0.0002958886806619268,
      "loss": 0.2354,
      "step": 2803
    },
    {
      "epoch": 1.3195294117647058,
      "grad_norm": 0.5538181251331912,
      "learning_rate": 0.00029587848156920564,
      "loss": 0.2578,
      "step": 2804
    },
    {
      "epoch": 1.32,
      "grad_norm": 0.5924317644133589,
      "learning_rate": 0.0002958682700177466,
      "loss": 0.2598,
      "step": 2805
    },
    {
      "epoch": 1.320470588235294,
      "grad_norm": 1.0066811654200514,
      "learning_rate": 0.00029585804600842187,
      "loss": 0.7617,
      "step": 2806
    },
    {
      "epoch": 1.3209411764705883,
      "grad_norm": 1.0070444864414556,
      "learning_rate": 0.00029584780954210454,
      "loss": 0.8516,
      "step": 2807
    },
    {
      "epoch": 1.3214117647058823,
      "grad_norm": 0.8261372104885344,
      "learning_rate": 0.0002958375606196689,
      "loss": 0.5039,
      "step": 2808
    },
    {
      "epoch": 1.3218823529411765,
      "grad_norm": 0.9144271554710904,
      "learning_rate": 0.0002958272992419903,
      "loss": 0.8711,
      "step": 2809
    },
    {
      "epoch": 1.3223529411764705,
      "grad_norm": 0.9746869530157763,
      "learning_rate": 0.000295817025409945,
      "loss": 0.5742,
      "step": 2810
    },
    {
      "epoch": 1.3228235294117647,
      "grad_norm": 1.0567261443511875,
      "learning_rate": 0.00029580673912441057,
      "loss": 0.7773,
      "step": 2811
    },
    {
      "epoch": 1.323294117647059,
      "grad_norm": 1.0697889031126648,
      "learning_rate": 0.00029579644038626546,
      "loss": 1.25,
      "step": 2812
    },
    {
      "epoch": 1.323764705882353,
      "grad_norm": 0.533372928355867,
      "learning_rate": 0.00029578612919638913,
      "loss": 0.2715,
      "step": 2813
    },
    {
      "epoch": 1.324235294117647,
      "grad_norm": 1.2417928906001188,
      "learning_rate": 0.0002957758055556623,
      "loss": 0.8398,
      "step": 2814
    },
    {
      "epoch": 1.3247058823529412,
      "grad_norm": 0.41036489677107446,
      "learning_rate": 0.0002957654694649667,
      "loss": 0.3223,
      "step": 2815
    },
    {
      "epoch": 1.3251764705882354,
      "grad_norm": 0.6114754868671489,
      "learning_rate": 0.000295755120925185,
      "loss": 0.25,
      "step": 2816
    },
    {
      "epoch": 1.3256470588235294,
      "grad_norm": 0.7634258448311515,
      "learning_rate": 0.0002957447599372011,
      "loss": 0.3086,
      "step": 2817
    },
    {
      "epoch": 1.3261176470588234,
      "grad_norm": 0.8050948762455548,
      "learning_rate": 0.00029573438650189974,
      "loss": 0.6406,
      "step": 2818
    },
    {
      "epoch": 1.3265882352941176,
      "grad_norm": 0.7532202029139122,
      "learning_rate": 0.000295724000620167,
      "loss": 0.2812,
      "step": 2819
    },
    {
      "epoch": 1.3270588235294118,
      "grad_norm": 0.5067769858225711,
      "learning_rate": 0.00029571360229288984,
      "loss": 0.1514,
      "step": 2820
    },
    {
      "epoch": 1.3275294117647058,
      "grad_norm": 0.9519375573059763,
      "learning_rate": 0.0002957031915209563,
      "loss": 0.7031,
      "step": 2821
    },
    {
      "epoch": 1.328,
      "grad_norm": 1.2217798922989802,
      "learning_rate": 0.00029569276830525553,
      "loss": 1.0312,
      "step": 2822
    },
    {
      "epoch": 1.328470588235294,
      "grad_norm": 0.638934485563932,
      "learning_rate": 0.00029568233264667775,
      "loss": 0.5156,
      "step": 2823
    },
    {
      "epoch": 1.3289411764705883,
      "grad_norm": 0.7605645067587199,
      "learning_rate": 0.00029567188454611415,
      "loss": 0.7109,
      "step": 2824
    },
    {
      "epoch": 1.3294117647058823,
      "grad_norm": 0.36383576272714435,
      "learning_rate": 0.00029566142400445716,
      "loss": 0.1396,
      "step": 2825
    },
    {
      "epoch": 1.3298823529411765,
      "grad_norm": 0.6930356741372713,
      "learning_rate": 0.0002956509510226001,
      "loss": 0.5078,
      "step": 2826
    },
    {
      "epoch": 1.3303529411764705,
      "grad_norm": 0.9107803001088586,
      "learning_rate": 0.00029564046560143735,
      "loss": 0.7227,
      "step": 2827
    },
    {
      "epoch": 1.3308235294117647,
      "grad_norm": 0.7244155089236692,
      "learning_rate": 0.0002956299677418645,
      "loss": 0.5,
      "step": 2828
    },
    {
      "epoch": 1.331294117647059,
      "grad_norm": 1.1508048213202031,
      "learning_rate": 0.000295619457444778,
      "loss": 0.9062,
      "step": 2829
    },
    {
      "epoch": 1.331764705882353,
      "grad_norm": 0.7987878156432573,
      "learning_rate": 0.00029560893471107573,
      "loss": 0.5938,
      "step": 2830
    },
    {
      "epoch": 1.332235294117647,
      "grad_norm": 0.8410106601537884,
      "learning_rate": 0.00029559839954165616,
      "loss": 0.8281,
      "step": 2831
    },
    {
      "epoch": 1.3327058823529412,
      "grad_norm": 0.6283304028619213,
      "learning_rate": 0.00029558785193741914,
      "loss": 0.3809,
      "step": 2832
    },
    {
      "epoch": 1.3331764705882354,
      "grad_norm": 1.820471771539865,
      "learning_rate": 0.0002955772918992654,
      "loss": 1.3672,
      "step": 2833
    },
    {
      "epoch": 1.3336470588235294,
      "grad_norm": 0.8512230731863196,
      "learning_rate": 0.00029556671942809694,
      "loss": 0.377,
      "step": 2834
    },
    {
      "epoch": 1.3341176470588234,
      "grad_norm": 0.55518411042656,
      "learning_rate": 0.00029555613452481657,
      "loss": 0.3535,
      "step": 2835
    },
    {
      "epoch": 1.3345882352941176,
      "grad_norm": 1.2256029995174744,
      "learning_rate": 0.00029554553719032844,
      "loss": 1.2031,
      "step": 2836
    },
    {
      "epoch": 1.3350588235294119,
      "grad_norm": 0.705927984588325,
      "learning_rate": 0.00029553492742553754,
      "loss": 0.4121,
      "step": 2837
    },
    {
      "epoch": 1.3355294117647059,
      "grad_norm": 0.8405313232457731,
      "learning_rate": 0.00029552430523134993,
      "loss": 0.6055,
      "step": 2838
    },
    {
      "epoch": 1.336,
      "grad_norm": 0.42227879595816054,
      "learning_rate": 0.0002955136706086729,
      "loss": 0.2109,
      "step": 2839
    },
    {
      "epoch": 1.336470588235294,
      "grad_norm": 1.100637199596241,
      "learning_rate": 0.0002955030235584146,
      "loss": 1.1016,
      "step": 2840
    },
    {
      "epoch": 1.3369411764705883,
      "grad_norm": 0.6878506495226467,
      "learning_rate": 0.00029549236408148453,
      "loss": 0.7227,
      "step": 2841
    },
    {
      "epoch": 1.3374117647058823,
      "grad_norm": 0.5890143025718589,
      "learning_rate": 0.00029548169217879285,
      "loss": 0.5664,
      "step": 2842
    },
    {
      "epoch": 1.3378823529411765,
      "grad_norm": 0.38298003331348907,
      "learning_rate": 0.000295471007851251,
      "loss": 0.1553,
      "step": 2843
    },
    {
      "epoch": 1.3383529411764705,
      "grad_norm": 0.5748058434849105,
      "learning_rate": 0.00029546031109977164,
      "loss": 0.2617,
      "step": 2844
    },
    {
      "epoch": 1.3388235294117647,
      "grad_norm": 0.8046328304889785,
      "learning_rate": 0.00029544960192526824,
      "loss": 0.3926,
      "step": 2845
    },
    {
      "epoch": 1.3392941176470587,
      "grad_norm": 1.2890053746888126,
      "learning_rate": 0.0002954388803286554,
      "loss": 0.6211,
      "step": 2846
    },
    {
      "epoch": 1.339764705882353,
      "grad_norm": 1.421052132564019,
      "learning_rate": 0.0002954281463108488,
      "loss": 1.4688,
      "step": 2847
    },
    {
      "epoch": 1.340235294117647,
      "grad_norm": 0.9927702201944428,
      "learning_rate": 0.0002954173998727652,
      "loss": 0.7031,
      "step": 2848
    },
    {
      "epoch": 1.3407058823529412,
      "grad_norm": 0.34601427182421246,
      "learning_rate": 0.00029540664101532227,
      "loss": 0.1377,
      "step": 2849
    },
    {
      "epoch": 1.3411764705882354,
      "grad_norm": 0.8188634685127872,
      "learning_rate": 0.0002953958697394391,
      "loss": 0.7422,
      "step": 2850
    },
    {
      "epoch": 1.3416470588235294,
      "grad_norm": 1.0009622973029124,
      "learning_rate": 0.0002953850860460355,
      "loss": 0.6758,
      "step": 2851
    },
    {
      "epoch": 1.3421176470588234,
      "grad_norm": 0.638732563038146,
      "learning_rate": 0.00029537428993603237,
      "loss": 0.4414,
      "step": 2852
    },
    {
      "epoch": 1.3425882352941176,
      "grad_norm": 1.1409263247912307,
      "learning_rate": 0.00029536348141035183,
      "loss": 0.7578,
      "step": 2853
    },
    {
      "epoch": 1.3430588235294119,
      "grad_norm": 1.5103610109592867,
      "learning_rate": 0.000295352660469917,
      "loss": 1.2812,
      "step": 2854
    },
    {
      "epoch": 1.3435294117647059,
      "grad_norm": 0.9187640166282875,
      "learning_rate": 0.00029534182711565203,
      "loss": 0.5156,
      "step": 2855
    },
    {
      "epoch": 1.3439999999999999,
      "grad_norm": 0.6583251079644065,
      "learning_rate": 0.0002953309813484821,
      "loss": 0.2891,
      "step": 2856
    },
    {
      "epoch": 1.344470588235294,
      "grad_norm": 0.7103923665348718,
      "learning_rate": 0.0002953201231693336,
      "loss": 0.3516,
      "step": 2857
    },
    {
      "epoch": 1.3449411764705883,
      "grad_norm": 0.6414557114245153,
      "learning_rate": 0.00029530925257913366,
      "loss": 0.334,
      "step": 2858
    },
    {
      "epoch": 1.3454117647058823,
      "grad_norm": 1.312465622376216,
      "learning_rate": 0.0002952983695788109,
      "loss": 1.0,
      "step": 2859
    },
    {
      "epoch": 1.3458823529411765,
      "grad_norm": 0.789839798611962,
      "learning_rate": 0.00029528747416929463,
      "loss": 0.5508,
      "step": 2860
    },
    {
      "epoch": 1.3463529411764705,
      "grad_norm": 0.9094568783729892,
      "learning_rate": 0.0002952765663515155,
      "loss": 0.4668,
      "step": 2861
    },
    {
      "epoch": 1.3468235294117648,
      "grad_norm": 0.9380178089123381,
      "learning_rate": 0.000295265646126405,
      "loss": 0.4551,
      "step": 2862
    },
    {
      "epoch": 1.3472941176470588,
      "grad_norm": 1.207212987123451,
      "learning_rate": 0.00029525471349489576,
      "loss": 0.8633,
      "step": 2863
    },
    {
      "epoch": 1.347764705882353,
      "grad_norm": 0.8321475566743822,
      "learning_rate": 0.00029524376845792153,
      "loss": 0.5703,
      "step": 2864
    },
    {
      "epoch": 1.348235294117647,
      "grad_norm": 0.8796857225047171,
      "learning_rate": 0.0002952328110164171,
      "loss": 0.7109,
      "step": 2865
    },
    {
      "epoch": 1.3487058823529412,
      "grad_norm": 0.9389762885129029,
      "learning_rate": 0.0002952218411713182,
      "loss": 0.9805,
      "step": 2866
    },
    {
      "epoch": 1.3491764705882354,
      "grad_norm": 0.816103900417602,
      "learning_rate": 0.0002952108589235618,
      "loss": 0.4531,
      "step": 2867
    },
    {
      "epoch": 1.3496470588235294,
      "grad_norm": 1.3217614885649491,
      "learning_rate": 0.0002951998642740858,
      "loss": 1.0156,
      "step": 2868
    },
    {
      "epoch": 1.3501176470588234,
      "grad_norm": 0.3025506074466686,
      "learning_rate": 0.00029518885722382913,
      "loss": 0.0537,
      "step": 2869
    },
    {
      "epoch": 1.3505882352941176,
      "grad_norm": 1.2896503892020457,
      "learning_rate": 0.00029517783777373194,
      "loss": 0.8242,
      "step": 2870
    },
    {
      "epoch": 1.3510588235294119,
      "grad_norm": 0.6725129398317088,
      "learning_rate": 0.0002951668059247353,
      "loss": 0.4629,
      "step": 2871
    },
    {
      "epoch": 1.3515294117647059,
      "grad_norm": 1.0046662419039085,
      "learning_rate": 0.0002951557616777814,
      "loss": 0.5938,
      "step": 2872
    },
    {
      "epoch": 1.3519999999999999,
      "grad_norm": 1.5522769794776243,
      "learning_rate": 0.0002951447050338135,
      "loss": 1.1719,
      "step": 2873
    },
    {
      "epoch": 1.352470588235294,
      "grad_norm": 0.8712784466685137,
      "learning_rate": 0.00029513363599377587,
      "loss": 0.7305,
      "step": 2874
    },
    {
      "epoch": 1.3529411764705883,
      "grad_norm": 0.6979078999024507,
      "learning_rate": 0.00029512255455861375,
      "loss": 0.6328,
      "step": 2875
    },
    {
      "epoch": 1.3534117647058823,
      "grad_norm": 0.6140151722171181,
      "learning_rate": 0.0002951114607292737,
      "loss": 0.3496,
      "step": 2876
    },
    {
      "epoch": 1.3538823529411765,
      "grad_norm": 0.8565499811137973,
      "learning_rate": 0.0002951003545067032,
      "loss": 0.6914,
      "step": 2877
    },
    {
      "epoch": 1.3543529411764705,
      "grad_norm": 0.13251488780950954,
      "learning_rate": 0.0002950892358918506,
      "loss": 0.0286,
      "step": 2878
    },
    {
      "epoch": 1.3548235294117648,
      "grad_norm": 1.0783206199369542,
      "learning_rate": 0.00029507810488566563,
      "loss": 1.2344,
      "step": 2879
    },
    {
      "epoch": 1.3552941176470588,
      "grad_norm": 0.942503475420565,
      "learning_rate": 0.000295066961489099,
      "loss": 0.5586,
      "step": 2880
    },
    {
      "epoch": 1.355764705882353,
      "grad_norm": 1.1051052876038272,
      "learning_rate": 0.0002950558057031022,
      "loss": 1.0859,
      "step": 2881
    },
    {
      "epoch": 1.356235294117647,
      "grad_norm": 0.872173430265138,
      "learning_rate": 0.0002950446375286281,
      "loss": 0.4941,
      "step": 2882
    },
    {
      "epoch": 1.3567058823529412,
      "grad_norm": 0.7582063975890123,
      "learning_rate": 0.00029503345696663054,
      "loss": 0.6523,
      "step": 2883
    },
    {
      "epoch": 1.3571764705882354,
      "grad_norm": 0.7848450265835281,
      "learning_rate": 0.0002950222640180644,
      "loss": 0.7266,
      "step": 2884
    },
    {
      "epoch": 1.3576470588235294,
      "grad_norm": 0.44447287762595367,
      "learning_rate": 0.0002950110586838855,
      "loss": 0.1387,
      "step": 2885
    },
    {
      "epoch": 1.3581176470588234,
      "grad_norm": 0.5436116853930372,
      "learning_rate": 0.00029499984096505107,
      "loss": 0.3945,
      "step": 2886
    },
    {
      "epoch": 1.3585882352941177,
      "grad_norm": 1.5809683103480308,
      "learning_rate": 0.00029498861086251887,
      "loss": 1.1562,
      "step": 2887
    },
    {
      "epoch": 1.3590588235294119,
      "grad_norm": 1.075157889433343,
      "learning_rate": 0.00029497736837724817,
      "loss": 0.7109,
      "step": 2888
    },
    {
      "epoch": 1.3595294117647059,
      "grad_norm": 0.45495867597525325,
      "learning_rate": 0.0002949661135101991,
      "loss": 0.1973,
      "step": 2889
    },
    {
      "epoch": 1.3599999999999999,
      "grad_norm": 0.5992927115927402,
      "learning_rate": 0.0002949548462623329,
      "loss": 0.3555,
      "step": 2890
    },
    {
      "epoch": 1.360470588235294,
      "grad_norm": 0.69426541148909,
      "learning_rate": 0.0002949435666346118,
      "loss": 0.3965,
      "step": 2891
    },
    {
      "epoch": 1.3609411764705883,
      "grad_norm": 1.1147924460925818,
      "learning_rate": 0.00029493227462799914,
      "loss": 0.7461,
      "step": 2892
    },
    {
      "epoch": 1.3614117647058823,
      "grad_norm": 0.7472755359477457,
      "learning_rate": 0.00029492097024345936,
      "loss": 0.5312,
      "step": 2893
    },
    {
      "epoch": 1.3618823529411765,
      "grad_norm": 0.780177266632701,
      "learning_rate": 0.00029490965348195794,
      "loss": 0.7344,
      "step": 2894
    },
    {
      "epoch": 1.3623529411764705,
      "grad_norm": 0.9761583289599205,
      "learning_rate": 0.00029489832434446126,
      "loss": 0.8555,
      "step": 2895
    },
    {
      "epoch": 1.3628235294117648,
      "grad_norm": 0.580020767676283,
      "learning_rate": 0.000294886982831937,
      "loss": 0.4277,
      "step": 2896
    },
    {
      "epoch": 1.3632941176470588,
      "grad_norm": 0.6221670107037656,
      "learning_rate": 0.00029487562894535375,
      "loss": 0.5938,
      "step": 2897
    },
    {
      "epoch": 1.363764705882353,
      "grad_norm": 0.8051307899224585,
      "learning_rate": 0.0002948642626856812,
      "loss": 0.7617,
      "step": 2898
    },
    {
      "epoch": 1.364235294117647,
      "grad_norm": 0.4299859317968286,
      "learning_rate": 0.00029485288405389,
      "loss": 0.1309,
      "step": 2899
    },
    {
      "epoch": 1.3647058823529412,
      "grad_norm": 0.8018324629140803,
      "learning_rate": 0.000294841493050952,
      "loss": 0.6094,
      "step": 2900
    },
    {
      "epoch": 1.3651764705882352,
      "grad_norm": 0.7909182137821784,
      "learning_rate": 0.00029483008967784007,
      "loss": 0.5586,
      "step": 2901
    },
    {
      "epoch": 1.3656470588235294,
      "grad_norm": 0.8582342328164578,
      "learning_rate": 0.0002948186739355281,
      "loss": 0.8164,
      "step": 2902
    },
    {
      "epoch": 1.3661176470588234,
      "grad_norm": 0.7971717233106302,
      "learning_rate": 0.00029480724582499107,
      "loss": 0.6719,
      "step": 2903
    },
    {
      "epoch": 1.3665882352941177,
      "grad_norm": 1.1684750531841481,
      "learning_rate": 0.0002947958053472049,
      "loss": 1.0234,
      "step": 2904
    },
    {
      "epoch": 1.3670588235294119,
      "grad_norm": 0.6369410324993975,
      "learning_rate": 0.0002947843525031468,
      "loss": 0.3086,
      "step": 2905
    },
    {
      "epoch": 1.3675294117647059,
      "grad_norm": 0.6641377453577186,
      "learning_rate": 0.0002947728872937948,
      "loss": 0.5117,
      "step": 2906
    },
    {
      "epoch": 1.3679999999999999,
      "grad_norm": 1.0185079265684214,
      "learning_rate": 0.00029476140972012815,
      "loss": 0.9609,
      "step": 2907
    },
    {
      "epoch": 1.368470588235294,
      "grad_norm": 0.6156340781521052,
      "learning_rate": 0.000294749919783127,
      "loss": 0.2168,
      "step": 2908
    },
    {
      "epoch": 1.3689411764705883,
      "grad_norm": 1.4201852105407229,
      "learning_rate": 0.00029473841748377277,
      "loss": 1.1172,
      "step": 2909
    },
    {
      "epoch": 1.3694117647058823,
      "grad_norm": 0.558510107980463,
      "learning_rate": 0.0002947269028230477,
      "loss": 0.3223,
      "step": 2910
    },
    {
      "epoch": 1.3698823529411766,
      "grad_norm": 0.5840545776591266,
      "learning_rate": 0.0002947153758019353,
      "loss": 0.25,
      "step": 2911
    },
    {
      "epoch": 1.3703529411764706,
      "grad_norm": 1.0035530846369083,
      "learning_rate": 0.00029470383642142,
      "loss": 0.8125,
      "step": 2912
    },
    {
      "epoch": 1.3708235294117648,
      "grad_norm": 0.49840391165915543,
      "learning_rate": 0.0002946922846824872,
      "loss": 0.3164,
      "step": 2913
    },
    {
      "epoch": 1.3712941176470588,
      "grad_norm": 0.7369458605779301,
      "learning_rate": 0.00029468072058612367,
      "loss": 0.4277,
      "step": 2914
    },
    {
      "epoch": 1.371764705882353,
      "grad_norm": 1.3125122567059815,
      "learning_rate": 0.0002946691441333169,
      "loss": 1.0234,
      "step": 2915
    },
    {
      "epoch": 1.372235294117647,
      "grad_norm": 0.862385358181648,
      "learning_rate": 0.0002946575553250556,
      "loss": 0.875,
      "step": 2916
    },
    {
      "epoch": 1.3727058823529412,
      "grad_norm": 0.7830274458589196,
      "learning_rate": 0.00029464595416232956,
      "loss": 0.5977,
      "step": 2917
    },
    {
      "epoch": 1.3731764705882352,
      "grad_norm": 0.38852825199656366,
      "learning_rate": 0.0002946343406461296,
      "loss": 0.2695,
      "step": 2918
    },
    {
      "epoch": 1.3736470588235294,
      "grad_norm": 0.6282534011418347,
      "learning_rate": 0.0002946227147774475,
      "loss": 0.1543,
      "step": 2919
    },
    {
      "epoch": 1.3741176470588234,
      "grad_norm": 0.9660308116978766,
      "learning_rate": 0.0002946110765572761,
      "loss": 0.6797,
      "step": 2920
    },
    {
      "epoch": 1.3745882352941177,
      "grad_norm": 0.37738050921581107,
      "learning_rate": 0.00029459942598660953,
      "loss": 0.0894,
      "step": 2921
    },
    {
      "epoch": 1.3750588235294119,
      "grad_norm": 1.3350741053986481,
      "learning_rate": 0.00029458776306644273,
      "loss": 0.8945,
      "step": 2922
    },
    {
      "epoch": 1.375529411764706,
      "grad_norm": 0.8580067208522447,
      "learning_rate": 0.0002945760877977717,
      "loss": 0.5977,
      "step": 2923
    },
    {
      "epoch": 1.376,
      "grad_norm": 0.5409364865931044,
      "learning_rate": 0.0002945644001815937,
      "loss": 0.2051,
      "step": 2924
    },
    {
      "epoch": 1.3764705882352941,
      "grad_norm": 1.219165709828649,
      "learning_rate": 0.0002945527002189068,
      "loss": 0.8359,
      "step": 2925
    },
    {
      "epoch": 1.3769411764705883,
      "grad_norm": 0.9451890753515141,
      "learning_rate": 0.0002945409879107103,
      "loss": 0.4551,
      "step": 2926
    },
    {
      "epoch": 1.3774117647058823,
      "grad_norm": 0.4848952763077009,
      "learning_rate": 0.00029452926325800445,
      "loss": 0.3281,
      "step": 2927
    },
    {
      "epoch": 1.3778823529411763,
      "grad_norm": 1.0882898387282771,
      "learning_rate": 0.0002945175262617906,
      "loss": 0.918,
      "step": 2928
    },
    {
      "epoch": 1.3783529411764706,
      "grad_norm": 0.9248774245733397,
      "learning_rate": 0.00029450577692307115,
      "loss": 0.5898,
      "step": 2929
    },
    {
      "epoch": 1.3788235294117648,
      "grad_norm": 0.8225236158241794,
      "learning_rate": 0.0002944940152428496,
      "loss": 0.668,
      "step": 2930
    },
    {
      "epoch": 1.3792941176470588,
      "grad_norm": 0.7362021133349334,
      "learning_rate": 0.0002944822412221303,
      "loss": 0.5039,
      "step": 2931
    },
    {
      "epoch": 1.379764705882353,
      "grad_norm": 0.5175440078492957,
      "learning_rate": 0.00029447045486191904,
      "loss": 0.4082,
      "step": 2932
    },
    {
      "epoch": 1.380235294117647,
      "grad_norm": 0.3645237235764819,
      "learning_rate": 0.0002944586561632223,
      "loss": 0.1235,
      "step": 2933
    },
    {
      "epoch": 1.3807058823529412,
      "grad_norm": 2.8999665449336383,
      "learning_rate": 0.0002944468451270477,
      "loss": 1.1953,
      "step": 2934
    },
    {
      "epoch": 1.3811764705882352,
      "grad_norm": 0.7417467674908269,
      "learning_rate": 0.000294435021754404,
      "loss": 0.4062,
      "step": 2935
    },
    {
      "epoch": 1.3816470588235295,
      "grad_norm": 0.7660853376514496,
      "learning_rate": 0.000294423186046301,
      "loss": 0.416,
      "step": 2936
    },
    {
      "epoch": 1.3821176470588235,
      "grad_norm": 1.3862512030485337,
      "learning_rate": 0.00029441133800374957,
      "loss": 1.0547,
      "step": 2937
    },
    {
      "epoch": 1.3825882352941177,
      "grad_norm": 0.5833051380514935,
      "learning_rate": 0.00029439947762776154,
      "loss": 0.25,
      "step": 2938
    },
    {
      "epoch": 1.383058823529412,
      "grad_norm": 1.2841377861864045,
      "learning_rate": 0.00029438760491934987,
      "loss": 0.9492,
      "step": 2939
    },
    {
      "epoch": 1.383529411764706,
      "grad_norm": 0.6360252161211858,
      "learning_rate": 0.0002943757198795284,
      "loss": 0.3418,
      "step": 2940
    },
    {
      "epoch": 1.384,
      "grad_norm": 0.3267382328320419,
      "learning_rate": 0.0002943638225093124,
      "loss": 0.1348,
      "step": 2941
    },
    {
      "epoch": 1.3844705882352941,
      "grad_norm": 1.1914288236321426,
      "learning_rate": 0.00029435191280971783,
      "loss": 0.7266,
      "step": 2942
    },
    {
      "epoch": 1.3849411764705883,
      "grad_norm": 0.36961360893039513,
      "learning_rate": 0.0002943399907817619,
      "loss": 0.1729,
      "step": 2943
    },
    {
      "epoch": 1.3854117647058823,
      "grad_norm": 0.7348855386032791,
      "learning_rate": 0.0002943280564264627,
      "loss": 0.291,
      "step": 2944
    },
    {
      "epoch": 1.3858823529411763,
      "grad_norm": 1.7390262816351378,
      "learning_rate": 0.0002943161097448396,
      "loss": 1.2266,
      "step": 2945
    },
    {
      "epoch": 1.3863529411764706,
      "grad_norm": 1.3836663519828079,
      "learning_rate": 0.0002943041507379128,
      "loss": 1.0703,
      "step": 2946
    },
    {
      "epoch": 1.3868235294117648,
      "grad_norm": 0.9418912258746135,
      "learning_rate": 0.00029429217940670383,
      "loss": 0.5898,
      "step": 2947
    },
    {
      "epoch": 1.3872941176470588,
      "grad_norm": 0.7807657572347532,
      "learning_rate": 0.0002942801957522349,
      "loss": 0.8359,
      "step": 2948
    },
    {
      "epoch": 1.387764705882353,
      "grad_norm": 0.4513621330858745,
      "learning_rate": 0.00029426819977552965,
      "loss": 0.208,
      "step": 2949
    },
    {
      "epoch": 1.388235294117647,
      "grad_norm": 0.9198045639183542,
      "learning_rate": 0.0002942561914776124,
      "loss": 0.7891,
      "step": 2950
    },
    {
      "epoch": 1.3887058823529412,
      "grad_norm": 0.4711583264567377,
      "learning_rate": 0.0002942441708595089,
      "loss": 0.2559,
      "step": 2951
    },
    {
      "epoch": 1.3891764705882352,
      "grad_norm": 1.2220371080783998,
      "learning_rate": 0.0002942321379222457,
      "loss": 0.8828,
      "step": 2952
    },
    {
      "epoch": 1.3896470588235295,
      "grad_norm": 0.8850807651339792,
      "learning_rate": 0.0002942200926668504,
      "loss": 0.6094,
      "step": 2953
    },
    {
      "epoch": 1.3901176470588235,
      "grad_norm": 0.5760669820706534,
      "learning_rate": 0.00029420803509435186,
      "loss": 0.21,
      "step": 2954
    },
    {
      "epoch": 1.3905882352941177,
      "grad_norm": 0.5493667584957393,
      "learning_rate": 0.00029419596520577975,
      "loss": 0.3496,
      "step": 2955
    },
    {
      "epoch": 1.391058823529412,
      "grad_norm": 0.8434235513865036,
      "learning_rate": 0.00029418388300216493,
      "loss": 0.6445,
      "step": 2956
    },
    {
      "epoch": 1.391529411764706,
      "grad_norm": 0.9485932691762506,
      "learning_rate": 0.0002941717884845393,
      "loss": 0.3906,
      "step": 2957
    },
    {
      "epoch": 1.392,
      "grad_norm": 0.5734614674709898,
      "learning_rate": 0.00029415968165393577,
      "loss": 0.4473,
      "step": 2958
    },
    {
      "epoch": 1.3924705882352941,
      "grad_norm": 0.7309064139103171,
      "learning_rate": 0.00029414756251138835,
      "loss": 0.3574,
      "step": 2959
    },
    {
      "epoch": 1.3929411764705883,
      "grad_norm": 0.989804173479212,
      "learning_rate": 0.000294135431057932,
      "loss": 0.5898,
      "step": 2960
    },
    {
      "epoch": 1.3934117647058823,
      "grad_norm": 0.8567220595190055,
      "learning_rate": 0.0002941232872946029,
      "loss": 0.6094,
      "step": 2961
    },
    {
      "epoch": 1.3938823529411764,
      "grad_norm": 0.5553339309103207,
      "learning_rate": 0.00029411113122243814,
      "loss": 0.25,
      "step": 2962
    },
    {
      "epoch": 1.3943529411764706,
      "grad_norm": 0.8083676700074488,
      "learning_rate": 0.0002940989628424759,
      "loss": 0.3887,
      "step": 2963
    },
    {
      "epoch": 1.3948235294117648,
      "grad_norm": 1.2530463369375542,
      "learning_rate": 0.0002940867821557554,
      "loss": 0.9375,
      "step": 2964
    },
    {
      "epoch": 1.3952941176470588,
      "grad_norm": 0.8872242618836018,
      "learning_rate": 0.00029407458916331705,
      "loss": 0.5391,
      "step": 2965
    },
    {
      "epoch": 1.395764705882353,
      "grad_norm": 0.49526527272108445,
      "learning_rate": 0.00029406238386620205,
      "loss": 0.2871,
      "step": 2966
    },
    {
      "epoch": 1.396235294117647,
      "grad_norm": 1.2333247290783367,
      "learning_rate": 0.0002940501662654528,
      "loss": 0.7305,
      "step": 2967
    },
    {
      "epoch": 1.3967058823529412,
      "grad_norm": 0.621001839972613,
      "learning_rate": 0.0002940379363621129,
      "loss": 0.3984,
      "step": 2968
    },
    {
      "epoch": 1.3971764705882352,
      "grad_norm": 0.7075036073192205,
      "learning_rate": 0.00029402569415722666,
      "loss": 0.4082,
      "step": 2969
    },
    {
      "epoch": 1.3976470588235295,
      "grad_norm": 0.39060068839959194,
      "learning_rate": 0.00029401343965183965,
      "loss": 0.1719,
      "step": 2970
    },
    {
      "epoch": 1.3981176470588235,
      "grad_norm": 1.2885703733368277,
      "learning_rate": 0.0002940011728469986,
      "loss": 0.8516,
      "step": 2971
    },
    {
      "epoch": 1.3985882352941177,
      "grad_norm": 0.9108088525714801,
      "learning_rate": 0.000293988893743751,
      "loss": 0.7344,
      "step": 2972
    },
    {
      "epoch": 1.3990588235294117,
      "grad_norm": 0.7847128915660587,
      "learning_rate": 0.0002939766023431457,
      "loss": 0.6602,
      "step": 2973
    },
    {
      "epoch": 1.399529411764706,
      "grad_norm": 0.8739150153414027,
      "learning_rate": 0.0002939642986462323,
      "loss": 0.6406,
      "step": 2974
    },
    {
      "epoch": 1.4,
      "grad_norm": 1.2300176080906047,
      "learning_rate": 0.0002939519826540616,
      "loss": 0.9102,
      "step": 2975
    },
    {
      "epoch": 1.4004705882352941,
      "grad_norm": 1.870659112354195,
      "learning_rate": 0.0002939396543676856,
      "loss": 1.75,
      "step": 2976
    },
    {
      "epoch": 1.4009411764705884,
      "grad_norm": 0.9121363669073584,
      "learning_rate": 0.00029392731378815706,
      "loss": 0.7383,
      "step": 2977
    },
    {
      "epoch": 1.4014117647058824,
      "grad_norm": 0.8712011831174732,
      "learning_rate": 0.00029391496091652993,
      "loss": 0.8281,
      "step": 2978
    },
    {
      "epoch": 1.4018823529411764,
      "grad_norm": 1.5793392542722946,
      "learning_rate": 0.0002939025957538593,
      "loss": 1.6719,
      "step": 2979
    },
    {
      "epoch": 1.4023529411764706,
      "grad_norm": 0.5553282012312283,
      "learning_rate": 0.0002938902183012011,
      "loss": 0.4434,
      "step": 2980
    },
    {
      "epoch": 1.4028235294117648,
      "grad_norm": 1.0644053803391733,
      "learning_rate": 0.00029387782855961253,
      "loss": 0.832,
      "step": 2981
    },
    {
      "epoch": 1.4032941176470588,
      "grad_norm": 0.951833330047515,
      "learning_rate": 0.00029386542653015167,
      "loss": 0.7461,
      "step": 2982
    },
    {
      "epoch": 1.4037647058823528,
      "grad_norm": 0.6199577452153509,
      "learning_rate": 0.0002938530122138777,
      "loss": 0.2891,
      "step": 2983
    },
    {
      "epoch": 1.404235294117647,
      "grad_norm": 0.889084790346587,
      "learning_rate": 0.0002938405856118509,
      "loss": 0.7969,
      "step": 2984
    },
    {
      "epoch": 1.4047058823529412,
      "grad_norm": 1.0097659638307928,
      "learning_rate": 0.0002938281467251326,
      "loss": 0.8711,
      "step": 2985
    },
    {
      "epoch": 1.4051764705882352,
      "grad_norm": 0.35485353360400457,
      "learning_rate": 0.00029381569555478505,
      "loss": 0.2256,
      "step": 2986
    },
    {
      "epoch": 1.4056470588235295,
      "grad_norm": 0.7211897850765567,
      "learning_rate": 0.00029380323210187166,
      "loss": 0.3457,
      "step": 2987
    },
    {
      "epoch": 1.4061176470588235,
      "grad_norm": 0.6827216294500663,
      "learning_rate": 0.00029379075636745694,
      "loss": 0.3086,
      "step": 2988
    },
    {
      "epoch": 1.4065882352941177,
      "grad_norm": 0.672184944768573,
      "learning_rate": 0.0002937782683526063,
      "loss": 0.4395,
      "step": 2989
    },
    {
      "epoch": 1.4070588235294117,
      "grad_norm": 0.36494565805619456,
      "learning_rate": 0.0002937657680583864,
      "loss": 0.1279,
      "step": 2990
    },
    {
      "epoch": 1.407529411764706,
      "grad_norm": 1.1073043794078274,
      "learning_rate": 0.0002937532554858647,
      "loss": 0.7383,
      "step": 2991
    },
    {
      "epoch": 1.408,
      "grad_norm": 0.6144811690038026,
      "learning_rate": 0.0002937407306361099,
      "loss": 0.2461,
      "step": 2992
    },
    {
      "epoch": 1.4084705882352941,
      "grad_norm": 0.9696284083830523,
      "learning_rate": 0.0002937281935101916,
      "loss": 0.7852,
      "step": 2993
    },
    {
      "epoch": 1.4089411764705884,
      "grad_norm": 0.9697694517659677,
      "learning_rate": 0.0002937156441091806,
      "loss": 0.6992,
      "step": 2994
    },
    {
      "epoch": 1.4094117647058824,
      "grad_norm": 0.5964907767711409,
      "learning_rate": 0.0002937030824341487,
      "loss": 0.3301,
      "step": 2995
    },
    {
      "epoch": 1.4098823529411764,
      "grad_norm": 0.907487688419984,
      "learning_rate": 0.00029369050848616863,
      "loss": 0.8164,
      "step": 2996
    },
    {
      "epoch": 1.4103529411764706,
      "grad_norm": 0.516260171725575,
      "learning_rate": 0.0002936779222663144,
      "loss": 0.2578,
      "step": 2997
    },
    {
      "epoch": 1.4108235294117648,
      "grad_norm": 0.7052750811009979,
      "learning_rate": 0.0002936653237756609,
      "loss": 0.3965,
      "step": 2998
    },
    {
      "epoch": 1.4112941176470588,
      "grad_norm": 0.8617895345222177,
      "learning_rate": 0.00029365271301528406,
      "loss": 0.8242,
      "step": 2999
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 1.0549455499458984,
      "learning_rate": 0.00029364008998626086,
      "loss": 0.7148,
      "step": 3000
    },
    {
      "epoch": 1.412235294117647,
      "grad_norm": 0.6400486069786544,
      "learning_rate": 0.00029362745468966946,
      "loss": 0.2871,
      "step": 3001
    },
    {
      "epoch": 1.4127058823529413,
      "grad_norm": 1.1477972578873998,
      "learning_rate": 0.0002936148071265889,
      "loss": 0.8945,
      "step": 3002
    },
    {
      "epoch": 1.4131764705882353,
      "grad_norm": 0.9997536175503997,
      "learning_rate": 0.00029360214729809946,
      "loss": 0.7148,
      "step": 3003
    },
    {
      "epoch": 1.4136470588235295,
      "grad_norm": 0.8956057073584853,
      "learning_rate": 0.00029358947520528227,
      "loss": 0.6641,
      "step": 3004
    },
    {
      "epoch": 1.4141176470588235,
      "grad_norm": 1.5000579158876977,
      "learning_rate": 0.0002935767908492196,
      "loss": 1.3359,
      "step": 3005
    },
    {
      "epoch": 1.4145882352941177,
      "grad_norm": 1.1441413461357754,
      "learning_rate": 0.0002935640942309947,
      "loss": 0.9219,
      "step": 3006
    },
    {
      "epoch": 1.4150588235294117,
      "grad_norm": 0.587119991424542,
      "learning_rate": 0.000293551385351692,
      "loss": 0.3887,
      "step": 3007
    },
    {
      "epoch": 1.415529411764706,
      "grad_norm": 0.8137468379650477,
      "learning_rate": 0.0002935386642123968,
      "loss": 0.5625,
      "step": 3008
    },
    {
      "epoch": 1.416,
      "grad_norm": 0.45682141438789864,
      "learning_rate": 0.00029352593081419573,
      "loss": 0.3359,
      "step": 3009
    },
    {
      "epoch": 1.4164705882352941,
      "grad_norm": 0.80543646524573,
      "learning_rate": 0.0002935131851581762,
      "loss": 0.416,
      "step": 3010
    },
    {
      "epoch": 1.4169411764705884,
      "grad_norm": 0.6246766511977281,
      "learning_rate": 0.00029350042724542657,
      "loss": 0.377,
      "step": 3011
    },
    {
      "epoch": 1.4174117647058824,
      "grad_norm": 0.49695583259913534,
      "learning_rate": 0.0002934876570770367,
      "loss": 0.293,
      "step": 3012
    },
    {
      "epoch": 1.4178823529411764,
      "grad_norm": 0.9516563121893202,
      "learning_rate": 0.00029347487465409714,
      "loss": 0.4805,
      "step": 3013
    },
    {
      "epoch": 1.4183529411764706,
      "grad_norm": 0.7978288919498021,
      "learning_rate": 0.0002934620799776995,
      "loss": 0.4082,
      "step": 3014
    },
    {
      "epoch": 1.4188235294117648,
      "grad_norm": 1.4083730916692487,
      "learning_rate": 0.00029344927304893653,
      "loss": 1.0703,
      "step": 3015
    },
    {
      "epoch": 1.4192941176470588,
      "grad_norm": 0.32316492058418017,
      "learning_rate": 0.00029343645386890204,
      "loss": 0.0718,
      "step": 3016
    },
    {
      "epoch": 1.4197647058823528,
      "grad_norm": 0.8967815859333583,
      "learning_rate": 0.00029342362243869093,
      "loss": 0.6523,
      "step": 3017
    },
    {
      "epoch": 1.420235294117647,
      "grad_norm": 0.4625623392142967,
      "learning_rate": 0.00029341077875939884,
      "loss": 0.2363,
      "step": 3018
    },
    {
      "epoch": 1.4207058823529413,
      "grad_norm": 1.1028778115137974,
      "learning_rate": 0.0002933979228321229,
      "loss": 0.6797,
      "step": 3019
    },
    {
      "epoch": 1.4211764705882353,
      "grad_norm": 0.7384481430744735,
      "learning_rate": 0.00029338505465796095,
      "loss": 0.4316,
      "step": 3020
    },
    {
      "epoch": 1.4216470588235295,
      "grad_norm": 0.7944242303193781,
      "learning_rate": 0.00029337217423801206,
      "loss": 0.4434,
      "step": 3021
    },
    {
      "epoch": 1.4221176470588235,
      "grad_norm": 0.43452106048924694,
      "learning_rate": 0.00029335928157337625,
      "loss": 0.4004,
      "step": 3022
    },
    {
      "epoch": 1.4225882352941177,
      "grad_norm": 0.4674304184451012,
      "learning_rate": 0.00029334637666515457,
      "loss": 0.1553,
      "step": 3023
    },
    {
      "epoch": 1.4230588235294117,
      "grad_norm": 1.154518297909369,
      "learning_rate": 0.00029333345951444923,
      "loss": 1.1094,
      "step": 3024
    },
    {
      "epoch": 1.423529411764706,
      "grad_norm": 0.849238810403484,
      "learning_rate": 0.00029332053012236345,
      "loss": 0.3613,
      "step": 3025
    },
    {
      "epoch": 1.424,
      "grad_norm": 0.7279267117945414,
      "learning_rate": 0.0002933075884900014,
      "loss": 0.5078,
      "step": 3026
    },
    {
      "epoch": 1.4244705882352942,
      "grad_norm": 0.7039298120341109,
      "learning_rate": 0.00029329463461846833,
      "loss": 0.6562,
      "step": 3027
    },
    {
      "epoch": 1.4249411764705882,
      "grad_norm": 1.3578699711987707,
      "learning_rate": 0.00029328166850887066,
      "loss": 1.6875,
      "step": 3028
    },
    {
      "epoch": 1.4254117647058824,
      "grad_norm": 0.7997955588278847,
      "learning_rate": 0.00029326869016231566,
      "loss": 0.7617,
      "step": 3029
    },
    {
      "epoch": 1.4258823529411764,
      "grad_norm": 0.4447373670981682,
      "learning_rate": 0.0002932556995799118,
      "loss": 0.2197,
      "step": 3030
    },
    {
      "epoch": 1.4263529411764706,
      "grad_norm": 0.8167133158703116,
      "learning_rate": 0.00029324269676276856,
      "loss": 0.6133,
      "step": 3031
    },
    {
      "epoch": 1.4268235294117648,
      "grad_norm": 1.077134101887308,
      "learning_rate": 0.0002932296817119964,
      "loss": 0.9492,
      "step": 3032
    },
    {
      "epoch": 1.4272941176470588,
      "grad_norm": 0.6654430991726861,
      "learning_rate": 0.0002932166544287069,
      "loss": 0.3203,
      "step": 3033
    },
    {
      "epoch": 1.4277647058823528,
      "grad_norm": 0.5765973008901019,
      "learning_rate": 0.00029320361491401263,
      "loss": 0.1963,
      "step": 3034
    },
    {
      "epoch": 1.428235294117647,
      "grad_norm": 1.007917654820607,
      "learning_rate": 0.00029319056316902723,
      "loss": 1.1172,
      "step": 3035
    },
    {
      "epoch": 1.4287058823529413,
      "grad_norm": 0.7962376926495064,
      "learning_rate": 0.0002931774991948654,
      "loss": 0.6406,
      "step": 3036
    },
    {
      "epoch": 1.4291764705882353,
      "grad_norm": 0.6691541925689288,
      "learning_rate": 0.0002931644229926428,
      "loss": 0.4258,
      "step": 3037
    },
    {
      "epoch": 1.4296470588235295,
      "grad_norm": 0.5191976962895297,
      "learning_rate": 0.00029315133456347633,
      "loss": 0.252,
      "step": 3038
    },
    {
      "epoch": 1.4301176470588235,
      "grad_norm": 0.7559422159989462,
      "learning_rate": 0.0002931382339084837,
      "loss": 0.5898,
      "step": 3039
    },
    {
      "epoch": 1.4305882352941177,
      "grad_norm": 0.3808194199857243,
      "learning_rate": 0.0002931251210287838,
      "loss": 0.1465,
      "step": 3040
    },
    {
      "epoch": 1.4310588235294117,
      "grad_norm": 1.2114283011941465,
      "learning_rate": 0.00029311199592549656,
      "loss": 0.9844,
      "step": 3041
    },
    {
      "epoch": 1.431529411764706,
      "grad_norm": 0.4329454763582818,
      "learning_rate": 0.00029309885859974294,
      "loss": 0.2012,
      "step": 3042
    },
    {
      "epoch": 1.432,
      "grad_norm": 0.8562653787438425,
      "learning_rate": 0.00029308570905264484,
      "loss": 0.3652,
      "step": 3043
    },
    {
      "epoch": 1.4324705882352942,
      "grad_norm": 1.2138807118507788,
      "learning_rate": 0.00029307254728532536,
      "loss": 0.8711,
      "step": 3044
    },
    {
      "epoch": 1.4329411764705882,
      "grad_norm": 0.9249088331150538,
      "learning_rate": 0.00029305937329890863,
      "loss": 0.6914,
      "step": 3045
    },
    {
      "epoch": 1.4334117647058824,
      "grad_norm": 1.3636399528556813,
      "learning_rate": 0.00029304618709451966,
      "loss": 1.125,
      "step": 3046
    },
    {
      "epoch": 1.4338823529411764,
      "grad_norm": 0.5752220650383408,
      "learning_rate": 0.0002930329886732847,
      "loss": 0.2988,
      "step": 3047
    },
    {
      "epoch": 1.4343529411764706,
      "grad_norm": 0.8101611101142487,
      "learning_rate": 0.00029301977803633094,
      "loss": 0.3535,
      "step": 3048
    },
    {
      "epoch": 1.4348235294117648,
      "grad_norm": 0.8899691675185828,
      "learning_rate": 0.0002930065551847866,
      "loss": 0.5039,
      "step": 3049
    },
    {
      "epoch": 1.4352941176470588,
      "grad_norm": 0.7179124844430599,
      "learning_rate": 0.00029299332011978107,
      "loss": 0.4688,
      "step": 3050
    },
    {
      "epoch": 1.4357647058823528,
      "grad_norm": 0.7364124062042244,
      "learning_rate": 0.00029298007284244456,
      "loss": 0.4785,
      "step": 3051
    },
    {
      "epoch": 1.436235294117647,
      "grad_norm": 0.5445067772297878,
      "learning_rate": 0.0002929668133539085,
      "loss": 0.2461,
      "step": 3052
    },
    {
      "epoch": 1.4367058823529413,
      "grad_norm": 1.238538811825628,
      "learning_rate": 0.0002929535416553054,
      "loss": 0.8711,
      "step": 3053
    },
    {
      "epoch": 1.4371764705882353,
      "grad_norm": 0.9607988483247937,
      "learning_rate": 0.0002929402577477686,
      "loss": 0.8672,
      "step": 3054
    },
    {
      "epoch": 1.4376470588235293,
      "grad_norm": 0.9032412498176795,
      "learning_rate": 0.0002929269616324327,
      "loss": 0.7109,
      "step": 3055
    },
    {
      "epoch": 1.4381176470588235,
      "grad_norm": 0.4590582387071062,
      "learning_rate": 0.0002929136533104332,
      "loss": 0.2432,
      "step": 3056
    },
    {
      "epoch": 1.4385882352941177,
      "grad_norm": 0.9550792198274408,
      "learning_rate": 0.0002929003327829067,
      "loss": 1.0547,
      "step": 3057
    },
    {
      "epoch": 1.4390588235294117,
      "grad_norm": 1.016938440593264,
      "learning_rate": 0.0002928870000509909,
      "loss": 0.9492,
      "step": 3058
    },
    {
      "epoch": 1.439529411764706,
      "grad_norm": 0.36334155212731656,
      "learning_rate": 0.0002928736551158244,
      "loss": 0.1943,
      "step": 3059
    },
    {
      "epoch": 1.44,
      "grad_norm": 0.799261173818633,
      "learning_rate": 0.00029286029797854704,
      "loss": 0.375,
      "step": 3060
    },
    {
      "epoch": 1.4404705882352942,
      "grad_norm": 0.9984714812138238,
      "learning_rate": 0.00029284692864029946,
      "loss": 0.6055,
      "step": 3061
    },
    {
      "epoch": 1.4409411764705882,
      "grad_norm": 0.753943861180155,
      "learning_rate": 0.0002928335471022235,
      "loss": 0.5586,
      "step": 3062
    },
    {
      "epoch": 1.4414117647058824,
      "grad_norm": 0.9160079805525155,
      "learning_rate": 0.00029282015336546205,
      "loss": 0.6289,
      "step": 3063
    },
    {
      "epoch": 1.4418823529411764,
      "grad_norm": 0.9541193984503541,
      "learning_rate": 0.00029280674743115896,
      "loss": 0.6172,
      "step": 3064
    },
    {
      "epoch": 1.4423529411764706,
      "grad_norm": 0.9399121188222109,
      "learning_rate": 0.0002927933293004592,
      "loss": 0.5273,
      "step": 3065
    },
    {
      "epoch": 1.4428235294117648,
      "grad_norm": 0.7626652707013352,
      "learning_rate": 0.00029277989897450867,
      "loss": 0.5156,
      "step": 3066
    },
    {
      "epoch": 1.4432941176470588,
      "grad_norm": 1.365924051186285,
      "learning_rate": 0.0002927664564544545,
      "loss": 0.8984,
      "step": 3067
    },
    {
      "epoch": 1.4437647058823528,
      "grad_norm": 1.2498202607740272,
      "learning_rate": 0.00029275300174144464,
      "loss": 1.3281,
      "step": 3068
    },
    {
      "epoch": 1.444235294117647,
      "grad_norm": 0.6912114038382895,
      "learning_rate": 0.0002927395348366282,
      "loss": 0.418,
      "step": 3069
    },
    {
      "epoch": 1.4447058823529413,
      "grad_norm": 0.5362657398476641,
      "learning_rate": 0.00029272605574115544,
      "loss": 0.2002,
      "step": 3070
    },
    {
      "epoch": 1.4451764705882353,
      "grad_norm": 0.959782602719673,
      "learning_rate": 0.00029271256445617737,
      "loss": 0.8125,
      "step": 3071
    },
    {
      "epoch": 1.4456470588235293,
      "grad_norm": 0.9859573180771553,
      "learning_rate": 0.0002926990609828463,
      "loss": 0.6602,
      "step": 3072
    },
    {
      "epoch": 1.4461176470588235,
      "grad_norm": 1.1892769803104748,
      "learning_rate": 0.00029268554532231556,
      "loss": 0.8945,
      "step": 3073
    },
    {
      "epoch": 1.4465882352941177,
      "grad_norm": 0.4712765952822137,
      "learning_rate": 0.00029267201747573934,
      "loss": 0.291,
      "step": 3074
    },
    {
      "epoch": 1.4470588235294117,
      "grad_norm": 1.090095886271055,
      "learning_rate": 0.00029265847744427303,
      "loss": 0.832,
      "step": 3075
    },
    {
      "epoch": 1.447529411764706,
      "grad_norm": 0.4450859977092717,
      "learning_rate": 0.000292644925229073,
      "loss": 0.2041,
      "step": 3076
    },
    {
      "epoch": 1.448,
      "grad_norm": 0.9124785362250023,
      "learning_rate": 0.00029263136083129666,
      "loss": 0.4258,
      "step": 3077
    },
    {
      "epoch": 1.4484705882352942,
      "grad_norm": 1.161180253144298,
      "learning_rate": 0.00029261778425210256,
      "loss": 0.8555,
      "step": 3078
    },
    {
      "epoch": 1.4489411764705882,
      "grad_norm": 0.6385080032749998,
      "learning_rate": 0.0002926041954926501,
      "loss": 0.2676,
      "step": 3079
    },
    {
      "epoch": 1.4494117647058824,
      "grad_norm": 0.679247244813809,
      "learning_rate": 0.00029259059455409987,
      "loss": 0.4512,
      "step": 3080
    },
    {
      "epoch": 1.4498823529411764,
      "grad_norm": 0.8752921505074228,
      "learning_rate": 0.00029257698143761354,
      "loss": 0.7891,
      "step": 3081
    },
    {
      "epoch": 1.4503529411764706,
      "grad_norm": 0.4706948098078892,
      "learning_rate": 0.00029256335614435356,
      "loss": 0.2559,
      "step": 3082
    },
    {
      "epoch": 1.4508235294117646,
      "grad_norm": 0.7352367620549192,
      "learning_rate": 0.00029254971867548373,
      "loss": 0.707,
      "step": 3083
    },
    {
      "epoch": 1.4512941176470588,
      "grad_norm": 1.0114241466179137,
      "learning_rate": 0.00029253606903216874,
      "loss": 0.8359,
      "step": 3084
    },
    {
      "epoch": 1.4517647058823528,
      "grad_norm": 0.9418373682791273,
      "learning_rate": 0.00029252240721557433,
      "loss": 0.7344,
      "step": 3085
    },
    {
      "epoch": 1.452235294117647,
      "grad_norm": 0.5208676373853071,
      "learning_rate": 0.00029250873322686727,
      "loss": 0.2598,
      "step": 3086
    },
    {
      "epoch": 1.4527058823529413,
      "grad_norm": 0.930851380957868,
      "learning_rate": 0.00029249504706721534,
      "loss": 0.6836,
      "step": 3087
    },
    {
      "epoch": 1.4531764705882353,
      "grad_norm": 0.8446699552040094,
      "learning_rate": 0.00029248134873778754,
      "loss": 0.7695,
      "step": 3088
    },
    {
      "epoch": 1.4536470588235293,
      "grad_norm": 0.44177665979637515,
      "learning_rate": 0.0002924676382397536,
      "loss": 0.2656,
      "step": 3089
    },
    {
      "epoch": 1.4541176470588235,
      "grad_norm": 1.0237443271039373,
      "learning_rate": 0.0002924539155742846,
      "loss": 1.2266,
      "step": 3090
    },
    {
      "epoch": 1.4545882352941177,
      "grad_norm": 0.9425465337635728,
      "learning_rate": 0.0002924401807425525,
      "loss": 0.6602,
      "step": 3091
    },
    {
      "epoch": 1.4550588235294117,
      "grad_norm": 0.5350871482460449,
      "learning_rate": 0.0002924264337457303,
      "loss": 0.3457,
      "step": 3092
    },
    {
      "epoch": 1.455529411764706,
      "grad_norm": 1.0975346696764259,
      "learning_rate": 0.000292412674584992,
      "loss": 1.0938,
      "step": 3093
    },
    {
      "epoch": 1.456,
      "grad_norm": 0.9353882269771309,
      "learning_rate": 0.0002923989032615128,
      "loss": 0.8008,
      "step": 3094
    },
    {
      "epoch": 1.4564705882352942,
      "grad_norm": 0.4996340204586132,
      "learning_rate": 0.0002923851197764688,
      "loss": 0.252,
      "step": 3095
    },
    {
      "epoch": 1.4569411764705882,
      "grad_norm": 0.5204575140370068,
      "learning_rate": 0.00029237132413103715,
      "loss": 0.2363,
      "step": 3096
    },
    {
      "epoch": 1.4574117647058824,
      "grad_norm": 0.6905546618283894,
      "learning_rate": 0.00029235751632639617,
      "loss": 0.4043,
      "step": 3097
    },
    {
      "epoch": 1.4578823529411764,
      "grad_norm": 0.5282869025009299,
      "learning_rate": 0.00029234369636372496,
      "loss": 0.2656,
      "step": 3098
    },
    {
      "epoch": 1.4583529411764706,
      "grad_norm": 0.6726148684486466,
      "learning_rate": 0.0002923298642442039,
      "loss": 0.3711,
      "step": 3099
    },
    {
      "epoch": 1.4588235294117646,
      "grad_norm": 0.5454382599416454,
      "learning_rate": 0.00029231601996901433,
      "loss": 0.3438,
      "step": 3100
    },
    {
      "epoch": 1.4592941176470589,
      "grad_norm": 0.9815439053062047,
      "learning_rate": 0.0002923021635393386,
      "loss": 0.8398,
      "step": 3101
    },
    {
      "epoch": 1.4597647058823529,
      "grad_norm": 0.7987133423107908,
      "learning_rate": 0.0002922882949563601,
      "loss": 0.8281,
      "step": 3102
    },
    {
      "epoch": 1.460235294117647,
      "grad_norm": 0.356354381000986,
      "learning_rate": 0.0002922744142212633,
      "loss": 0.1504,
      "step": 3103
    },
    {
      "epoch": 1.4607058823529413,
      "grad_norm": 0.7982012516291883,
      "learning_rate": 0.00029226052133523363,
      "loss": 0.5234,
      "step": 3104
    },
    {
      "epoch": 1.4611764705882353,
      "grad_norm": 0.5481559045808155,
      "learning_rate": 0.0002922466162994577,
      "loss": 0.3223,
      "step": 3105
    },
    {
      "epoch": 1.4616470588235293,
      "grad_norm": 1.0121941894803763,
      "learning_rate": 0.000292232699115123,
      "loss": 0.7383,
      "step": 3106
    },
    {
      "epoch": 1.4621176470588235,
      "grad_norm": 1.0257463778188989,
      "learning_rate": 0.0002922187697834182,
      "loss": 0.6719,
      "step": 3107
    },
    {
      "epoch": 1.4625882352941177,
      "grad_norm": 0.7506359931630818,
      "learning_rate": 0.00029220482830553284,
      "loss": 0.332,
      "step": 3108
    },
    {
      "epoch": 1.4630588235294117,
      "grad_norm": 0.5955979155237389,
      "learning_rate": 0.0002921908746826576,
      "loss": 0.1865,
      "step": 3109
    },
    {
      "epoch": 1.4635294117647057,
      "grad_norm": 0.7447013212721558,
      "learning_rate": 0.0002921769089159843,
      "loss": 0.7266,
      "step": 3110
    },
    {
      "epoch": 1.464,
      "grad_norm": 0.4471906752255671,
      "learning_rate": 0.00029216293100670554,
      "loss": 0.2773,
      "step": 3111
    },
    {
      "epoch": 1.4644705882352942,
      "grad_norm": 0.6406585000876265,
      "learning_rate": 0.0002921489409560152,
      "loss": 0.4375,
      "step": 3112
    },
    {
      "epoch": 1.4649411764705882,
      "grad_norm": 0.9590155215700602,
      "learning_rate": 0.00029213493876510807,
      "loss": 0.9961,
      "step": 3113
    },
    {
      "epoch": 1.4654117647058824,
      "grad_norm": 0.842404120921619,
      "learning_rate": 0.0002921209244351799,
      "loss": 0.7773,
      "step": 3114
    },
    {
      "epoch": 1.4658823529411764,
      "grad_norm": 0.4737453428961733,
      "learning_rate": 0.00029210689796742783,
      "loss": 0.1494,
      "step": 3115
    },
    {
      "epoch": 1.4663529411764706,
      "grad_norm": 0.5350767225174186,
      "learning_rate": 0.00029209285936304956,
      "loss": 0.4512,
      "step": 3116
    },
    {
      "epoch": 1.4668235294117646,
      "grad_norm": 0.3787368921491348,
      "learning_rate": 0.0002920788086232442,
      "loss": 0.1357,
      "step": 3117
    },
    {
      "epoch": 1.4672941176470589,
      "grad_norm": 1.2120321037053532,
      "learning_rate": 0.00029206474574921165,
      "loss": 0.9961,
      "step": 3118
    },
    {
      "epoch": 1.4677647058823529,
      "grad_norm": 0.4477243984694744,
      "learning_rate": 0.0002920506707421529,
      "loss": 0.1426,
      "step": 3119
    },
    {
      "epoch": 1.468235294117647,
      "grad_norm": 0.47660390093266913,
      "learning_rate": 0.0002920365836032703,
      "loss": 0.3789,
      "step": 3120
    },
    {
      "epoch": 1.4687058823529413,
      "grad_norm": 1.0390243528773182,
      "learning_rate": 0.00029202248433376663,
      "loss": 0.5039,
      "step": 3121
    },
    {
      "epoch": 1.4691764705882353,
      "grad_norm": 0.6253823761103464,
      "learning_rate": 0.00029200837293484627,
      "loss": 0.4141,
      "step": 3122
    },
    {
      "epoch": 1.4696470588235293,
      "grad_norm": 0.431297552590125,
      "learning_rate": 0.0002919942494077142,
      "loss": 0.207,
      "step": 3123
    },
    {
      "epoch": 1.4701176470588235,
      "grad_norm": 0.7845393201719867,
      "learning_rate": 0.0002919801137535769,
      "loss": 0.4121,
      "step": 3124
    },
    {
      "epoch": 1.4705882352941178,
      "grad_norm": 0.7972656950732148,
      "learning_rate": 0.0002919659659736414,
      "loss": 0.5977,
      "step": 3125
    },
    {
      "epoch": 1.4710588235294118,
      "grad_norm": 1.0915875565296136,
      "learning_rate": 0.00029195180606911606,
      "loss": 0.8125,
      "step": 3126
    },
    {
      "epoch": 1.4715294117647058,
      "grad_norm": 0.8978886626336576,
      "learning_rate": 0.0002919376340412102,
      "loss": 0.8555,
      "step": 3127
    },
    {
      "epoch": 1.472,
      "grad_norm": 0.3501971842408922,
      "learning_rate": 0.00029192344989113425,
      "loss": 0.1943,
      "step": 3128
    },
    {
      "epoch": 1.4724705882352942,
      "grad_norm": 0.16819911381219058,
      "learning_rate": 0.0002919092536200995,
      "loss": 0.0879,
      "step": 3129
    },
    {
      "epoch": 1.4729411764705882,
      "grad_norm": 1.146076483263405,
      "learning_rate": 0.00029189504522931847,
      "loss": 0.668,
      "step": 3130
    },
    {
      "epoch": 1.4734117647058824,
      "grad_norm": 0.6415815094556608,
      "learning_rate": 0.00029188082472000453,
      "loss": 0.418,
      "step": 3131
    },
    {
      "epoch": 1.4738823529411764,
      "grad_norm": 0.5156208738519833,
      "learning_rate": 0.0002918665920933722,
      "loss": 0.2676,
      "step": 3132
    },
    {
      "epoch": 1.4743529411764706,
      "grad_norm": 0.7204178191569977,
      "learning_rate": 0.0002918523473506372,
      "loss": 0.4434,
      "step": 3133
    },
    {
      "epoch": 1.4748235294117646,
      "grad_norm": 0.84311936063702,
      "learning_rate": 0.00029183809049301585,
      "loss": 0.6055,
      "step": 3134
    },
    {
      "epoch": 1.4752941176470589,
      "grad_norm": 0.6093846787365985,
      "learning_rate": 0.0002918238215217259,
      "loss": 0.543,
      "step": 3135
    },
    {
      "epoch": 1.4757647058823529,
      "grad_norm": 0.4684441695983158,
      "learning_rate": 0.00029180954043798595,
      "loss": 0.1196,
      "step": 3136
    },
    {
      "epoch": 1.476235294117647,
      "grad_norm": 0.9302891346088432,
      "learning_rate": 0.00029179524724301563,
      "loss": 0.8672,
      "step": 3137
    },
    {
      "epoch": 1.4767058823529413,
      "grad_norm": 0.4525657163648865,
      "learning_rate": 0.00029178094193803576,
      "loss": 0.416,
      "step": 3138
    },
    {
      "epoch": 1.4771764705882353,
      "grad_norm": 0.6379807191405619,
      "learning_rate": 0.00029176662452426796,
      "loss": 0.4316,
      "step": 3139
    },
    {
      "epoch": 1.4776470588235293,
      "grad_norm": 0.9182904314908454,
      "learning_rate": 0.0002917522950029351,
      "loss": 0.7969,
      "step": 3140
    },
    {
      "epoch": 1.4781176470588235,
      "grad_norm": 0.27434358377353196,
      "learning_rate": 0.00029173795337526093,
      "loss": 0.2598,
      "step": 3141
    },
    {
      "epoch": 1.4785882352941178,
      "grad_norm": 0.9784320508606443,
      "learning_rate": 0.0002917235996424703,
      "loss": 0.6719,
      "step": 3142
    },
    {
      "epoch": 1.4790588235294118,
      "grad_norm": 0.7758214137142613,
      "learning_rate": 0.0002917092338057892,
      "loss": 0.4785,
      "step": 3143
    },
    {
      "epoch": 1.4795294117647058,
      "grad_norm": 1.6116554466579307,
      "learning_rate": 0.00029169485586644434,
      "loss": 1.2656,
      "step": 3144
    },
    {
      "epoch": 1.48,
      "grad_norm": 0.4411117118575394,
      "learning_rate": 0.00029168046582566385,
      "loss": 0.168,
      "step": 3145
    },
    {
      "epoch": 1.4804705882352942,
      "grad_norm": 0.9460371460986967,
      "learning_rate": 0.00029166606368467667,
      "loss": 0.6719,
      "step": 3146
    },
    {
      "epoch": 1.4809411764705882,
      "grad_norm": 0.7987041977053106,
      "learning_rate": 0.0002916516494447127,
      "loss": 0.5781,
      "step": 3147
    },
    {
      "epoch": 1.4814117647058824,
      "grad_norm": 0.9710020874977177,
      "learning_rate": 0.0002916372231070031,
      "loss": 0.7383,
      "step": 3148
    },
    {
      "epoch": 1.4818823529411764,
      "grad_norm": 0.6479664592778328,
      "learning_rate": 0.00029162278467277994,
      "loss": 0.8164,
      "step": 3149
    },
    {
      "epoch": 1.4823529411764707,
      "grad_norm": 0.6479871895743305,
      "learning_rate": 0.0002916083341432763,
      "loss": 0.3574,
      "step": 3150
    },
    {
      "epoch": 1.4828235294117647,
      "grad_norm": 0.725267446836674,
      "learning_rate": 0.0002915938715197264,
      "loss": 0.6367,
      "step": 3151
    },
    {
      "epoch": 1.4832941176470589,
      "grad_norm": 0.5773179814969865,
      "learning_rate": 0.00029157939680336527,
      "loss": 0.2949,
      "step": 3152
    },
    {
      "epoch": 1.4837647058823529,
      "grad_norm": 1.1461794129614402,
      "learning_rate": 0.00029156490999542923,
      "loss": 0.7188,
      "step": 3153
    },
    {
      "epoch": 1.484235294117647,
      "grad_norm": 1.2874758736326881,
      "learning_rate": 0.0002915504110971555,
      "loss": 0.5977,
      "step": 3154
    },
    {
      "epoch": 1.484705882352941,
      "grad_norm": 0.756729303469366,
      "learning_rate": 0.0002915359001097824,
      "loss": 0.4941,
      "step": 3155
    },
    {
      "epoch": 1.4851764705882353,
      "grad_norm": 0.697172579933618,
      "learning_rate": 0.0002915213770345492,
      "loss": 0.3965,
      "step": 3156
    },
    {
      "epoch": 1.4856470588235293,
      "grad_norm": 0.8829128201054532,
      "learning_rate": 0.0002915068418726962,
      "loss": 0.6094,
      "step": 3157
    },
    {
      "epoch": 1.4861176470588235,
      "grad_norm": 0.2853417150974055,
      "learning_rate": 0.00029149229462546483,
      "loss": 0.0835,
      "step": 3158
    },
    {
      "epoch": 1.4865882352941178,
      "grad_norm": 0.2516597983391714,
      "learning_rate": 0.0002914777352940975,
      "loss": 0.0659,
      "step": 3159
    },
    {
      "epoch": 1.4870588235294118,
      "grad_norm": 0.9387166549252599,
      "learning_rate": 0.0002914631638798376,
      "loss": 0.9102,
      "step": 3160
    },
    {
      "epoch": 1.4875294117647058,
      "grad_norm": 0.7479364723716284,
      "learning_rate": 0.00029144858038392964,
      "loss": 0.4805,
      "step": 3161
    },
    {
      "epoch": 1.488,
      "grad_norm": 1.0319813440499817,
      "learning_rate": 0.00029143398480761915,
      "loss": 0.7734,
      "step": 3162
    },
    {
      "epoch": 1.4884705882352942,
      "grad_norm": 0.5369594106553616,
      "learning_rate": 0.00029141937715215256,
      "loss": 0.2793,
      "step": 3163
    },
    {
      "epoch": 1.4889411764705882,
      "grad_norm": 0.8053415066415631,
      "learning_rate": 0.0002914047574187776,
      "loss": 0.5625,
      "step": 3164
    },
    {
      "epoch": 1.4894117647058824,
      "grad_norm": 1.0041482290538226,
      "learning_rate": 0.0002913901256087427,
      "loss": 0.7305,
      "step": 3165
    },
    {
      "epoch": 1.4898823529411764,
      "grad_norm": 0.5246217078760552,
      "learning_rate": 0.00029137548172329756,
      "loss": 0.2598,
      "step": 3166
    },
    {
      "epoch": 1.4903529411764707,
      "grad_norm": 1.0212730212130583,
      "learning_rate": 0.0002913608257636929,
      "loss": 0.875,
      "step": 3167
    },
    {
      "epoch": 1.4908235294117647,
      "grad_norm": 0.7637871351538408,
      "learning_rate": 0.0002913461577311803,
      "loss": 0.7773,
      "step": 3168
    },
    {
      "epoch": 1.4912941176470589,
      "grad_norm": 0.45016860497846134,
      "learning_rate": 0.00029133147762701246,
      "loss": 0.2793,
      "step": 3169
    },
    {
      "epoch": 1.4917647058823529,
      "grad_norm": 0.8613353536006796,
      "learning_rate": 0.00029131678545244333,
      "loss": 0.6602,
      "step": 3170
    },
    {
      "epoch": 1.492235294117647,
      "grad_norm": 0.948672233660718,
      "learning_rate": 0.00029130208120872747,
      "loss": 0.8281,
      "step": 3171
    },
    {
      "epoch": 1.492705882352941,
      "grad_norm": 0.7981864786905704,
      "learning_rate": 0.00029128736489712087,
      "loss": 0.7461,
      "step": 3172
    },
    {
      "epoch": 1.4931764705882353,
      "grad_norm": 1.1371962892324778,
      "learning_rate": 0.0002912726365188803,
      "loss": 0.7031,
      "step": 3173
    },
    {
      "epoch": 1.4936470588235293,
      "grad_norm": 0.8716314236387359,
      "learning_rate": 0.0002912578960752636,
      "loss": 0.8516,
      "step": 3174
    },
    {
      "epoch": 1.4941176470588236,
      "grad_norm": 0.6696813680446202,
      "learning_rate": 0.00029124314356752967,
      "loss": 0.6602,
      "step": 3175
    },
    {
      "epoch": 1.4945882352941178,
      "grad_norm": 1.1441721727287846,
      "learning_rate": 0.00029122837899693847,
      "loss": 0.9023,
      "step": 3176
    },
    {
      "epoch": 1.4950588235294118,
      "grad_norm": 1.4623329645236682,
      "learning_rate": 0.0002912136023647511,
      "loss": 0.9609,
      "step": 3177
    },
    {
      "epoch": 1.4955294117647058,
      "grad_norm": 0.7193831845370292,
      "learning_rate": 0.0002911988136722293,
      "loss": 0.7148,
      "step": 3178
    },
    {
      "epoch": 1.496,
      "grad_norm": 0.5525644309306383,
      "learning_rate": 0.00029118401292063634,
      "loss": 0.3555,
      "step": 3179
    },
    {
      "epoch": 1.4964705882352942,
      "grad_norm": 0.7940750905131647,
      "learning_rate": 0.00029116920011123615,
      "loss": 0.7422,
      "step": 3180
    },
    {
      "epoch": 1.4969411764705882,
      "grad_norm": 0.3537394913295491,
      "learning_rate": 0.00029115437524529383,
      "loss": 0.1455,
      "step": 3181
    },
    {
      "epoch": 1.4974117647058822,
      "grad_norm": 0.7756768590905702,
      "learning_rate": 0.0002911395383240755,
      "loss": 0.7305,
      "step": 3182
    },
    {
      "epoch": 1.4978823529411764,
      "grad_norm": 0.8474529820044506,
      "learning_rate": 0.0002911246893488483,
      "loss": 0.5391,
      "step": 3183
    },
    {
      "epoch": 1.4983529411764707,
      "grad_norm": 0.937105300775336,
      "learning_rate": 0.0002911098283208805,
      "loss": 0.5859,
      "step": 3184
    },
    {
      "epoch": 1.4988235294117647,
      "grad_norm": 0.813402035643478,
      "learning_rate": 0.0002910949552414411,
      "loss": 0.8711,
      "step": 3185
    },
    {
      "epoch": 1.4992941176470589,
      "grad_norm": 1.0771062203914477,
      "learning_rate": 0.0002910800701118005,
      "loss": 0.6172,
      "step": 3186
    },
    {
      "epoch": 1.4997647058823529,
      "grad_norm": 0.8019121077023289,
      "learning_rate": 0.00029106517293323,
      "loss": 0.4453,
      "step": 3187
    },
    {
      "epoch": 1.5002352941176471,
      "grad_norm": 1.4807560274144047,
      "learning_rate": 0.0002910502637070017,
      "loss": 0.9102,
      "step": 3188
    },
    {
      "epoch": 1.5007058823529413,
      "grad_norm": 0.9510203390645469,
      "learning_rate": 0.0002910353424343891,
      "loss": 0.6992,
      "step": 3189
    },
    {
      "epoch": 1.5011764705882353,
      "grad_norm": 0.48059541667457534,
      "learning_rate": 0.00029102040911666645,
      "loss": 0.2559,
      "step": 3190
    },
    {
      "epoch": 1.5016470588235293,
      "grad_norm": 0.6340027012412923,
      "learning_rate": 0.0002910054637551092,
      "loss": 0.5234,
      "step": 3191
    },
    {
      "epoch": 1.5021176470588236,
      "grad_norm": 0.6471246639500149,
      "learning_rate": 0.0002909905063509936,
      "loss": 0.7031,
      "step": 3192
    },
    {
      "epoch": 1.5025882352941178,
      "grad_norm": 1.0571108323651346,
      "learning_rate": 0.0002909755369055973,
      "loss": 1.0234,
      "step": 3193
    },
    {
      "epoch": 1.5030588235294118,
      "grad_norm": 1.14049374480048,
      "learning_rate": 0.00029096055542019865,
      "loss": 0.7422,
      "step": 3194
    },
    {
      "epoch": 1.5035294117647058,
      "grad_norm": 0.39975401927529547,
      "learning_rate": 0.0002909455618960772,
      "loss": 0.209,
      "step": 3195
    },
    {
      "epoch": 1.504,
      "grad_norm": 1.2880167833924083,
      "learning_rate": 0.0002909305563345134,
      "loss": 0.9414,
      "step": 3196
    },
    {
      "epoch": 1.5044705882352942,
      "grad_norm": 0.7221166985624878,
      "learning_rate": 0.0002909155387367888,
      "loss": 0.3691,
      "step": 3197
    },
    {
      "epoch": 1.5049411764705882,
      "grad_norm": 0.7758712428816001,
      "learning_rate": 0.0002909005091041861,
      "loss": 0.6094,
      "step": 3198
    },
    {
      "epoch": 1.5054117647058822,
      "grad_norm": 0.6308649346459554,
      "learning_rate": 0.00029088546743798866,
      "loss": 0.5273,
      "step": 3199
    },
    {
      "epoch": 1.5058823529411764,
      "grad_norm": 0.5937911357085252,
      "learning_rate": 0.00029087041373948135,
      "loss": 0.2676,
      "step": 3200
    },
    {
      "epoch": 1.5063529411764707,
      "grad_norm": 0.6049716234758743,
      "learning_rate": 0.00029085534800994976,
      "loss": 0.2109,
      "step": 3201
    },
    {
      "epoch": 1.5068235294117647,
      "grad_norm": 0.8345993762721299,
      "learning_rate": 0.0002908402702506805,
      "loss": 0.5586,
      "step": 3202
    },
    {
      "epoch": 1.5072941176470587,
      "grad_norm": 0.8135872318986758,
      "learning_rate": 0.0002908251804629614,
      "loss": 0.7422,
      "step": 3203
    },
    {
      "epoch": 1.507764705882353,
      "grad_norm": 1.1237853188021663,
      "learning_rate": 0.0002908100786480811,
      "loss": 1.1406,
      "step": 3204
    },
    {
      "epoch": 1.5082352941176471,
      "grad_norm": 0.4732912685167568,
      "learning_rate": 0.0002907949648073294,
      "loss": 0.3555,
      "step": 3205
    },
    {
      "epoch": 1.5087058823529413,
      "grad_norm": 1.1778543831076873,
      "learning_rate": 0.00029077983894199716,
      "loss": 1.0469,
      "step": 3206
    },
    {
      "epoch": 1.5091764705882353,
      "grad_norm": 0.4844720366997301,
      "learning_rate": 0.00029076470105337615,
      "loss": 0.375,
      "step": 3207
    },
    {
      "epoch": 1.5096470588235293,
      "grad_norm": 0.8005567131070354,
      "learning_rate": 0.0002907495511427592,
      "loss": 0.6367,
      "step": 3208
    },
    {
      "epoch": 1.5101176470588236,
      "grad_norm": 0.8450484263867714,
      "learning_rate": 0.0002907343892114402,
      "loss": 0.5742,
      "step": 3209
    },
    {
      "epoch": 1.5105882352941178,
      "grad_norm": 0.8621598401038704,
      "learning_rate": 0.0002907192152607141,
      "loss": 0.5508,
      "step": 3210
    },
    {
      "epoch": 1.5110588235294118,
      "grad_norm": 0.980971360116892,
      "learning_rate": 0.00029070402929187673,
      "loss": 0.9883,
      "step": 3211
    },
    {
      "epoch": 1.5115294117647058,
      "grad_norm": 0.5456463180854729,
      "learning_rate": 0.0002906888313062252,
      "loss": 0.3652,
      "step": 3212
    },
    {
      "epoch": 1.512,
      "grad_norm": 1.3077076577390099,
      "learning_rate": 0.00029067362130505734,
      "loss": 0.957,
      "step": 3213
    },
    {
      "epoch": 1.5124705882352942,
      "grad_norm": 0.3158456679814356,
      "learning_rate": 0.00029065839928967224,
      "loss": 0.0991,
      "step": 3214
    },
    {
      "epoch": 1.5129411764705882,
      "grad_norm": 0.45877035486715834,
      "learning_rate": 0.0002906431652613699,
      "loss": 0.1602,
      "step": 3215
    },
    {
      "epoch": 1.5134117647058822,
      "grad_norm": 0.6884865985295004,
      "learning_rate": 0.0002906279192214514,
      "loss": 0.4141,
      "step": 3216
    },
    {
      "epoch": 1.5138823529411765,
      "grad_norm": 0.7957495798559275,
      "learning_rate": 0.0002906126611712188,
      "loss": 0.4395,
      "step": 3217
    },
    {
      "epoch": 1.5143529411764707,
      "grad_norm": 0.5761562931644942,
      "learning_rate": 0.00029059739111197524,
      "loss": 0.3086,
      "step": 3218
    },
    {
      "epoch": 1.5148235294117647,
      "grad_norm": 0.504461808596958,
      "learning_rate": 0.0002905821090450249,
      "loss": 0.2441,
      "step": 3219
    },
    {
      "epoch": 1.5152941176470587,
      "grad_norm": 1.0012099283848555,
      "learning_rate": 0.00029056681497167285,
      "loss": 0.6172,
      "step": 3220
    },
    {
      "epoch": 1.515764705882353,
      "grad_norm": 1.415671168340461,
      "learning_rate": 0.00029055150889322536,
      "loss": 1.3438,
      "step": 3221
    },
    {
      "epoch": 1.5162352941176471,
      "grad_norm": 0.718794687646925,
      "learning_rate": 0.0002905361908109896,
      "loss": 0.668,
      "step": 3222
    },
    {
      "epoch": 1.5167058823529411,
      "grad_norm": 0.32976215060596753,
      "learning_rate": 0.0002905208607262738,
      "loss": 0.1504,
      "step": 3223
    },
    {
      "epoch": 1.5171764705882353,
      "grad_norm": 0.775428039600159,
      "learning_rate": 0.0002905055186403873,
      "loss": 0.6836,
      "step": 3224
    },
    {
      "epoch": 1.5176470588235293,
      "grad_norm": 0.662487027406664,
      "learning_rate": 0.0002904901645546403,
      "loss": 0.3438,
      "step": 3225
    },
    {
      "epoch": 1.5181176470588236,
      "grad_norm": 0.3652417754193283,
      "learning_rate": 0.0002904747984703441,
      "loss": 0.1494,
      "step": 3226
    },
    {
      "epoch": 1.5185882352941178,
      "grad_norm": 1.2769918448093662,
      "learning_rate": 0.00029045942038881113,
      "loss": 0.8555,
      "step": 3227
    },
    {
      "epoch": 1.5190588235294118,
      "grad_norm": 0.42467068083791837,
      "learning_rate": 0.00029044403031135466,
      "loss": 0.332,
      "step": 3228
    },
    {
      "epoch": 1.5195294117647058,
      "grad_norm": 0.8710736445992879,
      "learning_rate": 0.0002904286282392892,
      "loss": 0.6562,
      "step": 3229
    },
    {
      "epoch": 1.52,
      "grad_norm": 0.7413648529539812,
      "learning_rate": 0.00029041321417393006,
      "loss": 0.6094,
      "step": 3230
    },
    {
      "epoch": 1.5204705882352942,
      "grad_norm": 1.1092345691550523,
      "learning_rate": 0.0002903977881165937,
      "loss": 0.707,
      "step": 3231
    },
    {
      "epoch": 1.5209411764705882,
      "grad_norm": 0.4582301897723165,
      "learning_rate": 0.0002903823500685976,
      "loss": 0.3809,
      "step": 3232
    },
    {
      "epoch": 1.5214117647058822,
      "grad_norm": 0.6918640589328529,
      "learning_rate": 0.0002903669000312602,
      "loss": 0.5625,
      "step": 3233
    },
    {
      "epoch": 1.5218823529411765,
      "grad_norm": 0.7025783933862936,
      "learning_rate": 0.0002903514380059011,
      "loss": 0.7852,
      "step": 3234
    },
    {
      "epoch": 1.5223529411764707,
      "grad_norm": 0.780577882814579,
      "learning_rate": 0.0002903359639938407,
      "loss": 0.4492,
      "step": 3235
    },
    {
      "epoch": 1.5228235294117647,
      "grad_norm": 0.6038651076934988,
      "learning_rate": 0.0002903204779964007,
      "loss": 0.2949,
      "step": 3236
    },
    {
      "epoch": 1.5232941176470587,
      "grad_norm": 0.6837900852144324,
      "learning_rate": 0.00029030498001490356,
      "loss": 0.5234,
      "step": 3237
    },
    {
      "epoch": 1.523764705882353,
      "grad_norm": 0.5616927913483393,
      "learning_rate": 0.00029028947005067297,
      "loss": 0.3438,
      "step": 3238
    },
    {
      "epoch": 1.5242352941176471,
      "grad_norm": 0.6093836478704988,
      "learning_rate": 0.00029027394810503346,
      "loss": 0.459,
      "step": 3239
    },
    {
      "epoch": 1.5247058823529411,
      "grad_norm": 0.5704753040502614,
      "learning_rate": 0.0002902584141793108,
      "loss": 0.3086,
      "step": 3240
    },
    {
      "epoch": 1.5251764705882351,
      "grad_norm": 0.4745585333293977,
      "learning_rate": 0.0002902428682748316,
      "loss": 0.2236,
      "step": 3241
    },
    {
      "epoch": 1.5256470588235294,
      "grad_norm": 0.9287079301734218,
      "learning_rate": 0.0002902273103929236,
      "loss": 0.6875,
      "step": 3242
    },
    {
      "epoch": 1.5261176470588236,
      "grad_norm": 0.42285542181889757,
      "learning_rate": 0.00029021174053491547,
      "loss": 0.2051,
      "step": 3243
    },
    {
      "epoch": 1.5265882352941178,
      "grad_norm": 0.6046263989815525,
      "learning_rate": 0.0002901961587021369,
      "loss": 0.4062,
      "step": 3244
    },
    {
      "epoch": 1.5270588235294118,
      "grad_norm": 0.5358019322299347,
      "learning_rate": 0.00029018056489591874,
      "loss": 0.3516,
      "step": 3245
    },
    {
      "epoch": 1.5275294117647058,
      "grad_norm": 0.5037045565254944,
      "learning_rate": 0.0002901649591175928,
      "loss": 0.2012,
      "step": 3246
    },
    {
      "epoch": 1.528,
      "grad_norm": 0.7449507781872409,
      "learning_rate": 0.0002901493413684918,
      "loss": 0.8281,
      "step": 3247
    },
    {
      "epoch": 1.5284705882352942,
      "grad_norm": 1.0228754865241085,
      "learning_rate": 0.0002901337116499496,
      "loss": 0.6914,
      "step": 3248
    },
    {
      "epoch": 1.5289411764705882,
      "grad_norm": 0.7400145206734291,
      "learning_rate": 0.0002901180699633011,
      "loss": 0.4883,
      "step": 3249
    },
    {
      "epoch": 1.5294117647058822,
      "grad_norm": 0.9743112229728506,
      "learning_rate": 0.00029010241630988217,
      "loss": 0.832,
      "step": 3250
    },
    {
      "epoch": 1.5298823529411765,
      "grad_norm": 0.8294413433993028,
      "learning_rate": 0.0002900867506910297,
      "loss": 0.6406,
      "step": 3251
    },
    {
      "epoch": 1.5303529411764707,
      "grad_norm": 0.5261010253987096,
      "learning_rate": 0.00029007107310808153,
      "loss": 0.2754,
      "step": 3252
    },
    {
      "epoch": 1.5308235294117647,
      "grad_norm": 1.305444409154303,
      "learning_rate": 0.00029005538356237674,
      "loss": 0.7461,
      "step": 3253
    },
    {
      "epoch": 1.5312941176470587,
      "grad_norm": 0.7606843973465984,
      "learning_rate": 0.00029003968205525523,
      "loss": 0.6953,
      "step": 3254
    },
    {
      "epoch": 1.531764705882353,
      "grad_norm": 0.8044527887531303,
      "learning_rate": 0.00029002396858805794,
      "loss": 0.5625,
      "step": 3255
    },
    {
      "epoch": 1.5322352941176471,
      "grad_norm": 0.8425518595981185,
      "learning_rate": 0.0002900082431621269,
      "loss": 0.5234,
      "step": 3256
    },
    {
      "epoch": 1.5327058823529411,
      "grad_norm": 1.1993817928111565,
      "learning_rate": 0.0002899925057788052,
      "loss": 0.9922,
      "step": 3257
    },
    {
      "epoch": 1.5331764705882351,
      "grad_norm": 0.9643167167923227,
      "learning_rate": 0.00028997675643943686,
      "loss": 0.8711,
      "step": 3258
    },
    {
      "epoch": 1.5336470588235294,
      "grad_norm": 0.9152185626115272,
      "learning_rate": 0.0002899609951453669,
      "loss": 0.8594,
      "step": 3259
    },
    {
      "epoch": 1.5341176470588236,
      "grad_norm": 0.815906134830871,
      "learning_rate": 0.0002899452218979415,
      "loss": 0.8047,
      "step": 3260
    },
    {
      "epoch": 1.5345882352941178,
      "grad_norm": 0.7702233603800038,
      "learning_rate": 0.0002899294366985076,
      "loss": 0.6211,
      "step": 3261
    },
    {
      "epoch": 1.5350588235294118,
      "grad_norm": 0.6738017100130809,
      "learning_rate": 0.0002899136395484136,
      "loss": 0.7344,
      "step": 3262
    },
    {
      "epoch": 1.5355294117647058,
      "grad_norm": 0.49549673774942277,
      "learning_rate": 0.00028989783044900845,
      "loss": 0.2812,
      "step": 3263
    },
    {
      "epoch": 1.536,
      "grad_norm": 0.5892968885915502,
      "learning_rate": 0.0002898820094016424,
      "loss": 0.4102,
      "step": 3264
    },
    {
      "epoch": 1.5364705882352943,
      "grad_norm": 0.3134293736258129,
      "learning_rate": 0.0002898661764076666,
      "loss": 0.0908,
      "step": 3265
    },
    {
      "epoch": 1.5369411764705883,
      "grad_norm": 1.1512472420901683,
      "learning_rate": 0.00028985033146843335,
      "loss": 0.8047,
      "step": 3266
    },
    {
      "epoch": 1.5374117647058823,
      "grad_norm": 0.8200153645368293,
      "learning_rate": 0.00028983447458529575,
      "loss": 0.6523,
      "step": 3267
    },
    {
      "epoch": 1.5378823529411765,
      "grad_norm": 0.8819367039704669,
      "learning_rate": 0.0002898186057596082,
      "loss": 0.7031,
      "step": 3268
    },
    {
      "epoch": 1.5383529411764707,
      "grad_norm": 0.9800621360276959,
      "learning_rate": 0.0002898027249927259,
      "loss": 0.7031,
      "step": 3269
    },
    {
      "epoch": 1.5388235294117647,
      "grad_norm": 0.8486455185928145,
      "learning_rate": 0.00028978683228600514,
      "loss": 0.6211,
      "step": 3270
    },
    {
      "epoch": 1.5392941176470587,
      "grad_norm": 1.1469071190495808,
      "learning_rate": 0.0002897709276408032,
      "loss": 0.8594,
      "step": 3271
    },
    {
      "epoch": 1.539764705882353,
      "grad_norm": 0.6830867059327113,
      "learning_rate": 0.00028975501105847856,
      "loss": 0.3223,
      "step": 3272
    },
    {
      "epoch": 1.5402352941176471,
      "grad_norm": 0.5721557790782602,
      "learning_rate": 0.00028973908254039044,
      "loss": 0.4004,
      "step": 3273
    },
    {
      "epoch": 1.5407058823529411,
      "grad_norm": 1.2202577840521482,
      "learning_rate": 0.00028972314208789926,
      "loss": 0.9531,
      "step": 3274
    },
    {
      "epoch": 1.5411764705882351,
      "grad_norm": 1.1914599960333454,
      "learning_rate": 0.00028970718970236644,
      "loss": 0.8555,
      "step": 3275
    },
    {
      "epoch": 1.5416470588235294,
      "grad_norm": 0.6618720258941797,
      "learning_rate": 0.00028969122538515435,
      "loss": 0.6211,
      "step": 3276
    },
    {
      "epoch": 1.5421176470588236,
      "grad_norm": 0.7333625435621157,
      "learning_rate": 0.00028967524913762636,
      "loss": 0.582,
      "step": 3277
    },
    {
      "epoch": 1.5425882352941176,
      "grad_norm": 0.48239907026753553,
      "learning_rate": 0.00028965926096114706,
      "loss": 0.2139,
      "step": 3278
    },
    {
      "epoch": 1.5430588235294118,
      "grad_norm": 0.6800667400817973,
      "learning_rate": 0.00028964326085708183,
      "loss": 0.2012,
      "step": 3279
    },
    {
      "epoch": 1.5435294117647058,
      "grad_norm": 0.9631967091222348,
      "learning_rate": 0.00028962724882679715,
      "loss": 0.7305,
      "step": 3280
    },
    {
      "epoch": 1.544,
      "grad_norm": 0.3848961889913865,
      "learning_rate": 0.00028961122487166063,
      "loss": 0.209,
      "step": 3281
    },
    {
      "epoch": 1.5444705882352943,
      "grad_norm": 0.6219008377268986,
      "learning_rate": 0.00028959518899304064,
      "loss": 0.3027,
      "step": 3282
    },
    {
      "epoch": 1.5449411764705883,
      "grad_norm": 0.5409792439045434,
      "learning_rate": 0.00028957914119230686,
      "loss": 0.3594,
      "step": 3283
    },
    {
      "epoch": 1.5454117647058823,
      "grad_norm": 0.9006864016103581,
      "learning_rate": 0.0002895630814708297,
      "loss": 0.7031,
      "step": 3284
    },
    {
      "epoch": 1.5458823529411765,
      "grad_norm": 0.40451262792408005,
      "learning_rate": 0.00028954700982998093,
      "loss": 0.1562,
      "step": 3285
    },
    {
      "epoch": 1.5463529411764707,
      "grad_norm": 0.7842489022139068,
      "learning_rate": 0.000289530926271133,
      "loss": 0.3477,
      "step": 3286
    },
    {
      "epoch": 1.5468235294117647,
      "grad_norm": 0.9375978491813001,
      "learning_rate": 0.0002895148307956596,
      "loss": 0.5625,
      "step": 3287
    },
    {
      "epoch": 1.5472941176470587,
      "grad_norm": 0.37171609075958467,
      "learning_rate": 0.00028949872340493525,
      "loss": 0.2451,
      "step": 3288
    },
    {
      "epoch": 1.547764705882353,
      "grad_norm": 1.03866774357236,
      "learning_rate": 0.00028948260410033576,
      "loss": 0.9102,
      "step": 3289
    },
    {
      "epoch": 1.5482352941176472,
      "grad_norm": 0.8880439526377862,
      "learning_rate": 0.00028946647288323766,
      "loss": 0.6211,
      "step": 3290
    },
    {
      "epoch": 1.5487058823529412,
      "grad_norm": 0.9996436394994087,
      "learning_rate": 0.00028945032975501877,
      "loss": 0.668,
      "step": 3291
    },
    {
      "epoch": 1.5491764705882352,
      "grad_norm": 0.7353351772234096,
      "learning_rate": 0.0002894341747170577,
      "loss": 0.6836,
      "step": 3292
    },
    {
      "epoch": 1.5496470588235294,
      "grad_norm": 0.664015460928567,
      "learning_rate": 0.00028941800777073414,
      "loss": 0.4258,
      "step": 3293
    },
    {
      "epoch": 1.5501176470588236,
      "grad_norm": 0.7048920066441566,
      "learning_rate": 0.0002894018289174289,
      "loss": 0.6484,
      "step": 3294
    },
    {
      "epoch": 1.5505882352941176,
      "grad_norm": 0.7258149036451597,
      "learning_rate": 0.0002893856381585237,
      "loss": 0.418,
      "step": 3295
    },
    {
      "epoch": 1.5510588235294116,
      "grad_norm": 0.7363897699721487,
      "learning_rate": 0.0002893694354954014,
      "loss": 0.5664,
      "step": 3296
    },
    {
      "epoch": 1.5515294117647058,
      "grad_norm": 0.7321070464314333,
      "learning_rate": 0.00028935322092944564,
      "loss": 0.5273,
      "step": 3297
    },
    {
      "epoch": 1.552,
      "grad_norm": 0.8685014596085707,
      "learning_rate": 0.00028933699446204126,
      "loss": 0.5781,
      "step": 3298
    },
    {
      "epoch": 1.5524705882352943,
      "grad_norm": 1.1922075100904244,
      "learning_rate": 0.0002893207560945742,
      "loss": 0.7695,
      "step": 3299
    },
    {
      "epoch": 1.5529411764705883,
      "grad_norm": 1.135623310384825,
      "learning_rate": 0.0002893045058284311,
      "loss": 0.8633,
      "step": 3300
    },
    {
      "epoch": 1.5534117647058823,
      "grad_norm": 0.5506507919552207,
      "learning_rate": 0.000289288243665,
      "loss": 0.2891,
      "step": 3301
    },
    {
      "epoch": 1.5538823529411765,
      "grad_norm": 0.43982749757886547,
      "learning_rate": 0.00028927196960566964,
      "loss": 0.3047,
      "step": 3302
    },
    {
      "epoch": 1.5543529411764707,
      "grad_norm": 1.0942804457589528,
      "learning_rate": 0.00028925568365182995,
      "loss": 0.875,
      "step": 3303
    },
    {
      "epoch": 1.5548235294117647,
      "grad_norm": 0.6511586498709077,
      "learning_rate": 0.00028923938580487183,
      "loss": 0.6133,
      "step": 3304
    },
    {
      "epoch": 1.5552941176470587,
      "grad_norm": 1.1234376057344804,
      "learning_rate": 0.0002892230760661872,
      "loss": 0.6445,
      "step": 3305
    },
    {
      "epoch": 1.555764705882353,
      "grad_norm": 0.400139850488488,
      "learning_rate": 0.000289206754437169,
      "loss": 0.1445,
      "step": 3306
    },
    {
      "epoch": 1.5562352941176472,
      "grad_norm": 0.7988878679139176,
      "learning_rate": 0.0002891904209192112,
      "loss": 0.5742,
      "step": 3307
    },
    {
      "epoch": 1.5567058823529412,
      "grad_norm": 0.6650111228685045,
      "learning_rate": 0.00028917407551370863,
      "loss": 0.5898,
      "step": 3308
    },
    {
      "epoch": 1.5571764705882352,
      "grad_norm": 0.9821826354328674,
      "learning_rate": 0.00028915771822205747,
      "loss": 0.5,
      "step": 3309
    },
    {
      "epoch": 1.5576470588235294,
      "grad_norm": 1.119875099311957,
      "learning_rate": 0.00028914134904565455,
      "loss": 0.9102,
      "step": 3310
    },
    {
      "epoch": 1.5581176470588236,
      "grad_norm": 0.7438575690816962,
      "learning_rate": 0.00028912496798589796,
      "loss": 0.6836,
      "step": 3311
    },
    {
      "epoch": 1.5585882352941176,
      "grad_norm": 0.35969934100697337,
      "learning_rate": 0.0002891085750441867,
      "loss": 0.2344,
      "step": 3312
    },
    {
      "epoch": 1.5590588235294116,
      "grad_norm": 0.8976381247955832,
      "learning_rate": 0.0002890921702219208,
      "loss": 0.7617,
      "step": 3313
    },
    {
      "epoch": 1.5595294117647058,
      "grad_norm": 0.7995820042924132,
      "learning_rate": 0.00028907575352050133,
      "loss": 0.7422,
      "step": 3314
    },
    {
      "epoch": 1.56,
      "grad_norm": 0.6987440510410342,
      "learning_rate": 0.00028905932494133034,
      "loss": 0.6016,
      "step": 3315
    },
    {
      "epoch": 1.5604705882352943,
      "grad_norm": 0.6997324910054653,
      "learning_rate": 0.0002890428844858109,
      "loss": 0.8008,
      "step": 3316
    },
    {
      "epoch": 1.5609411764705883,
      "grad_norm": 1.1721925082996036,
      "learning_rate": 0.0002890264321553471,
      "loss": 0.6992,
      "step": 3317
    },
    {
      "epoch": 1.5614117647058823,
      "grad_norm": 0.6521973725416851,
      "learning_rate": 0.00028900996795134414,
      "loss": 0.5195,
      "step": 3318
    },
    {
      "epoch": 1.5618823529411765,
      "grad_norm": 0.8218077563660742,
      "learning_rate": 0.000288993491875208,
      "loss": 0.5703,
      "step": 3319
    },
    {
      "epoch": 1.5623529411764707,
      "grad_norm": 0.7568443094162097,
      "learning_rate": 0.000288977003928346,
      "loss": 0.5,
      "step": 3320
    },
    {
      "epoch": 1.5628235294117647,
      "grad_norm": 0.9102825186803644,
      "learning_rate": 0.0002889605041121661,
      "loss": 0.6445,
      "step": 3321
    },
    {
      "epoch": 1.5632941176470587,
      "grad_norm": 0.6390914790148755,
      "learning_rate": 0.0002889439924280776,
      "loss": 0.4727,
      "step": 3322
    },
    {
      "epoch": 1.563764705882353,
      "grad_norm": 0.7989400098875774,
      "learning_rate": 0.0002889274688774906,
      "loss": 0.7773,
      "step": 3323
    },
    {
      "epoch": 1.5642352941176472,
      "grad_norm": 0.6397362876563306,
      "learning_rate": 0.00028891093346181633,
      "loss": 0.5352,
      "step": 3324
    },
    {
      "epoch": 1.5647058823529412,
      "grad_norm": 0.6675125941205322,
      "learning_rate": 0.000288894386182467,
      "loss": 0.543,
      "step": 3325
    },
    {
      "epoch": 1.5651764705882352,
      "grad_norm": 0.7652490265559518,
      "learning_rate": 0.0002888778270408558,
      "loss": 0.4375,
      "step": 3326
    },
    {
      "epoch": 1.5656470588235294,
      "grad_norm": 0.6863645574277041,
      "learning_rate": 0.000288861256038397,
      "loss": 0.6211,
      "step": 3327
    },
    {
      "epoch": 1.5661176470588236,
      "grad_norm": 0.4443755632557831,
      "learning_rate": 0.0002888446731765058,
      "loss": 0.2031,
      "step": 3328
    },
    {
      "epoch": 1.5665882352941176,
      "grad_norm": 0.6260800169005666,
      "learning_rate": 0.00028882807845659846,
      "loss": 0.5352,
      "step": 3329
    },
    {
      "epoch": 1.5670588235294116,
      "grad_norm": 0.6619922165869571,
      "learning_rate": 0.00028881147188009233,
      "loss": 0.4668,
      "step": 3330
    },
    {
      "epoch": 1.5675294117647058,
      "grad_norm": 0.49369273622890036,
      "learning_rate": 0.00028879485344840563,
      "loss": 0.375,
      "step": 3331
    },
    {
      "epoch": 1.568,
      "grad_norm": 0.7607805323218936,
      "learning_rate": 0.00028877822316295763,
      "loss": 0.7266,
      "step": 3332
    },
    {
      "epoch": 1.5684705882352943,
      "grad_norm": 0.44878508314481436,
      "learning_rate": 0.0002887615810251687,
      "loss": 0.2539,
      "step": 3333
    },
    {
      "epoch": 1.5689411764705883,
      "grad_norm": 0.7579802601703334,
      "learning_rate": 0.0002887449270364601,
      "loss": 0.7656,
      "step": 3334
    },
    {
      "epoch": 1.5694117647058823,
      "grad_norm": 0.5396785888445491,
      "learning_rate": 0.0002887282611982542,
      "loss": 0.2656,
      "step": 3335
    },
    {
      "epoch": 1.5698823529411765,
      "grad_norm": 0.6207090798941212,
      "learning_rate": 0.0002887115835119743,
      "loss": 0.4258,
      "step": 3336
    },
    {
      "epoch": 1.5703529411764707,
      "grad_norm": 0.5026119852779064,
      "learning_rate": 0.00028869489397904486,
      "loss": 0.3086,
      "step": 3337
    },
    {
      "epoch": 1.5708235294117647,
      "grad_norm": 0.7180450065246099,
      "learning_rate": 0.00028867819260089113,
      "loss": 0.6211,
      "step": 3338
    },
    {
      "epoch": 1.5712941176470587,
      "grad_norm": 0.5556209447743382,
      "learning_rate": 0.0002886614793789396,
      "loss": 0.2559,
      "step": 3339
    },
    {
      "epoch": 1.571764705882353,
      "grad_norm": 0.8038599059800572,
      "learning_rate": 0.0002886447543146175,
      "loss": 0.4941,
      "step": 3340
    },
    {
      "epoch": 1.5722352941176472,
      "grad_norm": 0.8865253125939264,
      "learning_rate": 0.0002886280174093534,
      "loss": 0.6016,
      "step": 3341
    },
    {
      "epoch": 1.5727058823529412,
      "grad_norm": 0.7513979390625958,
      "learning_rate": 0.0002886112686645766,
      "loss": 0.7695,
      "step": 3342
    },
    {
      "epoch": 1.5731764705882352,
      "grad_norm": 0.40098393252390013,
      "learning_rate": 0.00028859450808171763,
      "loss": 0.1035,
      "step": 3343
    },
    {
      "epoch": 1.5736470588235294,
      "grad_norm": 0.6486695580475887,
      "learning_rate": 0.00028857773566220784,
      "loss": 0.5664,
      "step": 3344
    },
    {
      "epoch": 1.5741176470588236,
      "grad_norm": 0.8570302543837988,
      "learning_rate": 0.0002885609514074797,
      "loss": 0.6055,
      "step": 3345
    },
    {
      "epoch": 1.5745882352941176,
      "grad_norm": 0.5788673955881779,
      "learning_rate": 0.0002885441553189667,
      "loss": 0.5078,
      "step": 3346
    },
    {
      "epoch": 1.5750588235294116,
      "grad_norm": 0.521908158815139,
      "learning_rate": 0.00028852734739810324,
      "loss": 0.375,
      "step": 3347
    },
    {
      "epoch": 1.5755294117647058,
      "grad_norm": 0.8868224183213661,
      "learning_rate": 0.00028851052764632487,
      "loss": 0.5859,
      "step": 3348
    },
    {
      "epoch": 1.576,
      "grad_norm": 0.6657723734028925,
      "learning_rate": 0.00028849369606506807,
      "loss": 0.459,
      "step": 3349
    },
    {
      "epoch": 1.576470588235294,
      "grad_norm": 0.7755290165306453,
      "learning_rate": 0.0002884768526557703,
      "loss": 0.7109,
      "step": 3350
    },
    {
      "epoch": 1.5769411764705883,
      "grad_norm": 1.413247899198657,
      "learning_rate": 0.00028845999741987003,
      "loss": 0.9492,
      "step": 3351
    },
    {
      "epoch": 1.5774117647058823,
      "grad_norm": 0.8101206428013311,
      "learning_rate": 0.0002884431303588069,
      "loss": 0.8477,
      "step": 3352
    },
    {
      "epoch": 1.5778823529411765,
      "grad_norm": 0.9016937783894072,
      "learning_rate": 0.00028842625147402134,
      "loss": 0.7148,
      "step": 3353
    },
    {
      "epoch": 1.5783529411764707,
      "grad_norm": 0.7582691471600709,
      "learning_rate": 0.00028840936076695493,
      "loss": 0.5703,
      "step": 3354
    },
    {
      "epoch": 1.5788235294117647,
      "grad_norm": 0.7629129920904122,
      "learning_rate": 0.0002883924582390503,
      "loss": 0.7109,
      "step": 3355
    },
    {
      "epoch": 1.5792941176470587,
      "grad_norm": 0.5699996561831852,
      "learning_rate": 0.0002883755438917508,
      "loss": 0.2559,
      "step": 3356
    },
    {
      "epoch": 1.579764705882353,
      "grad_norm": 0.8416076604494918,
      "learning_rate": 0.0002883586177265012,
      "loss": 0.75,
      "step": 3357
    },
    {
      "epoch": 1.5802352941176472,
      "grad_norm": 0.8552611138524068,
      "learning_rate": 0.000288341679744747,
      "loss": 0.5938,
      "step": 3358
    },
    {
      "epoch": 1.5807058823529412,
      "grad_norm": 0.5318127031355874,
      "learning_rate": 0.0002883247299479348,
      "loss": 0.3301,
      "step": 3359
    },
    {
      "epoch": 1.5811764705882352,
      "grad_norm": 0.7456753062144384,
      "learning_rate": 0.0002883077683375121,
      "loss": 0.3574,
      "step": 3360
    },
    {
      "epoch": 1.5816470588235294,
      "grad_norm": 0.686167316552025,
      "learning_rate": 0.0002882907949149277,
      "loss": 0.4629,
      "step": 3361
    },
    {
      "epoch": 1.5821176470588236,
      "grad_norm": 0.7919984162876091,
      "learning_rate": 0.000288273809681631,
      "loss": 0.6094,
      "step": 3362
    },
    {
      "epoch": 1.5825882352941176,
      "grad_norm": 0.664538427755424,
      "learning_rate": 0.00028825681263907276,
      "loss": 0.6992,
      "step": 3363
    },
    {
      "epoch": 1.5830588235294116,
      "grad_norm": 0.8864255663341604,
      "learning_rate": 0.00028823980378870457,
      "loss": 0.4258,
      "step": 3364
    },
    {
      "epoch": 1.5835294117647059,
      "grad_norm": 0.585804492923309,
      "learning_rate": 0.0002882227831319791,
      "loss": 0.4609,
      "step": 3365
    },
    {
      "epoch": 1.584,
      "grad_norm": 0.7201870888475127,
      "learning_rate": 0.00028820575067035,
      "loss": 0.2617,
      "step": 3366
    },
    {
      "epoch": 1.584470588235294,
      "grad_norm": 0.9132136452165728,
      "learning_rate": 0.00028818870640527183,
      "loss": 0.7734,
      "step": 3367
    },
    {
      "epoch": 1.584941176470588,
      "grad_norm": 0.7966096934481909,
      "learning_rate": 0.0002881716503382004,
      "loss": 0.4004,
      "step": 3368
    },
    {
      "epoch": 1.5854117647058823,
      "grad_norm": 0.6671213246661036,
      "learning_rate": 0.0002881545824705922,
      "loss": 0.3789,
      "step": 3369
    },
    {
      "epoch": 1.5858823529411765,
      "grad_norm": 0.8722393326828113,
      "learning_rate": 0.00028813750280390506,
      "loss": 0.8203,
      "step": 3370
    },
    {
      "epoch": 1.5863529411764707,
      "grad_norm": 0.7454570724419333,
      "learning_rate": 0.00028812041133959764,
      "loss": 0.5078,
      "step": 3371
    },
    {
      "epoch": 1.5868235294117647,
      "grad_norm": 0.6588486425025997,
      "learning_rate": 0.0002881033080791296,
      "loss": 0.5078,
      "step": 3372
    },
    {
      "epoch": 1.5872941176470587,
      "grad_norm": 0.4418319057391016,
      "learning_rate": 0.0002880861930239616,
      "loss": 0.1885,
      "step": 3373
    },
    {
      "epoch": 1.587764705882353,
      "grad_norm": 0.674225297164138,
      "learning_rate": 0.0002880690661755555,
      "loss": 0.4004,
      "step": 3374
    },
    {
      "epoch": 1.5882352941176472,
      "grad_norm": 0.8136711722329426,
      "learning_rate": 0.00028805192753537386,
      "loss": 0.6094,
      "step": 3375
    },
    {
      "epoch": 1.5887058823529412,
      "grad_norm": 0.8834631050654744,
      "learning_rate": 0.00028803477710488055,
      "loss": 0.7188,
      "step": 3376
    },
    {
      "epoch": 1.5891764705882352,
      "grad_norm": 0.9365323139500751,
      "learning_rate": 0.00028801761488554017,
      "loss": 0.5,
      "step": 3377
    },
    {
      "epoch": 1.5896470588235294,
      "grad_norm": 0.7506251718415827,
      "learning_rate": 0.0002880004408788185,
      "loss": 0.6562,
      "step": 3378
    },
    {
      "epoch": 1.5901176470588236,
      "grad_norm": 0.6876525138079673,
      "learning_rate": 0.0002879832550861823,
      "loss": 0.4844,
      "step": 3379
    },
    {
      "epoch": 1.5905882352941176,
      "grad_norm": 0.7361856198036372,
      "learning_rate": 0.0002879660575090993,
      "loss": 0.3887,
      "step": 3380
    },
    {
      "epoch": 1.5910588235294116,
      "grad_norm": 0.41174439831824,
      "learning_rate": 0.0002879488481490382,
      "loss": 0.3535,
      "step": 3381
    },
    {
      "epoch": 1.5915294117647059,
      "grad_norm": 1.0357965411767167,
      "learning_rate": 0.0002879316270074689,
      "loss": 0.9102,
      "step": 3382
    },
    {
      "epoch": 1.592,
      "grad_norm": 0.33114900441256556,
      "learning_rate": 0.00028791439408586215,
      "loss": 0.1104,
      "step": 3383
    },
    {
      "epoch": 1.592470588235294,
      "grad_norm": 0.8302128367699787,
      "learning_rate": 0.00028789714938568965,
      "loss": 0.6953,
      "step": 3384
    },
    {
      "epoch": 1.592941176470588,
      "grad_norm": 0.864456179681862,
      "learning_rate": 0.0002878798929084242,
      "loss": 0.5508,
      "step": 3385
    },
    {
      "epoch": 1.5934117647058823,
      "grad_norm": 1.0636010308051846,
      "learning_rate": 0.00028786262465553954,
      "loss": 1.0469,
      "step": 3386
    },
    {
      "epoch": 1.5938823529411765,
      "grad_norm": 0.6035860551774835,
      "learning_rate": 0.00028784534462851063,
      "loss": 0.4961,
      "step": 3387
    },
    {
      "epoch": 1.5943529411764708,
      "grad_norm": 0.8054893385712557,
      "learning_rate": 0.0002878280528288131,
      "loss": 0.543,
      "step": 3388
    },
    {
      "epoch": 1.5948235294117648,
      "grad_norm": 0.9094802417586425,
      "learning_rate": 0.00028781074925792377,
      "loss": 0.5234,
      "step": 3389
    },
    {
      "epoch": 1.5952941176470588,
      "grad_norm": 0.5491675961333706,
      "learning_rate": 0.00028779343391732053,
      "loss": 0.2773,
      "step": 3390
    },
    {
      "epoch": 1.595764705882353,
      "grad_norm": 1.3181653535427884,
      "learning_rate": 0.0002877761068084822,
      "loss": 1.0469,
      "step": 3391
    },
    {
      "epoch": 1.5962352941176472,
      "grad_norm": 0.46819016564255495,
      "learning_rate": 0.00028775876793288855,
      "loss": 0.2402,
      "step": 3392
    },
    {
      "epoch": 1.5967058823529412,
      "grad_norm": 0.7530868690239465,
      "learning_rate": 0.0002877414172920204,
      "loss": 0.4316,
      "step": 3393
    },
    {
      "epoch": 1.5971764705882352,
      "grad_norm": 0.765693478235934,
      "learning_rate": 0.00028772405488735963,
      "loss": 0.668,
      "step": 3394
    },
    {
      "epoch": 1.5976470588235294,
      "grad_norm": 0.54997260201142,
      "learning_rate": 0.000287706680720389,
      "loss": 0.2773,
      "step": 3395
    },
    {
      "epoch": 1.5981176470588236,
      "grad_norm": 0.7653365684430301,
      "learning_rate": 0.0002876892947925924,
      "loss": 0.7656,
      "step": 3396
    },
    {
      "epoch": 1.5985882352941176,
      "grad_norm": 1.1241486630767434,
      "learning_rate": 0.0002876718971054547,
      "loss": 0.8164,
      "step": 3397
    },
    {
      "epoch": 1.5990588235294116,
      "grad_norm": 0.4519206262729537,
      "learning_rate": 0.00028765448766046163,
      "loss": 0.249,
      "step": 3398
    },
    {
      "epoch": 1.5995294117647059,
      "grad_norm": 0.7166756415272412,
      "learning_rate": 0.00028763706645910014,
      "loss": 0.2598,
      "step": 3399
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.5147703593989259,
      "learning_rate": 0.0002876196335028581,
      "loss": 0.4258,
      "step": 3400
    },
    {
      "epoch": 1.600470588235294,
      "grad_norm": 0.05493698690263926,
      "learning_rate": 0.00028760218879322436,
      "loss": 0.011,
      "step": 3401
    },
    {
      "epoch": 1.600941176470588,
      "grad_norm": 0.7248648080469663,
      "learning_rate": 0.00028758473233168874,
      "loss": 0.543,
      "step": 3402
    },
    {
      "epoch": 1.6014117647058823,
      "grad_norm": 0.7704306566505371,
      "learning_rate": 0.0002875672641197421,
      "loss": 0.668,
      "step": 3403
    },
    {
      "epoch": 1.6018823529411765,
      "grad_norm": 0.867410842629948,
      "learning_rate": 0.00028754978415887635,
      "loss": 0.5547,
      "step": 3404
    },
    {
      "epoch": 1.6023529411764705,
      "grad_norm": 0.35215066120082883,
      "learning_rate": 0.0002875322924505844,
      "loss": 0.2949,
      "step": 3405
    },
    {
      "epoch": 1.6028235294117648,
      "grad_norm": 0.7162120026808144,
      "learning_rate": 0.00028751478899636,
      "loss": 0.5938,
      "step": 3406
    },
    {
      "epoch": 1.6032941176470588,
      "grad_norm": 1.1187042113274033,
      "learning_rate": 0.00028749727379769817,
      "loss": 0.7773,
      "step": 3407
    },
    {
      "epoch": 1.603764705882353,
      "grad_norm": 0.8164298682487021,
      "learning_rate": 0.0002874797468560948,
      "loss": 0.7227,
      "step": 3408
    },
    {
      "epoch": 1.6042352941176472,
      "grad_norm": 0.5907272361224564,
      "learning_rate": 0.0002874622081730466,
      "loss": 0.2197,
      "step": 3409
    },
    {
      "epoch": 1.6047058823529412,
      "grad_norm": 0.7992119557356046,
      "learning_rate": 0.0002874446577500516,
      "loss": 0.5938,
      "step": 3410
    },
    {
      "epoch": 1.6051764705882352,
      "grad_norm": 0.5731227151959111,
      "learning_rate": 0.0002874270955886087,
      "loss": 0.2891,
      "step": 3411
    },
    {
      "epoch": 1.6056470588235294,
      "grad_norm": 0.5772398324166471,
      "learning_rate": 0.0002874095216902178,
      "loss": 0.334,
      "step": 3412
    },
    {
      "epoch": 1.6061176470588236,
      "grad_norm": 1.2111601390138964,
      "learning_rate": 0.00028739193605637967,
      "loss": 0.9258,
      "step": 3413
    },
    {
      "epoch": 1.6065882352941177,
      "grad_norm": 1.0400185595621052,
      "learning_rate": 0.00028737433868859634,
      "loss": 1.1094,
      "step": 3414
    },
    {
      "epoch": 1.6070588235294117,
      "grad_norm": 0.43909238769586667,
      "learning_rate": 0.0002873567295883707,
      "loss": 0.167,
      "step": 3415
    },
    {
      "epoch": 1.6075294117647059,
      "grad_norm": 0.519381798500473,
      "learning_rate": 0.00028733910875720654,
      "loss": 0.3809,
      "step": 3416
    },
    {
      "epoch": 1.608,
      "grad_norm": 0.7583008502282581,
      "learning_rate": 0.00028732147619660883,
      "loss": 0.4473,
      "step": 3417
    },
    {
      "epoch": 1.608470588235294,
      "grad_norm": 0.5810906764531795,
      "learning_rate": 0.00028730383190808364,
      "loss": 0.4668,
      "step": 3418
    },
    {
      "epoch": 1.608941176470588,
      "grad_norm": 1.1095608653973124,
      "learning_rate": 0.0002872861758931376,
      "loss": 0.8477,
      "step": 3419
    },
    {
      "epoch": 1.6094117647058823,
      "grad_norm": 1.4449826213399701,
      "learning_rate": 0.0002872685081532788,
      "loss": 0.9141,
      "step": 3420
    },
    {
      "epoch": 1.6098823529411765,
      "grad_norm": 0.931488776870523,
      "learning_rate": 0.00028725082869001615,
      "loss": 0.9688,
      "step": 3421
    },
    {
      "epoch": 1.6103529411764705,
      "grad_norm": 0.8022791631957205,
      "learning_rate": 0.00028723313750485946,
      "loss": 0.6055,
      "step": 3422
    },
    {
      "epoch": 1.6108235294117645,
      "grad_norm": 0.46612975215487434,
      "learning_rate": 0.0002872154345993197,
      "loss": 0.4141,
      "step": 3423
    },
    {
      "epoch": 1.6112941176470588,
      "grad_norm": 0.704881079058541,
      "learning_rate": 0.0002871977199749088,
      "loss": 0.625,
      "step": 3424
    },
    {
      "epoch": 1.611764705882353,
      "grad_norm": 1.3192385056641904,
      "learning_rate": 0.00028717999363313967,
      "loss": 1.0234,
      "step": 3425
    },
    {
      "epoch": 1.6122352941176472,
      "grad_norm": 0.8880425154672361,
      "learning_rate": 0.0002871622555755262,
      "loss": 0.8359,
      "step": 3426
    },
    {
      "epoch": 1.6127058823529412,
      "grad_norm": 0.3630868289564518,
      "learning_rate": 0.00028714450580358334,
      "loss": 0.2178,
      "step": 3427
    },
    {
      "epoch": 1.6131764705882352,
      "grad_norm": 0.8417329497317854,
      "learning_rate": 0.000287126744318827,
      "loss": 0.6484,
      "step": 3428
    },
    {
      "epoch": 1.6136470588235294,
      "grad_norm": 0.7110124246871155,
      "learning_rate": 0.0002871089711227741,
      "loss": 0.5078,
      "step": 3429
    },
    {
      "epoch": 1.6141176470588237,
      "grad_norm": 0.7648526892841557,
      "learning_rate": 0.0002870911862169426,
      "loss": 0.6719,
      "step": 3430
    },
    {
      "epoch": 1.6145882352941177,
      "grad_norm": 0.5221198422754166,
      "learning_rate": 0.0002870733896028513,
      "loss": 0.1914,
      "step": 3431
    },
    {
      "epoch": 1.6150588235294117,
      "grad_norm": 0.8310040396623308,
      "learning_rate": 0.00028705558128202017,
      "loss": 0.498,
      "step": 3432
    },
    {
      "epoch": 1.6155294117647059,
      "grad_norm": 0.8445740102055443,
      "learning_rate": 0.0002870377612559702,
      "loss": 0.5898,
      "step": 3433
    },
    {
      "epoch": 1.616,
      "grad_norm": 0.5633798975930505,
      "learning_rate": 0.0002870199295262232,
      "loss": 0.5781,
      "step": 3434
    },
    {
      "epoch": 1.616470588235294,
      "grad_norm": 0.38760150998751025,
      "learning_rate": 0.00028700208609430217,
      "loss": 0.2539,
      "step": 3435
    },
    {
      "epoch": 1.616941176470588,
      "grad_norm": 0.7657381401386721,
      "learning_rate": 0.00028698423096173097,
      "loss": 0.7734,
      "step": 3436
    },
    {
      "epoch": 1.6174117647058823,
      "grad_norm": 0.6236916174994552,
      "learning_rate": 0.0002869663641300346,
      "loss": 0.6211,
      "step": 3437
    },
    {
      "epoch": 1.6178823529411765,
      "grad_norm": 0.8104891902993661,
      "learning_rate": 0.00028694848560073885,
      "loss": 0.6836,
      "step": 3438
    },
    {
      "epoch": 1.6183529411764705,
      "grad_norm": 0.8254942337098242,
      "learning_rate": 0.00028693059537537077,
      "loss": 0.2695,
      "step": 3439
    },
    {
      "epoch": 1.6188235294117646,
      "grad_norm": 1.0044313855666602,
      "learning_rate": 0.00028691269345545816,
      "loss": 0.5938,
      "step": 3440
    },
    {
      "epoch": 1.6192941176470588,
      "grad_norm": 0.639289846367356,
      "learning_rate": 0.00028689477984252996,
      "loss": 0.4922,
      "step": 3441
    },
    {
      "epoch": 1.619764705882353,
      "grad_norm": 0.5383478111679644,
      "learning_rate": 0.00028687685453811614,
      "loss": 0.5312,
      "step": 3442
    },
    {
      "epoch": 1.6202352941176472,
      "grad_norm": 0.8216051096174845,
      "learning_rate": 0.00028685891754374755,
      "loss": 0.7031,
      "step": 3443
    },
    {
      "epoch": 1.6207058823529412,
      "grad_norm": 0.5831894102555009,
      "learning_rate": 0.0002868409688609561,
      "loss": 0.4531,
      "step": 3444
    },
    {
      "epoch": 1.6211764705882352,
      "grad_norm": 0.5516106903672413,
      "learning_rate": 0.00028682300849127465,
      "loss": 0.207,
      "step": 3445
    },
    {
      "epoch": 1.6216470588235294,
      "grad_norm": 0.4527632360350347,
      "learning_rate": 0.0002868050364362373,
      "loss": 0.3301,
      "step": 3446
    },
    {
      "epoch": 1.6221176470588237,
      "grad_norm": 1.1360309856163573,
      "learning_rate": 0.00028678705269737873,
      "loss": 0.832,
      "step": 3447
    },
    {
      "epoch": 1.6225882352941177,
      "grad_norm": 0.7107755993947426,
      "learning_rate": 0.0002867690572762349,
      "loss": 0.293,
      "step": 3448
    },
    {
      "epoch": 1.6230588235294117,
      "grad_norm": 0.9954843999344225,
      "learning_rate": 0.0002867510501743428,
      "loss": 0.8359,
      "step": 3449
    },
    {
      "epoch": 1.6235294117647059,
      "grad_norm": 0.3208138014397753,
      "learning_rate": 0.0002867330313932402,
      "loss": 0.0767,
      "step": 3450
    },
    {
      "epoch": 1.624,
      "grad_norm": 0.9278866701654044,
      "learning_rate": 0.00028671500093446604,
      "loss": 0.4004,
      "step": 3451
    },
    {
      "epoch": 1.624470588235294,
      "grad_norm": 0.9143897821692593,
      "learning_rate": 0.0002866969587995602,
      "loss": 0.8555,
      "step": 3452
    },
    {
      "epoch": 1.6249411764705881,
      "grad_norm": 0.5642815749056929,
      "learning_rate": 0.00028667890499006364,
      "loss": 0.5312,
      "step": 3453
    },
    {
      "epoch": 1.6254117647058823,
      "grad_norm": 0.5230181817730231,
      "learning_rate": 0.00028666083950751824,
      "loss": 0.5,
      "step": 3454
    },
    {
      "epoch": 1.6258823529411766,
      "grad_norm": 0.5237990286782699,
      "learning_rate": 0.0002866427623534667,
      "loss": 0.2217,
      "step": 3455
    },
    {
      "epoch": 1.6263529411764706,
      "grad_norm": 0.9810019724687804,
      "learning_rate": 0.0002866246735294531,
      "loss": 0.9609,
      "step": 3456
    },
    {
      "epoch": 1.6268235294117646,
      "grad_norm": 0.6903077206197871,
      "learning_rate": 0.00028660657303702225,
      "loss": 0.4219,
      "step": 3457
    },
    {
      "epoch": 1.6272941176470588,
      "grad_norm": 0.5294184874172738,
      "learning_rate": 0.00028658846087771994,
      "loss": 0.4902,
      "step": 3458
    },
    {
      "epoch": 1.627764705882353,
      "grad_norm": 0.8610442852259976,
      "learning_rate": 0.0002865703370530932,
      "loss": 0.875,
      "step": 3459
    },
    {
      "epoch": 1.6282352941176472,
      "grad_norm": 0.5040313117090343,
      "learning_rate": 0.0002865522015646897,
      "loss": 0.4336,
      "step": 3460
    },
    {
      "epoch": 1.6287058823529412,
      "grad_norm": 0.7686671884716438,
      "learning_rate": 0.0002865340544140585,
      "loss": 0.4648,
      "step": 3461
    },
    {
      "epoch": 1.6291764705882352,
      "grad_norm": 0.6312334057099335,
      "learning_rate": 0.00028651589560274934,
      "loss": 0.4551,
      "step": 3462
    },
    {
      "epoch": 1.6296470588235294,
      "grad_norm": 0.8919431118897766,
      "learning_rate": 0.00028649772513231306,
      "loss": 0.8711,
      "step": 3463
    },
    {
      "epoch": 1.6301176470588237,
      "grad_norm": 0.4624161682199338,
      "learning_rate": 0.00028647954300430154,
      "loss": 0.375,
      "step": 3464
    },
    {
      "epoch": 1.6305882352941177,
      "grad_norm": 0.9745168570747599,
      "learning_rate": 0.0002864613492202676,
      "loss": 0.4883,
      "step": 3465
    },
    {
      "epoch": 1.6310588235294117,
      "grad_norm": 0.729580422805926,
      "learning_rate": 0.00028644314378176513,
      "loss": 0.4902,
      "step": 3466
    },
    {
      "epoch": 1.631529411764706,
      "grad_norm": 0.5753082263593453,
      "learning_rate": 0.00028642492669034894,
      "loss": 0.3301,
      "step": 3467
    },
    {
      "epoch": 1.6320000000000001,
      "grad_norm": 0.38369895462187875,
      "learning_rate": 0.0002864066979475748,
      "loss": 0.334,
      "step": 3468
    },
    {
      "epoch": 1.6324705882352941,
      "grad_norm": 0.8035925180365321,
      "learning_rate": 0.00028638845755499964,
      "loss": 0.4941,
      "step": 3469
    },
    {
      "epoch": 1.6329411764705881,
      "grad_norm": 0.8196492544507828,
      "learning_rate": 0.00028637020551418123,
      "loss": 0.4375,
      "step": 3470
    },
    {
      "epoch": 1.6334117647058823,
      "grad_norm": 0.44275445391018364,
      "learning_rate": 0.00028635194182667834,
      "loss": 0.2637,
      "step": 3471
    },
    {
      "epoch": 1.6338823529411766,
      "grad_norm": 0.7468354006182637,
      "learning_rate": 0.0002863336664940508,
      "loss": 0.5078,
      "step": 3472
    },
    {
      "epoch": 1.6343529411764706,
      "grad_norm": 0.681250085990525,
      "learning_rate": 0.00028631537951785945,
      "loss": 0.6719,
      "step": 3473
    },
    {
      "epoch": 1.6348235294117646,
      "grad_norm": 1.042948112623363,
      "learning_rate": 0.000286297080899666,
      "loss": 0.8711,
      "step": 3474
    },
    {
      "epoch": 1.6352941176470588,
      "grad_norm": 0.6261844615064271,
      "learning_rate": 0.0002862787706410334,
      "loss": 0.5,
      "step": 3475
    },
    {
      "epoch": 1.635764705882353,
      "grad_norm": 0.40062218018711,
      "learning_rate": 0.00028626044874352534,
      "loss": 0.1357,
      "step": 3476
    },
    {
      "epoch": 1.636235294117647,
      "grad_norm": 0.4338856129276994,
      "learning_rate": 0.0002862421152087066,
      "loss": 0.2207,
      "step": 3477
    },
    {
      "epoch": 1.636705882352941,
      "grad_norm": 0.7193783494152654,
      "learning_rate": 0.00028622377003814295,
      "loss": 0.416,
      "step": 3478
    },
    {
      "epoch": 1.6371764705882352,
      "grad_norm": 0.6151780563505225,
      "learning_rate": 0.00028620541323340115,
      "loss": 0.3184,
      "step": 3479
    },
    {
      "epoch": 1.6376470588235295,
      "grad_norm": 0.9131453800727708,
      "learning_rate": 0.00028618704479604895,
      "loss": 0.75,
      "step": 3480
    },
    {
      "epoch": 1.6381176470588237,
      "grad_norm": 0.43392510127199185,
      "learning_rate": 0.00028616866472765514,
      "loss": 0.2119,
      "step": 3481
    },
    {
      "epoch": 1.6385882352941177,
      "grad_norm": 0.46424110570801586,
      "learning_rate": 0.00028615027302978947,
      "loss": 0.1592,
      "step": 3482
    },
    {
      "epoch": 1.6390588235294117,
      "grad_norm": 1.3269570271815763,
      "learning_rate": 0.00028613186970402274,
      "loss": 1.3594,
      "step": 3483
    },
    {
      "epoch": 1.639529411764706,
      "grad_norm": 0.4833556538885346,
      "learning_rate": 0.0002861134547519265,
      "loss": 0.2812,
      "step": 3484
    },
    {
      "epoch": 1.6400000000000001,
      "grad_norm": 0.9539010895746259,
      "learning_rate": 0.0002860950281750737,
      "loss": 0.793,
      "step": 3485
    },
    {
      "epoch": 1.6404705882352941,
      "grad_norm": 1.0483857315568281,
      "learning_rate": 0.00028607658997503786,
      "loss": 1.0859,
      "step": 3486
    },
    {
      "epoch": 1.6409411764705881,
      "grad_norm": 0.6320854107752084,
      "learning_rate": 0.00028605814015339383,
      "loss": 0.582,
      "step": 3487
    },
    {
      "epoch": 1.6414117647058823,
      "grad_norm": 0.4361794795468205,
      "learning_rate": 0.0002860396787117173,
      "loss": 0.3184,
      "step": 3488
    },
    {
      "epoch": 1.6418823529411766,
      "grad_norm": 0.39207442115255536,
      "learning_rate": 0.00028602120565158486,
      "loss": 0.1504,
      "step": 3489
    },
    {
      "epoch": 1.6423529411764706,
      "grad_norm": 0.6066054035454492,
      "learning_rate": 0.0002860027209745743,
      "loss": 0.5586,
      "step": 3490
    },
    {
      "epoch": 1.6428235294117646,
      "grad_norm": 0.5252551703846989,
      "learning_rate": 0.00028598422468226425,
      "loss": 0.2559,
      "step": 3491
    },
    {
      "epoch": 1.6432941176470588,
      "grad_norm": 0.5355122108764591,
      "learning_rate": 0.0002859657167762345,
      "loss": 0.2695,
      "step": 3492
    },
    {
      "epoch": 1.643764705882353,
      "grad_norm": 0.37837646426431365,
      "learning_rate": 0.00028594719725806556,
      "loss": 0.1289,
      "step": 3493
    },
    {
      "epoch": 1.644235294117647,
      "grad_norm": 1.2053657471655812,
      "learning_rate": 0.0002859286661293392,
      "loss": 1.3125,
      "step": 3494
    },
    {
      "epoch": 1.644705882352941,
      "grad_norm": 0.8546080982777429,
      "learning_rate": 0.00028591012339163806,
      "loss": 0.7305,
      "step": 3495
    },
    {
      "epoch": 1.6451764705882352,
      "grad_norm": 0.6579753663956319,
      "learning_rate": 0.00028589156904654565,
      "loss": 0.4453,
      "step": 3496
    },
    {
      "epoch": 1.6456470588235295,
      "grad_norm": 0.4531224362926499,
      "learning_rate": 0.00028587300309564675,
      "loss": 0.3398,
      "step": 3497
    },
    {
      "epoch": 1.6461176470588237,
      "grad_norm": 1.1390723444531656,
      "learning_rate": 0.0002858544255405269,
      "loss": 1.0938,
      "step": 3498
    },
    {
      "epoch": 1.6465882352941177,
      "grad_norm": 0.5848363413427065,
      "learning_rate": 0.0002858358363827728,
      "loss": 0.4453,
      "step": 3499
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 0.9825737735174055,
      "learning_rate": 0.000285817235623972,
      "loss": 1.2266,
      "step": 3500
    },
    {
      "epoch": 1.647529411764706,
      "grad_norm": 0.49695887648090653,
      "learning_rate": 0.00028579862326571305,
      "loss": 0.3516,
      "step": 3501
    },
    {
      "epoch": 1.6480000000000001,
      "grad_norm": 0.8449870437014363,
      "learning_rate": 0.0002857799993095856,
      "loss": 0.7109,
      "step": 3502
    },
    {
      "epoch": 1.6484705882352941,
      "grad_norm": 0.5632568473936094,
      "learning_rate": 0.0002857613637571802,
      "loss": 0.6289,
      "step": 3503
    },
    {
      "epoch": 1.6489411764705881,
      "grad_norm": 0.4706432222151201,
      "learning_rate": 0.00028574271661008845,
      "loss": 0.373,
      "step": 3504
    },
    {
      "epoch": 1.6494117647058824,
      "grad_norm": 0.6921977923018484,
      "learning_rate": 0.00028572405786990294,
      "loss": 0.4297,
      "step": 3505
    },
    {
      "epoch": 1.6498823529411766,
      "grad_norm": 0.5819323478581838,
      "learning_rate": 0.00028570538753821704,
      "loss": 0.3672,
      "step": 3506
    },
    {
      "epoch": 1.6503529411764706,
      "grad_norm": 1.4685526409919771,
      "learning_rate": 0.0002856867056166255,
      "loss": 1.2031,
      "step": 3507
    },
    {
      "epoch": 1.6508235294117646,
      "grad_norm": 0.6301753442126323,
      "learning_rate": 0.00028566801210672376,
      "loss": 0.668,
      "step": 3508
    },
    {
      "epoch": 1.6512941176470588,
      "grad_norm": 0.8758572878680992,
      "learning_rate": 0.0002856493070101084,
      "loss": 0.8281,
      "step": 3509
    },
    {
      "epoch": 1.651764705882353,
      "grad_norm": 0.551288811904646,
      "learning_rate": 0.0002856305903283767,
      "loss": 0.4629,
      "step": 3510
    },
    {
      "epoch": 1.652235294117647,
      "grad_norm": 0.6229769229135661,
      "learning_rate": 0.00028561186206312744,
      "loss": 0.3418,
      "step": 3511
    },
    {
      "epoch": 1.652705882352941,
      "grad_norm": 0.5447243715363418,
      "learning_rate": 0.00028559312221595996,
      "loss": 0.3438,
      "step": 3512
    },
    {
      "epoch": 1.6531764705882352,
      "grad_norm": 0.34817772198367114,
      "learning_rate": 0.00028557437078847477,
      "loss": 0.1475,
      "step": 3513
    },
    {
      "epoch": 1.6536470588235295,
      "grad_norm": 0.46652945832134424,
      "learning_rate": 0.0002855556077822733,
      "loss": 0.2852,
      "step": 3514
    },
    {
      "epoch": 1.6541176470588237,
      "grad_norm": 0.8066378042265908,
      "learning_rate": 0.00028553683319895804,
      "loss": 0.4707,
      "step": 3515
    },
    {
      "epoch": 1.6545882352941177,
      "grad_norm": 0.70713453751126,
      "learning_rate": 0.00028551804704013244,
      "loss": 0.6172,
      "step": 3516
    },
    {
      "epoch": 1.6550588235294117,
      "grad_norm": 0.6325563555335357,
      "learning_rate": 0.0002854992493074009,
      "loss": 0.4648,
      "step": 3517
    },
    {
      "epoch": 1.655529411764706,
      "grad_norm": 0.6689633052458899,
      "learning_rate": 0.00028548044000236886,
      "loss": 0.4043,
      "step": 3518
    },
    {
      "epoch": 1.6560000000000001,
      "grad_norm": 0.6061054639518381,
      "learning_rate": 0.0002854616191266427,
      "loss": 0.3555,
      "step": 3519
    },
    {
      "epoch": 1.6564705882352941,
      "grad_norm": 0.6582728516819407,
      "learning_rate": 0.0002854427866818298,
      "loss": 0.6172,
      "step": 3520
    },
    {
      "epoch": 1.6569411764705881,
      "grad_norm": 0.9087080426467662,
      "learning_rate": 0.00028542394266953865,
      "loss": 0.875,
      "step": 3521
    },
    {
      "epoch": 1.6574117647058824,
      "grad_norm": 0.39210226759944,
      "learning_rate": 0.00028540508709137846,
      "loss": 0.2158,
      "step": 3522
    },
    {
      "epoch": 1.6578823529411766,
      "grad_norm": 0.37756127554893076,
      "learning_rate": 0.00028538621994895976,
      "loss": 0.2285,
      "step": 3523
    },
    {
      "epoch": 1.6583529411764706,
      "grad_norm": 0.5878870460537065,
      "learning_rate": 0.00028536734124389373,
      "loss": 0.3184,
      "step": 3524
    },
    {
      "epoch": 1.6588235294117646,
      "grad_norm": 0.2278426246918084,
      "learning_rate": 0.0002853484509777928,
      "loss": 0.0713,
      "step": 3525
    },
    {
      "epoch": 1.6592941176470588,
      "grad_norm": 0.38972741553085555,
      "learning_rate": 0.00028532954915227033,
      "loss": 0.2324,
      "step": 3526
    },
    {
      "epoch": 1.659764705882353,
      "grad_norm": 0.8179364517153702,
      "learning_rate": 0.0002853106357689405,
      "loss": 0.8945,
      "step": 3527
    },
    {
      "epoch": 1.660235294117647,
      "grad_norm": 0.6059318219731182,
      "learning_rate": 0.0002852917108294187,
      "loss": 0.3418,
      "step": 3528
    },
    {
      "epoch": 1.660705882352941,
      "grad_norm": 0.932767269319697,
      "learning_rate": 0.0002852727743353213,
      "loss": 0.7773,
      "step": 3529
    },
    {
      "epoch": 1.6611764705882353,
      "grad_norm": 0.673219321617026,
      "learning_rate": 0.00028525382628826536,
      "loss": 0.6641,
      "step": 3530
    },
    {
      "epoch": 1.6616470588235295,
      "grad_norm": 1.0592027745953239,
      "learning_rate": 0.0002852348666898692,
      "loss": 0.9141,
      "step": 3531
    },
    {
      "epoch": 1.6621176470588235,
      "grad_norm": 0.6420455170133296,
      "learning_rate": 0.0002852158955417522,
      "loss": 0.3711,
      "step": 3532
    },
    {
      "epoch": 1.6625882352941177,
      "grad_norm": 0.727712400323115,
      "learning_rate": 0.0002851969128455345,
      "loss": 0.5625,
      "step": 3533
    },
    {
      "epoch": 1.6630588235294117,
      "grad_norm": 0.3250600781422879,
      "learning_rate": 0.0002851779186028372,
      "loss": 0.1021,
      "step": 3534
    },
    {
      "epoch": 1.663529411764706,
      "grad_norm": 0.9799615186292521,
      "learning_rate": 0.0002851589128152827,
      "loss": 0.5273,
      "step": 3535
    },
    {
      "epoch": 1.6640000000000001,
      "grad_norm": 0.9214219000556669,
      "learning_rate": 0.000285139895484494,
      "loss": 0.6289,
      "step": 3536
    },
    {
      "epoch": 1.6644705882352941,
      "grad_norm": 1.4892879998923148,
      "learning_rate": 0.0002851208666120955,
      "loss": 0.832,
      "step": 3537
    },
    {
      "epoch": 1.6649411764705881,
      "grad_norm": 0.8652857113329901,
      "learning_rate": 0.0002851018261997121,
      "loss": 0.9102,
      "step": 3538
    },
    {
      "epoch": 1.6654117647058824,
      "grad_norm": 0.6692455209795538,
      "learning_rate": 0.00028508277424897015,
      "loss": 0.5234,
      "step": 3539
    },
    {
      "epoch": 1.6658823529411766,
      "grad_norm": 0.7370399618246862,
      "learning_rate": 0.00028506371076149665,
      "loss": 0.7031,
      "step": 3540
    },
    {
      "epoch": 1.6663529411764706,
      "grad_norm": 0.5615654526323977,
      "learning_rate": 0.0002850446357389198,
      "loss": 0.5938,
      "step": 3541
    },
    {
      "epoch": 1.6668235294117646,
      "grad_norm": 0.9245744566829716,
      "learning_rate": 0.00028502554918286867,
      "loss": 0.6484,
      "step": 3542
    },
    {
      "epoch": 1.6672941176470588,
      "grad_norm": 0.6638735888404956,
      "learning_rate": 0.0002850064510949733,
      "loss": 0.6289,
      "step": 3543
    },
    {
      "epoch": 1.667764705882353,
      "grad_norm": 0.9236228894431021,
      "learning_rate": 0.00028498734147686485,
      "loss": 0.5117,
      "step": 3544
    },
    {
      "epoch": 1.668235294117647,
      "grad_norm": 0.5690439708985374,
      "learning_rate": 0.0002849682203301753,
      "loss": 0.4375,
      "step": 3545
    },
    {
      "epoch": 1.668705882352941,
      "grad_norm": 0.46396024444816436,
      "learning_rate": 0.0002849490876565378,
      "loss": 0.4863,
      "step": 3546
    },
    {
      "epoch": 1.6691764705882353,
      "grad_norm": 1.535680275905703,
      "learning_rate": 0.0002849299434575862,
      "loss": 0.1963,
      "step": 3547
    },
    {
      "epoch": 1.6696470588235295,
      "grad_norm": 0.6187811072395218,
      "learning_rate": 0.00028491078773495564,
      "loss": 0.5156,
      "step": 3548
    },
    {
      "epoch": 1.6701176470588235,
      "grad_norm": 0.7672970216306173,
      "learning_rate": 0.00028489162049028205,
      "loss": 0.7695,
      "step": 3549
    },
    {
      "epoch": 1.6705882352941175,
      "grad_norm": 0.8564956385572626,
      "learning_rate": 0.00028487244172520246,
      "loss": 1.0547,
      "step": 3550
    },
    {
      "epoch": 1.6710588235294117,
      "grad_norm": 0.6910696026993304,
      "learning_rate": 0.0002848532514413548,
      "loss": 0.3965,
      "step": 3551
    },
    {
      "epoch": 1.671529411764706,
      "grad_norm": 1.0403488896295559,
      "learning_rate": 0.00028483404964037794,
      "loss": 0.8242,
      "step": 3552
    },
    {
      "epoch": 1.6720000000000002,
      "grad_norm": 0.9602440224345248,
      "learning_rate": 0.000284814836323912,
      "loss": 0.6641,
      "step": 3553
    },
    {
      "epoch": 1.6724705882352942,
      "grad_norm": 0.4114679086077065,
      "learning_rate": 0.0002847956114935977,
      "loss": 0.1138,
      "step": 3554
    },
    {
      "epoch": 1.6729411764705882,
      "grad_norm": 0.9403623113552889,
      "learning_rate": 0.000284776375151077,
      "loss": 0.5977,
      "step": 3555
    },
    {
      "epoch": 1.6734117647058824,
      "grad_norm": 0.39127010687042796,
      "learning_rate": 0.00028475712729799285,
      "loss": 0.1582,
      "step": 3556
    },
    {
      "epoch": 1.6738823529411766,
      "grad_norm": 0.7211387263906489,
      "learning_rate": 0.000284737867935989,
      "loss": 0.6953,
      "step": 3557
    },
    {
      "epoch": 1.6743529411764706,
      "grad_norm": 0.8841987672002491,
      "learning_rate": 0.0002847185970667104,
      "loss": 0.668,
      "step": 3558
    },
    {
      "epoch": 1.6748235294117646,
      "grad_norm": 2.907365621378162,
      "learning_rate": 0.00028469931469180275,
      "loss": 2.5156,
      "step": 3559
    },
    {
      "epoch": 1.6752941176470588,
      "grad_norm": 0.6805942785785574,
      "learning_rate": 0.00028468002081291296,
      "loss": 0.3984,
      "step": 3560
    },
    {
      "epoch": 1.675764705882353,
      "grad_norm": 0.927685980972349,
      "learning_rate": 0.0002846607154316888,
      "loss": 0.6719,
      "step": 3561
    },
    {
      "epoch": 1.676235294117647,
      "grad_norm": 0.5414086017525036,
      "learning_rate": 0.000284641398549779,
      "loss": 0.5273,
      "step": 3562
    },
    {
      "epoch": 1.676705882352941,
      "grad_norm": 0.9781411617604397,
      "learning_rate": 0.00028462207016883336,
      "loss": 0.6328,
      "step": 3563
    },
    {
      "epoch": 1.6771764705882353,
      "grad_norm": 0.726401503208951,
      "learning_rate": 0.0002846027302905026,
      "loss": 0.7266,
      "step": 3564
    },
    {
      "epoch": 1.6776470588235295,
      "grad_norm": 0.5826926231048266,
      "learning_rate": 0.0002845833789164385,
      "loss": 0.5234,
      "step": 3565
    },
    {
      "epoch": 1.6781176470588235,
      "grad_norm": 0.6057191533355558,
      "learning_rate": 0.00028456401604829367,
      "loss": 0.6016,
      "step": 3566
    },
    {
      "epoch": 1.6785882352941175,
      "grad_norm": 1.4313956962491703,
      "learning_rate": 0.00028454464168772186,
      "loss": 0.7656,
      "step": 3567
    },
    {
      "epoch": 1.6790588235294117,
      "grad_norm": 0.4214973456663243,
      "learning_rate": 0.00028452525583637774,
      "loss": 0.2031,
      "step": 3568
    },
    {
      "epoch": 1.679529411764706,
      "grad_norm": 1.0593097367549513,
      "learning_rate": 0.0002845058584959169,
      "loss": 0.5508,
      "step": 3569
    },
    {
      "epoch": 1.6800000000000002,
      "grad_norm": 0.6800184473507142,
      "learning_rate": 0.00028448644966799595,
      "loss": 0.5039,
      "step": 3570
    },
    {
      "epoch": 1.6804705882352942,
      "grad_norm": 0.7136917931051618,
      "learning_rate": 0.0002844670293542726,
      "loss": 0.5938,
      "step": 3571
    },
    {
      "epoch": 1.6809411764705882,
      "grad_norm": 0.4870386259419945,
      "learning_rate": 0.00028444759755640535,
      "loss": 0.3613,
      "step": 3572
    },
    {
      "epoch": 1.6814117647058824,
      "grad_norm": 1.8031090042329936,
      "learning_rate": 0.0002844281542760539,
      "loss": 1.3438,
      "step": 3573
    },
    {
      "epoch": 1.6818823529411766,
      "grad_norm": 1.1831789985082337,
      "learning_rate": 0.0002844086995148786,
      "loss": 0.9141,
      "step": 3574
    },
    {
      "epoch": 1.6823529411764706,
      "grad_norm": 0.4373532873959307,
      "learning_rate": 0.00028438923327454114,
      "loss": 0.1963,
      "step": 3575
    },
    {
      "epoch": 1.6828235294117646,
      "grad_norm": 1.1577089072330613,
      "learning_rate": 0.000284369755556704,
      "loss": 0.8945,
      "step": 3576
    },
    {
      "epoch": 1.6832941176470588,
      "grad_norm": 0.6059473805561175,
      "learning_rate": 0.0002843502663630306,
      "loss": 0.3848,
      "step": 3577
    },
    {
      "epoch": 1.683764705882353,
      "grad_norm": 0.7278292409091865,
      "learning_rate": 0.0002843307656951855,
      "loss": 0.8086,
      "step": 3578
    },
    {
      "epoch": 1.684235294117647,
      "grad_norm": 0.8037470078275394,
      "learning_rate": 0.0002843112535548341,
      "loss": 0.5547,
      "step": 3579
    },
    {
      "epoch": 1.684705882352941,
      "grad_norm": 0.46358285632106955,
      "learning_rate": 0.0002842917299436429,
      "loss": 0.1787,
      "step": 3580
    },
    {
      "epoch": 1.6851764705882353,
      "grad_norm": 0.41548631493576477,
      "learning_rate": 0.00028427219486327923,
      "loss": 0.1387,
      "step": 3581
    },
    {
      "epoch": 1.6856470588235295,
      "grad_norm": 0.610354764992946,
      "learning_rate": 0.00028425264831541153,
      "loss": 0.3691,
      "step": 3582
    },
    {
      "epoch": 1.6861176470588235,
      "grad_norm": 0.5211429216741488,
      "learning_rate": 0.0002842330903017091,
      "loss": 0.2617,
      "step": 3583
    },
    {
      "epoch": 1.6865882352941175,
      "grad_norm": 0.5835490365798376,
      "learning_rate": 0.00028421352082384245,
      "loss": 0.4961,
      "step": 3584
    },
    {
      "epoch": 1.6870588235294117,
      "grad_norm": 0.2511165993002892,
      "learning_rate": 0.00028419393988348275,
      "loss": 0.1475,
      "step": 3585
    },
    {
      "epoch": 1.687529411764706,
      "grad_norm": 0.7769420108198765,
      "learning_rate": 0.00028417434748230237,
      "loss": 0.498,
      "step": 3586
    },
    {
      "epoch": 1.688,
      "grad_norm": 1.1413470174165463,
      "learning_rate": 0.0002841547436219746,
      "loss": 0.6406,
      "step": 3587
    },
    {
      "epoch": 1.6884705882352942,
      "grad_norm": 0.4562798238696041,
      "learning_rate": 0.0002841351283041737,
      "loss": 0.3594,
      "step": 3588
    },
    {
      "epoch": 1.6889411764705882,
      "grad_norm": 0.4832462441682348,
      "learning_rate": 0.0002841155015305749,
      "loss": 0.3398,
      "step": 3589
    },
    {
      "epoch": 1.6894117647058824,
      "grad_norm": 0.46394473526566843,
      "learning_rate": 0.00028409586330285444,
      "loss": 0.1572,
      "step": 3590
    },
    {
      "epoch": 1.6898823529411766,
      "grad_norm": 0.5640544954747362,
      "learning_rate": 0.00028407621362268955,
      "loss": 0.2041,
      "step": 3591
    },
    {
      "epoch": 1.6903529411764706,
      "grad_norm": 0.9177081543984685,
      "learning_rate": 0.0002840565524917583,
      "loss": 0.498,
      "step": 3592
    },
    {
      "epoch": 1.6908235294117646,
      "grad_norm": 0.5917850045143818,
      "learning_rate": 0.00028403687991174003,
      "loss": 0.543,
      "step": 3593
    },
    {
      "epoch": 1.6912941176470588,
      "grad_norm": 0.6110948283625697,
      "learning_rate": 0.0002840171958843147,
      "loss": 0.3047,
      "step": 3594
    },
    {
      "epoch": 1.691764705882353,
      "grad_norm": 1.0050790118562445,
      "learning_rate": 0.0002839975004111635,
      "loss": 0.4336,
      "step": 3595
    },
    {
      "epoch": 1.692235294117647,
      "grad_norm": 0.7761360251883805,
      "learning_rate": 0.00028397779349396854,
      "loss": 0.4492,
      "step": 3596
    },
    {
      "epoch": 1.692705882352941,
      "grad_norm": 0.7018181853249601,
      "learning_rate": 0.0002839580751344128,
      "loss": 0.6914,
      "step": 3597
    },
    {
      "epoch": 1.6931764705882353,
      "grad_norm": 0.7034639577228935,
      "learning_rate": 0.00028393834533418043,
      "loss": 0.5859,
      "step": 3598
    },
    {
      "epoch": 1.6936470588235295,
      "grad_norm": 0.6019252590193531,
      "learning_rate": 0.0002839186040949564,
      "loss": 0.4082,
      "step": 3599
    },
    {
      "epoch": 1.6941176470588235,
      "grad_norm": 1.2359336575215827,
      "learning_rate": 0.0002838988514184267,
      "loss": 0.7812,
      "step": 3600
    },
    {
      "epoch": 1.6945882352941175,
      "grad_norm": 0.9942918540530794,
      "learning_rate": 0.0002838790873062784,
      "loss": 0.7734,
      "step": 3601
    },
    {
      "epoch": 1.6950588235294117,
      "grad_norm": 0.8296642253397282,
      "learning_rate": 0.0002838593117601992,
      "loss": 0.8789,
      "step": 3602
    },
    {
      "epoch": 1.695529411764706,
      "grad_norm": 0.9576442909175039,
      "learning_rate": 0.0002838395247818783,
      "loss": 0.8086,
      "step": 3603
    },
    {
      "epoch": 1.696,
      "grad_norm": 0.26233679771896445,
      "learning_rate": 0.0002838197263730055,
      "loss": 0.054,
      "step": 3604
    },
    {
      "epoch": 1.696470588235294,
      "grad_norm": 0.5781296364964492,
      "learning_rate": 0.0002837999165352717,
      "loss": 0.4824,
      "step": 3605
    },
    {
      "epoch": 1.6969411764705882,
      "grad_norm": 0.5826851908559701,
      "learning_rate": 0.0002837800952703687,
      "loss": 0.2754,
      "step": 3606
    },
    {
      "epoch": 1.6974117647058824,
      "grad_norm": 0.6716163123912666,
      "learning_rate": 0.00028376026257998947,
      "loss": 0.5469,
      "step": 3607
    },
    {
      "epoch": 1.6978823529411766,
      "grad_norm": 1.3938979876480468,
      "learning_rate": 0.0002837404184658276,
      "loss": 0.8594,
      "step": 3608
    },
    {
      "epoch": 1.6983529411764706,
      "grad_norm": 0.8797749955035246,
      "learning_rate": 0.00028372056292957806,
      "loss": 0.5664,
      "step": 3609
    },
    {
      "epoch": 1.6988235294117646,
      "grad_norm": 1.0182731695054896,
      "learning_rate": 0.0002837006959729366,
      "loss": 0.6328,
      "step": 3610
    },
    {
      "epoch": 1.6992941176470588,
      "grad_norm": 0.5850377038996714,
      "learning_rate": 0.00028368081759759983,
      "loss": 0.6367,
      "step": 3611
    },
    {
      "epoch": 1.699764705882353,
      "grad_norm": 0.47911058058085676,
      "learning_rate": 0.00028366092780526554,
      "loss": 0.2637,
      "step": 3612
    },
    {
      "epoch": 1.700235294117647,
      "grad_norm": 0.9973439561199943,
      "learning_rate": 0.00028364102659763247,
      "loss": 0.8398,
      "step": 3613
    },
    {
      "epoch": 1.700705882352941,
      "grad_norm": 0.647189472427287,
      "learning_rate": 0.00028362111397640016,
      "loss": 0.334,
      "step": 3614
    },
    {
      "epoch": 1.7011764705882353,
      "grad_norm": 0.3484083158254792,
      "learning_rate": 0.00028360118994326936,
      "loss": 0.1943,
      "step": 3615
    },
    {
      "epoch": 1.7016470588235295,
      "grad_norm": 0.9280562836014769,
      "learning_rate": 0.00028358125449994156,
      "loss": 0.6133,
      "step": 3616
    },
    {
      "epoch": 1.7021176470588235,
      "grad_norm": 1.2671931923502138,
      "learning_rate": 0.0002835613076481195,
      "loss": 0.7734,
      "step": 3617
    },
    {
      "epoch": 1.7025882352941175,
      "grad_norm": 0.7120260860802421,
      "learning_rate": 0.0002835413493895066,
      "loss": 0.4609,
      "step": 3618
    },
    {
      "epoch": 1.7030588235294117,
      "grad_norm": 0.8836871834418528,
      "learning_rate": 0.00028352137972580745,
      "loss": 0.7891,
      "step": 3619
    },
    {
      "epoch": 1.703529411764706,
      "grad_norm": 0.9351094183980451,
      "learning_rate": 0.0002835013986587276,
      "loss": 1.0547,
      "step": 3620
    },
    {
      "epoch": 1.704,
      "grad_norm": 0.3747962826832087,
      "learning_rate": 0.0002834814061899734,
      "loss": 0.3516,
      "step": 3621
    },
    {
      "epoch": 1.704470588235294,
      "grad_norm": 0.5911856100784645,
      "learning_rate": 0.0002834614023212525,
      "loss": 0.2949,
      "step": 3622
    },
    {
      "epoch": 1.7049411764705882,
      "grad_norm": 0.6841817116057363,
      "learning_rate": 0.0002834413870542731,
      "loss": 0.5039,
      "step": 3623
    },
    {
      "epoch": 1.7054117647058824,
      "grad_norm": 0.7051500422743797,
      "learning_rate": 0.00028342136039074476,
      "loss": 0.5039,
      "step": 3624
    },
    {
      "epoch": 1.7058823529411766,
      "grad_norm": 0.7003288525112249,
      "learning_rate": 0.0002834013223323778,
      "loss": 0.5078,
      "step": 3625
    },
    {
      "epoch": 1.7063529411764706,
      "grad_norm": 0.9491529563966796,
      "learning_rate": 0.0002833812728808836,
      "loss": 0.9336,
      "step": 3626
    },
    {
      "epoch": 1.7068235294117646,
      "grad_norm": 0.8358579282422002,
      "learning_rate": 0.00028336121203797446,
      "loss": 1.0156,
      "step": 3627
    },
    {
      "epoch": 1.7072941176470589,
      "grad_norm": 0.7331436975215098,
      "learning_rate": 0.0002833411398053637,
      "loss": 0.5,
      "step": 3628
    },
    {
      "epoch": 1.707764705882353,
      "grad_norm": 0.9432376146783822,
      "learning_rate": 0.00028332105618476556,
      "loss": 0.5195,
      "step": 3629
    },
    {
      "epoch": 1.708235294117647,
      "grad_norm": 0.7579183743579835,
      "learning_rate": 0.0002833009611778953,
      "loss": 0.3945,
      "step": 3630
    },
    {
      "epoch": 1.708705882352941,
      "grad_norm": 0.802296031431035,
      "learning_rate": 0.0002832808547864691,
      "loss": 0.3457,
      "step": 3631
    },
    {
      "epoch": 1.7091764705882353,
      "grad_norm": 0.7128906885312686,
      "learning_rate": 0.00028326073701220416,
      "loss": 0.6875,
      "step": 3632
    },
    {
      "epoch": 1.7096470588235295,
      "grad_norm": 0.4121236515358932,
      "learning_rate": 0.0002832406078568186,
      "loss": 0.3379,
      "step": 3633
    },
    {
      "epoch": 1.7101176470588235,
      "grad_norm": 0.6021962252469805,
      "learning_rate": 0.00028322046732203165,
      "loss": 0.4414,
      "step": 3634
    },
    {
      "epoch": 1.7105882352941175,
      "grad_norm": 0.5078494952120365,
      "learning_rate": 0.0002832003154095633,
      "loss": 0.4531,
      "step": 3635
    },
    {
      "epoch": 1.7110588235294117,
      "grad_norm": 0.5727810714992835,
      "learning_rate": 0.0002831801521211347,
      "loss": 0.2773,
      "step": 3636
    },
    {
      "epoch": 1.711529411764706,
      "grad_norm": 0.5441105340549813,
      "learning_rate": 0.0002831599774584679,
      "loss": 0.4434,
      "step": 3637
    },
    {
      "epoch": 1.712,
      "grad_norm": 0.7324037308713832,
      "learning_rate": 0.0002831397914232858,
      "loss": 0.4395,
      "step": 3638
    },
    {
      "epoch": 1.712470588235294,
      "grad_norm": 0.6817487502609612,
      "learning_rate": 0.0002831195940173125,
      "loss": 0.4609,
      "step": 3639
    },
    {
      "epoch": 1.7129411764705882,
      "grad_norm": 0.6918979988996429,
      "learning_rate": 0.00028309938524227296,
      "loss": 0.5625,
      "step": 3640
    },
    {
      "epoch": 1.7134117647058824,
      "grad_norm": 0.7749734595569372,
      "learning_rate": 0.000283079165099893,
      "loss": 0.7422,
      "step": 3641
    },
    {
      "epoch": 1.7138823529411766,
      "grad_norm": 0.8620858182242327,
      "learning_rate": 0.0002830589335918997,
      "loss": 0.9414,
      "step": 3642
    },
    {
      "epoch": 1.7143529411764706,
      "grad_norm": 0.5830877048349178,
      "learning_rate": 0.0002830386907200207,
      "loss": 0.5352,
      "step": 3643
    },
    {
      "epoch": 1.7148235294117646,
      "grad_norm": 0.3148430159630115,
      "learning_rate": 0.000283018436485985,
      "loss": 0.0889,
      "step": 3644
    },
    {
      "epoch": 1.7152941176470589,
      "grad_norm": 0.6584317745815224,
      "learning_rate": 0.0002829981708915224,
      "loss": 0.668,
      "step": 3645
    },
    {
      "epoch": 1.715764705882353,
      "grad_norm": 0.7471371682306498,
      "learning_rate": 0.0002829778939383636,
      "loss": 0.8516,
      "step": 3646
    },
    {
      "epoch": 1.716235294117647,
      "grad_norm": 0.6894930642803019,
      "learning_rate": 0.0002829576056282405,
      "loss": 0.2793,
      "step": 3647
    },
    {
      "epoch": 1.716705882352941,
      "grad_norm": 0.6949763926262815,
      "learning_rate": 0.00028293730596288565,
      "loss": 0.7305,
      "step": 3648
    },
    {
      "epoch": 1.7171764705882353,
      "grad_norm": 1.0017592727006894,
      "learning_rate": 0.00028291699494403284,
      "loss": 1.1094,
      "step": 3649
    },
    {
      "epoch": 1.7176470588235295,
      "grad_norm": 0.5366334659928277,
      "learning_rate": 0.0002828966725734167,
      "loss": 0.3691,
      "step": 3650
    },
    {
      "epoch": 1.7181176470588235,
      "grad_norm": 0.31919527016487553,
      "learning_rate": 0.0002828763388527729,
      "loss": 0.3809,
      "step": 3651
    },
    {
      "epoch": 1.7185882352941175,
      "grad_norm": 0.6655130426033885,
      "learning_rate": 0.000282855993783838,
      "loss": 0.3965,
      "step": 3652
    },
    {
      "epoch": 1.7190588235294118,
      "grad_norm": 0.591788990242633,
      "learning_rate": 0.0002828356373683495,
      "loss": 0.4668,
      "step": 3653
    },
    {
      "epoch": 1.719529411764706,
      "grad_norm": 0.5468028870312629,
      "learning_rate": 0.0002828152696080461,
      "loss": 0.3555,
      "step": 3654
    },
    {
      "epoch": 1.72,
      "grad_norm": 0.6991150001584625,
      "learning_rate": 0.0002827948905046672,
      "loss": 0.6133,
      "step": 3655
    },
    {
      "epoch": 1.720470588235294,
      "grad_norm": 0.7307735426417483,
      "learning_rate": 0.00028277450005995326,
      "loss": 0.5586,
      "step": 3656
    },
    {
      "epoch": 1.7209411764705882,
      "grad_norm": 0.45594323212042626,
      "learning_rate": 0.0002827540982756458,
      "loss": 0.2539,
      "step": 3657
    },
    {
      "epoch": 1.7214117647058824,
      "grad_norm": 0.7329179040665491,
      "learning_rate": 0.00028273368515348713,
      "loss": 0.7617,
      "step": 3658
    },
    {
      "epoch": 1.7218823529411764,
      "grad_norm": 0.540948415330195,
      "learning_rate": 0.00028271326069522076,
      "loss": 0.3164,
      "step": 3659
    },
    {
      "epoch": 1.7223529411764706,
      "grad_norm": 0.44175393115398426,
      "learning_rate": 0.0002826928249025909,
      "loss": 0.375,
      "step": 3660
    },
    {
      "epoch": 1.7228235294117646,
      "grad_norm": 0.3946040320741336,
      "learning_rate": 0.000282672377777343,
      "loss": 0.2256,
      "step": 3661
    },
    {
      "epoch": 1.7232941176470589,
      "grad_norm": 0.4342725650178296,
      "learning_rate": 0.0002826519193212233,
      "loss": 0.2656,
      "step": 3662
    },
    {
      "epoch": 1.723764705882353,
      "grad_norm": 0.7471414211108934,
      "learning_rate": 0.000282631449535979,
      "loss": 0.4648,
      "step": 3663
    },
    {
      "epoch": 1.724235294117647,
      "grad_norm": 0.8896513572431123,
      "learning_rate": 0.00028261096842335836,
      "loss": 0.5938,
      "step": 3664
    },
    {
      "epoch": 1.724705882352941,
      "grad_norm": 0.824423677338162,
      "learning_rate": 0.00028259047598511053,
      "loss": 0.7695,
      "step": 3665
    },
    {
      "epoch": 1.7251764705882353,
      "grad_norm": 0.553343867524248,
      "learning_rate": 0.0002825699722229858,
      "loss": 0.5391,
      "step": 3666
    },
    {
      "epoch": 1.7256470588235295,
      "grad_norm": 0.61876590765586,
      "learning_rate": 0.0002825494571387351,
      "loss": 0.4004,
      "step": 3667
    },
    {
      "epoch": 1.7261176470588235,
      "grad_norm": 0.4473630451296737,
      "learning_rate": 0.00028252893073411065,
      "loss": 0.3242,
      "step": 3668
    },
    {
      "epoch": 1.7265882352941175,
      "grad_norm": 0.44786293986367864,
      "learning_rate": 0.00028250839301086544,
      "loss": 0.2432,
      "step": 3669
    },
    {
      "epoch": 1.7270588235294118,
      "grad_norm": 0.5477591267776258,
      "learning_rate": 0.0002824878439707535,
      "loss": 0.3555,
      "step": 3670
    },
    {
      "epoch": 1.727529411764706,
      "grad_norm": 0.4257854708325125,
      "learning_rate": 0.0002824672836155299,
      "loss": 0.248,
      "step": 3671
    },
    {
      "epoch": 1.728,
      "grad_norm": 0.814558972044449,
      "learning_rate": 0.0002824467119469505,
      "loss": 0.5352,
      "step": 3672
    },
    {
      "epoch": 1.728470588235294,
      "grad_norm": 0.7423956860699735,
      "learning_rate": 0.0002824261289667722,
      "loss": 0.6055,
      "step": 3673
    },
    {
      "epoch": 1.7289411764705882,
      "grad_norm": 0.32564340090298166,
      "learning_rate": 0.000282405534676753,
      "loss": 0.1045,
      "step": 3674
    },
    {
      "epoch": 1.7294117647058824,
      "grad_norm": 0.8323880080943173,
      "learning_rate": 0.0002823849290786517,
      "loss": 0.668,
      "step": 3675
    },
    {
      "epoch": 1.7298823529411764,
      "grad_norm": 0.6757889729979161,
      "learning_rate": 0.0002823643121742281,
      "loss": 0.6406,
      "step": 3676
    },
    {
      "epoch": 1.7303529411764704,
      "grad_norm": 0.29166847420649716,
      "learning_rate": 0.000282343683965243,
      "loss": 0.1201,
      "step": 3677
    },
    {
      "epoch": 1.7308235294117646,
      "grad_norm": 0.9690650155808623,
      "learning_rate": 0.0002823230444534581,
      "loss": 0.7383,
      "step": 3678
    },
    {
      "epoch": 1.7312941176470589,
      "grad_norm": 1.3860021105824376,
      "learning_rate": 0.0002823023936406362,
      "loss": 1.7422,
      "step": 3679
    },
    {
      "epoch": 1.731764705882353,
      "grad_norm": 0.6391910883790365,
      "learning_rate": 0.0002822817315285409,
      "loss": 0.3594,
      "step": 3680
    },
    {
      "epoch": 1.732235294117647,
      "grad_norm": 0.6180349563136561,
      "learning_rate": 0.000282261058118937,
      "loss": 0.4941,
      "step": 3681
    },
    {
      "epoch": 1.732705882352941,
      "grad_norm": 0.5391141736982438,
      "learning_rate": 0.00028224037341358993,
      "loss": 0.4453,
      "step": 3682
    },
    {
      "epoch": 1.7331764705882353,
      "grad_norm": 0.846782439394446,
      "learning_rate": 0.0002822196774142663,
      "loss": 0.7148,
      "step": 3683
    },
    {
      "epoch": 1.7336470588235295,
      "grad_norm": 0.5817284508141206,
      "learning_rate": 0.0002821989701227337,
      "loss": 0.4043,
      "step": 3684
    },
    {
      "epoch": 1.7341176470588235,
      "grad_norm": 0.6294100737023718,
      "learning_rate": 0.00028217825154076067,
      "loss": 0.3574,
      "step": 3685
    },
    {
      "epoch": 1.7345882352941175,
      "grad_norm": 0.8300825969622244,
      "learning_rate": 0.0002821575216701166,
      "loss": 0.5352,
      "step": 3686
    },
    {
      "epoch": 1.7350588235294118,
      "grad_norm": 1.1492417927537653,
      "learning_rate": 0.00028213678051257194,
      "loss": 0.9492,
      "step": 3687
    },
    {
      "epoch": 1.735529411764706,
      "grad_norm": 0.7443039916315172,
      "learning_rate": 0.00028211602806989814,
      "loss": 0.3418,
      "step": 3688
    },
    {
      "epoch": 1.736,
      "grad_norm": 0.48518989928862477,
      "learning_rate": 0.00028209526434386744,
      "loss": 0.3008,
      "step": 3689
    },
    {
      "epoch": 1.736470588235294,
      "grad_norm": 0.847842298404246,
      "learning_rate": 0.0002820744893362533,
      "loss": 0.7344,
      "step": 3690
    },
    {
      "epoch": 1.7369411764705882,
      "grad_norm": 0.6751342919915696,
      "learning_rate": 0.0002820537030488299,
      "loss": 0.5117,
      "step": 3691
    },
    {
      "epoch": 1.7374117647058824,
      "grad_norm": 1.0711328013692276,
      "learning_rate": 0.00028203290548337257,
      "loss": 0.8359,
      "step": 3692
    },
    {
      "epoch": 1.7378823529411764,
      "grad_norm": 0.8120928475856347,
      "learning_rate": 0.0002820120966416575,
      "loss": 0.5234,
      "step": 3693
    },
    {
      "epoch": 1.7383529411764704,
      "grad_norm": 0.7350037626721889,
      "learning_rate": 0.0002819912765254619,
      "loss": 0.9141,
      "step": 3694
    },
    {
      "epoch": 1.7388235294117647,
      "grad_norm": 0.6032666199486313,
      "learning_rate": 0.0002819704451365638,
      "loss": 0.3477,
      "step": 3695
    },
    {
      "epoch": 1.7392941176470589,
      "grad_norm": 0.40823726092882817,
      "learning_rate": 0.0002819496024767424,
      "loss": 0.21,
      "step": 3696
    },
    {
      "epoch": 1.739764705882353,
      "grad_norm": 0.5928693833118931,
      "learning_rate": 0.0002819287485477778,
      "loss": 0.1934,
      "step": 3697
    },
    {
      "epoch": 1.740235294117647,
      "grad_norm": 0.664396660884856,
      "learning_rate": 0.0002819078833514509,
      "loss": 0.6133,
      "step": 3698
    },
    {
      "epoch": 1.740705882352941,
      "grad_norm": 0.5674275725124244,
      "learning_rate": 0.0002818870068895437,
      "loss": 0.2832,
      "step": 3699
    },
    {
      "epoch": 1.7411764705882353,
      "grad_norm": 0.939486031754798,
      "learning_rate": 0.0002818661191638393,
      "loss": 0.5508,
      "step": 3700
    },
    {
      "epoch": 1.7416470588235295,
      "grad_norm": 0.7255400980416136,
      "learning_rate": 0.0002818452201761215,
      "loss": 0.6055,
      "step": 3701
    },
    {
      "epoch": 1.7421176470588235,
      "grad_norm": 0.9647812477731322,
      "learning_rate": 0.0002818243099281752,
      "loss": 0.7812,
      "step": 3702
    },
    {
      "epoch": 1.7425882352941175,
      "grad_norm": 0.6446766958597875,
      "learning_rate": 0.00028180338842178623,
      "loss": 0.4297,
      "step": 3703
    },
    {
      "epoch": 1.7430588235294118,
      "grad_norm": 1.0960658373473529,
      "learning_rate": 0.0002817824556587414,
      "loss": 0.9102,
      "step": 3704
    },
    {
      "epoch": 1.743529411764706,
      "grad_norm": 0.9108929740025917,
      "learning_rate": 0.0002817615116408285,
      "loss": 0.7461,
      "step": 3705
    },
    {
      "epoch": 1.744,
      "grad_norm": 0.7229756208942536,
      "learning_rate": 0.0002817405563698362,
      "loss": 0.4512,
      "step": 3706
    },
    {
      "epoch": 1.744470588235294,
      "grad_norm": 0.6396034856908364,
      "learning_rate": 0.0002817195898475542,
      "loss": 0.4043,
      "step": 3707
    },
    {
      "epoch": 1.7449411764705882,
      "grad_norm": 0.6968944600784205,
      "learning_rate": 0.00028169861207577313,
      "loss": 0.5273,
      "step": 3708
    },
    {
      "epoch": 1.7454117647058824,
      "grad_norm": 0.5568019792768144,
      "learning_rate": 0.0002816776230562846,
      "loss": 0.5078,
      "step": 3709
    },
    {
      "epoch": 1.7458823529411764,
      "grad_norm": 0.755568853025853,
      "learning_rate": 0.0002816566227908812,
      "loss": 0.4238,
      "step": 3710
    },
    {
      "epoch": 1.7463529411764704,
      "grad_norm": 1.2863094476739614,
      "learning_rate": 0.00028163561128135645,
      "loss": 1.3438,
      "step": 3711
    },
    {
      "epoch": 1.7468235294117647,
      "grad_norm": 0.6289498483422276,
      "learning_rate": 0.0002816145885295048,
      "loss": 0.582,
      "step": 3712
    },
    {
      "epoch": 1.7472941176470589,
      "grad_norm": 0.5072557794718313,
      "learning_rate": 0.0002815935545371218,
      "loss": 0.4062,
      "step": 3713
    },
    {
      "epoch": 1.7477647058823529,
      "grad_norm": 0.7565083650540234,
      "learning_rate": 0.00028157250930600374,
      "loss": 0.6992,
      "step": 3714
    },
    {
      "epoch": 1.748235294117647,
      "grad_norm": 0.48642390693864623,
      "learning_rate": 0.00028155145283794797,
      "loss": 0.1895,
      "step": 3715
    },
    {
      "epoch": 1.748705882352941,
      "grad_norm": 0.5976598201663426,
      "learning_rate": 0.0002815303851347529,
      "loss": 0.3125,
      "step": 3716
    },
    {
      "epoch": 1.7491764705882353,
      "grad_norm": 0.5108377731656037,
      "learning_rate": 0.0002815093061982178,
      "loss": 0.2031,
      "step": 3717
    },
    {
      "epoch": 1.7496470588235296,
      "grad_norm": 0.7047214252158194,
      "learning_rate": 0.0002814882160301429,
      "loss": 0.5391,
      "step": 3718
    },
    {
      "epoch": 1.7501176470588236,
      "grad_norm": 0.38527919557133916,
      "learning_rate": 0.00028146711463232943,
      "loss": 0.1475,
      "step": 3719
    },
    {
      "epoch": 1.7505882352941176,
      "grad_norm": 0.7334236555748853,
      "learning_rate": 0.0002814460020065795,
      "loss": 0.7812,
      "step": 3720
    },
    {
      "epoch": 1.7510588235294118,
      "grad_norm": 0.597841530067331,
      "learning_rate": 0.00028142487815469627,
      "loss": 0.668,
      "step": 3721
    },
    {
      "epoch": 1.751529411764706,
      "grad_norm": 0.609648193789252,
      "learning_rate": 0.00028140374307848384,
      "loss": 0.4512,
      "step": 3722
    },
    {
      "epoch": 1.752,
      "grad_norm": 0.6131119783667652,
      "learning_rate": 0.00028138259677974714,
      "loss": 0.4785,
      "step": 3723
    },
    {
      "epoch": 1.752470588235294,
      "grad_norm": 0.7849409396050073,
      "learning_rate": 0.0002813614392602923,
      "loss": 0.7578,
      "step": 3724
    },
    {
      "epoch": 1.7529411764705882,
      "grad_norm": 0.5643439352605022,
      "learning_rate": 0.00028134027052192624,
      "loss": 0.5312,
      "step": 3725
    },
    {
      "epoch": 1.7534117647058824,
      "grad_norm": 0.6262696265182391,
      "learning_rate": 0.0002813190905664568,
      "loss": 0.5977,
      "step": 3726
    },
    {
      "epoch": 1.7538823529411764,
      "grad_norm": 0.7351473682265033,
      "learning_rate": 0.0002812978993956929,
      "loss": 0.7266,
      "step": 3727
    },
    {
      "epoch": 1.7543529411764704,
      "grad_norm": 1.133540961131897,
      "learning_rate": 0.0002812766970114445,
      "loss": 0.9883,
      "step": 3728
    },
    {
      "epoch": 1.7548235294117647,
      "grad_norm": 0.19933992788929247,
      "learning_rate": 0.00028125548341552213,
      "loss": 0.0312,
      "step": 3729
    },
    {
      "epoch": 1.755294117647059,
      "grad_norm": 0.43954111386014294,
      "learning_rate": 0.00028123425860973773,
      "loss": 0.3438,
      "step": 3730
    },
    {
      "epoch": 1.755764705882353,
      "grad_norm": 0.666564572923216,
      "learning_rate": 0.000281213022595904,
      "loss": 0.3164,
      "step": 3731
    },
    {
      "epoch": 1.756235294117647,
      "grad_norm": 0.4771968018951241,
      "learning_rate": 0.0002811917753758345,
      "loss": 0.3809,
      "step": 3732
    },
    {
      "epoch": 1.7567058823529411,
      "grad_norm": 0.5515750133886957,
      "learning_rate": 0.00028117051695134384,
      "loss": 0.5,
      "step": 3733
    },
    {
      "epoch": 1.7571764705882353,
      "grad_norm": 0.4887944913555378,
      "learning_rate": 0.0002811492473242477,
      "loss": 0.2812,
      "step": 3734
    },
    {
      "epoch": 1.7576470588235296,
      "grad_norm": 0.676638484080709,
      "learning_rate": 0.00028112796649636254,
      "loss": 0.4043,
      "step": 3735
    },
    {
      "epoch": 1.7581176470588236,
      "grad_norm": 0.6493296662317044,
      "learning_rate": 0.0002811066744695059,
      "loss": 0.3672,
      "step": 3736
    },
    {
      "epoch": 1.7585882352941176,
      "grad_norm": 0.4793829235455271,
      "learning_rate": 0.0002810853712454961,
      "loss": 0.2793,
      "step": 3737
    },
    {
      "epoch": 1.7590588235294118,
      "grad_norm": 0.2691949117807097,
      "learning_rate": 0.00028106405682615265,
      "loss": 0.0608,
      "step": 3738
    },
    {
      "epoch": 1.759529411764706,
      "grad_norm": 0.60952434753471,
      "learning_rate": 0.0002810427312132959,
      "loss": 0.5312,
      "step": 3739
    },
    {
      "epoch": 1.76,
      "grad_norm": 0.519192931302981,
      "learning_rate": 0.00028102139440874715,
      "loss": 0.375,
      "step": 3740
    },
    {
      "epoch": 1.760470588235294,
      "grad_norm": 0.295082709945018,
      "learning_rate": 0.0002810000464143287,
      "loss": 0.1611,
      "step": 3741
    },
    {
      "epoch": 1.7609411764705882,
      "grad_norm": 0.5655593088402526,
      "learning_rate": 0.0002809786872318637,
      "loss": 0.416,
      "step": 3742
    },
    {
      "epoch": 1.7614117647058825,
      "grad_norm": 0.6920446801032476,
      "learning_rate": 0.00028095731686317635,
      "loss": 0.5195,
      "step": 3743
    },
    {
      "epoch": 1.7618823529411765,
      "grad_norm": 0.5552860290232193,
      "learning_rate": 0.0002809359353100918,
      "loss": 0.4844,
      "step": 3744
    },
    {
      "epoch": 1.7623529411764705,
      "grad_norm": 0.36221101661848304,
      "learning_rate": 0.0002809145425744361,
      "loss": 0.1152,
      "step": 3745
    },
    {
      "epoch": 1.7628235294117647,
      "grad_norm": 0.7483869606850422,
      "learning_rate": 0.00028089313865803644,
      "loss": 0.4883,
      "step": 3746
    },
    {
      "epoch": 1.763294117647059,
      "grad_norm": 0.3570649746218718,
      "learning_rate": 0.0002808717235627206,
      "loss": 0.3145,
      "step": 3747
    },
    {
      "epoch": 1.763764705882353,
      "grad_norm": 1.051932917781932,
      "learning_rate": 0.0002808502972903177,
      "loss": 1.0938,
      "step": 3748
    },
    {
      "epoch": 1.764235294117647,
      "grad_norm": 0.5216496535609505,
      "learning_rate": 0.0002808288598426576,
      "loss": 0.3633,
      "step": 3749
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.5580384893205014,
      "learning_rate": 0.0002808074112215711,
      "loss": 0.5586,
      "step": 3750
    },
    {
      "epoch": 1.7651764705882353,
      "grad_norm": 0.6269584122944047,
      "learning_rate": 0.00028078595142889013,
      "loss": 0.4785,
      "step": 3751
    },
    {
      "epoch": 1.7656470588235296,
      "grad_norm": 0.3791264501836267,
      "learning_rate": 0.0002807644804664474,
      "loss": 0.2793,
      "step": 3752
    },
    {
      "epoch": 1.7661176470588236,
      "grad_norm": 0.20315148239719485,
      "learning_rate": 0.0002807429983360766,
      "loss": 0.0518,
      "step": 3753
    },
    {
      "epoch": 1.7665882352941176,
      "grad_norm": 0.4842005273166867,
      "learning_rate": 0.0002807215050396125,
      "loss": 0.2051,
      "step": 3754
    },
    {
      "epoch": 1.7670588235294118,
      "grad_norm": 0.38902343530146144,
      "learning_rate": 0.0002807000005788907,
      "loss": 0.1934,
      "step": 3755
    },
    {
      "epoch": 1.767529411764706,
      "grad_norm": 0.42721648246624944,
      "learning_rate": 0.00028067848495574773,
      "loss": 0.1904,
      "step": 3756
    },
    {
      "epoch": 1.768,
      "grad_norm": 1.1232467969981792,
      "learning_rate": 0.0002806569581720212,
      "loss": 0.9531,
      "step": 3757
    },
    {
      "epoch": 1.768470588235294,
      "grad_norm": 0.7967831778637525,
      "learning_rate": 0.00028063542022954956,
      "loss": 0.6875,
      "step": 3758
    },
    {
      "epoch": 1.7689411764705882,
      "grad_norm": 0.6512289052963648,
      "learning_rate": 0.00028061387113017226,
      "loss": 0.3301,
      "step": 3759
    },
    {
      "epoch": 1.7694117647058825,
      "grad_norm": 0.844192792795544,
      "learning_rate": 0.0002805923108757297,
      "loss": 0.6641,
      "step": 3760
    },
    {
      "epoch": 1.7698823529411765,
      "grad_norm": 0.5435297861150721,
      "learning_rate": 0.00028057073946806323,
      "loss": 0.5664,
      "step": 3761
    },
    {
      "epoch": 1.7703529411764705,
      "grad_norm": 0.5102500646983564,
      "learning_rate": 0.00028054915690901524,
      "loss": 0.5156,
      "step": 3762
    },
    {
      "epoch": 1.7708235294117647,
      "grad_norm": 1.1039478413241928,
      "learning_rate": 0.0002805275632004288,
      "loss": 1.125,
      "step": 3763
    },
    {
      "epoch": 1.771294117647059,
      "grad_norm": 0.6756759617277113,
      "learning_rate": 0.00028050595834414835,
      "loss": 0.6133,
      "step": 3764
    },
    {
      "epoch": 1.771764705882353,
      "grad_norm": 0.5423090312630486,
      "learning_rate": 0.00028048434234201883,
      "loss": 0.4648,
      "step": 3765
    },
    {
      "epoch": 1.772235294117647,
      "grad_norm": 0.5299642287080881,
      "learning_rate": 0.00028046271519588644,
      "loss": 0.3613,
      "step": 3766
    },
    {
      "epoch": 1.7727058823529411,
      "grad_norm": 0.7629841793764996,
      "learning_rate": 0.0002804410769075984,
      "loss": 0.4277,
      "step": 3767
    },
    {
      "epoch": 1.7731764705882354,
      "grad_norm": 0.5658015765559193,
      "learning_rate": 0.0002804194274790024,
      "loss": 0.5977,
      "step": 3768
    },
    {
      "epoch": 1.7736470588235296,
      "grad_norm": 1.5510897326787372,
      "learning_rate": 0.00028039776691194773,
      "loss": 1.0781,
      "step": 3769
    },
    {
      "epoch": 1.7741176470588236,
      "grad_norm": 0.8628697848054986,
      "learning_rate": 0.0002803760952082841,
      "loss": 0.7969,
      "step": 3770
    },
    {
      "epoch": 1.7745882352941176,
      "grad_norm": 0.8125861567489489,
      "learning_rate": 0.0002803544123698624,
      "loss": 0.8164,
      "step": 3771
    },
    {
      "epoch": 1.7750588235294118,
      "grad_norm": 0.7821954355132266,
      "learning_rate": 0.0002803327183985346,
      "loss": 0.4824,
      "step": 3772
    },
    {
      "epoch": 1.775529411764706,
      "grad_norm": 0.4483513400617652,
      "learning_rate": 0.00028031101329615334,
      "loss": 0.2314,
      "step": 3773
    },
    {
      "epoch": 1.776,
      "grad_norm": 0.6654076664100012,
      "learning_rate": 0.00028028929706457235,
      "loss": 0.4434,
      "step": 3774
    },
    {
      "epoch": 1.776470588235294,
      "grad_norm": 0.7023518089663361,
      "learning_rate": 0.0002802675697056463,
      "loss": 0.5078,
      "step": 3775
    },
    {
      "epoch": 1.7769411764705882,
      "grad_norm": 0.7208085318485062,
      "learning_rate": 0.00028024583122123094,
      "loss": 0.582,
      "step": 3776
    },
    {
      "epoch": 1.7774117647058825,
      "grad_norm": 0.6515885112741995,
      "learning_rate": 0.00028022408161318266,
      "loss": 0.3789,
      "step": 3777
    },
    {
      "epoch": 1.7778823529411765,
      "grad_norm": 0.6553135916284841,
      "learning_rate": 0.00028020232088335907,
      "loss": 0.6055,
      "step": 3778
    },
    {
      "epoch": 1.7783529411764705,
      "grad_norm": 0.5440265696060874,
      "learning_rate": 0.00028018054903361864,
      "loss": 0.3438,
      "step": 3779
    },
    {
      "epoch": 1.7788235294117647,
      "grad_norm": 0.9541072992808032,
      "learning_rate": 0.00028015876606582083,
      "loss": 1.1016,
      "step": 3780
    },
    {
      "epoch": 1.779294117647059,
      "grad_norm": 0.6954544002578454,
      "learning_rate": 0.0002801369719818259,
      "loss": 0.5352,
      "step": 3781
    },
    {
      "epoch": 1.779764705882353,
      "grad_norm": 0.8574657240795543,
      "learning_rate": 0.0002801151667834953,
      "loss": 1.0547,
      "step": 3782
    },
    {
      "epoch": 1.780235294117647,
      "grad_norm": 0.5391081305066799,
      "learning_rate": 0.00028009335047269126,
      "loss": 0.3047,
      "step": 3783
    },
    {
      "epoch": 1.7807058823529411,
      "grad_norm": 0.677621258143534,
      "learning_rate": 0.00028007152305127694,
      "loss": 0.7305,
      "step": 3784
    },
    {
      "epoch": 1.7811764705882354,
      "grad_norm": 0.6141835334112624,
      "learning_rate": 0.0002800496845211166,
      "loss": 0.6172,
      "step": 3785
    },
    {
      "epoch": 1.7816470588235294,
      "grad_norm": 0.4216263033455609,
      "learning_rate": 0.0002800278348840753,
      "loss": 0.2041,
      "step": 3786
    },
    {
      "epoch": 1.7821176470588236,
      "grad_norm": 0.3343733607261248,
      "learning_rate": 0.00028000597414201906,
      "loss": 0.25,
      "step": 3787
    },
    {
      "epoch": 1.7825882352941176,
      "grad_norm": 0.6019696313580349,
      "learning_rate": 0.000279984102296815,
      "loss": 0.5078,
      "step": 3788
    },
    {
      "epoch": 1.7830588235294118,
      "grad_norm": 0.487498437586612,
      "learning_rate": 0.000279962219350331,
      "loss": 0.332,
      "step": 3789
    },
    {
      "epoch": 1.783529411764706,
      "grad_norm": 0.7034631239679525,
      "learning_rate": 0.00027994032530443606,
      "loss": 0.4902,
      "step": 3790
    },
    {
      "epoch": 1.784,
      "grad_norm": 0.42267906248649706,
      "learning_rate": 0.00027991842016099994,
      "loss": 0.3027,
      "step": 3791
    },
    {
      "epoch": 1.784470588235294,
      "grad_norm": 0.5655232655986316,
      "learning_rate": 0.0002798965039218936,
      "loss": 0.4277,
      "step": 3792
    },
    {
      "epoch": 1.7849411764705883,
      "grad_norm": 0.4922524509820604,
      "learning_rate": 0.0002798745765889886,
      "loss": 0.3848,
      "step": 3793
    },
    {
      "epoch": 1.7854117647058825,
      "grad_norm": 0.4614641823250527,
      "learning_rate": 0.00027985263816415767,
      "loss": 0.416,
      "step": 3794
    },
    {
      "epoch": 1.7858823529411765,
      "grad_norm": 0.5628954043156744,
      "learning_rate": 0.0002798306886492746,
      "loss": 0.3379,
      "step": 3795
    },
    {
      "epoch": 1.7863529411764705,
      "grad_norm": 0.7801374505713482,
      "learning_rate": 0.00027980872804621385,
      "loss": 0.6211,
      "step": 3796
    },
    {
      "epoch": 1.7868235294117647,
      "grad_norm": 0.7176111436423998,
      "learning_rate": 0.0002797867563568511,
      "loss": 0.6836,
      "step": 3797
    },
    {
      "epoch": 1.787294117647059,
      "grad_norm": 0.4523970630596179,
      "learning_rate": 0.0002797647735830627,
      "loss": 0.4727,
      "step": 3798
    },
    {
      "epoch": 1.787764705882353,
      "grad_norm": 0.5538075541475401,
      "learning_rate": 0.00027974277972672614,
      "loss": 0.3672,
      "step": 3799
    },
    {
      "epoch": 1.788235294117647,
      "grad_norm": 0.5047438874369757,
      "learning_rate": 0.0002797207747897198,
      "loss": 0.2344,
      "step": 3800
    },
    {
      "epoch": 1.7887058823529411,
      "grad_norm": 0.3990982408667749,
      "learning_rate": 0.000279698758773923,
      "loss": 0.1914,
      "step": 3801
    },
    {
      "epoch": 1.7891764705882354,
      "grad_norm": 1.1746258268449252,
      "learning_rate": 0.0002796767316812161,
      "loss": 1.0859,
      "step": 3802
    },
    {
      "epoch": 1.7896470588235294,
      "grad_norm": 0.3824642856176878,
      "learning_rate": 0.0002796546935134802,
      "loss": 0.1816,
      "step": 3803
    },
    {
      "epoch": 1.7901176470588234,
      "grad_norm": 0.8320146938416694,
      "learning_rate": 0.00027963264427259755,
      "loss": 0.7305,
      "step": 3804
    },
    {
      "epoch": 1.7905882352941176,
      "grad_norm": 0.6038528752849655,
      "learning_rate": 0.00027961058396045124,
      "loss": 0.4727,
      "step": 3805
    },
    {
      "epoch": 1.7910588235294118,
      "grad_norm": 0.522219913215883,
      "learning_rate": 0.00027958851257892526,
      "loss": 0.3945,
      "step": 3806
    },
    {
      "epoch": 1.791529411764706,
      "grad_norm": 0.5082153953211784,
      "learning_rate": 0.0002795664301299047,
      "loss": 0.2031,
      "step": 3807
    },
    {
      "epoch": 1.792,
      "grad_norm": 0.41672184959059855,
      "learning_rate": 0.00027954433661527554,
      "loss": 0.2129,
      "step": 3808
    },
    {
      "epoch": 1.792470588235294,
      "grad_norm": 2.1847384829367305,
      "learning_rate": 0.0002795222320369246,
      "loss": 2.4531,
      "step": 3809
    },
    {
      "epoch": 1.7929411764705883,
      "grad_norm": 1.0313147594785548,
      "learning_rate": 0.00027950011639673976,
      "loss": 1.0078,
      "step": 3810
    },
    {
      "epoch": 1.7934117647058825,
      "grad_norm": 0.6971415110561388,
      "learning_rate": 0.00027947798969660975,
      "loss": 0.3965,
      "step": 3811
    },
    {
      "epoch": 1.7938823529411765,
      "grad_norm": 0.36637647644076293,
      "learning_rate": 0.00027945585193842433,
      "loss": 0.2969,
      "step": 3812
    },
    {
      "epoch": 1.7943529411764705,
      "grad_norm": 0.3714052821960347,
      "learning_rate": 0.00027943370312407415,
      "loss": 0.2236,
      "step": 3813
    },
    {
      "epoch": 1.7948235294117647,
      "grad_norm": 1.022696987928234,
      "learning_rate": 0.00027941154325545096,
      "loss": 0.9062,
      "step": 3814
    },
    {
      "epoch": 1.795294117647059,
      "grad_norm": 1.3423241766930316,
      "learning_rate": 0.00027938937233444713,
      "loss": 1.5234,
      "step": 3815
    },
    {
      "epoch": 1.795764705882353,
      "grad_norm": 0.6159244069588707,
      "learning_rate": 0.00027936719036295623,
      "loss": 0.5391,
      "step": 3816
    },
    {
      "epoch": 1.796235294117647,
      "grad_norm": 0.7200952153575689,
      "learning_rate": 0.0002793449973428728,
      "loss": 0.7891,
      "step": 3817
    },
    {
      "epoch": 1.7967058823529412,
      "grad_norm": 0.3017528087718082,
      "learning_rate": 0.0002793227932760921,
      "loss": 0.1562,
      "step": 3818
    },
    {
      "epoch": 1.7971764705882354,
      "grad_norm": 1.3120943574409196,
      "learning_rate": 0.0002793005781645106,
      "loss": 1.1719,
      "step": 3819
    },
    {
      "epoch": 1.7976470588235294,
      "grad_norm": 1.1717446353284193,
      "learning_rate": 0.00027927835201002545,
      "loss": 1.1562,
      "step": 3820
    },
    {
      "epoch": 1.7981176470588234,
      "grad_norm": 0.6264439248131722,
      "learning_rate": 0.00027925611481453497,
      "loss": 0.4531,
      "step": 3821
    },
    {
      "epoch": 1.7985882352941176,
      "grad_norm": 0.8418170790211937,
      "learning_rate": 0.0002792338665799383,
      "loss": 0.6836,
      "step": 3822
    },
    {
      "epoch": 1.7990588235294118,
      "grad_norm": 0.6503388518705518,
      "learning_rate": 0.00027921160730813544,
      "loss": 0.5859,
      "step": 3823
    },
    {
      "epoch": 1.799529411764706,
      "grad_norm": 0.4959426516492372,
      "learning_rate": 0.0002791893370010276,
      "loss": 0.4707,
      "step": 3824
    },
    {
      "epoch": 1.8,
      "grad_norm": 0.9680730328915859,
      "learning_rate": 0.00027916705566051676,
      "loss": 0.8398,
      "step": 3825
    },
    {
      "epoch": 1.800470588235294,
      "grad_norm": 0.8172031292660273,
      "learning_rate": 0.00027914476328850575,
      "loss": 0.7695,
      "step": 3826
    },
    {
      "epoch": 1.8009411764705883,
      "grad_norm": 0.7642243726137664,
      "learning_rate": 0.0002791224598868985,
      "loss": 0.7773,
      "step": 3827
    },
    {
      "epoch": 1.8014117647058825,
      "grad_norm": 0.3392013380421595,
      "learning_rate": 0.00027910014545759986,
      "loss": 0.1396,
      "step": 3828
    },
    {
      "epoch": 1.8018823529411765,
      "grad_norm": 0.39054183011062593,
      "learning_rate": 0.00027907782000251557,
      "loss": 0.2119,
      "step": 3829
    },
    {
      "epoch": 1.8023529411764705,
      "grad_norm": 0.22796444650879705,
      "learning_rate": 0.00027905548352355226,
      "loss": 0.0957,
      "step": 3830
    },
    {
      "epoch": 1.8028235294117647,
      "grad_norm": 0.4416602143285297,
      "learning_rate": 0.00027903313602261774,
      "loss": 0.25,
      "step": 3831
    },
    {
      "epoch": 1.803294117647059,
      "grad_norm": 0.866181967335557,
      "learning_rate": 0.00027901077750162043,
      "loss": 0.6367,
      "step": 3832
    },
    {
      "epoch": 1.803764705882353,
      "grad_norm": 0.5366913523301564,
      "learning_rate": 0.00027898840796246994,
      "loss": 0.2344,
      "step": 3833
    },
    {
      "epoch": 1.804235294117647,
      "grad_norm": 0.894622145748722,
      "learning_rate": 0.00027896602740707673,
      "loss": 0.6719,
      "step": 3834
    },
    {
      "epoch": 1.8047058823529412,
      "grad_norm": 0.644862491686989,
      "learning_rate": 0.0002789436358373522,
      "loss": 0.3516,
      "step": 3835
    },
    {
      "epoch": 1.8051764705882354,
      "grad_norm": 0.7810811318203131,
      "learning_rate": 0.0002789212332552088,
      "loss": 0.7344,
      "step": 3836
    },
    {
      "epoch": 1.8056470588235294,
      "grad_norm": 0.16229182335862963,
      "learning_rate": 0.0002788988196625596,
      "loss": 0.0981,
      "step": 3837
    },
    {
      "epoch": 1.8061176470588234,
      "grad_norm": 0.4133328316625947,
      "learning_rate": 0.000278876395061319,
      "loss": 0.2061,
      "step": 3838
    },
    {
      "epoch": 1.8065882352941176,
      "grad_norm": 0.472251405869169,
      "learning_rate": 0.0002788539594534021,
      "loss": 0.2451,
      "step": 3839
    },
    {
      "epoch": 1.8070588235294118,
      "grad_norm": 0.7621812378757438,
      "learning_rate": 0.0002788315128407251,
      "loss": 0.7227,
      "step": 3840
    },
    {
      "epoch": 1.8075294117647058,
      "grad_norm": 0.47471059400006205,
      "learning_rate": 0.0002788090552252049,
      "loss": 0.2637,
      "step": 3841
    },
    {
      "epoch": 1.808,
      "grad_norm": 0.7747904965939536,
      "learning_rate": 0.00027878658660875967,
      "loss": 0.6836,
      "step": 3842
    },
    {
      "epoch": 1.808470588235294,
      "grad_norm": 0.2744229030570199,
      "learning_rate": 0.0002787641069933082,
      "loss": 0.1582,
      "step": 3843
    },
    {
      "epoch": 1.8089411764705883,
      "grad_norm": 0.48069690072548255,
      "learning_rate": 0.0002787416163807704,
      "loss": 0.2832,
      "step": 3844
    },
    {
      "epoch": 1.8094117647058825,
      "grad_norm": 0.7336713221325567,
      "learning_rate": 0.0002787191147730671,
      "loss": 0.5195,
      "step": 3845
    },
    {
      "epoch": 1.8098823529411765,
      "grad_norm": 0.6394289831343852,
      "learning_rate": 0.00027869660217212005,
      "loss": 0.4551,
      "step": 3846
    },
    {
      "epoch": 1.8103529411764705,
      "grad_norm": 0.5319214735434046,
      "learning_rate": 0.0002786740785798519,
      "loss": 0.2168,
      "step": 3847
    },
    {
      "epoch": 1.8108235294117647,
      "grad_norm": 0.5630552120272935,
      "learning_rate": 0.00027865154399818627,
      "loss": 0.4863,
      "step": 3848
    },
    {
      "epoch": 1.811294117647059,
      "grad_norm": 0.5728543884977327,
      "learning_rate": 0.00027862899842904783,
      "loss": 0.4492,
      "step": 3849
    },
    {
      "epoch": 1.811764705882353,
      "grad_norm": 0.6387469496667497,
      "learning_rate": 0.00027860644187436195,
      "loss": 0.707,
      "step": 3850
    },
    {
      "epoch": 1.812235294117647,
      "grad_norm": 0.25693627818222886,
      "learning_rate": 0.0002785838743360551,
      "loss": 0.0825,
      "step": 3851
    },
    {
      "epoch": 1.8127058823529412,
      "grad_norm": 1.091362806313739,
      "learning_rate": 0.00027856129581605474,
      "loss": 1.0078,
      "step": 3852
    },
    {
      "epoch": 1.8131764705882354,
      "grad_norm": 0.5955095024209821,
      "learning_rate": 0.0002785387063162891,
      "loss": 0.3633,
      "step": 3853
    },
    {
      "epoch": 1.8136470588235294,
      "grad_norm": 0.39869467302212896,
      "learning_rate": 0.00027851610583868745,
      "loss": 0.2354,
      "step": 3854
    },
    {
      "epoch": 1.8141176470588234,
      "grad_norm": 0.6252133883088165,
      "learning_rate": 0.00027849349438518003,
      "loss": 0.4219,
      "step": 3855
    },
    {
      "epoch": 1.8145882352941176,
      "grad_norm": 0.6741101104633557,
      "learning_rate": 0.0002784708719576979,
      "loss": 0.8164,
      "step": 3856
    },
    {
      "epoch": 1.8150588235294118,
      "grad_norm": 0.5950161913565867,
      "learning_rate": 0.00027844823855817315,
      "loss": 0.4844,
      "step": 3857
    },
    {
      "epoch": 1.8155294117647058,
      "grad_norm": 0.23779855570946348,
      "learning_rate": 0.0002784255941885388,
      "loss": 0.1846,
      "step": 3858
    },
    {
      "epoch": 1.8159999999999998,
      "grad_norm": 0.5663447441811974,
      "learning_rate": 0.0002784029388507288,
      "loss": 0.7188,
      "step": 3859
    },
    {
      "epoch": 1.816470588235294,
      "grad_norm": 0.7184797441161199,
      "learning_rate": 0.000278380272546678,
      "loss": 0.5742,
      "step": 3860
    },
    {
      "epoch": 1.8169411764705883,
      "grad_norm": 0.5614924305515817,
      "learning_rate": 0.00027835759527832226,
      "loss": 0.5508,
      "step": 3861
    },
    {
      "epoch": 1.8174117647058825,
      "grad_norm": 0.38644324051522355,
      "learning_rate": 0.0002783349070475982,
      "loss": 0.248,
      "step": 3862
    },
    {
      "epoch": 1.8178823529411765,
      "grad_norm": 0.2607340263343552,
      "learning_rate": 0.00027831220785644366,
      "loss": 0.0571,
      "step": 3863
    },
    {
      "epoch": 1.8183529411764705,
      "grad_norm": 0.6161679324043288,
      "learning_rate": 0.0002782894977067972,
      "loss": 0.4141,
      "step": 3864
    },
    {
      "epoch": 1.8188235294117647,
      "grad_norm": 0.36507296877392953,
      "learning_rate": 0.0002782667766005984,
      "loss": 0.3633,
      "step": 3865
    },
    {
      "epoch": 1.819294117647059,
      "grad_norm": 0.34148035669627536,
      "learning_rate": 0.0002782440445397877,
      "loss": 0.1641,
      "step": 3866
    },
    {
      "epoch": 1.819764705882353,
      "grad_norm": 0.4439808579909843,
      "learning_rate": 0.00027822130152630657,
      "loss": 0.2832,
      "step": 3867
    },
    {
      "epoch": 1.820235294117647,
      "grad_norm": 0.5814260901901954,
      "learning_rate": 0.00027819854756209734,
      "loss": 0.3438,
      "step": 3868
    },
    {
      "epoch": 1.8207058823529412,
      "grad_norm": 0.7657550978323627,
      "learning_rate": 0.00027817578264910343,
      "loss": 0.4473,
      "step": 3869
    },
    {
      "epoch": 1.8211764705882354,
      "grad_norm": 0.5168541382455689,
      "learning_rate": 0.00027815300678926886,
      "loss": 0.168,
      "step": 3870
    },
    {
      "epoch": 1.8216470588235294,
      "grad_norm": 0.6429724159583073,
      "learning_rate": 0.000278130219984539,
      "loss": 0.5352,
      "step": 3871
    },
    {
      "epoch": 1.8221176470588234,
      "grad_norm": 1.655918577687699,
      "learning_rate": 0.00027810742223685983,
      "loss": 1.3672,
      "step": 3872
    },
    {
      "epoch": 1.8225882352941176,
      "grad_norm": 0.3120384659898095,
      "learning_rate": 0.00027808461354817847,
      "loss": 0.1934,
      "step": 3873
    },
    {
      "epoch": 1.8230588235294118,
      "grad_norm": 0.68919271569048,
      "learning_rate": 0.00027806179392044286,
      "loss": 0.6367,
      "step": 3874
    },
    {
      "epoch": 1.8235294117647058,
      "grad_norm": 0.4919641970053821,
      "learning_rate": 0.0002780389633556019,
      "loss": 0.2871,
      "step": 3875
    },
    {
      "epoch": 1.8239999999999998,
      "grad_norm": 0.43186443786827866,
      "learning_rate": 0.00027801612185560554,
      "loss": 0.5977,
      "step": 3876
    },
    {
      "epoch": 1.824470588235294,
      "grad_norm": 0.7357983960681725,
      "learning_rate": 0.00027799326942240434,
      "loss": 0.5117,
      "step": 3877
    },
    {
      "epoch": 1.8249411764705883,
      "grad_norm": 0.3787930247567851,
      "learning_rate": 0.0002779704060579501,
      "loss": 0.1514,
      "step": 3878
    },
    {
      "epoch": 1.8254117647058825,
      "grad_norm": 0.24027686417681998,
      "learning_rate": 0.00027794753176419557,
      "loss": 0.04,
      "step": 3879
    },
    {
      "epoch": 1.8258823529411765,
      "grad_norm": 0.4380668233479901,
      "learning_rate": 0.00027792464654309427,
      "loss": 0.1875,
      "step": 3880
    },
    {
      "epoch": 1.8263529411764705,
      "grad_norm": 0.5360056439844745,
      "learning_rate": 0.0002779017503966006,
      "loss": 0.5547,
      "step": 3881
    },
    {
      "epoch": 1.8268235294117647,
      "grad_norm": 0.5183706714691898,
      "learning_rate": 0.0002778788433266702,
      "loss": 0.4629,
      "step": 3882
    },
    {
      "epoch": 1.827294117647059,
      "grad_norm": 0.6013618647117075,
      "learning_rate": 0.00027785592533525924,
      "loss": 0.5547,
      "step": 3883
    },
    {
      "epoch": 1.827764705882353,
      "grad_norm": 0.3522701587914841,
      "learning_rate": 0.00027783299642432515,
      "loss": 0.2891,
      "step": 3884
    },
    {
      "epoch": 1.828235294117647,
      "grad_norm": 0.791816574909577,
      "learning_rate": 0.0002778100565958262,
      "loss": 1.0234,
      "step": 3885
    },
    {
      "epoch": 1.8287058823529412,
      "grad_norm": 0.8293688953572603,
      "learning_rate": 0.0002777871058517215,
      "loss": 0.7852,
      "step": 3886
    },
    {
      "epoch": 1.8291764705882354,
      "grad_norm": 0.6968985961828907,
      "learning_rate": 0.00027776414419397106,
      "loss": 0.543,
      "step": 3887
    },
    {
      "epoch": 1.8296470588235294,
      "grad_norm": 0.481629729071086,
      "learning_rate": 0.00027774117162453617,
      "loss": 0.4375,
      "step": 3888
    },
    {
      "epoch": 1.8301176470588234,
      "grad_norm": 0.4422046569980391,
      "learning_rate": 0.00027771818814537855,
      "loss": 0.4785,
      "step": 3889
    },
    {
      "epoch": 1.8305882352941176,
      "grad_norm": 0.3703009612282084,
      "learning_rate": 0.00027769519375846126,
      "loss": 0.3164,
      "step": 3890
    },
    {
      "epoch": 1.8310588235294118,
      "grad_norm": 0.34116838478725753,
      "learning_rate": 0.0002776721884657481,
      "loss": 0.418,
      "step": 3891
    },
    {
      "epoch": 1.8315294117647059,
      "grad_norm": 0.7593125000758456,
      "learning_rate": 0.00027764917226920377,
      "loss": 0.5664,
      "step": 3892
    },
    {
      "epoch": 1.8319999999999999,
      "grad_norm": 0.5337856164279724,
      "learning_rate": 0.000277626145170794,
      "loss": 0.5977,
      "step": 3893
    },
    {
      "epoch": 1.832470588235294,
      "grad_norm": 0.7997090101794577,
      "learning_rate": 0.0002776031071724855,
      "loss": 0.6523,
      "step": 3894
    },
    {
      "epoch": 1.8329411764705883,
      "grad_norm": 0.7786974898871846,
      "learning_rate": 0.0002775800582762457,
      "loss": 0.4746,
      "step": 3895
    },
    {
      "epoch": 1.8334117647058825,
      "grad_norm": 0.9244900503717735,
      "learning_rate": 0.00027755699848404315,
      "loss": 0.6523,
      "step": 3896
    },
    {
      "epoch": 1.8338823529411765,
      "grad_norm": 0.9014063269860924,
      "learning_rate": 0.0002775339277978473,
      "loss": 1.0234,
      "step": 3897
    },
    {
      "epoch": 1.8343529411764705,
      "grad_norm": 0.8945514524533574,
      "learning_rate": 0.0002775108462196284,
      "loss": 0.4727,
      "step": 3898
    },
    {
      "epoch": 1.8348235294117647,
      "grad_norm": 0.7894403446837085,
      "learning_rate": 0.0002774877537513579,
      "loss": 0.8008,
      "step": 3899
    },
    {
      "epoch": 1.835294117647059,
      "grad_norm": 0.37397425562347963,
      "learning_rate": 0.0002774646503950078,
      "loss": 0.1982,
      "step": 3900
    },
    {
      "epoch": 1.835764705882353,
      "grad_norm": 0.7469641526377534,
      "learning_rate": 0.0002774415361525514,
      "loss": 0.6758,
      "step": 3901
    },
    {
      "epoch": 1.836235294117647,
      "grad_norm": 0.6890415259927013,
      "learning_rate": 0.00027741841102596276,
      "loss": 0.6875,
      "step": 3902
    },
    {
      "epoch": 1.8367058823529412,
      "grad_norm": 0.7310814572999927,
      "learning_rate": 0.0002773952750172168,
      "loss": 1.1172,
      "step": 3903
    },
    {
      "epoch": 1.8371764705882354,
      "grad_norm": 0.6786771013174253,
      "learning_rate": 0.0002773721281282894,
      "loss": 0.5352,
      "step": 3904
    },
    {
      "epoch": 1.8376470588235294,
      "grad_norm": 0.5938229496250137,
      "learning_rate": 0.0002773489703611576,
      "loss": 0.4746,
      "step": 3905
    },
    {
      "epoch": 1.8381176470588234,
      "grad_norm": 0.9062197191099887,
      "learning_rate": 0.00027732580171779906,
      "loss": 0.4961,
      "step": 3906
    },
    {
      "epoch": 1.8385882352941176,
      "grad_norm": 0.873027136154233,
      "learning_rate": 0.0002773026222001925,
      "loss": 0.7305,
      "step": 3907
    },
    {
      "epoch": 1.8390588235294119,
      "grad_norm": 0.787482930942999,
      "learning_rate": 0.00027727943181031766,
      "loss": 0.6719,
      "step": 3908
    },
    {
      "epoch": 1.8395294117647059,
      "grad_norm": 0.6200634471520419,
      "learning_rate": 0.00027725623055015495,
      "loss": 0.4844,
      "step": 3909
    },
    {
      "epoch": 1.8399999999999999,
      "grad_norm": 0.38974259820484697,
      "learning_rate": 0.000277233018421686,
      "loss": 0.1445,
      "step": 3910
    },
    {
      "epoch": 1.840470588235294,
      "grad_norm": 0.6980801444126857,
      "learning_rate": 0.00027720979542689324,
      "loss": 0.6484,
      "step": 3911
    },
    {
      "epoch": 1.8409411764705883,
      "grad_norm": 0.5900989669251462,
      "learning_rate": 0.0002771865615677599,
      "loss": 0.3027,
      "step": 3912
    },
    {
      "epoch": 1.8414117647058823,
      "grad_norm": 0.8303671305551333,
      "learning_rate": 0.00027716331684627044,
      "loss": 0.6641,
      "step": 3913
    },
    {
      "epoch": 1.8418823529411765,
      "grad_norm": 1.2485811260799975,
      "learning_rate": 0.00027714006126441,
      "loss": 0.8359,
      "step": 3914
    },
    {
      "epoch": 1.8423529411764705,
      "grad_norm": 0.6824619995974542,
      "learning_rate": 0.00027711679482416465,
      "loss": 0.7109,
      "step": 3915
    },
    {
      "epoch": 1.8428235294117647,
      "grad_norm": 0.7263852289271132,
      "learning_rate": 0.0002770935175275215,
      "loss": 0.5781,
      "step": 3916
    },
    {
      "epoch": 1.843294117647059,
      "grad_norm": 0.32768151055534134,
      "learning_rate": 0.0002770702293764686,
      "loss": 0.1147,
      "step": 3917
    },
    {
      "epoch": 1.843764705882353,
      "grad_norm": 0.7586409543036547,
      "learning_rate": 0.0002770469303729949,
      "loss": 0.5117,
      "step": 3918
    },
    {
      "epoch": 1.844235294117647,
      "grad_norm": 0.4699866472248715,
      "learning_rate": 0.0002770236205190901,
      "loss": 0.4258,
      "step": 3919
    },
    {
      "epoch": 1.8447058823529412,
      "grad_norm": 0.655793056682084,
      "learning_rate": 0.00027700029981674506,
      "loss": 0.4668,
      "step": 3920
    },
    {
      "epoch": 1.8451764705882354,
      "grad_norm": 0.3584806707164074,
      "learning_rate": 0.0002769769682679515,
      "loss": 0.165,
      "step": 3921
    },
    {
      "epoch": 1.8456470588235294,
      "grad_norm": 0.4899691973024172,
      "learning_rate": 0.00027695362587470203,
      "loss": 0.3477,
      "step": 3922
    },
    {
      "epoch": 1.8461176470588234,
      "grad_norm": 0.6152275139039254,
      "learning_rate": 0.0002769302726389902,
      "loss": 0.4336,
      "step": 3923
    },
    {
      "epoch": 1.8465882352941176,
      "grad_norm": 1.2325820136295662,
      "learning_rate": 0.00027690690856281054,
      "loss": 0.7617,
      "step": 3924
    },
    {
      "epoch": 1.8470588235294119,
      "grad_norm": 0.51854381261653,
      "learning_rate": 0.00027688353364815834,
      "loss": 0.2676,
      "step": 3925
    },
    {
      "epoch": 1.8475294117647059,
      "grad_norm": 0.5780166795081341,
      "learning_rate": 0.0002768601478970301,
      "loss": 0.4863,
      "step": 3926
    },
    {
      "epoch": 1.8479999999999999,
      "grad_norm": 0.7129189035341507,
      "learning_rate": 0.00027683675131142294,
      "loss": 0.832,
      "step": 3927
    },
    {
      "epoch": 1.848470588235294,
      "grad_norm": 0.5847182802617522,
      "learning_rate": 0.000276813343893335,
      "loss": 0.5898,
      "step": 3928
    },
    {
      "epoch": 1.8489411764705883,
      "grad_norm": 0.39713683564785723,
      "learning_rate": 0.0002767899256447656,
      "loss": 0.334,
      "step": 3929
    },
    {
      "epoch": 1.8494117647058823,
      "grad_norm": 0.5987574725851941,
      "learning_rate": 0.0002767664965677146,
      "loss": 0.6875,
      "step": 3930
    },
    {
      "epoch": 1.8498823529411763,
      "grad_norm": 0.3268710432643319,
      "learning_rate": 0.000276743056664183,
      "loss": 0.0898,
      "step": 3931
    },
    {
      "epoch": 1.8503529411764705,
      "grad_norm": 0.32853357706786185,
      "learning_rate": 0.0002767196059361727,
      "loss": 0.126,
      "step": 3932
    },
    {
      "epoch": 1.8508235294117648,
      "grad_norm": 0.4666917574654503,
      "learning_rate": 0.0002766961443856865,
      "loss": 0.5234,
      "step": 3933
    },
    {
      "epoch": 1.851294117647059,
      "grad_norm": 0.7022391775896999,
      "learning_rate": 0.00027667267201472815,
      "loss": 0.5547,
      "step": 3934
    },
    {
      "epoch": 1.851764705882353,
      "grad_norm": 0.8132141381973707,
      "learning_rate": 0.00027664918882530225,
      "loss": 0.8633,
      "step": 3935
    },
    {
      "epoch": 1.852235294117647,
      "grad_norm": 0.6001508025528643,
      "learning_rate": 0.0002766256948194144,
      "loss": 0.4922,
      "step": 3936
    },
    {
      "epoch": 1.8527058823529412,
      "grad_norm": 0.7456427774825428,
      "learning_rate": 0.0002766021899990712,
      "loss": 0.6328,
      "step": 3937
    },
    {
      "epoch": 1.8531764705882354,
      "grad_norm": 0.4960640667497519,
      "learning_rate": 0.0002765786743662799,
      "loss": 0.208,
      "step": 3938
    },
    {
      "epoch": 1.8536470588235294,
      "grad_norm": 0.8789243670712015,
      "learning_rate": 0.0002765551479230489,
      "loss": 0.8203,
      "step": 3939
    },
    {
      "epoch": 1.8541176470588234,
      "grad_norm": 0.5386286590457142,
      "learning_rate": 0.00027653161067138764,
      "loss": 0.2793,
      "step": 3940
    },
    {
      "epoch": 1.8545882352941176,
      "grad_norm": 0.6066955248842688,
      "learning_rate": 0.00027650806261330616,
      "loss": 0.4512,
      "step": 3941
    },
    {
      "epoch": 1.8550588235294119,
      "grad_norm": 0.9940810944054727,
      "learning_rate": 0.0002764845037508156,
      "loss": 0.9336,
      "step": 3942
    },
    {
      "epoch": 1.8555294117647059,
      "grad_norm": 0.6156105674666339,
      "learning_rate": 0.00027646093408592804,
      "loss": 0.5469,
      "step": 3943
    },
    {
      "epoch": 1.8559999999999999,
      "grad_norm": 0.7237920845771668,
      "learning_rate": 0.0002764373536206564,
      "loss": 0.1621,
      "step": 3944
    },
    {
      "epoch": 1.856470588235294,
      "grad_norm": 0.5503778976657888,
      "learning_rate": 0.0002764137623570146,
      "loss": 0.2773,
      "step": 3945
    },
    {
      "epoch": 1.8569411764705883,
      "grad_norm": 0.39010950181034304,
      "learning_rate": 0.0002763901602970174,
      "loss": 0.1738,
      "step": 3946
    },
    {
      "epoch": 1.8574117647058823,
      "grad_norm": 1.0387936430908393,
      "learning_rate": 0.0002763665474426806,
      "loss": 1.2812,
      "step": 3947
    },
    {
      "epoch": 1.8578823529411763,
      "grad_norm": 0.8533205467467522,
      "learning_rate": 0.00027634292379602086,
      "loss": 0.668,
      "step": 3948
    },
    {
      "epoch": 1.8583529411764705,
      "grad_norm": 0.4159272841292245,
      "learning_rate": 0.00027631928935905567,
      "loss": 0.1934,
      "step": 3949
    },
    {
      "epoch": 1.8588235294117648,
      "grad_norm": 0.6021963537254054,
      "learning_rate": 0.0002762956441338036,
      "loss": 0.4316,
      "step": 3950
    },
    {
      "epoch": 1.859294117647059,
      "grad_norm": 0.7053793181072461,
      "learning_rate": 0.00027627198812228405,
      "loss": 0.4336,
      "step": 3951
    },
    {
      "epoch": 1.859764705882353,
      "grad_norm": 0.7097309206289655,
      "learning_rate": 0.0002762483213265173,
      "loss": 0.6133,
      "step": 3952
    },
    {
      "epoch": 1.860235294117647,
      "grad_norm": 0.5310776503679604,
      "learning_rate": 0.0002762246437485247,
      "loss": 0.2266,
      "step": 3953
    },
    {
      "epoch": 1.8607058823529412,
      "grad_norm": 0.6763274567049191,
      "learning_rate": 0.0002762009553903284,
      "loss": 0.7266,
      "step": 3954
    },
    {
      "epoch": 1.8611764705882354,
      "grad_norm": 0.8350701575135364,
      "learning_rate": 0.0002761772562539515,
      "loss": 0.7148,
      "step": 3955
    },
    {
      "epoch": 1.8616470588235294,
      "grad_norm": 0.6167527961793426,
      "learning_rate": 0.0002761535463414181,
      "loss": 0.6914,
      "step": 3956
    },
    {
      "epoch": 1.8621176470588234,
      "grad_norm": 0.4314255527828077,
      "learning_rate": 0.0002761298256547529,
      "loss": 0.2451,
      "step": 3957
    },
    {
      "epoch": 1.8625882352941177,
      "grad_norm": 0.6685826603209386,
      "learning_rate": 0.00027610609419598204,
      "loss": 1.0391,
      "step": 3958
    },
    {
      "epoch": 1.8630588235294119,
      "grad_norm": 0.7174059731772868,
      "learning_rate": 0.0002760823519671321,
      "loss": 0.6094,
      "step": 3959
    },
    {
      "epoch": 1.8635294117647059,
      "grad_norm": 0.7652942611723236,
      "learning_rate": 0.0002760585989702309,
      "loss": 0.625,
      "step": 3960
    },
    {
      "epoch": 1.8639999999999999,
      "grad_norm": 0.4731061824738193,
      "learning_rate": 0.0002760348352073071,
      "loss": 0.2012,
      "step": 3961
    },
    {
      "epoch": 1.864470588235294,
      "grad_norm": 0.5042307048071627,
      "learning_rate": 0.0002760110606803901,
      "loss": 0.2373,
      "step": 3962
    },
    {
      "epoch": 1.8649411764705883,
      "grad_norm": 0.8614675149704194,
      "learning_rate": 0.00027598727539151054,
      "loss": 0.5352,
      "step": 3963
    },
    {
      "epoch": 1.8654117647058823,
      "grad_norm": 0.8130215042342522,
      "learning_rate": 0.0002759634793426996,
      "loss": 0.6992,
      "step": 3964
    },
    {
      "epoch": 1.8658823529411763,
      "grad_norm": 0.45616358763552695,
      "learning_rate": 0.00027593967253598974,
      "loss": 0.2021,
      "step": 3965
    },
    {
      "epoch": 1.8663529411764705,
      "grad_norm": 0.7547290781655868,
      "learning_rate": 0.00027591585497341406,
      "loss": 0.7422,
      "step": 3966
    },
    {
      "epoch": 1.8668235294117648,
      "grad_norm": 0.32501372675611395,
      "learning_rate": 0.00027589202665700676,
      "loss": 0.3027,
      "step": 3967
    },
    {
      "epoch": 1.8672941176470588,
      "grad_norm": 0.41858268300616674,
      "learning_rate": 0.0002758681875888029,
      "loss": 0.209,
      "step": 3968
    },
    {
      "epoch": 1.867764705882353,
      "grad_norm": 0.7883302331585545,
      "learning_rate": 0.00027584433777083845,
      "loss": 0.5781,
      "step": 3969
    },
    {
      "epoch": 1.868235294117647,
      "grad_norm": 0.8298017383937216,
      "learning_rate": 0.0002758204772051503,
      "loss": 0.6992,
      "step": 3970
    },
    {
      "epoch": 1.8687058823529412,
      "grad_norm": 0.5743887820869097,
      "learning_rate": 0.00027579660589377626,
      "loss": 0.2793,
      "step": 3971
    },
    {
      "epoch": 1.8691764705882354,
      "grad_norm": 1.0739539078649853,
      "learning_rate": 0.00027577272383875497,
      "loss": 0.707,
      "step": 3972
    },
    {
      "epoch": 1.8696470588235294,
      "grad_norm": 0.7177942375675294,
      "learning_rate": 0.0002757488310421262,
      "loss": 0.6484,
      "step": 3973
    },
    {
      "epoch": 1.8701176470588234,
      "grad_norm": 0.731328135893135,
      "learning_rate": 0.0002757249275059305,
      "loss": 0.6055,
      "step": 3974
    },
    {
      "epoch": 1.8705882352941177,
      "grad_norm": 1.0585398607353789,
      "learning_rate": 0.00027570101323220927,
      "loss": 0.9688,
      "step": 3975
    },
    {
      "epoch": 1.8710588235294119,
      "grad_norm": 0.7411030214528306,
      "learning_rate": 0.00027567708822300496,
      "loss": 0.6328,
      "step": 3976
    },
    {
      "epoch": 1.8715294117647059,
      "grad_norm": 0.7053208789746354,
      "learning_rate": 0.0002756531524803609,
      "loss": 0.3887,
      "step": 3977
    },
    {
      "epoch": 1.8719999999999999,
      "grad_norm": 0.4547094323943761,
      "learning_rate": 0.00027562920600632127,
      "loss": 0.3945,
      "step": 3978
    },
    {
      "epoch": 1.872470588235294,
      "grad_norm": 0.6594210511038281,
      "learning_rate": 0.0002756052488029312,
      "loss": 0.7773,
      "step": 3979
    },
    {
      "epoch": 1.8729411764705883,
      "grad_norm": 0.5605126959567968,
      "learning_rate": 0.0002755812808722368,
      "loss": 0.4531,
      "step": 3980
    },
    {
      "epoch": 1.8734117647058823,
      "grad_norm": 0.417489031165668,
      "learning_rate": 0.0002755573022162851,
      "loss": 0.25,
      "step": 3981
    },
    {
      "epoch": 1.8738823529411763,
      "grad_norm": 0.49619170405527624,
      "learning_rate": 0.0002755333128371239,
      "loss": 0.4355,
      "step": 3982
    },
    {
      "epoch": 1.8743529411764706,
      "grad_norm": 0.8772351975268169,
      "learning_rate": 0.00027550931273680207,
      "loss": 1.0938,
      "step": 3983
    },
    {
      "epoch": 1.8748235294117648,
      "grad_norm": 0.8691992442182258,
      "learning_rate": 0.00027548530191736926,
      "loss": 0.582,
      "step": 3984
    },
    {
      "epoch": 1.8752941176470588,
      "grad_norm": 0.6490141202156395,
      "learning_rate": 0.0002754612803808762,
      "loss": 0.7617,
      "step": 3985
    },
    {
      "epoch": 1.8757647058823528,
      "grad_norm": 0.40349771326396483,
      "learning_rate": 0.00027543724812937443,
      "loss": 0.2432,
      "step": 3986
    },
    {
      "epoch": 1.876235294117647,
      "grad_norm": 0.7213044956019453,
      "learning_rate": 0.0002754132051649164,
      "loss": 0.6523,
      "step": 3987
    },
    {
      "epoch": 1.8767058823529412,
      "grad_norm": 0.777512832193699,
      "learning_rate": 0.00027538915148955545,
      "loss": 0.5625,
      "step": 3988
    },
    {
      "epoch": 1.8771764705882354,
      "grad_norm": 0.6784285336018275,
      "learning_rate": 0.000275365087105346,
      "loss": 0.7383,
      "step": 3989
    },
    {
      "epoch": 1.8776470588235294,
      "grad_norm": 0.6848531694241873,
      "learning_rate": 0.00027534101201434315,
      "loss": 0.4238,
      "step": 3990
    },
    {
      "epoch": 1.8781176470588234,
      "grad_norm": 1.087117252477535,
      "learning_rate": 0.00027531692621860307,
      "loss": 1.1016,
      "step": 3991
    },
    {
      "epoch": 1.8785882352941177,
      "grad_norm": 0.7549096127882149,
      "learning_rate": 0.00027529282972018287,
      "loss": 0.9062,
      "step": 3992
    },
    {
      "epoch": 1.879058823529412,
      "grad_norm": 0.670022664267468,
      "learning_rate": 0.0002752687225211404,
      "loss": 0.5312,
      "step": 3993
    },
    {
      "epoch": 1.879529411764706,
      "grad_norm": 0.43975575363710045,
      "learning_rate": 0.0002752446046235346,
      "loss": 0.2061,
      "step": 3994
    },
    {
      "epoch": 1.88,
      "grad_norm": 0.4024942368080892,
      "learning_rate": 0.00027522047602942525,
      "loss": 0.1514,
      "step": 3995
    },
    {
      "epoch": 1.8804705882352941,
      "grad_norm": 0.7244802839293341,
      "learning_rate": 0.00027519633674087305,
      "loss": 0.6914,
      "step": 3996
    },
    {
      "epoch": 1.8809411764705883,
      "grad_norm": 0.6704604092926951,
      "learning_rate": 0.0002751721867599396,
      "loss": 0.5391,
      "step": 3997
    },
    {
      "epoch": 1.8814117647058823,
      "grad_norm": 0.293547802002523,
      "learning_rate": 0.00027514802608868743,
      "loss": 0.1416,
      "step": 3998
    },
    {
      "epoch": 1.8818823529411763,
      "grad_norm": 0.6259440882785536,
      "learning_rate": 0.00027512385472918,
      "loss": 0.3691,
      "step": 3999
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 0.5339084928643123,
      "learning_rate": 0.0002750996726834817,
      "loss": 0.5352,
      "step": 4000
    },
    {
      "epoch": 1.8828235294117648,
      "grad_norm": 0.6557802393675713,
      "learning_rate": 0.00027507547995365766,
      "loss": 0.6484,
      "step": 4001
    },
    {
      "epoch": 1.8832941176470588,
      "grad_norm": 0.5047431523413416,
      "learning_rate": 0.00027505127654177413,
      "loss": 0.2617,
      "step": 4002
    },
    {
      "epoch": 1.8837647058823528,
      "grad_norm": 1.176575587761762,
      "learning_rate": 0.0002750270624498983,
      "loss": 0.7344,
      "step": 4003
    },
    {
      "epoch": 1.884235294117647,
      "grad_norm": 0.3471264085901988,
      "learning_rate": 0.000275002837680098,
      "loss": 0.1309,
      "step": 4004
    },
    {
      "epoch": 1.8847058823529412,
      "grad_norm": 0.8717034771891344,
      "learning_rate": 0.0002749786022344422,
      "loss": 0.832,
      "step": 4005
    },
    {
      "epoch": 1.8851764705882355,
      "grad_norm": 0.65800369262309,
      "learning_rate": 0.0002749543561150009,
      "loss": 0.3457,
      "step": 4006
    },
    {
      "epoch": 1.8856470588235295,
      "grad_norm": 0.3513336360982513,
      "learning_rate": 0.00027493009932384456,
      "loss": 0.2188,
      "step": 4007
    },
    {
      "epoch": 1.8861176470588235,
      "grad_norm": 0.4535870086119682,
      "learning_rate": 0.000274905831863045,
      "loss": 0.2441,
      "step": 4008
    },
    {
      "epoch": 1.8865882352941177,
      "grad_norm": 0.6248469038849802,
      "learning_rate": 0.00027488155373467474,
      "loss": 0.3301,
      "step": 4009
    },
    {
      "epoch": 1.887058823529412,
      "grad_norm": 0.5378859773560352,
      "learning_rate": 0.00027485726494080724,
      "loss": 0.2676,
      "step": 4010
    },
    {
      "epoch": 1.887529411764706,
      "grad_norm": 0.8225477265748647,
      "learning_rate": 0.0002748329654835169,
      "loss": 0.2178,
      "step": 4011
    },
    {
      "epoch": 1.888,
      "grad_norm": 1.1001186393129685,
      "learning_rate": 0.000274808655364879,
      "loss": 1.0391,
      "step": 4012
    },
    {
      "epoch": 1.8884705882352941,
      "grad_norm": 0.6468093211223629,
      "learning_rate": 0.00027478433458696976,
      "loss": 0.6367,
      "step": 4013
    },
    {
      "epoch": 1.8889411764705883,
      "grad_norm": 0.44383083248631483,
      "learning_rate": 0.0002747600031518663,
      "loss": 0.3574,
      "step": 4014
    },
    {
      "epoch": 1.8894117647058823,
      "grad_norm": 0.48344440634795244,
      "learning_rate": 0.0002747356610616466,
      "loss": 0.2207,
      "step": 4015
    },
    {
      "epoch": 1.8898823529411763,
      "grad_norm": 0.8212469252062574,
      "learning_rate": 0.0002747113083183896,
      "loss": 0.7852,
      "step": 4016
    },
    {
      "epoch": 1.8903529411764706,
      "grad_norm": 0.5375110301502413,
      "learning_rate": 0.00027468694492417515,
      "loss": 0.5,
      "step": 4017
    },
    {
      "epoch": 1.8908235294117648,
      "grad_norm": 0.6086126004377057,
      "learning_rate": 0.0002746625708810841,
      "loss": 0.5039,
      "step": 4018
    },
    {
      "epoch": 1.8912941176470588,
      "grad_norm": 1.027446028815243,
      "learning_rate": 0.00027463818619119794,
      "loss": 0.7852,
      "step": 4019
    },
    {
      "epoch": 1.8917647058823528,
      "grad_norm": 0.38614439084198093,
      "learning_rate": 0.0002746137908565994,
      "loss": 0.1504,
      "step": 4020
    },
    {
      "epoch": 1.892235294117647,
      "grad_norm": 0.4025511960323414,
      "learning_rate": 0.00027458938487937186,
      "loss": 0.3301,
      "step": 4021
    },
    {
      "epoch": 1.8927058823529412,
      "grad_norm": 0.5168401155811978,
      "learning_rate": 0.0002745649682615998,
      "loss": 0.2559,
      "step": 4022
    },
    {
      "epoch": 1.8931764705882355,
      "grad_norm": 1.2006092855258759,
      "learning_rate": 0.00027454054100536836,
      "loss": 0.7344,
      "step": 4023
    },
    {
      "epoch": 1.8936470588235295,
      "grad_norm": 0.33649389430983917,
      "learning_rate": 0.00027451610311276397,
      "loss": 0.1177,
      "step": 4024
    },
    {
      "epoch": 1.8941176470588235,
      "grad_norm": 1.0547161585055058,
      "learning_rate": 0.00027449165458587354,
      "loss": 0.793,
      "step": 4025
    },
    {
      "epoch": 1.8945882352941177,
      "grad_norm": 0.3716007206022911,
      "learning_rate": 0.0002744671954267852,
      "loss": 0.1748,
      "step": 4026
    },
    {
      "epoch": 1.895058823529412,
      "grad_norm": 0.9413736594162492,
      "learning_rate": 0.00027444272563758784,
      "loss": 0.7422,
      "step": 4027
    },
    {
      "epoch": 1.895529411764706,
      "grad_norm": 0.7948234638932443,
      "learning_rate": 0.00027441824522037137,
      "loss": 0.7344,
      "step": 4028
    },
    {
      "epoch": 1.896,
      "grad_norm": 0.9965553983668861,
      "learning_rate": 0.0002743937541772265,
      "loss": 0.9492,
      "step": 4029
    },
    {
      "epoch": 1.8964705882352941,
      "grad_norm": 0.6353893565290767,
      "learning_rate": 0.0002743692525102448,
      "loss": 0.543,
      "step": 4030
    },
    {
      "epoch": 1.8969411764705884,
      "grad_norm": 0.7715123952250846,
      "learning_rate": 0.000274344740221519,
      "loss": 0.7227,
      "step": 4031
    },
    {
      "epoch": 1.8974117647058824,
      "grad_norm": 0.6004283872845986,
      "learning_rate": 0.0002743202173131424,
      "loss": 0.3926,
      "step": 4032
    },
    {
      "epoch": 1.8978823529411764,
      "grad_norm": 0.3690191713028451,
      "learning_rate": 0.00027429568378720945,
      "loss": 0.1621,
      "step": 4033
    },
    {
      "epoch": 1.8983529411764706,
      "grad_norm": 0.5762600131852833,
      "learning_rate": 0.0002742711396458155,
      "loss": 0.7109,
      "step": 4034
    },
    {
      "epoch": 1.8988235294117648,
      "grad_norm": 1.0317129379367862,
      "learning_rate": 0.00027424658489105667,
      "loss": 0.7969,
      "step": 4035
    },
    {
      "epoch": 1.8992941176470588,
      "grad_norm": 0.52472123479117,
      "learning_rate": 0.00027422201952503005,
      "loss": 0.3574,
      "step": 4036
    },
    {
      "epoch": 1.8997647058823528,
      "grad_norm": 0.7164135531939753,
      "learning_rate": 0.00027419744354983364,
      "loss": 0.4512,
      "step": 4037
    },
    {
      "epoch": 1.900235294117647,
      "grad_norm": 0.8412985043639228,
      "learning_rate": 0.00027417285696756634,
      "loss": 0.9453,
      "step": 4038
    },
    {
      "epoch": 1.9007058823529412,
      "grad_norm": 0.7374483646025264,
      "learning_rate": 0.00027414825978032805,
      "loss": 0.6172,
      "step": 4039
    },
    {
      "epoch": 1.9011764705882352,
      "grad_norm": 0.768064612513805,
      "learning_rate": 0.0002741236519902194,
      "loss": 0.8125,
      "step": 4040
    },
    {
      "epoch": 1.9016470588235292,
      "grad_norm": 0.465929056751395,
      "learning_rate": 0.0002740990335993421,
      "loss": 0.5078,
      "step": 4041
    },
    {
      "epoch": 1.9021176470588235,
      "grad_norm": 0.6321360002331355,
      "learning_rate": 0.00027407440460979854,
      "loss": 0.4512,
      "step": 4042
    },
    {
      "epoch": 1.9025882352941177,
      "grad_norm": 0.5511423552458173,
      "learning_rate": 0.0002740497650236923,
      "loss": 0.4844,
      "step": 4043
    },
    {
      "epoch": 1.903058823529412,
      "grad_norm": 0.9568924626693439,
      "learning_rate": 0.00027402511484312764,
      "loss": 0.9062,
      "step": 4044
    },
    {
      "epoch": 1.903529411764706,
      "grad_norm": 0.5440124310681103,
      "learning_rate": 0.00027400045407020986,
      "loss": 0.3223,
      "step": 4045
    },
    {
      "epoch": 1.904,
      "grad_norm": 0.6220628066483593,
      "learning_rate": 0.00027397578270704504,
      "loss": 0.4297,
      "step": 4046
    },
    {
      "epoch": 1.9044705882352941,
      "grad_norm": 0.9429635266213207,
      "learning_rate": 0.0002739511007557403,
      "loss": 0.8125,
      "step": 4047
    },
    {
      "epoch": 1.9049411764705884,
      "grad_norm": 0.5350720769035296,
      "learning_rate": 0.00027392640821840365,
      "loss": 0.4961,
      "step": 4048
    },
    {
      "epoch": 1.9054117647058824,
      "grad_norm": 0.6544240454294412,
      "learning_rate": 0.0002739017050971438,
      "loss": 0.3906,
      "step": 4049
    },
    {
      "epoch": 1.9058823529411764,
      "grad_norm": 0.45990820777531755,
      "learning_rate": 0.0002738769913940706,
      "loss": 0.2402,
      "step": 4050
    },
    {
      "epoch": 1.9063529411764706,
      "grad_norm": 0.5514909873046844,
      "learning_rate": 0.00027385226711129477,
      "loss": 0.3164,
      "step": 4051
    },
    {
      "epoch": 1.9068235294117648,
      "grad_norm": 0.3626742970685476,
      "learning_rate": 0.00027382753225092777,
      "loss": 0.2314,
      "step": 4052
    },
    {
      "epoch": 1.9072941176470588,
      "grad_norm": 0.33851315156813916,
      "learning_rate": 0.0002738027868150822,
      "loss": 0.209,
      "step": 4053
    },
    {
      "epoch": 1.9077647058823528,
      "grad_norm": 0.5710961098963037,
      "learning_rate": 0.0002737780308058714,
      "loss": 0.4883,
      "step": 4054
    },
    {
      "epoch": 1.908235294117647,
      "grad_norm": 0.7166649412092536,
      "learning_rate": 0.00027375326422540957,
      "loss": 0.5078,
      "step": 4055
    },
    {
      "epoch": 1.9087058823529413,
      "grad_norm": 0.8058191824478242,
      "learning_rate": 0.000273728487075812,
      "loss": 0.5078,
      "step": 4056
    },
    {
      "epoch": 1.9091764705882353,
      "grad_norm": 0.6032015607461299,
      "learning_rate": 0.0002737036993591948,
      "loss": 0.334,
      "step": 4057
    },
    {
      "epoch": 1.9096470588235293,
      "grad_norm": 0.4711191693360901,
      "learning_rate": 0.0002736789010776749,
      "loss": 0.5586,
      "step": 4058
    },
    {
      "epoch": 1.9101176470588235,
      "grad_norm": 0.4800040056124118,
      "learning_rate": 0.00027365409223337016,
      "loss": 0.3594,
      "step": 4059
    },
    {
      "epoch": 1.9105882352941177,
      "grad_norm": 0.5397508483848866,
      "learning_rate": 0.0002736292728283994,
      "loss": 0.3555,
      "step": 4060
    },
    {
      "epoch": 1.911058823529412,
      "grad_norm": 0.6799926676586412,
      "learning_rate": 0.0002736044428648824,
      "loss": 0.8008,
      "step": 4061
    },
    {
      "epoch": 1.911529411764706,
      "grad_norm": 0.43823939968871567,
      "learning_rate": 0.0002735796023449397,
      "loss": 0.377,
      "step": 4062
    },
    {
      "epoch": 1.912,
      "grad_norm": 0.5852739095381241,
      "learning_rate": 0.00027355475127069275,
      "loss": 0.7188,
      "step": 4063
    },
    {
      "epoch": 1.9124705882352941,
      "grad_norm": 0.6462511111681506,
      "learning_rate": 0.0002735298896442641,
      "loss": 0.5273,
      "step": 4064
    },
    {
      "epoch": 1.9129411764705884,
      "grad_norm": 0.5328408761448448,
      "learning_rate": 0.0002735050174677769,
      "loss": 0.4668,
      "step": 4065
    },
    {
      "epoch": 1.9134117647058824,
      "grad_norm": 1.1524082704184864,
      "learning_rate": 0.00027348013474335546,
      "loss": 0.5664,
      "step": 4066
    },
    {
      "epoch": 1.9138823529411764,
      "grad_norm": 0.3259094733620401,
      "learning_rate": 0.0002734552414731248,
      "loss": 0.1357,
      "step": 4067
    },
    {
      "epoch": 1.9143529411764706,
      "grad_norm": 0.5665566413173382,
      "learning_rate": 0.000273430337659211,
      "loss": 0.3809,
      "step": 4068
    },
    {
      "epoch": 1.9148235294117648,
      "grad_norm": 0.6063916888144619,
      "learning_rate": 0.0002734054233037409,
      "loss": 0.6836,
      "step": 4069
    },
    {
      "epoch": 1.9152941176470588,
      "grad_norm": 0.6393549915472975,
      "learning_rate": 0.00027338049840884245,
      "loss": 0.5156,
      "step": 4070
    },
    {
      "epoch": 1.9157647058823528,
      "grad_norm": 0.6203006859295337,
      "learning_rate": 0.00027335556297664417,
      "loss": 0.5078,
      "step": 4071
    },
    {
      "epoch": 1.916235294117647,
      "grad_norm": 0.8381027302589332,
      "learning_rate": 0.00027333061700927577,
      "loss": 1.0234,
      "step": 4072
    },
    {
      "epoch": 1.9167058823529413,
      "grad_norm": 0.5364048475502152,
      "learning_rate": 0.00027330566050886776,
      "loss": 0.416,
      "step": 4073
    },
    {
      "epoch": 1.9171764705882353,
      "grad_norm": 0.7040029538736489,
      "learning_rate": 0.0002732806934775515,
      "loss": 0.7656,
      "step": 4074
    },
    {
      "epoch": 1.9176470588235293,
      "grad_norm": 0.5214229737399447,
      "learning_rate": 0.0002732557159174594,
      "loss": 0.4883,
      "step": 4075
    },
    {
      "epoch": 1.9181176470588235,
      "grad_norm": 0.780242322739733,
      "learning_rate": 0.0002732307278307246,
      "loss": 0.7109,
      "step": 4076
    },
    {
      "epoch": 1.9185882352941177,
      "grad_norm": 0.27618432041662117,
      "learning_rate": 0.0002732057292194811,
      "loss": 0.1108,
      "step": 4077
    },
    {
      "epoch": 1.919058823529412,
      "grad_norm": 0.7910541292618956,
      "learning_rate": 0.000273180720085864,
      "loss": 1.0703,
      "step": 4078
    },
    {
      "epoch": 1.919529411764706,
      "grad_norm": 0.7656884226499127,
      "learning_rate": 0.0002731557004320093,
      "loss": 0.3828,
      "step": 4079
    },
    {
      "epoch": 1.92,
      "grad_norm": 0.7513048720847957,
      "learning_rate": 0.0002731306702600537,
      "loss": 0.6797,
      "step": 4080
    },
    {
      "epoch": 1.9204705882352942,
      "grad_norm": 0.5522071143782589,
      "learning_rate": 0.00027310562957213487,
      "loss": 0.6836,
      "step": 4081
    },
    {
      "epoch": 1.9209411764705884,
      "grad_norm": 0.724288264878706,
      "learning_rate": 0.00027308057837039144,
      "loss": 0.832,
      "step": 4082
    },
    {
      "epoch": 1.9214117647058824,
      "grad_norm": 0.6915129914867442,
      "learning_rate": 0.0002730555166569629,
      "loss": 0.5781,
      "step": 4083
    },
    {
      "epoch": 1.9218823529411764,
      "grad_norm": 0.37617518575779263,
      "learning_rate": 0.00027303044443398967,
      "loss": 0.4082,
      "step": 4084
    },
    {
      "epoch": 1.9223529411764706,
      "grad_norm": 0.5424816939722388,
      "learning_rate": 0.00027300536170361307,
      "loss": 0.4863,
      "step": 4085
    },
    {
      "epoch": 1.9228235294117648,
      "grad_norm": 0.6137446193159389,
      "learning_rate": 0.00027298026846797515,
      "loss": 0.6719,
      "step": 4086
    },
    {
      "epoch": 1.9232941176470588,
      "grad_norm": 0.46762685760119,
      "learning_rate": 0.0002729551647292191,
      "loss": 0.3145,
      "step": 4087
    },
    {
      "epoch": 1.9237647058823528,
      "grad_norm": 0.38105576284660225,
      "learning_rate": 0.000272930050489489,
      "loss": 0.2197,
      "step": 4088
    },
    {
      "epoch": 1.924235294117647,
      "grad_norm": 0.6192017516770177,
      "learning_rate": 0.0002729049257509295,
      "loss": 0.3555,
      "step": 4089
    },
    {
      "epoch": 1.9247058823529413,
      "grad_norm": 0.5372210163262572,
      "learning_rate": 0.0002728797905156866,
      "loss": 0.4473,
      "step": 4090
    },
    {
      "epoch": 1.9251764705882353,
      "grad_norm": 0.4942545888604023,
      "learning_rate": 0.0002728546447859068,
      "loss": 0.2949,
      "step": 4091
    },
    {
      "epoch": 1.9256470588235293,
      "grad_norm": 0.3688919151331801,
      "learning_rate": 0.00027282948856373774,
      "loss": 0.3359,
      "step": 4092
    },
    {
      "epoch": 1.9261176470588235,
      "grad_norm": 0.4544820036837746,
      "learning_rate": 0.0002728043218513279,
      "loss": 0.4473,
      "step": 4093
    },
    {
      "epoch": 1.9265882352941177,
      "grad_norm": 0.6507609919918274,
      "learning_rate": 0.0002727791446508266,
      "loss": 0.5508,
      "step": 4094
    },
    {
      "epoch": 1.9270588235294117,
      "grad_norm": 0.4756363268452348,
      "learning_rate": 0.00027275395696438413,
      "loss": 0.2598,
      "step": 4095
    },
    {
      "epoch": 1.927529411764706,
      "grad_norm": 0.4280014977475402,
      "learning_rate": 0.0002727287587941516,
      "loss": 0.4062,
      "step": 4096
    },
    {
      "epoch": 1.928,
      "grad_norm": 0.5048138468817189,
      "learning_rate": 0.00027270355014228113,
      "loss": 0.4844,
      "step": 4097
    },
    {
      "epoch": 1.9284705882352942,
      "grad_norm": 0.43261978835693954,
      "learning_rate": 0.00027267833101092564,
      "loss": 0.2832,
      "step": 4098
    },
    {
      "epoch": 1.9289411764705884,
      "grad_norm": 0.598857190039284,
      "learning_rate": 0.0002726531014022389,
      "loss": 0.457,
      "step": 4099
    },
    {
      "epoch": 1.9294117647058824,
      "grad_norm": 0.41848308429433173,
      "learning_rate": 0.00027262786131837573,
      "loss": 0.4258,
      "step": 4100
    },
    {
      "epoch": 1.9298823529411764,
      "grad_norm": 0.6846371746251669,
      "learning_rate": 0.00027260261076149166,
      "loss": 0.6719,
      "step": 4101
    },
    {
      "epoch": 1.9303529411764706,
      "grad_norm": 0.7582752811626499,
      "learning_rate": 0.00027257734973374335,
      "loss": 0.75,
      "step": 4102
    },
    {
      "epoch": 1.9308235294117648,
      "grad_norm": 0.7051527801163808,
      "learning_rate": 0.0002725520782372881,
      "loss": 0.7227,
      "step": 4103
    },
    {
      "epoch": 1.9312941176470588,
      "grad_norm": 0.6294481071419034,
      "learning_rate": 0.00027252679627428423,
      "loss": 0.7656,
      "step": 4104
    },
    {
      "epoch": 1.9317647058823528,
      "grad_norm": 0.5563586494475451,
      "learning_rate": 0.00027250150384689094,
      "loss": 0.2871,
      "step": 4105
    },
    {
      "epoch": 1.932235294117647,
      "grad_norm": 0.45752666525941127,
      "learning_rate": 0.0002724762009572684,
      "loss": 0.3242,
      "step": 4106
    },
    {
      "epoch": 1.9327058823529413,
      "grad_norm": 0.8121460792896862,
      "learning_rate": 0.0002724508876075776,
      "loss": 0.8711,
      "step": 4107
    },
    {
      "epoch": 1.9331764705882353,
      "grad_norm": 0.7297387922317325,
      "learning_rate": 0.0002724255637999803,
      "loss": 0.8125,
      "step": 4108
    },
    {
      "epoch": 1.9336470588235293,
      "grad_norm": 0.666960299976239,
      "learning_rate": 0.0002724002295366394,
      "loss": 0.6289,
      "step": 4109
    },
    {
      "epoch": 1.9341176470588235,
      "grad_norm": 0.45840335579862684,
      "learning_rate": 0.0002723748848197185,
      "loss": 0.2559,
      "step": 4110
    },
    {
      "epoch": 1.9345882352941177,
      "grad_norm": 0.5945082223610539,
      "learning_rate": 0.00027234952965138223,
      "loss": 0.5664,
      "step": 4111
    },
    {
      "epoch": 1.9350588235294117,
      "grad_norm": 0.7010415825776645,
      "learning_rate": 0.000272324164033796,
      "loss": 0.5859,
      "step": 4112
    },
    {
      "epoch": 1.9355294117647057,
      "grad_norm": 0.6712353425484646,
      "learning_rate": 0.0002722987879691261,
      "loss": 0.5312,
      "step": 4113
    },
    {
      "epoch": 1.936,
      "grad_norm": 0.6741349109961332,
      "learning_rate": 0.00027227340145953994,
      "loss": 0.5781,
      "step": 4114
    },
    {
      "epoch": 1.9364705882352942,
      "grad_norm": 0.6428567447506448,
      "learning_rate": 0.0002722480045072055,
      "loss": 0.5156,
      "step": 4115
    },
    {
      "epoch": 1.9369411764705884,
      "grad_norm": 0.7291534637892629,
      "learning_rate": 0.0002722225971142918,
      "loss": 0.75,
      "step": 4116
    },
    {
      "epoch": 1.9374117647058824,
      "grad_norm": 0.5268672265008164,
      "learning_rate": 0.00027219717928296885,
      "loss": 0.3926,
      "step": 4117
    },
    {
      "epoch": 1.9378823529411764,
      "grad_norm": 0.6287032006017744,
      "learning_rate": 0.0002721717510154074,
      "loss": 0.3828,
      "step": 4118
    },
    {
      "epoch": 1.9383529411764706,
      "grad_norm": 0.20641598153286828,
      "learning_rate": 0.0002721463123137792,
      "loss": 0.0476,
      "step": 4119
    },
    {
      "epoch": 1.9388235294117648,
      "grad_norm": 0.6474917945297719,
      "learning_rate": 0.0002721208631802568,
      "loss": 0.6484,
      "step": 4120
    },
    {
      "epoch": 1.9392941176470588,
      "grad_norm": 0.6714843003888679,
      "learning_rate": 0.00027209540361701365,
      "loss": 0.4023,
      "step": 4121
    },
    {
      "epoch": 1.9397647058823528,
      "grad_norm": 0.20116383030582774,
      "learning_rate": 0.0002720699336262242,
      "loss": 0.0874,
      "step": 4122
    },
    {
      "epoch": 1.940235294117647,
      "grad_norm": 0.34729588417885165,
      "learning_rate": 0.00027204445321006365,
      "loss": 0.1436,
      "step": 4123
    },
    {
      "epoch": 1.9407058823529413,
      "grad_norm": 0.5976436181588669,
      "learning_rate": 0.00027201896237070815,
      "loss": 0.3301,
      "step": 4124
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 0.6382639143281464,
      "learning_rate": 0.0002719934611103348,
      "loss": 0.5195,
      "step": 4125
    },
    {
      "epoch": 1.9416470588235293,
      "grad_norm": 0.09001135700997824,
      "learning_rate": 0.00027196794943112147,
      "loss": 0.032,
      "step": 4126
    },
    {
      "epoch": 1.9421176470588235,
      "grad_norm": 0.500034462008103,
      "learning_rate": 0.000271942427335247,
      "loss": 0.3379,
      "step": 4127
    },
    {
      "epoch": 1.9425882352941177,
      "grad_norm": 0.7279015076298838,
      "learning_rate": 0.00027191689482489114,
      "loss": 0.625,
      "step": 4128
    },
    {
      "epoch": 1.9430588235294117,
      "grad_norm": 1.0767131566664918,
      "learning_rate": 0.0002718913519022344,
      "loss": 0.7734,
      "step": 4129
    },
    {
      "epoch": 1.9435294117647057,
      "grad_norm": 0.8329775137406091,
      "learning_rate": 0.00027186579856945843,
      "loss": 0.5898,
      "step": 4130
    },
    {
      "epoch": 1.944,
      "grad_norm": 0.5496421308787154,
      "learning_rate": 0.00027184023482874545,
      "loss": 0.3379,
      "step": 4131
    },
    {
      "epoch": 1.9444705882352942,
      "grad_norm": 0.6644963740924845,
      "learning_rate": 0.00027181466068227885,
      "loss": 0.3008,
      "step": 4132
    },
    {
      "epoch": 1.9449411764705884,
      "grad_norm": 0.6163038011499697,
      "learning_rate": 0.00027178907613224264,
      "loss": 0.4316,
      "step": 4133
    },
    {
      "epoch": 1.9454117647058824,
      "grad_norm": 0.39777520914299586,
      "learning_rate": 0.00027176348118082203,
      "loss": 0.1318,
      "step": 4134
    },
    {
      "epoch": 1.9458823529411764,
      "grad_norm": 0.4893150947900296,
      "learning_rate": 0.0002717378758302029,
      "loss": 0.1465,
      "step": 4135
    },
    {
      "epoch": 1.9463529411764706,
      "grad_norm": 0.32627862805634494,
      "learning_rate": 0.000271712260082572,
      "loss": 0.1758,
      "step": 4136
    },
    {
      "epoch": 1.9468235294117648,
      "grad_norm": 0.5965657288705131,
      "learning_rate": 0.0002716866339401171,
      "loss": 0.4512,
      "step": 4137
    },
    {
      "epoch": 1.9472941176470588,
      "grad_norm": 0.6737212501471217,
      "learning_rate": 0.0002716609974050268,
      "loss": 0.6367,
      "step": 4138
    },
    {
      "epoch": 1.9477647058823528,
      "grad_norm": 0.74705475271687,
      "learning_rate": 0.0002716353504794906,
      "loss": 0.875,
      "step": 4139
    },
    {
      "epoch": 1.948235294117647,
      "grad_norm": 0.7072141163605922,
      "learning_rate": 0.0002716096931656988,
      "loss": 0.7148,
      "step": 4140
    },
    {
      "epoch": 1.9487058823529413,
      "grad_norm": 1.2315914426288488,
      "learning_rate": 0.0002715840254658427,
      "loss": 1.2188,
      "step": 4141
    },
    {
      "epoch": 1.9491764705882353,
      "grad_norm": 0.4415742321063551,
      "learning_rate": 0.0002715583473821145,
      "loss": 0.2295,
      "step": 4142
    },
    {
      "epoch": 1.9496470588235293,
      "grad_norm": 0.5340274728762203,
      "learning_rate": 0.0002715326589167072,
      "loss": 0.625,
      "step": 4143
    },
    {
      "epoch": 1.9501176470588235,
      "grad_norm": 0.7736400571108276,
      "learning_rate": 0.0002715069600718147,
      "loss": 0.793,
      "step": 4144
    },
    {
      "epoch": 1.9505882352941177,
      "grad_norm": 0.7566141850157977,
      "learning_rate": 0.0002714812508496319,
      "loss": 0.707,
      "step": 4145
    },
    {
      "epoch": 1.9510588235294117,
      "grad_norm": 0.5524242482909183,
      "learning_rate": 0.0002714555312523543,
      "loss": 0.4551,
      "step": 4146
    },
    {
      "epoch": 1.9515294117647057,
      "grad_norm": 0.6780377997978625,
      "learning_rate": 0.00027142980128217864,
      "loss": 0.7891,
      "step": 4147
    },
    {
      "epoch": 1.952,
      "grad_norm": 0.44843792795771764,
      "learning_rate": 0.00027140406094130234,
      "loss": 0.2158,
      "step": 4148
    },
    {
      "epoch": 1.9524705882352942,
      "grad_norm": 0.4662407985902808,
      "learning_rate": 0.0002713783102319237,
      "loss": 0.3594,
      "step": 4149
    },
    {
      "epoch": 1.9529411764705882,
      "grad_norm": 0.6191019315892731,
      "learning_rate": 0.0002713525491562421,
      "loss": 0.8359,
      "step": 4150
    },
    {
      "epoch": 1.9534117647058824,
      "grad_norm": 0.864675464923986,
      "learning_rate": 0.0002713267777164575,
      "loss": 0.7031,
      "step": 4151
    },
    {
      "epoch": 1.9538823529411764,
      "grad_norm": 0.5440434172559419,
      "learning_rate": 0.00027130099591477104,
      "loss": 0.4629,
      "step": 4152
    },
    {
      "epoch": 1.9543529411764706,
      "grad_norm": 1.0579697216911372,
      "learning_rate": 0.0002712752037533845,
      "loss": 1.1953,
      "step": 4153
    },
    {
      "epoch": 1.9548235294117648,
      "grad_norm": 0.5703385384863524,
      "learning_rate": 0.00027124940123450075,
      "loss": 0.5273,
      "step": 4154
    },
    {
      "epoch": 1.9552941176470588,
      "grad_norm": 0.6073614598919403,
      "learning_rate": 0.00027122358836032337,
      "loss": 0.6719,
      "step": 4155
    },
    {
      "epoch": 1.9557647058823528,
      "grad_norm": 0.5087345904874195,
      "learning_rate": 0.000271197765133057,
      "loss": 0.2275,
      "step": 4156
    },
    {
      "epoch": 1.956235294117647,
      "grad_norm": 0.5540081482501631,
      "learning_rate": 0.00027117193155490695,
      "loss": 0.7344,
      "step": 4157
    },
    {
      "epoch": 1.9567058823529413,
      "grad_norm": 0.5834596558048663,
      "learning_rate": 0.00027114608762807965,
      "loss": 0.5781,
      "step": 4158
    },
    {
      "epoch": 1.9571764705882353,
      "grad_norm": 0.5464397806555018,
      "learning_rate": 0.0002711202333547822,
      "loss": 0.4512,
      "step": 4159
    },
    {
      "epoch": 1.9576470588235293,
      "grad_norm": 0.6776470330712817,
      "learning_rate": 0.00027109436873722277,
      "loss": 0.8164,
      "step": 4160
    },
    {
      "epoch": 1.9581176470588235,
      "grad_norm": 0.5638436003640133,
      "learning_rate": 0.00027106849377761023,
      "loss": 0.4531,
      "step": 4161
    },
    {
      "epoch": 1.9585882352941177,
      "grad_norm": 0.473271950317997,
      "learning_rate": 0.00027104260847815457,
      "loss": 0.4785,
      "step": 4162
    },
    {
      "epoch": 1.9590588235294117,
      "grad_norm": 0.5570113673292221,
      "learning_rate": 0.0002710167128410664,
      "loss": 0.4434,
      "step": 4163
    },
    {
      "epoch": 1.9595294117647057,
      "grad_norm": 0.666701989554519,
      "learning_rate": 0.00027099080686855735,
      "loss": 0.4336,
      "step": 4164
    },
    {
      "epoch": 1.96,
      "grad_norm": 0.34224678092273775,
      "learning_rate": 0.00027096489056283995,
      "loss": 0.1504,
      "step": 4165
    },
    {
      "epoch": 1.9604705882352942,
      "grad_norm": 0.6189177330935469,
      "learning_rate": 0.00027093896392612753,
      "loss": 0.4004,
      "step": 4166
    },
    {
      "epoch": 1.9609411764705882,
      "grad_norm": 0.5439110347148092,
      "learning_rate": 0.0002709130269606345,
      "loss": 0.4043,
      "step": 4167
    },
    {
      "epoch": 1.9614117647058822,
      "grad_norm": 0.3634695705099473,
      "learning_rate": 0.00027088707966857576,
      "loss": 0.2471,
      "step": 4168
    },
    {
      "epoch": 1.9618823529411764,
      "grad_norm": 0.5717275906232221,
      "learning_rate": 0.0002708611220521675,
      "loss": 0.543,
      "step": 4169
    },
    {
      "epoch": 1.9623529411764706,
      "grad_norm": 0.729860138961098,
      "learning_rate": 0.0002708351541136267,
      "loss": 0.9688,
      "step": 4170
    },
    {
      "epoch": 1.9628235294117649,
      "grad_norm": 0.7147377248458574,
      "learning_rate": 0.0002708091758551709,
      "loss": 0.5938,
      "step": 4171
    },
    {
      "epoch": 1.9632941176470589,
      "grad_norm": 0.8705084688078278,
      "learning_rate": 0.000270783187279019,
      "loss": 1.1875,
      "step": 4172
    },
    {
      "epoch": 1.9637647058823529,
      "grad_norm": 0.5192027476460945,
      "learning_rate": 0.00027075718838739043,
      "loss": 0.6602,
      "step": 4173
    },
    {
      "epoch": 1.964235294117647,
      "grad_norm": 0.49726765194997147,
      "learning_rate": 0.0002707311791825057,
      "loss": 0.6289,
      "step": 4174
    },
    {
      "epoch": 1.9647058823529413,
      "grad_norm": 0.5926160403954986,
      "learning_rate": 0.00027070515966658604,
      "loss": 0.5,
      "step": 4175
    },
    {
      "epoch": 1.9651764705882353,
      "grad_norm": 0.46312010020674127,
      "learning_rate": 0.00027067912984185376,
      "loss": 0.5508,
      "step": 4176
    },
    {
      "epoch": 1.9656470588235293,
      "grad_norm": 0.5435311081179011,
      "learning_rate": 0.0002706530897105318,
      "loss": 0.5586,
      "step": 4177
    },
    {
      "epoch": 1.9661176470588235,
      "grad_norm": 0.6595733959708091,
      "learning_rate": 0.0002706270392748442,
      "loss": 0.9219,
      "step": 4178
    },
    {
      "epoch": 1.9665882352941177,
      "grad_norm": 0.4061080169693764,
      "learning_rate": 0.00027060097853701576,
      "loss": 0.2305,
      "step": 4179
    },
    {
      "epoch": 1.9670588235294117,
      "grad_norm": 0.6241831274270139,
      "learning_rate": 0.0002705749074992722,
      "loss": 0.4531,
      "step": 4180
    },
    {
      "epoch": 1.9675294117647057,
      "grad_norm": 0.6090864306558611,
      "learning_rate": 0.00027054882616384015,
      "loss": 0.4688,
      "step": 4181
    },
    {
      "epoch": 1.968,
      "grad_norm": 0.5462262324732557,
      "learning_rate": 0.0002705227345329471,
      "loss": 0.6758,
      "step": 4182
    },
    {
      "epoch": 1.9684705882352942,
      "grad_norm": 1.1025468469711601,
      "learning_rate": 0.00027049663260882124,
      "loss": 0.7188,
      "step": 4183
    },
    {
      "epoch": 1.9689411764705882,
      "grad_norm": 0.6689007782408438,
      "learning_rate": 0.00027047052039369204,
      "loss": 0.5586,
      "step": 4184
    },
    {
      "epoch": 1.9694117647058822,
      "grad_norm": 1.0224588980894367,
      "learning_rate": 0.0002704443978897894,
      "loss": 0.8086,
      "step": 4185
    },
    {
      "epoch": 1.9698823529411764,
      "grad_norm": 0.48775854716360423,
      "learning_rate": 0.0002704182650993445,
      "loss": 0.4297,
      "step": 4186
    },
    {
      "epoch": 1.9703529411764706,
      "grad_norm": 0.7877070523373413,
      "learning_rate": 0.00027039212202458905,
      "loss": 1.25,
      "step": 4187
    },
    {
      "epoch": 1.9708235294117649,
      "grad_norm": 0.3669622122447087,
      "learning_rate": 0.0002703659686677559,
      "loss": 0.3633,
      "step": 4188
    },
    {
      "epoch": 1.9712941176470589,
      "grad_norm": 0.6318962336780939,
      "learning_rate": 0.00027033980503107856,
      "loss": 0.6602,
      "step": 4189
    },
    {
      "epoch": 1.9717647058823529,
      "grad_norm": 0.6993623409525851,
      "learning_rate": 0.00027031363111679164,
      "loss": 0.4512,
      "step": 4190
    },
    {
      "epoch": 1.972235294117647,
      "grad_norm": 0.24095192867909282,
      "learning_rate": 0.0002702874469271305,
      "loss": 0.1895,
      "step": 4191
    },
    {
      "epoch": 1.9727058823529413,
      "grad_norm": 0.4906876700140123,
      "learning_rate": 0.00027026125246433144,
      "loss": 0.3652,
      "step": 4192
    },
    {
      "epoch": 1.9731764705882353,
      "grad_norm": 0.2108203503599911,
      "learning_rate": 0.0002702350477306315,
      "loss": 0.0869,
      "step": 4193
    },
    {
      "epoch": 1.9736470588235293,
      "grad_norm": 0.8808315782028759,
      "learning_rate": 0.0002702088327282687,
      "loss": 0.6641,
      "step": 4194
    },
    {
      "epoch": 1.9741176470588235,
      "grad_norm": 0.8123054144101733,
      "learning_rate": 0.00027018260745948196,
      "loss": 0.8047,
      "step": 4195
    },
    {
      "epoch": 1.9745882352941178,
      "grad_norm": 0.6820028257001692,
      "learning_rate": 0.0002701563719265111,
      "loss": 0.6211,
      "step": 4196
    },
    {
      "epoch": 1.9750588235294118,
      "grad_norm": 0.40799677287012154,
      "learning_rate": 0.0002701301261315967,
      "loss": 0.1865,
      "step": 4197
    },
    {
      "epoch": 1.9755294117647058,
      "grad_norm": 0.37158464194941454,
      "learning_rate": 0.0002701038700769803,
      "loss": 0.1455,
      "step": 4198
    },
    {
      "epoch": 1.976,
      "grad_norm": 0.5252733410604213,
      "learning_rate": 0.00027007760376490425,
      "loss": 0.3418,
      "step": 4199
    },
    {
      "epoch": 1.9764705882352942,
      "grad_norm": 1.3322398595234004,
      "learning_rate": 0.0002700513271976119,
      "loss": 0.8242,
      "step": 4200
    },
    {
      "epoch": 1.9769411764705882,
      "grad_norm": 0.687938172124599,
      "learning_rate": 0.0002700250403773473,
      "loss": 0.459,
      "step": 4201
    },
    {
      "epoch": 1.9774117647058822,
      "grad_norm": 0.905812727141934,
      "learning_rate": 0.0002699987433063556,
      "loss": 1.1094,
      "step": 4202
    },
    {
      "epoch": 1.9778823529411764,
      "grad_norm": 0.5317604361706185,
      "learning_rate": 0.0002699724359868826,
      "loss": 0.4395,
      "step": 4203
    },
    {
      "epoch": 1.9783529411764706,
      "grad_norm": 0.5766111705854692,
      "learning_rate": 0.0002699461184211751,
      "loss": 0.4355,
      "step": 4204
    },
    {
      "epoch": 1.9788235294117649,
      "grad_norm": 0.576442708691518,
      "learning_rate": 0.00026991979061148077,
      "loss": 0.5195,
      "step": 4205
    },
    {
      "epoch": 1.9792941176470589,
      "grad_norm": 1.0955070619229417,
      "learning_rate": 0.00026989345256004814,
      "loss": 0.9062,
      "step": 4206
    },
    {
      "epoch": 1.9797647058823529,
      "grad_norm": 0.6651399047309557,
      "learning_rate": 0.00026986710426912656,
      "loss": 0.8828,
      "step": 4207
    },
    {
      "epoch": 1.980235294117647,
      "grad_norm": 0.5959926265971957,
      "learning_rate": 0.0002698407457409663,
      "loss": 0.498,
      "step": 4208
    },
    {
      "epoch": 1.9807058823529413,
      "grad_norm": 0.39809938352030827,
      "learning_rate": 0.00026981437697781854,
      "loss": 0.3633,
      "step": 4209
    },
    {
      "epoch": 1.9811764705882353,
      "grad_norm": 0.8712961009748557,
      "learning_rate": 0.0002697879979819354,
      "loss": 0.8789,
      "step": 4210
    },
    {
      "epoch": 1.9816470588235293,
      "grad_norm": 0.5383634986874908,
      "learning_rate": 0.0002697616087555696,
      "loss": 0.6406,
      "step": 4211
    },
    {
      "epoch": 1.9821176470588235,
      "grad_norm": 0.6324364178079317,
      "learning_rate": 0.00026973520930097494,
      "loss": 0.3477,
      "step": 4212
    },
    {
      "epoch": 1.9825882352941178,
      "grad_norm": 0.7444935122086357,
      "learning_rate": 0.0002697087996204062,
      "loss": 0.3574,
      "step": 4213
    },
    {
      "epoch": 1.9830588235294118,
      "grad_norm": 0.8575481872962593,
      "learning_rate": 0.0002696823797161187,
      "loss": 0.7969,
      "step": 4214
    },
    {
      "epoch": 1.9835294117647058,
      "grad_norm": 0.9136902400545179,
      "learning_rate": 0.00026965594959036906,
      "loss": 0.9922,
      "step": 4215
    },
    {
      "epoch": 1.984,
      "grad_norm": 0.3919170749132662,
      "learning_rate": 0.00026962950924541435,
      "loss": 0.2256,
      "step": 4216
    },
    {
      "epoch": 1.9844705882352942,
      "grad_norm": 0.5314757341298305,
      "learning_rate": 0.0002696030586835128,
      "loss": 0.3691,
      "step": 4217
    },
    {
      "epoch": 1.9849411764705882,
      "grad_norm": 0.6622144035050918,
      "learning_rate": 0.00026957659790692335,
      "loss": 0.582,
      "step": 4218
    },
    {
      "epoch": 1.9854117647058822,
      "grad_norm": 0.6285166087943888,
      "learning_rate": 0.0002695501269179059,
      "loss": 0.2871,
      "step": 4219
    },
    {
      "epoch": 1.9858823529411764,
      "grad_norm": 0.5347316605458725,
      "learning_rate": 0.00026952364571872125,
      "loss": 0.5469,
      "step": 4220
    },
    {
      "epoch": 1.9863529411764707,
      "grad_norm": 0.4926298756802735,
      "learning_rate": 0.000269497154311631,
      "loss": 0.3438,
      "step": 4221
    },
    {
      "epoch": 1.9868235294117647,
      "grad_norm": 0.690251045654084,
      "learning_rate": 0.00026947065269889765,
      "loss": 0.7148,
      "step": 4222
    },
    {
      "epoch": 1.9872941176470589,
      "grad_norm": 0.32119518943776193,
      "learning_rate": 0.00026944414088278453,
      "loss": 0.1328,
      "step": 4223
    },
    {
      "epoch": 1.9877647058823529,
      "grad_norm": 0.9131156066701479,
      "learning_rate": 0.00026941761886555595,
      "loss": 0.9805,
      "step": 4224
    },
    {
      "epoch": 1.988235294117647,
      "grad_norm": 0.818519079236489,
      "learning_rate": 0.00026939108664947694,
      "loss": 0.8555,
      "step": 4225
    },
    {
      "epoch": 1.9887058823529413,
      "grad_norm": 0.6839593605610562,
      "learning_rate": 0.0002693645442368135,
      "loss": 0.6758,
      "step": 4226
    },
    {
      "epoch": 1.9891764705882353,
      "grad_norm": 0.34497982727064685,
      "learning_rate": 0.0002693379916298326,
      "loss": 0.167,
      "step": 4227
    },
    {
      "epoch": 1.9896470588235293,
      "grad_norm": 0.6460183321281365,
      "learning_rate": 0.00026931142883080176,
      "loss": 0.5508,
      "step": 4228
    },
    {
      "epoch": 1.9901176470588235,
      "grad_norm": 0.7204011177517077,
      "learning_rate": 0.00026928485584198973,
      "loss": 0.75,
      "step": 4229
    },
    {
      "epoch": 1.9905882352941178,
      "grad_norm": 0.5433199700219933,
      "learning_rate": 0.00026925827266566595,
      "loss": 0.3906,
      "step": 4230
    },
    {
      "epoch": 1.9910588235294118,
      "grad_norm": 0.3660516219789815,
      "learning_rate": 0.0002692316793041007,
      "loss": 0.4238,
      "step": 4231
    },
    {
      "epoch": 1.9915294117647058,
      "grad_norm": 0.5269477815425992,
      "learning_rate": 0.0002692050757595652,
      "loss": 0.4277,
      "step": 4232
    },
    {
      "epoch": 1.992,
      "grad_norm": 0.5143172066430923,
      "learning_rate": 0.00026917846203433155,
      "loss": 0.2891,
      "step": 4233
    },
    {
      "epoch": 1.9924705882352942,
      "grad_norm": 0.5918591770554029,
      "learning_rate": 0.00026915183813067264,
      "loss": 0.5078,
      "step": 4234
    },
    {
      "epoch": 1.9929411764705882,
      "grad_norm": 0.8935248475177143,
      "learning_rate": 0.0002691252040508623,
      "loss": 0.9414,
      "step": 4235
    },
    {
      "epoch": 1.9934117647058822,
      "grad_norm": 0.5843488046914562,
      "learning_rate": 0.0002690985597971753,
      "loss": 0.5703,
      "step": 4236
    },
    {
      "epoch": 1.9938823529411764,
      "grad_norm": 0.6049511593210308,
      "learning_rate": 0.00026907190537188707,
      "loss": 0.2441,
      "step": 4237
    },
    {
      "epoch": 1.9943529411764707,
      "grad_norm": 0.41300992232841804,
      "learning_rate": 0.0002690452407772741,
      "loss": 0.2129,
      "step": 4238
    },
    {
      "epoch": 1.9948235294117647,
      "grad_norm": 0.51255743765977,
      "learning_rate": 0.0002690185660156136,
      "loss": 0.3984,
      "step": 4239
    },
    {
      "epoch": 1.9952941176470587,
      "grad_norm": 0.32248067077643533,
      "learning_rate": 0.00026899188108918387,
      "loss": 0.2305,
      "step": 4240
    },
    {
      "epoch": 1.9957647058823529,
      "grad_norm": 0.5777232615665824,
      "learning_rate": 0.00026896518600026377,
      "loss": 0.5352,
      "step": 4241
    },
    {
      "epoch": 1.996235294117647,
      "grad_norm": 0.8125209506092438,
      "learning_rate": 0.00026893848075113327,
      "loss": 0.6484,
      "step": 4242
    },
    {
      "epoch": 1.9967058823529413,
      "grad_norm": 0.4179002335232279,
      "learning_rate": 0.00026891176534407315,
      "loss": 0.1797,
      "step": 4243
    },
    {
      "epoch": 1.9971764705882353,
      "grad_norm": 0.6895793588539851,
      "learning_rate": 0.00026888503978136506,
      "loss": 0.4375,
      "step": 4244
    },
    {
      "epoch": 1.9976470588235293,
      "grad_norm": 0.5730980886773281,
      "learning_rate": 0.00026885830406529136,
      "loss": 0.3262,
      "step": 4245
    },
    {
      "epoch": 1.9981176470588236,
      "grad_norm": 0.7659673565713988,
      "learning_rate": 0.0002688315581981355,
      "loss": 0.5234,
      "step": 4246
    },
    {
      "epoch": 1.9985882352941178,
      "grad_norm": 0.8587689035462222,
      "learning_rate": 0.0002688048021821817,
      "loss": 1.0312,
      "step": 4247
    },
    {
      "epoch": 1.9990588235294118,
      "grad_norm": 0.4512006678760764,
      "learning_rate": 0.00026877803601971516,
      "loss": 0.3145,
      "step": 4248
    },
    {
      "epoch": 1.9995294117647058,
      "grad_norm": 0.4581821693217795,
      "learning_rate": 0.00026875125971302167,
      "loss": 0.3418,
      "step": 4249
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.41160759796004903,
      "learning_rate": 0.0002687244732643881,
      "loss": 0.2754,
      "step": 4250
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.4967285096645355,
      "eval_runtime": 44.163,
      "eval_samples_per_second": 16.983,
      "eval_steps_per_second": 8.491,
      "step": 4250
    },
    {
      "epoch": 2.0004705882352942,
      "grad_norm": 0.5486064565378,
      "learning_rate": 0.0002686976766761022,
      "loss": 0.4355,
      "step": 4251
    },
    {
      "epoch": 2.0009411764705884,
      "grad_norm": 0.9648942957184907,
      "learning_rate": 0.0002686708699504525,
      "loss": 0.8516,
      "step": 4252
    },
    {
      "epoch": 2.0014117647058822,
      "grad_norm": 0.6242696127343348,
      "learning_rate": 0.0002686440530897284,
      "loss": 0.3438,
      "step": 4253
    },
    {
      "epoch": 2.0018823529411764,
      "grad_norm": 0.6858524424130792,
      "learning_rate": 0.00026861722609622027,
      "loss": 0.8828,
      "step": 4254
    },
    {
      "epoch": 2.0023529411764707,
      "grad_norm": 0.6681424860029618,
      "learning_rate": 0.0002685903889722192,
      "loss": 0.5312,
      "step": 4255
    },
    {
      "epoch": 2.002823529411765,
      "grad_norm": 0.34685587764038356,
      "learning_rate": 0.00026856354172001724,
      "loss": 0.2676,
      "step": 4256
    },
    {
      "epoch": 2.0032941176470587,
      "grad_norm": 0.687299589281167,
      "learning_rate": 0.0002685366843419073,
      "loss": 0.3789,
      "step": 4257
    },
    {
      "epoch": 2.003764705882353,
      "grad_norm": 0.7085749828166507,
      "learning_rate": 0.000268509816840183,
      "loss": 0.582,
      "step": 4258
    },
    {
      "epoch": 2.004235294117647,
      "grad_norm": 0.7677905255050071,
      "learning_rate": 0.00026848293921713913,
      "loss": 0.4902,
      "step": 4259
    },
    {
      "epoch": 2.0047058823529413,
      "grad_norm": 0.5783289669626686,
      "learning_rate": 0.0002684560514750711,
      "loss": 0.6289,
      "step": 4260
    },
    {
      "epoch": 2.005176470588235,
      "grad_norm": 0.6093926608298043,
      "learning_rate": 0.00026842915361627523,
      "loss": 0.3125,
      "step": 4261
    },
    {
      "epoch": 2.0056470588235293,
      "grad_norm": 0.5281194122025569,
      "learning_rate": 0.0002684022456430487,
      "loss": 0.4727,
      "step": 4262
    },
    {
      "epoch": 2.0061176470588236,
      "grad_norm": 0.556003096737819,
      "learning_rate": 0.0002683753275576897,
      "loss": 0.3184,
      "step": 4263
    },
    {
      "epoch": 2.006588235294118,
      "grad_norm": 0.6913862399093079,
      "learning_rate": 0.0002683483993624971,
      "loss": 0.8711,
      "step": 4264
    },
    {
      "epoch": 2.0070588235294116,
      "grad_norm": 0.40156945039504677,
      "learning_rate": 0.0002683214610597706,
      "loss": 0.1572,
      "step": 4265
    },
    {
      "epoch": 2.007529411764706,
      "grad_norm": 0.049409282969105975,
      "learning_rate": 0.0002682945126518111,
      "loss": 0.0199,
      "step": 4266
    },
    {
      "epoch": 2.008,
      "grad_norm": 0.7241761886113016,
      "learning_rate": 0.00026826755414091993,
      "loss": 0.6797,
      "step": 4267
    },
    {
      "epoch": 2.0084705882352942,
      "grad_norm": 0.9543886988675269,
      "learning_rate": 0.00026824058552939954,
      "loss": 0.5898,
      "step": 4268
    },
    {
      "epoch": 2.0089411764705885,
      "grad_norm": 0.41653925657221036,
      "learning_rate": 0.00026821360681955316,
      "loss": 0.1807,
      "step": 4269
    },
    {
      "epoch": 2.0094117647058822,
      "grad_norm": 0.18553434722444365,
      "learning_rate": 0.00026818661801368494,
      "loss": 0.0947,
      "step": 4270
    },
    {
      "epoch": 2.0098823529411765,
      "grad_norm": 0.49669342587509824,
      "learning_rate": 0.0002681596191140998,
      "loss": 0.248,
      "step": 4271
    },
    {
      "epoch": 2.0103529411764707,
      "grad_norm": 0.7596773356419102,
      "learning_rate": 0.00026813261012310365,
      "loss": 0.7812,
      "step": 4272
    },
    {
      "epoch": 2.010823529411765,
      "grad_norm": 0.3942067229319076,
      "learning_rate": 0.00026810559104300315,
      "loss": 0.3125,
      "step": 4273
    },
    {
      "epoch": 2.0112941176470587,
      "grad_norm": 0.4545484440595272,
      "learning_rate": 0.0002680785618761059,
      "loss": 0.3457,
      "step": 4274
    },
    {
      "epoch": 2.011764705882353,
      "grad_norm": 0.6070414805373118,
      "learning_rate": 0.00026805152262472025,
      "loss": 0.6523,
      "step": 4275
    },
    {
      "epoch": 2.012235294117647,
      "grad_norm": 0.6641458894588185,
      "learning_rate": 0.00026802447329115557,
      "loss": 0.9336,
      "step": 4276
    },
    {
      "epoch": 2.0127058823529413,
      "grad_norm": 0.1600558247708568,
      "learning_rate": 0.00026799741387772193,
      "loss": 0.0491,
      "step": 4277
    },
    {
      "epoch": 2.013176470588235,
      "grad_norm": 0.5513059978511294,
      "learning_rate": 0.0002679703443867304,
      "loss": 0.3672,
      "step": 4278
    },
    {
      "epoch": 2.0136470588235293,
      "grad_norm": 0.4965551899025064,
      "learning_rate": 0.0002679432648204928,
      "loss": 0.2539,
      "step": 4279
    },
    {
      "epoch": 2.0141176470588236,
      "grad_norm": 0.8884087126925498,
      "learning_rate": 0.00026791617518132186,
      "loss": 0.7578,
      "step": 4280
    },
    {
      "epoch": 2.014588235294118,
      "grad_norm": 0.5566555816880808,
      "learning_rate": 0.00026788907547153126,
      "loss": 0.5703,
      "step": 4281
    },
    {
      "epoch": 2.0150588235294116,
      "grad_norm": 0.674063889112442,
      "learning_rate": 0.00026786196569343525,
      "loss": 0.4707,
      "step": 4282
    },
    {
      "epoch": 2.015529411764706,
      "grad_norm": 0.42902738176124405,
      "learning_rate": 0.0002678348458493494,
      "loss": 0.5703,
      "step": 4283
    },
    {
      "epoch": 2.016,
      "grad_norm": 0.6207258826700355,
      "learning_rate": 0.00026780771594158966,
      "loss": 0.5664,
      "step": 4284
    },
    {
      "epoch": 2.0164705882352942,
      "grad_norm": 0.6254302525531609,
      "learning_rate": 0.0002677805759724731,
      "loss": 0.8359,
      "step": 4285
    },
    {
      "epoch": 2.016941176470588,
      "grad_norm": 0.6939507494119996,
      "learning_rate": 0.0002677534259443177,
      "loss": 0.6055,
      "step": 4286
    },
    {
      "epoch": 2.0174117647058822,
      "grad_norm": 0.439779977868431,
      "learning_rate": 0.00026772626585944213,
      "loss": 0.4395,
      "step": 4287
    },
    {
      "epoch": 2.0178823529411765,
      "grad_norm": 0.7709524186588053,
      "learning_rate": 0.00026769909572016605,
      "loss": 0.9336,
      "step": 4288
    },
    {
      "epoch": 2.0183529411764707,
      "grad_norm": 0.5737641298597498,
      "learning_rate": 0.0002676719155288098,
      "loss": 0.6797,
      "step": 4289
    },
    {
      "epoch": 2.018823529411765,
      "grad_norm": 0.4502919357655913,
      "learning_rate": 0.00026764472528769485,
      "loss": 0.5352,
      "step": 4290
    },
    {
      "epoch": 2.0192941176470587,
      "grad_norm": 0.4020799825790959,
      "learning_rate": 0.0002676175249991433,
      "loss": 0.2217,
      "step": 4291
    },
    {
      "epoch": 2.019764705882353,
      "grad_norm": 0.35632225269554346,
      "learning_rate": 0.0002675903146654782,
      "loss": 0.2002,
      "step": 4292
    },
    {
      "epoch": 2.020235294117647,
      "grad_norm": 0.6441818079664476,
      "learning_rate": 0.00026756309428902343,
      "loss": 0.75,
      "step": 4293
    },
    {
      "epoch": 2.0207058823529414,
      "grad_norm": 0.49559301578788106,
      "learning_rate": 0.0002675358638721038,
      "loss": 0.4238,
      "step": 4294
    },
    {
      "epoch": 2.021176470588235,
      "grad_norm": 0.664730306901123,
      "learning_rate": 0.0002675086234170449,
      "loss": 0.6172,
      "step": 4295
    },
    {
      "epoch": 2.0216470588235294,
      "grad_norm": 0.5907859251658529,
      "learning_rate": 0.0002674813729261731,
      "loss": 0.5664,
      "step": 4296
    },
    {
      "epoch": 2.0221176470588236,
      "grad_norm": 0.5623979694201932,
      "learning_rate": 0.0002674541124018159,
      "loss": 0.6133,
      "step": 4297
    },
    {
      "epoch": 2.022588235294118,
      "grad_norm": 0.5535323334952645,
      "learning_rate": 0.00026742684184630135,
      "loss": 0.6133,
      "step": 4298
    },
    {
      "epoch": 2.0230588235294116,
      "grad_norm": 0.6186267592766225,
      "learning_rate": 0.00026739956126195855,
      "loss": 0.4824,
      "step": 4299
    },
    {
      "epoch": 2.023529411764706,
      "grad_norm": 0.7343806824318303,
      "learning_rate": 0.0002673722706511174,
      "loss": 0.5664,
      "step": 4300
    },
    {
      "epoch": 2.024,
      "grad_norm": 0.44956106240296106,
      "learning_rate": 0.00026734497001610854,
      "loss": 0.3242,
      "step": 4301
    },
    {
      "epoch": 2.0244705882352942,
      "grad_norm": 0.5877285531132485,
      "learning_rate": 0.00026731765935926374,
      "loss": 0.3848,
      "step": 4302
    },
    {
      "epoch": 2.024941176470588,
      "grad_norm": 0.5340658241904955,
      "learning_rate": 0.00026729033868291537,
      "loss": 0.5273,
      "step": 4303
    },
    {
      "epoch": 2.0254117647058822,
      "grad_norm": 0.16804581358610998,
      "learning_rate": 0.00026726300798939675,
      "loss": 0.0615,
      "step": 4304
    },
    {
      "epoch": 2.0258823529411765,
      "grad_norm": 0.26868341312078275,
      "learning_rate": 0.0002672356672810421,
      "loss": 0.1074,
      "step": 4305
    },
    {
      "epoch": 2.0263529411764707,
      "grad_norm": 0.7752232008022466,
      "learning_rate": 0.00026720831656018644,
      "loss": 1.0391,
      "step": 4306
    },
    {
      "epoch": 2.026823529411765,
      "grad_norm": 0.48297186321409813,
      "learning_rate": 0.0002671809558291656,
      "loss": 0.2393,
      "step": 4307
    },
    {
      "epoch": 2.0272941176470587,
      "grad_norm": 0.5394670378042928,
      "learning_rate": 0.00026715358509031643,
      "loss": 0.5078,
      "step": 4308
    },
    {
      "epoch": 2.027764705882353,
      "grad_norm": 0.6601588100595589,
      "learning_rate": 0.00026712620434597636,
      "loss": 0.6211,
      "step": 4309
    },
    {
      "epoch": 2.028235294117647,
      "grad_norm": 0.765434942170978,
      "learning_rate": 0.00026709881359848406,
      "loss": 0.416,
      "step": 4310
    },
    {
      "epoch": 2.0287058823529414,
      "grad_norm": 0.3878987134105604,
      "learning_rate": 0.0002670714128501786,
      "loss": 0.2471,
      "step": 4311
    },
    {
      "epoch": 2.029176470588235,
      "grad_norm": 0.6346975917100255,
      "learning_rate": 0.0002670440021034003,
      "loss": 0.5547,
      "step": 4312
    },
    {
      "epoch": 2.0296470588235294,
      "grad_norm": 0.49314689706217646,
      "learning_rate": 0.00026701658136049015,
      "loss": 0.4941,
      "step": 4313
    },
    {
      "epoch": 2.0301176470588236,
      "grad_norm": 0.6790547876463046,
      "learning_rate": 0.00026698915062378994,
      "loss": 0.5039,
      "step": 4314
    },
    {
      "epoch": 2.030588235294118,
      "grad_norm": 0.6474290061394405,
      "learning_rate": 0.00026696170989564245,
      "loss": 0.5859,
      "step": 4315
    },
    {
      "epoch": 2.0310588235294116,
      "grad_norm": 0.48762146205241996,
      "learning_rate": 0.0002669342591783913,
      "loss": 0.5391,
      "step": 4316
    },
    {
      "epoch": 2.031529411764706,
      "grad_norm": 0.5517311762052681,
      "learning_rate": 0.00026690679847438077,
      "loss": 0.5234,
      "step": 4317
    },
    {
      "epoch": 2.032,
      "grad_norm": 0.5014599574203525,
      "learning_rate": 0.0002668793277859563,
      "loss": 0.5312,
      "step": 4318
    },
    {
      "epoch": 2.0324705882352943,
      "grad_norm": 0.44552409728018083,
      "learning_rate": 0.0002668518471154639,
      "loss": 0.5273,
      "step": 4319
    },
    {
      "epoch": 2.032941176470588,
      "grad_norm": 0.6548060130352544,
      "learning_rate": 0.0002668243564652506,
      "loss": 0.7305,
      "step": 4320
    },
    {
      "epoch": 2.0334117647058823,
      "grad_norm": 0.6445140948616688,
      "learning_rate": 0.0002667968558376642,
      "loss": 0.5312,
      "step": 4321
    },
    {
      "epoch": 2.0338823529411765,
      "grad_norm": 0.6961719496101866,
      "learning_rate": 0.0002667693452350535,
      "loss": 0.5703,
      "step": 4322
    },
    {
      "epoch": 2.0343529411764707,
      "grad_norm": 0.5649580122189826,
      "learning_rate": 0.0002667418246597679,
      "loss": 0.2988,
      "step": 4323
    },
    {
      "epoch": 2.034823529411765,
      "grad_norm": 0.391190966397928,
      "learning_rate": 0.0002667142941141579,
      "loss": 0.1201,
      "step": 4324
    },
    {
      "epoch": 2.0352941176470587,
      "grad_norm": 0.5567073759379981,
      "learning_rate": 0.0002666867536005747,
      "loss": 0.1699,
      "step": 4325
    },
    {
      "epoch": 2.035764705882353,
      "grad_norm": 1.1831806248715768,
      "learning_rate": 0.00026665920312137045,
      "loss": 0.8945,
      "step": 4326
    },
    {
      "epoch": 2.036235294117647,
      "grad_norm": 0.5543668629620537,
      "learning_rate": 0.000266631642678898,
      "loss": 0.3418,
      "step": 4327
    },
    {
      "epoch": 2.0367058823529414,
      "grad_norm": 0.5341414796982584,
      "learning_rate": 0.00026660407227551113,
      "loss": 0.5195,
      "step": 4328
    },
    {
      "epoch": 2.037176470588235,
      "grad_norm": 0.7514522204504379,
      "learning_rate": 0.0002665764919135646,
      "loss": 0.582,
      "step": 4329
    },
    {
      "epoch": 2.0376470588235294,
      "grad_norm": 0.9341896102077042,
      "learning_rate": 0.00026654890159541386,
      "loss": 0.7266,
      "step": 4330
    },
    {
      "epoch": 2.0381176470588236,
      "grad_norm": 0.46947773112308927,
      "learning_rate": 0.0002665213013234153,
      "loss": 0.3145,
      "step": 4331
    },
    {
      "epoch": 2.038588235294118,
      "grad_norm": 0.28574186650522126,
      "learning_rate": 0.00026649369109992596,
      "loss": 0.1396,
      "step": 4332
    },
    {
      "epoch": 2.0390588235294116,
      "grad_norm": 0.8509507554863581,
      "learning_rate": 0.0002664660709273041,
      "loss": 1.0547,
      "step": 4333
    },
    {
      "epoch": 2.039529411764706,
      "grad_norm": 0.6730540091676184,
      "learning_rate": 0.0002664384408079085,
      "loss": 0.4668,
      "step": 4334
    },
    {
      "epoch": 2.04,
      "grad_norm": 0.38415739217202555,
      "learning_rate": 0.0002664108007440989,
      "loss": 0.2441,
      "step": 4335
    },
    {
      "epoch": 2.0404705882352943,
      "grad_norm": 0.3723348006289008,
      "learning_rate": 0.00026638315073823594,
      "loss": 0.2354,
      "step": 4336
    },
    {
      "epoch": 2.040941176470588,
      "grad_norm": 0.8221667234877678,
      "learning_rate": 0.00026635549079268104,
      "loss": 0.8711,
      "step": 4337
    },
    {
      "epoch": 2.0414117647058823,
      "grad_norm": 0.33556014755125685,
      "learning_rate": 0.00026632782090979653,
      "loss": 0.1855,
      "step": 4338
    },
    {
      "epoch": 2.0418823529411765,
      "grad_norm": 0.506769301125026,
      "learning_rate": 0.0002663001410919455,
      "loss": 0.332,
      "step": 4339
    },
    {
      "epoch": 2.0423529411764707,
      "grad_norm": 0.5778800062364113,
      "learning_rate": 0.000266272451341492,
      "loss": 0.3359,
      "step": 4340
    },
    {
      "epoch": 2.0428235294117645,
      "grad_norm": 0.5339715897461264,
      "learning_rate": 0.0002662447516608008,
      "loss": 0.5273,
      "step": 4341
    },
    {
      "epoch": 2.0432941176470587,
      "grad_norm": 0.9416320493493575,
      "learning_rate": 0.0002662170420522377,
      "loss": 0.8125,
      "step": 4342
    },
    {
      "epoch": 2.043764705882353,
      "grad_norm": 0.6099524659743776,
      "learning_rate": 0.0002661893225181691,
      "loss": 0.5703,
      "step": 4343
    },
    {
      "epoch": 2.044235294117647,
      "grad_norm": 0.6722322603407672,
      "learning_rate": 0.0002661615930609625,
      "loss": 0.4062,
      "step": 4344
    },
    {
      "epoch": 2.0447058823529414,
      "grad_norm": 0.2788715719380632,
      "learning_rate": 0.000266133853682986,
      "loss": 0.1572,
      "step": 4345
    },
    {
      "epoch": 2.045176470588235,
      "grad_norm": 0.28559309895137636,
      "learning_rate": 0.0002661061043866089,
      "loss": 0.1768,
      "step": 4346
    },
    {
      "epoch": 2.0456470588235294,
      "grad_norm": 0.512995735688563,
      "learning_rate": 0.00026607834517420085,
      "loss": 0.291,
      "step": 4347
    },
    {
      "epoch": 2.0461176470588236,
      "grad_norm": 0.37220852057394355,
      "learning_rate": 0.00026605057604813285,
      "loss": 0.2344,
      "step": 4348
    },
    {
      "epoch": 2.046588235294118,
      "grad_norm": 0.36372681606162305,
      "learning_rate": 0.00026602279701077637,
      "loss": 0.1396,
      "step": 4349
    },
    {
      "epoch": 2.0470588235294116,
      "grad_norm": 0.5948784504171364,
      "learning_rate": 0.000265995008064504,
      "loss": 0.4883,
      "step": 4350
    },
    {
      "epoch": 2.047529411764706,
      "grad_norm": 0.8377941087684021,
      "learning_rate": 0.00026596720921168897,
      "loss": 0.9766,
      "step": 4351
    },
    {
      "epoch": 2.048,
      "grad_norm": 0.6391776481348349,
      "learning_rate": 0.00026593940045470545,
      "loss": 0.6797,
      "step": 4352
    },
    {
      "epoch": 2.0484705882352943,
      "grad_norm": 0.4669023312575711,
      "learning_rate": 0.00026591158179592846,
      "loss": 0.3496,
      "step": 4353
    },
    {
      "epoch": 2.048941176470588,
      "grad_norm": 0.6730919542640031,
      "learning_rate": 0.0002658837532377338,
      "loss": 0.7695,
      "step": 4354
    },
    {
      "epoch": 2.0494117647058823,
      "grad_norm": 0.4374884089916385,
      "learning_rate": 0.00026585591478249824,
      "loss": 0.2656,
      "step": 4355
    },
    {
      "epoch": 2.0498823529411765,
      "grad_norm": 0.5934651782742754,
      "learning_rate": 0.0002658280664325993,
      "loss": 0.8398,
      "step": 4356
    },
    {
      "epoch": 2.0503529411764707,
      "grad_norm": 0.6182235155959157,
      "learning_rate": 0.00026580020819041533,
      "loss": 0.4258,
      "step": 4357
    },
    {
      "epoch": 2.0508235294117645,
      "grad_norm": 0.5806190577939929,
      "learning_rate": 0.0002657723400583256,
      "loss": 0.3086,
      "step": 4358
    },
    {
      "epoch": 2.0512941176470587,
      "grad_norm": 0.5919717784452762,
      "learning_rate": 0.00026574446203871013,
      "loss": 0.4316,
      "step": 4359
    },
    {
      "epoch": 2.051764705882353,
      "grad_norm": 0.5351704853300485,
      "learning_rate": 0.0002657165741339499,
      "loss": 0.459,
      "step": 4360
    },
    {
      "epoch": 2.052235294117647,
      "grad_norm": 0.5196256020787429,
      "learning_rate": 0.00026568867634642665,
      "loss": 0.5781,
      "step": 4361
    },
    {
      "epoch": 2.0527058823529414,
      "grad_norm": 0.601702722358359,
      "learning_rate": 0.00026566076867852304,
      "loss": 0.7969,
      "step": 4362
    },
    {
      "epoch": 2.053176470588235,
      "grad_norm": 0.6631658769039718,
      "learning_rate": 0.0002656328511326224,
      "loss": 0.6445,
      "step": 4363
    },
    {
      "epoch": 2.0536470588235294,
      "grad_norm": 0.419923803873229,
      "learning_rate": 0.00026560492371110913,
      "loss": 0.1904,
      "step": 4364
    },
    {
      "epoch": 2.0541176470588236,
      "grad_norm": 0.35448626357076196,
      "learning_rate": 0.00026557698641636835,
      "loss": 0.2109,
      "step": 4365
    },
    {
      "epoch": 2.054588235294118,
      "grad_norm": 0.6501701895736027,
      "learning_rate": 0.000265549039250786,
      "loss": 0.5,
      "step": 4366
    },
    {
      "epoch": 2.0550588235294116,
      "grad_norm": 0.28055681387535003,
      "learning_rate": 0.000265521082216749,
      "loss": 0.1826,
      "step": 4367
    },
    {
      "epoch": 2.055529411764706,
      "grad_norm": 0.35051996082498343,
      "learning_rate": 0.0002654931153166449,
      "loss": 0.2178,
      "step": 4368
    },
    {
      "epoch": 2.056,
      "grad_norm": 0.6494678971951912,
      "learning_rate": 0.0002654651385528622,
      "loss": 0.7305,
      "step": 4369
    },
    {
      "epoch": 2.0564705882352943,
      "grad_norm": 0.8481598299944222,
      "learning_rate": 0.00026543715192779043,
      "loss": 0.793,
      "step": 4370
    },
    {
      "epoch": 2.056941176470588,
      "grad_norm": 0.5964385212057283,
      "learning_rate": 0.0002654091554438196,
      "loss": 0.5273,
      "step": 4371
    },
    {
      "epoch": 2.0574117647058823,
      "grad_norm": 1.800157715080433,
      "learning_rate": 0.0002653811491033408,
      "loss": 1.4688,
      "step": 4372
    },
    {
      "epoch": 2.0578823529411765,
      "grad_norm": 0.8564289835716495,
      "learning_rate": 0.00026535313290874593,
      "loss": 0.6719,
      "step": 4373
    },
    {
      "epoch": 2.0583529411764707,
      "grad_norm": 0.7246915077358448,
      "learning_rate": 0.0002653251068624278,
      "loss": 0.5117,
      "step": 4374
    },
    {
      "epoch": 2.0588235294117645,
      "grad_norm": 0.5817599395492827,
      "learning_rate": 0.00026529707096677977,
      "loss": 0.3945,
      "step": 4375
    },
    {
      "epoch": 2.0592941176470587,
      "grad_norm": 0.25369372066618245,
      "learning_rate": 0.00026526902522419637,
      "loss": 0.168,
      "step": 4376
    },
    {
      "epoch": 2.059764705882353,
      "grad_norm": 0.4935823151105354,
      "learning_rate": 0.0002652409696370729,
      "loss": 0.418,
      "step": 4377
    },
    {
      "epoch": 2.060235294117647,
      "grad_norm": 0.4304991080455059,
      "learning_rate": 0.0002652129042078053,
      "loss": 0.2598,
      "step": 4378
    },
    {
      "epoch": 2.0607058823529414,
      "grad_norm": 0.6507933005348169,
      "learning_rate": 0.00026518482893879056,
      "loss": 0.5898,
      "step": 4379
    },
    {
      "epoch": 2.061176470588235,
      "grad_norm": 0.587156815834741,
      "learning_rate": 0.0002651567438324265,
      "loss": 0.5938,
      "step": 4380
    },
    {
      "epoch": 2.0616470588235294,
      "grad_norm": 0.5202118303896175,
      "learning_rate": 0.00026512864889111163,
      "loss": 0.2559,
      "step": 4381
    },
    {
      "epoch": 2.0621176470588236,
      "grad_norm": 0.865645696479338,
      "learning_rate": 0.00026510054411724545,
      "loss": 0.6211,
      "step": 4382
    },
    {
      "epoch": 2.062588235294118,
      "grad_norm": 0.48106640463595596,
      "learning_rate": 0.00026507242951322825,
      "loss": 0.4434,
      "step": 4383
    },
    {
      "epoch": 2.0630588235294116,
      "grad_norm": 0.4793679224945372,
      "learning_rate": 0.00026504430508146116,
      "loss": 0.3809,
      "step": 4384
    },
    {
      "epoch": 2.063529411764706,
      "grad_norm": 0.6663815321350542,
      "learning_rate": 0.00026501617082434607,
      "loss": 0.6094,
      "step": 4385
    },
    {
      "epoch": 2.064,
      "grad_norm": 0.4994883090662756,
      "learning_rate": 0.0002649880267442859,
      "loss": 0.3184,
      "step": 4386
    },
    {
      "epoch": 2.0644705882352943,
      "grad_norm": 0.7954610006892138,
      "learning_rate": 0.00026495987284368415,
      "loss": 0.6445,
      "step": 4387
    },
    {
      "epoch": 2.064941176470588,
      "grad_norm": 0.7220130033554153,
      "learning_rate": 0.0002649317091249455,
      "loss": 0.625,
      "step": 4388
    },
    {
      "epoch": 2.0654117647058823,
      "grad_norm": 0.38744969872903795,
      "learning_rate": 0.00026490353559047505,
      "loss": 0.1157,
      "step": 4389
    },
    {
      "epoch": 2.0658823529411765,
      "grad_norm": 0.47972439472382583,
      "learning_rate": 0.0002648753522426791,
      "loss": 0.2373,
      "step": 4390
    },
    {
      "epoch": 2.0663529411764707,
      "grad_norm": 0.7367014295429342,
      "learning_rate": 0.0002648471590839646,
      "loss": 0.8672,
      "step": 4391
    },
    {
      "epoch": 2.0668235294117645,
      "grad_norm": 0.5250278452571803,
      "learning_rate": 0.00026481895611673944,
      "loss": 0.3848,
      "step": 4392
    },
    {
      "epoch": 2.0672941176470587,
      "grad_norm": 0.7430425105171127,
      "learning_rate": 0.0002647907433434122,
      "loss": 0.4824,
      "step": 4393
    },
    {
      "epoch": 2.067764705882353,
      "grad_norm": 0.5457785764256958,
      "learning_rate": 0.00026476252076639245,
      "loss": 0.4531,
      "step": 4394
    },
    {
      "epoch": 2.068235294117647,
      "grad_norm": 0.7227932421069151,
      "learning_rate": 0.0002647342883880905,
      "loss": 0.7148,
      "step": 4395
    },
    {
      "epoch": 2.0687058823529414,
      "grad_norm": 0.55779323231783,
      "learning_rate": 0.0002647060462109176,
      "loss": 0.6797,
      "step": 4396
    },
    {
      "epoch": 2.069176470588235,
      "grad_norm": 0.5738608067884284,
      "learning_rate": 0.00026467779423728564,
      "loss": 0.5625,
      "step": 4397
    },
    {
      "epoch": 2.0696470588235294,
      "grad_norm": 0.646013735761748,
      "learning_rate": 0.0002646495324696076,
      "loss": 0.4863,
      "step": 4398
    },
    {
      "epoch": 2.0701176470588236,
      "grad_norm": 0.43584467314659137,
      "learning_rate": 0.0002646212609102972,
      "loss": 0.2637,
      "step": 4399
    },
    {
      "epoch": 2.070588235294118,
      "grad_norm": 0.8736477180827968,
      "learning_rate": 0.00026459297956176885,
      "loss": 0.9727,
      "step": 4400
    },
    {
      "epoch": 2.0710588235294116,
      "grad_norm": 0.6010549481848975,
      "learning_rate": 0.000264564688426438,
      "loss": 0.4219,
      "step": 4401
    },
    {
      "epoch": 2.071529411764706,
      "grad_norm": 0.5651395214331164,
      "learning_rate": 0.0002645363875067208,
      "loss": 0.2334,
      "step": 4402
    },
    {
      "epoch": 2.072,
      "grad_norm": 0.7163435010124525,
      "learning_rate": 0.0002645080768050342,
      "loss": 0.7539,
      "step": 4403
    },
    {
      "epoch": 2.0724705882352943,
      "grad_norm": 0.6815862018575961,
      "learning_rate": 0.00026447975632379637,
      "loss": 0.5586,
      "step": 4404
    },
    {
      "epoch": 2.072941176470588,
      "grad_norm": 0.42369136440841487,
      "learning_rate": 0.0002644514260654257,
      "loss": 0.166,
      "step": 4405
    },
    {
      "epoch": 2.0734117647058823,
      "grad_norm": 0.7913444027962176,
      "learning_rate": 0.0002644230860323419,
      "loss": 0.6797,
      "step": 4406
    },
    {
      "epoch": 2.0738823529411765,
      "grad_norm": 0.615622427490397,
      "learning_rate": 0.00026439473622696534,
      "loss": 0.459,
      "step": 4407
    },
    {
      "epoch": 2.0743529411764707,
      "grad_norm": 0.43077031250988246,
      "learning_rate": 0.0002643663766517172,
      "loss": 0.3477,
      "step": 4408
    },
    {
      "epoch": 2.0748235294117645,
      "grad_norm": 0.4574819394684922,
      "learning_rate": 0.00026433800730901947,
      "loss": 0.3145,
      "step": 4409
    },
    {
      "epoch": 2.0752941176470587,
      "grad_norm": 0.56410004157919,
      "learning_rate": 0.00026430962820129513,
      "loss": 0.4316,
      "step": 4410
    },
    {
      "epoch": 2.075764705882353,
      "grad_norm": 0.6295724090966404,
      "learning_rate": 0.00026428123933096786,
      "loss": 0.7422,
      "step": 4411
    },
    {
      "epoch": 2.076235294117647,
      "grad_norm": 0.8707475707310174,
      "learning_rate": 0.0002642528407004622,
      "loss": 1.2031,
      "step": 4412
    },
    {
      "epoch": 2.0767058823529414,
      "grad_norm": 0.5909968698819146,
      "learning_rate": 0.00026422443231220353,
      "loss": 0.6797,
      "step": 4413
    },
    {
      "epoch": 2.077176470588235,
      "grad_norm": 0.4071186031345012,
      "learning_rate": 0.0002641960141686181,
      "loss": 0.5039,
      "step": 4414
    },
    {
      "epoch": 2.0776470588235294,
      "grad_norm": 0.41222685743093906,
      "learning_rate": 0.0002641675862721329,
      "loss": 0.3691,
      "step": 4415
    },
    {
      "epoch": 2.0781176470588236,
      "grad_norm": 0.8808396341500025,
      "learning_rate": 0.00026413914862517585,
      "loss": 0.9258,
      "step": 4416
    },
    {
      "epoch": 2.078588235294118,
      "grad_norm": 0.7514126338344955,
      "learning_rate": 0.00026411070123017567,
      "loss": 0.6484,
      "step": 4417
    },
    {
      "epoch": 2.0790588235294116,
      "grad_norm": 0.4749449376970795,
      "learning_rate": 0.0002640822440895619,
      "loss": 0.2578,
      "step": 4418
    },
    {
      "epoch": 2.079529411764706,
      "grad_norm": 0.4045869513247027,
      "learning_rate": 0.0002640537772057649,
      "loss": 0.5,
      "step": 4419
    },
    {
      "epoch": 2.08,
      "grad_norm": 0.539834223270227,
      "learning_rate": 0.0002640253005812159,
      "loss": 0.2598,
      "step": 4420
    },
    {
      "epoch": 2.0804705882352943,
      "grad_norm": 0.64281978465338,
      "learning_rate": 0.00026399681421834694,
      "loss": 0.5234,
      "step": 4421
    },
    {
      "epoch": 2.080941176470588,
      "grad_norm": 0.4149099092429481,
      "learning_rate": 0.0002639683181195909,
      "loss": 0.2969,
      "step": 4422
    },
    {
      "epoch": 2.0814117647058823,
      "grad_norm": 0.5139151211908326,
      "learning_rate": 0.00026393981228738144,
      "loss": 0.4434,
      "step": 4423
    },
    {
      "epoch": 2.0818823529411765,
      "grad_norm": 0.8449838689573945,
      "learning_rate": 0.0002639112967241532,
      "loss": 0.7188,
      "step": 4424
    },
    {
      "epoch": 2.0823529411764707,
      "grad_norm": 0.4210295858835139,
      "learning_rate": 0.00026388277143234146,
      "loss": 0.375,
      "step": 4425
    },
    {
      "epoch": 2.0828235294117645,
      "grad_norm": 0.5316717859748868,
      "learning_rate": 0.0002638542364143825,
      "loss": 0.4043,
      "step": 4426
    },
    {
      "epoch": 2.0832941176470587,
      "grad_norm": 0.2967415525319554,
      "learning_rate": 0.00026382569167271317,
      "loss": 0.124,
      "step": 4427
    },
    {
      "epoch": 2.083764705882353,
      "grad_norm": 1.053726428800365,
      "learning_rate": 0.00026379713720977156,
      "loss": 0.75,
      "step": 4428
    },
    {
      "epoch": 2.084235294117647,
      "grad_norm": 1.0415095537016206,
      "learning_rate": 0.0002637685730279962,
      "loss": 0.9023,
      "step": 4429
    },
    {
      "epoch": 2.084705882352941,
      "grad_norm": 0.8284544293598962,
      "learning_rate": 0.00026373999912982676,
      "loss": 0.7812,
      "step": 4430
    },
    {
      "epoch": 2.085176470588235,
      "grad_norm": 0.3524992647087984,
      "learning_rate": 0.00026371141551770344,
      "loss": 0.2891,
      "step": 4431
    },
    {
      "epoch": 2.0856470588235294,
      "grad_norm": 0.5458904690921348,
      "learning_rate": 0.0002636828221940675,
      "loss": 0.3672,
      "step": 4432
    },
    {
      "epoch": 2.0861176470588236,
      "grad_norm": 0.7649952904108709,
      "learning_rate": 0.00026365421916136093,
      "loss": 0.625,
      "step": 4433
    },
    {
      "epoch": 2.086588235294118,
      "grad_norm": 0.5809203446846052,
      "learning_rate": 0.00026362560642202655,
      "loss": 0.418,
      "step": 4434
    },
    {
      "epoch": 2.0870588235294116,
      "grad_norm": 0.6224341864903149,
      "learning_rate": 0.0002635969839785081,
      "loss": 0.9453,
      "step": 4435
    },
    {
      "epoch": 2.087529411764706,
      "grad_norm": 0.3745984757806806,
      "learning_rate": 0.00026356835183325,
      "loss": 0.3594,
      "step": 4436
    },
    {
      "epoch": 2.088,
      "grad_norm": 0.8768067693329815,
      "learning_rate": 0.00026353970998869763,
      "loss": 0.6719,
      "step": 4437
    },
    {
      "epoch": 2.0884705882352943,
      "grad_norm": 0.7178376016503228,
      "learning_rate": 0.0002635110584472971,
      "loss": 0.8164,
      "step": 4438
    },
    {
      "epoch": 2.088941176470588,
      "grad_norm": 0.8621606225090624,
      "learning_rate": 0.00026348239721149535,
      "loss": 1.0,
      "step": 4439
    },
    {
      "epoch": 2.0894117647058823,
      "grad_norm": 0.3910442010275159,
      "learning_rate": 0.0002634537262837403,
      "loss": 0.1992,
      "step": 4440
    },
    {
      "epoch": 2.0898823529411765,
      "grad_norm": 0.6528063665881492,
      "learning_rate": 0.00026342504566648055,
      "loss": 0.5547,
      "step": 4441
    },
    {
      "epoch": 2.0903529411764707,
      "grad_norm": 0.38999398949557607,
      "learning_rate": 0.0002633963553621655,
      "loss": 0.3594,
      "step": 4442
    },
    {
      "epoch": 2.0908235294117645,
      "grad_norm": 0.48315506613629444,
      "learning_rate": 0.0002633676553732456,
      "loss": 0.2559,
      "step": 4443
    },
    {
      "epoch": 2.0912941176470587,
      "grad_norm": 0.5106488383548241,
      "learning_rate": 0.0002633389457021718,
      "loss": 0.249,
      "step": 4444
    },
    {
      "epoch": 2.091764705882353,
      "grad_norm": 0.6792614069011539,
      "learning_rate": 0.0002633102263513961,
      "loss": 0.5859,
      "step": 4445
    },
    {
      "epoch": 2.092235294117647,
      "grad_norm": 0.506595969685194,
      "learning_rate": 0.0002632814973233713,
      "loss": 0.4883,
      "step": 4446
    },
    {
      "epoch": 2.092705882352941,
      "grad_norm": 0.7800753069673577,
      "learning_rate": 0.00026325275862055104,
      "loss": 0.9141,
      "step": 4447
    },
    {
      "epoch": 2.093176470588235,
      "grad_norm": 0.694429721053364,
      "learning_rate": 0.0002632240102453896,
      "loss": 0.7773,
      "step": 4448
    },
    {
      "epoch": 2.0936470588235294,
      "grad_norm": 0.5486464669706381,
      "learning_rate": 0.0002631952522003424,
      "loss": 0.3242,
      "step": 4449
    },
    {
      "epoch": 2.0941176470588236,
      "grad_norm": 0.3425537117877426,
      "learning_rate": 0.00026316648448786536,
      "loss": 0.165,
      "step": 4450
    },
    {
      "epoch": 2.094588235294118,
      "grad_norm": 0.5357157792940483,
      "learning_rate": 0.00026313770711041556,
      "loss": 0.3594,
      "step": 4451
    },
    {
      "epoch": 2.0950588235294116,
      "grad_norm": 0.697325702409607,
      "learning_rate": 0.0002631089200704506,
      "loss": 0.5352,
      "step": 4452
    },
    {
      "epoch": 2.095529411764706,
      "grad_norm": 0.23934914326159187,
      "learning_rate": 0.00026308012337042906,
      "loss": 0.0688,
      "step": 4453
    },
    {
      "epoch": 2.096,
      "grad_norm": 0.22129148530850615,
      "learning_rate": 0.0002630513170128103,
      "loss": 0.0796,
      "step": 4454
    },
    {
      "epoch": 2.0964705882352943,
      "grad_norm": 0.8545452839347777,
      "learning_rate": 0.0002630225010000546,
      "loss": 0.7383,
      "step": 4455
    },
    {
      "epoch": 2.096941176470588,
      "grad_norm": 0.48372050885826795,
      "learning_rate": 0.0002629936753346229,
      "loss": 0.3262,
      "step": 4456
    },
    {
      "epoch": 2.0974117647058823,
      "grad_norm": 0.657482287419575,
      "learning_rate": 0.00026296484001897715,
      "loss": 0.7148,
      "step": 4457
    },
    {
      "epoch": 2.0978823529411765,
      "grad_norm": 0.5451356806763562,
      "learning_rate": 0.00026293599505558,
      "loss": 0.4277,
      "step": 4458
    },
    {
      "epoch": 2.0983529411764708,
      "grad_norm": 0.6351181501305951,
      "learning_rate": 0.00026290714044689483,
      "loss": 0.6016,
      "step": 4459
    },
    {
      "epoch": 2.0988235294117645,
      "grad_norm": 0.4081203991858068,
      "learning_rate": 0.0002628782761953861,
      "loss": 0.2891,
      "step": 4460
    },
    {
      "epoch": 2.0992941176470588,
      "grad_norm": 0.5524291974008506,
      "learning_rate": 0.0002628494023035189,
      "loss": 0.6016,
      "step": 4461
    },
    {
      "epoch": 2.099764705882353,
      "grad_norm": 0.6606601868726256,
      "learning_rate": 0.00026282051877375923,
      "loss": 0.6094,
      "step": 4462
    },
    {
      "epoch": 2.100235294117647,
      "grad_norm": 0.507610651458036,
      "learning_rate": 0.0002627916256085739,
      "loss": 0.4258,
      "step": 4463
    },
    {
      "epoch": 2.100705882352941,
      "grad_norm": 0.5097771299681934,
      "learning_rate": 0.0002627627228104305,
      "loss": 0.4121,
      "step": 4464
    },
    {
      "epoch": 2.101176470588235,
      "grad_norm": 0.5657190517422125,
      "learning_rate": 0.00026273381038179744,
      "loss": 0.582,
      "step": 4465
    },
    {
      "epoch": 2.1016470588235294,
      "grad_norm": 0.6864770514769638,
      "learning_rate": 0.000262704888325144,
      "loss": 0.5312,
      "step": 4466
    },
    {
      "epoch": 2.1021176470588236,
      "grad_norm": 0.5973058195767167,
      "learning_rate": 0.0002626759566429404,
      "loss": 0.3926,
      "step": 4467
    },
    {
      "epoch": 2.1025882352941174,
      "grad_norm": 0.546313022047002,
      "learning_rate": 0.0002626470153376573,
      "loss": 0.2637,
      "step": 4468
    },
    {
      "epoch": 2.1030588235294116,
      "grad_norm": 0.3670816759891915,
      "learning_rate": 0.00026261806441176666,
      "loss": 0.1514,
      "step": 4469
    },
    {
      "epoch": 2.103529411764706,
      "grad_norm": 0.6300116718968787,
      "learning_rate": 0.0002625891038677408,
      "loss": 0.4668,
      "step": 4470
    },
    {
      "epoch": 2.104,
      "grad_norm": 0.2942098661748279,
      "learning_rate": 0.0002625601337080533,
      "loss": 0.3672,
      "step": 4471
    },
    {
      "epoch": 2.1044705882352943,
      "grad_norm": 0.38923484726708757,
      "learning_rate": 0.00026253115393517834,
      "loss": 0.3789,
      "step": 4472
    },
    {
      "epoch": 2.104941176470588,
      "grad_norm": 0.6921368471875585,
      "learning_rate": 0.0002625021645515908,
      "loss": 0.8008,
      "step": 4473
    },
    {
      "epoch": 2.1054117647058823,
      "grad_norm": 0.4312865945048761,
      "learning_rate": 0.00026247316555976657,
      "loss": 0.4316,
      "step": 4474
    },
    {
      "epoch": 2.1058823529411765,
      "grad_norm": 0.2703728845887438,
      "learning_rate": 0.00026244415696218237,
      "loss": 0.167,
      "step": 4475
    },
    {
      "epoch": 2.1063529411764708,
      "grad_norm": 0.7637765744679236,
      "learning_rate": 0.00026241513876131564,
      "loss": 0.6602,
      "step": 4476
    },
    {
      "epoch": 2.1068235294117645,
      "grad_norm": 1.4698738840753076,
      "learning_rate": 0.00026238611095964466,
      "loss": 1.5625,
      "step": 4477
    },
    {
      "epoch": 2.1072941176470588,
      "grad_norm": 0.44864100441624494,
      "learning_rate": 0.0002623570735596485,
      "loss": 0.5117,
      "step": 4478
    },
    {
      "epoch": 2.107764705882353,
      "grad_norm": 0.5631560492767534,
      "learning_rate": 0.0002623280265638072,
      "loss": 0.2715,
      "step": 4479
    },
    {
      "epoch": 2.108235294117647,
      "grad_norm": 0.7968198656359903,
      "learning_rate": 0.0002622989699746014,
      "loss": 1.1016,
      "step": 4480
    },
    {
      "epoch": 2.108705882352941,
      "grad_norm": 0.5269050259603293,
      "learning_rate": 0.00026226990379451284,
      "loss": 0.4961,
      "step": 4481
    },
    {
      "epoch": 2.109176470588235,
      "grad_norm": 0.6192678748743397,
      "learning_rate": 0.00026224082802602373,
      "loss": 0.5117,
      "step": 4482
    },
    {
      "epoch": 2.1096470588235294,
      "grad_norm": 0.8222855086659604,
      "learning_rate": 0.00026221174267161744,
      "loss": 0.9258,
      "step": 4483
    },
    {
      "epoch": 2.1101176470588237,
      "grad_norm": 1.1853840574122236,
      "learning_rate": 0.0002621826477337779,
      "loss": 1.0859,
      "step": 4484
    },
    {
      "epoch": 2.1105882352941174,
      "grad_norm": 0.488722196403573,
      "learning_rate": 0.00026215354321498997,
      "loss": 0.3828,
      "step": 4485
    },
    {
      "epoch": 2.1110588235294117,
      "grad_norm": 0.5067280453170315,
      "learning_rate": 0.00026212442911773936,
      "loss": 0.2793,
      "step": 4486
    },
    {
      "epoch": 2.111529411764706,
      "grad_norm": 0.29201242596908916,
      "learning_rate": 0.0002620953054445125,
      "loss": 0.2559,
      "step": 4487
    },
    {
      "epoch": 2.112,
      "grad_norm": 0.4867032285806062,
      "learning_rate": 0.0002620661721977968,
      "loss": 0.2852,
      "step": 4488
    },
    {
      "epoch": 2.1124705882352943,
      "grad_norm": 0.6183108319880906,
      "learning_rate": 0.00026203702938008025,
      "loss": 0.4844,
      "step": 4489
    },
    {
      "epoch": 2.112941176470588,
      "grad_norm": 0.5995390010255607,
      "learning_rate": 0.00026200787699385186,
      "loss": 0.5039,
      "step": 4490
    },
    {
      "epoch": 2.1134117647058823,
      "grad_norm": 0.8476655449395945,
      "learning_rate": 0.0002619787150416014,
      "loss": 0.9688,
      "step": 4491
    },
    {
      "epoch": 2.1138823529411765,
      "grad_norm": 0.5943222804243082,
      "learning_rate": 0.00026194954352581944,
      "loss": 0.4082,
      "step": 4492
    },
    {
      "epoch": 2.1143529411764708,
      "grad_norm": 0.4158823997774111,
      "learning_rate": 0.0002619203624489973,
      "loss": 0.2363,
      "step": 4493
    },
    {
      "epoch": 2.1148235294117645,
      "grad_norm": 0.44177127807814376,
      "learning_rate": 0.00026189117181362733,
      "loss": 0.2227,
      "step": 4494
    },
    {
      "epoch": 2.1152941176470588,
      "grad_norm": 0.7496629088471293,
      "learning_rate": 0.0002618619716222024,
      "loss": 0.5234,
      "step": 4495
    },
    {
      "epoch": 2.115764705882353,
      "grad_norm": 0.571929234108239,
      "learning_rate": 0.00026183276187721635,
      "loss": 0.5391,
      "step": 4496
    },
    {
      "epoch": 2.116235294117647,
      "grad_norm": 0.5957085582291936,
      "learning_rate": 0.000261803542581164,
      "loss": 0.625,
      "step": 4497
    },
    {
      "epoch": 2.116705882352941,
      "grad_norm": 0.5627466969740502,
      "learning_rate": 0.0002617743137365407,
      "loss": 0.4551,
      "step": 4498
    },
    {
      "epoch": 2.117176470588235,
      "grad_norm": 0.3826784191543446,
      "learning_rate": 0.0002617450753458427,
      "loss": 0.207,
      "step": 4499
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 0.34443013881241274,
      "learning_rate": 0.00026171582741156725,
      "loss": 0.1924,
      "step": 4500
    },
    {
      "epoch": 2.1181176470588237,
      "grad_norm": 0.5682121441925181,
      "learning_rate": 0.0002616865699362121,
      "loss": 0.3984,
      "step": 4501
    },
    {
      "epoch": 2.1185882352941174,
      "grad_norm": 0.5648519765383978,
      "learning_rate": 0.0002616573029222761,
      "loss": 0.5078,
      "step": 4502
    },
    {
      "epoch": 2.1190588235294117,
      "grad_norm": 0.5503630549716446,
      "learning_rate": 0.0002616280263722588,
      "loss": 0.5781,
      "step": 4503
    },
    {
      "epoch": 2.119529411764706,
      "grad_norm": 0.6838652288772278,
      "learning_rate": 0.00026159874028866044,
      "loss": 0.5117,
      "step": 4504
    },
    {
      "epoch": 2.12,
      "grad_norm": 0.7317125472382465,
      "learning_rate": 0.0002615694446739823,
      "loss": 0.4766,
      "step": 4505
    },
    {
      "epoch": 2.1204705882352943,
      "grad_norm": 0.4725440898122109,
      "learning_rate": 0.0002615401395307263,
      "loss": 0.4551,
      "step": 4506
    },
    {
      "epoch": 2.120941176470588,
      "grad_norm": 0.4338531703312182,
      "learning_rate": 0.00026151082486139534,
      "loss": 0.416,
      "step": 4507
    },
    {
      "epoch": 2.1214117647058823,
      "grad_norm": 0.2096945849080278,
      "learning_rate": 0.0002614815006684929,
      "loss": 0.126,
      "step": 4508
    },
    {
      "epoch": 2.1218823529411766,
      "grad_norm": 0.2612862830435826,
      "learning_rate": 0.0002614521669545236,
      "loss": 0.1011,
      "step": 4509
    },
    {
      "epoch": 2.1223529411764708,
      "grad_norm": 0.6240283244795406,
      "learning_rate": 0.0002614228237219925,
      "loss": 0.4297,
      "step": 4510
    },
    {
      "epoch": 2.1228235294117646,
      "grad_norm": 0.6825857850309905,
      "learning_rate": 0.00026139347097340574,
      "loss": 0.6172,
      "step": 4511
    },
    {
      "epoch": 2.1232941176470588,
      "grad_norm": 0.8079956266420634,
      "learning_rate": 0.00026136410871127017,
      "loss": 0.6992,
      "step": 4512
    },
    {
      "epoch": 2.123764705882353,
      "grad_norm": 0.6073935561244862,
      "learning_rate": 0.0002613347369380935,
      "loss": 0.3535,
      "step": 4513
    },
    {
      "epoch": 2.124235294117647,
      "grad_norm": 0.8005418945149209,
      "learning_rate": 0.00026130535565638415,
      "loss": 0.7734,
      "step": 4514
    },
    {
      "epoch": 2.124705882352941,
      "grad_norm": 0.5303975478741595,
      "learning_rate": 0.00026127596486865155,
      "loss": 0.6406,
      "step": 4515
    },
    {
      "epoch": 2.1251764705882352,
      "grad_norm": 0.616359956612679,
      "learning_rate": 0.00026124656457740574,
      "loss": 0.5703,
      "step": 4516
    },
    {
      "epoch": 2.1256470588235294,
      "grad_norm": 0.38699071535177726,
      "learning_rate": 0.00026121715478515757,
      "loss": 0.4219,
      "step": 4517
    },
    {
      "epoch": 2.1261176470588237,
      "grad_norm": 0.8512702943216562,
      "learning_rate": 0.0002611877354944189,
      "loss": 0.8477,
      "step": 4518
    },
    {
      "epoch": 2.1265882352941174,
      "grad_norm": 0.6191685081604534,
      "learning_rate": 0.00026115830670770225,
      "loss": 0.459,
      "step": 4519
    },
    {
      "epoch": 2.1270588235294117,
      "grad_norm": 0.36686841842931256,
      "learning_rate": 0.00026112886842752104,
      "loss": 0.1982,
      "step": 4520
    },
    {
      "epoch": 2.127529411764706,
      "grad_norm": 0.578422176633533,
      "learning_rate": 0.0002610994206563893,
      "loss": 0.5469,
      "step": 4521
    },
    {
      "epoch": 2.128,
      "grad_norm": 0.6685072411347361,
      "learning_rate": 0.0002610699633968222,
      "loss": 0.4688,
      "step": 4522
    },
    {
      "epoch": 2.1284705882352943,
      "grad_norm": 0.6002185970999901,
      "learning_rate": 0.00026104049665133533,
      "loss": 0.4844,
      "step": 4523
    },
    {
      "epoch": 2.128941176470588,
      "grad_norm": 0.9148578292827221,
      "learning_rate": 0.00026101102042244543,
      "loss": 0.8477,
      "step": 4524
    },
    {
      "epoch": 2.1294117647058823,
      "grad_norm": 0.4107327687209083,
      "learning_rate": 0.0002609815347126699,
      "loss": 0.3281,
      "step": 4525
    },
    {
      "epoch": 2.1298823529411766,
      "grad_norm": 0.45366899887161655,
      "learning_rate": 0.00026095203952452694,
      "loss": 0.5391,
      "step": 4526
    },
    {
      "epoch": 2.130352941176471,
      "grad_norm": 0.34839588203714833,
      "learning_rate": 0.00026092253486053555,
      "loss": 0.1484,
      "step": 4527
    },
    {
      "epoch": 2.1308235294117646,
      "grad_norm": 0.5079827834705443,
      "learning_rate": 0.0002608930207232156,
      "loss": 0.4121,
      "step": 4528
    },
    {
      "epoch": 2.131294117647059,
      "grad_norm": 0.5468906408347735,
      "learning_rate": 0.0002608634971150878,
      "loss": 0.2539,
      "step": 4529
    },
    {
      "epoch": 2.131764705882353,
      "grad_norm": 0.46385226859010037,
      "learning_rate": 0.0002608339640386735,
      "loss": 0.5039,
      "step": 4530
    },
    {
      "epoch": 2.1322352941176472,
      "grad_norm": 0.805880364034414,
      "learning_rate": 0.00026080442149649516,
      "loss": 0.4902,
      "step": 4531
    },
    {
      "epoch": 2.132705882352941,
      "grad_norm": 0.35629187578865207,
      "learning_rate": 0.0002607748694910756,
      "loss": 0.166,
      "step": 4532
    },
    {
      "epoch": 2.1331764705882352,
      "grad_norm": 0.47951376223772263,
      "learning_rate": 0.0002607453080249389,
      "loss": 0.5078,
      "step": 4533
    },
    {
      "epoch": 2.1336470588235295,
      "grad_norm": 0.26567692454015834,
      "learning_rate": 0.0002607157371006097,
      "loss": 0.2773,
      "step": 4534
    },
    {
      "epoch": 2.1341176470588237,
      "grad_norm": 0.4672214601096073,
      "learning_rate": 0.00026068615672061353,
      "loss": 0.6523,
      "step": 4535
    },
    {
      "epoch": 2.1345882352941175,
      "grad_norm": 0.6294362039029532,
      "learning_rate": 0.0002606565668874766,
      "loss": 0.3477,
      "step": 4536
    },
    {
      "epoch": 2.1350588235294117,
      "grad_norm": 0.40923276061539154,
      "learning_rate": 0.00026062696760372606,
      "loss": 0.2305,
      "step": 4537
    },
    {
      "epoch": 2.135529411764706,
      "grad_norm": 0.6339981200169077,
      "learning_rate": 0.0002605973588718899,
      "loss": 0.5078,
      "step": 4538
    },
    {
      "epoch": 2.136,
      "grad_norm": 0.491524710533055,
      "learning_rate": 0.00026056774069449685,
      "loss": 0.4102,
      "step": 4539
    },
    {
      "epoch": 2.1364705882352943,
      "grad_norm": 0.6260145736921263,
      "learning_rate": 0.00026053811307407635,
      "loss": 0.2871,
      "step": 4540
    },
    {
      "epoch": 2.136941176470588,
      "grad_norm": 0.47487988545494714,
      "learning_rate": 0.00026050847601315886,
      "loss": 0.3945,
      "step": 4541
    },
    {
      "epoch": 2.1374117647058823,
      "grad_norm": 0.48280173046538344,
      "learning_rate": 0.00026047882951427543,
      "loss": 0.3691,
      "step": 4542
    },
    {
      "epoch": 2.1378823529411766,
      "grad_norm": 0.6775284365070031,
      "learning_rate": 0.00026044917357995813,
      "loss": 0.5625,
      "step": 4543
    },
    {
      "epoch": 2.138352941176471,
      "grad_norm": 0.44823860392795767,
      "learning_rate": 0.00026041950821273965,
      "loss": 0.1973,
      "step": 4544
    },
    {
      "epoch": 2.1388235294117646,
      "grad_norm": 0.4929910556258834,
      "learning_rate": 0.0002603898334151535,
      "loss": 0.4727,
      "step": 4545
    },
    {
      "epoch": 2.139294117647059,
      "grad_norm": 0.4758426120483008,
      "learning_rate": 0.00026036014918973414,
      "loss": 0.4844,
      "step": 4546
    },
    {
      "epoch": 2.139764705882353,
      "grad_norm": 0.41394902843361403,
      "learning_rate": 0.00026033045553901674,
      "loss": 0.1494,
      "step": 4547
    },
    {
      "epoch": 2.1402352941176472,
      "grad_norm": 0.75087499122002,
      "learning_rate": 0.0002603007524655373,
      "loss": 0.459,
      "step": 4548
    },
    {
      "epoch": 2.140705882352941,
      "grad_norm": 0.5612041627476565,
      "learning_rate": 0.0002602710399718325,
      "loss": 0.6055,
      "step": 4549
    },
    {
      "epoch": 2.1411764705882352,
      "grad_norm": 0.7133394767103015,
      "learning_rate": 0.0002602413180604401,
      "loss": 0.6445,
      "step": 4550
    },
    {
      "epoch": 2.1416470588235295,
      "grad_norm": 0.5807579018616598,
      "learning_rate": 0.0002602115867338983,
      "loss": 0.3027,
      "step": 4551
    },
    {
      "epoch": 2.1421176470588237,
      "grad_norm": 0.7750312668063777,
      "learning_rate": 0.0002601818459947465,
      "loss": 0.7695,
      "step": 4552
    },
    {
      "epoch": 2.1425882352941175,
      "grad_norm": 0.5739452535648946,
      "learning_rate": 0.0002601520958455246,
      "loss": 0.5195,
      "step": 4553
    },
    {
      "epoch": 2.1430588235294117,
      "grad_norm": 0.49086108795715283,
      "learning_rate": 0.00026012233628877336,
      "loss": 0.2393,
      "step": 4554
    },
    {
      "epoch": 2.143529411764706,
      "grad_norm": 0.6365066335945561,
      "learning_rate": 0.0002600925673270345,
      "loss": 0.4141,
      "step": 4555
    },
    {
      "epoch": 2.144,
      "grad_norm": 0.39877461165327,
      "learning_rate": 0.0002600627889628504,
      "loss": 0.2305,
      "step": 4556
    },
    {
      "epoch": 2.1444705882352944,
      "grad_norm": 0.604583050248581,
      "learning_rate": 0.00026003300119876424,
      "loss": 0.6016,
      "step": 4557
    },
    {
      "epoch": 2.144941176470588,
      "grad_norm": 0.43299541385182144,
      "learning_rate": 0.0002600032040373201,
      "loss": 0.2969,
      "step": 4558
    },
    {
      "epoch": 2.1454117647058824,
      "grad_norm": 0.5110693046072966,
      "learning_rate": 0.00025997339748106275,
      "loss": 0.2988,
      "step": 4559
    },
    {
      "epoch": 2.1458823529411766,
      "grad_norm": 0.5763764024566413,
      "learning_rate": 0.00025994358153253784,
      "loss": 0.6875,
      "step": 4560
    },
    {
      "epoch": 2.1463529411764704,
      "grad_norm": 0.38086848787024447,
      "learning_rate": 0.0002599137561942918,
      "loss": 0.2559,
      "step": 4561
    },
    {
      "epoch": 2.1468235294117646,
      "grad_norm": 0.4731494908490144,
      "learning_rate": 0.00025988392146887183,
      "loss": 0.2793,
      "step": 4562
    },
    {
      "epoch": 2.147294117647059,
      "grad_norm": 0.4898165875993446,
      "learning_rate": 0.00025985407735882606,
      "loss": 0.2988,
      "step": 4563
    },
    {
      "epoch": 2.147764705882353,
      "grad_norm": 0.51861177135845,
      "learning_rate": 0.0002598242238667032,
      "loss": 0.3887,
      "step": 4564
    },
    {
      "epoch": 2.1482352941176472,
      "grad_norm": 0.25791007747631073,
      "learning_rate": 0.00025979436099505296,
      "loss": 0.1367,
      "step": 4565
    },
    {
      "epoch": 2.148705882352941,
      "grad_norm": 0.4233528131823742,
      "learning_rate": 0.0002597644887464257,
      "loss": 0.3164,
      "step": 4566
    },
    {
      "epoch": 2.1491764705882352,
      "grad_norm": 0.496983255818866,
      "learning_rate": 0.00025973460712337275,
      "loss": 0.4258,
      "step": 4567
    },
    {
      "epoch": 2.1496470588235295,
      "grad_norm": 0.36834589846778076,
      "learning_rate": 0.0002597047161284461,
      "loss": 0.168,
      "step": 4568
    },
    {
      "epoch": 2.1501176470588237,
      "grad_norm": 0.9244189005480623,
      "learning_rate": 0.00025967481576419864,
      "loss": 0.8477,
      "step": 4569
    },
    {
      "epoch": 2.1505882352941175,
      "grad_norm": 0.4599347587517413,
      "learning_rate": 0.0002596449060331839,
      "loss": 0.3066,
      "step": 4570
    },
    {
      "epoch": 2.1510588235294117,
      "grad_norm": 0.4759263022403602,
      "learning_rate": 0.0002596149869379564,
      "loss": 0.6328,
      "step": 4571
    },
    {
      "epoch": 2.151529411764706,
      "grad_norm": 0.41142954423779965,
      "learning_rate": 0.00025958505848107133,
      "loss": 0.3789,
      "step": 4572
    },
    {
      "epoch": 2.152,
      "grad_norm": 0.39317733620433776,
      "learning_rate": 0.00025955512066508477,
      "loss": 0.1992,
      "step": 4573
    },
    {
      "epoch": 2.152470588235294,
      "grad_norm": 0.213866777478681,
      "learning_rate": 0.0002595251734925536,
      "loss": 0.1211,
      "step": 4574
    },
    {
      "epoch": 2.152941176470588,
      "grad_norm": 0.6224086195435299,
      "learning_rate": 0.00025949521696603536,
      "loss": 0.3125,
      "step": 4575
    },
    {
      "epoch": 2.1534117647058824,
      "grad_norm": 0.5480503291468086,
      "learning_rate": 0.0002594652510880885,
      "loss": 0.5078,
      "step": 4576
    },
    {
      "epoch": 2.1538823529411766,
      "grad_norm": 0.5103614238136905,
      "learning_rate": 0.0002594352758612723,
      "loss": 0.3086,
      "step": 4577
    },
    {
      "epoch": 2.1543529411764704,
      "grad_norm": 0.8272477148851224,
      "learning_rate": 0.0002594052912881467,
      "loss": 0.5234,
      "step": 4578
    },
    {
      "epoch": 2.1548235294117646,
      "grad_norm": 0.5668428170776236,
      "learning_rate": 0.00025937529737127263,
      "loss": 0.5742,
      "step": 4579
    },
    {
      "epoch": 2.155294117647059,
      "grad_norm": 0.49049644403904713,
      "learning_rate": 0.0002593452941132117,
      "loss": 0.3594,
      "step": 4580
    },
    {
      "epoch": 2.155764705882353,
      "grad_norm": 0.4983097780501627,
      "learning_rate": 0.0002593152815165263,
      "loss": 0.2852,
      "step": 4581
    },
    {
      "epoch": 2.1562352941176472,
      "grad_norm": 0.34409938120778294,
      "learning_rate": 0.0002592852595837797,
      "loss": 0.2988,
      "step": 4582
    },
    {
      "epoch": 2.156705882352941,
      "grad_norm": 0.31625449476996453,
      "learning_rate": 0.00025925522831753584,
      "loss": 0.543,
      "step": 4583
    },
    {
      "epoch": 2.1571764705882353,
      "grad_norm": 0.24636018928432912,
      "learning_rate": 0.00025922518772035963,
      "loss": 0.1836,
      "step": 4584
    },
    {
      "epoch": 2.1576470588235295,
      "grad_norm": 0.6062864236395132,
      "learning_rate": 0.0002591951377948166,
      "loss": 0.6055,
      "step": 4585
    },
    {
      "epoch": 2.1581176470588237,
      "grad_norm": 0.9088114387381037,
      "learning_rate": 0.00025916507854347327,
      "loss": 0.7031,
      "step": 4586
    },
    {
      "epoch": 2.1585882352941175,
      "grad_norm": 0.3609378485764631,
      "learning_rate": 0.00025913500996889677,
      "loss": 0.3711,
      "step": 4587
    },
    {
      "epoch": 2.1590588235294117,
      "grad_norm": 0.5454830203840353,
      "learning_rate": 0.00025910493207365504,
      "loss": 0.3105,
      "step": 4588
    },
    {
      "epoch": 2.159529411764706,
      "grad_norm": 0.6357964752222928,
      "learning_rate": 0.000259074844860317,
      "loss": 0.5195,
      "step": 4589
    },
    {
      "epoch": 2.16,
      "grad_norm": 0.37277088920269014,
      "learning_rate": 0.00025904474833145225,
      "loss": 0.1602,
      "step": 4590
    },
    {
      "epoch": 2.160470588235294,
      "grad_norm": 0.5389448915254579,
      "learning_rate": 0.00025901464248963106,
      "loss": 0.2422,
      "step": 4591
    },
    {
      "epoch": 2.160941176470588,
      "grad_norm": 0.7418147266455108,
      "learning_rate": 0.00025898452733742473,
      "loss": 0.7305,
      "step": 4592
    },
    {
      "epoch": 2.1614117647058824,
      "grad_norm": 0.43624543167904073,
      "learning_rate": 0.00025895440287740514,
      "loss": 0.2051,
      "step": 4593
    },
    {
      "epoch": 2.1618823529411766,
      "grad_norm": 0.6265529001704931,
      "learning_rate": 0.00025892426911214517,
      "loss": 0.5664,
      "step": 4594
    },
    {
      "epoch": 2.1623529411764704,
      "grad_norm": 0.5562908136063804,
      "learning_rate": 0.0002588941260442183,
      "loss": 0.334,
      "step": 4595
    },
    {
      "epoch": 2.1628235294117646,
      "grad_norm": 0.620936059400018,
      "learning_rate": 0.00025886397367619904,
      "loss": 0.4531,
      "step": 4596
    },
    {
      "epoch": 2.163294117647059,
      "grad_norm": 0.49062798672685487,
      "learning_rate": 0.00025883381201066236,
      "loss": 0.2383,
      "step": 4597
    },
    {
      "epoch": 2.163764705882353,
      "grad_norm": 0.6490667968493191,
      "learning_rate": 0.0002588036410501844,
      "loss": 0.5,
      "step": 4598
    },
    {
      "epoch": 2.1642352941176473,
      "grad_norm": 1.0395680025891052,
      "learning_rate": 0.00025877346079734175,
      "loss": 1.1953,
      "step": 4599
    },
    {
      "epoch": 2.164705882352941,
      "grad_norm": 0.6330600419024494,
      "learning_rate": 0.000258743271254712,
      "loss": 0.6289,
      "step": 4600
    },
    {
      "epoch": 2.1651764705882353,
      "grad_norm": 0.6042689575902254,
      "learning_rate": 0.00025871307242487354,
      "loss": 0.4043,
      "step": 4601
    },
    {
      "epoch": 2.1656470588235295,
      "grad_norm": 0.38280047635034714,
      "learning_rate": 0.00025868286431040546,
      "loss": 0.2559,
      "step": 4602
    },
    {
      "epoch": 2.1661176470588237,
      "grad_norm": 0.3006464045368231,
      "learning_rate": 0.0002586526469138877,
      "loss": 0.1562,
      "step": 4603
    },
    {
      "epoch": 2.1665882352941175,
      "grad_norm": 0.460368408871988,
      "learning_rate": 0.0002586224202379009,
      "loss": 0.3066,
      "step": 4604
    },
    {
      "epoch": 2.1670588235294117,
      "grad_norm": 0.595865021481656,
      "learning_rate": 0.00025859218428502667,
      "loss": 0.5195,
      "step": 4605
    },
    {
      "epoch": 2.167529411764706,
      "grad_norm": 0.3184738033183429,
      "learning_rate": 0.00025856193905784725,
      "loss": 0.2852,
      "step": 4606
    },
    {
      "epoch": 2.168,
      "grad_norm": 0.9287463630139501,
      "learning_rate": 0.00025853168455894566,
      "loss": 1.1719,
      "step": 4607
    },
    {
      "epoch": 2.168470588235294,
      "grad_norm": 0.9077259559511451,
      "learning_rate": 0.000258501420790906,
      "loss": 0.9297,
      "step": 4608
    },
    {
      "epoch": 2.168941176470588,
      "grad_norm": 0.5002774027295798,
      "learning_rate": 0.0002584711477563127,
      "loss": 0.3281,
      "step": 4609
    },
    {
      "epoch": 2.1694117647058824,
      "grad_norm": 0.4227697080155563,
      "learning_rate": 0.0002584408654577514,
      "loss": 0.4258,
      "step": 4610
    },
    {
      "epoch": 2.1698823529411766,
      "grad_norm": 0.5829228329286302,
      "learning_rate": 0.0002584105738978083,
      "loss": 0.5352,
      "step": 4611
    },
    {
      "epoch": 2.1703529411764704,
      "grad_norm": 0.8105676869771716,
      "learning_rate": 0.00025838027307907037,
      "loss": 0.6055,
      "step": 4612
    },
    {
      "epoch": 2.1708235294117646,
      "grad_norm": 0.17359658221802388,
      "learning_rate": 0.00025834996300412556,
      "loss": 0.0405,
      "step": 4613
    },
    {
      "epoch": 2.171294117647059,
      "grad_norm": 0.3467456180220078,
      "learning_rate": 0.0002583196436755624,
      "loss": 0.1602,
      "step": 4614
    },
    {
      "epoch": 2.171764705882353,
      "grad_norm": 0.6567305540355813,
      "learning_rate": 0.00025828931509597045,
      "loss": 0.4492,
      "step": 4615
    },
    {
      "epoch": 2.1722352941176473,
      "grad_norm": 0.5591394386090078,
      "learning_rate": 0.0002582589772679398,
      "loss": 0.373,
      "step": 4616
    },
    {
      "epoch": 2.172705882352941,
      "grad_norm": 0.474173987779546,
      "learning_rate": 0.0002582286301940615,
      "loss": 0.249,
      "step": 4617
    },
    {
      "epoch": 2.1731764705882353,
      "grad_norm": 0.6226875195269982,
      "learning_rate": 0.00025819827387692733,
      "loss": 0.5586,
      "step": 4618
    },
    {
      "epoch": 2.1736470588235295,
      "grad_norm": 0.6481704072335323,
      "learning_rate": 0.0002581679083191298,
      "loss": 0.6055,
      "step": 4619
    },
    {
      "epoch": 2.1741176470588237,
      "grad_norm": 0.6385402094146627,
      "learning_rate": 0.0002581375335232624,
      "loss": 0.4766,
      "step": 4620
    },
    {
      "epoch": 2.1745882352941175,
      "grad_norm": 0.8042841353152016,
      "learning_rate": 0.00025810714949191916,
      "loss": 0.9219,
      "step": 4621
    },
    {
      "epoch": 2.1750588235294117,
      "grad_norm": 0.2738803961410565,
      "learning_rate": 0.00025807675622769514,
      "loss": 0.1982,
      "step": 4622
    },
    {
      "epoch": 2.175529411764706,
      "grad_norm": 0.5615840868278879,
      "learning_rate": 0.00025804635373318603,
      "loss": 0.5781,
      "step": 4623
    },
    {
      "epoch": 2.176,
      "grad_norm": 0.534064538373418,
      "learning_rate": 0.0002580159420109883,
      "loss": 0.3203,
      "step": 4624
    },
    {
      "epoch": 2.176470588235294,
      "grad_norm": 0.494021017350798,
      "learning_rate": 0.00025798552106369937,
      "loss": 0.2139,
      "step": 4625
    },
    {
      "epoch": 2.176941176470588,
      "grad_norm": 0.7418102896204116,
      "learning_rate": 0.00025795509089391726,
      "loss": 0.7383,
      "step": 4626
    },
    {
      "epoch": 2.1774117647058824,
      "grad_norm": 0.33567195911835584,
      "learning_rate": 0.00025792465150424083,
      "loss": 0.3105,
      "step": 4627
    },
    {
      "epoch": 2.1778823529411766,
      "grad_norm": 0.6094557806273464,
      "learning_rate": 0.00025789420289726983,
      "loss": 0.75,
      "step": 4628
    },
    {
      "epoch": 2.1783529411764704,
      "grad_norm": 0.6506059153907325,
      "learning_rate": 0.00025786374507560467,
      "loss": 0.6211,
      "step": 4629
    },
    {
      "epoch": 2.1788235294117646,
      "grad_norm": 0.44313644501543425,
      "learning_rate": 0.0002578332780418466,
      "loss": 0.4746,
      "step": 4630
    },
    {
      "epoch": 2.179294117647059,
      "grad_norm": 0.5858765990254171,
      "learning_rate": 0.0002578028017985977,
      "loss": 0.6367,
      "step": 4631
    },
    {
      "epoch": 2.179764705882353,
      "grad_norm": 0.5487409768487513,
      "learning_rate": 0.00025777231634846074,
      "loss": 0.5547,
      "step": 4632
    },
    {
      "epoch": 2.1802352941176473,
      "grad_norm": 0.4153544792591298,
      "learning_rate": 0.00025774182169403933,
      "loss": 0.4844,
      "step": 4633
    },
    {
      "epoch": 2.180705882352941,
      "grad_norm": 0.6816842245797069,
      "learning_rate": 0.00025771131783793787,
      "loss": 0.5312,
      "step": 4634
    },
    {
      "epoch": 2.1811764705882353,
      "grad_norm": 0.977285542899838,
      "learning_rate": 0.00025768080478276163,
      "loss": 0.4004,
      "step": 4635
    },
    {
      "epoch": 2.1816470588235295,
      "grad_norm": 0.4967986723326878,
      "learning_rate": 0.0002576502825311164,
      "loss": 0.3125,
      "step": 4636
    },
    {
      "epoch": 2.1821176470588237,
      "grad_norm": 0.6192227283874776,
      "learning_rate": 0.000257619751085609,
      "loss": 0.6953,
      "step": 4637
    },
    {
      "epoch": 2.1825882352941175,
      "grad_norm": 0.509410482941071,
      "learning_rate": 0.00025758921044884706,
      "loss": 0.6641,
      "step": 4638
    },
    {
      "epoch": 2.1830588235294117,
      "grad_norm": 0.5551912570925963,
      "learning_rate": 0.0002575586606234388,
      "loss": 0.4414,
      "step": 4639
    },
    {
      "epoch": 2.183529411764706,
      "grad_norm": 0.5510210457459473,
      "learning_rate": 0.0002575281016119933,
      "loss": 0.5312,
      "step": 4640
    },
    {
      "epoch": 2.184,
      "grad_norm": 0.44649839573470346,
      "learning_rate": 0.00025749753341712056,
      "loss": 0.3672,
      "step": 4641
    },
    {
      "epoch": 2.184470588235294,
      "grad_norm": 0.5778036032070772,
      "learning_rate": 0.0002574669560414312,
      "loss": 0.6016,
      "step": 4642
    },
    {
      "epoch": 2.184941176470588,
      "grad_norm": 0.5353720855475039,
      "learning_rate": 0.0002574363694875367,
      "loss": 0.373,
      "step": 4643
    },
    {
      "epoch": 2.1854117647058824,
      "grad_norm": 0.5358885715556854,
      "learning_rate": 0.0002574057737580492,
      "loss": 0.4004,
      "step": 4644
    },
    {
      "epoch": 2.1858823529411766,
      "grad_norm": 0.5956402586966274,
      "learning_rate": 0.0002573751688555818,
      "loss": 0.3672,
      "step": 4645
    },
    {
      "epoch": 2.1863529411764704,
      "grad_norm": 0.8072290785086871,
      "learning_rate": 0.0002573445547827484,
      "loss": 0.8086,
      "step": 4646
    },
    {
      "epoch": 2.1868235294117646,
      "grad_norm": 0.6258696983178585,
      "learning_rate": 0.0002573139315421634,
      "loss": 0.7188,
      "step": 4647
    },
    {
      "epoch": 2.187294117647059,
      "grad_norm": 0.3819126554081378,
      "learning_rate": 0.0002572832991364423,
      "loss": 0.1602,
      "step": 4648
    },
    {
      "epoch": 2.187764705882353,
      "grad_norm": 0.5717356536381497,
      "learning_rate": 0.00025725265756820125,
      "loss": 0.5156,
      "step": 4649
    },
    {
      "epoch": 2.1882352941176473,
      "grad_norm": 0.5473467176176838,
      "learning_rate": 0.00025722200684005715,
      "loss": 0.6875,
      "step": 4650
    },
    {
      "epoch": 2.188705882352941,
      "grad_norm": 0.5416755007723242,
      "learning_rate": 0.0002571913469546277,
      "loss": 0.5859,
      "step": 4651
    },
    {
      "epoch": 2.1891764705882353,
      "grad_norm": 0.6051007636488295,
      "learning_rate": 0.00025716067791453157,
      "loss": 0.4414,
      "step": 4652
    },
    {
      "epoch": 2.1896470588235295,
      "grad_norm": 0.5531174286657609,
      "learning_rate": 0.0002571299997223878,
      "loss": 0.5391,
      "step": 4653
    },
    {
      "epoch": 2.1901176470588237,
      "grad_norm": 0.5292056961301418,
      "learning_rate": 0.0002570993123808167,
      "loss": 0.7852,
      "step": 4654
    },
    {
      "epoch": 2.1905882352941175,
      "grad_norm": 0.5949349279386033,
      "learning_rate": 0.00025706861589243894,
      "loss": 0.6602,
      "step": 4655
    },
    {
      "epoch": 2.1910588235294117,
      "grad_norm": 0.6983320393907568,
      "learning_rate": 0.0002570379102598762,
      "loss": 0.6602,
      "step": 4656
    },
    {
      "epoch": 2.191529411764706,
      "grad_norm": 0.5834977578850148,
      "learning_rate": 0.000257007195485751,
      "loss": 0.5977,
      "step": 4657
    },
    {
      "epoch": 2.192,
      "grad_norm": 0.6186669647877616,
      "learning_rate": 0.00025697647157268635,
      "loss": 0.8047,
      "step": 4658
    },
    {
      "epoch": 2.192470588235294,
      "grad_norm": 0.45605079546777544,
      "learning_rate": 0.0002569457385233063,
      "loss": 0.3633,
      "step": 4659
    },
    {
      "epoch": 2.192941176470588,
      "grad_norm": 0.6919978249667137,
      "learning_rate": 0.00025691499634023564,
      "loss": 0.7383,
      "step": 4660
    },
    {
      "epoch": 2.1934117647058824,
      "grad_norm": 0.45934483208444954,
      "learning_rate": 0.0002568842450260999,
      "loss": 0.3535,
      "step": 4661
    },
    {
      "epoch": 2.1938823529411766,
      "grad_norm": 0.7394222175224198,
      "learning_rate": 0.0002568534845835253,
      "loss": 0.6562,
      "step": 4662
    },
    {
      "epoch": 2.1943529411764704,
      "grad_norm": 0.6941523224224644,
      "learning_rate": 0.0002568227150151391,
      "loss": 0.6602,
      "step": 4663
    },
    {
      "epoch": 2.1948235294117646,
      "grad_norm": 0.51723179088871,
      "learning_rate": 0.000256791936323569,
      "loss": 0.4004,
      "step": 4664
    },
    {
      "epoch": 2.195294117647059,
      "grad_norm": 0.599312752712392,
      "learning_rate": 0.00025676114851144375,
      "loss": 0.5469,
      "step": 4665
    },
    {
      "epoch": 2.195764705882353,
      "grad_norm": 0.5114814897596465,
      "learning_rate": 0.0002567303515813928,
      "loss": 0.3711,
      "step": 4666
    },
    {
      "epoch": 2.1962352941176473,
      "grad_norm": 0.38477383259491577,
      "learning_rate": 0.0002566995455360463,
      "loss": 0.21,
      "step": 4667
    },
    {
      "epoch": 2.196705882352941,
      "grad_norm": 0.4972422883216076,
      "learning_rate": 0.00025666873037803517,
      "loss": 0.3906,
      "step": 4668
    },
    {
      "epoch": 2.1971764705882353,
      "grad_norm": 0.40959582165796726,
      "learning_rate": 0.00025663790610999134,
      "loss": 0.2578,
      "step": 4669
    },
    {
      "epoch": 2.1976470588235295,
      "grad_norm": 0.43847524482382705,
      "learning_rate": 0.0002566070727345473,
      "loss": 0.3477,
      "step": 4670
    },
    {
      "epoch": 2.1981176470588237,
      "grad_norm": 0.6486054416217321,
      "learning_rate": 0.00025657623025433627,
      "loss": 0.6523,
      "step": 4671
    },
    {
      "epoch": 2.1985882352941175,
      "grad_norm": 1.1847519818322796,
      "learning_rate": 0.0002565453786719924,
      "loss": 1.2891,
      "step": 4672
    },
    {
      "epoch": 2.1990588235294117,
      "grad_norm": 0.6167365655003213,
      "learning_rate": 0.00025651451799015067,
      "loss": 0.6953,
      "step": 4673
    },
    {
      "epoch": 2.199529411764706,
      "grad_norm": 0.3903600328344621,
      "learning_rate": 0.0002564836482114466,
      "loss": 0.4883,
      "step": 4674
    },
    {
      "epoch": 2.2,
      "grad_norm": 0.6712434570382724,
      "learning_rate": 0.00025645276933851667,
      "loss": 0.7148,
      "step": 4675
    },
    {
      "epoch": 2.200470588235294,
      "grad_norm": 0.785729133167382,
      "learning_rate": 0.00025642188137399815,
      "loss": 0.4219,
      "step": 4676
    },
    {
      "epoch": 2.200941176470588,
      "grad_norm": 0.5029715274131354,
      "learning_rate": 0.00025639098432052895,
      "loss": 0.4863,
      "step": 4677
    },
    {
      "epoch": 2.2014117647058824,
      "grad_norm": 0.40620254250592824,
      "learning_rate": 0.0002563600781807478,
      "loss": 0.1924,
      "step": 4678
    },
    {
      "epoch": 2.2018823529411766,
      "grad_norm": 0.8011879762001856,
      "learning_rate": 0.0002563291629572943,
      "loss": 0.8516,
      "step": 4679
    },
    {
      "epoch": 2.2023529411764704,
      "grad_norm": 0.6282727950422572,
      "learning_rate": 0.00025629823865280874,
      "loss": 0.5391,
      "step": 4680
    },
    {
      "epoch": 2.2028235294117646,
      "grad_norm": 0.41432308442481763,
      "learning_rate": 0.00025626730526993223,
      "loss": 0.4375,
      "step": 4681
    },
    {
      "epoch": 2.203294117647059,
      "grad_norm": 0.5072163798041255,
      "learning_rate": 0.00025623636281130657,
      "loss": 0.5469,
      "step": 4682
    },
    {
      "epoch": 2.203764705882353,
      "grad_norm": 0.6014899296450597,
      "learning_rate": 0.0002562054112795745,
      "loss": 0.668,
      "step": 4683
    },
    {
      "epoch": 2.2042352941176473,
      "grad_norm": 0.41803425216342366,
      "learning_rate": 0.00025617445067737933,
      "loss": 0.3027,
      "step": 4684
    },
    {
      "epoch": 2.204705882352941,
      "grad_norm": 0.3265646839082366,
      "learning_rate": 0.0002561434810073653,
      "loss": 0.293,
      "step": 4685
    },
    {
      "epoch": 2.2051764705882353,
      "grad_norm": 0.6285160616016466,
      "learning_rate": 0.00025611250227217737,
      "loss": 0.7422,
      "step": 4686
    },
    {
      "epoch": 2.2056470588235295,
      "grad_norm": 0.4161634133294261,
      "learning_rate": 0.0002560815144744612,
      "loss": 0.3203,
      "step": 4687
    },
    {
      "epoch": 2.2061176470588233,
      "grad_norm": 0.2535566291137702,
      "learning_rate": 0.00025605051761686345,
      "loss": 0.2598,
      "step": 4688
    },
    {
      "epoch": 2.2065882352941175,
      "grad_norm": 0.47050702092325325,
      "learning_rate": 0.0002560195117020313,
      "loss": 0.3965,
      "step": 4689
    },
    {
      "epoch": 2.2070588235294117,
      "grad_norm": 0.49260763873462443,
      "learning_rate": 0.0002559884967326128,
      "loss": 0.5,
      "step": 4690
    },
    {
      "epoch": 2.207529411764706,
      "grad_norm": 0.5906988829007608,
      "learning_rate": 0.0002559574727112569,
      "loss": 0.4922,
      "step": 4691
    },
    {
      "epoch": 2.208,
      "grad_norm": 0.5524704385623527,
      "learning_rate": 0.000255926439640613,
      "loss": 0.4375,
      "step": 4692
    },
    {
      "epoch": 2.208470588235294,
      "grad_norm": 0.6661250434838879,
      "learning_rate": 0.00025589539752333165,
      "loss": 0.8633,
      "step": 4693
    },
    {
      "epoch": 2.208941176470588,
      "grad_norm": 0.5648644353052868,
      "learning_rate": 0.0002558643463620639,
      "loss": 0.5586,
      "step": 4694
    },
    {
      "epoch": 2.2094117647058824,
      "grad_norm": 0.44082424746919674,
      "learning_rate": 0.0002558332861594617,
      "loss": 0.2695,
      "step": 4695
    },
    {
      "epoch": 2.2098823529411766,
      "grad_norm": 0.41424352570045725,
      "learning_rate": 0.0002558022169181778,
      "loss": 0.1387,
      "step": 4696
    },
    {
      "epoch": 2.2103529411764704,
      "grad_norm": 0.3597849822322065,
      "learning_rate": 0.00025577113864086557,
      "loss": 0.2676,
      "step": 4697
    },
    {
      "epoch": 2.2108235294117646,
      "grad_norm": 0.3863403624583322,
      "learning_rate": 0.0002557400513301794,
      "loss": 0.2393,
      "step": 4698
    },
    {
      "epoch": 2.211294117647059,
      "grad_norm": 0.35576280574810015,
      "learning_rate": 0.00025570895498877404,
      "loss": 0.2578,
      "step": 4699
    },
    {
      "epoch": 2.211764705882353,
      "grad_norm": 0.4631819083794413,
      "learning_rate": 0.00025567784961930546,
      "loss": 0.498,
      "step": 4700
    },
    {
      "epoch": 2.212235294117647,
      "grad_norm": 0.9422828600114094,
      "learning_rate": 0.0002556467352244302,
      "loss": 1.0703,
      "step": 4701
    },
    {
      "epoch": 2.212705882352941,
      "grad_norm": 0.3297236794937075,
      "learning_rate": 0.00025561561180680563,
      "loss": 0.1963,
      "step": 4702
    },
    {
      "epoch": 2.2131764705882353,
      "grad_norm": 0.8099726940082717,
      "learning_rate": 0.0002555844793690897,
      "loss": 0.7734,
      "step": 4703
    },
    {
      "epoch": 2.2136470588235295,
      "grad_norm": 0.3474192382730098,
      "learning_rate": 0.0002555533379139414,
      "loss": 0.2422,
      "step": 4704
    },
    {
      "epoch": 2.2141176470588233,
      "grad_norm": 0.3147048555448843,
      "learning_rate": 0.00025552218744402026,
      "loss": 0.2715,
      "step": 4705
    },
    {
      "epoch": 2.2145882352941175,
      "grad_norm": 0.5364831915028744,
      "learning_rate": 0.00025549102796198673,
      "loss": 0.6289,
      "step": 4706
    },
    {
      "epoch": 2.2150588235294117,
      "grad_norm": 0.3791645471044389,
      "learning_rate": 0.00025545985947050197,
      "loss": 0.3379,
      "step": 4707
    },
    {
      "epoch": 2.215529411764706,
      "grad_norm": 0.42730595652831016,
      "learning_rate": 0.00025542868197222797,
      "loss": 0.252,
      "step": 4708
    },
    {
      "epoch": 2.216,
      "grad_norm": 0.3450461556736601,
      "learning_rate": 0.00025539749546982736,
      "loss": 0.1709,
      "step": 4709
    },
    {
      "epoch": 2.216470588235294,
      "grad_norm": 0.4876132432132117,
      "learning_rate": 0.00025536629996596367,
      "loss": 0.3418,
      "step": 4710
    },
    {
      "epoch": 2.216941176470588,
      "grad_norm": 0.5615775691666713,
      "learning_rate": 0.0002553350954633012,
      "loss": 0.4434,
      "step": 4711
    },
    {
      "epoch": 2.2174117647058824,
      "grad_norm": 0.7640600061289432,
      "learning_rate": 0.00025530388196450487,
      "loss": 0.5781,
      "step": 4712
    },
    {
      "epoch": 2.2178823529411766,
      "grad_norm": 0.5776594883047445,
      "learning_rate": 0.0002552726594722405,
      "loss": 0.4922,
      "step": 4713
    },
    {
      "epoch": 2.2183529411764704,
      "grad_norm": 0.7647681721244859,
      "learning_rate": 0.00025524142798917466,
      "loss": 0.668,
      "step": 4714
    },
    {
      "epoch": 2.2188235294117646,
      "grad_norm": 0.354209468673459,
      "learning_rate": 0.00025521018751797464,
      "loss": 0.1963,
      "step": 4715
    },
    {
      "epoch": 2.219294117647059,
      "grad_norm": 0.40221504586565054,
      "learning_rate": 0.0002551789380613086,
      "loss": 0.168,
      "step": 4716
    },
    {
      "epoch": 2.219764705882353,
      "grad_norm": 0.8189785350147063,
      "learning_rate": 0.00025514767962184535,
      "loss": 0.9062,
      "step": 4717
    },
    {
      "epoch": 2.220235294117647,
      "grad_norm": 0.5469480011468023,
      "learning_rate": 0.0002551164122022545,
      "loss": 0.4707,
      "step": 4718
    },
    {
      "epoch": 2.220705882352941,
      "grad_norm": 0.6337945792444262,
      "learning_rate": 0.00025508513580520644,
      "loss": 0.5352,
      "step": 4719
    },
    {
      "epoch": 2.2211764705882353,
      "grad_norm": 0.3254760483514532,
      "learning_rate": 0.00025505385043337236,
      "loss": 0.373,
      "step": 4720
    },
    {
      "epoch": 2.2216470588235295,
      "grad_norm": 0.5049601781517146,
      "learning_rate": 0.0002550225560894241,
      "loss": 0.4258,
      "step": 4721
    },
    {
      "epoch": 2.2221176470588233,
      "grad_norm": 0.46161687268407753,
      "learning_rate": 0.00025499125277603446,
      "loss": 0.3418,
      "step": 4722
    },
    {
      "epoch": 2.2225882352941175,
      "grad_norm": 0.5364634579849639,
      "learning_rate": 0.0002549599404958769,
      "loss": 0.543,
      "step": 4723
    },
    {
      "epoch": 2.2230588235294118,
      "grad_norm": 0.5514358530914742,
      "learning_rate": 0.0002549286192516256,
      "loss": 0.5234,
      "step": 4724
    },
    {
      "epoch": 2.223529411764706,
      "grad_norm": 0.1274833722035149,
      "learning_rate": 0.0002548972890459555,
      "loss": 0.0378,
      "step": 4725
    },
    {
      "epoch": 2.224,
      "grad_norm": 0.6778228370549233,
      "learning_rate": 0.00025486594988154235,
      "loss": 0.6719,
      "step": 4726
    },
    {
      "epoch": 2.224470588235294,
      "grad_norm": 0.7995440581468375,
      "learning_rate": 0.0002548346017610628,
      "loss": 0.6953,
      "step": 4727
    },
    {
      "epoch": 2.224941176470588,
      "grad_norm": 0.6100287166743275,
      "learning_rate": 0.000254803244687194,
      "loss": 0.6914,
      "step": 4728
    },
    {
      "epoch": 2.2254117647058824,
      "grad_norm": 0.7487466296235754,
      "learning_rate": 0.0002547718786626141,
      "loss": 0.5312,
      "step": 4729
    },
    {
      "epoch": 2.2258823529411766,
      "grad_norm": 0.5556641740824063,
      "learning_rate": 0.0002547405036900018,
      "loss": 0.5078,
      "step": 4730
    },
    {
      "epoch": 2.2263529411764704,
      "grad_norm": 0.5711240316582132,
      "learning_rate": 0.0002547091197720368,
      "loss": 0.4766,
      "step": 4731
    },
    {
      "epoch": 2.2268235294117646,
      "grad_norm": 0.5491464025860697,
      "learning_rate": 0.0002546777269113993,
      "loss": 0.6602,
      "step": 4732
    },
    {
      "epoch": 2.227294117647059,
      "grad_norm": 0.6329312201610895,
      "learning_rate": 0.0002546463251107705,
      "loss": 0.6523,
      "step": 4733
    },
    {
      "epoch": 2.227764705882353,
      "grad_norm": 0.4292959933003736,
      "learning_rate": 0.0002546149143728323,
      "loss": 0.4277,
      "step": 4734
    },
    {
      "epoch": 2.228235294117647,
      "grad_norm": 0.4585909086252167,
      "learning_rate": 0.0002545834947002672,
      "loss": 0.5078,
      "step": 4735
    },
    {
      "epoch": 2.228705882352941,
      "grad_norm": 0.4636491188005646,
      "learning_rate": 0.00025455206609575876,
      "loss": 0.2617,
      "step": 4736
    },
    {
      "epoch": 2.2291764705882353,
      "grad_norm": 0.3990076204033606,
      "learning_rate": 0.000254520628561991,
      "loss": 0.1953,
      "step": 4737
    },
    {
      "epoch": 2.2296470588235295,
      "grad_norm": 0.43349031034529484,
      "learning_rate": 0.00025448918210164886,
      "loss": 0.293,
      "step": 4738
    },
    {
      "epoch": 2.2301176470588233,
      "grad_norm": 0.8425376503403846,
      "learning_rate": 0.00025445772671741816,
      "loss": 0.9492,
      "step": 4739
    },
    {
      "epoch": 2.2305882352941175,
      "grad_norm": 0.8846783431593574,
      "learning_rate": 0.00025442626241198513,
      "loss": 0.7656,
      "step": 4740
    },
    {
      "epoch": 2.2310588235294118,
      "grad_norm": 0.7600668967210176,
      "learning_rate": 0.00025439478918803715,
      "loss": 0.7188,
      "step": 4741
    },
    {
      "epoch": 2.231529411764706,
      "grad_norm": 0.7952444373150923,
      "learning_rate": 0.00025436330704826205,
      "loss": 1.2969,
      "step": 4742
    },
    {
      "epoch": 2.232,
      "grad_norm": 0.5984663772408362,
      "learning_rate": 0.0002543318159953487,
      "loss": 0.5117,
      "step": 4743
    },
    {
      "epoch": 2.232470588235294,
      "grad_norm": 0.5839718817291628,
      "learning_rate": 0.00025430031603198644,
      "loss": 0.5195,
      "step": 4744
    },
    {
      "epoch": 2.232941176470588,
      "grad_norm": 0.5880376848621552,
      "learning_rate": 0.0002542688071608657,
      "loss": 0.7539,
      "step": 4745
    },
    {
      "epoch": 2.2334117647058824,
      "grad_norm": 0.3034718165507242,
      "learning_rate": 0.0002542372893846773,
      "loss": 0.2432,
      "step": 4746
    },
    {
      "epoch": 2.2338823529411767,
      "grad_norm": 0.6805604884651651,
      "learning_rate": 0.00025420576270611313,
      "loss": 0.4023,
      "step": 4747
    },
    {
      "epoch": 2.2343529411764704,
      "grad_norm": 0.5807881658187246,
      "learning_rate": 0.00025417422712786575,
      "loss": 0.5234,
      "step": 4748
    },
    {
      "epoch": 2.2348235294117647,
      "grad_norm": 0.47003757941142155,
      "learning_rate": 0.00025414268265262833,
      "loss": 0.2676,
      "step": 4749
    },
    {
      "epoch": 2.235294117647059,
      "grad_norm": 0.7782981777259841,
      "learning_rate": 0.0002541111292830951,
      "loss": 0.6953,
      "step": 4750
    },
    {
      "epoch": 2.235764705882353,
      "grad_norm": 0.43556212174977676,
      "learning_rate": 0.0002540795670219606,
      "loss": 0.3652,
      "step": 4751
    },
    {
      "epoch": 2.236235294117647,
      "grad_norm": 0.6565502348561176,
      "learning_rate": 0.0002540479958719207,
      "loss": 0.5703,
      "step": 4752
    },
    {
      "epoch": 2.236705882352941,
      "grad_norm": 1.0631079510377632,
      "learning_rate": 0.0002540164158356716,
      "loss": 1.25,
      "step": 4753
    },
    {
      "epoch": 2.2371764705882353,
      "grad_norm": 0.5019378517709951,
      "learning_rate": 0.0002539848269159104,
      "loss": 0.4473,
      "step": 4754
    },
    {
      "epoch": 2.2376470588235295,
      "grad_norm": 0.6645011386693095,
      "learning_rate": 0.00025395322911533484,
      "loss": 0.6328,
      "step": 4755
    },
    {
      "epoch": 2.2381176470588233,
      "grad_norm": 0.6144125927688279,
      "learning_rate": 0.0002539216224366437,
      "loss": 0.7734,
      "step": 4756
    },
    {
      "epoch": 2.2385882352941175,
      "grad_norm": 0.6499999213764176,
      "learning_rate": 0.00025389000688253626,
      "loss": 0.6562,
      "step": 4757
    },
    {
      "epoch": 2.2390588235294118,
      "grad_norm": 0.41705597275174183,
      "learning_rate": 0.00025385838245571264,
      "loss": 0.2676,
      "step": 4758
    },
    {
      "epoch": 2.239529411764706,
      "grad_norm": 0.5599773430925044,
      "learning_rate": 0.00025382674915887374,
      "loss": 0.7148,
      "step": 4759
    },
    {
      "epoch": 2.24,
      "grad_norm": 0.565872773680918,
      "learning_rate": 0.0002537951069947212,
      "loss": 0.6992,
      "step": 4760
    },
    {
      "epoch": 2.240470588235294,
      "grad_norm": 0.4214065922242212,
      "learning_rate": 0.00025376345596595743,
      "loss": 0.2217,
      "step": 4761
    },
    {
      "epoch": 2.240941176470588,
      "grad_norm": 0.4691081742897868,
      "learning_rate": 0.0002537317960752855,
      "loss": 0.2988,
      "step": 4762
    },
    {
      "epoch": 2.2414117647058824,
      "grad_norm": 0.8038233972163957,
      "learning_rate": 0.0002537001273254094,
      "loss": 0.9922,
      "step": 4763
    },
    {
      "epoch": 2.2418823529411767,
      "grad_norm": 0.6326591012016582,
      "learning_rate": 0.00025366844971903385,
      "loss": 0.5156,
      "step": 4764
    },
    {
      "epoch": 2.2423529411764704,
      "grad_norm": 0.7620481171715653,
      "learning_rate": 0.00025363676325886416,
      "loss": 1.0312,
      "step": 4765
    },
    {
      "epoch": 2.2428235294117647,
      "grad_norm": 0.5468115588463813,
      "learning_rate": 0.00025360506794760653,
      "loss": 0.3496,
      "step": 4766
    },
    {
      "epoch": 2.243294117647059,
      "grad_norm": 0.5189403861375624,
      "learning_rate": 0.00025357336378796795,
      "loss": 0.3633,
      "step": 4767
    },
    {
      "epoch": 2.243764705882353,
      "grad_norm": 0.27024825379069994,
      "learning_rate": 0.0002535416507826561,
      "loss": 0.0972,
      "step": 4768
    },
    {
      "epoch": 2.244235294117647,
      "grad_norm": 0.6237069107375527,
      "learning_rate": 0.0002535099289343793,
      "loss": 0.5742,
      "step": 4769
    },
    {
      "epoch": 2.244705882352941,
      "grad_norm": 0.4712130763316326,
      "learning_rate": 0.00025347819824584695,
      "loss": 0.4453,
      "step": 4770
    },
    {
      "epoch": 2.2451764705882353,
      "grad_norm": 0.5657084191351914,
      "learning_rate": 0.0002534464587197689,
      "loss": 0.2324,
      "step": 4771
    },
    {
      "epoch": 2.2456470588235296,
      "grad_norm": 0.38193782791478714,
      "learning_rate": 0.00025341471035885587,
      "loss": 0.3184,
      "step": 4772
    },
    {
      "epoch": 2.2461176470588233,
      "grad_norm": 0.5535248630774564,
      "learning_rate": 0.0002533829531658193,
      "loss": 0.793,
      "step": 4773
    },
    {
      "epoch": 2.2465882352941176,
      "grad_norm": 0.6045599483315213,
      "learning_rate": 0.00025335118714337146,
      "loss": 0.5664,
      "step": 4774
    },
    {
      "epoch": 2.2470588235294118,
      "grad_norm": 0.5014863581777941,
      "learning_rate": 0.0002533194122942253,
      "loss": 0.5352,
      "step": 4775
    },
    {
      "epoch": 2.247529411764706,
      "grad_norm": 0.6467535903112364,
      "learning_rate": 0.00025328762862109457,
      "loss": 0.8945,
      "step": 4776
    },
    {
      "epoch": 2.248,
      "grad_norm": 0.2407265438219574,
      "learning_rate": 0.00025325583612669373,
      "loss": 0.0791,
      "step": 4777
    },
    {
      "epoch": 2.248470588235294,
      "grad_norm": 0.6680208564805495,
      "learning_rate": 0.00025322403481373797,
      "loss": 0.4902,
      "step": 4778
    },
    {
      "epoch": 2.248941176470588,
      "grad_norm": 0.6015426749226206,
      "learning_rate": 0.00025319222468494335,
      "loss": 0.6562,
      "step": 4779
    },
    {
      "epoch": 2.2494117647058824,
      "grad_norm": 0.3060787567853923,
      "learning_rate": 0.00025316040574302663,
      "loss": 0.2236,
      "step": 4780
    },
    {
      "epoch": 2.2498823529411767,
      "grad_norm": 0.5108562399354737,
      "learning_rate": 0.00025312857799070524,
      "loss": 0.5117,
      "step": 4781
    },
    {
      "epoch": 2.2503529411764704,
      "grad_norm": 0.39638828922516733,
      "learning_rate": 0.0002530967414306974,
      "loss": 0.3652,
      "step": 4782
    },
    {
      "epoch": 2.2508235294117647,
      "grad_norm": 0.3098522073921604,
      "learning_rate": 0.0002530648960657222,
      "loss": 0.2793,
      "step": 4783
    },
    {
      "epoch": 2.251294117647059,
      "grad_norm": 0.5562282967250922,
      "learning_rate": 0.0002530330418984993,
      "loss": 0.3379,
      "step": 4784
    },
    {
      "epoch": 2.251764705882353,
      "grad_norm": 0.37552643287006127,
      "learning_rate": 0.00025300117893174934,
      "loss": 0.1455,
      "step": 4785
    },
    {
      "epoch": 2.252235294117647,
      "grad_norm": 0.5669388357722107,
      "learning_rate": 0.0002529693071681934,
      "loss": 0.6055,
      "step": 4786
    },
    {
      "epoch": 2.252705882352941,
      "grad_norm": 0.24938091880229474,
      "learning_rate": 0.00025293742661055364,
      "loss": 0.1885,
      "step": 4787
    },
    {
      "epoch": 2.2531764705882353,
      "grad_norm": 0.6377524130544341,
      "learning_rate": 0.0002529055372615527,
      "loss": 0.6172,
      "step": 4788
    },
    {
      "epoch": 2.2536470588235296,
      "grad_norm": 0.3100126581311615,
      "learning_rate": 0.00025287363912391403,
      "loss": 0.1216,
      "step": 4789
    },
    {
      "epoch": 2.2541176470588233,
      "grad_norm": 0.9031956417974011,
      "learning_rate": 0.00025284173220036215,
      "loss": 1.1094,
      "step": 4790
    },
    {
      "epoch": 2.2545882352941176,
      "grad_norm": 0.5521750100595488,
      "learning_rate": 0.00025280981649362184,
      "loss": 0.6875,
      "step": 4791
    },
    {
      "epoch": 2.255058823529412,
      "grad_norm": 0.6210622950081289,
      "learning_rate": 0.0002527778920064189,
      "loss": 0.5898,
      "step": 4792
    },
    {
      "epoch": 2.255529411764706,
      "grad_norm": 0.698496377519439,
      "learning_rate": 0.00025274595874147984,
      "loss": 0.6406,
      "step": 4793
    },
    {
      "epoch": 2.2560000000000002,
      "grad_norm": 0.500421469720562,
      "learning_rate": 0.000252714016701532,
      "loss": 0.25,
      "step": 4794
    },
    {
      "epoch": 2.256470588235294,
      "grad_norm": 0.35702723179834395,
      "learning_rate": 0.0002526820658893033,
      "loss": 0.2891,
      "step": 4795
    },
    {
      "epoch": 2.2569411764705882,
      "grad_norm": 0.5678562497465466,
      "learning_rate": 0.0002526501063075225,
      "loss": 0.5938,
      "step": 4796
    },
    {
      "epoch": 2.2574117647058825,
      "grad_norm": 0.49312511072846094,
      "learning_rate": 0.00025261813795891916,
      "loss": 0.4688,
      "step": 4797
    },
    {
      "epoch": 2.2578823529411762,
      "grad_norm": 0.4991868106852379,
      "learning_rate": 0.00025258616084622345,
      "loss": 0.5039,
      "step": 4798
    },
    {
      "epoch": 2.2583529411764705,
      "grad_norm": 0.6452611634051185,
      "learning_rate": 0.00025255417497216647,
      "loss": 0.7383,
      "step": 4799
    },
    {
      "epoch": 2.2588235294117647,
      "grad_norm": 0.5736045640062332,
      "learning_rate": 0.00025252218033947993,
      "loss": 0.6055,
      "step": 4800
    },
    {
      "epoch": 2.259294117647059,
      "grad_norm": 0.5614289898502112,
      "learning_rate": 0.0002524901769508963,
      "loss": 0.5117,
      "step": 4801
    },
    {
      "epoch": 2.259764705882353,
      "grad_norm": 0.4823537725839248,
      "learning_rate": 0.0002524581648091488,
      "loss": 0.5625,
      "step": 4802
    },
    {
      "epoch": 2.260235294117647,
      "grad_norm": 0.38854258073613673,
      "learning_rate": 0.00025242614391697153,
      "loss": 0.2373,
      "step": 4803
    },
    {
      "epoch": 2.260705882352941,
      "grad_norm": 0.22198622123772982,
      "learning_rate": 0.00025239411427709915,
      "loss": 0.0762,
      "step": 4804
    },
    {
      "epoch": 2.2611764705882353,
      "grad_norm": 0.14570712229420926,
      "learning_rate": 0.0002523620758922672,
      "loss": 0.0938,
      "step": 4805
    },
    {
      "epoch": 2.2616470588235296,
      "grad_norm": 0.6110989923750207,
      "learning_rate": 0.00025233002876521183,
      "loss": 0.25,
      "step": 4806
    },
    {
      "epoch": 2.2621176470588233,
      "grad_norm": 0.5290938674481777,
      "learning_rate": 0.0002522979728986701,
      "loss": 0.4707,
      "step": 4807
    },
    {
      "epoch": 2.2625882352941176,
      "grad_norm": 0.8085669541079211,
      "learning_rate": 0.00025226590829537977,
      "loss": 0.5664,
      "step": 4808
    },
    {
      "epoch": 2.263058823529412,
      "grad_norm": 0.4666047088136573,
      "learning_rate": 0.0002522338349580792,
      "loss": 0.3418,
      "step": 4809
    },
    {
      "epoch": 2.263529411764706,
      "grad_norm": 0.3527417482243147,
      "learning_rate": 0.00025220175288950765,
      "loss": 0.3262,
      "step": 4810
    },
    {
      "epoch": 2.2640000000000002,
      "grad_norm": 0.695549786569643,
      "learning_rate": 0.00025216966209240516,
      "loss": 0.5859,
      "step": 4811
    },
    {
      "epoch": 2.264470588235294,
      "grad_norm": 0.5724795815664626,
      "learning_rate": 0.00025213756256951234,
      "loss": 0.4316,
      "step": 4812
    },
    {
      "epoch": 2.2649411764705882,
      "grad_norm": 0.3256485452306068,
      "learning_rate": 0.00025210545432357076,
      "loss": 0.1621,
      "step": 4813
    },
    {
      "epoch": 2.2654117647058825,
      "grad_norm": 0.3498054353045448,
      "learning_rate": 0.0002520733373573225,
      "loss": 0.2988,
      "step": 4814
    },
    {
      "epoch": 2.2658823529411762,
      "grad_norm": 0.22748834682824148,
      "learning_rate": 0.0002520412116735106,
      "loss": 0.0791,
      "step": 4815
    },
    {
      "epoch": 2.2663529411764705,
      "grad_norm": 0.5074969377761122,
      "learning_rate": 0.0002520090772748787,
      "loss": 0.3047,
      "step": 4816
    },
    {
      "epoch": 2.2668235294117647,
      "grad_norm": 0.4994642296386455,
      "learning_rate": 0.00025197693416417126,
      "loss": 0.3379,
      "step": 4817
    },
    {
      "epoch": 2.267294117647059,
      "grad_norm": 0.46039243030780086,
      "learning_rate": 0.0002519447823441334,
      "loss": 0.4609,
      "step": 4818
    },
    {
      "epoch": 2.267764705882353,
      "grad_norm": 0.5858537390157668,
      "learning_rate": 0.00025191262181751113,
      "loss": 0.4785,
      "step": 4819
    },
    {
      "epoch": 2.268235294117647,
      "grad_norm": 0.6690329185937054,
      "learning_rate": 0.00025188045258705107,
      "loss": 0.3789,
      "step": 4820
    },
    {
      "epoch": 2.268705882352941,
      "grad_norm": 0.88477768987527,
      "learning_rate": 0.0002518482746555007,
      "loss": 1.0156,
      "step": 4821
    },
    {
      "epoch": 2.2691764705882354,
      "grad_norm": 0.2610297389015804,
      "learning_rate": 0.000251816088025608,
      "loss": 0.2314,
      "step": 4822
    },
    {
      "epoch": 2.2696470588235296,
      "grad_norm": 0.4696405999281979,
      "learning_rate": 0.00025178389270012205,
      "loss": 0.418,
      "step": 4823
    },
    {
      "epoch": 2.2701176470588234,
      "grad_norm": 0.557979358629659,
      "learning_rate": 0.0002517516886817924,
      "loss": 0.334,
      "step": 4824
    },
    {
      "epoch": 2.2705882352941176,
      "grad_norm": 0.6952222592432911,
      "learning_rate": 0.00025171947597336946,
      "loss": 0.5508,
      "step": 4825
    },
    {
      "epoch": 2.271058823529412,
      "grad_norm": 0.7118608576469394,
      "learning_rate": 0.00025168725457760434,
      "loss": 0.5508,
      "step": 4826
    },
    {
      "epoch": 2.271529411764706,
      "grad_norm": 0.33803615981837365,
      "learning_rate": 0.0002516550244972489,
      "loss": 0.2393,
      "step": 4827
    },
    {
      "epoch": 2.2720000000000002,
      "grad_norm": 0.6431103794126961,
      "learning_rate": 0.00025162278573505575,
      "loss": 0.5117,
      "step": 4828
    },
    {
      "epoch": 2.272470588235294,
      "grad_norm": 0.5715814215703803,
      "learning_rate": 0.0002515905382937782,
      "loss": 0.6211,
      "step": 4829
    },
    {
      "epoch": 2.2729411764705882,
      "grad_norm": 0.45641426701172466,
      "learning_rate": 0.0002515582821761704,
      "loss": 0.332,
      "step": 4830
    },
    {
      "epoch": 2.2734117647058825,
      "grad_norm": 0.5115570600367919,
      "learning_rate": 0.0002515260173849872,
      "loss": 0.5781,
      "step": 4831
    },
    {
      "epoch": 2.2738823529411762,
      "grad_norm": 0.3769067173015736,
      "learning_rate": 0.00025149374392298407,
      "loss": 0.1865,
      "step": 4832
    },
    {
      "epoch": 2.2743529411764705,
      "grad_norm": 0.9229992303198541,
      "learning_rate": 0.00025146146179291746,
      "loss": 1.1328,
      "step": 4833
    },
    {
      "epoch": 2.2748235294117647,
      "grad_norm": 0.5415128833345699,
      "learning_rate": 0.0002514291709975443,
      "loss": 0.5508,
      "step": 4834
    },
    {
      "epoch": 2.275294117647059,
      "grad_norm": 0.5684073863875051,
      "learning_rate": 0.00025139687153962246,
      "loss": 0.5508,
      "step": 4835
    },
    {
      "epoch": 2.275764705882353,
      "grad_norm": 0.33693092158657095,
      "learning_rate": 0.00025136456342191043,
      "loss": 0.2422,
      "step": 4836
    },
    {
      "epoch": 2.276235294117647,
      "grad_norm": 0.47884463946109557,
      "learning_rate": 0.0002513322466471675,
      "loss": 0.5703,
      "step": 4837
    },
    {
      "epoch": 2.276705882352941,
      "grad_norm": 0.6209502661837735,
      "learning_rate": 0.00025129992121815365,
      "loss": 0.5859,
      "step": 4838
    },
    {
      "epoch": 2.2771764705882354,
      "grad_norm": 0.4346435529071709,
      "learning_rate": 0.00025126758713762964,
      "loss": 0.3125,
      "step": 4839
    },
    {
      "epoch": 2.2776470588235296,
      "grad_norm": 0.5255343328749668,
      "learning_rate": 0.000251235244408357,
      "loss": 0.6875,
      "step": 4840
    },
    {
      "epoch": 2.2781176470588234,
      "grad_norm": 0.4804016463572148,
      "learning_rate": 0.00025120289303309795,
      "loss": 0.5781,
      "step": 4841
    },
    {
      "epoch": 2.2785882352941176,
      "grad_norm": 0.49305073036704483,
      "learning_rate": 0.00025117053301461546,
      "loss": 0.5234,
      "step": 4842
    },
    {
      "epoch": 2.279058823529412,
      "grad_norm": 0.3510102032011162,
      "learning_rate": 0.00025113816435567315,
      "loss": 0.2207,
      "step": 4843
    },
    {
      "epoch": 2.279529411764706,
      "grad_norm": 0.4894179493852025,
      "learning_rate": 0.00025110578705903557,
      "loss": 0.2373,
      "step": 4844
    },
    {
      "epoch": 2.2800000000000002,
      "grad_norm": 0.5521802865345787,
      "learning_rate": 0.0002510734011274678,
      "loss": 0.5352,
      "step": 4845
    },
    {
      "epoch": 2.280470588235294,
      "grad_norm": 0.5628092821705656,
      "learning_rate": 0.00025104100656373585,
      "loss": 0.4961,
      "step": 4846
    },
    {
      "epoch": 2.2809411764705882,
      "grad_norm": 0.3713467391323399,
      "learning_rate": 0.0002510086033706063,
      "loss": 0.1279,
      "step": 4847
    },
    {
      "epoch": 2.2814117647058825,
      "grad_norm": 0.5711824224260695,
      "learning_rate": 0.0002509761915508466,
      "loss": 0.4102,
      "step": 4848
    },
    {
      "epoch": 2.2818823529411763,
      "grad_norm": 0.49257038448787693,
      "learning_rate": 0.0002509437711072249,
      "loss": 0.2676,
      "step": 4849
    },
    {
      "epoch": 2.2823529411764705,
      "grad_norm": 0.2857121889039428,
      "learning_rate": 0.00025091134204250997,
      "loss": 0.1318,
      "step": 4850
    },
    {
      "epoch": 2.2828235294117647,
      "grad_norm": 0.4180795542534351,
      "learning_rate": 0.0002508789043594715,
      "loss": 0.2432,
      "step": 4851
    },
    {
      "epoch": 2.283294117647059,
      "grad_norm": 0.7360560367756671,
      "learning_rate": 0.00025084645806087974,
      "loss": 0.6719,
      "step": 4852
    },
    {
      "epoch": 2.283764705882353,
      "grad_norm": 0.6385837635662431,
      "learning_rate": 0.00025081400314950586,
      "loss": 0.5703,
      "step": 4853
    },
    {
      "epoch": 2.284235294117647,
      "grad_norm": 0.7719581114519443,
      "learning_rate": 0.0002507815396281216,
      "loss": 0.8281,
      "step": 4854
    },
    {
      "epoch": 2.284705882352941,
      "grad_norm": 0.6798443311955135,
      "learning_rate": 0.0002507490674994995,
      "loss": 0.7461,
      "step": 4855
    },
    {
      "epoch": 2.2851764705882354,
      "grad_norm": 0.5283674915957259,
      "learning_rate": 0.0002507165867664129,
      "loss": 0.4609,
      "step": 4856
    },
    {
      "epoch": 2.2856470588235296,
      "grad_norm": 0.4442693523541522,
      "learning_rate": 0.0002506840974316357,
      "loss": 0.3164,
      "step": 4857
    },
    {
      "epoch": 2.2861176470588234,
      "grad_norm": 0.44116437829641114,
      "learning_rate": 0.00025065159949794283,
      "loss": 0.2598,
      "step": 4858
    },
    {
      "epoch": 2.2865882352941176,
      "grad_norm": 0.45812379749460597,
      "learning_rate": 0.00025061909296810966,
      "loss": 0.3438,
      "step": 4859
    },
    {
      "epoch": 2.287058823529412,
      "grad_norm": 0.6005226915493933,
      "learning_rate": 0.0002505865778449124,
      "loss": 0.6445,
      "step": 4860
    },
    {
      "epoch": 2.287529411764706,
      "grad_norm": 0.44010506219826306,
      "learning_rate": 0.000250554054131128,
      "loss": 0.2793,
      "step": 4861
    },
    {
      "epoch": 2.288,
      "grad_norm": 0.6727951503085008,
      "learning_rate": 0.00025052152182953414,
      "loss": 0.6367,
      "step": 4862
    },
    {
      "epoch": 2.288470588235294,
      "grad_norm": 0.4164089284625522,
      "learning_rate": 0.0002504889809429093,
      "loss": 0.166,
      "step": 4863
    },
    {
      "epoch": 2.2889411764705883,
      "grad_norm": 1.0659300813016581,
      "learning_rate": 0.0002504564314740326,
      "loss": 1.1016,
      "step": 4864
    },
    {
      "epoch": 2.2894117647058825,
      "grad_norm": 0.5057424647891264,
      "learning_rate": 0.0002504238734256839,
      "loss": 0.6094,
      "step": 4865
    },
    {
      "epoch": 2.2898823529411763,
      "grad_norm": 0.6682499380057625,
      "learning_rate": 0.0002503913068006438,
      "loss": 0.4551,
      "step": 4866
    },
    {
      "epoch": 2.2903529411764705,
      "grad_norm": 0.5696758417278159,
      "learning_rate": 0.00025035873160169373,
      "loss": 0.4883,
      "step": 4867
    },
    {
      "epoch": 2.2908235294117647,
      "grad_norm": 0.27195660505859376,
      "learning_rate": 0.0002503261478316157,
      "loss": 0.1641,
      "step": 4868
    },
    {
      "epoch": 2.291294117647059,
      "grad_norm": 0.5142894490885546,
      "learning_rate": 0.0002502935554931926,
      "loss": 0.6328,
      "step": 4869
    },
    {
      "epoch": 2.291764705882353,
      "grad_norm": 0.5269900887769777,
      "learning_rate": 0.0002502609545892079,
      "loss": 0.582,
      "step": 4870
    },
    {
      "epoch": 2.292235294117647,
      "grad_norm": 0.5181949285836019,
      "learning_rate": 0.00025022834512244583,
      "loss": 0.5039,
      "step": 4871
    },
    {
      "epoch": 2.292705882352941,
      "grad_norm": 0.7083948372345779,
      "learning_rate": 0.00025019572709569153,
      "loss": 0.6641,
      "step": 4872
    },
    {
      "epoch": 2.2931764705882354,
      "grad_norm": 0.5346003700820869,
      "learning_rate": 0.0002501631005117307,
      "loss": 0.6016,
      "step": 4873
    },
    {
      "epoch": 2.2936470588235296,
      "grad_norm": 0.5629621736485146,
      "learning_rate": 0.0002501304653733497,
      "loss": 0.291,
      "step": 4874
    },
    {
      "epoch": 2.2941176470588234,
      "grad_norm": 0.4153039282287993,
      "learning_rate": 0.0002500978216833359,
      "loss": 0.3457,
      "step": 4875
    },
    {
      "epoch": 2.2945882352941176,
      "grad_norm": 0.689466259508461,
      "learning_rate": 0.0002500651694444771,
      "loss": 0.7109,
      "step": 4876
    },
    {
      "epoch": 2.295058823529412,
      "grad_norm": 0.45696268364849363,
      "learning_rate": 0.00025003250865956203,
      "loss": 0.4297,
      "step": 4877
    },
    {
      "epoch": 2.295529411764706,
      "grad_norm": 0.4401352032780643,
      "learning_rate": 0.0002499998393313801,
      "loss": 0.4609,
      "step": 4878
    },
    {
      "epoch": 2.296,
      "grad_norm": 0.47477851861662856,
      "learning_rate": 0.00024996716146272137,
      "loss": 0.4297,
      "step": 4879
    },
    {
      "epoch": 2.296470588235294,
      "grad_norm": 1.054300012910902,
      "learning_rate": 0.0002499344750563767,
      "loss": 1.2031,
      "step": 4880
    },
    {
      "epoch": 2.2969411764705883,
      "grad_norm": 0.2898023440783706,
      "learning_rate": 0.0002499017801151377,
      "loss": 0.252,
      "step": 4881
    },
    {
      "epoch": 2.2974117647058825,
      "grad_norm": 0.4916797125595257,
      "learning_rate": 0.0002498690766417967,
      "loss": 0.3145,
      "step": 4882
    },
    {
      "epoch": 2.2978823529411763,
      "grad_norm": 0.45161802939102613,
      "learning_rate": 0.00024983636463914665,
      "loss": 0.4121,
      "step": 4883
    },
    {
      "epoch": 2.2983529411764705,
      "grad_norm": 0.4786337889273318,
      "learning_rate": 0.00024980364410998133,
      "loss": 0.6211,
      "step": 4884
    },
    {
      "epoch": 2.2988235294117647,
      "grad_norm": 0.3779286150612372,
      "learning_rate": 0.0002497709150570953,
      "loss": 0.4707,
      "step": 4885
    },
    {
      "epoch": 2.299294117647059,
      "grad_norm": 0.4225568829712211,
      "learning_rate": 0.0002497381774832838,
      "loss": 0.5469,
      "step": 4886
    },
    {
      "epoch": 2.299764705882353,
      "grad_norm": 0.3818486983911503,
      "learning_rate": 0.00024970543139134275,
      "loss": 0.1934,
      "step": 4887
    },
    {
      "epoch": 2.300235294117647,
      "grad_norm": 0.27470856680450256,
      "learning_rate": 0.0002496726767840687,
      "loss": 0.0781,
      "step": 4888
    },
    {
      "epoch": 2.300705882352941,
      "grad_norm": 0.649270840142484,
      "learning_rate": 0.0002496399136642592,
      "loss": 0.5742,
      "step": 4889
    },
    {
      "epoch": 2.3011764705882354,
      "grad_norm": 0.056556393752884807,
      "learning_rate": 0.0002496071420347124,
      "loss": 0.0154,
      "step": 4890
    },
    {
      "epoch": 2.3016470588235296,
      "grad_norm": 0.2532425229788149,
      "learning_rate": 0.00024957436189822706,
      "loss": 0.1494,
      "step": 4891
    },
    {
      "epoch": 2.3021176470588234,
      "grad_norm": 0.5733783724420775,
      "learning_rate": 0.0002495415732576028,
      "loss": 0.7266,
      "step": 4892
    },
    {
      "epoch": 2.3025882352941176,
      "grad_norm": 0.6600689714893817,
      "learning_rate": 0.00024950877611564,
      "loss": 0.832,
      "step": 4893
    },
    {
      "epoch": 2.303058823529412,
      "grad_norm": 0.6438289747911669,
      "learning_rate": 0.0002494759704751395,
      "loss": 0.6602,
      "step": 4894
    },
    {
      "epoch": 2.303529411764706,
      "grad_norm": 0.2994814203481865,
      "learning_rate": 0.00024944315633890327,
      "loss": 0.1182,
      "step": 4895
    },
    {
      "epoch": 2.304,
      "grad_norm": 0.6077509551801137,
      "learning_rate": 0.0002494103337097337,
      "loss": 0.6406,
      "step": 4896
    },
    {
      "epoch": 2.304470588235294,
      "grad_norm": 0.6145622619567569,
      "learning_rate": 0.00024937750259043407,
      "loss": 0.5664,
      "step": 4897
    },
    {
      "epoch": 2.3049411764705883,
      "grad_norm": 0.44782560461041576,
      "learning_rate": 0.00024934466298380824,
      "loss": 0.543,
      "step": 4898
    },
    {
      "epoch": 2.3054117647058825,
      "grad_norm": 0.7480653099371988,
      "learning_rate": 0.0002493118148926609,
      "loss": 0.6445,
      "step": 4899
    },
    {
      "epoch": 2.3058823529411763,
      "grad_norm": 0.7702191019610566,
      "learning_rate": 0.00024927895831979745,
      "loss": 0.3574,
      "step": 4900
    },
    {
      "epoch": 2.3063529411764705,
      "grad_norm": 0.6988697429500782,
      "learning_rate": 0.000249246093268024,
      "loss": 0.3203,
      "step": 4901
    },
    {
      "epoch": 2.3068235294117647,
      "grad_norm": 0.6128714683043236,
      "learning_rate": 0.0002492132197401474,
      "loss": 0.7969,
      "step": 4902
    },
    {
      "epoch": 2.307294117647059,
      "grad_norm": 0.5532343465523978,
      "learning_rate": 0.0002491803377389752,
      "loss": 0.4746,
      "step": 4903
    },
    {
      "epoch": 2.307764705882353,
      "grad_norm": 0.6165566766552083,
      "learning_rate": 0.0002491474472673157,
      "loss": 0.6172,
      "step": 4904
    },
    {
      "epoch": 2.308235294117647,
      "grad_norm": 0.6254676751749914,
      "learning_rate": 0.00024911454832797784,
      "loss": 0.6562,
      "step": 4905
    },
    {
      "epoch": 2.308705882352941,
      "grad_norm": 0.8970150970386876,
      "learning_rate": 0.0002490816409237714,
      "loss": 0.6133,
      "step": 4906
    },
    {
      "epoch": 2.3091764705882354,
      "grad_norm": 0.48841780836331006,
      "learning_rate": 0.0002490487250575069,
      "loss": 0.4199,
      "step": 4907
    },
    {
      "epoch": 2.309647058823529,
      "grad_norm": 0.3752623055540824,
      "learning_rate": 0.00024901580073199536,
      "loss": 0.1885,
      "step": 4908
    },
    {
      "epoch": 2.3101176470588234,
      "grad_norm": 0.2922147630385861,
      "learning_rate": 0.0002489828679500488,
      "loss": 0.104,
      "step": 4909
    },
    {
      "epoch": 2.3105882352941176,
      "grad_norm": 0.39758324787118166,
      "learning_rate": 0.00024894992671447983,
      "loss": 0.249,
      "step": 4910
    },
    {
      "epoch": 2.311058823529412,
      "grad_norm": 0.6476142372961317,
      "learning_rate": 0.0002489169770281018,
      "loss": 0.5469,
      "step": 4911
    },
    {
      "epoch": 2.311529411764706,
      "grad_norm": 0.49754355540670997,
      "learning_rate": 0.0002488840188937287,
      "loss": 0.459,
      "step": 4912
    },
    {
      "epoch": 2.312,
      "grad_norm": 0.4359074373143244,
      "learning_rate": 0.00024885105231417536,
      "loss": 0.4531,
      "step": 4913
    },
    {
      "epoch": 2.312470588235294,
      "grad_norm": 0.565080746942003,
      "learning_rate": 0.00024881807729225727,
      "loss": 0.4551,
      "step": 4914
    },
    {
      "epoch": 2.3129411764705883,
      "grad_norm": 0.4210034203634548,
      "learning_rate": 0.00024878509383079075,
      "loss": 0.5273,
      "step": 4915
    },
    {
      "epoch": 2.3134117647058825,
      "grad_norm": 0.5613551738716313,
      "learning_rate": 0.0002487521019325926,
      "loss": 0.5273,
      "step": 4916
    },
    {
      "epoch": 2.3138823529411763,
      "grad_norm": 0.4805377419239064,
      "learning_rate": 0.00024871910160048063,
      "loss": 0.4902,
      "step": 4917
    },
    {
      "epoch": 2.3143529411764705,
      "grad_norm": 0.42854932100686965,
      "learning_rate": 0.00024868609283727316,
      "loss": 0.5195,
      "step": 4918
    },
    {
      "epoch": 2.3148235294117647,
      "grad_norm": 0.5745212842489957,
      "learning_rate": 0.00024865307564578927,
      "loss": 0.4258,
      "step": 4919
    },
    {
      "epoch": 2.315294117647059,
      "grad_norm": 0.4719354732102307,
      "learning_rate": 0.00024862005002884886,
      "loss": 0.4453,
      "step": 4920
    },
    {
      "epoch": 2.315764705882353,
      "grad_norm": 0.6971884152395469,
      "learning_rate": 0.0002485870159892724,
      "loss": 0.9102,
      "step": 4921
    },
    {
      "epoch": 2.316235294117647,
      "grad_norm": 1.0456075178448927,
      "learning_rate": 0.00024855397352988127,
      "loss": 1.3047,
      "step": 4922
    },
    {
      "epoch": 2.316705882352941,
      "grad_norm": 0.5625465882874423,
      "learning_rate": 0.0002485209226534974,
      "loss": 0.6367,
      "step": 4923
    },
    {
      "epoch": 2.3171764705882354,
      "grad_norm": 0.4988121798855817,
      "learning_rate": 0.00024848786336294346,
      "loss": 0.6172,
      "step": 4924
    },
    {
      "epoch": 2.317647058823529,
      "grad_norm": 0.37951539505931575,
      "learning_rate": 0.0002484547956610429,
      "loss": 0.3828,
      "step": 4925
    },
    {
      "epoch": 2.3181176470588234,
      "grad_norm": 0.7895000775993251,
      "learning_rate": 0.0002484217195506199,
      "loss": 0.4102,
      "step": 4926
    },
    {
      "epoch": 2.3185882352941176,
      "grad_norm": 0.376752818837293,
      "learning_rate": 0.00024838863503449926,
      "loss": 0.25,
      "step": 4927
    },
    {
      "epoch": 2.319058823529412,
      "grad_norm": 0.5428687405434834,
      "learning_rate": 0.00024835554211550664,
      "loss": 0.5742,
      "step": 4928
    },
    {
      "epoch": 2.319529411764706,
      "grad_norm": 0.571284023861206,
      "learning_rate": 0.00024832244079646827,
      "loss": 0.6172,
      "step": 4929
    },
    {
      "epoch": 2.32,
      "grad_norm": 0.24230344490812927,
      "learning_rate": 0.00024828933108021116,
      "loss": 0.1006,
      "step": 4930
    },
    {
      "epoch": 2.320470588235294,
      "grad_norm": 0.5325453333872602,
      "learning_rate": 0.00024825621296956313,
      "loss": 0.2871,
      "step": 4931
    },
    {
      "epoch": 2.3209411764705883,
      "grad_norm": 0.4880349945253872,
      "learning_rate": 0.00024822308646735254,
      "loss": 0.5703,
      "step": 4932
    },
    {
      "epoch": 2.3214117647058825,
      "grad_norm": 0.21779966012224247,
      "learning_rate": 0.00024818995157640855,
      "loss": 0.2119,
      "step": 4933
    },
    {
      "epoch": 2.3218823529411763,
      "grad_norm": 0.772230560674521,
      "learning_rate": 0.00024815680829956115,
      "loss": 0.8477,
      "step": 4934
    },
    {
      "epoch": 2.3223529411764705,
      "grad_norm": 0.19109112690548888,
      "learning_rate": 0.0002481236566396408,
      "loss": 0.1973,
      "step": 4935
    },
    {
      "epoch": 2.3228235294117647,
      "grad_norm": 0.3560765572317951,
      "learning_rate": 0.0002480904965994789,
      "loss": 0.2314,
      "step": 4936
    },
    {
      "epoch": 2.323294117647059,
      "grad_norm": 0.4723426735088004,
      "learning_rate": 0.00024805732818190747,
      "loss": 0.4219,
      "step": 4937
    },
    {
      "epoch": 2.323764705882353,
      "grad_norm": 0.439140198113067,
      "learning_rate": 0.00024802415138975924,
      "loss": 0.3242,
      "step": 4938
    },
    {
      "epoch": 2.324235294117647,
      "grad_norm": 0.40991373749481397,
      "learning_rate": 0.00024799096622586767,
      "loss": 0.3262,
      "step": 4939
    },
    {
      "epoch": 2.324705882352941,
      "grad_norm": 0.5128722230739303,
      "learning_rate": 0.00024795777269306696,
      "loss": 0.5078,
      "step": 4940
    },
    {
      "epoch": 2.3251764705882354,
      "grad_norm": 0.38721354078142084,
      "learning_rate": 0.0002479245707941919,
      "loss": 0.4336,
      "step": 4941
    },
    {
      "epoch": 2.325647058823529,
      "grad_norm": 0.5917196575521565,
      "learning_rate": 0.0002478913605320783,
      "loss": 0.6875,
      "step": 4942
    },
    {
      "epoch": 2.3261176470588234,
      "grad_norm": 0.5252817281526558,
      "learning_rate": 0.0002478581419095623,
      "loss": 0.5859,
      "step": 4943
    },
    {
      "epoch": 2.3265882352941176,
      "grad_norm": 0.5647173536542485,
      "learning_rate": 0.0002478249149294809,
      "loss": 0.6914,
      "step": 4944
    },
    {
      "epoch": 2.327058823529412,
      "grad_norm": 0.5796370065094514,
      "learning_rate": 0.00024779167959467207,
      "loss": 0.7891,
      "step": 4945
    },
    {
      "epoch": 2.327529411764706,
      "grad_norm": 0.8031727508845895,
      "learning_rate": 0.0002477584359079741,
      "loss": 0.9258,
      "step": 4946
    },
    {
      "epoch": 2.328,
      "grad_norm": 0.4525457554595316,
      "learning_rate": 0.00024772518387222616,
      "loss": 0.3398,
      "step": 4947
    },
    {
      "epoch": 2.328470588235294,
      "grad_norm": 0.6489661300067717,
      "learning_rate": 0.0002476919234902682,
      "loss": 0.5,
      "step": 4948
    },
    {
      "epoch": 2.3289411764705883,
      "grad_norm": 0.4896539986873784,
      "learning_rate": 0.0002476586547649408,
      "loss": 0.4551,
      "step": 4949
    },
    {
      "epoch": 2.3294117647058825,
      "grad_norm": 0.748679954721286,
      "learning_rate": 0.00024762537769908535,
      "loss": 0.7188,
      "step": 4950
    },
    {
      "epoch": 2.3298823529411763,
      "grad_norm": 0.38481935337936624,
      "learning_rate": 0.0002475920922955437,
      "loss": 0.2314,
      "step": 4951
    },
    {
      "epoch": 2.3303529411764705,
      "grad_norm": 0.46208217155071124,
      "learning_rate": 0.0002475587985571587,
      "loss": 0.4727,
      "step": 4952
    },
    {
      "epoch": 2.3308235294117647,
      "grad_norm": 0.29969064292186326,
      "learning_rate": 0.0002475254964867738,
      "loss": 0.1118,
      "step": 4953
    },
    {
      "epoch": 2.331294117647059,
      "grad_norm": 0.48284035060933456,
      "learning_rate": 0.0002474921860872331,
      "loss": 0.4434,
      "step": 4954
    },
    {
      "epoch": 2.331764705882353,
      "grad_norm": 0.46776998902485073,
      "learning_rate": 0.00024745886736138154,
      "loss": 0.5703,
      "step": 4955
    },
    {
      "epoch": 2.332235294117647,
      "grad_norm": 0.5081555687004523,
      "learning_rate": 0.0002474255403120647,
      "loss": 0.6094,
      "step": 4956
    },
    {
      "epoch": 2.332705882352941,
      "grad_norm": 0.3752678653215363,
      "learning_rate": 0.0002473922049421288,
      "loss": 0.2178,
      "step": 4957
    },
    {
      "epoch": 2.3331764705882354,
      "grad_norm": 0.38158837687169445,
      "learning_rate": 0.0002473588612544209,
      "loss": 0.1953,
      "step": 4958
    },
    {
      "epoch": 2.333647058823529,
      "grad_norm": 0.21098717783348322,
      "learning_rate": 0.0002473255092517887,
      "loss": 0.1865,
      "step": 4959
    },
    {
      "epoch": 2.3341176470588234,
      "grad_norm": 0.6525740281528442,
      "learning_rate": 0.0002472921489370807,
      "loss": 0.5508,
      "step": 4960
    },
    {
      "epoch": 2.3345882352941176,
      "grad_norm": 0.3956621374248933,
      "learning_rate": 0.0002472587803131459,
      "loss": 0.248,
      "step": 4961
    },
    {
      "epoch": 2.335058823529412,
      "grad_norm": 0.4596495419180343,
      "learning_rate": 0.0002472254033828342,
      "loss": 0.5703,
      "step": 4962
    },
    {
      "epoch": 2.335529411764706,
      "grad_norm": 0.5445981722403163,
      "learning_rate": 0.00024719201814899617,
      "loss": 0.4902,
      "step": 4963
    },
    {
      "epoch": 2.336,
      "grad_norm": 0.7933547853218459,
      "learning_rate": 0.0002471586246144831,
      "loss": 0.6172,
      "step": 4964
    },
    {
      "epoch": 2.336470588235294,
      "grad_norm": 0.6235557431456362,
      "learning_rate": 0.0002471252227821469,
      "loss": 1.0703,
      "step": 4965
    },
    {
      "epoch": 2.3369411764705883,
      "grad_norm": 0.45865336713344984,
      "learning_rate": 0.0002470918126548403,
      "loss": 0.3809,
      "step": 4966
    },
    {
      "epoch": 2.3374117647058825,
      "grad_norm": 0.2691894779793019,
      "learning_rate": 0.00024705839423541664,
      "loss": 0.2578,
      "step": 4967
    },
    {
      "epoch": 2.3378823529411763,
      "grad_norm": 0.6247274071128284,
      "learning_rate": 0.00024702496752673005,
      "loss": 0.6758,
      "step": 4968
    },
    {
      "epoch": 2.3383529411764705,
      "grad_norm": 0.5872255453570652,
      "learning_rate": 0.00024699153253163533,
      "loss": 0.5312,
      "step": 4969
    },
    {
      "epoch": 2.3388235294117647,
      "grad_norm": 0.4398182707613349,
      "learning_rate": 0.000246958089252988,
      "loss": 0.4922,
      "step": 4970
    },
    {
      "epoch": 2.339294117647059,
      "grad_norm": 0.6710953478579095,
      "learning_rate": 0.0002469246376936443,
      "loss": 0.7656,
      "step": 4971
    },
    {
      "epoch": 2.339764705882353,
      "grad_norm": 0.6131246975152995,
      "learning_rate": 0.00024689117785646106,
      "loss": 0.7734,
      "step": 4972
    },
    {
      "epoch": 2.340235294117647,
      "grad_norm": 0.8869595834613958,
      "learning_rate": 0.000246857709744296,
      "loss": 0.7188,
      "step": 4973
    },
    {
      "epoch": 2.340705882352941,
      "grad_norm": 0.2795566483355374,
      "learning_rate": 0.00024682423336000746,
      "loss": 0.1836,
      "step": 4974
    },
    {
      "epoch": 2.3411764705882354,
      "grad_norm": 0.5439837878500853,
      "learning_rate": 0.00024679074870645446,
      "loss": 0.5273,
      "step": 4975
    },
    {
      "epoch": 2.341647058823529,
      "grad_norm": 0.4264493500639976,
      "learning_rate": 0.0002467572557864968,
      "loss": 0.2637,
      "step": 4976
    },
    {
      "epoch": 2.3421176470588234,
      "grad_norm": 0.4635012855206506,
      "learning_rate": 0.0002467237546029949,
      "loss": 0.3828,
      "step": 4977
    },
    {
      "epoch": 2.3425882352941176,
      "grad_norm": 0.49409658842300896,
      "learning_rate": 0.0002466902451588099,
      "loss": 0.3828,
      "step": 4978
    },
    {
      "epoch": 2.343058823529412,
      "grad_norm": 0.37049891653102973,
      "learning_rate": 0.0002466567274568037,
      "loss": 0.1807,
      "step": 4979
    },
    {
      "epoch": 2.343529411764706,
      "grad_norm": 0.5026983993024015,
      "learning_rate": 0.0002466232014998389,
      "loss": 0.5781,
      "step": 4980
    },
    {
      "epoch": 2.344,
      "grad_norm": 0.3488368648952396,
      "learning_rate": 0.00024658966729077875,
      "loss": 0.25,
      "step": 4981
    },
    {
      "epoch": 2.344470588235294,
      "grad_norm": 0.245088523202375,
      "learning_rate": 0.00024655612483248727,
      "loss": 0.105,
      "step": 4982
    },
    {
      "epoch": 2.3449411764705883,
      "grad_norm": 0.43622740394881404,
      "learning_rate": 0.00024652257412782904,
      "loss": 0.3848,
      "step": 4983
    },
    {
      "epoch": 2.3454117647058825,
      "grad_norm": 0.550206818731648,
      "learning_rate": 0.0002464890151796696,
      "loss": 0.8984,
      "step": 4984
    },
    {
      "epoch": 2.3458823529411763,
      "grad_norm": 0.3669679416475437,
      "learning_rate": 0.000246455447990875,
      "loss": 0.1934,
      "step": 4985
    },
    {
      "epoch": 2.3463529411764705,
      "grad_norm": 0.19511902275447093,
      "learning_rate": 0.000246421872564312,
      "loss": 0.2402,
      "step": 4986
    },
    {
      "epoch": 2.3468235294117648,
      "grad_norm": 0.45653171424706795,
      "learning_rate": 0.0002463882889028481,
      "loss": 0.5977,
      "step": 4987
    },
    {
      "epoch": 2.347294117647059,
      "grad_norm": 0.6561742239205719,
      "learning_rate": 0.0002463546970093516,
      "loss": 0.6641,
      "step": 4988
    },
    {
      "epoch": 2.3477647058823528,
      "grad_norm": 0.6188045758888574,
      "learning_rate": 0.00024632109688669135,
      "loss": 0.3945,
      "step": 4989
    },
    {
      "epoch": 2.348235294117647,
      "grad_norm": 0.21489625051261457,
      "learning_rate": 0.000246287488537737,
      "loss": 0.2539,
      "step": 4990
    },
    {
      "epoch": 2.348705882352941,
      "grad_norm": 0.6168516091689875,
      "learning_rate": 0.00024625387196535885,
      "loss": 0.6484,
      "step": 4991
    },
    {
      "epoch": 2.3491764705882354,
      "grad_norm": 0.4218355354578522,
      "learning_rate": 0.0002462202471724278,
      "loss": 0.209,
      "step": 4992
    },
    {
      "epoch": 2.349647058823529,
      "grad_norm": 0.5064419151746388,
      "learning_rate": 0.0002461866141618158,
      "loss": 0.3535,
      "step": 4993
    },
    {
      "epoch": 2.3501176470588234,
      "grad_norm": 0.5013353188199993,
      "learning_rate": 0.0002461529729363951,
      "loss": 0.4512,
      "step": 4994
    },
    {
      "epoch": 2.3505882352941176,
      "grad_norm": 0.6419670528642579,
      "learning_rate": 0.0002461193234990389,
      "loss": 0.5781,
      "step": 4995
    },
    {
      "epoch": 2.351058823529412,
      "grad_norm": 0.6377120284587791,
      "learning_rate": 0.00024608566585262103,
      "loss": 0.4805,
      "step": 4996
    },
    {
      "epoch": 2.351529411764706,
      "grad_norm": 5.62376726722049,
      "learning_rate": 0.000246052000000016,
      "loss": 1.0391,
      "step": 4997
    },
    {
      "epoch": 2.352,
      "grad_norm": 0.4942110755120728,
      "learning_rate": 0.00024601832594409905,
      "loss": 0.7227,
      "step": 4998
    },
    {
      "epoch": 2.352470588235294,
      "grad_norm": 0.48871126863966374,
      "learning_rate": 0.00024598464368774604,
      "loss": 0.6602,
      "step": 4999
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 0.3330525630047989,
      "learning_rate": 0.00024595095323383365,
      "loss": 0.165,
      "step": 5000
    },
    {
      "epoch": 2.3534117647058825,
      "grad_norm": 0.5339582292534168,
      "learning_rate": 0.0002459172545852393,
      "loss": 0.5586,
      "step": 5001
    },
    {
      "epoch": 2.3538823529411763,
      "grad_norm": 0.38695651026378075,
      "learning_rate": 0.0002458835477448408,
      "loss": 0.4297,
      "step": 5002
    },
    {
      "epoch": 2.3543529411764705,
      "grad_norm": 0.2979778727259982,
      "learning_rate": 0.00024584983271551717,
      "loss": 0.332,
      "step": 5003
    },
    {
      "epoch": 2.3548235294117648,
      "grad_norm": 0.3989676197337178,
      "learning_rate": 0.0002458161095001476,
      "loss": 0.2432,
      "step": 5004
    },
    {
      "epoch": 2.355294117647059,
      "grad_norm": 0.760386644317298,
      "learning_rate": 0.0002457823781016123,
      "loss": 0.418,
      "step": 5005
    },
    {
      "epoch": 2.3557647058823528,
      "grad_norm": 0.636868122986148,
      "learning_rate": 0.0002457486385227922,
      "loss": 0.3398,
      "step": 5006
    },
    {
      "epoch": 2.356235294117647,
      "grad_norm": 0.23357689672355586,
      "learning_rate": 0.00024571489076656865,
      "loss": 0.0991,
      "step": 5007
    },
    {
      "epoch": 2.356705882352941,
      "grad_norm": 2.47390013294144,
      "learning_rate": 0.0002456811348358239,
      "loss": 2.3906,
      "step": 5008
    },
    {
      "epoch": 2.3571764705882354,
      "grad_norm": 0.40286444750189787,
      "learning_rate": 0.000245647370733441,
      "loss": 0.2236,
      "step": 5009
    },
    {
      "epoch": 2.357647058823529,
      "grad_norm": 0.3596075106372916,
      "learning_rate": 0.0002456135984623034,
      "loss": 0.375,
      "step": 5010
    },
    {
      "epoch": 2.3581176470588234,
      "grad_norm": 0.5813649616687967,
      "learning_rate": 0.0002455798180252956,
      "loss": 0.4102,
      "step": 5011
    },
    {
      "epoch": 2.3585882352941177,
      "grad_norm": 0.6522489297970003,
      "learning_rate": 0.0002455460294253025,
      "loss": 0.8984,
      "step": 5012
    },
    {
      "epoch": 2.359058823529412,
      "grad_norm": 0.6901396887224759,
      "learning_rate": 0.00024551223266520983,
      "loss": 0.793,
      "step": 5013
    },
    {
      "epoch": 2.359529411764706,
      "grad_norm": 0.6327221908201955,
      "learning_rate": 0.00024547842774790403,
      "loss": 0.8359,
      "step": 5014
    },
    {
      "epoch": 2.36,
      "grad_norm": 0.3086451260599053,
      "learning_rate": 0.0002454446146762721,
      "loss": 0.3555,
      "step": 5015
    },
    {
      "epoch": 2.360470588235294,
      "grad_norm": 0.385260647488788,
      "learning_rate": 0.000245410793453202,
      "loss": 0.1689,
      "step": 5016
    },
    {
      "epoch": 2.3609411764705883,
      "grad_norm": 0.8700021767931543,
      "learning_rate": 0.00024537696408158215,
      "loss": 0.8047,
      "step": 5017
    },
    {
      "epoch": 2.3614117647058825,
      "grad_norm": 0.762160201012782,
      "learning_rate": 0.0002453431265643017,
      "loss": 0.8086,
      "step": 5018
    },
    {
      "epoch": 2.3618823529411763,
      "grad_norm": 0.35116342153156094,
      "learning_rate": 0.0002453092809042507,
      "loss": 0.1689,
      "step": 5019
    },
    {
      "epoch": 2.3623529411764705,
      "grad_norm": 0.5300369421576191,
      "learning_rate": 0.0002452754271043195,
      "loss": 0.582,
      "step": 5020
    },
    {
      "epoch": 2.3628235294117648,
      "grad_norm": 0.45958921563008465,
      "learning_rate": 0.0002452415651673995,
      "loss": 0.4082,
      "step": 5021
    },
    {
      "epoch": 2.363294117647059,
      "grad_norm": 0.43777884781058324,
      "learning_rate": 0.00024520769509638274,
      "loss": 0.4336,
      "step": 5022
    },
    {
      "epoch": 2.3637647058823528,
      "grad_norm": 0.4352612685451371,
      "learning_rate": 0.0002451738168941618,
      "loss": 0.2334,
      "step": 5023
    },
    {
      "epoch": 2.364235294117647,
      "grad_norm": 0.8637261549361441,
      "learning_rate": 0.0002451399305636301,
      "loss": 0.6602,
      "step": 5024
    },
    {
      "epoch": 2.364705882352941,
      "grad_norm": 0.390320615518816,
      "learning_rate": 0.00024510603610768166,
      "loss": 0.4121,
      "step": 5025
    },
    {
      "epoch": 2.3651764705882354,
      "grad_norm": 0.549537085935599,
      "learning_rate": 0.0002450721335292112,
      "loss": 0.5781,
      "step": 5026
    },
    {
      "epoch": 2.365647058823529,
      "grad_norm": 0.445773612649251,
      "learning_rate": 0.00024503822283111423,
      "loss": 0.4453,
      "step": 5027
    },
    {
      "epoch": 2.3661176470588234,
      "grad_norm": 0.3285971295740033,
      "learning_rate": 0.0002450043040162869,
      "loss": 0.2949,
      "step": 5028
    },
    {
      "epoch": 2.3665882352941177,
      "grad_norm": 0.5374218561305252,
      "learning_rate": 0.000244970377087626,
      "loss": 0.5273,
      "step": 5029
    },
    {
      "epoch": 2.367058823529412,
      "grad_norm": 0.5399873550270445,
      "learning_rate": 0.00024493644204802906,
      "loss": 0.4922,
      "step": 5030
    },
    {
      "epoch": 2.367529411764706,
      "grad_norm": 0.5041256247457896,
      "learning_rate": 0.0002449024989003943,
      "loss": 0.4219,
      "step": 5031
    },
    {
      "epoch": 2.368,
      "grad_norm": 1.178866809867554,
      "learning_rate": 0.0002448685476476207,
      "loss": 0.957,
      "step": 5032
    },
    {
      "epoch": 2.368470588235294,
      "grad_norm": 0.38621120465587905,
      "learning_rate": 0.0002448345882926077,
      "loss": 0.3691,
      "step": 5033
    },
    {
      "epoch": 2.3689411764705883,
      "grad_norm": 0.7427285231285022,
      "learning_rate": 0.0002448006208382558,
      "loss": 0.8789,
      "step": 5034
    },
    {
      "epoch": 2.369411764705882,
      "grad_norm": 0.4215072240861038,
      "learning_rate": 0.0002447666452874658,
      "loss": 0.4961,
      "step": 5035
    },
    {
      "epoch": 2.3698823529411763,
      "grad_norm": 0.3692151681101024,
      "learning_rate": 0.00024473266164313953,
      "loss": 0.1758,
      "step": 5036
    },
    {
      "epoch": 2.3703529411764706,
      "grad_norm": 0.7748266114669858,
      "learning_rate": 0.00024469866990817923,
      "loss": 0.7578,
      "step": 5037
    },
    {
      "epoch": 2.3708235294117648,
      "grad_norm": 0.44028966936190833,
      "learning_rate": 0.00024466467008548814,
      "loss": 0.3359,
      "step": 5038
    },
    {
      "epoch": 2.371294117647059,
      "grad_norm": 0.5513384521467735,
      "learning_rate": 0.0002446306621779698,
      "loss": 0.9062,
      "step": 5039
    },
    {
      "epoch": 2.3717647058823528,
      "grad_norm": 0.2982346782271909,
      "learning_rate": 0.00024459664618852877,
      "loss": 0.1445,
      "step": 5040
    },
    {
      "epoch": 2.372235294117647,
      "grad_norm": 0.38164144877481604,
      "learning_rate": 0.0002445626221200702,
      "loss": 0.334,
      "step": 5041
    },
    {
      "epoch": 2.372705882352941,
      "grad_norm": 0.40865424198292294,
      "learning_rate": 0.0002445285899754998,
      "loss": 0.4043,
      "step": 5042
    },
    {
      "epoch": 2.3731764705882354,
      "grad_norm": 0.22822774836049914,
      "learning_rate": 0.0002444945497577242,
      "loss": 0.1445,
      "step": 5043
    },
    {
      "epoch": 2.373647058823529,
      "grad_norm": 0.5684032729895108,
      "learning_rate": 0.0002444605014696506,
      "loss": 0.6055,
      "step": 5044
    },
    {
      "epoch": 2.3741176470588234,
      "grad_norm": 0.7226042589484735,
      "learning_rate": 0.0002444264451141868,
      "loss": 0.4844,
      "step": 5045
    },
    {
      "epoch": 2.3745882352941177,
      "grad_norm": 1.3332168067862185,
      "learning_rate": 0.0002443923806942414,
      "loss": 1.0859,
      "step": 5046
    },
    {
      "epoch": 2.375058823529412,
      "grad_norm": 0.554327586509812,
      "learning_rate": 0.0002443583082127237,
      "loss": 0.7031,
      "step": 5047
    },
    {
      "epoch": 2.375529411764706,
      "grad_norm": 0.6262245624858868,
      "learning_rate": 0.0002443242276725437,
      "loss": 0.7266,
      "step": 5048
    },
    {
      "epoch": 2.376,
      "grad_norm": 0.669271998276399,
      "learning_rate": 0.00024429013907661197,
      "loss": 0.8398,
      "step": 5049
    },
    {
      "epoch": 2.376470588235294,
      "grad_norm": 0.5547972176816807,
      "learning_rate": 0.0002442560424278399,
      "loss": 0.6094,
      "step": 5050
    },
    {
      "epoch": 2.3769411764705883,
      "grad_norm": 0.3989037778626395,
      "learning_rate": 0.0002442219377291395,
      "loss": 0.4688,
      "step": 5051
    },
    {
      "epoch": 2.377411764705882,
      "grad_norm": 0.5699441987959791,
      "learning_rate": 0.00024418782498342336,
      "loss": 0.6055,
      "step": 5052
    },
    {
      "epoch": 2.3778823529411763,
      "grad_norm": 0.6670428996124322,
      "learning_rate": 0.00024415370419360506,
      "loss": 0.6992,
      "step": 5053
    },
    {
      "epoch": 2.3783529411764706,
      "grad_norm": 0.4857881733304547,
      "learning_rate": 0.00024411957536259857,
      "loss": 0.4141,
      "step": 5054
    },
    {
      "epoch": 2.378823529411765,
      "grad_norm": 0.6678000087139206,
      "learning_rate": 0.0002440854384933187,
      "loss": 0.8789,
      "step": 5055
    },
    {
      "epoch": 2.379294117647059,
      "grad_norm": 0.5864084012390973,
      "learning_rate": 0.0002440512935886809,
      "loss": 0.4883,
      "step": 5056
    },
    {
      "epoch": 2.379764705882353,
      "grad_norm": 0.5406297889773919,
      "learning_rate": 0.0002440171406516013,
      "loss": 0.5703,
      "step": 5057
    },
    {
      "epoch": 2.380235294117647,
      "grad_norm": 0.3270618641379837,
      "learning_rate": 0.00024398297968499677,
      "loss": 0.1226,
      "step": 5058
    },
    {
      "epoch": 2.3807058823529412,
      "grad_norm": 0.5972982961590172,
      "learning_rate": 0.00024394881069178477,
      "loss": 0.5,
      "step": 5059
    },
    {
      "epoch": 2.3811764705882354,
      "grad_norm": 0.2775386509105466,
      "learning_rate": 0.00024391463367488348,
      "loss": 0.1436,
      "step": 5060
    },
    {
      "epoch": 2.3816470588235292,
      "grad_norm": 0.7896674498265138,
      "learning_rate": 0.00024388044863721188,
      "loss": 0.6484,
      "step": 5061
    },
    {
      "epoch": 2.3821176470588235,
      "grad_norm": 0.48164695222702314,
      "learning_rate": 0.00024384625558168948,
      "loss": 0.6289,
      "step": 5062
    },
    {
      "epoch": 2.3825882352941177,
      "grad_norm": 0.42088691072211415,
      "learning_rate": 0.0002438120545112365,
      "loss": 0.5312,
      "step": 5063
    },
    {
      "epoch": 2.383058823529412,
      "grad_norm": 0.6063243289049963,
      "learning_rate": 0.000243777845428774,
      "loss": 0.4336,
      "step": 5064
    },
    {
      "epoch": 2.383529411764706,
      "grad_norm": 0.442439014995722,
      "learning_rate": 0.00024374362833722344,
      "loss": 0.4277,
      "step": 5065
    },
    {
      "epoch": 2.384,
      "grad_norm": 0.6501934788269054,
      "learning_rate": 0.00024370940323950724,
      "loss": 0.7383,
      "step": 5066
    },
    {
      "epoch": 2.384470588235294,
      "grad_norm": 0.6322157126716044,
      "learning_rate": 0.0002436751701385484,
      "loss": 0.6445,
      "step": 5067
    },
    {
      "epoch": 2.3849411764705883,
      "grad_norm": 0.4852193237375101,
      "learning_rate": 0.0002436409290372705,
      "loss": 0.3379,
      "step": 5068
    },
    {
      "epoch": 2.385411764705882,
      "grad_norm": 0.8093277030453881,
      "learning_rate": 0.000243606679938598,
      "loss": 0.9688,
      "step": 5069
    },
    {
      "epoch": 2.3858823529411763,
      "grad_norm": 0.4592506631132281,
      "learning_rate": 0.00024357242284545587,
      "loss": 0.457,
      "step": 5070
    },
    {
      "epoch": 2.3863529411764706,
      "grad_norm": 0.2995281316880581,
      "learning_rate": 0.00024353815776076985,
      "loss": 0.2578,
      "step": 5071
    },
    {
      "epoch": 2.386823529411765,
      "grad_norm": 0.30676384469439527,
      "learning_rate": 0.00024350388468746635,
      "loss": 0.1309,
      "step": 5072
    },
    {
      "epoch": 2.387294117647059,
      "grad_norm": 0.6607365100507316,
      "learning_rate": 0.00024346960362847247,
      "loss": 0.9258,
      "step": 5073
    },
    {
      "epoch": 2.387764705882353,
      "grad_norm": 0.6092259329764445,
      "learning_rate": 0.000243435314586716,
      "loss": 0.7383,
      "step": 5074
    },
    {
      "epoch": 2.388235294117647,
      "grad_norm": 0.5967223653887869,
      "learning_rate": 0.00024340101756512532,
      "loss": 0.5898,
      "step": 5075
    },
    {
      "epoch": 2.3887058823529412,
      "grad_norm": 0.419617063498048,
      "learning_rate": 0.00024336671256662963,
      "loss": 0.2773,
      "step": 5076
    },
    {
      "epoch": 2.3891764705882355,
      "grad_norm": 0.14932736397783858,
      "learning_rate": 0.00024333239959415873,
      "loss": 0.0381,
      "step": 5077
    },
    {
      "epoch": 2.3896470588235292,
      "grad_norm": 0.24161807782710976,
      "learning_rate": 0.0002432980786506431,
      "loss": 0.1279,
      "step": 5078
    },
    {
      "epoch": 2.3901176470588235,
      "grad_norm": 0.5133835485067241,
      "learning_rate": 0.00024326374973901395,
      "loss": 0.4609,
      "step": 5079
    },
    {
      "epoch": 2.3905882352941177,
      "grad_norm": 0.5883545281912803,
      "learning_rate": 0.0002432294128622031,
      "loss": 0.5469,
      "step": 5080
    },
    {
      "epoch": 2.391058823529412,
      "grad_norm": 0.7459988328809268,
      "learning_rate": 0.0002431950680231431,
      "loss": 0.4922,
      "step": 5081
    },
    {
      "epoch": 2.391529411764706,
      "grad_norm": 0.31345830355180593,
      "learning_rate": 0.00024316071522476718,
      "loss": 0.2334,
      "step": 5082
    },
    {
      "epoch": 2.392,
      "grad_norm": 0.48140168051107146,
      "learning_rate": 0.0002431263544700093,
      "loss": 0.2617,
      "step": 5083
    },
    {
      "epoch": 2.392470588235294,
      "grad_norm": 0.7091135348162225,
      "learning_rate": 0.00024309198576180388,
      "loss": 0.9219,
      "step": 5084
    },
    {
      "epoch": 2.3929411764705883,
      "grad_norm": 0.5287364203484575,
      "learning_rate": 0.00024305760910308627,
      "loss": 0.6289,
      "step": 5085
    },
    {
      "epoch": 2.393411764705882,
      "grad_norm": 0.44934623960104975,
      "learning_rate": 0.00024302322449679245,
      "loss": 0.334,
      "step": 5086
    },
    {
      "epoch": 2.3938823529411764,
      "grad_norm": 0.6907613488906043,
      "learning_rate": 0.0002429888319458589,
      "loss": 0.7461,
      "step": 5087
    },
    {
      "epoch": 2.3943529411764706,
      "grad_norm": 0.49315421821656624,
      "learning_rate": 0.00024295443145322303,
      "loss": 0.5156,
      "step": 5088
    },
    {
      "epoch": 2.394823529411765,
      "grad_norm": 0.3090250143984182,
      "learning_rate": 0.0002429200230218228,
      "loss": 0.1123,
      "step": 5089
    },
    {
      "epoch": 2.395294117647059,
      "grad_norm": 0.6911496947568558,
      "learning_rate": 0.00024288560665459679,
      "loss": 0.707,
      "step": 5090
    },
    {
      "epoch": 2.395764705882353,
      "grad_norm": 0.3536864759947301,
      "learning_rate": 0.0002428511823544844,
      "loss": 0.2734,
      "step": 5091
    },
    {
      "epoch": 2.396235294117647,
      "grad_norm": 0.38860818919772516,
      "learning_rate": 0.00024281675012442563,
      "loss": 0.1719,
      "step": 5092
    },
    {
      "epoch": 2.3967058823529412,
      "grad_norm": 0.7326893474063206,
      "learning_rate": 0.00024278230996736109,
      "loss": 0.8555,
      "step": 5093
    },
    {
      "epoch": 2.3971764705882355,
      "grad_norm": 0.4763275332147563,
      "learning_rate": 0.0002427478618862322,
      "loss": 0.5586,
      "step": 5094
    },
    {
      "epoch": 2.3976470588235292,
      "grad_norm": 0.5084180090670908,
      "learning_rate": 0.00024271340588398097,
      "loss": 0.5391,
      "step": 5095
    },
    {
      "epoch": 2.3981176470588235,
      "grad_norm": 0.3836842664460454,
      "learning_rate": 0.00024267894196355015,
      "loss": 0.2275,
      "step": 5096
    },
    {
      "epoch": 2.3985882352941177,
      "grad_norm": 0.45153597107205035,
      "learning_rate": 0.00024264447012788308,
      "loss": 0.3711,
      "step": 5097
    },
    {
      "epoch": 2.399058823529412,
      "grad_norm": 0.4074319057810799,
      "learning_rate": 0.00024260999037992384,
      "loss": 0.3926,
      "step": 5098
    },
    {
      "epoch": 2.3995294117647057,
      "grad_norm": 0.4057745264161298,
      "learning_rate": 0.00024257550272261714,
      "loss": 0.4336,
      "step": 5099
    },
    {
      "epoch": 2.4,
      "grad_norm": 0.4640640977214595,
      "learning_rate": 0.00024254100715890846,
      "loss": 0.2314,
      "step": 5100
    },
    {
      "epoch": 2.400470588235294,
      "grad_norm": 0.8288437945695799,
      "learning_rate": 0.00024250650369174386,
      "loss": 0.8438,
      "step": 5101
    },
    {
      "epoch": 2.4009411764705884,
      "grad_norm": 0.4550644235659725,
      "learning_rate": 0.0002424719923240701,
      "loss": 0.252,
      "step": 5102
    },
    {
      "epoch": 2.401411764705882,
      "grad_norm": 0.28756313600150135,
      "learning_rate": 0.00024243747305883461,
      "loss": 0.125,
      "step": 5103
    },
    {
      "epoch": 2.4018823529411764,
      "grad_norm": 0.43696755307018964,
      "learning_rate": 0.00024240294589898554,
      "loss": 0.334,
      "step": 5104
    },
    {
      "epoch": 2.4023529411764706,
      "grad_norm": 0.5844840680913983,
      "learning_rate": 0.00024236841084747165,
      "loss": 0.6875,
      "step": 5105
    },
    {
      "epoch": 2.402823529411765,
      "grad_norm": 0.36606204094474337,
      "learning_rate": 0.00024233386790724242,
      "loss": 0.3379,
      "step": 5106
    },
    {
      "epoch": 2.403294117647059,
      "grad_norm": 0.582260512842493,
      "learning_rate": 0.000242299317081248,
      "loss": 0.5195,
      "step": 5107
    },
    {
      "epoch": 2.403764705882353,
      "grad_norm": 1.1968664453925728,
      "learning_rate": 0.00024226475837243908,
      "loss": 0.375,
      "step": 5108
    },
    {
      "epoch": 2.404235294117647,
      "grad_norm": 0.3594766495374509,
      "learning_rate": 0.00024223019178376732,
      "loss": 0.2109,
      "step": 5109
    },
    {
      "epoch": 2.4047058823529412,
      "grad_norm": 0.6019451793812234,
      "learning_rate": 0.00024219561731818475,
      "loss": 0.5,
      "step": 5110
    },
    {
      "epoch": 2.4051764705882355,
      "grad_norm": 0.5551194393812728,
      "learning_rate": 0.00024216103497864426,
      "loss": 0.6875,
      "step": 5111
    },
    {
      "epoch": 2.4056470588235292,
      "grad_norm": 0.40987104493695764,
      "learning_rate": 0.00024212644476809934,
      "loss": 0.3027,
      "step": 5112
    },
    {
      "epoch": 2.4061176470588235,
      "grad_norm": 0.24802829956407368,
      "learning_rate": 0.0002420918466895042,
      "loss": 0.0986,
      "step": 5113
    },
    {
      "epoch": 2.4065882352941177,
      "grad_norm": 0.5502916815159954,
      "learning_rate": 0.0002420572407458136,
      "loss": 0.4375,
      "step": 5114
    },
    {
      "epoch": 2.407058823529412,
      "grad_norm": 0.6374021704635094,
      "learning_rate": 0.0002420226269399831,
      "loss": 0.6914,
      "step": 5115
    },
    {
      "epoch": 2.4075294117647057,
      "grad_norm": 0.5323332469865636,
      "learning_rate": 0.00024198800527496889,
      "loss": 0.5039,
      "step": 5116
    },
    {
      "epoch": 2.408,
      "grad_norm": 0.5690400810698915,
      "learning_rate": 0.00024195337575372788,
      "loss": 0.6289,
      "step": 5117
    },
    {
      "epoch": 2.408470588235294,
      "grad_norm": 0.48158155290079235,
      "learning_rate": 0.00024191873837921748,
      "loss": 0.4609,
      "step": 5118
    },
    {
      "epoch": 2.4089411764705884,
      "grad_norm": 0.3372465704208769,
      "learning_rate": 0.000241884093154396,
      "loss": 0.1289,
      "step": 5119
    },
    {
      "epoch": 2.409411764705882,
      "grad_norm": 0.5937438590878024,
      "learning_rate": 0.00024184944008222227,
      "loss": 0.7461,
      "step": 5120
    },
    {
      "epoch": 2.4098823529411764,
      "grad_norm": 0.6805573929405101,
      "learning_rate": 0.00024181477916565582,
      "loss": 0.4336,
      "step": 5121
    },
    {
      "epoch": 2.4103529411764706,
      "grad_norm": 0.5521288012735619,
      "learning_rate": 0.0002417801104076569,
      "loss": 0.4062,
      "step": 5122
    },
    {
      "epoch": 2.410823529411765,
      "grad_norm": 0.32036517835434847,
      "learning_rate": 0.00024174543381118638,
      "loss": 0.1465,
      "step": 5123
    },
    {
      "epoch": 2.411294117647059,
      "grad_norm": 0.6071223134943511,
      "learning_rate": 0.00024171074937920579,
      "loss": 0.3809,
      "step": 5124
    },
    {
      "epoch": 2.411764705882353,
      "grad_norm": 0.4558709589940154,
      "learning_rate": 0.00024167605711467738,
      "loss": 0.3203,
      "step": 5125
    },
    {
      "epoch": 2.412235294117647,
      "grad_norm": 0.5377863072779808,
      "learning_rate": 0.000241641357020564,
      "loss": 0.2598,
      "step": 5126
    },
    {
      "epoch": 2.4127058823529413,
      "grad_norm": 0.6365798226256534,
      "learning_rate": 0.00024160664909982928,
      "loss": 0.5703,
      "step": 5127
    },
    {
      "epoch": 2.4131764705882355,
      "grad_norm": 0.3601711761313518,
      "learning_rate": 0.00024157193335543734,
      "loss": 0.2539,
      "step": 5128
    },
    {
      "epoch": 2.4136470588235293,
      "grad_norm": 0.5639282958816914,
      "learning_rate": 0.00024153720979035317,
      "loss": 0.418,
      "step": 5129
    },
    {
      "epoch": 2.4141176470588235,
      "grad_norm": 0.468386139297663,
      "learning_rate": 0.00024150247840754228,
      "loss": 0.4453,
      "step": 5130
    },
    {
      "epoch": 2.4145882352941177,
      "grad_norm": 0.4460588355159692,
      "learning_rate": 0.00024146773920997091,
      "loss": 0.3496,
      "step": 5131
    },
    {
      "epoch": 2.415058823529412,
      "grad_norm": 0.6222720980582133,
      "learning_rate": 0.000241432992200606,
      "loss": 0.8867,
      "step": 5132
    },
    {
      "epoch": 2.4155294117647057,
      "grad_norm": 0.8888741074487964,
      "learning_rate": 0.000241398237382415,
      "loss": 0.9414,
      "step": 5133
    },
    {
      "epoch": 2.416,
      "grad_norm": 0.5829218777282449,
      "learning_rate": 0.00024136347475836633,
      "loss": 0.5312,
      "step": 5134
    },
    {
      "epoch": 2.416470588235294,
      "grad_norm": 0.4172063879852054,
      "learning_rate": 0.0002413287043314287,
      "loss": 0.248,
      "step": 5135
    },
    {
      "epoch": 2.4169411764705884,
      "grad_norm": 0.28371511122457255,
      "learning_rate": 0.00024129392610457176,
      "loss": 0.1543,
      "step": 5136
    },
    {
      "epoch": 2.417411764705882,
      "grad_norm": 0.5551105179292902,
      "learning_rate": 0.0002412591400807658,
      "loss": 0.6133,
      "step": 5137
    },
    {
      "epoch": 2.4178823529411764,
      "grad_norm": 0.342433002203451,
      "learning_rate": 0.0002412243462629816,
      "loss": 0.1904,
      "step": 5138
    },
    {
      "epoch": 2.4183529411764706,
      "grad_norm": 0.5420515306690216,
      "learning_rate": 0.0002411895446541908,
      "loss": 0.4824,
      "step": 5139
    },
    {
      "epoch": 2.418823529411765,
      "grad_norm": 0.6729264704041911,
      "learning_rate": 0.00024115473525736561,
      "loss": 0.6797,
      "step": 5140
    },
    {
      "epoch": 2.419294117647059,
      "grad_norm": 0.3543413299109693,
      "learning_rate": 0.0002411199180754789,
      "loss": 0.4062,
      "step": 5141
    },
    {
      "epoch": 2.419764705882353,
      "grad_norm": 0.4732055644664826,
      "learning_rate": 0.00024108509311150422,
      "loss": 0.4531,
      "step": 5142
    },
    {
      "epoch": 2.420235294117647,
      "grad_norm": 0.46765215850071334,
      "learning_rate": 0.00024105026036841582,
      "loss": 0.5938,
      "step": 5143
    },
    {
      "epoch": 2.4207058823529413,
      "grad_norm": 0.33644119401600864,
      "learning_rate": 0.00024101541984918864,
      "loss": 0.1855,
      "step": 5144
    },
    {
      "epoch": 2.4211764705882355,
      "grad_norm": 0.3560242534026641,
      "learning_rate": 0.00024098057155679813,
      "loss": 0.21,
      "step": 5145
    },
    {
      "epoch": 2.4216470588235293,
      "grad_norm": 0.45702597720359617,
      "learning_rate": 0.0002409457154942206,
      "loss": 0.3945,
      "step": 5146
    },
    {
      "epoch": 2.4221176470588235,
      "grad_norm": 0.535822800963764,
      "learning_rate": 0.00024091085166443285,
      "loss": 0.5859,
      "step": 5147
    },
    {
      "epoch": 2.4225882352941177,
      "grad_norm": 0.42475985732091204,
      "learning_rate": 0.00024087598007041243,
      "loss": 0.3848,
      "step": 5148
    },
    {
      "epoch": 2.423058823529412,
      "grad_norm": 0.7789757941250844,
      "learning_rate": 0.0002408411007151376,
      "loss": 0.7539,
      "step": 5149
    },
    {
      "epoch": 2.4235294117647057,
      "grad_norm": 0.3496719287536856,
      "learning_rate": 0.00024080621360158717,
      "loss": 0.3574,
      "step": 5150
    },
    {
      "epoch": 2.424,
      "grad_norm": 0.3116930499853952,
      "learning_rate": 0.00024077131873274072,
      "loss": 0.3809,
      "step": 5151
    },
    {
      "epoch": 2.424470588235294,
      "grad_norm": 0.3559464375465677,
      "learning_rate": 0.0002407364161115784,
      "loss": 0.332,
      "step": 5152
    },
    {
      "epoch": 2.4249411764705884,
      "grad_norm": 0.4212957138238931,
      "learning_rate": 0.00024070150574108113,
      "loss": 0.4805,
      "step": 5153
    },
    {
      "epoch": 2.425411764705882,
      "grad_norm": 0.42986325097437983,
      "learning_rate": 0.00024066658762423034,
      "loss": 0.498,
      "step": 5154
    },
    {
      "epoch": 2.4258823529411764,
      "grad_norm": 0.7410311009156061,
      "learning_rate": 0.00024063166176400825,
      "loss": 1.0547,
      "step": 5155
    },
    {
      "epoch": 2.4263529411764706,
      "grad_norm": 0.5029200763831753,
      "learning_rate": 0.0002405967281633977,
      "loss": 0.4531,
      "step": 5156
    },
    {
      "epoch": 2.426823529411765,
      "grad_norm": 0.33526744230663363,
      "learning_rate": 0.00024056178682538222,
      "loss": 0.3652,
      "step": 5157
    },
    {
      "epoch": 2.427294117647059,
      "grad_norm": 0.544042114445247,
      "learning_rate": 0.00024052683775294592,
      "loss": 0.4805,
      "step": 5158
    },
    {
      "epoch": 2.427764705882353,
      "grad_norm": 0.5376228714812729,
      "learning_rate": 0.00024049188094907363,
      "loss": 0.6992,
      "step": 5159
    },
    {
      "epoch": 2.428235294117647,
      "grad_norm": 0.22005921621329952,
      "learning_rate": 0.00024045691641675086,
      "loss": 0.1089,
      "step": 5160
    },
    {
      "epoch": 2.4287058823529413,
      "grad_norm": 0.8543980395738469,
      "learning_rate": 0.00024042194415896376,
      "loss": 0.5547,
      "step": 5161
    },
    {
      "epoch": 2.429176470588235,
      "grad_norm": 0.5456514769703317,
      "learning_rate": 0.00024038696417869908,
      "loss": 0.3711,
      "step": 5162
    },
    {
      "epoch": 2.4296470588235293,
      "grad_norm": 0.7773388492264534,
      "learning_rate": 0.00024035197647894434,
      "loss": 0.793,
      "step": 5163
    },
    {
      "epoch": 2.4301176470588235,
      "grad_norm": 0.9881591253870129,
      "learning_rate": 0.00024031698106268763,
      "loss": 1.0547,
      "step": 5164
    },
    {
      "epoch": 2.4305882352941177,
      "grad_norm": 0.5531014620634345,
      "learning_rate": 0.0002402819779329177,
      "loss": 0.5273,
      "step": 5165
    },
    {
      "epoch": 2.431058823529412,
      "grad_norm": 0.8365193343258623,
      "learning_rate": 0.00024024696709262406,
      "loss": 0.6875,
      "step": 5166
    },
    {
      "epoch": 2.4315294117647057,
      "grad_norm": 0.6502241779746702,
      "learning_rate": 0.00024021194854479674,
      "loss": 0.6523,
      "step": 5167
    },
    {
      "epoch": 2.432,
      "grad_norm": 0.5608631445043079,
      "learning_rate": 0.00024017692229242655,
      "loss": 0.5781,
      "step": 5168
    },
    {
      "epoch": 2.432470588235294,
      "grad_norm": 0.5905514636167678,
      "learning_rate": 0.00024014188833850486,
      "loss": 0.5352,
      "step": 5169
    },
    {
      "epoch": 2.4329411764705884,
      "grad_norm": 0.4833892500273297,
      "learning_rate": 0.0002401068466860238,
      "loss": 0.6406,
      "step": 5170
    },
    {
      "epoch": 2.433411764705882,
      "grad_norm": 0.5164426184650883,
      "learning_rate": 0.00024007179733797604,
      "loss": 0.6523,
      "step": 5171
    },
    {
      "epoch": 2.4338823529411764,
      "grad_norm": 0.6599482590380157,
      "learning_rate": 0.00024003674029735497,
      "loss": 0.5273,
      "step": 5172
    },
    {
      "epoch": 2.4343529411764706,
      "grad_norm": 0.9044478467714402,
      "learning_rate": 0.00024000167556715468,
      "loss": 0.7422,
      "step": 5173
    },
    {
      "epoch": 2.434823529411765,
      "grad_norm": 0.5342350028320267,
      "learning_rate": 0.00023996660315036986,
      "loss": 0.5547,
      "step": 5174
    },
    {
      "epoch": 2.435294117647059,
      "grad_norm": 0.6110872128314631,
      "learning_rate": 0.00023993152304999582,
      "loss": 0.7109,
      "step": 5175
    },
    {
      "epoch": 2.435764705882353,
      "grad_norm": 0.2721038931062111,
      "learning_rate": 0.0002398964352690286,
      "loss": 0.2754,
      "step": 5176
    },
    {
      "epoch": 2.436235294117647,
      "grad_norm": 0.3718529643376319,
      "learning_rate": 0.00023986133981046485,
      "loss": 0.1924,
      "step": 5177
    },
    {
      "epoch": 2.4367058823529413,
      "grad_norm": 0.45215478279485927,
      "learning_rate": 0.00023982623667730195,
      "loss": 0.4629,
      "step": 5178
    },
    {
      "epoch": 2.437176470588235,
      "grad_norm": 0.5211828795024603,
      "learning_rate": 0.0002397911258725378,
      "loss": 0.6172,
      "step": 5179
    },
    {
      "epoch": 2.4376470588235293,
      "grad_norm": 0.35376170323169176,
      "learning_rate": 0.00023975600739917113,
      "loss": 0.3867,
      "step": 5180
    },
    {
      "epoch": 2.4381176470588235,
      "grad_norm": 0.3415232314597138,
      "learning_rate": 0.00023972088126020114,
      "loss": 0.2256,
      "step": 5181
    },
    {
      "epoch": 2.4385882352941177,
      "grad_norm": 0.6065362809034968,
      "learning_rate": 0.00023968574745862782,
      "loss": 0.543,
      "step": 5182
    },
    {
      "epoch": 2.439058823529412,
      "grad_norm": 0.6044248084936793,
      "learning_rate": 0.0002396506059974518,
      "loss": 0.8398,
      "step": 5183
    },
    {
      "epoch": 2.4395294117647057,
      "grad_norm": 0.49616986901598015,
      "learning_rate": 0.00023961545687967422,
      "loss": 0.4375,
      "step": 5184
    },
    {
      "epoch": 2.44,
      "grad_norm": 0.3678753531536614,
      "learning_rate": 0.0002395803001082971,
      "loss": 0.252,
      "step": 5185
    },
    {
      "epoch": 2.440470588235294,
      "grad_norm": 0.39694967651949875,
      "learning_rate": 0.00023954513568632296,
      "loss": 0.1396,
      "step": 5186
    },
    {
      "epoch": 2.4409411764705884,
      "grad_norm": 0.3913895369170229,
      "learning_rate": 0.000239509963616755,
      "loss": 0.3379,
      "step": 5187
    },
    {
      "epoch": 2.441411764705882,
      "grad_norm": 0.3416804569697358,
      "learning_rate": 0.00023947478390259712,
      "loss": 0.2061,
      "step": 5188
    },
    {
      "epoch": 2.4418823529411764,
      "grad_norm": 0.38267258613618566,
      "learning_rate": 0.00023943959654685383,
      "loss": 0.459,
      "step": 5189
    },
    {
      "epoch": 2.4423529411764706,
      "grad_norm": 0.3268593243250194,
      "learning_rate": 0.00023940440155253024,
      "loss": 0.1455,
      "step": 5190
    },
    {
      "epoch": 2.442823529411765,
      "grad_norm": 0.5686852831641184,
      "learning_rate": 0.0002393691989226323,
      "loss": 0.6836,
      "step": 5191
    },
    {
      "epoch": 2.443294117647059,
      "grad_norm": 0.5675121592029512,
      "learning_rate": 0.00023933398866016642,
      "loss": 0.4473,
      "step": 5192
    },
    {
      "epoch": 2.443764705882353,
      "grad_norm": 0.5344570766822185,
      "learning_rate": 0.0002392987707681397,
      "loss": 0.4316,
      "step": 5193
    },
    {
      "epoch": 2.444235294117647,
      "grad_norm": 0.30966027425598225,
      "learning_rate": 0.00023926354524955994,
      "loss": 0.4453,
      "step": 5194
    },
    {
      "epoch": 2.4447058823529413,
      "grad_norm": 0.4418882521557837,
      "learning_rate": 0.0002392283121074356,
      "loss": 0.4297,
      "step": 5195
    },
    {
      "epoch": 2.445176470588235,
      "grad_norm": 0.49852609778948,
      "learning_rate": 0.00023919307134477572,
      "loss": 0.668,
      "step": 5196
    },
    {
      "epoch": 2.4456470588235293,
      "grad_norm": 0.5792263731197076,
      "learning_rate": 0.00023915782296459007,
      "loss": 0.5469,
      "step": 5197
    },
    {
      "epoch": 2.4461176470588235,
      "grad_norm": 0.4182461604574684,
      "learning_rate": 0.00023912256696988904,
      "loss": 0.3477,
      "step": 5198
    },
    {
      "epoch": 2.4465882352941177,
      "grad_norm": 0.28749949374273714,
      "learning_rate": 0.00023908730336368364,
      "loss": 0.2061,
      "step": 5199
    },
    {
      "epoch": 2.447058823529412,
      "grad_norm": 0.3357900117919782,
      "learning_rate": 0.00023905203214898558,
      "loss": 0.1289,
      "step": 5200
    },
    {
      "epoch": 2.4475294117647057,
      "grad_norm": 0.5221146958942925,
      "learning_rate": 0.00023901675332880715,
      "loss": 0.3516,
      "step": 5201
    },
    {
      "epoch": 2.448,
      "grad_norm": 0.43569723279086736,
      "learning_rate": 0.00023898146690616134,
      "loss": 0.6641,
      "step": 5202
    },
    {
      "epoch": 2.448470588235294,
      "grad_norm": 0.4716734664694001,
      "learning_rate": 0.0002389461728840618,
      "loss": 0.5039,
      "step": 5203
    },
    {
      "epoch": 2.4489411764705884,
      "grad_norm": 0.5267513015515028,
      "learning_rate": 0.00023891087126552292,
      "loss": 0.5625,
      "step": 5204
    },
    {
      "epoch": 2.449411764705882,
      "grad_norm": 0.31920188796644844,
      "learning_rate": 0.00023887556205355945,
      "loss": 0.1562,
      "step": 5205
    },
    {
      "epoch": 2.4498823529411764,
      "grad_norm": 0.644392638795947,
      "learning_rate": 0.00023884024525118702,
      "loss": 0.7148,
      "step": 5206
    },
    {
      "epoch": 2.4503529411764706,
      "grad_norm": 0.440265603900035,
      "learning_rate": 0.00023880492086142193,
      "loss": 0.3867,
      "step": 5207
    },
    {
      "epoch": 2.450823529411765,
      "grad_norm": 0.7187282045481249,
      "learning_rate": 0.00023876958888728098,
      "loss": 0.8438,
      "step": 5208
    },
    {
      "epoch": 2.451294117647059,
      "grad_norm": 0.3876798574749979,
      "learning_rate": 0.00023873424933178178,
      "loss": 0.5352,
      "step": 5209
    },
    {
      "epoch": 2.451764705882353,
      "grad_norm": 0.5171739047255329,
      "learning_rate": 0.00023869890219794238,
      "loss": 0.5,
      "step": 5210
    },
    {
      "epoch": 2.452235294117647,
      "grad_norm": 0.29579840038482597,
      "learning_rate": 0.0002386635474887817,
      "loss": 0.166,
      "step": 5211
    },
    {
      "epoch": 2.4527058823529413,
      "grad_norm": 0.504354904067854,
      "learning_rate": 0.0002386281852073192,
      "loss": 0.2949,
      "step": 5212
    },
    {
      "epoch": 2.453176470588235,
      "grad_norm": 0.28363420171840464,
      "learning_rate": 0.00023859281535657496,
      "loss": 0.1836,
      "step": 5213
    },
    {
      "epoch": 2.4536470588235293,
      "grad_norm": 0.6176775042580959,
      "learning_rate": 0.00023855743793956967,
      "loss": 0.7422,
      "step": 5214
    },
    {
      "epoch": 2.4541176470588235,
      "grad_norm": 0.42615286805853764,
      "learning_rate": 0.0002385220529593249,
      "loss": 0.3828,
      "step": 5215
    },
    {
      "epoch": 2.4545882352941177,
      "grad_norm": 0.2895055077225304,
      "learning_rate": 0.00023848666041886258,
      "loss": 0.2363,
      "step": 5216
    },
    {
      "epoch": 2.455058823529412,
      "grad_norm": 0.3709748509110612,
      "learning_rate": 0.00023845126032120539,
      "loss": 0.3711,
      "step": 5217
    },
    {
      "epoch": 2.4555294117647057,
      "grad_norm": 0.35634921762578364,
      "learning_rate": 0.00023841585266937677,
      "loss": 0.4453,
      "step": 5218
    },
    {
      "epoch": 2.456,
      "grad_norm": 0.4442894478054506,
      "learning_rate": 0.00023838043746640065,
      "loss": 0.3711,
      "step": 5219
    },
    {
      "epoch": 2.456470588235294,
      "grad_norm": 0.7721437765527831,
      "learning_rate": 0.00023834501471530165,
      "loss": 0.7227,
      "step": 5220
    },
    {
      "epoch": 2.4569411764705884,
      "grad_norm": 0.3462502559556214,
      "learning_rate": 0.0002383095844191051,
      "loss": 0.2617,
      "step": 5221
    },
    {
      "epoch": 2.457411764705882,
      "grad_norm": 0.2558410331122316,
      "learning_rate": 0.0002382741465808368,
      "loss": 0.1079,
      "step": 5222
    },
    {
      "epoch": 2.4578823529411764,
      "grad_norm": 0.5238462054180684,
      "learning_rate": 0.0002382387012035235,
      "loss": 0.6328,
      "step": 5223
    },
    {
      "epoch": 2.4583529411764706,
      "grad_norm": 0.24810638344757466,
      "learning_rate": 0.0002382032482901922,
      "loss": 0.1016,
      "step": 5224
    },
    {
      "epoch": 2.458823529411765,
      "grad_norm": 0.28060272829143673,
      "learning_rate": 0.00023816778784387094,
      "loss": 0.2754,
      "step": 5225
    },
    {
      "epoch": 2.4592941176470586,
      "grad_norm": 0.6419742498999491,
      "learning_rate": 0.00023813231986758815,
      "loss": 0.4004,
      "step": 5226
    },
    {
      "epoch": 2.459764705882353,
      "grad_norm": 0.5629959909178418,
      "learning_rate": 0.00023809684436437294,
      "loss": 0.459,
      "step": 5227
    },
    {
      "epoch": 2.460235294117647,
      "grad_norm": 0.4059533077790589,
      "learning_rate": 0.00023806136133725506,
      "loss": 0.3809,
      "step": 5228
    },
    {
      "epoch": 2.4607058823529413,
      "grad_norm": 0.6285116221500937,
      "learning_rate": 0.00023802587078926505,
      "loss": 0.7383,
      "step": 5229
    },
    {
      "epoch": 2.461176470588235,
      "grad_norm": 0.17746663716693237,
      "learning_rate": 0.00023799037272343387,
      "loss": 0.0698,
      "step": 5230
    },
    {
      "epoch": 2.4616470588235293,
      "grad_norm": 0.5458433105574949,
      "learning_rate": 0.00023795486714279327,
      "loss": 0.3418,
      "step": 5231
    },
    {
      "epoch": 2.4621176470588235,
      "grad_norm": 0.5149965723107435,
      "learning_rate": 0.0002379193540503756,
      "loss": 0.3594,
      "step": 5232
    },
    {
      "epoch": 2.4625882352941177,
      "grad_norm": 0.6435347419423769,
      "learning_rate": 0.00023788383344921384,
      "loss": 0.3594,
      "step": 5233
    },
    {
      "epoch": 2.463058823529412,
      "grad_norm": 0.2910732997075613,
      "learning_rate": 0.0002378483053423417,
      "loss": 0.3105,
      "step": 5234
    },
    {
      "epoch": 2.4635294117647057,
      "grad_norm": 0.5313827918342208,
      "learning_rate": 0.00023781276973279335,
      "loss": 0.5078,
      "step": 5235
    },
    {
      "epoch": 2.464,
      "grad_norm": 0.789132381058848,
      "learning_rate": 0.0002377772266236037,
      "loss": 0.5312,
      "step": 5236
    },
    {
      "epoch": 2.464470588235294,
      "grad_norm": 0.4884204666213358,
      "learning_rate": 0.0002377416760178084,
      "loss": 0.5234,
      "step": 5237
    },
    {
      "epoch": 2.4649411764705884,
      "grad_norm": 0.35708427246769403,
      "learning_rate": 0.00023770611791844364,
      "loss": 0.2363,
      "step": 5238
    },
    {
      "epoch": 2.465411764705882,
      "grad_norm": 0.36956220607787704,
      "learning_rate": 0.00023767055232854615,
      "loss": 0.2734,
      "step": 5239
    },
    {
      "epoch": 2.4658823529411764,
      "grad_norm": 0.30480274410389024,
      "learning_rate": 0.00023763497925115347,
      "loss": 0.2217,
      "step": 5240
    },
    {
      "epoch": 2.4663529411764706,
      "grad_norm": 0.5391504090958406,
      "learning_rate": 0.00023759939868930373,
      "loss": 0.418,
      "step": 5241
    },
    {
      "epoch": 2.466823529411765,
      "grad_norm": 0.47565869073915473,
      "learning_rate": 0.00023756381064603564,
      "loss": 0.3496,
      "step": 5242
    },
    {
      "epoch": 2.4672941176470586,
      "grad_norm": 0.703954679813863,
      "learning_rate": 0.0002375282151243887,
      "loss": 1.0,
      "step": 5243
    },
    {
      "epoch": 2.467764705882353,
      "grad_norm": 0.659869019858893,
      "learning_rate": 0.00023749261212740278,
      "loss": 0.6211,
      "step": 5244
    },
    {
      "epoch": 2.468235294117647,
      "grad_norm": 0.4340726551833708,
      "learning_rate": 0.00023745700165811868,
      "loss": 0.4297,
      "step": 5245
    },
    {
      "epoch": 2.4687058823529413,
      "grad_norm": 0.5111821304810747,
      "learning_rate": 0.00023742138371957765,
      "loss": 0.8398,
      "step": 5246
    },
    {
      "epoch": 2.469176470588235,
      "grad_norm": 0.4590495851326233,
      "learning_rate": 0.00023738575831482164,
      "loss": 0.4453,
      "step": 5247
    },
    {
      "epoch": 2.4696470588235293,
      "grad_norm": 1.1692870291350153,
      "learning_rate": 0.0002373501254468933,
      "loss": 1.0312,
      "step": 5248
    },
    {
      "epoch": 2.4701176470588235,
      "grad_norm": 0.6926427152408384,
      "learning_rate": 0.00023731448511883575,
      "loss": 0.8906,
      "step": 5249
    },
    {
      "epoch": 2.4705882352941178,
      "grad_norm": 0.37457775450496933,
      "learning_rate": 0.00023727883733369292,
      "loss": 0.2559,
      "step": 5250
    },
    {
      "epoch": 2.471058823529412,
      "grad_norm": 0.35698749099725097,
      "learning_rate": 0.00023724318209450925,
      "loss": 0.2471,
      "step": 5251
    },
    {
      "epoch": 2.4715294117647058,
      "grad_norm": 0.5196440213768055,
      "learning_rate": 0.00023720751940432992,
      "loss": 0.4824,
      "step": 5252
    },
    {
      "epoch": 2.472,
      "grad_norm": 0.44182102680434054,
      "learning_rate": 0.0002371718492662007,
      "loss": 0.3965,
      "step": 5253
    },
    {
      "epoch": 2.472470588235294,
      "grad_norm": 0.5728988739486188,
      "learning_rate": 0.00023713617168316797,
      "loss": 0.5938,
      "step": 5254
    },
    {
      "epoch": 2.4729411764705884,
      "grad_norm": 0.4778508218500466,
      "learning_rate": 0.00023710048665827872,
      "loss": 0.6562,
      "step": 5255
    },
    {
      "epoch": 2.473411764705882,
      "grad_norm": 0.600923290109437,
      "learning_rate": 0.00023706479419458075,
      "loss": 0.8164,
      "step": 5256
    },
    {
      "epoch": 2.4738823529411764,
      "grad_norm": 0.6444123213608032,
      "learning_rate": 0.00023702909429512228,
      "loss": 0.7266,
      "step": 5257
    },
    {
      "epoch": 2.4743529411764706,
      "grad_norm": 0.5915403400118662,
      "learning_rate": 0.00023699338696295228,
      "loss": 0.7422,
      "step": 5258
    },
    {
      "epoch": 2.474823529411765,
      "grad_norm": 0.30780499951212137,
      "learning_rate": 0.00023695767220112033,
      "loss": 0.1533,
      "step": 5259
    },
    {
      "epoch": 2.4752941176470586,
      "grad_norm": 0.3559206084669439,
      "learning_rate": 0.0002369219500126767,
      "loss": 0.2949,
      "step": 5260
    },
    {
      "epoch": 2.475764705882353,
      "grad_norm": 0.40979792880257543,
      "learning_rate": 0.00023688622040067217,
      "loss": 0.3145,
      "step": 5261
    },
    {
      "epoch": 2.476235294117647,
      "grad_norm": 0.36974789505695593,
      "learning_rate": 0.0002368504833681582,
      "loss": 0.1924,
      "step": 5262
    },
    {
      "epoch": 2.4767058823529413,
      "grad_norm": 1.5880748570670302,
      "learning_rate": 0.00023681473891818699,
      "loss": 1.1016,
      "step": 5263
    },
    {
      "epoch": 2.477176470588235,
      "grad_norm": 0.5014425821785162,
      "learning_rate": 0.00023677898705381123,
      "loss": 0.3691,
      "step": 5264
    },
    {
      "epoch": 2.4776470588235293,
      "grad_norm": 0.7577010456582535,
      "learning_rate": 0.00023674322777808432,
      "loss": 0.6211,
      "step": 5265
    },
    {
      "epoch": 2.4781176470588235,
      "grad_norm": 0.6372813758255633,
      "learning_rate": 0.0002367074610940603,
      "loss": 0.4375,
      "step": 5266
    },
    {
      "epoch": 2.4785882352941178,
      "grad_norm": 0.41060726200901326,
      "learning_rate": 0.00023667168700479383,
      "loss": 0.2422,
      "step": 5267
    },
    {
      "epoch": 2.479058823529412,
      "grad_norm": 0.4267863770989624,
      "learning_rate": 0.0002366359055133401,
      "loss": 0.2656,
      "step": 5268
    },
    {
      "epoch": 2.4795294117647058,
      "grad_norm": 0.5144427676143948,
      "learning_rate": 0.00023660011662275516,
      "loss": 0.5195,
      "step": 5269
    },
    {
      "epoch": 2.48,
      "grad_norm": 0.3892700283898112,
      "learning_rate": 0.00023656432033609544,
      "loss": 0.3066,
      "step": 5270
    },
    {
      "epoch": 2.480470588235294,
      "grad_norm": 0.6360870382727757,
      "learning_rate": 0.0002365285166564182,
      "loss": 0.4863,
      "step": 5271
    },
    {
      "epoch": 2.4809411764705884,
      "grad_norm": 0.6452432811254672,
      "learning_rate": 0.00023649270558678123,
      "loss": 1.0781,
      "step": 5272
    },
    {
      "epoch": 2.481411764705882,
      "grad_norm": 0.54201273630318,
      "learning_rate": 0.00023645688713024295,
      "loss": 0.6445,
      "step": 5273
    },
    {
      "epoch": 2.4818823529411764,
      "grad_norm": 0.5901543935517208,
      "learning_rate": 0.0002364210612898624,
      "loss": 0.6836,
      "step": 5274
    },
    {
      "epoch": 2.4823529411764707,
      "grad_norm": 0.5996117252857484,
      "learning_rate": 0.00023638522806869935,
      "loss": 0.457,
      "step": 5275
    },
    {
      "epoch": 2.482823529411765,
      "grad_norm": 0.47902852085726566,
      "learning_rate": 0.0002363493874698141,
      "loss": 0.6836,
      "step": 5276
    },
    {
      "epoch": 2.4832941176470587,
      "grad_norm": 0.3618554584008877,
      "learning_rate": 0.0002363135394962676,
      "loss": 0.3125,
      "step": 5277
    },
    {
      "epoch": 2.483764705882353,
      "grad_norm": 0.7702016169721119,
      "learning_rate": 0.00023627768415112154,
      "loss": 0.8008,
      "step": 5278
    },
    {
      "epoch": 2.484235294117647,
      "grad_norm": 0.45620034572508084,
      "learning_rate": 0.00023624182143743797,
      "loss": 0.3574,
      "step": 5279
    },
    {
      "epoch": 2.4847058823529413,
      "grad_norm": 0.5637894544812907,
      "learning_rate": 0.00023620595135827985,
      "loss": 0.5273,
      "step": 5280
    },
    {
      "epoch": 2.485176470588235,
      "grad_norm": 0.5618411156977391,
      "learning_rate": 0.0002361700739167107,
      "loss": 0.5352,
      "step": 5281
    },
    {
      "epoch": 2.4856470588235293,
      "grad_norm": 0.36043670870175376,
      "learning_rate": 0.00023613418911579454,
      "loss": 0.249,
      "step": 5282
    },
    {
      "epoch": 2.4861176470588235,
      "grad_norm": 0.4929035089059198,
      "learning_rate": 0.00023609829695859618,
      "loss": 0.2852,
      "step": 5283
    },
    {
      "epoch": 2.4865882352941178,
      "grad_norm": 0.5778056376239905,
      "learning_rate": 0.00023606239744818094,
      "loss": 0.6914,
      "step": 5284
    },
    {
      "epoch": 2.487058823529412,
      "grad_norm": 0.4392621179759993,
      "learning_rate": 0.0002360264905876148,
      "loss": 0.5508,
      "step": 5285
    },
    {
      "epoch": 2.4875294117647058,
      "grad_norm": 0.4399866354646326,
      "learning_rate": 0.0002359905763799644,
      "loss": 0.3672,
      "step": 5286
    },
    {
      "epoch": 2.488,
      "grad_norm": 0.3416362431959906,
      "learning_rate": 0.00023595465482829697,
      "loss": 0.2637,
      "step": 5287
    },
    {
      "epoch": 2.488470588235294,
      "grad_norm": 0.5309669152208132,
      "learning_rate": 0.0002359187259356805,
      "loss": 0.5703,
      "step": 5288
    },
    {
      "epoch": 2.488941176470588,
      "grad_norm": 0.36696601207594914,
      "learning_rate": 0.00023588278970518333,
      "loss": 0.3301,
      "step": 5289
    },
    {
      "epoch": 2.489411764705882,
      "grad_norm": 0.5487159872621125,
      "learning_rate": 0.00023584684613987472,
      "loss": 0.625,
      "step": 5290
    },
    {
      "epoch": 2.4898823529411764,
      "grad_norm": 0.5331731528996672,
      "learning_rate": 0.00023581089524282432,
      "loss": 0.5195,
      "step": 5291
    },
    {
      "epoch": 2.4903529411764707,
      "grad_norm": 0.3430430108183226,
      "learning_rate": 0.0002357749370171026,
      "loss": 0.2578,
      "step": 5292
    },
    {
      "epoch": 2.490823529411765,
      "grad_norm": 0.4333417177235198,
      "learning_rate": 0.00023573897146578053,
      "loss": 0.4355,
      "step": 5293
    },
    {
      "epoch": 2.4912941176470587,
      "grad_norm": 0.4396007302587013,
      "learning_rate": 0.00023570299859192976,
      "loss": 0.4004,
      "step": 5294
    },
    {
      "epoch": 2.491764705882353,
      "grad_norm": 0.3512460158101692,
      "learning_rate": 0.00023566701839862253,
      "loss": 0.2334,
      "step": 5295
    },
    {
      "epoch": 2.492235294117647,
      "grad_norm": 0.4823635056439816,
      "learning_rate": 0.00023563103088893172,
      "loss": 0.5898,
      "step": 5296
    },
    {
      "epoch": 2.4927058823529413,
      "grad_norm": 0.5416841951244573,
      "learning_rate": 0.00023559503606593088,
      "loss": 0.5312,
      "step": 5297
    },
    {
      "epoch": 2.493176470588235,
      "grad_norm": 0.5137505436455556,
      "learning_rate": 0.0002355590339326941,
      "loss": 0.3887,
      "step": 5298
    },
    {
      "epoch": 2.4936470588235293,
      "grad_norm": 0.4360127909820285,
      "learning_rate": 0.00023552302449229612,
      "loss": 0.3184,
      "step": 5299
    },
    {
      "epoch": 2.4941176470588236,
      "grad_norm": 0.3759905004160892,
      "learning_rate": 0.00023548700774781242,
      "loss": 0.4004,
      "step": 5300
    },
    {
      "epoch": 2.4945882352941178,
      "grad_norm": 0.5006224087685626,
      "learning_rate": 0.00023545098370231893,
      "loss": 0.6172,
      "step": 5301
    },
    {
      "epoch": 2.495058823529412,
      "grad_norm": 0.35623478480815357,
      "learning_rate": 0.00023541495235889227,
      "loss": 0.2637,
      "step": 5302
    },
    {
      "epoch": 2.4955294117647058,
      "grad_norm": 0.5229004133897857,
      "learning_rate": 0.0002353789137206098,
      "loss": 0.582,
      "step": 5303
    },
    {
      "epoch": 2.496,
      "grad_norm": 0.3814802780898505,
      "learning_rate": 0.00023534286779054923,
      "loss": 0.2812,
      "step": 5304
    },
    {
      "epoch": 2.496470588235294,
      "grad_norm": 0.24909350128013513,
      "learning_rate": 0.00023530681457178916,
      "loss": 0.1816,
      "step": 5305
    },
    {
      "epoch": 2.496941176470588,
      "grad_norm": 0.28814054863035227,
      "learning_rate": 0.00023527075406740876,
      "loss": 0.1543,
      "step": 5306
    },
    {
      "epoch": 2.497411764705882,
      "grad_norm": 0.5897875936070586,
      "learning_rate": 0.00023523468628048764,
      "loss": 0.7227,
      "step": 5307
    },
    {
      "epoch": 2.4978823529411764,
      "grad_norm": 0.49954912334553037,
      "learning_rate": 0.0002351986112141063,
      "loss": 0.2891,
      "step": 5308
    },
    {
      "epoch": 2.4983529411764707,
      "grad_norm": 0.5009166290717678,
      "learning_rate": 0.00023516252887134558,
      "loss": 0.7617,
      "step": 5309
    },
    {
      "epoch": 2.498823529411765,
      "grad_norm": 0.16489242767303874,
      "learning_rate": 0.00023512643925528725,
      "loss": 0.0471,
      "step": 5310
    },
    {
      "epoch": 2.4992941176470587,
      "grad_norm": 0.3938196438582706,
      "learning_rate": 0.00023509034236901348,
      "loss": 0.3242,
      "step": 5311
    },
    {
      "epoch": 2.499764705882353,
      "grad_norm": 0.5054002779327917,
      "learning_rate": 0.00023505423821560705,
      "loss": 0.5977,
      "step": 5312
    },
    {
      "epoch": 2.500235294117647,
      "grad_norm": 0.48971590480316224,
      "learning_rate": 0.00023501812679815152,
      "loss": 0.3359,
      "step": 5313
    },
    {
      "epoch": 2.5007058823529413,
      "grad_norm": 0.5717813202250767,
      "learning_rate": 0.00023498200811973092,
      "loss": 0.5234,
      "step": 5314
    },
    {
      "epoch": 2.501176470588235,
      "grad_norm": 0.4162095371075988,
      "learning_rate": 0.00023494588218343003,
      "loss": 0.5273,
      "step": 5315
    },
    {
      "epoch": 2.5016470588235293,
      "grad_norm": 0.37183160773048424,
      "learning_rate": 0.0002349097489923341,
      "loss": 0.2031,
      "step": 5316
    },
    {
      "epoch": 2.5021176470588236,
      "grad_norm": 0.4284945429477692,
      "learning_rate": 0.00023487360854952915,
      "loss": 0.4316,
      "step": 5317
    },
    {
      "epoch": 2.502588235294118,
      "grad_norm": 0.5160343082406672,
      "learning_rate": 0.00023483746085810176,
      "loss": 0.5352,
      "step": 5318
    },
    {
      "epoch": 2.503058823529412,
      "grad_norm": 0.5343278296960139,
      "learning_rate": 0.00023480130592113904,
      "loss": 0.4336,
      "step": 5319
    },
    {
      "epoch": 2.503529411764706,
      "grad_norm": 0.6090038507494161,
      "learning_rate": 0.00023476514374172887,
      "loss": 0.4414,
      "step": 5320
    },
    {
      "epoch": 2.504,
      "grad_norm": 0.22847507170539424,
      "learning_rate": 0.0002347289743229596,
      "loss": 0.166,
      "step": 5321
    },
    {
      "epoch": 2.5044705882352942,
      "grad_norm": 0.38653446440345623,
      "learning_rate": 0.00023469279766792037,
      "loss": 0.5586,
      "step": 5322
    },
    {
      "epoch": 2.504941176470588,
      "grad_norm": 0.4278082023838277,
      "learning_rate": 0.0002346566137797008,
      "loss": 0.3672,
      "step": 5323
    },
    {
      "epoch": 2.5054117647058822,
      "grad_norm": 0.7721672112252203,
      "learning_rate": 0.0002346204226613912,
      "loss": 0.5508,
      "step": 5324
    },
    {
      "epoch": 2.5058823529411764,
      "grad_norm": 0.19261218650309028,
      "learning_rate": 0.0002345842243160824,
      "loss": 0.0552,
      "step": 5325
    },
    {
      "epoch": 2.5063529411764707,
      "grad_norm": 0.40278432611792253,
      "learning_rate": 0.00023454801874686596,
      "loss": 0.5195,
      "step": 5326
    },
    {
      "epoch": 2.506823529411765,
      "grad_norm": 0.40956669206850466,
      "learning_rate": 0.00023451180595683409,
      "loss": 0.457,
      "step": 5327
    },
    {
      "epoch": 2.5072941176470587,
      "grad_norm": 0.39280065779410894,
      "learning_rate": 0.0002344755859490794,
      "loss": 0.4043,
      "step": 5328
    },
    {
      "epoch": 2.507764705882353,
      "grad_norm": 0.653950650367909,
      "learning_rate": 0.00023443935872669534,
      "loss": 0.5117,
      "step": 5329
    },
    {
      "epoch": 2.508235294117647,
      "grad_norm": 0.5678502976798848,
      "learning_rate": 0.00023440312429277583,
      "loss": 0.8008,
      "step": 5330
    },
    {
      "epoch": 2.5087058823529413,
      "grad_norm": 0.49845589391709083,
      "learning_rate": 0.00023436688265041555,
      "loss": 0.291,
      "step": 5331
    },
    {
      "epoch": 2.509176470588235,
      "grad_norm": 0.5069974465693426,
      "learning_rate": 0.0002343306338027097,
      "loss": 0.7031,
      "step": 5332
    },
    {
      "epoch": 2.5096470588235293,
      "grad_norm": 0.4381069818432234,
      "learning_rate": 0.00023429437775275406,
      "loss": 0.5352,
      "step": 5333
    },
    {
      "epoch": 2.5101176470588236,
      "grad_norm": 0.38186543893795033,
      "learning_rate": 0.00023425811450364508,
      "loss": 0.1543,
      "step": 5334
    },
    {
      "epoch": 2.510588235294118,
      "grad_norm": 0.5655535325191441,
      "learning_rate": 0.00023422184405847985,
      "loss": 0.6094,
      "step": 5335
    },
    {
      "epoch": 2.511058823529412,
      "grad_norm": 0.5058026667880414,
      "learning_rate": 0.00023418556642035602,
      "loss": 0.4004,
      "step": 5336
    },
    {
      "epoch": 2.511529411764706,
      "grad_norm": 0.6576344853022666,
      "learning_rate": 0.0002341492815923719,
      "loss": 0.8242,
      "step": 5337
    },
    {
      "epoch": 2.512,
      "grad_norm": 0.3493725838502919,
      "learning_rate": 0.00023411298957762636,
      "loss": 0.3516,
      "step": 5338
    },
    {
      "epoch": 2.5124705882352942,
      "grad_norm": 0.4055712851404567,
      "learning_rate": 0.00023407669037921902,
      "loss": 0.4512,
      "step": 5339
    },
    {
      "epoch": 2.512941176470588,
      "grad_norm": 0.3736858349275942,
      "learning_rate": 0.00023404038400024984,
      "loss": 0.249,
      "step": 5340
    },
    {
      "epoch": 2.5134117647058822,
      "grad_norm": 0.452071503642508,
      "learning_rate": 0.00023400407044381965,
      "loss": 0.3418,
      "step": 5341
    },
    {
      "epoch": 2.5138823529411765,
      "grad_norm": 0.7726982710014744,
      "learning_rate": 0.00023396774971302985,
      "loss": 1.0156,
      "step": 5342
    },
    {
      "epoch": 2.5143529411764707,
      "grad_norm": 0.5413971479478211,
      "learning_rate": 0.00023393142181098235,
      "loss": 0.7852,
      "step": 5343
    },
    {
      "epoch": 2.514823529411765,
      "grad_norm": 0.49290234693855167,
      "learning_rate": 0.00023389508674077973,
      "loss": 0.5781,
      "step": 5344
    },
    {
      "epoch": 2.5152941176470587,
      "grad_norm": 0.6359503398451585,
      "learning_rate": 0.00023385874450552522,
      "loss": 0.5352,
      "step": 5345
    },
    {
      "epoch": 2.515764705882353,
      "grad_norm": 0.5678135724457478,
      "learning_rate": 0.00023382239510832262,
      "loss": 0.4785,
      "step": 5346
    },
    {
      "epoch": 2.516235294117647,
      "grad_norm": 0.5756175923342586,
      "learning_rate": 0.0002337860385522763,
      "loss": 0.75,
      "step": 5347
    },
    {
      "epoch": 2.516705882352941,
      "grad_norm": 0.5808883926754147,
      "learning_rate": 0.00023374967484049132,
      "loss": 0.625,
      "step": 5348
    },
    {
      "epoch": 2.517176470588235,
      "grad_norm": 0.3525775360858978,
      "learning_rate": 0.00023371330397607335,
      "loss": 0.4062,
      "step": 5349
    },
    {
      "epoch": 2.5176470588235293,
      "grad_norm": 0.5756454993886361,
      "learning_rate": 0.00023367692596212858,
      "loss": 0.7695,
      "step": 5350
    },
    {
      "epoch": 2.5181176470588236,
      "grad_norm": 0.41662109555156346,
      "learning_rate": 0.0002336405408017639,
      "loss": 0.3105,
      "step": 5351
    },
    {
      "epoch": 2.518588235294118,
      "grad_norm": 0.7155275173834992,
      "learning_rate": 0.0002336041484980868,
      "loss": 0.8047,
      "step": 5352
    },
    {
      "epoch": 2.519058823529412,
      "grad_norm": 0.47352239799292084,
      "learning_rate": 0.0002335677490542053,
      "loss": 0.582,
      "step": 5353
    },
    {
      "epoch": 2.519529411764706,
      "grad_norm": 0.4893499124023613,
      "learning_rate": 0.0002335313424732282,
      "loss": 0.6719,
      "step": 5354
    },
    {
      "epoch": 2.52,
      "grad_norm": 0.6126328797698543,
      "learning_rate": 0.0002334949287582647,
      "loss": 0.6133,
      "step": 5355
    },
    {
      "epoch": 2.5204705882352942,
      "grad_norm": 0.3175495727485476,
      "learning_rate": 0.00023345850791242477,
      "loss": 0.334,
      "step": 5356
    },
    {
      "epoch": 2.520941176470588,
      "grad_norm": 0.31407674670645563,
      "learning_rate": 0.00023342207993881883,
      "loss": 0.377,
      "step": 5357
    },
    {
      "epoch": 2.5214117647058822,
      "grad_norm": 0.4293027350777103,
      "learning_rate": 0.0002333856448405581,
      "loss": 0.3965,
      "step": 5358
    },
    {
      "epoch": 2.5218823529411765,
      "grad_norm": 0.33675526396809097,
      "learning_rate": 0.00023334920262075437,
      "loss": 0.2734,
      "step": 5359
    },
    {
      "epoch": 2.5223529411764707,
      "grad_norm": 0.40301577297350055,
      "learning_rate": 0.00023331275328251983,
      "loss": 0.1973,
      "step": 5360
    },
    {
      "epoch": 2.522823529411765,
      "grad_norm": 0.2720688820797423,
      "learning_rate": 0.00023327629682896758,
      "loss": 0.1523,
      "step": 5361
    },
    {
      "epoch": 2.5232941176470587,
      "grad_norm": 0.7256020542860574,
      "learning_rate": 0.00023323983326321106,
      "loss": 0.7305,
      "step": 5362
    },
    {
      "epoch": 2.523764705882353,
      "grad_norm": 0.5354143729518537,
      "learning_rate": 0.0002332033625883645,
      "loss": 0.5781,
      "step": 5363
    },
    {
      "epoch": 2.524235294117647,
      "grad_norm": 0.9038608979972643,
      "learning_rate": 0.00023316688480754267,
      "loss": 0.4844,
      "step": 5364
    },
    {
      "epoch": 2.524705882352941,
      "grad_norm": 0.7317052990454294,
      "learning_rate": 0.00023313039992386094,
      "loss": 0.5547,
      "step": 5365
    },
    {
      "epoch": 2.525176470588235,
      "grad_norm": 0.5572144714892493,
      "learning_rate": 0.00023309390794043532,
      "loss": 0.9258,
      "step": 5366
    },
    {
      "epoch": 2.5256470588235294,
      "grad_norm": 0.44668077941778206,
      "learning_rate": 0.00023305740886038235,
      "loss": 0.1855,
      "step": 5367
    },
    {
      "epoch": 2.5261176470588236,
      "grad_norm": 0.5810261771935437,
      "learning_rate": 0.00023302090268681932,
      "loss": 0.5,
      "step": 5368
    },
    {
      "epoch": 2.526588235294118,
      "grad_norm": 0.30551883021565895,
      "learning_rate": 0.00023298438942286395,
      "loss": 0.1895,
      "step": 5369
    },
    {
      "epoch": 2.527058823529412,
      "grad_norm": 0.6111230697231841,
      "learning_rate": 0.00023294786907163473,
      "loss": 0.6094,
      "step": 5370
    },
    {
      "epoch": 2.527529411764706,
      "grad_norm": 0.4433509226865134,
      "learning_rate": 0.0002329113416362506,
      "loss": 0.4336,
      "step": 5371
    },
    {
      "epoch": 2.528,
      "grad_norm": 0.4626130421939739,
      "learning_rate": 0.00023287480711983125,
      "loss": 0.5625,
      "step": 5372
    },
    {
      "epoch": 2.5284705882352942,
      "grad_norm": 0.4256333876574161,
      "learning_rate": 0.00023283826552549682,
      "loss": 0.3223,
      "step": 5373
    },
    {
      "epoch": 2.528941176470588,
      "grad_norm": 0.4395626569311707,
      "learning_rate": 0.00023280171685636822,
      "loss": 0.5234,
      "step": 5374
    },
    {
      "epoch": 2.5294117647058822,
      "grad_norm": 0.3188066895183788,
      "learning_rate": 0.0002327651611155669,
      "loss": 0.1426,
      "step": 5375
    },
    {
      "epoch": 2.5298823529411765,
      "grad_norm": 0.383895587466269,
      "learning_rate": 0.0002327285983062148,
      "loss": 0.4785,
      "step": 5376
    },
    {
      "epoch": 2.5303529411764707,
      "grad_norm": 0.38450172316129566,
      "learning_rate": 0.00023269202843143462,
      "loss": 0.6289,
      "step": 5377
    },
    {
      "epoch": 2.530823529411765,
      "grad_norm": 0.4945991016731509,
      "learning_rate": 0.00023265545149434968,
      "loss": 0.377,
      "step": 5378
    },
    {
      "epoch": 2.5312941176470587,
      "grad_norm": 0.5190338572874614,
      "learning_rate": 0.0002326188674980837,
      "loss": 0.5742,
      "step": 5379
    },
    {
      "epoch": 2.531764705882353,
      "grad_norm": 0.512209697561833,
      "learning_rate": 0.0002325822764457612,
      "loss": 0.3633,
      "step": 5380
    },
    {
      "epoch": 2.532235294117647,
      "grad_norm": 0.5184639325949999,
      "learning_rate": 0.0002325456783405072,
      "loss": 0.4727,
      "step": 5381
    },
    {
      "epoch": 2.532705882352941,
      "grad_norm": 0.41613207589094087,
      "learning_rate": 0.00023250907318544744,
      "loss": 0.377,
      "step": 5382
    },
    {
      "epoch": 2.533176470588235,
      "grad_norm": 0.5544411832552749,
      "learning_rate": 0.0002324724609837081,
      "loss": 0.4668,
      "step": 5383
    },
    {
      "epoch": 2.5336470588235294,
      "grad_norm": 0.5117331503836614,
      "learning_rate": 0.0002324358417384161,
      "loss": 0.4023,
      "step": 5384
    },
    {
      "epoch": 2.5341176470588236,
      "grad_norm": 0.31178493924097345,
      "learning_rate": 0.00023239921545269884,
      "loss": 0.293,
      "step": 5385
    },
    {
      "epoch": 2.534588235294118,
      "grad_norm": 0.5075941586672349,
      "learning_rate": 0.00023236258212968445,
      "loss": 0.416,
      "step": 5386
    },
    {
      "epoch": 2.535058823529412,
      "grad_norm": 0.47271309437609205,
      "learning_rate": 0.00023232594177250148,
      "loss": 0.5156,
      "step": 5387
    },
    {
      "epoch": 2.535529411764706,
      "grad_norm": 0.25999474375014653,
      "learning_rate": 0.00023228929438427932,
      "loss": 0.105,
      "step": 5388
    },
    {
      "epoch": 2.536,
      "grad_norm": 0.8057641004995917,
      "learning_rate": 0.0002322526399681478,
      "loss": 0.7305,
      "step": 5389
    },
    {
      "epoch": 2.5364705882352943,
      "grad_norm": 0.4671999088752133,
      "learning_rate": 0.0002322159785272374,
      "loss": 0.2246,
      "step": 5390
    },
    {
      "epoch": 2.536941176470588,
      "grad_norm": 0.2921875060354508,
      "learning_rate": 0.00023217931006467912,
      "loss": 0.2559,
      "step": 5391
    },
    {
      "epoch": 2.5374117647058823,
      "grad_norm": 0.15266571797482556,
      "learning_rate": 0.00023214263458360467,
      "loss": 0.0483,
      "step": 5392
    },
    {
      "epoch": 2.5378823529411765,
      "grad_norm": 0.6168401073263733,
      "learning_rate": 0.00023210595208714635,
      "loss": 0.5586,
      "step": 5393
    },
    {
      "epoch": 2.5383529411764707,
      "grad_norm": 0.3178616026670035,
      "learning_rate": 0.00023206926257843702,
      "loss": 0.2197,
      "step": 5394
    },
    {
      "epoch": 2.538823529411765,
      "grad_norm": 0.2602023201562123,
      "learning_rate": 0.00023203256606061008,
      "loss": 0.2295,
      "step": 5395
    },
    {
      "epoch": 2.5392941176470587,
      "grad_norm": 0.47484731132605457,
      "learning_rate": 0.00023199586253679964,
      "loss": 0.5195,
      "step": 5396
    },
    {
      "epoch": 2.539764705882353,
      "grad_norm": 0.4561435875804841,
      "learning_rate": 0.00023195915201014035,
      "loss": 0.4023,
      "step": 5397
    },
    {
      "epoch": 2.540235294117647,
      "grad_norm": 0.8244995425248492,
      "learning_rate": 0.00023192243448376745,
      "loss": 1.0078,
      "step": 5398
    },
    {
      "epoch": 2.540705882352941,
      "grad_norm": 0.450473567584631,
      "learning_rate": 0.00023188570996081687,
      "loss": 0.4121,
      "step": 5399
    },
    {
      "epoch": 2.541176470588235,
      "grad_norm": 0.40970331861751846,
      "learning_rate": 0.00023184897844442495,
      "loss": 0.5039,
      "step": 5400
    },
    {
      "epoch": 2.5416470588235294,
      "grad_norm": 0.3128844075117423,
      "learning_rate": 0.00023181223993772886,
      "loss": 0.1709,
      "step": 5401
    },
    {
      "epoch": 2.5421176470588236,
      "grad_norm": 0.3822804399618392,
      "learning_rate": 0.00023177549444386618,
      "loss": 0.25,
      "step": 5402
    },
    {
      "epoch": 2.542588235294118,
      "grad_norm": 0.38408189482198596,
      "learning_rate": 0.00023173874196597515,
      "loss": 0.373,
      "step": 5403
    },
    {
      "epoch": 2.543058823529412,
      "grad_norm": 0.5887227533099608,
      "learning_rate": 0.00023170198250719464,
      "loss": 0.5742,
      "step": 5404
    },
    {
      "epoch": 2.543529411764706,
      "grad_norm": 0.6605515792431261,
      "learning_rate": 0.00023166521607066409,
      "loss": 0.7461,
      "step": 5405
    },
    {
      "epoch": 2.544,
      "grad_norm": 0.4305941449080143,
      "learning_rate": 0.00023162844265952348,
      "loss": 0.1245,
      "step": 5406
    },
    {
      "epoch": 2.5444705882352943,
      "grad_norm": 0.37890060826020266,
      "learning_rate": 0.0002315916622769135,
      "loss": 0.2451,
      "step": 5407
    },
    {
      "epoch": 2.544941176470588,
      "grad_norm": 0.6653818511970446,
      "learning_rate": 0.00023155487492597536,
      "loss": 0.7383,
      "step": 5408
    },
    {
      "epoch": 2.5454117647058823,
      "grad_norm": 0.49649546652997373,
      "learning_rate": 0.0002315180806098509,
      "loss": 0.3242,
      "step": 5409
    },
    {
      "epoch": 2.5458823529411765,
      "grad_norm": 0.44114939342468784,
      "learning_rate": 0.00023148127933168245,
      "loss": 0.377,
      "step": 5410
    },
    {
      "epoch": 2.5463529411764707,
      "grad_norm": 0.6878611262276634,
      "learning_rate": 0.00023144447109461314,
      "loss": 0.6875,
      "step": 5411
    },
    {
      "epoch": 2.546823529411765,
      "grad_norm": 0.8218166592932025,
      "learning_rate": 0.0002314076559017865,
      "loss": 0.8906,
      "step": 5412
    },
    {
      "epoch": 2.5472941176470587,
      "grad_norm": 0.7890240970806901,
      "learning_rate": 0.00023137083375634676,
      "loss": 1.125,
      "step": 5413
    },
    {
      "epoch": 2.547764705882353,
      "grad_norm": 0.21151945211262377,
      "learning_rate": 0.00023133400466143867,
      "loss": 0.21,
      "step": 5414
    },
    {
      "epoch": 2.548235294117647,
      "grad_norm": 0.7244407658236937,
      "learning_rate": 0.00023129716862020766,
      "loss": 0.6172,
      "step": 5415
    },
    {
      "epoch": 2.548705882352941,
      "grad_norm": 0.39484081552284184,
      "learning_rate": 0.0002312603256357997,
      "loss": 0.2188,
      "step": 5416
    },
    {
      "epoch": 2.549176470588235,
      "grad_norm": 0.5441042817453974,
      "learning_rate": 0.00023122347571136127,
      "loss": 0.4766,
      "step": 5417
    },
    {
      "epoch": 2.5496470588235294,
      "grad_norm": 0.30678425214831206,
      "learning_rate": 0.00023118661885003969,
      "loss": 0.1514,
      "step": 5418
    },
    {
      "epoch": 2.5501176470588236,
      "grad_norm": 0.5187711065195829,
      "learning_rate": 0.00023114975505498264,
      "loss": 0.5,
      "step": 5419
    },
    {
      "epoch": 2.550588235294118,
      "grad_norm": 0.6417505105411327,
      "learning_rate": 0.00023111288432933847,
      "loss": 0.9961,
      "step": 5420
    },
    {
      "epoch": 2.5510588235294116,
      "grad_norm": 0.5099746498131066,
      "learning_rate": 0.0002310760066762561,
      "loss": 0.582,
      "step": 5421
    },
    {
      "epoch": 2.551529411764706,
      "grad_norm": 0.4998632833326201,
      "learning_rate": 0.00023103912209888512,
      "loss": 0.6172,
      "step": 5422
    },
    {
      "epoch": 2.552,
      "grad_norm": 0.3674739797080477,
      "learning_rate": 0.00023100223060037558,
      "loss": 0.2344,
      "step": 5423
    },
    {
      "epoch": 2.5524705882352943,
      "grad_norm": 0.5938940894135704,
      "learning_rate": 0.00023096533218387827,
      "loss": 0.6133,
      "step": 5424
    },
    {
      "epoch": 2.552941176470588,
      "grad_norm": 0.3340632150663209,
      "learning_rate": 0.00023092842685254442,
      "loss": 0.0884,
      "step": 5425
    },
    {
      "epoch": 2.5534117647058823,
      "grad_norm": 0.4318832574179834,
      "learning_rate": 0.000230891514609526,
      "loss": 0.2734,
      "step": 5426
    },
    {
      "epoch": 2.5538823529411765,
      "grad_norm": 0.4766464153126551,
      "learning_rate": 0.00023085459545797542,
      "loss": 0.2256,
      "step": 5427
    },
    {
      "epoch": 2.5543529411764707,
      "grad_norm": 0.3498255588538622,
      "learning_rate": 0.00023081766940104585,
      "loss": 0.3164,
      "step": 5428
    },
    {
      "epoch": 2.554823529411765,
      "grad_norm": 0.6424817878857937,
      "learning_rate": 0.00023078073644189085,
      "loss": 0.6875,
      "step": 5429
    },
    {
      "epoch": 2.5552941176470587,
      "grad_norm": 0.2651276175741101,
      "learning_rate": 0.00023074379658366478,
      "loss": 0.1367,
      "step": 5430
    },
    {
      "epoch": 2.555764705882353,
      "grad_norm": 0.38454187612050467,
      "learning_rate": 0.0002307068498295224,
      "loss": 0.3535,
      "step": 5431
    },
    {
      "epoch": 2.556235294117647,
      "grad_norm": 0.5418641559998786,
      "learning_rate": 0.00023066989618261919,
      "loss": 0.5078,
      "step": 5432
    },
    {
      "epoch": 2.556705882352941,
      "grad_norm": 0.4852298090980221,
      "learning_rate": 0.00023063293564611114,
      "loss": 0.3301,
      "step": 5433
    },
    {
      "epoch": 2.557176470588235,
      "grad_norm": 0.2936064710130091,
      "learning_rate": 0.00023059596822315491,
      "loss": 0.2344,
      "step": 5434
    },
    {
      "epoch": 2.5576470588235294,
      "grad_norm": 0.2845780365732703,
      "learning_rate": 0.00023055899391690766,
      "loss": 0.1592,
      "step": 5435
    },
    {
      "epoch": 2.5581176470588236,
      "grad_norm": 0.37063206694060424,
      "learning_rate": 0.00023052201273052723,
      "loss": 0.2871,
      "step": 5436
    },
    {
      "epoch": 2.558588235294118,
      "grad_norm": 0.5450040086826979,
      "learning_rate": 0.00023048502466717192,
      "loss": 0.2217,
      "step": 5437
    },
    {
      "epoch": 2.5590588235294116,
      "grad_norm": 0.5306505924515108,
      "learning_rate": 0.00023044802973000073,
      "loss": 0.6445,
      "step": 5438
    },
    {
      "epoch": 2.559529411764706,
      "grad_norm": 0.486026679035221,
      "learning_rate": 0.00023041102792217323,
      "loss": 0.4277,
      "step": 5439
    },
    {
      "epoch": 2.56,
      "grad_norm": 0.7350568874855147,
      "learning_rate": 0.00023037401924684946,
      "loss": 0.793,
      "step": 5440
    },
    {
      "epoch": 2.5604705882352943,
      "grad_norm": 0.4917822818109823,
      "learning_rate": 0.00023033700370719026,
      "loss": 0.373,
      "step": 5441
    },
    {
      "epoch": 2.560941176470588,
      "grad_norm": 0.4768596325040291,
      "learning_rate": 0.00023029998130635692,
      "loss": 0.3457,
      "step": 5442
    },
    {
      "epoch": 2.5614117647058823,
      "grad_norm": 0.6007821601982417,
      "learning_rate": 0.00023026295204751126,
      "loss": 0.8086,
      "step": 5443
    },
    {
      "epoch": 2.5618823529411765,
      "grad_norm": 0.5378293924334453,
      "learning_rate": 0.00023022591593381582,
      "loss": 0.498,
      "step": 5444
    },
    {
      "epoch": 2.5623529411764707,
      "grad_norm": 0.4376307738884531,
      "learning_rate": 0.0002301888729684337,
      "loss": 0.3203,
      "step": 5445
    },
    {
      "epoch": 2.562823529411765,
      "grad_norm": 0.27210781691523767,
      "learning_rate": 0.00023015182315452843,
      "loss": 0.3516,
      "step": 5446
    },
    {
      "epoch": 2.5632941176470587,
      "grad_norm": 0.7559989463126302,
      "learning_rate": 0.00023011476649526438,
      "loss": 0.6758,
      "step": 5447
    },
    {
      "epoch": 2.563764705882353,
      "grad_norm": 0.5815023165694269,
      "learning_rate": 0.0002300777029938063,
      "loss": 0.4688,
      "step": 5448
    },
    {
      "epoch": 2.564235294117647,
      "grad_norm": 0.5972371759486464,
      "learning_rate": 0.0002300406326533196,
      "loss": 0.6445,
      "step": 5449
    },
    {
      "epoch": 2.564705882352941,
      "grad_norm": 0.4180880820150069,
      "learning_rate": 0.00023000355547697027,
      "loss": 0.5391,
      "step": 5450
    },
    {
      "epoch": 2.565176470588235,
      "grad_norm": 0.46942606177242013,
      "learning_rate": 0.00022996647146792488,
      "loss": 0.2041,
      "step": 5451
    },
    {
      "epoch": 2.5656470588235294,
      "grad_norm": 0.5509800552403217,
      "learning_rate": 0.0002299293806293506,
      "loss": 0.291,
      "step": 5452
    },
    {
      "epoch": 2.5661176470588236,
      "grad_norm": 0.33043996878560533,
      "learning_rate": 0.00022989228296441515,
      "loss": 0.1924,
      "step": 5453
    },
    {
      "epoch": 2.566588235294118,
      "grad_norm": 0.466969830144488,
      "learning_rate": 0.00022985517847628686,
      "loss": 0.4785,
      "step": 5454
    },
    {
      "epoch": 2.5670588235294116,
      "grad_norm": 0.5186921651524544,
      "learning_rate": 0.0002298180671681347,
      "loss": 0.3574,
      "step": 5455
    },
    {
      "epoch": 2.567529411764706,
      "grad_norm": 0.6171668950021092,
      "learning_rate": 0.00022978094904312803,
      "loss": 0.5938,
      "step": 5456
    },
    {
      "epoch": 2.568,
      "grad_norm": 0.45308478339896713,
      "learning_rate": 0.000229743824104437,
      "loss": 0.3281,
      "step": 5457
    },
    {
      "epoch": 2.5684705882352943,
      "grad_norm": 0.3096501782151317,
      "learning_rate": 0.00022970669235523224,
      "loss": 0.1846,
      "step": 5458
    },
    {
      "epoch": 2.568941176470588,
      "grad_norm": 0.7075511540808915,
      "learning_rate": 0.00022966955379868502,
      "loss": 0.8047,
      "step": 5459
    },
    {
      "epoch": 2.5694117647058823,
      "grad_norm": 0.5514215300348002,
      "learning_rate": 0.00022963240843796714,
      "loss": 0.4902,
      "step": 5460
    },
    {
      "epoch": 2.5698823529411765,
      "grad_norm": 0.44778943850848485,
      "learning_rate": 0.000229595256276251,
      "loss": 0.3828,
      "step": 5461
    },
    {
      "epoch": 2.5703529411764707,
      "grad_norm": 0.4121420050321488,
      "learning_rate": 0.00022955809731670955,
      "loss": 0.2598,
      "step": 5462
    },
    {
      "epoch": 2.570823529411765,
      "grad_norm": 0.5104258454685339,
      "learning_rate": 0.00022952093156251635,
      "loss": 0.4004,
      "step": 5463
    },
    {
      "epoch": 2.5712941176470587,
      "grad_norm": 0.3190176043387539,
      "learning_rate": 0.0002294837590168455,
      "loss": 0.166,
      "step": 5464
    },
    {
      "epoch": 2.571764705882353,
      "grad_norm": 0.8451888657048999,
      "learning_rate": 0.00022944657968287184,
      "loss": 1.0781,
      "step": 5465
    },
    {
      "epoch": 2.572235294117647,
      "grad_norm": 0.5362683421667177,
      "learning_rate": 0.00022940939356377052,
      "loss": 0.4297,
      "step": 5466
    },
    {
      "epoch": 2.572705882352941,
      "grad_norm": 0.35849219616042,
      "learning_rate": 0.00022937220066271752,
      "loss": 0.1699,
      "step": 5467
    },
    {
      "epoch": 2.573176470588235,
      "grad_norm": 0.41325775847056906,
      "learning_rate": 0.0002293350009828893,
      "loss": 0.4141,
      "step": 5468
    },
    {
      "epoch": 2.5736470588235294,
      "grad_norm": 0.6564315797050955,
      "learning_rate": 0.00022929779452746283,
      "loss": 0.6172,
      "step": 5469
    },
    {
      "epoch": 2.5741176470588236,
      "grad_norm": 0.5217441081179996,
      "learning_rate": 0.0002292605812996157,
      "loss": 0.5156,
      "step": 5470
    },
    {
      "epoch": 2.574588235294118,
      "grad_norm": 0.4730458859866846,
      "learning_rate": 0.00022922336130252622,
      "loss": 0.4668,
      "step": 5471
    },
    {
      "epoch": 2.5750588235294116,
      "grad_norm": 0.6940170291145175,
      "learning_rate": 0.00022918613453937314,
      "loss": 0.7734,
      "step": 5472
    },
    {
      "epoch": 2.575529411764706,
      "grad_norm": 0.4983108096372724,
      "learning_rate": 0.00022914890101333573,
      "loss": 0.6211,
      "step": 5473
    },
    {
      "epoch": 2.576,
      "grad_norm": 0.5162711250340759,
      "learning_rate": 0.00022911166072759386,
      "loss": 0.2832,
      "step": 5474
    },
    {
      "epoch": 2.576470588235294,
      "grad_norm": 0.3366485038105727,
      "learning_rate": 0.00022907441368532822,
      "loss": 0.3398,
      "step": 5475
    },
    {
      "epoch": 2.576941176470588,
      "grad_norm": 0.28966048767928954,
      "learning_rate": 0.00022903715988971973,
      "loss": 0.2871,
      "step": 5476
    },
    {
      "epoch": 2.5774117647058823,
      "grad_norm": 0.5258476416000882,
      "learning_rate": 0.0002289998993439502,
      "loss": 0.3984,
      "step": 5477
    },
    {
      "epoch": 2.5778823529411765,
      "grad_norm": 0.4871484921766546,
      "learning_rate": 0.00022896263205120167,
      "loss": 0.4922,
      "step": 5478
    },
    {
      "epoch": 2.5783529411764707,
      "grad_norm": 0.6983897919584956,
      "learning_rate": 0.0002289253580146571,
      "loss": 0.668,
      "step": 5479
    },
    {
      "epoch": 2.578823529411765,
      "grad_norm": 0.48054975240432096,
      "learning_rate": 0.00022888807723749988,
      "loss": 0.4629,
      "step": 5480
    },
    {
      "epoch": 2.5792941176470587,
      "grad_norm": 0.24659486865077665,
      "learning_rate": 0.0002288507897229138,
      "loss": 0.1836,
      "step": 5481
    },
    {
      "epoch": 2.579764705882353,
      "grad_norm": 0.37494956856522627,
      "learning_rate": 0.00022881349547408363,
      "loss": 0.3457,
      "step": 5482
    },
    {
      "epoch": 2.580235294117647,
      "grad_norm": 0.5749000298016228,
      "learning_rate": 0.00022877619449419436,
      "loss": 0.5586,
      "step": 5483
    },
    {
      "epoch": 2.580705882352941,
      "grad_norm": 0.8438055552516057,
      "learning_rate": 0.00022873888678643164,
      "loss": 1.0,
      "step": 5484
    },
    {
      "epoch": 2.581176470588235,
      "grad_norm": 0.4213378845993771,
      "learning_rate": 0.0002287015723539818,
      "loss": 0.2598,
      "step": 5485
    },
    {
      "epoch": 2.5816470588235294,
      "grad_norm": 0.6591177092594707,
      "learning_rate": 0.00022866425120003165,
      "loss": 0.6836,
      "step": 5486
    },
    {
      "epoch": 2.5821176470588236,
      "grad_norm": 0.18741061720464874,
      "learning_rate": 0.0002286269233277686,
      "loss": 0.1216,
      "step": 5487
    },
    {
      "epoch": 2.582588235294118,
      "grad_norm": 0.47718127863314325,
      "learning_rate": 0.0002285895887403806,
      "loss": 0.5312,
      "step": 5488
    },
    {
      "epoch": 2.5830588235294116,
      "grad_norm": 0.48073081507011334,
      "learning_rate": 0.00022855224744105625,
      "loss": 0.2969,
      "step": 5489
    },
    {
      "epoch": 2.583529411764706,
      "grad_norm": 0.6023045215652062,
      "learning_rate": 0.00022851489943298472,
      "loss": 0.4297,
      "step": 5490
    },
    {
      "epoch": 2.584,
      "grad_norm": 0.7994185740922293,
      "learning_rate": 0.0002284775447193556,
      "loss": 0.875,
      "step": 5491
    },
    {
      "epoch": 2.584470588235294,
      "grad_norm": 0.7439963461057669,
      "learning_rate": 0.00022844018330335928,
      "loss": 0.7969,
      "step": 5492
    },
    {
      "epoch": 2.584941176470588,
      "grad_norm": 0.6001586114681773,
      "learning_rate": 0.00022840281518818657,
      "loss": 0.6914,
      "step": 5493
    },
    {
      "epoch": 2.5854117647058823,
      "grad_norm": 0.38577159948090584,
      "learning_rate": 0.00022836544037702883,
      "loss": 0.4453,
      "step": 5494
    },
    {
      "epoch": 2.5858823529411765,
      "grad_norm": 0.44525649527835925,
      "learning_rate": 0.00022832805887307815,
      "loss": 0.3457,
      "step": 5495
    },
    {
      "epoch": 2.5863529411764707,
      "grad_norm": 0.5449018605903478,
      "learning_rate": 0.00022829067067952702,
      "loss": 0.6328,
      "step": 5496
    },
    {
      "epoch": 2.586823529411765,
      "grad_norm": 0.387825329941419,
      "learning_rate": 0.00022825327579956863,
      "loss": 0.249,
      "step": 5497
    },
    {
      "epoch": 2.5872941176470587,
      "grad_norm": 0.5840179838583313,
      "learning_rate": 0.0002282158742363966,
      "loss": 0.7227,
      "step": 5498
    },
    {
      "epoch": 2.587764705882353,
      "grad_norm": 0.5156342657326729,
      "learning_rate": 0.00022817846599320528,
      "loss": 0.4395,
      "step": 5499
    },
    {
      "epoch": 2.588235294117647,
      "grad_norm": 0.494418325390938,
      "learning_rate": 0.00022814105107318952,
      "loss": 0.6016,
      "step": 5500
    },
    {
      "epoch": 2.588705882352941,
      "grad_norm": 0.549651112870386,
      "learning_rate": 0.00022810362947954475,
      "loss": 0.5391,
      "step": 5501
    },
    {
      "epoch": 2.589176470588235,
      "grad_norm": 0.37907343422452033,
      "learning_rate": 0.00022806620121546684,
      "loss": 0.4883,
      "step": 5502
    },
    {
      "epoch": 2.5896470588235294,
      "grad_norm": 0.5610340435143393,
      "learning_rate": 0.0002280287662841525,
      "loss": 0.6758,
      "step": 5503
    },
    {
      "epoch": 2.5901176470588236,
      "grad_norm": 0.32868769834635836,
      "learning_rate": 0.00022799132468879878,
      "loss": 0.3164,
      "step": 5504
    },
    {
      "epoch": 2.590588235294118,
      "grad_norm": 0.2803480211467101,
      "learning_rate": 0.00022795387643260336,
      "loss": 0.1514,
      "step": 5505
    },
    {
      "epoch": 2.5910588235294116,
      "grad_norm": 0.34190230300434965,
      "learning_rate": 0.00022791642151876455,
      "loss": 0.1836,
      "step": 5506
    },
    {
      "epoch": 2.591529411764706,
      "grad_norm": 0.5085250579098011,
      "learning_rate": 0.00022787895995048118,
      "loss": 0.6445,
      "step": 5507
    },
    {
      "epoch": 2.592,
      "grad_norm": 0.38530519641550925,
      "learning_rate": 0.00022784149173095262,
      "loss": 0.168,
      "step": 5508
    },
    {
      "epoch": 2.592470588235294,
      "grad_norm": 0.4801331448695324,
      "learning_rate": 0.00022780401686337885,
      "loss": 0.8516,
      "step": 5509
    },
    {
      "epoch": 2.592941176470588,
      "grad_norm": 0.6051940130494253,
      "learning_rate": 0.00022776653535096044,
      "loss": 0.9453,
      "step": 5510
    },
    {
      "epoch": 2.5934117647058823,
      "grad_norm": 0.3318455682167656,
      "learning_rate": 0.00022772904719689844,
      "loss": 0.2061,
      "step": 5511
    },
    {
      "epoch": 2.5938823529411765,
      "grad_norm": 0.49068277812806393,
      "learning_rate": 0.00022769155240439455,
      "loss": 0.4023,
      "step": 5512
    },
    {
      "epoch": 2.5943529411764708,
      "grad_norm": 0.31207597936645,
      "learning_rate": 0.00022765405097665103,
      "loss": 0.2871,
      "step": 5513
    },
    {
      "epoch": 2.594823529411765,
      "grad_norm": 0.347433006720433,
      "learning_rate": 0.00022761654291687065,
      "loss": 0.3672,
      "step": 5514
    },
    {
      "epoch": 2.5952941176470588,
      "grad_norm": 0.31688509698713024,
      "learning_rate": 0.00022757902822825683,
      "loss": 0.1543,
      "step": 5515
    },
    {
      "epoch": 2.595764705882353,
      "grad_norm": 0.3752779708024356,
      "learning_rate": 0.00022754150691401345,
      "loss": 0.3711,
      "step": 5516
    },
    {
      "epoch": 2.596235294117647,
      "grad_norm": 0.3960604197884332,
      "learning_rate": 0.00022750397897734504,
      "loss": 0.2129,
      "step": 5517
    },
    {
      "epoch": 2.596705882352941,
      "grad_norm": 0.5245133590329883,
      "learning_rate": 0.00022746644442145665,
      "loss": 0.4648,
      "step": 5518
    },
    {
      "epoch": 2.597176470588235,
      "grad_norm": 0.567713084454323,
      "learning_rate": 0.00022742890324955396,
      "loss": 0.5586,
      "step": 5519
    },
    {
      "epoch": 2.5976470588235294,
      "grad_norm": 0.5343217429377248,
      "learning_rate": 0.00022739135546484315,
      "loss": 0.293,
      "step": 5520
    },
    {
      "epoch": 2.5981176470588236,
      "grad_norm": 0.6208978250845663,
      "learning_rate": 0.00022735380107053094,
      "loss": 0.8008,
      "step": 5521
    },
    {
      "epoch": 2.598588235294118,
      "grad_norm": 0.5220387883876593,
      "learning_rate": 0.00022731624006982473,
      "loss": 0.6914,
      "step": 5522
    },
    {
      "epoch": 2.5990588235294116,
      "grad_norm": 0.4847235464944967,
      "learning_rate": 0.0002272786724659324,
      "loss": 0.3438,
      "step": 5523
    },
    {
      "epoch": 2.599529411764706,
      "grad_norm": 0.34901999067647416,
      "learning_rate": 0.00022724109826206235,
      "loss": 0.4004,
      "step": 5524
    },
    {
      "epoch": 2.6,
      "grad_norm": 0.3037566129927558,
      "learning_rate": 0.00022720351746142366,
      "loss": 0.252,
      "step": 5525
    },
    {
      "epoch": 2.600470588235294,
      "grad_norm": 0.6413521542633457,
      "learning_rate": 0.00022716593006722594,
      "loss": 0.4902,
      "step": 5526
    },
    {
      "epoch": 2.600941176470588,
      "grad_norm": 0.4269317100610131,
      "learning_rate": 0.0002271283360826793,
      "loss": 0.6602,
      "step": 5527
    },
    {
      "epoch": 2.6014117647058823,
      "grad_norm": 0.38102427487814094,
      "learning_rate": 0.0002270907355109944,
      "loss": 0.4102,
      "step": 5528
    },
    {
      "epoch": 2.6018823529411765,
      "grad_norm": 0.49933679797637326,
      "learning_rate": 0.00022705312835538257,
      "loss": 0.3672,
      "step": 5529
    },
    {
      "epoch": 2.6023529411764708,
      "grad_norm": 0.6651616064601547,
      "learning_rate": 0.00022701551461905563,
      "loss": 0.8281,
      "step": 5530
    },
    {
      "epoch": 2.602823529411765,
      "grad_norm": 0.5292474582563568,
      "learning_rate": 0.00022697789430522602,
      "loss": 0.5273,
      "step": 5531
    },
    {
      "epoch": 2.6032941176470588,
      "grad_norm": 0.3918636448567008,
      "learning_rate": 0.00022694026741710657,
      "loss": 0.3867,
      "step": 5532
    },
    {
      "epoch": 2.603764705882353,
      "grad_norm": 0.2779295529027058,
      "learning_rate": 0.00022690263395791095,
      "loss": 0.0825,
      "step": 5533
    },
    {
      "epoch": 2.604235294117647,
      "grad_norm": 0.4018426340813571,
      "learning_rate": 0.00022686499393085322,
      "loss": 0.2988,
      "step": 5534
    },
    {
      "epoch": 2.604705882352941,
      "grad_norm": 0.38358687550763154,
      "learning_rate": 0.00022682734733914792,
      "loss": 0.4102,
      "step": 5535
    },
    {
      "epoch": 2.605176470588235,
      "grad_norm": 0.5367452325647063,
      "learning_rate": 0.00022678969418601038,
      "loss": 0.5703,
      "step": 5536
    },
    {
      "epoch": 2.6056470588235294,
      "grad_norm": 0.7522602874974444,
      "learning_rate": 0.0002267520344746563,
      "loss": 0.8633,
      "step": 5537
    },
    {
      "epoch": 2.6061176470588236,
      "grad_norm": 0.5046905260735767,
      "learning_rate": 0.000226714368208302,
      "loss": 0.6289,
      "step": 5538
    },
    {
      "epoch": 2.606588235294118,
      "grad_norm": 0.3873534483158038,
      "learning_rate": 0.00022667669539016435,
      "loss": 0.3379,
      "step": 5539
    },
    {
      "epoch": 2.6070588235294117,
      "grad_norm": 0.23019881825526003,
      "learning_rate": 0.00022663901602346082,
      "loss": 0.3418,
      "step": 5540
    },
    {
      "epoch": 2.607529411764706,
      "grad_norm": 0.17470426720167911,
      "learning_rate": 0.00022660133011140942,
      "loss": 0.0435,
      "step": 5541
    },
    {
      "epoch": 2.608,
      "grad_norm": 0.4654437219308544,
      "learning_rate": 0.00022656363765722872,
      "loss": 0.3652,
      "step": 5542
    },
    {
      "epoch": 2.608470588235294,
      "grad_norm": 0.4287600249905735,
      "learning_rate": 0.00022652593866413775,
      "loss": 0.5703,
      "step": 5543
    },
    {
      "epoch": 2.608941176470588,
      "grad_norm": 0.38310244044705716,
      "learning_rate": 0.00022648823313535635,
      "loss": 0.3125,
      "step": 5544
    },
    {
      "epoch": 2.6094117647058823,
      "grad_norm": 0.5462043704996566,
      "learning_rate": 0.00022645052107410463,
      "loss": 0.6367,
      "step": 5545
    },
    {
      "epoch": 2.6098823529411765,
      "grad_norm": 0.4352490680264524,
      "learning_rate": 0.0002264128024836034,
      "loss": 0.4609,
      "step": 5546
    },
    {
      "epoch": 2.6103529411764708,
      "grad_norm": 0.46568423093680494,
      "learning_rate": 0.00022637507736707408,
      "loss": 0.4355,
      "step": 5547
    },
    {
      "epoch": 2.6108235294117645,
      "grad_norm": 0.497795638440595,
      "learning_rate": 0.00022633734572773852,
      "loss": 0.4004,
      "step": 5548
    },
    {
      "epoch": 2.6112941176470588,
      "grad_norm": 0.4952985072593159,
      "learning_rate": 0.00022629960756881916,
      "loss": 0.3145,
      "step": 5549
    },
    {
      "epoch": 2.611764705882353,
      "grad_norm": 0.45042464083879014,
      "learning_rate": 0.00022626186289353913,
      "loss": 0.3301,
      "step": 5550
    },
    {
      "epoch": 2.612235294117647,
      "grad_norm": 0.47261567476205435,
      "learning_rate": 0.00022622411170512193,
      "loss": 0.3516,
      "step": 5551
    },
    {
      "epoch": 2.612705882352941,
      "grad_norm": 0.7765957229724576,
      "learning_rate": 0.0002261863540067917,
      "loss": 0.8945,
      "step": 5552
    },
    {
      "epoch": 2.613176470588235,
      "grad_norm": 0.4367535075066605,
      "learning_rate": 0.00022614858980177314,
      "loss": 0.3086,
      "step": 5553
    },
    {
      "epoch": 2.6136470588235294,
      "grad_norm": 0.6399629944778391,
      "learning_rate": 0.00022611081909329153,
      "loss": 0.6953,
      "step": 5554
    },
    {
      "epoch": 2.6141176470588237,
      "grad_norm": 0.5970578639511452,
      "learning_rate": 0.00022607304188457264,
      "loss": 0.8516,
      "step": 5555
    },
    {
      "epoch": 2.614588235294118,
      "grad_norm": 0.6329080304445089,
      "learning_rate": 0.0002260352581788428,
      "loss": 0.4023,
      "step": 5556
    },
    {
      "epoch": 2.6150588235294117,
      "grad_norm": 0.31904159728574893,
      "learning_rate": 0.000225997467979329,
      "loss": 0.3242,
      "step": 5557
    },
    {
      "epoch": 2.615529411764706,
      "grad_norm": 0.3214748374013503,
      "learning_rate": 0.00022595967128925866,
      "loss": 0.2012,
      "step": 5558
    },
    {
      "epoch": 2.616,
      "grad_norm": 0.534393248620517,
      "learning_rate": 0.0002259218681118598,
      "loss": 0.6758,
      "step": 5559
    },
    {
      "epoch": 2.616470588235294,
      "grad_norm": 0.30761186888128833,
      "learning_rate": 0.000225884058450361,
      "loss": 0.3516,
      "step": 5560
    },
    {
      "epoch": 2.616941176470588,
      "grad_norm": 0.3906838207847682,
      "learning_rate": 0.00022584624230799144,
      "loss": 0.2773,
      "step": 5561
    },
    {
      "epoch": 2.6174117647058823,
      "grad_norm": 0.31112934555951144,
      "learning_rate": 0.00022580841968798072,
      "loss": 0.2871,
      "step": 5562
    },
    {
      "epoch": 2.6178823529411765,
      "grad_norm": 0.6984414714642695,
      "learning_rate": 0.00022577059059355914,
      "loss": 0.8125,
      "step": 5563
    },
    {
      "epoch": 2.6183529411764708,
      "grad_norm": 0.4411438378524668,
      "learning_rate": 0.00022573275502795747,
      "loss": 0.3672,
      "step": 5564
    },
    {
      "epoch": 2.6188235294117646,
      "grad_norm": 0.8500249327682686,
      "learning_rate": 0.00022569491299440704,
      "loss": 1.0781,
      "step": 5565
    },
    {
      "epoch": 2.6192941176470588,
      "grad_norm": 0.4554538736155196,
      "learning_rate": 0.00022565706449613978,
      "loss": 0.4121,
      "step": 5566
    },
    {
      "epoch": 2.619764705882353,
      "grad_norm": 0.29107788707013293,
      "learning_rate": 0.0002256192095363881,
      "loss": 0.334,
      "step": 5567
    },
    {
      "epoch": 2.620235294117647,
      "grad_norm": 0.6037484826128292,
      "learning_rate": 0.00022558134811838504,
      "loss": 0.8867,
      "step": 5568
    },
    {
      "epoch": 2.620705882352941,
      "grad_norm": 0.3594447336697031,
      "learning_rate": 0.00022554348024536413,
      "loss": 0.5508,
      "step": 5569
    },
    {
      "epoch": 2.621176470588235,
      "grad_norm": 0.47048149767807923,
      "learning_rate": 0.00022550560592055942,
      "loss": 0.5977,
      "step": 5570
    },
    {
      "epoch": 2.6216470588235294,
      "grad_norm": 0.22499867056964915,
      "learning_rate": 0.00022546772514720566,
      "loss": 0.1226,
      "step": 5571
    },
    {
      "epoch": 2.6221176470588237,
      "grad_norm": 0.3921660507469553,
      "learning_rate": 0.00022542983792853797,
      "loss": 0.3887,
      "step": 5572
    },
    {
      "epoch": 2.622588235294118,
      "grad_norm": 0.3304653629480917,
      "learning_rate": 0.00022539194426779216,
      "loss": 0.2314,
      "step": 5573
    },
    {
      "epoch": 2.6230588235294117,
      "grad_norm": 0.5668372862210067,
      "learning_rate": 0.00022535404416820453,
      "loss": 0.8672,
      "step": 5574
    },
    {
      "epoch": 2.623529411764706,
      "grad_norm": 0.5260626851682225,
      "learning_rate": 0.0002253161376330119,
      "loss": 0.4473,
      "step": 5575
    },
    {
      "epoch": 2.624,
      "grad_norm": 0.6712139849288861,
      "learning_rate": 0.00022527822466545164,
      "loss": 0.5664,
      "step": 5576
    },
    {
      "epoch": 2.624470588235294,
      "grad_norm": 0.5703893773722032,
      "learning_rate": 0.00022524030526876183,
      "loss": 0.4531,
      "step": 5577
    },
    {
      "epoch": 2.624941176470588,
      "grad_norm": 0.34246948577413416,
      "learning_rate": 0.0002252023794461809,
      "loss": 0.2441,
      "step": 5578
    },
    {
      "epoch": 2.6254117647058823,
      "grad_norm": 0.4047708814028661,
      "learning_rate": 0.00022516444720094785,
      "loss": 0.3047,
      "step": 5579
    },
    {
      "epoch": 2.6258823529411766,
      "grad_norm": 0.5016875454446045,
      "learning_rate": 0.00022512650853630234,
      "loss": 0.8477,
      "step": 5580
    },
    {
      "epoch": 2.626352941176471,
      "grad_norm": 0.41088331184288535,
      "learning_rate": 0.00022508856345548455,
      "loss": 0.3906,
      "step": 5581
    },
    {
      "epoch": 2.6268235294117646,
      "grad_norm": 0.43313037264275794,
      "learning_rate": 0.00022505061196173513,
      "loss": 0.4941,
      "step": 5582
    },
    {
      "epoch": 2.627294117647059,
      "grad_norm": 0.5709755257423578,
      "learning_rate": 0.00022501265405829526,
      "loss": 0.5664,
      "step": 5583
    },
    {
      "epoch": 2.627764705882353,
      "grad_norm": 0.47859889701088204,
      "learning_rate": 0.00022497468974840687,
      "loss": 0.3223,
      "step": 5584
    },
    {
      "epoch": 2.6282352941176472,
      "grad_norm": 0.4653720114367266,
      "learning_rate": 0.00022493671903531218,
      "loss": 0.7305,
      "step": 5585
    },
    {
      "epoch": 2.628705882352941,
      "grad_norm": 0.38133783966703805,
      "learning_rate": 0.00022489874192225414,
      "loss": 0.4043,
      "step": 5586
    },
    {
      "epoch": 2.6291764705882352,
      "grad_norm": 0.7002399481511986,
      "learning_rate": 0.00022486075841247617,
      "loss": 0.8359,
      "step": 5587
    },
    {
      "epoch": 2.6296470588235294,
      "grad_norm": 0.5184896478914827,
      "learning_rate": 0.0002248227685092222,
      "loss": 0.6172,
      "step": 5588
    },
    {
      "epoch": 2.6301176470588237,
      "grad_norm": 0.5268300709138554,
      "learning_rate": 0.00022478477221573687,
      "loss": 0.4668,
      "step": 5589
    },
    {
      "epoch": 2.630588235294118,
      "grad_norm": 0.08838557728839305,
      "learning_rate": 0.00022474676953526508,
      "loss": 0.0156,
      "step": 5590
    },
    {
      "epoch": 2.6310588235294117,
      "grad_norm": 0.4384782171251776,
      "learning_rate": 0.00022470876047105262,
      "loss": 0.4141,
      "step": 5591
    },
    {
      "epoch": 2.631529411764706,
      "grad_norm": 0.5069957224464414,
      "learning_rate": 0.00022467074502634558,
      "loss": 0.459,
      "step": 5592
    },
    {
      "epoch": 2.632,
      "grad_norm": 0.4737306545307117,
      "learning_rate": 0.00022463272320439063,
      "loss": 0.5117,
      "step": 5593
    },
    {
      "epoch": 2.632470588235294,
      "grad_norm": 0.7780648001592386,
      "learning_rate": 0.00022459469500843502,
      "loss": 1.1094,
      "step": 5594
    },
    {
      "epoch": 2.632941176470588,
      "grad_norm": 0.35081606112321184,
      "learning_rate": 0.0002245566604417266,
      "loss": 0.3633,
      "step": 5595
    },
    {
      "epoch": 2.6334117647058823,
      "grad_norm": 0.4491777091090566,
      "learning_rate": 0.0002245186195075137,
      "loss": 0.4863,
      "step": 5596
    },
    {
      "epoch": 2.6338823529411766,
      "grad_norm": 0.6336303281407815,
      "learning_rate": 0.0002244805722090451,
      "loss": 1.0312,
      "step": 5597
    },
    {
      "epoch": 2.634352941176471,
      "grad_norm": 0.3611799107589588,
      "learning_rate": 0.0002244425185495704,
      "loss": 0.3105,
      "step": 5598
    },
    {
      "epoch": 2.6348235294117646,
      "grad_norm": 0.3093350710467691,
      "learning_rate": 0.00022440445853233945,
      "loss": 0.3867,
      "step": 5599
    },
    {
      "epoch": 2.635294117647059,
      "grad_norm": 0.4313692779723592,
      "learning_rate": 0.00022436639216060275,
      "loss": 0.5,
      "step": 5600
    },
    {
      "epoch": 2.635764705882353,
      "grad_norm": 0.4490303979637152,
      "learning_rate": 0.00022432831943761146,
      "loss": 0.6016,
      "step": 5601
    },
    {
      "epoch": 2.636235294117647,
      "grad_norm": 0.4740883383423453,
      "learning_rate": 0.00022429024036661708,
      "loss": 0.6914,
      "step": 5602
    },
    {
      "epoch": 2.636705882352941,
      "grad_norm": 0.4210760150737795,
      "learning_rate": 0.00022425215495087177,
      "loss": 0.4883,
      "step": 5603
    },
    {
      "epoch": 2.6371764705882352,
      "grad_norm": 0.47643607407955174,
      "learning_rate": 0.00022421406319362823,
      "loss": 0.4375,
      "step": 5604
    },
    {
      "epoch": 2.6376470588235295,
      "grad_norm": 0.4487208011936612,
      "learning_rate": 0.00022417596509813968,
      "loss": 0.6445,
      "step": 5605
    },
    {
      "epoch": 2.6381176470588237,
      "grad_norm": 0.5779392099520014,
      "learning_rate": 0.0002241378606676599,
      "loss": 0.6992,
      "step": 5606
    },
    {
      "epoch": 2.638588235294118,
      "grad_norm": 0.4236828537769855,
      "learning_rate": 0.00022409974990544312,
      "loss": 0.5195,
      "step": 5607
    },
    {
      "epoch": 2.6390588235294117,
      "grad_norm": 0.5101579616411691,
      "learning_rate": 0.0002240616328147443,
      "loss": 0.4844,
      "step": 5608
    },
    {
      "epoch": 2.639529411764706,
      "grad_norm": 0.4913547079764609,
      "learning_rate": 0.00022402350939881873,
      "loss": 0.4727,
      "step": 5609
    },
    {
      "epoch": 2.64,
      "grad_norm": 0.35376426517621706,
      "learning_rate": 0.00022398537966092236,
      "loss": 0.2656,
      "step": 5610
    },
    {
      "epoch": 2.640470588235294,
      "grad_norm": 0.49924759464765583,
      "learning_rate": 0.00022394724360431167,
      "loss": 0.3984,
      "step": 5611
    },
    {
      "epoch": 2.640941176470588,
      "grad_norm": 0.5053797133368172,
      "learning_rate": 0.00022390910123224373,
      "loss": 0.7227,
      "step": 5612
    },
    {
      "epoch": 2.6414117647058823,
      "grad_norm": 0.30486832218025367,
      "learning_rate": 0.00022387095254797597,
      "loss": 0.3105,
      "step": 5613
    },
    {
      "epoch": 2.6418823529411766,
      "grad_norm": 0.2008046180100013,
      "learning_rate": 0.00022383279755476651,
      "loss": 0.2383,
      "step": 5614
    },
    {
      "epoch": 2.642352941176471,
      "grad_norm": 0.520213356601482,
      "learning_rate": 0.00022379463625587406,
      "loss": 0.6875,
      "step": 5615
    },
    {
      "epoch": 2.6428235294117646,
      "grad_norm": 0.4127767030284262,
      "learning_rate": 0.00022375646865455766,
      "loss": 0.2754,
      "step": 5616
    },
    {
      "epoch": 2.643294117647059,
      "grad_norm": 0.2663399853692618,
      "learning_rate": 0.00022371829475407713,
      "loss": 0.2021,
      "step": 5617
    },
    {
      "epoch": 2.643764705882353,
      "grad_norm": 0.24983584227290692,
      "learning_rate": 0.00022368011455769251,
      "loss": 0.1953,
      "step": 5618
    },
    {
      "epoch": 2.644235294117647,
      "grad_norm": 0.6670195475494834,
      "learning_rate": 0.00022364192806866481,
      "loss": 0.8672,
      "step": 5619
    },
    {
      "epoch": 2.644705882352941,
      "grad_norm": 0.46331247541337306,
      "learning_rate": 0.00022360373529025523,
      "loss": 0.498,
      "step": 5620
    },
    {
      "epoch": 2.6451764705882352,
      "grad_norm": 0.29060325946705373,
      "learning_rate": 0.00022356553622572564,
      "loss": 0.1592,
      "step": 5621
    },
    {
      "epoch": 2.6456470588235295,
      "grad_norm": 0.3341056095930405,
      "learning_rate": 0.00022352733087833838,
      "loss": 0.3086,
      "step": 5622
    },
    {
      "epoch": 2.6461176470588237,
      "grad_norm": 0.30710351603790126,
      "learning_rate": 0.00022348911925135646,
      "loss": 0.293,
      "step": 5623
    },
    {
      "epoch": 2.646588235294118,
      "grad_norm": 0.40165886179792604,
      "learning_rate": 0.00022345090134804325,
      "loss": 0.416,
      "step": 5624
    },
    {
      "epoch": 2.6470588235294117,
      "grad_norm": 0.44019564531938243,
      "learning_rate": 0.0002234126771716628,
      "loss": 0.4902,
      "step": 5625
    },
    {
      "epoch": 2.647529411764706,
      "grad_norm": 0.3415191423797517,
      "learning_rate": 0.00022337444672547967,
      "loss": 0.4238,
      "step": 5626
    },
    {
      "epoch": 2.648,
      "grad_norm": 0.37921797090196513,
      "learning_rate": 0.00022333621001275884,
      "loss": 0.2754,
      "step": 5627
    },
    {
      "epoch": 2.648470588235294,
      "grad_norm": 0.4744870381410847,
      "learning_rate": 0.00022329796703676597,
      "loss": 0.4922,
      "step": 5628
    },
    {
      "epoch": 2.648941176470588,
      "grad_norm": 0.44741489752137825,
      "learning_rate": 0.00022325971780076721,
      "loss": 0.5781,
      "step": 5629
    },
    {
      "epoch": 2.6494117647058824,
      "grad_norm": 0.8261754244965673,
      "learning_rate": 0.00022322146230802924,
      "loss": 0.9102,
      "step": 5630
    },
    {
      "epoch": 2.6498823529411766,
      "grad_norm": 0.38101902510737934,
      "learning_rate": 0.0002231832005618192,
      "loss": 0.5469,
      "step": 5631
    },
    {
      "epoch": 2.650352941176471,
      "grad_norm": 0.5093980556802485,
      "learning_rate": 0.00022314493256540485,
      "loss": 0.3184,
      "step": 5632
    },
    {
      "epoch": 2.6508235294117646,
      "grad_norm": 0.33518539388276913,
      "learning_rate": 0.0002231066583220545,
      "loss": 0.3105,
      "step": 5633
    },
    {
      "epoch": 2.651294117647059,
      "grad_norm": 0.3670151520199061,
      "learning_rate": 0.00022306837783503698,
      "loss": 0.3086,
      "step": 5634
    },
    {
      "epoch": 2.651764705882353,
      "grad_norm": 0.6291439966785194,
      "learning_rate": 0.00022303009110762154,
      "loss": 0.7656,
      "step": 5635
    },
    {
      "epoch": 2.652235294117647,
      "grad_norm": 0.4033092211791966,
      "learning_rate": 0.00022299179814307818,
      "loss": 0.4922,
      "step": 5636
    },
    {
      "epoch": 2.652705882352941,
      "grad_norm": 0.5437923460165474,
      "learning_rate": 0.0002229534989446772,
      "loss": 0.3477,
      "step": 5637
    },
    {
      "epoch": 2.6531764705882352,
      "grad_norm": 0.33366810002692143,
      "learning_rate": 0.00022291519351568955,
      "loss": 0.2617,
      "step": 5638
    },
    {
      "epoch": 2.6536470588235295,
      "grad_norm": 0.47331666284388213,
      "learning_rate": 0.00022287688185938673,
      "loss": 0.6055,
      "step": 5639
    },
    {
      "epoch": 2.6541176470588237,
      "grad_norm": 0.30257469982877866,
      "learning_rate": 0.00022283856397904077,
      "loss": 0.1758,
      "step": 5640
    },
    {
      "epoch": 2.654588235294118,
      "grad_norm": 0.617002417310379,
      "learning_rate": 0.00022280023987792416,
      "loss": 0.6953,
      "step": 5641
    },
    {
      "epoch": 2.6550588235294117,
      "grad_norm": 0.4744013063660423,
      "learning_rate": 0.00022276190955930998,
      "loss": 0.4316,
      "step": 5642
    },
    {
      "epoch": 2.655529411764706,
      "grad_norm": 0.30005471092495783,
      "learning_rate": 0.00022272357302647187,
      "loss": 0.1206,
      "step": 5643
    },
    {
      "epoch": 2.656,
      "grad_norm": 0.3910576501571263,
      "learning_rate": 0.00022268523028268385,
      "loss": 0.3809,
      "step": 5644
    },
    {
      "epoch": 2.656470588235294,
      "grad_norm": 0.32237595796379764,
      "learning_rate": 0.00022264688133122069,
      "loss": 0.2988,
      "step": 5645
    },
    {
      "epoch": 2.656941176470588,
      "grad_norm": 0.49481494950991967,
      "learning_rate": 0.00022260852617535752,
      "loss": 0.4062,
      "step": 5646
    },
    {
      "epoch": 2.6574117647058824,
      "grad_norm": 0.2636819269122374,
      "learning_rate": 0.00022257016481837006,
      "loss": 0.0806,
      "step": 5647
    },
    {
      "epoch": 2.6578823529411766,
      "grad_norm": 0.33171301227348,
      "learning_rate": 0.00022253179726353458,
      "loss": 0.2148,
      "step": 5648
    },
    {
      "epoch": 2.658352941176471,
      "grad_norm": 0.4319258192635268,
      "learning_rate": 0.00022249342351412783,
      "loss": 0.3906,
      "step": 5649
    },
    {
      "epoch": 2.6588235294117646,
      "grad_norm": 0.209818344280518,
      "learning_rate": 0.00022245504357342716,
      "loss": 0.0737,
      "step": 5650
    },
    {
      "epoch": 2.659294117647059,
      "grad_norm": 0.2797011154995776,
      "learning_rate": 0.00022241665744471037,
      "loss": 0.1064,
      "step": 5651
    },
    {
      "epoch": 2.659764705882353,
      "grad_norm": 0.475082861095803,
      "learning_rate": 0.00022237826513125584,
      "loss": 0.6172,
      "step": 5652
    },
    {
      "epoch": 2.660235294117647,
      "grad_norm": 0.10463559148003747,
      "learning_rate": 0.00022233986663634241,
      "loss": 0.0306,
      "step": 5653
    },
    {
      "epoch": 2.660705882352941,
      "grad_norm": 0.4954562560945669,
      "learning_rate": 0.0002223014619632496,
      "loss": 0.5195,
      "step": 5654
    },
    {
      "epoch": 2.6611764705882353,
      "grad_norm": 0.38497543508660026,
      "learning_rate": 0.00022226305111525726,
      "loss": 0.4375,
      "step": 5655
    },
    {
      "epoch": 2.6616470588235295,
      "grad_norm": 0.46722733796982513,
      "learning_rate": 0.00022222463409564595,
      "loss": 0.2754,
      "step": 5656
    },
    {
      "epoch": 2.6621176470588237,
      "grad_norm": 0.3707647807604953,
      "learning_rate": 0.0002221862109076966,
      "loss": 0.4062,
      "step": 5657
    },
    {
      "epoch": 2.662588235294118,
      "grad_norm": 0.20102885723941427,
      "learning_rate": 0.00022214778155469077,
      "loss": 0.1582,
      "step": 5658
    },
    {
      "epoch": 2.6630588235294117,
      "grad_norm": 0.6328861209660277,
      "learning_rate": 0.0002221093460399105,
      "loss": 0.7266,
      "step": 5659
    },
    {
      "epoch": 2.663529411764706,
      "grad_norm": 0.3246832593582326,
      "learning_rate": 0.0002220709043666384,
      "loss": 0.3516,
      "step": 5660
    },
    {
      "epoch": 2.664,
      "grad_norm": 0.659412783451354,
      "learning_rate": 0.0002220324565381576,
      "loss": 0.8828,
      "step": 5661
    },
    {
      "epoch": 2.664470588235294,
      "grad_norm": 0.666837226147973,
      "learning_rate": 0.0002219940025577516,
      "loss": 0.543,
      "step": 5662
    },
    {
      "epoch": 2.664941176470588,
      "grad_norm": 0.4877337847441209,
      "learning_rate": 0.00022195554242870473,
      "loss": 0.4277,
      "step": 5663
    },
    {
      "epoch": 2.6654117647058824,
      "grad_norm": 0.3636826680248081,
      "learning_rate": 0.0002219170761543016,
      "loss": 0.3281,
      "step": 5664
    },
    {
      "epoch": 2.6658823529411766,
      "grad_norm": 0.5072843190538464,
      "learning_rate": 0.00022187860373782737,
      "loss": 0.4102,
      "step": 5665
    },
    {
      "epoch": 2.666352941176471,
      "grad_norm": 0.543307326864823,
      "learning_rate": 0.00022184012518256785,
      "loss": 0.668,
      "step": 5666
    },
    {
      "epoch": 2.6668235294117646,
      "grad_norm": 0.38636342146736424,
      "learning_rate": 0.00022180164049180926,
      "loss": 0.2812,
      "step": 5667
    },
    {
      "epoch": 2.667294117647059,
      "grad_norm": 0.670704480457243,
      "learning_rate": 0.00022176314966883837,
      "loss": 0.3066,
      "step": 5668
    },
    {
      "epoch": 2.667764705882353,
      "grad_norm": 0.5322533124260773,
      "learning_rate": 0.0002217246527169425,
      "loss": 0.5391,
      "step": 5669
    },
    {
      "epoch": 2.668235294117647,
      "grad_norm": 1.1004353803181741,
      "learning_rate": 0.00022168614963940952,
      "loss": 1.5,
      "step": 5670
    },
    {
      "epoch": 2.668705882352941,
      "grad_norm": 0.3468203494360885,
      "learning_rate": 0.00022164764043952772,
      "loss": 0.375,
      "step": 5671
    },
    {
      "epoch": 2.6691764705882353,
      "grad_norm": 0.4120164041477013,
      "learning_rate": 0.00022160912512058598,
      "loss": 0.4082,
      "step": 5672
    },
    {
      "epoch": 2.6696470588235295,
      "grad_norm": 0.4043848357564468,
      "learning_rate": 0.0002215706036858737,
      "loss": 0.4668,
      "step": 5673
    },
    {
      "epoch": 2.6701176470588237,
      "grad_norm": 0.7148970973988342,
      "learning_rate": 0.00022153207613868087,
      "loss": 0.8711,
      "step": 5674
    },
    {
      "epoch": 2.6705882352941175,
      "grad_norm": 0.5234527074039532,
      "learning_rate": 0.00022149354248229784,
      "loss": 0.6992,
      "step": 5675
    },
    {
      "epoch": 2.6710588235294117,
      "grad_norm": 0.5330774188849307,
      "learning_rate": 0.00022145500272001558,
      "loss": 0.4961,
      "step": 5676
    },
    {
      "epoch": 2.671529411764706,
      "grad_norm": 0.6966611686031892,
      "learning_rate": 0.00022141645685512564,
      "loss": 0.8516,
      "step": 5677
    },
    {
      "epoch": 2.672,
      "grad_norm": 0.5047469133175535,
      "learning_rate": 0.00022137790489091999,
      "loss": 0.5156,
      "step": 5678
    },
    {
      "epoch": 2.672470588235294,
      "grad_norm": 0.8977937199153992,
      "learning_rate": 0.00022133934683069115,
      "loss": 0.9492,
      "step": 5679
    },
    {
      "epoch": 2.672941176470588,
      "grad_norm": 0.27850437343633766,
      "learning_rate": 0.00022130078267773218,
      "loss": 0.1196,
      "step": 5680
    },
    {
      "epoch": 2.6734117647058824,
      "grad_norm": 0.3490324377461639,
      "learning_rate": 0.00022126221243533664,
      "loss": 0.1953,
      "step": 5681
    },
    {
      "epoch": 2.6738823529411766,
      "grad_norm": 0.2863864612058429,
      "learning_rate": 0.00022122363610679863,
      "loss": 0.1309,
      "step": 5682
    },
    {
      "epoch": 2.674352941176471,
      "grad_norm": 0.6898540879051347,
      "learning_rate": 0.0002211850536954127,
      "loss": 0.4551,
      "step": 5683
    },
    {
      "epoch": 2.6748235294117646,
      "grad_norm": 0.7428367458288417,
      "learning_rate": 0.00022114646520447406,
      "loss": 0.8047,
      "step": 5684
    },
    {
      "epoch": 2.675294117647059,
      "grad_norm": 0.702487435745668,
      "learning_rate": 0.00022110787063727832,
      "loss": 0.5508,
      "step": 5685
    },
    {
      "epoch": 2.675764705882353,
      "grad_norm": 0.4564707554689176,
      "learning_rate": 0.00022106926999712165,
      "loss": 0.5625,
      "step": 5686
    },
    {
      "epoch": 2.676235294117647,
      "grad_norm": 0.5872802001528862,
      "learning_rate": 0.0002210306632873007,
      "loss": 0.5898,
      "step": 5687
    },
    {
      "epoch": 2.676705882352941,
      "grad_norm": 0.46368883829067714,
      "learning_rate": 0.00022099205051111272,
      "loss": 0.3066,
      "step": 5688
    },
    {
      "epoch": 2.6771764705882353,
      "grad_norm": 0.2952905112526489,
      "learning_rate": 0.0002209534316718554,
      "loss": 0.1416,
      "step": 5689
    },
    {
      "epoch": 2.6776470588235295,
      "grad_norm": 1.017029177824626,
      "learning_rate": 0.000220914806772827,
      "loss": 1.25,
      "step": 5690
    },
    {
      "epoch": 2.6781176470588237,
      "grad_norm": 0.6931987642186909,
      "learning_rate": 0.0002208761758173263,
      "loss": 0.6914,
      "step": 5691
    },
    {
      "epoch": 2.6785882352941175,
      "grad_norm": 0.4358043772667271,
      "learning_rate": 0.0002208375388086525,
      "loss": 0.1709,
      "step": 5692
    },
    {
      "epoch": 2.6790588235294117,
      "grad_norm": 0.37624661593659736,
      "learning_rate": 0.00022079889575010543,
      "loss": 0.2559,
      "step": 5693
    },
    {
      "epoch": 2.679529411764706,
      "grad_norm": 0.458960169003368,
      "learning_rate": 0.0002207602466449854,
      "loss": 0.2598,
      "step": 5694
    },
    {
      "epoch": 2.68,
      "grad_norm": 0.3968137785158616,
      "learning_rate": 0.00022072159149659327,
      "loss": 0.3359,
      "step": 5695
    },
    {
      "epoch": 2.680470588235294,
      "grad_norm": 0.3495458878409794,
      "learning_rate": 0.00022068293030823026,
      "loss": 0.1953,
      "step": 5696
    },
    {
      "epoch": 2.680941176470588,
      "grad_norm": 0.6602547454309817,
      "learning_rate": 0.00022064426308319838,
      "loss": 0.4785,
      "step": 5697
    },
    {
      "epoch": 2.6814117647058824,
      "grad_norm": 0.9597304998364133,
      "learning_rate": 0.0002206055898247999,
      "loss": 0.9961,
      "step": 5698
    },
    {
      "epoch": 2.6818823529411766,
      "grad_norm": 0.5696548658729349,
      "learning_rate": 0.0002205669105363377,
      "loss": 0.5234,
      "step": 5699
    },
    {
      "epoch": 2.682352941176471,
      "grad_norm": 0.5230067294709504,
      "learning_rate": 0.00022052822522111522,
      "loss": 0.6992,
      "step": 5700
    },
    {
      "epoch": 2.6828235294117646,
      "grad_norm": 0.6678324112848377,
      "learning_rate": 0.0002204895338824364,
      "loss": 0.6172,
      "step": 5701
    },
    {
      "epoch": 2.683294117647059,
      "grad_norm": 0.313190190359293,
      "learning_rate": 0.0002204508365236056,
      "loss": 0.1484,
      "step": 5702
    },
    {
      "epoch": 2.683764705882353,
      "grad_norm": 0.2638162940036779,
      "learning_rate": 0.0002204121331479278,
      "loss": 0.3008,
      "step": 5703
    },
    {
      "epoch": 2.684235294117647,
      "grad_norm": 0.43176996785514443,
      "learning_rate": 0.0002203734237587085,
      "loss": 0.2031,
      "step": 5704
    },
    {
      "epoch": 2.684705882352941,
      "grad_norm": 0.41819167460174816,
      "learning_rate": 0.00022033470835925361,
      "loss": 0.4453,
      "step": 5705
    },
    {
      "epoch": 2.6851764705882353,
      "grad_norm": 0.34141835247143576,
      "learning_rate": 0.00022029598695286963,
      "loss": 0.1973,
      "step": 5706
    },
    {
      "epoch": 2.6856470588235295,
      "grad_norm": 0.4961287134754011,
      "learning_rate": 0.00022025725954286356,
      "loss": 0.5703,
      "step": 5707
    },
    {
      "epoch": 2.6861176470588237,
      "grad_norm": 0.34247139729518544,
      "learning_rate": 0.00022021852613254297,
      "loss": 0.4102,
      "step": 5708
    },
    {
      "epoch": 2.6865882352941175,
      "grad_norm": 0.41378937961368806,
      "learning_rate": 0.00022017978672521577,
      "loss": 0.3379,
      "step": 5709
    },
    {
      "epoch": 2.6870588235294117,
      "grad_norm": 0.43040855602627914,
      "learning_rate": 0.00022014104132419055,
      "loss": 0.332,
      "step": 5710
    },
    {
      "epoch": 2.687529411764706,
      "grad_norm": 0.46290831105427466,
      "learning_rate": 0.0002201022899327764,
      "loss": 0.373,
      "step": 5711
    },
    {
      "epoch": 2.6879999999999997,
      "grad_norm": 0.6037399703355084,
      "learning_rate": 0.00022006353255428281,
      "loss": 0.8984,
      "step": 5712
    },
    {
      "epoch": 2.688470588235294,
      "grad_norm": 0.545943253313731,
      "learning_rate": 0.00022002476919201987,
      "loss": 0.6289,
      "step": 5713
    },
    {
      "epoch": 2.688941176470588,
      "grad_norm": 0.20449627318479427,
      "learning_rate": 0.00021998599984929818,
      "loss": 0.1006,
      "step": 5714
    },
    {
      "epoch": 2.6894117647058824,
      "grad_norm": 0.3859186933212679,
      "learning_rate": 0.00021994722452942886,
      "loss": 0.3828,
      "step": 5715
    },
    {
      "epoch": 2.6898823529411766,
      "grad_norm": 0.4783428470982351,
      "learning_rate": 0.00021990844323572346,
      "loss": 0.5,
      "step": 5716
    },
    {
      "epoch": 2.690352941176471,
      "grad_norm": 0.4515111302297094,
      "learning_rate": 0.0002198696559714941,
      "loss": 0.6367,
      "step": 5717
    },
    {
      "epoch": 2.6908235294117646,
      "grad_norm": 0.8171663950444725,
      "learning_rate": 0.0002198308627400534,
      "loss": 0.5625,
      "step": 5718
    },
    {
      "epoch": 2.691294117647059,
      "grad_norm": 0.47710282950280664,
      "learning_rate": 0.00021979206354471453,
      "loss": 0.4102,
      "step": 5719
    },
    {
      "epoch": 2.691764705882353,
      "grad_norm": 0.387751816796568,
      "learning_rate": 0.00021975325838879107,
      "loss": 0.3594,
      "step": 5720
    },
    {
      "epoch": 2.692235294117647,
      "grad_norm": 0.3553111966930887,
      "learning_rate": 0.0002197144472755972,
      "loss": 0.2246,
      "step": 5721
    },
    {
      "epoch": 2.692705882352941,
      "grad_norm": 0.18587636038930047,
      "learning_rate": 0.00021967563020844762,
      "loss": 0.0659,
      "step": 5722
    },
    {
      "epoch": 2.6931764705882353,
      "grad_norm": 0.3618698656341896,
      "learning_rate": 0.00021963680719065746,
      "loss": 0.3184,
      "step": 5723
    },
    {
      "epoch": 2.6936470588235295,
      "grad_norm": 0.34338646407878387,
      "learning_rate": 0.0002195979782255424,
      "loss": 0.1777,
      "step": 5724
    },
    {
      "epoch": 2.6941176470588237,
      "grad_norm": 0.701274436135329,
      "learning_rate": 0.0002195591433164186,
      "loss": 0.8086,
      "step": 5725
    },
    {
      "epoch": 2.6945882352941175,
      "grad_norm": 0.3768080040713405,
      "learning_rate": 0.00021952030246660283,
      "loss": 0.2412,
      "step": 5726
    },
    {
      "epoch": 2.6950588235294117,
      "grad_norm": 0.5274597409029741,
      "learning_rate": 0.00021948145567941217,
      "loss": 0.7031,
      "step": 5727
    },
    {
      "epoch": 2.695529411764706,
      "grad_norm": 0.4451677437178509,
      "learning_rate": 0.0002194426029581644,
      "loss": 0.3887,
      "step": 5728
    },
    {
      "epoch": 2.6959999999999997,
      "grad_norm": 0.4004714860145041,
      "learning_rate": 0.00021940374430617777,
      "loss": 0.2773,
      "step": 5729
    },
    {
      "epoch": 2.696470588235294,
      "grad_norm": 0.6155334670436019,
      "learning_rate": 0.0002193648797267709,
      "loss": 0.7734,
      "step": 5730
    },
    {
      "epoch": 2.696941176470588,
      "grad_norm": 0.5632952023258654,
      "learning_rate": 0.0002193260092232631,
      "loss": 0.6836,
      "step": 5731
    },
    {
      "epoch": 2.6974117647058824,
      "grad_norm": 0.3371805416444304,
      "learning_rate": 0.00021928713279897408,
      "loss": 0.4043,
      "step": 5732
    },
    {
      "epoch": 2.6978823529411766,
      "grad_norm": 0.4666002672111229,
      "learning_rate": 0.00021924825045722404,
      "loss": 0.4707,
      "step": 5733
    },
    {
      "epoch": 2.698352941176471,
      "grad_norm": 0.44225422667065273,
      "learning_rate": 0.00021920936220133373,
      "loss": 0.4473,
      "step": 5734
    },
    {
      "epoch": 2.6988235294117646,
      "grad_norm": 0.4786061856085557,
      "learning_rate": 0.0002191704680346245,
      "loss": 0.3203,
      "step": 5735
    },
    {
      "epoch": 2.699294117647059,
      "grad_norm": 0.4447035947139761,
      "learning_rate": 0.00021913156796041794,
      "loss": 0.6094,
      "step": 5736
    },
    {
      "epoch": 2.699764705882353,
      "grad_norm": 0.5594220097302212,
      "learning_rate": 0.0002190926619820364,
      "loss": 0.5234,
      "step": 5737
    },
    {
      "epoch": 2.700235294117647,
      "grad_norm": 0.32492978218798585,
      "learning_rate": 0.00021905375010280266,
      "loss": 0.1699,
      "step": 5738
    },
    {
      "epoch": 2.700705882352941,
      "grad_norm": 0.6628247921303547,
      "learning_rate": 0.00021901483232603995,
      "loss": 0.6289,
      "step": 5739
    },
    {
      "epoch": 2.7011764705882353,
      "grad_norm": 0.4134164571623913,
      "learning_rate": 0.000218975908655072,
      "loss": 0.3477,
      "step": 5740
    },
    {
      "epoch": 2.7016470588235295,
      "grad_norm": 0.574742237882419,
      "learning_rate": 0.0002189369790932232,
      "loss": 0.5664,
      "step": 5741
    },
    {
      "epoch": 2.7021176470588237,
      "grad_norm": 0.5491966732894259,
      "learning_rate": 0.0002188980436438182,
      "loss": 0.8164,
      "step": 5742
    },
    {
      "epoch": 2.7025882352941175,
      "grad_norm": 0.6834726155278792,
      "learning_rate": 0.0002188591023101823,
      "loss": 0.8281,
      "step": 5743
    },
    {
      "epoch": 2.7030588235294117,
      "grad_norm": 0.4930148428106541,
      "learning_rate": 0.00021882015509564132,
      "loss": 0.3828,
      "step": 5744
    },
    {
      "epoch": 2.703529411764706,
      "grad_norm": 0.4593161642023077,
      "learning_rate": 0.00021878120200352158,
      "loss": 0.4355,
      "step": 5745
    },
    {
      "epoch": 2.7039999999999997,
      "grad_norm": 0.5204115498201721,
      "learning_rate": 0.00021874224303714983,
      "loss": 0.7305,
      "step": 5746
    },
    {
      "epoch": 2.704470588235294,
      "grad_norm": 0.3608681697355452,
      "learning_rate": 0.00021870327819985331,
      "loss": 0.1719,
      "step": 5747
    },
    {
      "epoch": 2.704941176470588,
      "grad_norm": 0.48692111715780384,
      "learning_rate": 0.00021866430749495987,
      "loss": 0.4219,
      "step": 5748
    },
    {
      "epoch": 2.7054117647058824,
      "grad_norm": 0.43697520647690186,
      "learning_rate": 0.00021862533092579774,
      "loss": 0.375,
      "step": 5749
    },
    {
      "epoch": 2.7058823529411766,
      "grad_norm": 0.23309119678433754,
      "learning_rate": 0.00021858634849569576,
      "loss": 0.1123,
      "step": 5750
    },
    {
      "epoch": 2.706352941176471,
      "grad_norm": 0.4426750463410463,
      "learning_rate": 0.00021854736020798319,
      "loss": 0.4668,
      "step": 5751
    },
    {
      "epoch": 2.7068235294117646,
      "grad_norm": 0.42009287203611767,
      "learning_rate": 0.00021850836606598987,
      "loss": 0.5,
      "step": 5752
    },
    {
      "epoch": 2.707294117647059,
      "grad_norm": 0.5097152232405595,
      "learning_rate": 0.00021846936607304603,
      "loss": 0.3867,
      "step": 5753
    },
    {
      "epoch": 2.707764705882353,
      "grad_norm": 0.08689253800609673,
      "learning_rate": 0.00021843036023248247,
      "loss": 0.0352,
      "step": 5754
    },
    {
      "epoch": 2.708235294117647,
      "grad_norm": 0.3442580393964267,
      "learning_rate": 0.00021839134854763055,
      "loss": 0.1914,
      "step": 5755
    },
    {
      "epoch": 2.708705882352941,
      "grad_norm": 0.48272237237618204,
      "learning_rate": 0.00021835233102182199,
      "loss": 0.4922,
      "step": 5756
    },
    {
      "epoch": 2.7091764705882353,
      "grad_norm": 0.7480300583270812,
      "learning_rate": 0.00021831330765838903,
      "loss": 0.8047,
      "step": 5757
    },
    {
      "epoch": 2.7096470588235295,
      "grad_norm": 0.4592847957797846,
      "learning_rate": 0.00021827427846066462,
      "loss": 0.3984,
      "step": 5758
    },
    {
      "epoch": 2.7101176470588237,
      "grad_norm": 0.65076023478976,
      "learning_rate": 0.0002182352434319819,
      "loss": 0.5156,
      "step": 5759
    },
    {
      "epoch": 2.7105882352941175,
      "grad_norm": 0.5304689473089341,
      "learning_rate": 0.0002181962025756747,
      "loss": 0.5352,
      "step": 5760
    },
    {
      "epoch": 2.7110588235294117,
      "grad_norm": 0.3283580734891656,
      "learning_rate": 0.00021815715589507728,
      "loss": 0.373,
      "step": 5761
    },
    {
      "epoch": 2.711529411764706,
      "grad_norm": 0.4412275705418654,
      "learning_rate": 0.00021811810339352446,
      "loss": 0.3066,
      "step": 5762
    },
    {
      "epoch": 2.7119999999999997,
      "grad_norm": 0.40417528366254096,
      "learning_rate": 0.0002180790450743515,
      "loss": 0.2969,
      "step": 5763
    },
    {
      "epoch": 2.712470588235294,
      "grad_norm": 0.2939677302321997,
      "learning_rate": 0.0002180399809408941,
      "loss": 0.2559,
      "step": 5764
    },
    {
      "epoch": 2.712941176470588,
      "grad_norm": 0.31501309090729424,
      "learning_rate": 0.00021800091099648865,
      "loss": 0.1807,
      "step": 5765
    },
    {
      "epoch": 2.7134117647058824,
      "grad_norm": 0.6223291402089028,
      "learning_rate": 0.0002179618352444719,
      "loss": 0.582,
      "step": 5766
    },
    {
      "epoch": 2.7138823529411766,
      "grad_norm": 0.24239152898247204,
      "learning_rate": 0.00021792275368818102,
      "loss": 0.0879,
      "step": 5767
    },
    {
      "epoch": 2.714352941176471,
      "grad_norm": 0.4204150472051753,
      "learning_rate": 0.00021788366633095375,
      "loss": 0.1719,
      "step": 5768
    },
    {
      "epoch": 2.7148235294117646,
      "grad_norm": 0.5694934982329414,
      "learning_rate": 0.00021784457317612852,
      "loss": 0.4551,
      "step": 5769
    },
    {
      "epoch": 2.715294117647059,
      "grad_norm": 0.4266148297783364,
      "learning_rate": 0.0002178054742270439,
      "loss": 0.4453,
      "step": 5770
    },
    {
      "epoch": 2.715764705882353,
      "grad_norm": 0.6779152807808982,
      "learning_rate": 0.0002177663694870392,
      "loss": 0.9492,
      "step": 5771
    },
    {
      "epoch": 2.716235294117647,
      "grad_norm": 0.8499071082366265,
      "learning_rate": 0.00021772725895945422,
      "loss": 0.6484,
      "step": 5772
    },
    {
      "epoch": 2.716705882352941,
      "grad_norm": 0.5410021199317938,
      "learning_rate": 0.0002176881426476291,
      "loss": 0.6055,
      "step": 5773
    },
    {
      "epoch": 2.7171764705882353,
      "grad_norm": 0.32711718509978616,
      "learning_rate": 0.00021764902055490456,
      "loss": 0.2793,
      "step": 5774
    },
    {
      "epoch": 2.7176470588235295,
      "grad_norm": 0.48601866944269717,
      "learning_rate": 0.00021760989268462182,
      "loss": 0.4043,
      "step": 5775
    },
    {
      "epoch": 2.7181176470588237,
      "grad_norm": 0.2921177845437657,
      "learning_rate": 0.00021757075904012266,
      "loss": 0.3203,
      "step": 5776
    },
    {
      "epoch": 2.7185882352941175,
      "grad_norm": 0.3302519487369796,
      "learning_rate": 0.00021753161962474924,
      "loss": 0.2207,
      "step": 5777
    },
    {
      "epoch": 2.7190588235294118,
      "grad_norm": 0.4342042808193346,
      "learning_rate": 0.00021749247444184422,
      "loss": 0.4121,
      "step": 5778
    },
    {
      "epoch": 2.719529411764706,
      "grad_norm": 0.7356011768052735,
      "learning_rate": 0.00021745332349475092,
      "loss": 0.4688,
      "step": 5779
    },
    {
      "epoch": 2.7199999999999998,
      "grad_norm": 0.3703396588407391,
      "learning_rate": 0.0002174141667868129,
      "loss": 0.5742,
      "step": 5780
    },
    {
      "epoch": 2.720470588235294,
      "grad_norm": 0.5494593398360886,
      "learning_rate": 0.00021737500432137435,
      "loss": 0.252,
      "step": 5781
    },
    {
      "epoch": 2.720941176470588,
      "grad_norm": 0.395509937235939,
      "learning_rate": 0.00021733583610177997,
      "loss": 0.3809,
      "step": 5782
    },
    {
      "epoch": 2.7214117647058824,
      "grad_norm": 0.4008137082855877,
      "learning_rate": 0.00021729666213137494,
      "loss": 0.1904,
      "step": 5783
    },
    {
      "epoch": 2.7218823529411766,
      "grad_norm": 0.30106732411734705,
      "learning_rate": 0.00021725748241350486,
      "loss": 0.1602,
      "step": 5784
    },
    {
      "epoch": 2.722352941176471,
      "grad_norm": 0.6356920166564782,
      "learning_rate": 0.00021721829695151585,
      "loss": 0.582,
      "step": 5785
    },
    {
      "epoch": 2.7228235294117646,
      "grad_norm": 0.5773677901122111,
      "learning_rate": 0.00021717910574875458,
      "loss": 0.7891,
      "step": 5786
    },
    {
      "epoch": 2.723294117647059,
      "grad_norm": 0.7306559722119965,
      "learning_rate": 0.00021713990880856818,
      "loss": 0.7461,
      "step": 5787
    },
    {
      "epoch": 2.723764705882353,
      "grad_norm": 0.5746457771866594,
      "learning_rate": 0.0002171007061343043,
      "loss": 0.7695,
      "step": 5788
    },
    {
      "epoch": 2.724235294117647,
      "grad_norm": 0.4603210235768996,
      "learning_rate": 0.0002170614977293109,
      "loss": 0.3652,
      "step": 5789
    },
    {
      "epoch": 2.724705882352941,
      "grad_norm": 0.40591629342804236,
      "learning_rate": 0.00021702228359693672,
      "loss": 0.5195,
      "step": 5790
    },
    {
      "epoch": 2.7251764705882353,
      "grad_norm": 0.5765629813446779,
      "learning_rate": 0.00021698306374053075,
      "loss": 0.8047,
      "step": 5791
    },
    {
      "epoch": 2.7256470588235295,
      "grad_norm": 0.1917361690637917,
      "learning_rate": 0.00021694383816344258,
      "loss": 0.0491,
      "step": 5792
    },
    {
      "epoch": 2.7261176470588238,
      "grad_norm": 0.3931424024707498,
      "learning_rate": 0.0002169046068690223,
      "loss": 0.4336,
      "step": 5793
    },
    {
      "epoch": 2.7265882352941175,
      "grad_norm": 0.5613536918656258,
      "learning_rate": 0.00021686536986062043,
      "loss": 0.6641,
      "step": 5794
    },
    {
      "epoch": 2.7270588235294118,
      "grad_norm": 0.336663322737468,
      "learning_rate": 0.000216826127141588,
      "loss": 0.1982,
      "step": 5795
    },
    {
      "epoch": 2.727529411764706,
      "grad_norm": 0.38290225738632855,
      "learning_rate": 0.0002167868787152765,
      "loss": 0.248,
      "step": 5796
    },
    {
      "epoch": 2.7279999999999998,
      "grad_norm": 0.37488907910114583,
      "learning_rate": 0.000216747624585038,
      "loss": 0.3633,
      "step": 5797
    },
    {
      "epoch": 2.728470588235294,
      "grad_norm": 0.7245474735102622,
      "learning_rate": 0.00021670836475422498,
      "loss": 0.6602,
      "step": 5798
    },
    {
      "epoch": 2.728941176470588,
      "grad_norm": 0.5253718819264639,
      "learning_rate": 0.00021666909922619037,
      "loss": 0.5547,
      "step": 5799
    },
    {
      "epoch": 2.7294117647058824,
      "grad_norm": 0.40751100922514333,
      "learning_rate": 0.0002166298280042877,
      "loss": 0.2793,
      "step": 5800
    },
    {
      "epoch": 2.7298823529411766,
      "grad_norm": 0.38174507228944354,
      "learning_rate": 0.00021659055109187088,
      "loss": 0.3398,
      "step": 5801
    },
    {
      "epoch": 2.7303529411764704,
      "grad_norm": 0.3901248594212028,
      "learning_rate": 0.0002165512684922944,
      "loss": 0.2832,
      "step": 5802
    },
    {
      "epoch": 2.7308235294117646,
      "grad_norm": 0.5506331559869905,
      "learning_rate": 0.0002165119802089131,
      "loss": 0.375,
      "step": 5803
    },
    {
      "epoch": 2.731294117647059,
      "grad_norm": 0.43343496204419546,
      "learning_rate": 0.00021647268624508254,
      "loss": 0.2295,
      "step": 5804
    },
    {
      "epoch": 2.731764705882353,
      "grad_norm": 0.782807427127028,
      "learning_rate": 0.00021643338660415844,
      "loss": 0.9102,
      "step": 5805
    },
    {
      "epoch": 2.732235294117647,
      "grad_norm": 2.165186705276986,
      "learning_rate": 0.0002163940812894973,
      "loss": 2.25,
      "step": 5806
    },
    {
      "epoch": 2.732705882352941,
      "grad_norm": 0.5042197908209956,
      "learning_rate": 0.00021635477030445597,
      "loss": 0.5156,
      "step": 5807
    },
    {
      "epoch": 2.7331764705882353,
      "grad_norm": 0.2677577787900803,
      "learning_rate": 0.0002163154536523918,
      "loss": 0.1562,
      "step": 5808
    },
    {
      "epoch": 2.7336470588235295,
      "grad_norm": 0.24836133009140904,
      "learning_rate": 0.0002162761313366626,
      "loss": 0.1182,
      "step": 5809
    },
    {
      "epoch": 2.7341176470588238,
      "grad_norm": 0.521886230279867,
      "learning_rate": 0.0002162368033606267,
      "loss": 0.4883,
      "step": 5810
    },
    {
      "epoch": 2.7345882352941175,
      "grad_norm": 0.5469477485873295,
      "learning_rate": 0.00021619746972764288,
      "loss": 0.3828,
      "step": 5811
    },
    {
      "epoch": 2.7350588235294118,
      "grad_norm": 0.38693687805535787,
      "learning_rate": 0.00021615813044107045,
      "loss": 0.3672,
      "step": 5812
    },
    {
      "epoch": 2.735529411764706,
      "grad_norm": 0.4626532945529421,
      "learning_rate": 0.00021611878550426922,
      "loss": 0.4746,
      "step": 5813
    },
    {
      "epoch": 2.7359999999999998,
      "grad_norm": 0.3494056298791717,
      "learning_rate": 0.00021607943492059936,
      "loss": 0.2227,
      "step": 5814
    },
    {
      "epoch": 2.736470588235294,
      "grad_norm": 0.3761889222162181,
      "learning_rate": 0.00021604007869342168,
      "loss": 0.4512,
      "step": 5815
    },
    {
      "epoch": 2.736941176470588,
      "grad_norm": 0.5515220078272007,
      "learning_rate": 0.00021600071682609733,
      "loss": 0.7266,
      "step": 5816
    },
    {
      "epoch": 2.7374117647058824,
      "grad_norm": 0.3676598494607415,
      "learning_rate": 0.00021596134932198803,
      "loss": 0.2178,
      "step": 5817
    },
    {
      "epoch": 2.7378823529411767,
      "grad_norm": 0.3102686879741099,
      "learning_rate": 0.00021592197618445596,
      "loss": 0.2305,
      "step": 5818
    },
    {
      "epoch": 2.7383529411764704,
      "grad_norm": 0.5170887669226463,
      "learning_rate": 0.00021588259741686373,
      "loss": 0.5195,
      "step": 5819
    },
    {
      "epoch": 2.7388235294117647,
      "grad_norm": 0.3059117348309109,
      "learning_rate": 0.00021584321302257464,
      "loss": 0.1729,
      "step": 5820
    },
    {
      "epoch": 2.739294117647059,
      "grad_norm": 0.42766878829204813,
      "learning_rate": 0.00021580382300495215,
      "loss": 0.2734,
      "step": 5821
    },
    {
      "epoch": 2.739764705882353,
      "grad_norm": 0.48109891322999343,
      "learning_rate": 0.00021576442736736032,
      "loss": 0.3203,
      "step": 5822
    },
    {
      "epoch": 2.740235294117647,
      "grad_norm": 0.32004471865962997,
      "learning_rate": 0.00021572502611316393,
      "loss": 0.1328,
      "step": 5823
    },
    {
      "epoch": 2.740705882352941,
      "grad_norm": 0.24447862719534022,
      "learning_rate": 0.0002156856192457279,
      "loss": 0.104,
      "step": 5824
    },
    {
      "epoch": 2.7411764705882353,
      "grad_norm": 0.4368374703944882,
      "learning_rate": 0.00021564620676841782,
      "loss": 0.4688,
      "step": 5825
    },
    {
      "epoch": 2.7416470588235295,
      "grad_norm": 0.3102341785713059,
      "learning_rate": 0.0002156067886845996,
      "loss": 0.2695,
      "step": 5826
    },
    {
      "epoch": 2.7421176470588238,
      "grad_norm": 0.6947595125972911,
      "learning_rate": 0.00021556736499763992,
      "loss": 1.1094,
      "step": 5827
    },
    {
      "epoch": 2.7425882352941175,
      "grad_norm": 0.493880146426631,
      "learning_rate": 0.00021552793571090568,
      "loss": 0.5391,
      "step": 5828
    },
    {
      "epoch": 2.7430588235294118,
      "grad_norm": 0.2621259916458612,
      "learning_rate": 0.0002154885008277642,
      "loss": 0.1133,
      "step": 5829
    },
    {
      "epoch": 2.743529411764706,
      "grad_norm": 0.3854403739155651,
      "learning_rate": 0.0002154490603515836,
      "loss": 0.4004,
      "step": 5830
    },
    {
      "epoch": 2.7439999999999998,
      "grad_norm": 0.43001762012175576,
      "learning_rate": 0.0002154096142857323,
      "loss": 0.4883,
      "step": 5831
    },
    {
      "epoch": 2.744470588235294,
      "grad_norm": 0.32162233985981975,
      "learning_rate": 0.000215370162633579,
      "loss": 0.2871,
      "step": 5832
    },
    {
      "epoch": 2.744941176470588,
      "grad_norm": 0.29838479767619597,
      "learning_rate": 0.00021533070539849316,
      "loss": 0.1562,
      "step": 5833
    },
    {
      "epoch": 2.7454117647058824,
      "grad_norm": 0.4140814312732915,
      "learning_rate": 0.00021529124258384466,
      "loss": 0.2227,
      "step": 5834
    },
    {
      "epoch": 2.7458823529411767,
      "grad_norm": 0.255990063176985,
      "learning_rate": 0.00021525177419300383,
      "loss": 0.1177,
      "step": 5835
    },
    {
      "epoch": 2.7463529411764704,
      "grad_norm": 0.3967815167912101,
      "learning_rate": 0.0002152123002293414,
      "loss": 0.3438,
      "step": 5836
    },
    {
      "epoch": 2.7468235294117647,
      "grad_norm": 0.8802725034800085,
      "learning_rate": 0.0002151728206962287,
      "loss": 1.0469,
      "step": 5837
    },
    {
      "epoch": 2.747294117647059,
      "grad_norm": 0.21274325770312455,
      "learning_rate": 0.00021513333559703738,
      "loss": 0.1211,
      "step": 5838
    },
    {
      "epoch": 2.7477647058823527,
      "grad_norm": 0.5331696452097434,
      "learning_rate": 0.00021509384493513978,
      "loss": 0.6992,
      "step": 5839
    },
    {
      "epoch": 2.748235294117647,
      "grad_norm": 0.3408004286938784,
      "learning_rate": 0.00021505434871390848,
      "loss": 0.3516,
      "step": 5840
    },
    {
      "epoch": 2.748705882352941,
      "grad_norm": 0.2794244365313138,
      "learning_rate": 0.00021501484693671677,
      "loss": 0.1934,
      "step": 5841
    },
    {
      "epoch": 2.7491764705882353,
      "grad_norm": 0.810287823876288,
      "learning_rate": 0.00021497533960693823,
      "loss": 1.1875,
      "step": 5842
    },
    {
      "epoch": 2.7496470588235296,
      "grad_norm": 0.3720937031662669,
      "learning_rate": 0.00021493582672794693,
      "loss": 0.252,
      "step": 5843
    },
    {
      "epoch": 2.7501176470588238,
      "grad_norm": 0.6100120716933144,
      "learning_rate": 0.00021489630830311758,
      "loss": 0.7461,
      "step": 5844
    },
    {
      "epoch": 2.7505882352941176,
      "grad_norm": 0.5910508400884038,
      "learning_rate": 0.00021485678433582516,
      "loss": 0.75,
      "step": 5845
    },
    {
      "epoch": 2.751058823529412,
      "grad_norm": 0.4315214909677429,
      "learning_rate": 0.00021481725482944518,
      "loss": 0.498,
      "step": 5846
    },
    {
      "epoch": 2.751529411764706,
      "grad_norm": 0.3102342111909219,
      "learning_rate": 0.0002147777197873538,
      "loss": 0.5273,
      "step": 5847
    },
    {
      "epoch": 2.752,
      "grad_norm": 0.40857368356124574,
      "learning_rate": 0.0002147381792129274,
      "loss": 0.4277,
      "step": 5848
    },
    {
      "epoch": 2.752470588235294,
      "grad_norm": 0.48988928714251523,
      "learning_rate": 0.00021469863310954296,
      "loss": 0.2598,
      "step": 5849
    },
    {
      "epoch": 2.7529411764705882,
      "grad_norm": 0.4411380599066867,
      "learning_rate": 0.00021465908148057787,
      "loss": 0.3789,
      "step": 5850
    },
    {
      "epoch": 2.7534117647058824,
      "grad_norm": 0.1954701260521231,
      "learning_rate": 0.00021461952432941006,
      "loss": 0.0728,
      "step": 5851
    },
    {
      "epoch": 2.7538823529411767,
      "grad_norm": 0.34969474962928254,
      "learning_rate": 0.00021457996165941798,
      "loss": 0.2695,
      "step": 5852
    },
    {
      "epoch": 2.7543529411764704,
      "grad_norm": 0.35072907799911285,
      "learning_rate": 0.00021454039347398033,
      "loss": 0.2363,
      "step": 5853
    },
    {
      "epoch": 2.7548235294117647,
      "grad_norm": 0.3695702937353654,
      "learning_rate": 0.00021450081977647658,
      "loss": 0.373,
      "step": 5854
    },
    {
      "epoch": 2.755294117647059,
      "grad_norm": 0.42148244586655464,
      "learning_rate": 0.00021446124057028642,
      "loss": 0.2217,
      "step": 5855
    },
    {
      "epoch": 2.7557647058823527,
      "grad_norm": 0.26197466796128976,
      "learning_rate": 0.00021442165585879014,
      "loss": 0.126,
      "step": 5856
    },
    {
      "epoch": 2.756235294117647,
      "grad_norm": 0.6290373052060171,
      "learning_rate": 0.00021438206564536845,
      "loss": 0.6562,
      "step": 5857
    },
    {
      "epoch": 2.756705882352941,
      "grad_norm": 0.672602465177169,
      "learning_rate": 0.00021434246993340256,
      "loss": 0.8789,
      "step": 5858
    },
    {
      "epoch": 2.7571764705882353,
      "grad_norm": 0.661010341422546,
      "learning_rate": 0.00021430286872627417,
      "loss": 0.7969,
      "step": 5859
    },
    {
      "epoch": 2.7576470588235296,
      "grad_norm": 0.2866663090228659,
      "learning_rate": 0.00021426326202736532,
      "loss": 0.1758,
      "step": 5860
    },
    {
      "epoch": 2.758117647058824,
      "grad_norm": 0.30342905620650956,
      "learning_rate": 0.00021422364984005876,
      "loss": 0.2109,
      "step": 5861
    },
    {
      "epoch": 2.7585882352941176,
      "grad_norm": 0.424122396457882,
      "learning_rate": 0.00021418403216773746,
      "loss": 0.3047,
      "step": 5862
    },
    {
      "epoch": 2.759058823529412,
      "grad_norm": 0.39088747528242296,
      "learning_rate": 0.000214144409013785,
      "loss": 0.2852,
      "step": 5863
    },
    {
      "epoch": 2.759529411764706,
      "grad_norm": 0.409084087523257,
      "learning_rate": 0.00021410478038158537,
      "loss": 0.4863,
      "step": 5864
    },
    {
      "epoch": 2.76,
      "grad_norm": 0.4546155284178143,
      "learning_rate": 0.00021406514627452306,
      "loss": 0.4883,
      "step": 5865
    },
    {
      "epoch": 2.760470588235294,
      "grad_norm": 0.6222565130739477,
      "learning_rate": 0.00021402550669598303,
      "loss": 0.5586,
      "step": 5866
    },
    {
      "epoch": 2.7609411764705882,
      "grad_norm": 0.34165078782520464,
      "learning_rate": 0.00021398586164935067,
      "loss": 0.4219,
      "step": 5867
    },
    {
      "epoch": 2.7614117647058825,
      "grad_norm": 0.40032266681741924,
      "learning_rate": 0.0002139462111380119,
      "loss": 0.3652,
      "step": 5868
    },
    {
      "epoch": 2.7618823529411767,
      "grad_norm": 0.3890550093125683,
      "learning_rate": 0.00021390655516535304,
      "loss": 0.4473,
      "step": 5869
    },
    {
      "epoch": 2.7623529411764705,
      "grad_norm": 0.3134510432565094,
      "learning_rate": 0.00021386689373476087,
      "loss": 0.1787,
      "step": 5870
    },
    {
      "epoch": 2.7628235294117647,
      "grad_norm": 0.5090273618577678,
      "learning_rate": 0.00021382722684962276,
      "loss": 0.5664,
      "step": 5871
    },
    {
      "epoch": 2.763294117647059,
      "grad_norm": 0.47950220072474364,
      "learning_rate": 0.00021378755451332637,
      "loss": 0.4062,
      "step": 5872
    },
    {
      "epoch": 2.7637647058823527,
      "grad_norm": 0.559893872647295,
      "learning_rate": 0.00021374787672926,
      "loss": 0.6094,
      "step": 5873
    },
    {
      "epoch": 2.764235294117647,
      "grad_norm": 1.1095491752533255,
      "learning_rate": 0.00021370819350081222,
      "loss": 0.6328,
      "step": 5874
    },
    {
      "epoch": 2.764705882352941,
      "grad_norm": 0.44183094090776015,
      "learning_rate": 0.00021366850483137226,
      "loss": 0.4922,
      "step": 5875
    },
    {
      "epoch": 2.7651764705882353,
      "grad_norm": 0.3188901867788013,
      "learning_rate": 0.00021362881072432965,
      "loss": 0.457,
      "step": 5876
    },
    {
      "epoch": 2.7656470588235296,
      "grad_norm": 0.5383693641016852,
      "learning_rate": 0.00021358911118307455,
      "loss": 0.582,
      "step": 5877
    },
    {
      "epoch": 2.766117647058824,
      "grad_norm": 0.7538587444644145,
      "learning_rate": 0.00021354940621099745,
      "loss": 1.0625,
      "step": 5878
    },
    {
      "epoch": 2.7665882352941176,
      "grad_norm": 0.32037631097801916,
      "learning_rate": 0.00021350969581148934,
      "loss": 0.4199,
      "step": 5879
    },
    {
      "epoch": 2.767058823529412,
      "grad_norm": 0.256885693881512,
      "learning_rate": 0.00021346997998794172,
      "loss": 0.0845,
      "step": 5880
    },
    {
      "epoch": 2.767529411764706,
      "grad_norm": 0.4742674051883084,
      "learning_rate": 0.00021343025874374645,
      "loss": 0.3867,
      "step": 5881
    },
    {
      "epoch": 2.768,
      "grad_norm": 0.48232271944768346,
      "learning_rate": 0.000213390532082296,
      "loss": 0.543,
      "step": 5882
    },
    {
      "epoch": 2.768470588235294,
      "grad_norm": 0.47283643969507183,
      "learning_rate": 0.00021335080000698317,
      "loss": 0.6289,
      "step": 5883
    },
    {
      "epoch": 2.7689411764705882,
      "grad_norm": 0.5221286009170798,
      "learning_rate": 0.00021331106252120126,
      "loss": 0.6602,
      "step": 5884
    },
    {
      "epoch": 2.7694117647058825,
      "grad_norm": 0.6455992783291792,
      "learning_rate": 0.00021327131962834407,
      "loss": 0.5234,
      "step": 5885
    },
    {
      "epoch": 2.7698823529411767,
      "grad_norm": 0.3769748654443334,
      "learning_rate": 0.00021323157133180587,
      "loss": 0.3047,
      "step": 5886
    },
    {
      "epoch": 2.7703529411764705,
      "grad_norm": 0.5319437923990852,
      "learning_rate": 0.0002131918176349813,
      "loss": 0.4766,
      "step": 5887
    },
    {
      "epoch": 2.7708235294117647,
      "grad_norm": 0.43114171885059505,
      "learning_rate": 0.0002131520585412656,
      "loss": 0.3301,
      "step": 5888
    },
    {
      "epoch": 2.771294117647059,
      "grad_norm": 0.3986114202266955,
      "learning_rate": 0.0002131122940540543,
      "loss": 0.3887,
      "step": 5889
    },
    {
      "epoch": 2.7717647058823527,
      "grad_norm": 0.5646847725113249,
      "learning_rate": 0.00021307252417674356,
      "loss": 0.5039,
      "step": 5890
    },
    {
      "epoch": 2.772235294117647,
      "grad_norm": 0.6733272551721283,
      "learning_rate": 0.0002130327489127298,
      "loss": 0.7031,
      "step": 5891
    },
    {
      "epoch": 2.772705882352941,
      "grad_norm": 0.5456438923712102,
      "learning_rate": 0.0002129929682654102,
      "loss": 0.6133,
      "step": 5892
    },
    {
      "epoch": 2.7731764705882354,
      "grad_norm": 0.5718029389330961,
      "learning_rate": 0.0002129531822381821,
      "loss": 0.5625,
      "step": 5893
    },
    {
      "epoch": 2.7736470588235296,
      "grad_norm": 0.24047753548263145,
      "learning_rate": 0.00021291339083444344,
      "loss": 0.1689,
      "step": 5894
    },
    {
      "epoch": 2.774117647058824,
      "grad_norm": 0.5043905744584255,
      "learning_rate": 0.00021287359405759265,
      "loss": 0.3594,
      "step": 5895
    },
    {
      "epoch": 2.7745882352941176,
      "grad_norm": 1.1586512485525913,
      "learning_rate": 0.0002128337919110285,
      "loss": 0.9297,
      "step": 5896
    },
    {
      "epoch": 2.775058823529412,
      "grad_norm": 0.4857910012573639,
      "learning_rate": 0.00021279398439815034,
      "loss": 0.4824,
      "step": 5897
    },
    {
      "epoch": 2.775529411764706,
      "grad_norm": 0.3688683857174464,
      "learning_rate": 0.00021275417152235787,
      "loss": 0.2383,
      "step": 5898
    },
    {
      "epoch": 2.776,
      "grad_norm": 0.4825324280888145,
      "learning_rate": 0.00021271435328705138,
      "loss": 0.6641,
      "step": 5899
    },
    {
      "epoch": 2.776470588235294,
      "grad_norm": 0.5618840517429423,
      "learning_rate": 0.00021267452969563153,
      "loss": 0.3652,
      "step": 5900
    },
    {
      "epoch": 2.7769411764705882,
      "grad_norm": 0.380190848168775,
      "learning_rate": 0.00021263470075149937,
      "loss": 0.3496,
      "step": 5901
    },
    {
      "epoch": 2.7774117647058825,
      "grad_norm": 0.515838051501627,
      "learning_rate": 0.00021259486645805662,
      "loss": 0.4238,
      "step": 5902
    },
    {
      "epoch": 2.7778823529411767,
      "grad_norm": 0.5931038397792856,
      "learning_rate": 0.00021255502681870518,
      "loss": 0.373,
      "step": 5903
    },
    {
      "epoch": 2.7783529411764705,
      "grad_norm": 0.4739820663462951,
      "learning_rate": 0.00021251518183684767,
      "loss": 0.334,
      "step": 5904
    },
    {
      "epoch": 2.7788235294117647,
      "grad_norm": 0.21628232604895875,
      "learning_rate": 0.00021247533151588693,
      "loss": 0.0327,
      "step": 5905
    },
    {
      "epoch": 2.779294117647059,
      "grad_norm": 0.7471749982928497,
      "learning_rate": 0.0002124354758592265,
      "loss": 0.6133,
      "step": 5906
    },
    {
      "epoch": 2.7797647058823527,
      "grad_norm": 0.28896207894183323,
      "learning_rate": 0.00021239561487027017,
      "loss": 0.1299,
      "step": 5907
    },
    {
      "epoch": 2.780235294117647,
      "grad_norm": 0.549037239832453,
      "learning_rate": 0.00021235574855242226,
      "loss": 0.6602,
      "step": 5908
    },
    {
      "epoch": 2.780705882352941,
      "grad_norm": 0.42022596120387645,
      "learning_rate": 0.00021231587690908758,
      "loss": 0.4824,
      "step": 5909
    },
    {
      "epoch": 2.7811764705882354,
      "grad_norm": 0.31449641360204333,
      "learning_rate": 0.00021227599994367137,
      "loss": 0.1245,
      "step": 5910
    },
    {
      "epoch": 2.7816470588235296,
      "grad_norm": 0.49755623413719796,
      "learning_rate": 0.0002122361176595793,
      "loss": 0.2178,
      "step": 5911
    },
    {
      "epoch": 2.782117647058824,
      "grad_norm": 0.44717337619806147,
      "learning_rate": 0.0002121962300602175,
      "loss": 0.457,
      "step": 5912
    },
    {
      "epoch": 2.7825882352941176,
      "grad_norm": 0.514906517220915,
      "learning_rate": 0.00021215633714899264,
      "loss": 0.4629,
      "step": 5913
    },
    {
      "epoch": 2.783058823529412,
      "grad_norm": 0.54811846627168,
      "learning_rate": 0.00021211643892931167,
      "loss": 0.5352,
      "step": 5914
    },
    {
      "epoch": 2.783529411764706,
      "grad_norm": 0.5207238712451612,
      "learning_rate": 0.00021207653540458209,
      "loss": 0.5312,
      "step": 5915
    },
    {
      "epoch": 2.784,
      "grad_norm": 0.5640122775374861,
      "learning_rate": 0.00021203662657821198,
      "loss": 0.5977,
      "step": 5916
    },
    {
      "epoch": 2.784470588235294,
      "grad_norm": 0.42808677960246755,
      "learning_rate": 0.00021199671245360964,
      "loss": 0.2852,
      "step": 5917
    },
    {
      "epoch": 2.7849411764705883,
      "grad_norm": 0.7901552335653054,
      "learning_rate": 0.0002119567930341839,
      "loss": 0.7539,
      "step": 5918
    },
    {
      "epoch": 2.7854117647058825,
      "grad_norm": 0.47422199729004216,
      "learning_rate": 0.00021191686832334423,
      "loss": 0.291,
      "step": 5919
    },
    {
      "epoch": 2.7858823529411767,
      "grad_norm": 0.37624532072871697,
      "learning_rate": 0.0002118769383245003,
      "loss": 0.2695,
      "step": 5920
    },
    {
      "epoch": 2.7863529411764705,
      "grad_norm": 0.417365244186573,
      "learning_rate": 0.0002118370030410623,
      "loss": 0.3496,
      "step": 5921
    },
    {
      "epoch": 2.7868235294117647,
      "grad_norm": 0.3129630526201595,
      "learning_rate": 0.0002117970624764409,
      "loss": 0.1602,
      "step": 5922
    },
    {
      "epoch": 2.787294117647059,
      "grad_norm": 0.2625550708899677,
      "learning_rate": 0.00021175711663404728,
      "loss": 0.3047,
      "step": 5923
    },
    {
      "epoch": 2.7877647058823527,
      "grad_norm": 0.4025577787051295,
      "learning_rate": 0.000211717165517293,
      "loss": 0.4453,
      "step": 5924
    },
    {
      "epoch": 2.788235294117647,
      "grad_norm": 0.7753052343617094,
      "learning_rate": 0.00021167720912959004,
      "loss": 0.6562,
      "step": 5925
    },
    {
      "epoch": 2.788705882352941,
      "grad_norm": 0.5285670542447494,
      "learning_rate": 0.00021163724747435091,
      "loss": 0.5586,
      "step": 5926
    },
    {
      "epoch": 2.7891764705882354,
      "grad_norm": 0.28388926607912895,
      "learning_rate": 0.00021159728055498856,
      "loss": 0.1475,
      "step": 5927
    },
    {
      "epoch": 2.7896470588235296,
      "grad_norm": 0.2552979799670438,
      "learning_rate": 0.00021155730837491625,
      "loss": 0.1445,
      "step": 5928
    },
    {
      "epoch": 2.7901176470588234,
      "grad_norm": 0.3855814755085989,
      "learning_rate": 0.0002115173309375479,
      "loss": 0.3164,
      "step": 5929
    },
    {
      "epoch": 2.7905882352941176,
      "grad_norm": 0.5856200165824715,
      "learning_rate": 0.00021147734824629772,
      "loss": 0.5273,
      "step": 5930
    },
    {
      "epoch": 2.791058823529412,
      "grad_norm": 0.26871871618714505,
      "learning_rate": 0.0002114373603045805,
      "loss": 0.106,
      "step": 5931
    },
    {
      "epoch": 2.791529411764706,
      "grad_norm": 0.3975157994634951,
      "learning_rate": 0.00021139736711581132,
      "loss": 0.6016,
      "step": 5932
    },
    {
      "epoch": 2.792,
      "grad_norm": 0.3478525849295157,
      "learning_rate": 0.00021135736868340587,
      "loss": 0.2363,
      "step": 5933
    },
    {
      "epoch": 2.792470588235294,
      "grad_norm": 0.4382329769340859,
      "learning_rate": 0.00021131736501078018,
      "loss": 0.4824,
      "step": 5934
    },
    {
      "epoch": 2.7929411764705883,
      "grad_norm": 0.49584702754191423,
      "learning_rate": 0.0002112773561013507,
      "loss": 0.6055,
      "step": 5935
    },
    {
      "epoch": 2.7934117647058825,
      "grad_norm": 0.4091246789316494,
      "learning_rate": 0.0002112373419585345,
      "loss": 0.4043,
      "step": 5936
    },
    {
      "epoch": 2.7938823529411767,
      "grad_norm": 0.6824223040494218,
      "learning_rate": 0.00021119732258574893,
      "loss": 1.0312,
      "step": 5937
    },
    {
      "epoch": 2.7943529411764705,
      "grad_norm": 0.6227175806621011,
      "learning_rate": 0.0002111572979864118,
      "loss": 0.9297,
      "step": 5938
    },
    {
      "epoch": 2.7948235294117647,
      "grad_norm": 0.4157599900926538,
      "learning_rate": 0.00021111726816394147,
      "loss": 0.3945,
      "step": 5939
    },
    {
      "epoch": 2.795294117647059,
      "grad_norm": 0.5038787700312994,
      "learning_rate": 0.00021107723312175665,
      "loss": 0.625,
      "step": 5940
    },
    {
      "epoch": 2.7957647058823527,
      "grad_norm": 0.34012978190642595,
      "learning_rate": 0.00021103719286327658,
      "loss": 0.1846,
      "step": 5941
    },
    {
      "epoch": 2.796235294117647,
      "grad_norm": 0.5760015683714379,
      "learning_rate": 0.0002109971473919208,
      "loss": 0.6797,
      "step": 5942
    },
    {
      "epoch": 2.796705882352941,
      "grad_norm": 0.18479962238946193,
      "learning_rate": 0.00021095709671110942,
      "loss": 0.1177,
      "step": 5943
    },
    {
      "epoch": 2.7971764705882354,
      "grad_norm": 0.28711410947950766,
      "learning_rate": 0.00021091704082426308,
      "loss": 0.3047,
      "step": 5944
    },
    {
      "epoch": 2.7976470588235296,
      "grad_norm": 0.5669996056985289,
      "learning_rate": 0.0002108769797348026,
      "loss": 0.5273,
      "step": 5945
    },
    {
      "epoch": 2.7981176470588234,
      "grad_norm": 0.27122671170099616,
      "learning_rate": 0.00021083691344614947,
      "loss": 0.3086,
      "step": 5946
    },
    {
      "epoch": 2.7985882352941176,
      "grad_norm": 0.5765987562598542,
      "learning_rate": 0.00021079684196172554,
      "loss": 0.5977,
      "step": 5947
    },
    {
      "epoch": 2.799058823529412,
      "grad_norm": 0.4059812458943533,
      "learning_rate": 0.00021075676528495307,
      "loss": 0.2656,
      "step": 5948
    },
    {
      "epoch": 2.799529411764706,
      "grad_norm": 0.36365817160670194,
      "learning_rate": 0.00021071668341925482,
      "loss": 0.291,
      "step": 5949
    },
    {
      "epoch": 2.8,
      "grad_norm": 0.5734016333283005,
      "learning_rate": 0.00021067659636805403,
      "loss": 0.6055,
      "step": 5950
    },
    {
      "epoch": 2.800470588235294,
      "grad_norm": 0.3699192706705496,
      "learning_rate": 0.00021063650413477428,
      "loss": 0.5117,
      "step": 5951
    },
    {
      "epoch": 2.8009411764705883,
      "grad_norm": 0.31797946187877385,
      "learning_rate": 0.00021059640672283964,
      "loss": 0.2031,
      "step": 5952
    },
    {
      "epoch": 2.8014117647058825,
      "grad_norm": 0.5970438427938591,
      "learning_rate": 0.0002105563041356747,
      "loss": 0.7617,
      "step": 5953
    },
    {
      "epoch": 2.8018823529411767,
      "grad_norm": 0.36501351358259515,
      "learning_rate": 0.00021051619637670431,
      "loss": 0.2441,
      "step": 5954
    },
    {
      "epoch": 2.8023529411764705,
      "grad_norm": 0.4440459652582336,
      "learning_rate": 0.00021047608344935395,
      "loss": 0.2637,
      "step": 5955
    },
    {
      "epoch": 2.8028235294117647,
      "grad_norm": 0.4548854299562373,
      "learning_rate": 0.0002104359653570494,
      "loss": 0.1992,
      "step": 5956
    },
    {
      "epoch": 2.803294117647059,
      "grad_norm": 0.4147623316510566,
      "learning_rate": 0.00021039584210321699,
      "loss": 0.3418,
      "step": 5957
    },
    {
      "epoch": 2.8037647058823527,
      "grad_norm": 0.4943753216658368,
      "learning_rate": 0.00021035571369128341,
      "loss": 0.6289,
      "step": 5958
    },
    {
      "epoch": 2.804235294117647,
      "grad_norm": 0.6108692224635377,
      "learning_rate": 0.00021031558012467582,
      "loss": 0.4023,
      "step": 5959
    },
    {
      "epoch": 2.804705882352941,
      "grad_norm": 0.28576051922126067,
      "learning_rate": 0.00021027544140682188,
      "loss": 0.1104,
      "step": 5960
    },
    {
      "epoch": 2.8051764705882354,
      "grad_norm": 0.42672786874499363,
      "learning_rate": 0.00021023529754114956,
      "loss": 0.2383,
      "step": 5961
    },
    {
      "epoch": 2.8056470588235296,
      "grad_norm": 0.623708474929322,
      "learning_rate": 0.0002101951485310874,
      "loss": 0.8398,
      "step": 5962
    },
    {
      "epoch": 2.8061176470588234,
      "grad_norm": 0.40142343904946426,
      "learning_rate": 0.0002101549943800643,
      "loss": 0.2852,
      "step": 5963
    },
    {
      "epoch": 2.8065882352941176,
      "grad_norm": 0.5467886139708698,
      "learning_rate": 0.00021011483509150963,
      "loss": 0.5859,
      "step": 5964
    },
    {
      "epoch": 2.807058823529412,
      "grad_norm": 0.5060509377978899,
      "learning_rate": 0.00021007467066885314,
      "loss": 0.416,
      "step": 5965
    },
    {
      "epoch": 2.8075294117647056,
      "grad_norm": 0.35053302442273754,
      "learning_rate": 0.00021003450111552514,
      "loss": 0.3047,
      "step": 5966
    },
    {
      "epoch": 2.808,
      "grad_norm": 0.4220512783173255,
      "learning_rate": 0.00020999432643495626,
      "loss": 0.2988,
      "step": 5967
    },
    {
      "epoch": 2.808470588235294,
      "grad_norm": 0.3276728798133122,
      "learning_rate": 0.00020995414663057768,
      "loss": 0.1816,
      "step": 5968
    },
    {
      "epoch": 2.8089411764705883,
      "grad_norm": 0.507865810068967,
      "learning_rate": 0.00020991396170582088,
      "loss": 0.3828,
      "step": 5969
    },
    {
      "epoch": 2.8094117647058825,
      "grad_norm": 0.2515417043876185,
      "learning_rate": 0.00020987377166411785,
      "loss": 0.249,
      "step": 5970
    },
    {
      "epoch": 2.8098823529411767,
      "grad_norm": 0.3618339595085638,
      "learning_rate": 0.00020983357650890106,
      "loss": 0.4492,
      "step": 5971
    },
    {
      "epoch": 2.8103529411764705,
      "grad_norm": 0.36301981642528525,
      "learning_rate": 0.00020979337624360336,
      "loss": 0.377,
      "step": 5972
    },
    {
      "epoch": 2.8108235294117647,
      "grad_norm": 0.486331495261555,
      "learning_rate": 0.000209753170871658,
      "loss": 0.5781,
      "step": 5973
    },
    {
      "epoch": 2.811294117647059,
      "grad_norm": 0.15982494942694872,
      "learning_rate": 0.00020971296039649882,
      "loss": 0.0991,
      "step": 5974
    },
    {
      "epoch": 2.8117647058823527,
      "grad_norm": 0.6159241823875691,
      "learning_rate": 0.00020967274482155992,
      "loss": 0.6055,
      "step": 5975
    },
    {
      "epoch": 2.812235294117647,
      "grad_norm": 0.309232584925751,
      "learning_rate": 0.00020963252415027588,
      "loss": 0.3125,
      "step": 5976
    },
    {
      "epoch": 2.812705882352941,
      "grad_norm": 0.34920446741406586,
      "learning_rate": 0.0002095922983860818,
      "loss": 0.4082,
      "step": 5977
    },
    {
      "epoch": 2.8131764705882354,
      "grad_norm": 0.3611012514877233,
      "learning_rate": 0.00020955206753241315,
      "loss": 0.3887,
      "step": 5978
    },
    {
      "epoch": 2.8136470588235296,
      "grad_norm": 0.28380506781077663,
      "learning_rate": 0.00020951183159270586,
      "loss": 0.1592,
      "step": 5979
    },
    {
      "epoch": 2.8141176470588234,
      "grad_norm": 0.36015327834156385,
      "learning_rate": 0.0002094715905703962,
      "loss": 0.1973,
      "step": 5980
    },
    {
      "epoch": 2.8145882352941176,
      "grad_norm": 0.3809574122266932,
      "learning_rate": 0.00020943134446892102,
      "loss": 0.4102,
      "step": 5981
    },
    {
      "epoch": 2.815058823529412,
      "grad_norm": 0.3296851734424861,
      "learning_rate": 0.00020939109329171752,
      "loss": 0.1973,
      "step": 5982
    },
    {
      "epoch": 2.8155294117647056,
      "grad_norm": 0.5002839515456352,
      "learning_rate": 0.00020935083704222331,
      "loss": 0.5469,
      "step": 5983
    },
    {
      "epoch": 2.816,
      "grad_norm": 14.979693638134776,
      "learning_rate": 0.00020931057572387655,
      "loss": 11.5,
      "step": 5984
    },
    {
      "epoch": 2.816470588235294,
      "grad_norm": 1.3799334660592306,
      "learning_rate": 0.00020927030934011568,
      "loss": 0.3828,
      "step": 5985
    },
    {
      "epoch": 2.8169411764705883,
      "grad_norm": 17.504013178085774,
      "learning_rate": 0.00020923003789437967,
      "loss": 1.9531,
      "step": 5986
    },
    {
      "epoch": 2.8174117647058825,
      "grad_norm": 0.853522028541606,
      "learning_rate": 0.00020918976139010784,
      "loss": 0.6797,
      "step": 5987
    },
    {
      "epoch": 2.8178823529411767,
      "grad_norm": 0.9753029455866288,
      "learning_rate": 0.00020914947983074016,
      "loss": 0.6172,
      "step": 5988
    },
    {
      "epoch": 2.8183529411764705,
      "grad_norm": 0.36175966095430784,
      "learning_rate": 0.00020910919321971675,
      "loss": 0.7227,
      "step": 5989
    },
    {
      "epoch": 2.8188235294117647,
      "grad_norm": 0.46802845273249505,
      "learning_rate": 0.00020906890156047826,
      "loss": 0.5273,
      "step": 5990
    },
    {
      "epoch": 2.819294117647059,
      "grad_norm": 0.44987782301070106,
      "learning_rate": 0.0002090286048564659,
      "loss": 0.3086,
      "step": 5991
    },
    {
      "epoch": 2.8197647058823527,
      "grad_norm": 0.5069352159617397,
      "learning_rate": 0.00020898830311112118,
      "loss": 0.543,
      "step": 5992
    },
    {
      "epoch": 2.820235294117647,
      "grad_norm": 0.8570809348537196,
      "learning_rate": 0.00020894799632788596,
      "loss": 0.8203,
      "step": 5993
    },
    {
      "epoch": 2.820705882352941,
      "grad_norm": 1.2269192976015122,
      "learning_rate": 0.0002089076845102028,
      "loss": 0.2793,
      "step": 5994
    },
    {
      "epoch": 2.8211764705882354,
      "grad_norm": 160.35480568884634,
      "learning_rate": 0.00020886736766151438,
      "loss": 1.1875,
      "step": 5995
    },
    {
      "epoch": 2.8216470588235296,
      "grad_norm": 7.466885353328061,
      "learning_rate": 0.00020882704578526406,
      "loss": 1.3281,
      "step": 5996
    },
    {
      "epoch": 2.8221176470588234,
      "grad_norm": 2.426158070653201,
      "learning_rate": 0.0002087867188848954,
      "loss": 0.4805,
      "step": 5997
    },
    {
      "epoch": 2.8225882352941176,
      "grad_norm": 0.7739853349431269,
      "learning_rate": 0.0002087463869638527,
      "loss": 0.5664,
      "step": 5998
    },
    {
      "epoch": 2.823058823529412,
      "grad_norm": 0.4016060371265751,
      "learning_rate": 0.00020870605002558038,
      "loss": 0.4648,
      "step": 5999
    },
    {
      "epoch": 2.8235294117647056,
      "grad_norm": 0.7532047759780836,
      "learning_rate": 0.00020866570807352337,
      "loss": 0.2266,
      "step": 6000
    },
    {
      "epoch": 2.824,
      "grad_norm": 3.0975291212491376,
      "learning_rate": 0.0002086253611111272,
      "loss": 0.418,
      "step": 6001
    },
    {
      "epoch": 2.824470588235294,
      "grad_norm": 1.0124899859313554,
      "learning_rate": 0.00020858500914183763,
      "loss": 0.6914,
      "step": 6002
    },
    {
      "epoch": 2.8249411764705883,
      "grad_norm": 0.5600165030070815,
      "learning_rate": 0.0002085446521691009,
      "loss": 0.4062,
      "step": 6003
    },
    {
      "epoch": 2.8254117647058825,
      "grad_norm": 0.4157850173642932,
      "learning_rate": 0.00020850429019636372,
      "loss": 0.2012,
      "step": 6004
    },
    {
      "epoch": 2.8258823529411767,
      "grad_norm": 0.43664245329690965,
      "learning_rate": 0.00020846392322707323,
      "loss": 0.2949,
      "step": 6005
    },
    {
      "epoch": 2.8263529411764705,
      "grad_norm": 0.3646766332211682,
      "learning_rate": 0.00020842355126467693,
      "loss": 0.1377,
      "step": 6006
    },
    {
      "epoch": 2.8268235294117647,
      "grad_norm": 0.4430250642304003,
      "learning_rate": 0.00020838317431262274,
      "loss": 0.2207,
      "step": 6007
    },
    {
      "epoch": 2.827294117647059,
      "grad_norm": 0.4813182810807427,
      "learning_rate": 0.0002083427923743591,
      "loss": 0.1777,
      "step": 6008
    },
    {
      "epoch": 2.8277647058823527,
      "grad_norm": 0.5674292848841483,
      "learning_rate": 0.00020830240545333489,
      "loss": 0.3613,
      "step": 6009
    },
    {
      "epoch": 2.828235294117647,
      "grad_norm": 0.6478707025187849,
      "learning_rate": 0.00020826201355299918,
      "loss": 0.7148,
      "step": 6010
    },
    {
      "epoch": 2.828705882352941,
      "grad_norm": 0.5703632244732422,
      "learning_rate": 0.0002082216166768018,
      "loss": 0.3652,
      "step": 6011
    },
    {
      "epoch": 2.8291764705882354,
      "grad_norm": 0.5331414029061782,
      "learning_rate": 0.00020818121482819277,
      "loss": 0.6094,
      "step": 6012
    },
    {
      "epoch": 2.8296470588235296,
      "grad_norm": 0.4657525675520976,
      "learning_rate": 0.00020814080801062263,
      "loss": 0.4395,
      "step": 6013
    },
    {
      "epoch": 2.8301176470588234,
      "grad_norm": 0.46474834644207136,
      "learning_rate": 0.00020810039622754228,
      "loss": 0.4082,
      "step": 6014
    },
    {
      "epoch": 2.8305882352941176,
      "grad_norm": 0.6559812523554689,
      "learning_rate": 0.00020805997948240314,
      "loss": 0.4707,
      "step": 6015
    },
    {
      "epoch": 2.831058823529412,
      "grad_norm": 0.44214153877774487,
      "learning_rate": 0.00020801955777865694,
      "loss": 0.3184,
      "step": 6016
    },
    {
      "epoch": 2.8315294117647056,
      "grad_norm": 0.31995984345821493,
      "learning_rate": 0.00020797913111975593,
      "loss": 0.1504,
      "step": 6017
    },
    {
      "epoch": 2.832,
      "grad_norm": 0.7500681055887358,
      "learning_rate": 0.00020793869950915276,
      "loss": 0.625,
      "step": 6018
    },
    {
      "epoch": 2.832470588235294,
      "grad_norm": 0.4409748469432246,
      "learning_rate": 0.00020789826295030046,
      "loss": 0.5312,
      "step": 6019
    },
    {
      "epoch": 2.8329411764705883,
      "grad_norm": 0.35662235354073,
      "learning_rate": 0.0002078578214466525,
      "loss": 0.4688,
      "step": 6020
    },
    {
      "epoch": 2.8334117647058825,
      "grad_norm": 1.6390917270530598,
      "learning_rate": 0.00020781737500166279,
      "loss": 0.7188,
      "step": 6021
    },
    {
      "epoch": 2.8338823529411767,
      "grad_norm": 0.46656967051245735,
      "learning_rate": 0.00020777692361878569,
      "loss": 0.3906,
      "step": 6022
    },
    {
      "epoch": 2.8343529411764705,
      "grad_norm": 0.2444360116128899,
      "learning_rate": 0.0002077364673014759,
      "loss": 0.1582,
      "step": 6023
    },
    {
      "epoch": 2.8348235294117647,
      "grad_norm": 0.5193546527604246,
      "learning_rate": 0.00020769600605318857,
      "loss": 0.5898,
      "step": 6024
    },
    {
      "epoch": 2.835294117647059,
      "grad_norm": 0.8523138111013165,
      "learning_rate": 0.0002076555398773794,
      "loss": 0.7773,
      "step": 6025
    },
    {
      "epoch": 2.8357647058823527,
      "grad_norm": 0.6524254099946037,
      "learning_rate": 0.00020761506877750428,
      "loss": 0.9766,
      "step": 6026
    },
    {
      "epoch": 2.836235294117647,
      "grad_norm": 0.6983840962110687,
      "learning_rate": 0.00020757459275701966,
      "loss": 0.5625,
      "step": 6027
    },
    {
      "epoch": 2.836705882352941,
      "grad_norm": 0.5846693816203493,
      "learning_rate": 0.0002075341118193825,
      "loss": 0.5508,
      "step": 6028
    },
    {
      "epoch": 2.8371764705882354,
      "grad_norm": 0.4929778394446135,
      "learning_rate": 0.00020749362596804994,
      "loss": 0.3984,
      "step": 6029
    },
    {
      "epoch": 2.8376470588235296,
      "grad_norm": 0.7153515284498538,
      "learning_rate": 0.00020745313520647971,
      "loss": 0.6797,
      "step": 6030
    },
    {
      "epoch": 2.8381176470588234,
      "grad_norm": 0.5552048275631905,
      "learning_rate": 0.00020741263953812992,
      "loss": 0.8633,
      "step": 6031
    },
    {
      "epoch": 2.8385882352941176,
      "grad_norm": 0.35968699395608555,
      "learning_rate": 0.0002073721389664591,
      "loss": 0.3086,
      "step": 6032
    },
    {
      "epoch": 2.839058823529412,
      "grad_norm": 0.47139447784612715,
      "learning_rate": 0.00020733163349492625,
      "loss": 0.4043,
      "step": 6033
    },
    {
      "epoch": 2.8395294117647056,
      "grad_norm": 0.6417074768735125,
      "learning_rate": 0.00020729112312699066,
      "loss": 0.7891,
      "step": 6034
    },
    {
      "epoch": 2.84,
      "grad_norm": 0.34081379445643634,
      "learning_rate": 0.00020725060786611212,
      "loss": 0.1768,
      "step": 6035
    },
    {
      "epoch": 2.840470588235294,
      "grad_norm": 0.46616951015810837,
      "learning_rate": 0.0002072100877157509,
      "loss": 0.3516,
      "step": 6036
    },
    {
      "epoch": 2.8409411764705883,
      "grad_norm": 0.6313629609776432,
      "learning_rate": 0.00020716956267936759,
      "loss": 0.6797,
      "step": 6037
    },
    {
      "epoch": 2.8414117647058825,
      "grad_norm": 0.35993378173384705,
      "learning_rate": 0.0002071290327604231,
      "loss": 0.1904,
      "step": 6038
    },
    {
      "epoch": 2.8418823529411767,
      "grad_norm": 0.41439894906456687,
      "learning_rate": 0.00020708849796237914,
      "loss": 0.3379,
      "step": 6039
    },
    {
      "epoch": 2.8423529411764705,
      "grad_norm": 0.8214066429577612,
      "learning_rate": 0.00020704795828869736,
      "loss": 1.0078,
      "step": 6040
    },
    {
      "epoch": 2.8428235294117647,
      "grad_norm": 0.3596382441934766,
      "learning_rate": 0.00020700741374284014,
      "loss": 0.1787,
      "step": 6041
    },
    {
      "epoch": 2.843294117647059,
      "grad_norm": 0.40552716001594497,
      "learning_rate": 0.00020696686432827017,
      "loss": 0.3047,
      "step": 6042
    },
    {
      "epoch": 2.8437647058823528,
      "grad_norm": 0.4746065561096378,
      "learning_rate": 0.00020692631004845055,
      "loss": 0.416,
      "step": 6043
    },
    {
      "epoch": 2.844235294117647,
      "grad_norm": 0.35476563347432016,
      "learning_rate": 0.00020688575090684484,
      "loss": 0.3145,
      "step": 6044
    },
    {
      "epoch": 2.844705882352941,
      "grad_norm": 0.47561845409810144,
      "learning_rate": 0.00020684518690691696,
      "loss": 0.3848,
      "step": 6045
    },
    {
      "epoch": 2.8451764705882354,
      "grad_norm": 0.4684732661325496,
      "learning_rate": 0.00020680461805213133,
      "loss": 0.6562,
      "step": 6046
    },
    {
      "epoch": 2.8456470588235296,
      "grad_norm": 0.48707192605449545,
      "learning_rate": 0.00020676404434595265,
      "loss": 0.625,
      "step": 6047
    },
    {
      "epoch": 2.8461176470588234,
      "grad_norm": 0.3201311827600174,
      "learning_rate": 0.00020672346579184617,
      "loss": 0.2441,
      "step": 6048
    },
    {
      "epoch": 2.8465882352941176,
      "grad_norm": 0.3941079166796153,
      "learning_rate": 0.00020668288239327753,
      "loss": 0.2559,
      "step": 6049
    },
    {
      "epoch": 2.847058823529412,
      "grad_norm": 0.4038173790355036,
      "learning_rate": 0.00020664229415371266,
      "loss": 0.1895,
      "step": 6050
    },
    {
      "epoch": 2.8475294117647056,
      "grad_norm": 0.3264052937556375,
      "learning_rate": 0.00020660170107661803,
      "loss": 0.2402,
      "step": 6051
    },
    {
      "epoch": 2.848,
      "grad_norm": 0.42692780343603876,
      "learning_rate": 0.00020656110316546054,
      "loss": 0.2715,
      "step": 6052
    },
    {
      "epoch": 2.848470588235294,
      "grad_norm": 0.5469827216165865,
      "learning_rate": 0.00020652050042370743,
      "loss": 0.6055,
      "step": 6053
    },
    {
      "epoch": 2.8489411764705883,
      "grad_norm": 0.5239225858607279,
      "learning_rate": 0.0002064798928548263,
      "loss": 0.4766,
      "step": 6054
    },
    {
      "epoch": 2.8494117647058825,
      "grad_norm": 0.409786741572499,
      "learning_rate": 0.00020643928046228525,
      "loss": 0.4453,
      "step": 6055
    },
    {
      "epoch": 2.8498823529411763,
      "grad_norm": 0.2953622353312739,
      "learning_rate": 0.00020639866324955285,
      "loss": 0.3105,
      "step": 6056
    },
    {
      "epoch": 2.8503529411764705,
      "grad_norm": 0.41213505924221955,
      "learning_rate": 0.00020635804122009802,
      "loss": 0.3047,
      "step": 6057
    },
    {
      "epoch": 2.8508235294117648,
      "grad_norm": 0.4233404028728083,
      "learning_rate": 0.00020631741437738998,
      "loss": 0.5,
      "step": 6058
    },
    {
      "epoch": 2.851294117647059,
      "grad_norm": 0.2503487218173818,
      "learning_rate": 0.00020627678272489854,
      "loss": 0.2471,
      "step": 6059
    },
    {
      "epoch": 2.8517647058823528,
      "grad_norm": 0.3270974819301623,
      "learning_rate": 0.00020623614626609385,
      "loss": 0.2461,
      "step": 6060
    },
    {
      "epoch": 2.852235294117647,
      "grad_norm": 0.4679489768126148,
      "learning_rate": 0.0002061955050044464,
      "loss": 0.3301,
      "step": 6061
    },
    {
      "epoch": 2.852705882352941,
      "grad_norm": 0.5121971368157111,
      "learning_rate": 0.00020615485894342716,
      "loss": 0.7148,
      "step": 6062
    },
    {
      "epoch": 2.8531764705882354,
      "grad_norm": 0.7664600487909845,
      "learning_rate": 0.00020611420808650758,
      "loss": 0.8828,
      "step": 6063
    },
    {
      "epoch": 2.8536470588235296,
      "grad_norm": 0.623731133627249,
      "learning_rate": 0.00020607355243715936,
      "loss": 0.6758,
      "step": 6064
    },
    {
      "epoch": 2.8541176470588234,
      "grad_norm": 0.4656636363378075,
      "learning_rate": 0.00020603289199885473,
      "loss": 0.416,
      "step": 6065
    },
    {
      "epoch": 2.8545882352941176,
      "grad_norm": 0.3066864788088581,
      "learning_rate": 0.00020599222677506628,
      "loss": 0.1885,
      "step": 6066
    },
    {
      "epoch": 2.855058823529412,
      "grad_norm": 0.2815912545736045,
      "learning_rate": 0.00020595155676926704,
      "loss": 0.1006,
      "step": 6067
    },
    {
      "epoch": 2.8555294117647056,
      "grad_norm": 0.2455130980124076,
      "learning_rate": 0.0002059108819849304,
      "loss": 0.2969,
      "step": 6068
    },
    {
      "epoch": 2.856,
      "grad_norm": 0.6090199425168622,
      "learning_rate": 0.0002058702024255302,
      "loss": 0.5,
      "step": 6069
    },
    {
      "epoch": 2.856470588235294,
      "grad_norm": 0.42676326241942575,
      "learning_rate": 0.00020582951809454067,
      "loss": 0.4023,
      "step": 6070
    },
    {
      "epoch": 2.8569411764705883,
      "grad_norm": 0.5642284691437058,
      "learning_rate": 0.00020578882899543644,
      "loss": 0.4492,
      "step": 6071
    },
    {
      "epoch": 2.8574117647058825,
      "grad_norm": 0.31365759787339814,
      "learning_rate": 0.0002057481351316926,
      "loss": 0.2422,
      "step": 6072
    },
    {
      "epoch": 2.8578823529411763,
      "grad_norm": 0.547188185056948,
      "learning_rate": 0.00020570743650678454,
      "loss": 0.6016,
      "step": 6073
    },
    {
      "epoch": 2.8583529411764705,
      "grad_norm": 0.6617172570116137,
      "learning_rate": 0.00020566673312418817,
      "loss": 0.4707,
      "step": 6074
    },
    {
      "epoch": 2.8588235294117648,
      "grad_norm": 0.2923351619727783,
      "learning_rate": 0.00020562602498737975,
      "loss": 0.2305,
      "step": 6075
    },
    {
      "epoch": 2.859294117647059,
      "grad_norm": 0.5441200557465111,
      "learning_rate": 0.00020558531209983593,
      "loss": 0.5938,
      "step": 6076
    },
    {
      "epoch": 2.8597647058823528,
      "grad_norm": 0.5104221535627136,
      "learning_rate": 0.00020554459446503386,
      "loss": 0.3047,
      "step": 6077
    },
    {
      "epoch": 2.860235294117647,
      "grad_norm": 0.39585967031320846,
      "learning_rate": 0.00020550387208645093,
      "loss": 0.2012,
      "step": 6078
    },
    {
      "epoch": 2.860705882352941,
      "grad_norm": 0.2837931663238173,
      "learning_rate": 0.0002054631449675651,
      "loss": 0.2432,
      "step": 6079
    },
    {
      "epoch": 2.8611764705882354,
      "grad_norm": 0.4852358727809272,
      "learning_rate": 0.00020542241311185467,
      "loss": 0.582,
      "step": 6080
    },
    {
      "epoch": 2.8616470588235297,
      "grad_norm": 0.2881589821150506,
      "learning_rate": 0.0002053816765227983,
      "loss": 0.3555,
      "step": 6081
    },
    {
      "epoch": 2.8621176470588234,
      "grad_norm": 0.355552519096458,
      "learning_rate": 0.0002053409352038751,
      "loss": 0.3145,
      "step": 6082
    },
    {
      "epoch": 2.8625882352941177,
      "grad_norm": 0.36579731529206466,
      "learning_rate": 0.00020530018915856457,
      "loss": 0.2559,
      "step": 6083
    },
    {
      "epoch": 2.863058823529412,
      "grad_norm": 0.3060204275412933,
      "learning_rate": 0.00020525943839034668,
      "loss": 0.1543,
      "step": 6084
    },
    {
      "epoch": 2.8635294117647057,
      "grad_norm": 0.4842367397124674,
      "learning_rate": 0.0002052186829027017,
      "loss": 0.2969,
      "step": 6085
    },
    {
      "epoch": 2.864,
      "grad_norm": 0.5544817286406211,
      "learning_rate": 0.00020517792269911033,
      "loss": 0.5781,
      "step": 6086
    },
    {
      "epoch": 2.864470588235294,
      "grad_norm": 0.382934542804686,
      "learning_rate": 0.0002051371577830537,
      "loss": 0.207,
      "step": 6087
    },
    {
      "epoch": 2.8649411764705883,
      "grad_norm": 0.44684784771345276,
      "learning_rate": 0.00020509638815801342,
      "loss": 0.4062,
      "step": 6088
    },
    {
      "epoch": 2.8654117647058825,
      "grad_norm": 0.3537996595723599,
      "learning_rate": 0.00020505561382747127,
      "loss": 0.2197,
      "step": 6089
    },
    {
      "epoch": 2.8658823529411763,
      "grad_norm": 0.41581689334960403,
      "learning_rate": 0.00020501483479490968,
      "loss": 0.2773,
      "step": 6090
    },
    {
      "epoch": 2.8663529411764705,
      "grad_norm": 0.30668448465922077,
      "learning_rate": 0.00020497405106381138,
      "loss": 0.3984,
      "step": 6091
    },
    {
      "epoch": 2.8668235294117648,
      "grad_norm": 0.5113049995838701,
      "learning_rate": 0.0002049332626376594,
      "loss": 0.7227,
      "step": 6092
    },
    {
      "epoch": 2.8672941176470585,
      "grad_norm": 0.4738197836890734,
      "learning_rate": 0.00020489246951993742,
      "loss": 0.4062,
      "step": 6093
    },
    {
      "epoch": 2.8677647058823528,
      "grad_norm": 0.31138762412721016,
      "learning_rate": 0.00020485167171412926,
      "loss": 0.1436,
      "step": 6094
    },
    {
      "epoch": 2.868235294117647,
      "grad_norm": 0.6097307758371742,
      "learning_rate": 0.00020481086922371935,
      "loss": 0.5586,
      "step": 6095
    },
    {
      "epoch": 2.868705882352941,
      "grad_norm": 0.2684554462426144,
      "learning_rate": 0.00020477006205219228,
      "loss": 0.1191,
      "step": 6096
    },
    {
      "epoch": 2.8691764705882354,
      "grad_norm": 0.4505899516729099,
      "learning_rate": 0.0002047292502030333,
      "loss": 0.4434,
      "step": 6097
    },
    {
      "epoch": 2.8696470588235297,
      "grad_norm": 0.2844694391606763,
      "learning_rate": 0.00020468843367972786,
      "loss": 0.2178,
      "step": 6098
    },
    {
      "epoch": 2.8701176470588234,
      "grad_norm": 0.6057489217868705,
      "learning_rate": 0.00020464761248576203,
      "loss": 0.6875,
      "step": 6099
    },
    {
      "epoch": 2.8705882352941177,
      "grad_norm": 0.8175366926804607,
      "learning_rate": 0.00020460678662462194,
      "loss": 1.0547,
      "step": 6100
    },
    {
      "epoch": 2.871058823529412,
      "grad_norm": 0.3807798026259815,
      "learning_rate": 0.0002045659560997945,
      "loss": 0.4395,
      "step": 6101
    },
    {
      "epoch": 2.8715294117647057,
      "grad_norm": 0.44374528981923717,
      "learning_rate": 0.00020452512091476678,
      "loss": 0.3027,
      "step": 6102
    },
    {
      "epoch": 2.872,
      "grad_norm": 0.544644547531857,
      "learning_rate": 0.0002044842810730262,
      "loss": 0.6602,
      "step": 6103
    },
    {
      "epoch": 2.872470588235294,
      "grad_norm": 0.51798001049649,
      "learning_rate": 0.00020444343657806087,
      "loss": 0.6992,
      "step": 6104
    },
    {
      "epoch": 2.8729411764705883,
      "grad_norm": 0.32160152033373296,
      "learning_rate": 0.000204402587433359,
      "loss": 0.2715,
      "step": 6105
    },
    {
      "epoch": 2.8734117647058826,
      "grad_norm": 0.3941962649644655,
      "learning_rate": 0.0002043617336424093,
      "loss": 0.1865,
      "step": 6106
    },
    {
      "epoch": 2.8738823529411763,
      "grad_norm": 0.898230746000177,
      "learning_rate": 0.00020432087520870086,
      "loss": 1.25,
      "step": 6107
    },
    {
      "epoch": 2.8743529411764706,
      "grad_norm": 0.4423915580704798,
      "learning_rate": 0.0002042800121357233,
      "loss": 0.2422,
      "step": 6108
    },
    {
      "epoch": 2.8748235294117648,
      "grad_norm": 0.37320398425437207,
      "learning_rate": 0.00020423914442696642,
      "loss": 0.3418,
      "step": 6109
    },
    {
      "epoch": 2.8752941176470586,
      "grad_norm": 0.5339818795373303,
      "learning_rate": 0.00020419827208592058,
      "loss": 0.4297,
      "step": 6110
    },
    {
      "epoch": 2.875764705882353,
      "grad_norm": 0.4942913657411978,
      "learning_rate": 0.00020415739511607653,
      "loss": 0.6406,
      "step": 6111
    },
    {
      "epoch": 2.876235294117647,
      "grad_norm": 0.5305191293891404,
      "learning_rate": 0.0002041165135209252,
      "loss": 0.3496,
      "step": 6112
    },
    {
      "epoch": 2.8767058823529412,
      "grad_norm": 0.25429848779104236,
      "learning_rate": 0.0002040756273039582,
      "loss": 0.104,
      "step": 6113
    },
    {
      "epoch": 2.8771764705882354,
      "grad_norm": 0.3769584299847467,
      "learning_rate": 0.0002040347364686674,
      "loss": 0.208,
      "step": 6114
    },
    {
      "epoch": 2.8776470588235297,
      "grad_norm": 0.8456947949582442,
      "learning_rate": 0.00020399384101854506,
      "loss": 0.5312,
      "step": 6115
    },
    {
      "epoch": 2.8781176470588234,
      "grad_norm": 0.4843860617202702,
      "learning_rate": 0.00020395294095708387,
      "loss": 0.375,
      "step": 6116
    },
    {
      "epoch": 2.8785882352941177,
      "grad_norm": 0.6487423258109585,
      "learning_rate": 0.00020391203628777687,
      "loss": 0.8438,
      "step": 6117
    },
    {
      "epoch": 2.879058823529412,
      "grad_norm": 0.5998376963180221,
      "learning_rate": 0.00020387112701411754,
      "loss": 0.6367,
      "step": 6118
    },
    {
      "epoch": 2.8795294117647057,
      "grad_norm": 0.2594963067821364,
      "learning_rate": 0.00020383021313959972,
      "loss": 0.1123,
      "step": 6119
    },
    {
      "epoch": 2.88,
      "grad_norm": 0.4104791909553077,
      "learning_rate": 0.00020378929466771767,
      "loss": 0.2695,
      "step": 6120
    },
    {
      "epoch": 2.880470588235294,
      "grad_norm": 0.6438277197048915,
      "learning_rate": 0.000203748371601966,
      "loss": 0.6719,
      "step": 6121
    },
    {
      "epoch": 2.8809411764705883,
      "grad_norm": 0.48224335555633274,
      "learning_rate": 0.0002037074439458398,
      "loss": 0.4941,
      "step": 6122
    },
    {
      "epoch": 2.8814117647058826,
      "grad_norm": 0.5237913535304509,
      "learning_rate": 0.0002036665117028344,
      "loss": 0.4434,
      "step": 6123
    },
    {
      "epoch": 2.8818823529411763,
      "grad_norm": 0.5140802838047484,
      "learning_rate": 0.0002036255748764457,
      "loss": 0.6289,
      "step": 6124
    },
    {
      "epoch": 2.8823529411764706,
      "grad_norm": 0.5496309296115075,
      "learning_rate": 0.00020358463347016988,
      "loss": 0.5391,
      "step": 6125
    },
    {
      "epoch": 2.882823529411765,
      "grad_norm": 0.4324439037817502,
      "learning_rate": 0.00020354368748750357,
      "loss": 0.4805,
      "step": 6126
    },
    {
      "epoch": 2.8832941176470586,
      "grad_norm": 0.5350686385395483,
      "learning_rate": 0.00020350273693194363,
      "loss": 0.4707,
      "step": 6127
    },
    {
      "epoch": 2.883764705882353,
      "grad_norm": 0.394014414719312,
      "learning_rate": 0.00020346178180698758,
      "loss": 0.5117,
      "step": 6128
    },
    {
      "epoch": 2.884235294117647,
      "grad_norm": 0.4711700074880198,
      "learning_rate": 0.00020342082211613312,
      "loss": 0.4668,
      "step": 6129
    },
    {
      "epoch": 2.8847058823529412,
      "grad_norm": 0.1889004157789404,
      "learning_rate": 0.00020337985786287842,
      "loss": 0.0942,
      "step": 6130
    },
    {
      "epoch": 2.8851764705882355,
      "grad_norm": 0.3543645921252973,
      "learning_rate": 0.00020333888905072208,
      "loss": 0.2227,
      "step": 6131
    },
    {
      "epoch": 2.8856470588235297,
      "grad_norm": 0.733016106707705,
      "learning_rate": 0.00020329791568316294,
      "loss": 0.4629,
      "step": 6132
    },
    {
      "epoch": 2.8861176470588235,
      "grad_norm": 0.3156268796504413,
      "learning_rate": 0.00020325693776370037,
      "loss": 0.1895,
      "step": 6133
    },
    {
      "epoch": 2.8865882352941177,
      "grad_norm": 0.5150203087869527,
      "learning_rate": 0.0002032159552958341,
      "loss": 0.8789,
      "step": 6134
    },
    {
      "epoch": 2.887058823529412,
      "grad_norm": 0.7258141477996896,
      "learning_rate": 0.00020317496828306427,
      "loss": 0.3926,
      "step": 6135
    },
    {
      "epoch": 2.8875294117647057,
      "grad_norm": 0.26026533598558893,
      "learning_rate": 0.00020313397672889132,
      "loss": 0.2715,
      "step": 6136
    },
    {
      "epoch": 2.888,
      "grad_norm": 0.3849306250517165,
      "learning_rate": 0.00020309298063681607,
      "loss": 0.3105,
      "step": 6137
    },
    {
      "epoch": 2.888470588235294,
      "grad_norm": 0.37897731603693025,
      "learning_rate": 0.00020305198001033988,
      "loss": 0.209,
      "step": 6138
    },
    {
      "epoch": 2.8889411764705883,
      "grad_norm": 0.6656904950064902,
      "learning_rate": 0.00020301097485296443,
      "loss": 1.0547,
      "step": 6139
    },
    {
      "epoch": 2.8894117647058826,
      "grad_norm": 0.5066014239401345,
      "learning_rate": 0.00020296996516819163,
      "loss": 0.6211,
      "step": 6140
    },
    {
      "epoch": 2.8898823529411763,
      "grad_norm": 0.4310627894473288,
      "learning_rate": 0.00020292895095952402,
      "loss": 0.4023,
      "step": 6141
    },
    {
      "epoch": 2.8903529411764706,
      "grad_norm": 0.35746114963639325,
      "learning_rate": 0.0002028879322304644,
      "loss": 0.3926,
      "step": 6142
    },
    {
      "epoch": 2.890823529411765,
      "grad_norm": 0.31277590305807923,
      "learning_rate": 0.00020284690898451593,
      "loss": 0.1641,
      "step": 6143
    },
    {
      "epoch": 2.8912941176470586,
      "grad_norm": 0.4724731199360303,
      "learning_rate": 0.00020280588122518223,
      "loss": 0.2949,
      "step": 6144
    },
    {
      "epoch": 2.891764705882353,
      "grad_norm": 0.3765020832351325,
      "learning_rate": 0.00020276484895596726,
      "loss": 0.2871,
      "step": 6145
    },
    {
      "epoch": 2.892235294117647,
      "grad_norm": 0.2248738349339727,
      "learning_rate": 0.00020272381218037542,
      "loss": 0.1338,
      "step": 6146
    },
    {
      "epoch": 2.8927058823529412,
      "grad_norm": 0.46001802333481084,
      "learning_rate": 0.00020268277090191133,
      "loss": 0.4043,
      "step": 6147
    },
    {
      "epoch": 2.8931764705882355,
      "grad_norm": 0.8504686667151602,
      "learning_rate": 0.00020264172512408026,
      "loss": 0.582,
      "step": 6148
    },
    {
      "epoch": 2.8936470588235297,
      "grad_norm": 0.41100596125459293,
      "learning_rate": 0.00020260067485038767,
      "loss": 0.2852,
      "step": 6149
    },
    {
      "epoch": 2.8941176470588235,
      "grad_norm": 0.3657922427035591,
      "learning_rate": 0.0002025596200843394,
      "loss": 0.2715,
      "step": 6150
    },
    {
      "epoch": 2.8945882352941177,
      "grad_norm": 0.4765633823430536,
      "learning_rate": 0.00020251856082944178,
      "loss": 0.707,
      "step": 6151
    },
    {
      "epoch": 2.895058823529412,
      "grad_norm": 0.397879623746015,
      "learning_rate": 0.00020247749708920146,
      "loss": 0.3672,
      "step": 6152
    },
    {
      "epoch": 2.8955294117647057,
      "grad_norm": 0.4363532657982845,
      "learning_rate": 0.00020243642886712553,
      "loss": 0.5977,
      "step": 6153
    },
    {
      "epoch": 2.896,
      "grad_norm": 0.9608368345305981,
      "learning_rate": 0.0002023953561667213,
      "loss": 0.7891,
      "step": 6154
    },
    {
      "epoch": 2.896470588235294,
      "grad_norm": 0.37407402259712286,
      "learning_rate": 0.0002023542789914967,
      "loss": 0.1787,
      "step": 6155
    },
    {
      "epoch": 2.8969411764705884,
      "grad_norm": 0.3279830163758217,
      "learning_rate": 0.0002023131973449599,
      "loss": 0.25,
      "step": 6156
    },
    {
      "epoch": 2.8974117647058826,
      "grad_norm": 0.46386756935580964,
      "learning_rate": 0.00020227211123061942,
      "loss": 0.5664,
      "step": 6157
    },
    {
      "epoch": 2.8978823529411764,
      "grad_norm": 0.6077997387151325,
      "learning_rate": 0.00020223102065198426,
      "loss": 0.6523,
      "step": 6158
    },
    {
      "epoch": 2.8983529411764706,
      "grad_norm": 0.13147067054299727,
      "learning_rate": 0.00020218992561256375,
      "loss": 0.1074,
      "step": 6159
    },
    {
      "epoch": 2.898823529411765,
      "grad_norm": 0.6851581007650299,
      "learning_rate": 0.0002021488261158676,
      "loss": 0.4062,
      "step": 6160
    },
    {
      "epoch": 2.8992941176470586,
      "grad_norm": 0.28569711562668504,
      "learning_rate": 0.0002021077221654059,
      "loss": 0.1602,
      "step": 6161
    },
    {
      "epoch": 2.899764705882353,
      "grad_norm": 0.6938382695198518,
      "learning_rate": 0.00020206661376468912,
      "loss": 0.5742,
      "step": 6162
    },
    {
      "epoch": 2.900235294117647,
      "grad_norm": 0.6283886479216665,
      "learning_rate": 0.00020202550091722818,
      "loss": 0.6484,
      "step": 6163
    },
    {
      "epoch": 2.9007058823529412,
      "grad_norm": 0.36882368305268454,
      "learning_rate": 0.00020198438362653426,
      "loss": 0.208,
      "step": 6164
    },
    {
      "epoch": 2.9011764705882355,
      "grad_norm": 0.7319770350315087,
      "learning_rate": 0.00020194326189611897,
      "loss": 0.668,
      "step": 6165
    },
    {
      "epoch": 2.9016470588235292,
      "grad_norm": 0.4511072808152249,
      "learning_rate": 0.00020190213572949435,
      "loss": 0.2432,
      "step": 6166
    },
    {
      "epoch": 2.9021176470588235,
      "grad_norm": 0.22462737484511386,
      "learning_rate": 0.0002018610051301728,
      "loss": 0.051,
      "step": 6167
    },
    {
      "epoch": 2.9025882352941177,
      "grad_norm": 0.49831619436967,
      "learning_rate": 0.00020181987010166699,
      "loss": 0.4922,
      "step": 6168
    },
    {
      "epoch": 2.903058823529412,
      "grad_norm": 0.4949228029396825,
      "learning_rate": 0.0002017787306474901,
      "loss": 0.2031,
      "step": 6169
    },
    {
      "epoch": 2.9035294117647057,
      "grad_norm": 1.7125053112605044,
      "learning_rate": 0.00020173758677115568,
      "loss": 1.5312,
      "step": 6170
    },
    {
      "epoch": 2.904,
      "grad_norm": 0.370014863015624,
      "learning_rate": 0.00020169643847617753,
      "loss": 0.3477,
      "step": 6171
    },
    {
      "epoch": 2.904470588235294,
      "grad_norm": 0.39053018707159143,
      "learning_rate": 0.00020165528576607,
      "loss": 0.1934,
      "step": 6172
    },
    {
      "epoch": 2.9049411764705884,
      "grad_norm": 0.4562108203312454,
      "learning_rate": 0.0002016141286443477,
      "loss": 0.5703,
      "step": 6173
    },
    {
      "epoch": 2.9054117647058826,
      "grad_norm": 0.4520301788610755,
      "learning_rate": 0.0002015729671145256,
      "loss": 0.498,
      "step": 6174
    },
    {
      "epoch": 2.9058823529411764,
      "grad_norm": 0.5035069538113025,
      "learning_rate": 0.0002015318011801192,
      "loss": 0.5547,
      "step": 6175
    },
    {
      "epoch": 2.9063529411764706,
      "grad_norm": 0.46825787983042866,
      "learning_rate": 0.0002014906308446442,
      "loss": 0.4023,
      "step": 6176
    },
    {
      "epoch": 2.906823529411765,
      "grad_norm": 0.25419578250011254,
      "learning_rate": 0.0002014494561116168,
      "loss": 0.2305,
      "step": 6177
    },
    {
      "epoch": 2.9072941176470586,
      "grad_norm": 0.5721694773847535,
      "learning_rate": 0.00020140827698455338,
      "loss": 0.2559,
      "step": 6178
    },
    {
      "epoch": 2.907764705882353,
      "grad_norm": 0.5585384242629117,
      "learning_rate": 0.00020136709346697107,
      "loss": 0.5469,
      "step": 6179
    },
    {
      "epoch": 2.908235294117647,
      "grad_norm": 0.31589837302661583,
      "learning_rate": 0.00020132590556238696,
      "loss": 0.1729,
      "step": 6180
    },
    {
      "epoch": 2.9087058823529413,
      "grad_norm": 0.5591590617805676,
      "learning_rate": 0.00020128471327431877,
      "loss": 0.5352,
      "step": 6181
    },
    {
      "epoch": 2.9091764705882355,
      "grad_norm": 0.38811065590910976,
      "learning_rate": 0.0002012435166062845,
      "loss": 0.2871,
      "step": 6182
    },
    {
      "epoch": 2.9096470588235293,
      "grad_norm": 0.8266349369688737,
      "learning_rate": 0.00020120231556180262,
      "loss": 0.6328,
      "step": 6183
    },
    {
      "epoch": 2.9101176470588235,
      "grad_norm": 0.7618679347421312,
      "learning_rate": 0.00020116111014439178,
      "loss": 0.8008,
      "step": 6184
    },
    {
      "epoch": 2.9105882352941177,
      "grad_norm": 0.4495191713404086,
      "learning_rate": 0.00020111990035757125,
      "loss": 0.4727,
      "step": 6185
    },
    {
      "epoch": 2.911058823529412,
      "grad_norm": 0.31241766790583914,
      "learning_rate": 0.00020107868620486047,
      "loss": 0.209,
      "step": 6186
    },
    {
      "epoch": 2.9115294117647057,
      "grad_norm": 0.287071895163265,
      "learning_rate": 0.0002010374676897793,
      "loss": 0.0938,
      "step": 6187
    },
    {
      "epoch": 2.912,
      "grad_norm": 0.6056256571651228,
      "learning_rate": 0.00020099624481584807,
      "loss": 0.5469,
      "step": 6188
    },
    {
      "epoch": 2.912470588235294,
      "grad_norm": 0.6319988616977638,
      "learning_rate": 0.0002009550175865874,
      "loss": 0.5859,
      "step": 6189
    },
    {
      "epoch": 2.9129411764705884,
      "grad_norm": 0.3790665990800272,
      "learning_rate": 0.00020091378600551834,
      "loss": 0.2188,
      "step": 6190
    },
    {
      "epoch": 2.9134117647058826,
      "grad_norm": 0.15656071799018337,
      "learning_rate": 0.00020087255007616222,
      "loss": 0.025,
      "step": 6191
    },
    {
      "epoch": 2.9138823529411764,
      "grad_norm": 0.38025733506729326,
      "learning_rate": 0.00020083130980204074,
      "loss": 0.2031,
      "step": 6192
    },
    {
      "epoch": 2.9143529411764706,
      "grad_norm": 0.47005593170590537,
      "learning_rate": 0.0002007900651866761,
      "loss": 0.4902,
      "step": 6193
    },
    {
      "epoch": 2.914823529411765,
      "grad_norm": 0.3173463792669277,
      "learning_rate": 0.00020074881623359083,
      "loss": 0.291,
      "step": 6194
    },
    {
      "epoch": 2.9152941176470586,
      "grad_norm": 0.43574100617991063,
      "learning_rate": 0.00020070756294630765,
      "loss": 0.5625,
      "step": 6195
    },
    {
      "epoch": 2.915764705882353,
      "grad_norm": 0.608402274348054,
      "learning_rate": 0.0002006663053283499,
      "loss": 1.0391,
      "step": 6196
    },
    {
      "epoch": 2.916235294117647,
      "grad_norm": 0.3800516645180085,
      "learning_rate": 0.00020062504338324116,
      "loss": 0.3574,
      "step": 6197
    },
    {
      "epoch": 2.9167058823529413,
      "grad_norm": 0.7848388830015287,
      "learning_rate": 0.0002005837771145054,
      "loss": 0.0918,
      "step": 6198
    },
    {
      "epoch": 2.9171764705882355,
      "grad_norm": 0.291385228708208,
      "learning_rate": 0.00020054250652566696,
      "loss": 0.1963,
      "step": 6199
    },
    {
      "epoch": 2.9176470588235293,
      "grad_norm": 0.5947617437789864,
      "learning_rate": 0.0002005012316202506,
      "loss": 0.7188,
      "step": 6200
    },
    {
      "epoch": 2.9181176470588235,
      "grad_norm": 0.46199762729117505,
      "learning_rate": 0.00020045995240178133,
      "loss": 0.6992,
      "step": 6201
    },
    {
      "epoch": 2.9185882352941177,
      "grad_norm": 0.30028923698132387,
      "learning_rate": 0.00020041866887378457,
      "loss": 0.1514,
      "step": 6202
    },
    {
      "epoch": 2.919058823529412,
      "grad_norm": 0.3671505795339758,
      "learning_rate": 0.0002003773810397862,
      "loss": 0.2656,
      "step": 6203
    },
    {
      "epoch": 2.9195294117647057,
      "grad_norm": 0.5442192339793496,
      "learning_rate": 0.00020033608890331246,
      "loss": 0.8516,
      "step": 6204
    },
    {
      "epoch": 2.92,
      "grad_norm": 0.46785097626714023,
      "learning_rate": 0.00020029479246788973,
      "loss": 0.5586,
      "step": 6205
    },
    {
      "epoch": 2.920470588235294,
      "grad_norm": 0.20165401895864052,
      "learning_rate": 0.00020025349173704508,
      "loss": 0.1865,
      "step": 6206
    },
    {
      "epoch": 2.9209411764705884,
      "grad_norm": 0.7062505310660585,
      "learning_rate": 0.00020021218671430573,
      "loss": 0.2695,
      "step": 6207
    },
    {
      "epoch": 2.9214117647058826,
      "grad_norm": 0.24133169638229587,
      "learning_rate": 0.00020017087740319936,
      "loss": 0.0942,
      "step": 6208
    },
    {
      "epoch": 2.9218823529411764,
      "grad_norm": 0.3933794174229782,
      "learning_rate": 0.00020012956380725393,
      "loss": 0.375,
      "step": 6209
    },
    {
      "epoch": 2.9223529411764706,
      "grad_norm": 0.35637827649839,
      "learning_rate": 0.00020008824592999795,
      "loss": 0.1689,
      "step": 6210
    },
    {
      "epoch": 2.922823529411765,
      "grad_norm": 0.27712146290991174,
      "learning_rate": 0.00020004692377496004,
      "loss": 0.1953,
      "step": 6211
    },
    {
      "epoch": 2.9232941176470586,
      "grad_norm": 0.5639160841198524,
      "learning_rate": 0.00020000559734566933,
      "loss": 0.3594,
      "step": 6212
    },
    {
      "epoch": 2.923764705882353,
      "grad_norm": 0.5221662874133302,
      "learning_rate": 0.00019996426664565538,
      "loss": 0.6289,
      "step": 6213
    },
    {
      "epoch": 2.924235294117647,
      "grad_norm": 0.3922649415174257,
      "learning_rate": 0.000199922931678448,
      "loss": 0.2354,
      "step": 6214
    },
    {
      "epoch": 2.9247058823529413,
      "grad_norm": 0.2534320528783178,
      "learning_rate": 0.00019988159244757736,
      "loss": 0.2158,
      "step": 6215
    },
    {
      "epoch": 2.9251764705882355,
      "grad_norm": 0.38638840317729345,
      "learning_rate": 0.000199840248956574,
      "loss": 0.5117,
      "step": 6216
    },
    {
      "epoch": 2.9256470588235293,
      "grad_norm": 0.34148311034080286,
      "learning_rate": 0.000199798901208969,
      "loss": 0.3691,
      "step": 6217
    },
    {
      "epoch": 2.9261176470588235,
      "grad_norm": 0.37326569532624077,
      "learning_rate": 0.0001997575492082935,
      "loss": 0.3672,
      "step": 6218
    },
    {
      "epoch": 2.9265882352941177,
      "grad_norm": 0.3871953879684111,
      "learning_rate": 0.0001997161929580793,
      "loss": 0.1816,
      "step": 6219
    },
    {
      "epoch": 2.9270588235294115,
      "grad_norm": 0.3921603848580387,
      "learning_rate": 0.00019967483246185834,
      "loss": 0.2432,
      "step": 6220
    },
    {
      "epoch": 2.9275294117647057,
      "grad_norm": 0.29923373595077546,
      "learning_rate": 0.00019963346772316307,
      "loss": 0.2158,
      "step": 6221
    },
    {
      "epoch": 2.928,
      "grad_norm": 0.4563160265201589,
      "learning_rate": 0.00019959209874552612,
      "loss": 0.7305,
      "step": 6222
    },
    {
      "epoch": 2.928470588235294,
      "grad_norm": 0.49829711329418724,
      "learning_rate": 0.00019955072553248075,
      "loss": 0.6172,
      "step": 6223
    },
    {
      "epoch": 2.9289411764705884,
      "grad_norm": 0.35723967395039014,
      "learning_rate": 0.0001995093480875604,
      "loss": 0.1816,
      "step": 6224
    },
    {
      "epoch": 2.9294117647058826,
      "grad_norm": 0.37417155926932083,
      "learning_rate": 0.00019946796641429884,
      "loss": 0.6758,
      "step": 6225
    },
    {
      "epoch": 2.9298823529411764,
      "grad_norm": 0.3094256235876702,
      "learning_rate": 0.0001994265805162303,
      "loss": 0.2119,
      "step": 6226
    },
    {
      "epoch": 2.9303529411764706,
      "grad_norm": 0.3282576665402077,
      "learning_rate": 0.00019938519039688937,
      "loss": 0.1226,
      "step": 6227
    },
    {
      "epoch": 2.930823529411765,
      "grad_norm": 0.38169240284890105,
      "learning_rate": 0.00019934379605981097,
      "loss": 0.3008,
      "step": 6228
    },
    {
      "epoch": 2.9312941176470586,
      "grad_norm": 0.5036564242270467,
      "learning_rate": 0.0001993023975085303,
      "loss": 0.4707,
      "step": 6229
    },
    {
      "epoch": 2.931764705882353,
      "grad_norm": 0.4669764401579118,
      "learning_rate": 0.00019926099474658305,
      "loss": 0.6211,
      "step": 6230
    },
    {
      "epoch": 2.932235294117647,
      "grad_norm": 0.30317085704429325,
      "learning_rate": 0.0001992195877775053,
      "loss": 0.3496,
      "step": 6231
    },
    {
      "epoch": 2.9327058823529413,
      "grad_norm": 0.31688490732134955,
      "learning_rate": 0.00019917817660483327,
      "loss": 0.3867,
      "step": 6232
    },
    {
      "epoch": 2.9331764705882355,
      "grad_norm": 0.47621500536772676,
      "learning_rate": 0.00019913676123210373,
      "loss": 0.4355,
      "step": 6233
    },
    {
      "epoch": 2.9336470588235293,
      "grad_norm": 0.3871570343894499,
      "learning_rate": 0.00019909534166285382,
      "loss": 0.5703,
      "step": 6234
    },
    {
      "epoch": 2.9341176470588235,
      "grad_norm": 0.4418773507147949,
      "learning_rate": 0.00019905391790062092,
      "loss": 0.2812,
      "step": 6235
    },
    {
      "epoch": 2.9345882352941177,
      "grad_norm": 0.36788904864824146,
      "learning_rate": 0.00019901248994894276,
      "loss": 0.2812,
      "step": 6236
    },
    {
      "epoch": 2.9350588235294115,
      "grad_norm": 0.20476286359086232,
      "learning_rate": 0.0001989710578113576,
      "loss": 0.1484,
      "step": 6237
    },
    {
      "epoch": 2.9355294117647057,
      "grad_norm": 0.3278624781926666,
      "learning_rate": 0.00019892962149140392,
      "loss": 0.1245,
      "step": 6238
    },
    {
      "epoch": 2.936,
      "grad_norm": 0.4201304818104453,
      "learning_rate": 0.0001988881809926205,
      "loss": 0.127,
      "step": 6239
    },
    {
      "epoch": 2.936470588235294,
      "grad_norm": 0.2931045643175828,
      "learning_rate": 0.00019884673631854672,
      "loss": 0.1455,
      "step": 6240
    },
    {
      "epoch": 2.9369411764705884,
      "grad_norm": 0.38890297019288805,
      "learning_rate": 0.000198805287472722,
      "loss": 0.1768,
      "step": 6241
    },
    {
      "epoch": 2.9374117647058826,
      "grad_norm": 0.5366829650951331,
      "learning_rate": 0.0001987638344586864,
      "loss": 0.4922,
      "step": 6242
    },
    {
      "epoch": 2.9378823529411764,
      "grad_norm": 0.416016690738008,
      "learning_rate": 0.00019872237727998004,
      "loss": 0.2539,
      "step": 6243
    },
    {
      "epoch": 2.9383529411764706,
      "grad_norm": 0.3654156688862927,
      "learning_rate": 0.0001986809159401438,
      "loss": 0.1245,
      "step": 6244
    },
    {
      "epoch": 2.938823529411765,
      "grad_norm": 0.6288092577729679,
      "learning_rate": 0.00019863945044271846,
      "loss": 0.5234,
      "step": 6245
    },
    {
      "epoch": 2.9392941176470586,
      "grad_norm": 0.5135055046812711,
      "learning_rate": 0.00019859798079124554,
      "loss": 0.4902,
      "step": 6246
    },
    {
      "epoch": 2.939764705882353,
      "grad_norm": 0.362284640172922,
      "learning_rate": 0.00019855650698926666,
      "loss": 0.3418,
      "step": 6247
    },
    {
      "epoch": 2.940235294117647,
      "grad_norm": 0.48514203786934895,
      "learning_rate": 0.00019851502904032396,
      "loss": 0.4863,
      "step": 6248
    },
    {
      "epoch": 2.9407058823529413,
      "grad_norm": 0.4802986580995788,
      "learning_rate": 0.00019847354694795978,
      "loss": 0.5781,
      "step": 6249
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 0.2570704578491745,
      "learning_rate": 0.00019843206071571692,
      "loss": 0.1318,
      "step": 6250
    },
    {
      "epoch": 2.9416470588235293,
      "grad_norm": 0.44121861771256765,
      "learning_rate": 0.00019839057034713858,
      "loss": 0.2812,
      "step": 6251
    },
    {
      "epoch": 2.9421176470588235,
      "grad_norm": 0.7943674952134011,
      "learning_rate": 0.00019834907584576812,
      "loss": 0.5547,
      "step": 6252
    },
    {
      "epoch": 2.9425882352941177,
      "grad_norm": 0.5336281731341955,
      "learning_rate": 0.00019830757721514943,
      "loss": 0.3926,
      "step": 6253
    },
    {
      "epoch": 2.9430588235294115,
      "grad_norm": 0.6327839850436283,
      "learning_rate": 0.00019826607445882667,
      "loss": 0.7461,
      "step": 6254
    },
    {
      "epoch": 2.9435294117647057,
      "grad_norm": 0.252129131516349,
      "learning_rate": 0.00019822456758034445,
      "loss": 0.1523,
      "step": 6255
    },
    {
      "epoch": 2.944,
      "grad_norm": 0.2513703226358202,
      "learning_rate": 0.0001981830565832476,
      "loss": 0.2598,
      "step": 6256
    },
    {
      "epoch": 2.944470588235294,
      "grad_norm": 0.5124375824237115,
      "learning_rate": 0.0001981415414710814,
      "loss": 0.5195,
      "step": 6257
    },
    {
      "epoch": 2.9449411764705884,
      "grad_norm": 0.29028893361313146,
      "learning_rate": 0.00019810002224739142,
      "loss": 0.1465,
      "step": 6258
    },
    {
      "epoch": 2.9454117647058826,
      "grad_norm": 0.6006336392997661,
      "learning_rate": 0.00019805849891572363,
      "loss": 0.418,
      "step": 6259
    },
    {
      "epoch": 2.9458823529411764,
      "grad_norm": 0.4818669079323836,
      "learning_rate": 0.00019801697147962428,
      "loss": 0.3926,
      "step": 6260
    },
    {
      "epoch": 2.9463529411764706,
      "grad_norm": 1.1024306317332806,
      "learning_rate": 0.00019797543994264004,
      "loss": 0.6367,
      "step": 6261
    },
    {
      "epoch": 2.946823529411765,
      "grad_norm": 0.43071675198080345,
      "learning_rate": 0.00019793390430831794,
      "loss": 0.5938,
      "step": 6262
    },
    {
      "epoch": 2.9472941176470586,
      "grad_norm": 0.419507221462926,
      "learning_rate": 0.00019789236458020527,
      "loss": 0.4551,
      "step": 6263
    },
    {
      "epoch": 2.947764705882353,
      "grad_norm": 0.39528634657080913,
      "learning_rate": 0.00019785082076184977,
      "loss": 0.5039,
      "step": 6264
    },
    {
      "epoch": 2.948235294117647,
      "grad_norm": 0.4210301430134144,
      "learning_rate": 0.00019780927285679945,
      "loss": 0.4707,
      "step": 6265
    },
    {
      "epoch": 2.9487058823529413,
      "grad_norm": 0.4064081225550216,
      "learning_rate": 0.00019776772086860277,
      "loss": 0.4961,
      "step": 6266
    },
    {
      "epoch": 2.9491764705882355,
      "grad_norm": 0.3596001672382849,
      "learning_rate": 0.0001977261648008084,
      "loss": 0.2793,
      "step": 6267
    },
    {
      "epoch": 2.9496470588235293,
      "grad_norm": 0.3770195381750296,
      "learning_rate": 0.00019768460465696545,
      "loss": 0.4902,
      "step": 6268
    },
    {
      "epoch": 2.9501176470588235,
      "grad_norm": 0.19744759858868477,
      "learning_rate": 0.0001976430404406234,
      "loss": 0.125,
      "step": 6269
    },
    {
      "epoch": 2.9505882352941177,
      "grad_norm": 0.47165758908801064,
      "learning_rate": 0.00019760147215533197,
      "loss": 0.5234,
      "step": 6270
    },
    {
      "epoch": 2.9510588235294115,
      "grad_norm": 0.44388639261698176,
      "learning_rate": 0.00019755989980464137,
      "loss": 0.1738,
      "step": 6271
    },
    {
      "epoch": 2.9515294117647057,
      "grad_norm": 0.5067861059894353,
      "learning_rate": 0.00019751832339210204,
      "loss": 0.3066,
      "step": 6272
    },
    {
      "epoch": 2.952,
      "grad_norm": 0.4948851881902311,
      "learning_rate": 0.00019747674292126477,
      "loss": 0.6797,
      "step": 6273
    },
    {
      "epoch": 2.952470588235294,
      "grad_norm": 0.5062425378946187,
      "learning_rate": 0.00019743515839568083,
      "loss": 0.4648,
      "step": 6274
    },
    {
      "epoch": 2.9529411764705884,
      "grad_norm": 0.5390151208451751,
      "learning_rate": 0.00019739356981890167,
      "loss": 0.4238,
      "step": 6275
    },
    {
      "epoch": 2.9534117647058826,
      "grad_norm": 0.46770751839140734,
      "learning_rate": 0.00019735197719447916,
      "loss": 0.3555,
      "step": 6276
    },
    {
      "epoch": 2.9538823529411764,
      "grad_norm": 0.3668697396207256,
      "learning_rate": 0.00019731038052596558,
      "loss": 0.2891,
      "step": 6277
    },
    {
      "epoch": 2.9543529411764706,
      "grad_norm": 0.6875744023708885,
      "learning_rate": 0.00019726877981691347,
      "loss": 0.6602,
      "step": 6278
    },
    {
      "epoch": 2.954823529411765,
      "grad_norm": 0.6162624601706177,
      "learning_rate": 0.0001972271750708757,
      "loss": 0.8789,
      "step": 6279
    },
    {
      "epoch": 2.9552941176470586,
      "grad_norm": 0.37778067396662296,
      "learning_rate": 0.0001971855662914055,
      "loss": 0.3145,
      "step": 6280
    },
    {
      "epoch": 2.955764705882353,
      "grad_norm": 0.1818361529589192,
      "learning_rate": 0.00019714395348205649,
      "loss": 0.0737,
      "step": 6281
    },
    {
      "epoch": 2.956235294117647,
      "grad_norm": 0.2394919325278451,
      "learning_rate": 0.00019710233664638264,
      "loss": 0.1289,
      "step": 6282
    },
    {
      "epoch": 2.9567058823529413,
      "grad_norm": 0.8321317396022494,
      "learning_rate": 0.00019706071578793819,
      "loss": 0.6641,
      "step": 6283
    },
    {
      "epoch": 2.9571764705882355,
      "grad_norm": 0.6114347466222264,
      "learning_rate": 0.00019701909091027778,
      "loss": 0.9023,
      "step": 6284
    },
    {
      "epoch": 2.9576470588235293,
      "grad_norm": 0.448483859117895,
      "learning_rate": 0.0001969774620169564,
      "loss": 0.457,
      "step": 6285
    },
    {
      "epoch": 2.9581176470588235,
      "grad_norm": 0.34567904434256835,
      "learning_rate": 0.00019693582911152935,
      "loss": 0.2617,
      "step": 6286
    },
    {
      "epoch": 2.9585882352941177,
      "grad_norm": 0.5907521164031996,
      "learning_rate": 0.0001968941921975522,
      "loss": 0.3438,
      "step": 6287
    },
    {
      "epoch": 2.9590588235294115,
      "grad_norm": 0.12437849741839482,
      "learning_rate": 0.00019685255127858107,
      "loss": 0.0767,
      "step": 6288
    },
    {
      "epoch": 2.9595294117647057,
      "grad_norm": 0.4214523541387562,
      "learning_rate": 0.00019681090635817228,
      "loss": 0.249,
      "step": 6289
    },
    {
      "epoch": 2.96,
      "grad_norm": 0.38362848591285387,
      "learning_rate": 0.00019676925743988244,
      "loss": 0.6172,
      "step": 6290
    },
    {
      "epoch": 2.960470588235294,
      "grad_norm": 0.30022654925941894,
      "learning_rate": 0.00019672760452726863,
      "loss": 0.1011,
      "step": 6291
    },
    {
      "epoch": 2.9609411764705884,
      "grad_norm": 0.5564259048347803,
      "learning_rate": 0.0001966859476238882,
      "loss": 0.6367,
      "step": 6292
    },
    {
      "epoch": 2.961411764705882,
      "grad_norm": 0.320769067119535,
      "learning_rate": 0.00019664428673329883,
      "loss": 0.207,
      "step": 6293
    },
    {
      "epoch": 2.9618823529411764,
      "grad_norm": 0.3921983717744418,
      "learning_rate": 0.00019660262185905856,
      "loss": 0.2344,
      "step": 6294
    },
    {
      "epoch": 2.9623529411764706,
      "grad_norm": 0.3505937364108029,
      "learning_rate": 0.0001965609530047259,
      "loss": 0.1553,
      "step": 6295
    },
    {
      "epoch": 2.962823529411765,
      "grad_norm": 0.2942179295699984,
      "learning_rate": 0.0001965192801738594,
      "loss": 0.2217,
      "step": 6296
    },
    {
      "epoch": 2.9632941176470586,
      "grad_norm": 0.3589041607139194,
      "learning_rate": 0.0001964776033700182,
      "loss": 0.2295,
      "step": 6297
    },
    {
      "epoch": 2.963764705882353,
      "grad_norm": 0.5245939101378103,
      "learning_rate": 0.00019643592259676164,
      "loss": 0.7148,
      "step": 6298
    },
    {
      "epoch": 2.964235294117647,
      "grad_norm": 0.9151107261287388,
      "learning_rate": 0.0001963942378576496,
      "loss": 1.2734,
      "step": 6299
    },
    {
      "epoch": 2.9647058823529413,
      "grad_norm": 0.3547466588301367,
      "learning_rate": 0.0001963525491562421,
      "loss": 0.291,
      "step": 6300
    },
    {
      "epoch": 2.9651764705882355,
      "grad_norm": 0.25277284992233057,
      "learning_rate": 0.0001963108564960995,
      "loss": 0.2051,
      "step": 6301
    },
    {
      "epoch": 2.9656470588235293,
      "grad_norm": 0.42489245980789747,
      "learning_rate": 0.00019626915988078263,
      "loss": 0.4375,
      "step": 6302
    },
    {
      "epoch": 2.9661176470588235,
      "grad_norm": 0.5128688490961837,
      "learning_rate": 0.00019622745931385254,
      "loss": 0.4961,
      "step": 6303
    },
    {
      "epoch": 2.9665882352941177,
      "grad_norm": 0.3537096143310463,
      "learning_rate": 0.00019618575479887066,
      "loss": 0.2393,
      "step": 6304
    },
    {
      "epoch": 2.9670588235294115,
      "grad_norm": 0.3490528879313775,
      "learning_rate": 0.00019614404633939886,
      "loss": 0.2188,
      "step": 6305
    },
    {
      "epoch": 2.9675294117647057,
      "grad_norm": 0.3767569626597855,
      "learning_rate": 0.0001961023339389991,
      "loss": 0.3965,
      "step": 6306
    },
    {
      "epoch": 2.968,
      "grad_norm": 0.809450156066381,
      "learning_rate": 0.00019606061760123392,
      "loss": 0.8594,
      "step": 6307
    },
    {
      "epoch": 2.968470588235294,
      "grad_norm": 0.3591102441348753,
      "learning_rate": 0.00019601889732966604,
      "loss": 0.2139,
      "step": 6308
    },
    {
      "epoch": 2.9689411764705884,
      "grad_norm": 0.5113749157977154,
      "learning_rate": 0.0001959771731278586,
      "loss": 0.4043,
      "step": 6309
    },
    {
      "epoch": 2.969411764705882,
      "grad_norm": 0.48340835116851355,
      "learning_rate": 0.0001959354449993751,
      "loss": 0.4375,
      "step": 6310
    },
    {
      "epoch": 2.9698823529411764,
      "grad_norm": 0.35836023832519504,
      "learning_rate": 0.00019589371294777925,
      "loss": 0.2148,
      "step": 6311
    },
    {
      "epoch": 2.9703529411764706,
      "grad_norm": 0.5312952244837904,
      "learning_rate": 0.00019585197697663526,
      "loss": 0.0791,
      "step": 6312
    },
    {
      "epoch": 2.970823529411765,
      "grad_norm": 0.46391307807886095,
      "learning_rate": 0.0001958102370895075,
      "loss": 0.4277,
      "step": 6313
    },
    {
      "epoch": 2.9712941176470586,
      "grad_norm": 0.5624137348670271,
      "learning_rate": 0.00019576849328996075,
      "loss": 0.498,
      "step": 6314
    },
    {
      "epoch": 2.971764705882353,
      "grad_norm": 0.5622673054044823,
      "learning_rate": 0.00019572674558156024,
      "loss": 0.5234,
      "step": 6315
    },
    {
      "epoch": 2.972235294117647,
      "grad_norm": 0.6193713141630856,
      "learning_rate": 0.00019568499396787134,
      "loss": 0.418,
      "step": 6316
    },
    {
      "epoch": 2.9727058823529413,
      "grad_norm": 0.58699355721668,
      "learning_rate": 0.0001956432384524599,
      "loss": 0.4141,
      "step": 6317
    },
    {
      "epoch": 2.9731764705882355,
      "grad_norm": 0.44726585449570827,
      "learning_rate": 0.000195601479038892,
      "loss": 0.4766,
      "step": 6318
    },
    {
      "epoch": 2.9736470588235293,
      "grad_norm": 0.5980989727462234,
      "learning_rate": 0.0001955597157307341,
      "loss": 0.7461,
      "step": 6319
    },
    {
      "epoch": 2.9741176470588235,
      "grad_norm": 0.29150053052157676,
      "learning_rate": 0.000195517948531553,
      "loss": 0.3594,
      "step": 6320
    },
    {
      "epoch": 2.9745882352941178,
      "grad_norm": 0.5395253778393257,
      "learning_rate": 0.00019547617744491587,
      "loss": 0.6172,
      "step": 6321
    },
    {
      "epoch": 2.9750588235294115,
      "grad_norm": 0.46114656494965267,
      "learning_rate": 0.00019543440247439008,
      "loss": 0.5039,
      "step": 6322
    },
    {
      "epoch": 2.9755294117647058,
      "grad_norm": 0.3146922399280419,
      "learning_rate": 0.00019539262362354353,
      "loss": 0.1289,
      "step": 6323
    },
    {
      "epoch": 2.976,
      "grad_norm": 0.6509466000153208,
      "learning_rate": 0.00019535084089594422,
      "loss": 0.8867,
      "step": 6324
    },
    {
      "epoch": 2.976470588235294,
      "grad_norm": 0.6291230591031489,
      "learning_rate": 0.00019530905429516066,
      "loss": 0.7852,
      "step": 6325
    },
    {
      "epoch": 2.9769411764705884,
      "grad_norm": 0.4966065534095906,
      "learning_rate": 0.00019526726382476165,
      "loss": 0.4199,
      "step": 6326
    },
    {
      "epoch": 2.977411764705882,
      "grad_norm": 0.3909219703017449,
      "learning_rate": 0.00019522546948831627,
      "loss": 0.373,
      "step": 6327
    },
    {
      "epoch": 2.9778823529411764,
      "grad_norm": 0.4403953579707658,
      "learning_rate": 0.00019518367128939392,
      "loss": 0.5508,
      "step": 6328
    },
    {
      "epoch": 2.9783529411764706,
      "grad_norm": 0.25610752701769546,
      "learning_rate": 0.0001951418692315645,
      "loss": 0.1108,
      "step": 6329
    },
    {
      "epoch": 2.978823529411765,
      "grad_norm": 0.3303249288445271,
      "learning_rate": 0.00019510006331839794,
      "loss": 0.1826,
      "step": 6330
    },
    {
      "epoch": 2.9792941176470586,
      "grad_norm": 0.43683447798386094,
      "learning_rate": 0.00019505825355346483,
      "loss": 0.5859,
      "step": 6331
    },
    {
      "epoch": 2.979764705882353,
      "grad_norm": 0.5423829931359905,
      "learning_rate": 0.0001950164399403358,
      "loss": 0.4102,
      "step": 6332
    },
    {
      "epoch": 2.980235294117647,
      "grad_norm": 0.2958351822316111,
      "learning_rate": 0.00019497462248258202,
      "loss": 0.1377,
      "step": 6333
    },
    {
      "epoch": 2.9807058823529413,
      "grad_norm": 0.48409958681014215,
      "learning_rate": 0.0001949328011837749,
      "loss": 0.6484,
      "step": 6334
    },
    {
      "epoch": 2.9811764705882355,
      "grad_norm": 0.35642466582741145,
      "learning_rate": 0.00019489097604748613,
      "loss": 0.4922,
      "step": 6335
    },
    {
      "epoch": 2.9816470588235293,
      "grad_norm": 0.31095987126751995,
      "learning_rate": 0.00019484914707728785,
      "loss": 0.1309,
      "step": 6336
    },
    {
      "epoch": 2.9821176470588235,
      "grad_norm": 0.337164840821853,
      "learning_rate": 0.00019480731427675238,
      "loss": 0.2305,
      "step": 6337
    },
    {
      "epoch": 2.9825882352941178,
      "grad_norm": 0.6266846844487193,
      "learning_rate": 0.0001947654776494525,
      "loss": 0.6211,
      "step": 6338
    },
    {
      "epoch": 2.9830588235294115,
      "grad_norm": 0.5163975669783775,
      "learning_rate": 0.0001947236371989613,
      "loss": 0.6328,
      "step": 6339
    },
    {
      "epoch": 2.9835294117647058,
      "grad_norm": 0.9354430976841602,
      "learning_rate": 0.0001946817929288521,
      "loss": 1.2109,
      "step": 6340
    },
    {
      "epoch": 2.984,
      "grad_norm": 0.39228643496039434,
      "learning_rate": 0.0001946399448426986,
      "loss": 0.1514,
      "step": 6341
    },
    {
      "epoch": 2.984470588235294,
      "grad_norm": 0.3664636592167137,
      "learning_rate": 0.00019459809294407486,
      "loss": 0.3281,
      "step": 6342
    },
    {
      "epoch": 2.9849411764705884,
      "grad_norm": 0.3296556047356443,
      "learning_rate": 0.00019455623723655521,
      "loss": 0.1543,
      "step": 6343
    },
    {
      "epoch": 2.985411764705882,
      "grad_norm": 0.6773659397587778,
      "learning_rate": 0.00019451437772371438,
      "loss": 0.5,
      "step": 6344
    },
    {
      "epoch": 2.9858823529411764,
      "grad_norm": 0.18463766171111157,
      "learning_rate": 0.00019447251440912736,
      "loss": 0.0452,
      "step": 6345
    },
    {
      "epoch": 2.9863529411764707,
      "grad_norm": 0.5212099108432072,
      "learning_rate": 0.00019443064729636947,
      "loss": 0.457,
      "step": 6346
    },
    {
      "epoch": 2.9868235294117644,
      "grad_norm": 0.4341970181900948,
      "learning_rate": 0.00019438877638901635,
      "loss": 0.3652,
      "step": 6347
    },
    {
      "epoch": 2.9872941176470587,
      "grad_norm": 0.36458042197096563,
      "learning_rate": 0.00019434690169064402,
      "loss": 0.1406,
      "step": 6348
    },
    {
      "epoch": 2.987764705882353,
      "grad_norm": 0.6416736026865081,
      "learning_rate": 0.00019430502320482871,
      "loss": 0.5078,
      "step": 6349
    },
    {
      "epoch": 2.988235294117647,
      "grad_norm": 0.479398879582295,
      "learning_rate": 0.00019426314093514717,
      "loss": 0.4531,
      "step": 6350
    },
    {
      "epoch": 2.9887058823529413,
      "grad_norm": 0.36947352909350056,
      "learning_rate": 0.0001942212548851763,
      "loss": 0.1846,
      "step": 6351
    },
    {
      "epoch": 2.9891764705882355,
      "grad_norm": 0.30269096610785334,
      "learning_rate": 0.00019417936505849328,
      "loss": 0.2139,
      "step": 6352
    },
    {
      "epoch": 2.9896470588235293,
      "grad_norm": 0.41929824156657886,
      "learning_rate": 0.00019413747145867583,
      "loss": 0.3828,
      "step": 6353
    },
    {
      "epoch": 2.9901176470588235,
      "grad_norm": 0.4071523373046449,
      "learning_rate": 0.00019409557408930184,
      "loss": 0.4023,
      "step": 6354
    },
    {
      "epoch": 2.9905882352941178,
      "grad_norm": 0.32817970304898325,
      "learning_rate": 0.0001940536729539495,
      "loss": 0.2402,
      "step": 6355
    },
    {
      "epoch": 2.9910588235294115,
      "grad_norm": 0.3993456017194407,
      "learning_rate": 0.00019401176805619746,
      "loss": 0.2695,
      "step": 6356
    },
    {
      "epoch": 2.9915294117647058,
      "grad_norm": 0.9018033474711146,
      "learning_rate": 0.0001939698593996245,
      "loss": 0.9102,
      "step": 6357
    },
    {
      "epoch": 2.992,
      "grad_norm": 0.3850396824784858,
      "learning_rate": 0.00019392794698780988,
      "loss": 0.418,
      "step": 6358
    },
    {
      "epoch": 2.992470588235294,
      "grad_norm": 0.6526694954600742,
      "learning_rate": 0.00019388603082433315,
      "loss": 0.6992,
      "step": 6359
    },
    {
      "epoch": 2.9929411764705884,
      "grad_norm": 0.32762635939094187,
      "learning_rate": 0.00019384411091277412,
      "loss": 0.4121,
      "step": 6360
    },
    {
      "epoch": 2.993411764705882,
      "grad_norm": 0.6026696983400404,
      "learning_rate": 0.00019380218725671298,
      "loss": 0.6055,
      "step": 6361
    },
    {
      "epoch": 2.9938823529411764,
      "grad_norm": 0.4741659014482883,
      "learning_rate": 0.00019376025985973013,
      "loss": 0.498,
      "step": 6362
    },
    {
      "epoch": 2.9943529411764707,
      "grad_norm": 0.6465058758318808,
      "learning_rate": 0.00019371832872540652,
      "loss": 0.582,
      "step": 6363
    },
    {
      "epoch": 2.9948235294117644,
      "grad_norm": 0.5944210295351604,
      "learning_rate": 0.00019367639385732316,
      "loss": 0.5352,
      "step": 6364
    },
    {
      "epoch": 2.9952941176470587,
      "grad_norm": 0.4691756538356365,
      "learning_rate": 0.00019363445525906152,
      "loss": 0.5625,
      "step": 6365
    },
    {
      "epoch": 2.995764705882353,
      "grad_norm": 0.3863889577921748,
      "learning_rate": 0.00019359251293420336,
      "loss": 0.4082,
      "step": 6366
    },
    {
      "epoch": 2.996235294117647,
      "grad_norm": 0.490998133176891,
      "learning_rate": 0.00019355056688633086,
      "loss": 0.3613,
      "step": 6367
    },
    {
      "epoch": 2.9967058823529413,
      "grad_norm": 0.5184334545011122,
      "learning_rate": 0.0001935086171190263,
      "loss": 0.3555,
      "step": 6368
    },
    {
      "epoch": 2.9971764705882356,
      "grad_norm": 1.4423115604056636,
      "learning_rate": 0.00019346666363587238,
      "loss": 0.7461,
      "step": 6369
    },
    {
      "epoch": 2.9976470588235293,
      "grad_norm": 0.27001096275726755,
      "learning_rate": 0.00019342470644045223,
      "loss": 0.1963,
      "step": 6370
    },
    {
      "epoch": 2.9981176470588236,
      "grad_norm": 0.4402337652348266,
      "learning_rate": 0.00019338274553634913,
      "loss": 0.3945,
      "step": 6371
    },
    {
      "epoch": 2.9985882352941178,
      "grad_norm": 0.3736036430789037,
      "learning_rate": 0.00019334078092714683,
      "loss": 0.2695,
      "step": 6372
    },
    {
      "epoch": 2.9990588235294116,
      "grad_norm": 0.5492825288956905,
      "learning_rate": 0.00019329881261642918,
      "loss": 0.4316,
      "step": 6373
    },
    {
      "epoch": 2.9995294117647058,
      "grad_norm": 0.6456380396985603,
      "learning_rate": 0.00019325684060778056,
      "loss": 0.6328,
      "step": 6374
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.6608824753958732,
      "learning_rate": 0.00019321486490478563,
      "loss": 0.3691,
      "step": 6375
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.427168607711792,
      "eval_runtime": 44.1404,
      "eval_samples_per_second": 16.991,
      "eval_steps_per_second": 8.496,
      "step": 6375
    },
    {
      "epoch": 3.0004705882352942,
      "grad_norm": 1.0549255375031714,
      "learning_rate": 0.0001931728855110292,
      "loss": 0.4941,
      "step": 6376
    },
    {
      "epoch": 3.0009411764705884,
      "grad_norm": 0.4184990947448964,
      "learning_rate": 0.00019313090243009665,
      "loss": 0.3242,
      "step": 6377
    },
    {
      "epoch": 3.0014117647058822,
      "grad_norm": 0.23841525855568538,
      "learning_rate": 0.00019308891566557345,
      "loss": 0.2139,
      "step": 6378
    },
    {
      "epoch": 3.0018823529411764,
      "grad_norm": 0.4793905190192389,
      "learning_rate": 0.0001930469252210455,
      "loss": 0.6641,
      "step": 6379
    },
    {
      "epoch": 3.0023529411764707,
      "grad_norm": 0.34413309721789986,
      "learning_rate": 0.000193004931100099,
      "loss": 0.1729,
      "step": 6380
    },
    {
      "epoch": 3.002823529411765,
      "grad_norm": 0.28454913410260924,
      "learning_rate": 0.00019296293330632047,
      "loss": 0.1152,
      "step": 6381
    },
    {
      "epoch": 3.0032941176470587,
      "grad_norm": 0.35907509943585714,
      "learning_rate": 0.0001929209318432967,
      "loss": 0.1504,
      "step": 6382
    },
    {
      "epoch": 3.003764705882353,
      "grad_norm": 0.3591572902616618,
      "learning_rate": 0.00019287892671461475,
      "loss": 0.1582,
      "step": 6383
    },
    {
      "epoch": 3.004235294117647,
      "grad_norm": 0.24558945137064286,
      "learning_rate": 0.0001928369179238622,
      "loss": 0.1904,
      "step": 6384
    },
    {
      "epoch": 3.0047058823529413,
      "grad_norm": 0.33747421002797007,
      "learning_rate": 0.00019279490547462674,
      "loss": 0.2969,
      "step": 6385
    },
    {
      "epoch": 3.005176470588235,
      "grad_norm": 0.6948449631675184,
      "learning_rate": 0.0001927528893704964,
      "loss": 0.6172,
      "step": 6386
    },
    {
      "epoch": 3.0056470588235293,
      "grad_norm": 0.7017559132662022,
      "learning_rate": 0.00019271086961505964,
      "loss": 0.3906,
      "step": 6387
    },
    {
      "epoch": 3.0061176470588236,
      "grad_norm": 0.5953973217522845,
      "learning_rate": 0.0001926688462119051,
      "loss": 0.5625,
      "step": 6388
    },
    {
      "epoch": 3.006588235294118,
      "grad_norm": 0.31126905153459405,
      "learning_rate": 0.00019262681916462187,
      "loss": 0.1318,
      "step": 6389
    },
    {
      "epoch": 3.0070588235294116,
      "grad_norm": 0.48430969444190713,
      "learning_rate": 0.00019258478847679914,
      "loss": 0.4219,
      "step": 6390
    },
    {
      "epoch": 3.007529411764706,
      "grad_norm": 0.3913625293683232,
      "learning_rate": 0.0001925427541520266,
      "loss": 0.3867,
      "step": 6391
    },
    {
      "epoch": 3.008,
      "grad_norm": 0.2672176469707961,
      "learning_rate": 0.00019250071619389422,
      "loss": 0.1729,
      "step": 6392
    },
    {
      "epoch": 3.0084705882352942,
      "grad_norm": 0.5519859505154142,
      "learning_rate": 0.00019245867460599217,
      "loss": 0.7969,
      "step": 6393
    },
    {
      "epoch": 3.0089411764705885,
      "grad_norm": 0.5214067501684692,
      "learning_rate": 0.00019241662939191106,
      "loss": 0.6094,
      "step": 6394
    },
    {
      "epoch": 3.0094117647058822,
      "grad_norm": 0.4287916444173287,
      "learning_rate": 0.00019237458055524178,
      "loss": 0.2891,
      "step": 6395
    },
    {
      "epoch": 3.0098823529411765,
      "grad_norm": 0.583730837197956,
      "learning_rate": 0.00019233252809957546,
      "loss": 0.832,
      "step": 6396
    },
    {
      "epoch": 3.0103529411764707,
      "grad_norm": 0.2643222391638411,
      "learning_rate": 0.00019229047202850357,
      "loss": 0.124,
      "step": 6397
    },
    {
      "epoch": 3.010823529411765,
      "grad_norm": 0.4469144102162807,
      "learning_rate": 0.00019224841234561797,
      "loss": 0.3418,
      "step": 6398
    },
    {
      "epoch": 3.0112941176470587,
      "grad_norm": 0.6240889879744909,
      "learning_rate": 0.00019220634905451074,
      "loss": 0.8164,
      "step": 6399
    },
    {
      "epoch": 3.011764705882353,
      "grad_norm": 0.4613035621826292,
      "learning_rate": 0.00019216428215877425,
      "loss": 0.3438,
      "step": 6400
    },
    {
      "epoch": 3.012235294117647,
      "grad_norm": 0.4242758695465476,
      "learning_rate": 0.00019212221166200124,
      "loss": 0.3652,
      "step": 6401
    },
    {
      "epoch": 3.0127058823529413,
      "grad_norm": 0.325650685513944,
      "learning_rate": 0.00019208013756778476,
      "loss": 0.377,
      "step": 6402
    },
    {
      "epoch": 3.013176470588235,
      "grad_norm": 0.224936858780345,
      "learning_rate": 0.0001920380598797181,
      "loss": 0.1367,
      "step": 6403
    },
    {
      "epoch": 3.0136470588235293,
      "grad_norm": 0.42982422952107263,
      "learning_rate": 0.000191995978601395,
      "loss": 0.208,
      "step": 6404
    },
    {
      "epoch": 3.0141176470588236,
      "grad_norm": 0.5808489975073284,
      "learning_rate": 0.0001919538937364093,
      "loss": 0.5469,
      "step": 6405
    },
    {
      "epoch": 3.014588235294118,
      "grad_norm": 0.23018527961256968,
      "learning_rate": 0.00019191180528835527,
      "loss": 0.127,
      "step": 6406
    },
    {
      "epoch": 3.0150588235294116,
      "grad_norm": 0.2948256997398966,
      "learning_rate": 0.00019186971326082752,
      "loss": 0.1748,
      "step": 6407
    },
    {
      "epoch": 3.015529411764706,
      "grad_norm": 0.6217459171580368,
      "learning_rate": 0.0001918276176574209,
      "loss": 0.4727,
      "step": 6408
    },
    {
      "epoch": 3.016,
      "grad_norm": 0.4891347562647384,
      "learning_rate": 0.00019178551848173054,
      "loss": 0.4746,
      "step": 6409
    },
    {
      "epoch": 3.0164705882352942,
      "grad_norm": 0.542864025452893,
      "learning_rate": 0.00019174341573735196,
      "loss": 0.5859,
      "step": 6410
    },
    {
      "epoch": 3.016941176470588,
      "grad_norm": 0.5842241095532854,
      "learning_rate": 0.0001917013094278809,
      "loss": 0.6445,
      "step": 6411
    },
    {
      "epoch": 3.0174117647058822,
      "grad_norm": 0.4272245738974219,
      "learning_rate": 0.0001916591995569135,
      "loss": 0.2441,
      "step": 6412
    },
    {
      "epoch": 3.0178823529411765,
      "grad_norm": 0.3913206846887119,
      "learning_rate": 0.00019161708612804615,
      "loss": 0.3887,
      "step": 6413
    },
    {
      "epoch": 3.0183529411764707,
      "grad_norm": 0.3984458610063163,
      "learning_rate": 0.00019157496914487545,
      "loss": 0.2236,
      "step": 6414
    },
    {
      "epoch": 3.018823529411765,
      "grad_norm": 0.31314999818188893,
      "learning_rate": 0.00019153284861099848,
      "loss": 0.1582,
      "step": 6415
    },
    {
      "epoch": 3.0192941176470587,
      "grad_norm": 0.4186178524608014,
      "learning_rate": 0.00019149072453001254,
      "loss": 0.2793,
      "step": 6416
    },
    {
      "epoch": 3.019764705882353,
      "grad_norm": 0.27153871862577567,
      "learning_rate": 0.00019144859690551518,
      "loss": 0.2012,
      "step": 6417
    },
    {
      "epoch": 3.020235294117647,
      "grad_norm": 0.4789452946928273,
      "learning_rate": 0.00019140646574110438,
      "loss": 0.3906,
      "step": 6418
    },
    {
      "epoch": 3.0207058823529414,
      "grad_norm": 0.8171236510420306,
      "learning_rate": 0.0001913643310403783,
      "loss": 0.6641,
      "step": 6419
    },
    {
      "epoch": 3.021176470588235,
      "grad_norm": 0.6538888238986177,
      "learning_rate": 0.00019132219280693544,
      "loss": 0.7539,
      "step": 6420
    },
    {
      "epoch": 3.0216470588235294,
      "grad_norm": 0.7415628646844066,
      "learning_rate": 0.00019128005104437462,
      "loss": 0.7148,
      "step": 6421
    },
    {
      "epoch": 3.0221176470588236,
      "grad_norm": 0.27437948541963536,
      "learning_rate": 0.00019123790575629503,
      "loss": 0.1211,
      "step": 6422
    },
    {
      "epoch": 3.022588235294118,
      "grad_norm": 0.46401286825082066,
      "learning_rate": 0.00019119575694629594,
      "loss": 0.4941,
      "step": 6423
    },
    {
      "epoch": 3.0230588235294116,
      "grad_norm": 0.49172742100311234,
      "learning_rate": 0.00019115360461797716,
      "loss": 0.4082,
      "step": 6424
    },
    {
      "epoch": 3.023529411764706,
      "grad_norm": 0.2666429100731579,
      "learning_rate": 0.00019111144877493873,
      "loss": 0.0913,
      "step": 6425
    },
    {
      "epoch": 3.024,
      "grad_norm": 0.5376801600431526,
      "learning_rate": 0.0001910692894207809,
      "loss": 0.8047,
      "step": 6426
    },
    {
      "epoch": 3.0244705882352942,
      "grad_norm": 0.43432931640918787,
      "learning_rate": 0.00019102712655910426,
      "loss": 0.2461,
      "step": 6427
    },
    {
      "epoch": 3.024941176470588,
      "grad_norm": 0.388150029678066,
      "learning_rate": 0.00019098496019350985,
      "loss": 0.3418,
      "step": 6428
    },
    {
      "epoch": 3.0254117647058822,
      "grad_norm": 0.5116190070496669,
      "learning_rate": 0.0001909427903275988,
      "loss": 0.332,
      "step": 6429
    },
    {
      "epoch": 3.0258823529411765,
      "grad_norm": 0.3987451026926863,
      "learning_rate": 0.00019090061696497258,
      "loss": 0.2871,
      "step": 6430
    },
    {
      "epoch": 3.0263529411764707,
      "grad_norm": 0.19624942046350716,
      "learning_rate": 0.0001908584401092331,
      "loss": 0.0962,
      "step": 6431
    },
    {
      "epoch": 3.026823529411765,
      "grad_norm": 0.3030038886112521,
      "learning_rate": 0.0001908162597639824,
      "loss": 0.2754,
      "step": 6432
    },
    {
      "epoch": 3.0272941176470587,
      "grad_norm": 0.496445651578139,
      "learning_rate": 0.00019077407593282294,
      "loss": 0.4238,
      "step": 6433
    },
    {
      "epoch": 3.027764705882353,
      "grad_norm": 0.44946830676736127,
      "learning_rate": 0.00019073188861935735,
      "loss": 0.5078,
      "step": 6434
    },
    {
      "epoch": 3.028235294117647,
      "grad_norm": 0.6626460213754729,
      "learning_rate": 0.00019068969782718874,
      "loss": 0.582,
      "step": 6435
    },
    {
      "epoch": 3.0287058823529414,
      "grad_norm": 0.45980257550885223,
      "learning_rate": 0.00019064750355992036,
      "loss": 0.3047,
      "step": 6436
    },
    {
      "epoch": 3.029176470588235,
      "grad_norm": 2.4479582153556643,
      "learning_rate": 0.00019060530582115578,
      "loss": 0.6562,
      "step": 6437
    },
    {
      "epoch": 3.0296470588235294,
      "grad_norm": 0.5157422085040494,
      "learning_rate": 0.00019056310461449885,
      "loss": 0.4688,
      "step": 6438
    },
    {
      "epoch": 3.0301176470588236,
      "grad_norm": 0.30002157267622537,
      "learning_rate": 0.00019052089994355394,
      "loss": 0.2188,
      "step": 6439
    },
    {
      "epoch": 3.030588235294118,
      "grad_norm": 0.5250937102092531,
      "learning_rate": 0.00019047869181192535,
      "loss": 0.4375,
      "step": 6440
    },
    {
      "epoch": 3.0310588235294116,
      "grad_norm": 0.7671405303575043,
      "learning_rate": 0.00019043648022321793,
      "loss": 0.7383,
      "step": 6441
    },
    {
      "epoch": 3.031529411764706,
      "grad_norm": 0.4969339696997743,
      "learning_rate": 0.00019039426518103678,
      "loss": 0.3516,
      "step": 6442
    },
    {
      "epoch": 3.032,
      "grad_norm": 0.8615046729335795,
      "learning_rate": 0.00019035204668898725,
      "loss": 0.7188,
      "step": 6443
    },
    {
      "epoch": 3.0324705882352943,
      "grad_norm": 0.41724878680765504,
      "learning_rate": 0.00019030982475067498,
      "loss": 0.4844,
      "step": 6444
    },
    {
      "epoch": 3.032941176470588,
      "grad_norm": 0.304695055570821,
      "learning_rate": 0.00019026759936970598,
      "loss": 0.3047,
      "step": 6445
    },
    {
      "epoch": 3.0334117647058823,
      "grad_norm": 0.5895336444980886,
      "learning_rate": 0.0001902253705496865,
      "loss": 0.4941,
      "step": 6446
    },
    {
      "epoch": 3.0338823529411765,
      "grad_norm": 0.6180785047775622,
      "learning_rate": 0.00019018313829422307,
      "loss": 0.543,
      "step": 6447
    },
    {
      "epoch": 3.0343529411764707,
      "grad_norm": 0.34906529095575406,
      "learning_rate": 0.0001901409026069225,
      "loss": 0.1123,
      "step": 6448
    },
    {
      "epoch": 3.034823529411765,
      "grad_norm": 0.523747320933613,
      "learning_rate": 0.000190098663491392,
      "loss": 0.5195,
      "step": 6449
    },
    {
      "epoch": 3.0352941176470587,
      "grad_norm": 0.5560228950054644,
      "learning_rate": 0.00019005642095123895,
      "loss": 0.6328,
      "step": 6450
    },
    {
      "epoch": 3.035764705882353,
      "grad_norm": 0.45169601466203885,
      "learning_rate": 0.00019001417499007103,
      "loss": 0.3965,
      "step": 6451
    },
    {
      "epoch": 3.036235294117647,
      "grad_norm": 0.44381148289310307,
      "learning_rate": 0.0001899719256114963,
      "loss": 0.375,
      "step": 6452
    },
    {
      "epoch": 3.0367058823529414,
      "grad_norm": 0.39004715290324316,
      "learning_rate": 0.00018992967281912312,
      "loss": 0.4727,
      "step": 6453
    },
    {
      "epoch": 3.037176470588235,
      "grad_norm": 0.3865618736827882,
      "learning_rate": 0.00018988741661655996,
      "loss": 0.2246,
      "step": 6454
    },
    {
      "epoch": 3.0376470588235294,
      "grad_norm": 0.46685627059923157,
      "learning_rate": 0.00018984515700741579,
      "loss": 0.4551,
      "step": 6455
    },
    {
      "epoch": 3.0381176470588236,
      "grad_norm": 0.6596269627043639,
      "learning_rate": 0.0001898028939952998,
      "loss": 0.625,
      "step": 6456
    },
    {
      "epoch": 3.038588235294118,
      "grad_norm": 0.44662931577714454,
      "learning_rate": 0.00018976062758382144,
      "loss": 0.4844,
      "step": 6457
    },
    {
      "epoch": 3.0390588235294116,
      "grad_norm": 0.46976792415615226,
      "learning_rate": 0.0001897183577765904,
      "loss": 0.4727,
      "step": 6458
    },
    {
      "epoch": 3.039529411764706,
      "grad_norm": 0.42676796512216764,
      "learning_rate": 0.00018967608457721684,
      "loss": 0.3359,
      "step": 6459
    },
    {
      "epoch": 3.04,
      "grad_norm": 0.7498462659391273,
      "learning_rate": 0.00018963380798931102,
      "loss": 0.3574,
      "step": 6460
    },
    {
      "epoch": 3.0404705882352943,
      "grad_norm": 0.25288366651112604,
      "learning_rate": 0.00018959152801648364,
      "loss": 0.252,
      "step": 6461
    },
    {
      "epoch": 3.040941176470588,
      "grad_norm": 0.42229950811602285,
      "learning_rate": 0.00018954924466234554,
      "loss": 0.4629,
      "step": 6462
    },
    {
      "epoch": 3.0414117647058823,
      "grad_norm": 0.30644321917849826,
      "learning_rate": 0.00018950695793050796,
      "loss": 0.1406,
      "step": 6463
    },
    {
      "epoch": 3.0418823529411765,
      "grad_norm": 0.4406486512616738,
      "learning_rate": 0.0001894646678245824,
      "loss": 0.5039,
      "step": 6464
    },
    {
      "epoch": 3.0423529411764707,
      "grad_norm": 0.47676672841632994,
      "learning_rate": 0.00018942237434818062,
      "loss": 0.2676,
      "step": 6465
    },
    {
      "epoch": 3.0428235294117645,
      "grad_norm": 0.4456299258740864,
      "learning_rate": 0.00018938007750491468,
      "loss": 0.3828,
      "step": 6466
    },
    {
      "epoch": 3.0432941176470587,
      "grad_norm": 0.4025842465771094,
      "learning_rate": 0.00018933777729839701,
      "loss": 0.3789,
      "step": 6467
    },
    {
      "epoch": 3.043764705882353,
      "grad_norm": 0.3803690761082853,
      "learning_rate": 0.0001892954737322402,
      "loss": 0.4707,
      "step": 6468
    },
    {
      "epoch": 3.044235294117647,
      "grad_norm": 0.35409092011063353,
      "learning_rate": 0.00018925316681005718,
      "loss": 0.3145,
      "step": 6469
    },
    {
      "epoch": 3.0447058823529414,
      "grad_norm": 0.46147461070832463,
      "learning_rate": 0.0001892108565354612,
      "loss": 0.3398,
      "step": 6470
    },
    {
      "epoch": 3.045176470588235,
      "grad_norm": 0.612284160113774,
      "learning_rate": 0.0001891685429120657,
      "loss": 0.5352,
      "step": 6471
    },
    {
      "epoch": 3.0456470588235294,
      "grad_norm": 0.3911516727573484,
      "learning_rate": 0.00018912622594348454,
      "loss": 0.3789,
      "step": 6472
    },
    {
      "epoch": 3.0461176470588236,
      "grad_norm": 0.15099043992196123,
      "learning_rate": 0.00018908390563333175,
      "loss": 0.0713,
      "step": 6473
    },
    {
      "epoch": 3.046588235294118,
      "grad_norm": 0.4738251984007542,
      "learning_rate": 0.00018904158198522177,
      "loss": 0.4902,
      "step": 6474
    },
    {
      "epoch": 3.0470588235294116,
      "grad_norm": 0.2695937959761277,
      "learning_rate": 0.00018899925500276912,
      "loss": 0.1768,
      "step": 6475
    },
    {
      "epoch": 3.047529411764706,
      "grad_norm": 0.39486582199557185,
      "learning_rate": 0.00018895692468958884,
      "loss": 0.6406,
      "step": 6476
    },
    {
      "epoch": 3.048,
      "grad_norm": 0.44692079756320063,
      "learning_rate": 0.00018891459104929608,
      "loss": 0.4941,
      "step": 6477
    },
    {
      "epoch": 3.0484705882352943,
      "grad_norm": 0.47144280843257735,
      "learning_rate": 0.0001888722540855064,
      "loss": 0.668,
      "step": 6478
    },
    {
      "epoch": 3.048941176470588,
      "grad_norm": 0.2868506842315221,
      "learning_rate": 0.00018882991380183556,
      "loss": 0.1436,
      "step": 6479
    },
    {
      "epoch": 3.0494117647058823,
      "grad_norm": 0.4295507296928397,
      "learning_rate": 0.0001887875702018996,
      "loss": 0.3965,
      "step": 6480
    },
    {
      "epoch": 3.0498823529411765,
      "grad_norm": 0.35554091291794154,
      "learning_rate": 0.00018874522328931492,
      "loss": 0.4551,
      "step": 6481
    },
    {
      "epoch": 3.0503529411764707,
      "grad_norm": 0.3063777850901421,
      "learning_rate": 0.0001887028730676981,
      "loss": 0.2129,
      "step": 6482
    },
    {
      "epoch": 3.0508235294117645,
      "grad_norm": 0.609609719681232,
      "learning_rate": 0.00018866051954066608,
      "loss": 0.75,
      "step": 6483
    },
    {
      "epoch": 3.0512941176470587,
      "grad_norm": 0.5368676015164708,
      "learning_rate": 0.00018861816271183613,
      "loss": 0.7695,
      "step": 6484
    },
    {
      "epoch": 3.051764705882353,
      "grad_norm": 0.32508069987076815,
      "learning_rate": 0.00018857580258482557,
      "loss": 0.2949,
      "step": 6485
    },
    {
      "epoch": 3.052235294117647,
      "grad_norm": 0.5303775749987396,
      "learning_rate": 0.0001885334391632523,
      "loss": 0.668,
      "step": 6486
    },
    {
      "epoch": 3.0527058823529414,
      "grad_norm": 0.4463738841621934,
      "learning_rate": 0.00018849107245073432,
      "loss": 0.4102,
      "step": 6487
    },
    {
      "epoch": 3.053176470588235,
      "grad_norm": 0.44906676465705925,
      "learning_rate": 0.00018844870245088996,
      "loss": 0.4355,
      "step": 6488
    },
    {
      "epoch": 3.0536470588235294,
      "grad_norm": 0.29531278564654434,
      "learning_rate": 0.00018840632916733777,
      "loss": 0.1631,
      "step": 6489
    },
    {
      "epoch": 3.0541176470588236,
      "grad_norm": 0.40559376817250864,
      "learning_rate": 0.00018836395260369676,
      "loss": 0.3711,
      "step": 6490
    },
    {
      "epoch": 3.054588235294118,
      "grad_norm": 0.44582910875477993,
      "learning_rate": 0.000188321572763586,
      "loss": 0.2988,
      "step": 6491
    },
    {
      "epoch": 3.0550588235294116,
      "grad_norm": 0.33254900887236505,
      "learning_rate": 0.00018827918965062492,
      "loss": 0.291,
      "step": 6492
    },
    {
      "epoch": 3.055529411764706,
      "grad_norm": 0.4630848535068412,
      "learning_rate": 0.00018823680326843334,
      "loss": 0.4453,
      "step": 6493
    },
    {
      "epoch": 3.056,
      "grad_norm": 0.7266032625784392,
      "learning_rate": 0.00018819441362063122,
      "loss": 0.6914,
      "step": 6494
    },
    {
      "epoch": 3.0564705882352943,
      "grad_norm": 0.2387781944251436,
      "learning_rate": 0.00018815202071083878,
      "loss": 0.2344,
      "step": 6495
    },
    {
      "epoch": 3.056941176470588,
      "grad_norm": 0.18366751416573765,
      "learning_rate": 0.00018810962454267666,
      "loss": 0.0669,
      "step": 6496
    },
    {
      "epoch": 3.0574117647058823,
      "grad_norm": 0.5777787334718518,
      "learning_rate": 0.00018806722511976567,
      "loss": 0.6797,
      "step": 6497
    },
    {
      "epoch": 3.0578823529411765,
      "grad_norm": 0.34501360440611817,
      "learning_rate": 0.00018802482244572693,
      "loss": 0.2236,
      "step": 6498
    },
    {
      "epoch": 3.0583529411764707,
      "grad_norm": 0.3578180053819686,
      "learning_rate": 0.00018798241652418182,
      "loss": 0.1777,
      "step": 6499
    },
    {
      "epoch": 3.0588235294117645,
      "grad_norm": 0.41948872071331994,
      "learning_rate": 0.00018794000735875208,
      "loss": 0.2363,
      "step": 6500
    },
    {
      "epoch": 3.0592941176470587,
      "grad_norm": 0.31330189973430916,
      "learning_rate": 0.00018789759495305963,
      "loss": 0.1094,
      "step": 6501
    },
    {
      "epoch": 3.059764705882353,
      "grad_norm": 0.3309759012732637,
      "learning_rate": 0.00018785517931072668,
      "loss": 0.3555,
      "step": 6502
    },
    {
      "epoch": 3.060235294117647,
      "grad_norm": 0.6742452321116572,
      "learning_rate": 0.0001878127604353757,
      "loss": 0.5039,
      "step": 6503
    },
    {
      "epoch": 3.0607058823529414,
      "grad_norm": 0.34818546327776506,
      "learning_rate": 0.00018777033833062954,
      "loss": 0.2207,
      "step": 6504
    },
    {
      "epoch": 3.061176470588235,
      "grad_norm": 0.8536760384270238,
      "learning_rate": 0.00018772791300011122,
      "loss": 0.668,
      "step": 6505
    },
    {
      "epoch": 3.0616470588235294,
      "grad_norm": 0.5353354025720518,
      "learning_rate": 0.00018768548444744404,
      "loss": 0.6484,
      "step": 6506
    },
    {
      "epoch": 3.0621176470588236,
      "grad_norm": 0.6059341441956212,
      "learning_rate": 0.00018764305267625169,
      "loss": 0.3887,
      "step": 6507
    },
    {
      "epoch": 3.062588235294118,
      "grad_norm": 0.49700924667172885,
      "learning_rate": 0.000187600617690158,
      "loss": 0.5078,
      "step": 6508
    },
    {
      "epoch": 3.0630588235294116,
      "grad_norm": 0.44596420941517634,
      "learning_rate": 0.00018755817949278706,
      "loss": 0.4805,
      "step": 6509
    },
    {
      "epoch": 3.063529411764706,
      "grad_norm": 0.3564087951958234,
      "learning_rate": 0.00018751573808776346,
      "loss": 0.2158,
      "step": 6510
    },
    {
      "epoch": 3.064,
      "grad_norm": 0.4409311378991516,
      "learning_rate": 0.00018747329347871178,
      "loss": 0.2891,
      "step": 6511
    },
    {
      "epoch": 3.0644705882352943,
      "grad_norm": 0.7190179419236025,
      "learning_rate": 0.00018743084566925702,
      "loss": 0.3477,
      "step": 6512
    },
    {
      "epoch": 3.064941176470588,
      "grad_norm": 0.2568982533458036,
      "learning_rate": 0.00018738839466302443,
      "loss": 0.1357,
      "step": 6513
    },
    {
      "epoch": 3.0654117647058823,
      "grad_norm": 0.36869135408766185,
      "learning_rate": 0.00018734594046363956,
      "loss": 0.291,
      "step": 6514
    },
    {
      "epoch": 3.0658823529411765,
      "grad_norm": 0.5546109268445596,
      "learning_rate": 0.00018730348307472824,
      "loss": 0.5547,
      "step": 6515
    },
    {
      "epoch": 3.0663529411764707,
      "grad_norm": 0.2673789677255203,
      "learning_rate": 0.0001872610224999164,
      "loss": 0.1533,
      "step": 6516
    },
    {
      "epoch": 3.0668235294117645,
      "grad_norm": 0.5630328332454851,
      "learning_rate": 0.00018721855874283058,
      "loss": 0.3496,
      "step": 6517
    },
    {
      "epoch": 3.0672941176470587,
      "grad_norm": 0.3948907318349721,
      "learning_rate": 0.00018717609180709721,
      "loss": 0.6367,
      "step": 6518
    },
    {
      "epoch": 3.067764705882353,
      "grad_norm": 0.3453534423127653,
      "learning_rate": 0.00018713362169634325,
      "loss": 0.3945,
      "step": 6519
    },
    {
      "epoch": 3.068235294117647,
      "grad_norm": 0.483725204088955,
      "learning_rate": 0.00018709114841419588,
      "loss": 0.4766,
      "step": 6520
    },
    {
      "epoch": 3.0687058823529414,
      "grad_norm": 0.20487414651390648,
      "learning_rate": 0.0001870486719642825,
      "loss": 0.1187,
      "step": 6521
    },
    {
      "epoch": 3.069176470588235,
      "grad_norm": 0.5937122083543982,
      "learning_rate": 0.00018700619235023085,
      "loss": 0.5195,
      "step": 6522
    },
    {
      "epoch": 3.0696470588235294,
      "grad_norm": 0.7901585052447908,
      "learning_rate": 0.00018696370957566882,
      "loss": 0.8008,
      "step": 6523
    },
    {
      "epoch": 3.0701176470588236,
      "grad_norm": 0.6251093777216314,
      "learning_rate": 0.0001869212236442247,
      "loss": 0.7266,
      "step": 6524
    },
    {
      "epoch": 3.070588235294118,
      "grad_norm": 0.5049158256959468,
      "learning_rate": 0.00018687873455952697,
      "loss": 0.3145,
      "step": 6525
    },
    {
      "epoch": 3.0710588235294116,
      "grad_norm": 0.3999706660809434,
      "learning_rate": 0.00018683624232520446,
      "loss": 0.2734,
      "step": 6526
    },
    {
      "epoch": 3.071529411764706,
      "grad_norm": 0.3737133070281763,
      "learning_rate": 0.00018679374694488608,
      "loss": 0.4453,
      "step": 6527
    },
    {
      "epoch": 3.072,
      "grad_norm": 1.0310523091055013,
      "learning_rate": 0.00018675124842220135,
      "loss": 0.8281,
      "step": 6528
    },
    {
      "epoch": 3.0724705882352943,
      "grad_norm": 0.2541189825787483,
      "learning_rate": 0.0001867087467607797,
      "loss": 0.2734,
      "step": 6529
    },
    {
      "epoch": 3.072941176470588,
      "grad_norm": 0.624509340129677,
      "learning_rate": 0.00018666624196425098,
      "loss": 0.4238,
      "step": 6530
    },
    {
      "epoch": 3.0734117647058823,
      "grad_norm": 0.519495982913243,
      "learning_rate": 0.00018662373403624538,
      "loss": 0.4434,
      "step": 6531
    },
    {
      "epoch": 3.0738823529411765,
      "grad_norm": 0.45765661494455145,
      "learning_rate": 0.00018658122298039327,
      "loss": 0.4355,
      "step": 6532
    },
    {
      "epoch": 3.0743529411764707,
      "grad_norm": 0.23456055928220662,
      "learning_rate": 0.00018653870880032527,
      "loss": 0.0845,
      "step": 6533
    },
    {
      "epoch": 3.0748235294117645,
      "grad_norm": 0.47515778441215145,
      "learning_rate": 0.00018649619149967232,
      "loss": 0.5469,
      "step": 6534
    },
    {
      "epoch": 3.0752941176470587,
      "grad_norm": 0.2076060018502123,
      "learning_rate": 0.00018645367108206565,
      "loss": 0.053,
      "step": 6535
    },
    {
      "epoch": 3.075764705882353,
      "grad_norm": 0.6308671830135824,
      "learning_rate": 0.00018641114755113658,
      "loss": 0.3086,
      "step": 6536
    },
    {
      "epoch": 3.076235294117647,
      "grad_norm": 0.5332550927995271,
      "learning_rate": 0.00018636862091051697,
      "loss": 0.3555,
      "step": 6537
    },
    {
      "epoch": 3.0767058823529414,
      "grad_norm": 0.3324266544824948,
      "learning_rate": 0.00018632609116383874,
      "loss": 0.1289,
      "step": 6538
    },
    {
      "epoch": 3.077176470588235,
      "grad_norm": 0.34466446519231847,
      "learning_rate": 0.00018628355831473416,
      "loss": 0.3027,
      "step": 6539
    },
    {
      "epoch": 3.0776470588235294,
      "grad_norm": 0.4096424017815468,
      "learning_rate": 0.00018624102236683567,
      "loss": 0.4141,
      "step": 6540
    },
    {
      "epoch": 3.0781176470588236,
      "grad_norm": 0.41677438931319355,
      "learning_rate": 0.00018619848332377615,
      "loss": 0.1533,
      "step": 6541
    },
    {
      "epoch": 3.078588235294118,
      "grad_norm": 0.5371093319671301,
      "learning_rate": 0.00018615594118918862,
      "loss": 0.4375,
      "step": 6542
    },
    {
      "epoch": 3.0790588235294116,
      "grad_norm": 0.30345999959074504,
      "learning_rate": 0.00018611339596670632,
      "loss": 0.3301,
      "step": 6543
    },
    {
      "epoch": 3.079529411764706,
      "grad_norm": 0.44506128131994366,
      "learning_rate": 0.00018607084765996285,
      "loss": 0.3906,
      "step": 6544
    },
    {
      "epoch": 3.08,
      "grad_norm": 0.5342395971434281,
      "learning_rate": 0.00018602829627259213,
      "loss": 0.4863,
      "step": 6545
    },
    {
      "epoch": 3.0804705882352943,
      "grad_norm": 0.46566809958803534,
      "learning_rate": 0.00018598574180822818,
      "loss": 0.4219,
      "step": 6546
    },
    {
      "epoch": 3.080941176470588,
      "grad_norm": 0.48128358318508746,
      "learning_rate": 0.00018594318427050537,
      "loss": 0.4238,
      "step": 6547
    },
    {
      "epoch": 3.0814117647058823,
      "grad_norm": 0.4977103634469761,
      "learning_rate": 0.00018590062366305828,
      "loss": 0.5742,
      "step": 6548
    },
    {
      "epoch": 3.0818823529411765,
      "grad_norm": 0.5582134753469252,
      "learning_rate": 0.0001858580599895219,
      "loss": 0.373,
      "step": 6549
    },
    {
      "epoch": 3.0823529411764707,
      "grad_norm": 0.4069511541544268,
      "learning_rate": 0.00018581549325353126,
      "loss": 0.4316,
      "step": 6550
    },
    {
      "epoch": 3.0828235294117645,
      "grad_norm": 0.3428526900117534,
      "learning_rate": 0.00018577292345872188,
      "loss": 0.2139,
      "step": 6551
    },
    {
      "epoch": 3.0832941176470587,
      "grad_norm": 0.5476815769763196,
      "learning_rate": 0.00018573035060872933,
      "loss": 0.4492,
      "step": 6552
    },
    {
      "epoch": 3.083764705882353,
      "grad_norm": 0.33161406701725504,
      "learning_rate": 0.0001856877747071896,
      "loss": 0.4199,
      "step": 6553
    },
    {
      "epoch": 3.084235294117647,
      "grad_norm": 0.273057015561803,
      "learning_rate": 0.00018564519575773884,
      "loss": 0.0967,
      "step": 6554
    },
    {
      "epoch": 3.084705882352941,
      "grad_norm": 0.3234166270588043,
      "learning_rate": 0.00018560261376401356,
      "loss": 0.3809,
      "step": 6555
    },
    {
      "epoch": 3.085176470588235,
      "grad_norm": 0.37084715822834896,
      "learning_rate": 0.0001855600287296504,
      "loss": 0.1875,
      "step": 6556
    },
    {
      "epoch": 3.0856470588235294,
      "grad_norm": 0.6153442298636812,
      "learning_rate": 0.00018551744065828634,
      "loss": 0.457,
      "step": 6557
    },
    {
      "epoch": 3.0861176470588236,
      "grad_norm": 0.6415170796988818,
      "learning_rate": 0.0001854748495535587,
      "loss": 0.6445,
      "step": 6558
    },
    {
      "epoch": 3.086588235294118,
      "grad_norm": 0.46400634207952435,
      "learning_rate": 0.0001854322554191049,
      "loss": 0.25,
      "step": 6559
    },
    {
      "epoch": 3.0870588235294116,
      "grad_norm": 0.5403757005239229,
      "learning_rate": 0.00018538965825856264,
      "loss": 0.5508,
      "step": 6560
    },
    {
      "epoch": 3.087529411764706,
      "grad_norm": 0.180853479717156,
      "learning_rate": 0.00018534705807557003,
      "loss": 0.053,
      "step": 6561
    },
    {
      "epoch": 3.088,
      "grad_norm": 0.25703499163373483,
      "learning_rate": 0.00018530445487376524,
      "loss": 0.0737,
      "step": 6562
    },
    {
      "epoch": 3.0884705882352943,
      "grad_norm": 1.7701668396038608,
      "learning_rate": 0.0001852618486567869,
      "loss": 2.0938,
      "step": 6563
    },
    {
      "epoch": 3.088941176470588,
      "grad_norm": 0.5562147134916513,
      "learning_rate": 0.00018521923942827362,
      "loss": 0.4727,
      "step": 6564
    },
    {
      "epoch": 3.0894117647058823,
      "grad_norm": 0.4154165405607453,
      "learning_rate": 0.00018517662719186464,
      "loss": 0.2676,
      "step": 6565
    },
    {
      "epoch": 3.0898823529411765,
      "grad_norm": 0.4913483761253304,
      "learning_rate": 0.0001851340119511991,
      "loss": 0.5586,
      "step": 6566
    },
    {
      "epoch": 3.0903529411764707,
      "grad_norm": 0.4652703939328818,
      "learning_rate": 0.00018509139370991666,
      "loss": 0.5703,
      "step": 6567
    },
    {
      "epoch": 3.0908235294117645,
      "grad_norm": 0.2088596172254034,
      "learning_rate": 0.000185048772471657,
      "loss": 0.0728,
      "step": 6568
    },
    {
      "epoch": 3.0912941176470587,
      "grad_norm": 0.3903830245528807,
      "learning_rate": 0.00018500614824006035,
      "loss": 0.3613,
      "step": 6569
    },
    {
      "epoch": 3.091764705882353,
      "grad_norm": 0.4166783122164419,
      "learning_rate": 0.0001849635210187669,
      "loss": 0.3906,
      "step": 6570
    },
    {
      "epoch": 3.092235294117647,
      "grad_norm": 0.5246166093795865,
      "learning_rate": 0.00018492089081141724,
      "loss": 0.4629,
      "step": 6571
    },
    {
      "epoch": 3.092705882352941,
      "grad_norm": 0.52399835578277,
      "learning_rate": 0.00018487825762165223,
      "loss": 0.2969,
      "step": 6572
    },
    {
      "epoch": 3.093176470588235,
      "grad_norm": 0.42131054436757154,
      "learning_rate": 0.00018483562145311296,
      "loss": 0.293,
      "step": 6573
    },
    {
      "epoch": 3.0936470588235294,
      "grad_norm": 0.48447193767904567,
      "learning_rate": 0.00018479298230944071,
      "loss": 0.582,
      "step": 6574
    },
    {
      "epoch": 3.0941176470588236,
      "grad_norm": 0.5128573335354649,
      "learning_rate": 0.00018475034019427717,
      "loss": 0.4727,
      "step": 6575
    },
    {
      "epoch": 3.094588235294118,
      "grad_norm": 0.24828854953500748,
      "learning_rate": 0.00018470769511126409,
      "loss": 0.105,
      "step": 6576
    },
    {
      "epoch": 3.0950588235294116,
      "grad_norm": 0.35335334415544867,
      "learning_rate": 0.00018466504706404362,
      "loss": 0.4492,
      "step": 6577
    },
    {
      "epoch": 3.095529411764706,
      "grad_norm": 0.5780474317007441,
      "learning_rate": 0.00018462239605625808,
      "loss": 0.375,
      "step": 6578
    },
    {
      "epoch": 3.096,
      "grad_norm": 0.4994347778012882,
      "learning_rate": 0.0001845797420915501,
      "loss": 0.4375,
      "step": 6579
    },
    {
      "epoch": 3.0964705882352943,
      "grad_norm": 0.306803260930917,
      "learning_rate": 0.00018453708517356256,
      "loss": 0.1079,
      "step": 6580
    },
    {
      "epoch": 3.096941176470588,
      "grad_norm": 0.4426179948579079,
      "learning_rate": 0.0001844944253059385,
      "loss": 0.4863,
      "step": 6581
    },
    {
      "epoch": 3.0974117647058823,
      "grad_norm": 0.40238778403649234,
      "learning_rate": 0.00018445176249232134,
      "loss": 0.3555,
      "step": 6582
    },
    {
      "epoch": 3.0978823529411765,
      "grad_norm": 0.45453436194260677,
      "learning_rate": 0.0001844090967363547,
      "loss": 0.6406,
      "step": 6583
    },
    {
      "epoch": 3.0983529411764708,
      "grad_norm": 0.5366610068697091,
      "learning_rate": 0.0001843664280416824,
      "loss": 0.6328,
      "step": 6584
    },
    {
      "epoch": 3.0988235294117645,
      "grad_norm": 0.4206165830168188,
      "learning_rate": 0.0001843237564119486,
      "loss": 0.498,
      "step": 6585
    },
    {
      "epoch": 3.0992941176470588,
      "grad_norm": 0.30709921777627647,
      "learning_rate": 0.00018428108185079763,
      "loss": 0.1475,
      "step": 6586
    },
    {
      "epoch": 3.099764705882353,
      "grad_norm": 0.1400803263540728,
      "learning_rate": 0.00018423840436187412,
      "loss": 0.1016,
      "step": 6587
    },
    {
      "epoch": 3.100235294117647,
      "grad_norm": 0.6802602237855578,
      "learning_rate": 0.0001841957239488229,
      "loss": 0.5859,
      "step": 6588
    },
    {
      "epoch": 3.100705882352941,
      "grad_norm": 0.46079318882608533,
      "learning_rate": 0.00018415304061528918,
      "loss": 0.4355,
      "step": 6589
    },
    {
      "epoch": 3.101176470588235,
      "grad_norm": 0.37557617090751366,
      "learning_rate": 0.00018411035436491827,
      "loss": 0.4102,
      "step": 6590
    },
    {
      "epoch": 3.1016470588235294,
      "grad_norm": 0.37812036083811873,
      "learning_rate": 0.0001840676652013558,
      "loss": 0.6875,
      "step": 6591
    },
    {
      "epoch": 3.1021176470588236,
      "grad_norm": 0.15367349172369213,
      "learning_rate": 0.00018402497312824758,
      "loss": 0.0574,
      "step": 6592
    },
    {
      "epoch": 3.1025882352941174,
      "grad_norm": 0.38823327823297404,
      "learning_rate": 0.0001839822781492398,
      "loss": 0.5469,
      "step": 6593
    },
    {
      "epoch": 3.1030588235294116,
      "grad_norm": 0.1549810130110472,
      "learning_rate": 0.00018393958026797878,
      "loss": 0.063,
      "step": 6594
    },
    {
      "epoch": 3.103529411764706,
      "grad_norm": 0.6633605733704865,
      "learning_rate": 0.0001838968794881111,
      "loss": 0.4551,
      "step": 6595
    },
    {
      "epoch": 3.104,
      "grad_norm": 0.48367742300079325,
      "learning_rate": 0.00018385417581328364,
      "loss": 0.3125,
      "step": 6596
    },
    {
      "epoch": 3.1044705882352943,
      "grad_norm": 0.529850726688539,
      "learning_rate": 0.00018381146924714356,
      "loss": 0.793,
      "step": 6597
    },
    {
      "epoch": 3.104941176470588,
      "grad_norm": 0.3401336628028912,
      "learning_rate": 0.00018376875979333812,
      "loss": 0.1807,
      "step": 6598
    },
    {
      "epoch": 3.1054117647058823,
      "grad_norm": 3.41206162715995,
      "learning_rate": 0.000183726047455515,
      "loss": 2.0312,
      "step": 6599
    },
    {
      "epoch": 3.1058823529411765,
      "grad_norm": 1.3612109036044506,
      "learning_rate": 0.000183683332237322,
      "loss": 1.3047,
      "step": 6600
    },
    {
      "epoch": 3.1063529411764708,
      "grad_norm": 0.5134492339274216,
      "learning_rate": 0.00018364061414240717,
      "loss": 0.6719,
      "step": 6601
    },
    {
      "epoch": 3.1068235294117645,
      "grad_norm": 0.4105260141800564,
      "learning_rate": 0.00018359789317441892,
      "loss": 0.2061,
      "step": 6602
    },
    {
      "epoch": 3.1072941176470588,
      "grad_norm": 0.5316050375778808,
      "learning_rate": 0.00018355516933700578,
      "loss": 0.3789,
      "step": 6603
    },
    {
      "epoch": 3.107764705882353,
      "grad_norm": 0.5498674454193215,
      "learning_rate": 0.0001835124426338166,
      "loss": 0.7891,
      "step": 6604
    },
    {
      "epoch": 3.108235294117647,
      "grad_norm": 0.678742396614753,
      "learning_rate": 0.00018346971306850045,
      "loss": 0.5859,
      "step": 6605
    },
    {
      "epoch": 3.108705882352941,
      "grad_norm": 0.4435151706494922,
      "learning_rate": 0.00018342698064470662,
      "loss": 0.2275,
      "step": 6606
    },
    {
      "epoch": 3.109176470588235,
      "grad_norm": 0.9358751327537346,
      "learning_rate": 0.0001833842453660847,
      "loss": 0.8125,
      "step": 6607
    },
    {
      "epoch": 3.1096470588235294,
      "grad_norm": 0.8480742524862136,
      "learning_rate": 0.00018334150723628442,
      "loss": 0.3066,
      "step": 6608
    },
    {
      "epoch": 3.1101176470588237,
      "grad_norm": 0.41044297235778726,
      "learning_rate": 0.00018329876625895592,
      "loss": 0.2578,
      "step": 6609
    },
    {
      "epoch": 3.1105882352941174,
      "grad_norm": 0.41774472412623015,
      "learning_rate": 0.0001832560224377495,
      "loss": 0.4316,
      "step": 6610
    },
    {
      "epoch": 3.1110588235294117,
      "grad_norm": 0.2420265110617991,
      "learning_rate": 0.0001832132757763156,
      "loss": 0.2197,
      "step": 6611
    },
    {
      "epoch": 3.111529411764706,
      "grad_norm": 0.5004849511444127,
      "learning_rate": 0.000183170526278305,
      "loss": 0.6406,
      "step": 6612
    },
    {
      "epoch": 3.112,
      "grad_norm": 0.2975581410673076,
      "learning_rate": 0.0001831277739473688,
      "loss": 0.1104,
      "step": 6613
    },
    {
      "epoch": 3.1124705882352943,
      "grad_norm": 0.43376390665725806,
      "learning_rate": 0.00018308501878715823,
      "loss": 0.4668,
      "step": 6614
    },
    {
      "epoch": 3.112941176470588,
      "grad_norm": 0.484313261823687,
      "learning_rate": 0.00018304226080132472,
      "loss": 0.5078,
      "step": 6615
    },
    {
      "epoch": 3.1134117647058823,
      "grad_norm": 0.23366828429715655,
      "learning_rate": 0.00018299949999352013,
      "loss": 0.1138,
      "step": 6616
    },
    {
      "epoch": 3.1138823529411765,
      "grad_norm": 0.523626353598207,
      "learning_rate": 0.00018295673636739635,
      "loss": 0.8398,
      "step": 6617
    },
    {
      "epoch": 3.1143529411764708,
      "grad_norm": 0.31264480504614345,
      "learning_rate": 0.00018291396992660565,
      "loss": 0.1973,
      "step": 6618
    },
    {
      "epoch": 3.1148235294117645,
      "grad_norm": 0.2908010744092448,
      "learning_rate": 0.00018287120067480046,
      "loss": 0.209,
      "step": 6619
    },
    {
      "epoch": 3.1152941176470588,
      "grad_norm": 0.21760149982560745,
      "learning_rate": 0.0001828284286156335,
      "loss": 0.0688,
      "step": 6620
    },
    {
      "epoch": 3.115764705882353,
      "grad_norm": 0.29058733496494615,
      "learning_rate": 0.0001827856537527577,
      "loss": 0.1416,
      "step": 6621
    },
    {
      "epoch": 3.116235294117647,
      "grad_norm": 0.40507599969690633,
      "learning_rate": 0.00018274287608982626,
      "loss": 0.5078,
      "step": 6622
    },
    {
      "epoch": 3.116705882352941,
      "grad_norm": 0.2978648519074462,
      "learning_rate": 0.00018270009563049264,
      "loss": 0.126,
      "step": 6623
    },
    {
      "epoch": 3.117176470588235,
      "grad_norm": 0.30545270242519296,
      "learning_rate": 0.00018265731237841043,
      "loss": 0.1582,
      "step": 6624
    },
    {
      "epoch": 3.1176470588235294,
      "grad_norm": 0.4390386623619695,
      "learning_rate": 0.00018261452633723356,
      "loss": 0.2754,
      "step": 6625
    },
    {
      "epoch": 3.1181176470588237,
      "grad_norm": 0.5512663241501958,
      "learning_rate": 0.00018257173751061613,
      "loss": 0.7305,
      "step": 6626
    },
    {
      "epoch": 3.1185882352941174,
      "grad_norm": 0.3821876822271587,
      "learning_rate": 0.00018252894590221262,
      "loss": 0.2109,
      "step": 6627
    },
    {
      "epoch": 3.1190588235294117,
      "grad_norm": 0.5860307754032507,
      "learning_rate": 0.00018248615151567754,
      "loss": 0.7383,
      "step": 6628
    },
    {
      "epoch": 3.119529411764706,
      "grad_norm": 0.6281837266358681,
      "learning_rate": 0.00018244335435466577,
      "loss": 0.8945,
      "step": 6629
    },
    {
      "epoch": 3.12,
      "grad_norm": 0.46651128653528784,
      "learning_rate": 0.00018240055442283242,
      "loss": 0.6914,
      "step": 6630
    },
    {
      "epoch": 3.1204705882352943,
      "grad_norm": 0.3936117286008597,
      "learning_rate": 0.00018235775172383277,
      "loss": 0.5352,
      "step": 6631
    },
    {
      "epoch": 3.120941176470588,
      "grad_norm": 0.5536482631799661,
      "learning_rate": 0.00018231494626132243,
      "loss": 0.6602,
      "step": 6632
    },
    {
      "epoch": 3.1214117647058823,
      "grad_norm": 0.33560294113882355,
      "learning_rate": 0.00018227213803895717,
      "loss": 0.1748,
      "step": 6633
    },
    {
      "epoch": 3.1218823529411766,
      "grad_norm": 0.5164316329936188,
      "learning_rate": 0.00018222932706039307,
      "loss": 0.377,
      "step": 6634
    },
    {
      "epoch": 3.1223529411764708,
      "grad_norm": 0.2584639425588727,
      "learning_rate": 0.0001821865133292863,
      "loss": 0.2832,
      "step": 6635
    },
    {
      "epoch": 3.1228235294117646,
      "grad_norm": 0.5944885692924261,
      "learning_rate": 0.00018214369684929347,
      "loss": 0.832,
      "step": 6636
    },
    {
      "epoch": 3.1232941176470588,
      "grad_norm": 0.3616268332378625,
      "learning_rate": 0.00018210087762407122,
      "loss": 0.2148,
      "step": 6637
    },
    {
      "epoch": 3.123764705882353,
      "grad_norm": 0.2658562155261841,
      "learning_rate": 0.00018205805565727658,
      "loss": 0.1318,
      "step": 6638
    },
    {
      "epoch": 3.124235294117647,
      "grad_norm": 0.3536790604335437,
      "learning_rate": 0.00018201523095256672,
      "loss": 0.1885,
      "step": 6639
    },
    {
      "epoch": 3.124705882352941,
      "grad_norm": 0.31805423845070746,
      "learning_rate": 0.00018197240351359913,
      "loss": 0.1484,
      "step": 6640
    },
    {
      "epoch": 3.1251764705882352,
      "grad_norm": 0.35021649821370143,
      "learning_rate": 0.0001819295733440315,
      "loss": 0.3145,
      "step": 6641
    },
    {
      "epoch": 3.1256470588235294,
      "grad_norm": 0.3459665876065629,
      "learning_rate": 0.00018188674044752163,
      "loss": 0.1846,
      "step": 6642
    },
    {
      "epoch": 3.1261176470588237,
      "grad_norm": 0.3641578383103261,
      "learning_rate": 0.00018184390482772774,
      "loss": 0.2021,
      "step": 6643
    },
    {
      "epoch": 3.1265882352941174,
      "grad_norm": 0.5450793116597854,
      "learning_rate": 0.0001818010664883082,
      "loss": 0.6016,
      "step": 6644
    },
    {
      "epoch": 3.1270588235294117,
      "grad_norm": 0.3775561991343749,
      "learning_rate": 0.0001817582254329216,
      "loss": 0.3418,
      "step": 6645
    },
    {
      "epoch": 3.127529411764706,
      "grad_norm": 0.6220154653344324,
      "learning_rate": 0.0001817153816652267,
      "loss": 0.7422,
      "step": 6646
    },
    {
      "epoch": 3.128,
      "grad_norm": 1.0355539034009393,
      "learning_rate": 0.0001816725351888827,
      "loss": 0.7852,
      "step": 6647
    },
    {
      "epoch": 3.1284705882352943,
      "grad_norm": 0.5373987081899858,
      "learning_rate": 0.00018162968600754888,
      "loss": 0.4355,
      "step": 6648
    },
    {
      "epoch": 3.128941176470588,
      "grad_norm": 0.3344012012767351,
      "learning_rate": 0.00018158683412488466,
      "loss": 0.1553,
      "step": 6649
    },
    {
      "epoch": 3.1294117647058823,
      "grad_norm": 0.3833622984882053,
      "learning_rate": 0.00018154397954454993,
      "loss": 0.1709,
      "step": 6650
    },
    {
      "epoch": 3.1298823529411766,
      "grad_norm": 0.5264121445711082,
      "learning_rate": 0.0001815011222702046,
      "loss": 0.5273,
      "step": 6651
    },
    {
      "epoch": 3.130352941176471,
      "grad_norm": 0.37721986218895015,
      "learning_rate": 0.00018145826230550892,
      "loss": 0.167,
      "step": 6652
    },
    {
      "epoch": 3.1308235294117646,
      "grad_norm": 0.4227877463590345,
      "learning_rate": 0.00018141539965412334,
      "loss": 0.3574,
      "step": 6653
    },
    {
      "epoch": 3.131294117647059,
      "grad_norm": 0.21960118557703412,
      "learning_rate": 0.00018137253431970853,
      "loss": 0.0442,
      "step": 6654
    },
    {
      "epoch": 3.131764705882353,
      "grad_norm": 0.31717175779310275,
      "learning_rate": 0.0001813296663059254,
      "loss": 0.2793,
      "step": 6655
    },
    {
      "epoch": 3.1322352941176472,
      "grad_norm": 0.377003345599537,
      "learning_rate": 0.00018128679561643508,
      "loss": 0.3496,
      "step": 6656
    },
    {
      "epoch": 3.132705882352941,
      "grad_norm": 0.2079959292758527,
      "learning_rate": 0.000181243922254899,
      "loss": 0.0413,
      "step": 6657
    },
    {
      "epoch": 3.1331764705882352,
      "grad_norm": 0.4388227868992414,
      "learning_rate": 0.0001812010462249787,
      "loss": 0.1279,
      "step": 6658
    },
    {
      "epoch": 3.1336470588235295,
      "grad_norm": 0.3795549728029416,
      "learning_rate": 0.00018115816753033601,
      "loss": 0.2852,
      "step": 6659
    },
    {
      "epoch": 3.1341176470588237,
      "grad_norm": 0.6862599449306204,
      "learning_rate": 0.00018111528617463296,
      "loss": 0.4941,
      "step": 6660
    },
    {
      "epoch": 3.1345882352941175,
      "grad_norm": 0.4974852566522185,
      "learning_rate": 0.00018107240216153184,
      "loss": 0.3887,
      "step": 6661
    },
    {
      "epoch": 3.1350588235294117,
      "grad_norm": 0.5586495267121997,
      "learning_rate": 0.00018102951549469522,
      "loss": 0.7695,
      "step": 6662
    },
    {
      "epoch": 3.135529411764706,
      "grad_norm": 0.626032014918633,
      "learning_rate": 0.00018098662617778573,
      "loss": 0.6094,
      "step": 6663
    },
    {
      "epoch": 3.136,
      "grad_norm": 0.543377477370249,
      "learning_rate": 0.0001809437342144664,
      "loss": 0.6445,
      "step": 6664
    },
    {
      "epoch": 3.1364705882352943,
      "grad_norm": 0.5216940303825817,
      "learning_rate": 0.00018090083960840042,
      "loss": 0.3105,
      "step": 6665
    },
    {
      "epoch": 3.136941176470588,
      "grad_norm": 0.4868817681564056,
      "learning_rate": 0.0001808579423632511,
      "loss": 0.4277,
      "step": 6666
    },
    {
      "epoch": 3.1374117647058823,
      "grad_norm": 0.3934712296477174,
      "learning_rate": 0.0001808150424826822,
      "loss": 0.4082,
      "step": 6667
    },
    {
      "epoch": 3.1378823529411766,
      "grad_norm": 0.3043254765052008,
      "learning_rate": 0.0001807721399703575,
      "loss": 0.1045,
      "step": 6668
    },
    {
      "epoch": 3.138352941176471,
      "grad_norm": 0.30081686164951876,
      "learning_rate": 0.0001807292348299411,
      "loss": 0.1357,
      "step": 6669
    },
    {
      "epoch": 3.1388235294117646,
      "grad_norm": 0.5516950344550796,
      "learning_rate": 0.00018068632706509733,
      "loss": 0.6055,
      "step": 6670
    },
    {
      "epoch": 3.139294117647059,
      "grad_norm": 0.4265065112953929,
      "learning_rate": 0.00018064341667949074,
      "loss": 0.4492,
      "step": 6671
    },
    {
      "epoch": 3.139764705882353,
      "grad_norm": 0.6167103084382161,
      "learning_rate": 0.00018060050367678607,
      "loss": 0.6758,
      "step": 6672
    },
    {
      "epoch": 3.1402352941176472,
      "grad_norm": 0.2708006058138402,
      "learning_rate": 0.00018055758806064822,
      "loss": 0.2637,
      "step": 6673
    },
    {
      "epoch": 3.140705882352941,
      "grad_norm": 0.5085376787833807,
      "learning_rate": 0.00018051466983474252,
      "loss": 0.4219,
      "step": 6674
    },
    {
      "epoch": 3.1411764705882352,
      "grad_norm": 0.4303745708599374,
      "learning_rate": 0.00018047174900273435,
      "loss": 0.5547,
      "step": 6675
    },
    {
      "epoch": 3.1416470588235295,
      "grad_norm": 0.27345769790696606,
      "learning_rate": 0.00018042882556828927,
      "loss": 0.1758,
      "step": 6676
    },
    {
      "epoch": 3.1421176470588237,
      "grad_norm": 35.4284609273203,
      "learning_rate": 0.0001803858995350733,
      "loss": 1.9141,
      "step": 6677
    },
    {
      "epoch": 3.1425882352941175,
      "grad_norm": 0.48627032457982955,
      "learning_rate": 0.00018034297090675246,
      "loss": 0.3438,
      "step": 6678
    },
    {
      "epoch": 3.1430588235294117,
      "grad_norm": 4.465147826329886,
      "learning_rate": 0.00018030003968699308,
      "loss": 0.6758,
      "step": 6679
    },
    {
      "epoch": 3.143529411764706,
      "grad_norm": 0.6207161840022892,
      "learning_rate": 0.00018025710587946166,
      "loss": 0.1299,
      "step": 6680
    },
    {
      "epoch": 3.144,
      "grad_norm": 5.951569237043299,
      "learning_rate": 0.00018021416948782503,
      "loss": 0.8125,
      "step": 6681
    },
    {
      "epoch": 3.1444705882352944,
      "grad_norm": 8.147621521228984,
      "learning_rate": 0.00018017123051575012,
      "loss": 1.4375,
      "step": 6682
    },
    {
      "epoch": 3.144941176470588,
      "grad_norm": 0.393858487427852,
      "learning_rate": 0.0001801282889669041,
      "loss": 0.3496,
      "step": 6683
    },
    {
      "epoch": 3.1454117647058824,
      "grad_norm": 0.36922908579149205,
      "learning_rate": 0.00018008534484495445,
      "loss": 0.3633,
      "step": 6684
    },
    {
      "epoch": 3.1458823529411766,
      "grad_norm": 0.6185090525444036,
      "learning_rate": 0.0001800423981535688,
      "loss": 0.6016,
      "step": 6685
    },
    {
      "epoch": 3.1463529411764704,
      "grad_norm": 0.4209134122539208,
      "learning_rate": 0.000179999448896415,
      "loss": 0.2734,
      "step": 6686
    },
    {
      "epoch": 3.1468235294117646,
      "grad_norm": 0.7809016764496919,
      "learning_rate": 0.00017995649707716103,
      "loss": 0.6133,
      "step": 6687
    },
    {
      "epoch": 3.147294117647059,
      "grad_norm": 4.7342655149744575,
      "learning_rate": 0.00017991354269947533,
      "loss": 0.625,
      "step": 6688
    },
    {
      "epoch": 3.147764705882353,
      "grad_norm": 1.054866837632762,
      "learning_rate": 0.00017987058576702638,
      "loss": 0.1836,
      "step": 6689
    },
    {
      "epoch": 3.1482352941176472,
      "grad_norm": 0.5007448634099979,
      "learning_rate": 0.00017982762628348286,
      "loss": 0.2129,
      "step": 6690
    },
    {
      "epoch": 3.148705882352941,
      "grad_norm": 1.0615412320128927,
      "learning_rate": 0.00017978466425251375,
      "loss": 0.8945,
      "step": 6691
    },
    {
      "epoch": 3.1491764705882352,
      "grad_norm": 0.6764125759675217,
      "learning_rate": 0.00017974169967778825,
      "loss": 0.6367,
      "step": 6692
    },
    {
      "epoch": 3.1496470588235295,
      "grad_norm": 1.0445663862910661,
      "learning_rate": 0.00017969873256297572,
      "loss": 0.8438,
      "step": 6693
    },
    {
      "epoch": 3.1501176470588237,
      "grad_norm": 0.4838300736137761,
      "learning_rate": 0.0001796557629117457,
      "loss": 0.3496,
      "step": 6694
    },
    {
      "epoch": 3.1505882352941175,
      "grad_norm": 0.6775341257090123,
      "learning_rate": 0.00017961279072776812,
      "loss": 0.4902,
      "step": 6695
    },
    {
      "epoch": 3.1510588235294117,
      "grad_norm": 0.6782337429169029,
      "learning_rate": 0.00017956981601471294,
      "loss": 0.4863,
      "step": 6696
    },
    {
      "epoch": 3.151529411764706,
      "grad_norm": 0.6646148314230556,
      "learning_rate": 0.00017952683877625042,
      "loss": 0.4375,
      "step": 6697
    },
    {
      "epoch": 3.152,
      "grad_norm": 3.932765560034739,
      "learning_rate": 0.00017948385901605106,
      "loss": 0.7461,
      "step": 6698
    },
    {
      "epoch": 3.152470588235294,
      "grad_norm": 0.512145691621178,
      "learning_rate": 0.0001794408767377855,
      "loss": 0.2734,
      "step": 6699
    },
    {
      "epoch": 3.152941176470588,
      "grad_norm": 0.45644030024478793,
      "learning_rate": 0.00017939789194512472,
      "loss": 0.3984,
      "step": 6700
    },
    {
      "epoch": 3.1534117647058824,
      "grad_norm": 0.6848594159028834,
      "learning_rate": 0.0001793549046417397,
      "loss": 0.4121,
      "step": 6701
    },
    {
      "epoch": 3.1538823529411766,
      "grad_norm": 0.7293557690112936,
      "learning_rate": 0.00017931191483130186,
      "loss": 0.5391,
      "step": 6702
    },
    {
      "epoch": 3.1543529411764704,
      "grad_norm": 0.44162923219098843,
      "learning_rate": 0.00017926892251748274,
      "loss": 0.2373,
      "step": 6703
    },
    {
      "epoch": 3.1548235294117646,
      "grad_norm": 0.5088773203009701,
      "learning_rate": 0.00017922592770395405,
      "loss": 0.3672,
      "step": 6704
    },
    {
      "epoch": 3.155294117647059,
      "grad_norm": 0.6335379576531883,
      "learning_rate": 0.00017918293039438781,
      "loss": 0.5312,
      "step": 6705
    },
    {
      "epoch": 3.155764705882353,
      "grad_norm": 0.6968738271048888,
      "learning_rate": 0.00017913993059245616,
      "loss": 0.5703,
      "step": 6706
    },
    {
      "epoch": 3.1562352941176472,
      "grad_norm": 0.43361016751392434,
      "learning_rate": 0.00017909692830183152,
      "loss": 0.3809,
      "step": 6707
    },
    {
      "epoch": 3.156705882352941,
      "grad_norm": 0.4056927469574187,
      "learning_rate": 0.00017905392352618647,
      "loss": 0.3613,
      "step": 6708
    },
    {
      "epoch": 3.1571764705882353,
      "grad_norm": 0.46285301114377725,
      "learning_rate": 0.00017901091626919383,
      "loss": 0.2334,
      "step": 6709
    },
    {
      "epoch": 3.1576470588235295,
      "grad_norm": 0.4869408186800729,
      "learning_rate": 0.00017896790653452668,
      "loss": 0.5195,
      "step": 6710
    },
    {
      "epoch": 3.1581176470588237,
      "grad_norm": 0.2675093260864713,
      "learning_rate": 0.00017892489432585817,
      "loss": 0.0586,
      "step": 6711
    },
    {
      "epoch": 3.1585882352941175,
      "grad_norm": 0.5808890346850393,
      "learning_rate": 0.00017888187964686182,
      "loss": 0.7969,
      "step": 6712
    },
    {
      "epoch": 3.1590588235294117,
      "grad_norm": 0.7244435414997008,
      "learning_rate": 0.00017883886250121133,
      "loss": 0.7539,
      "step": 6713
    },
    {
      "epoch": 3.159529411764706,
      "grad_norm": 0.7603695600453277,
      "learning_rate": 0.00017879584289258048,
      "loss": 0.7305,
      "step": 6714
    },
    {
      "epoch": 3.16,
      "grad_norm": 0.3782995966336427,
      "learning_rate": 0.00017875282082464343,
      "loss": 0.3262,
      "step": 6715
    },
    {
      "epoch": 3.160470588235294,
      "grad_norm": 1.1534664656336666,
      "learning_rate": 0.00017870979630107446,
      "loss": 0.9805,
      "step": 6716
    },
    {
      "epoch": 3.160941176470588,
      "grad_norm": 0.26163214136950697,
      "learning_rate": 0.00017866676932554802,
      "loss": 0.1338,
      "step": 6717
    },
    {
      "epoch": 3.1614117647058824,
      "grad_norm": 0.5885318971862532,
      "learning_rate": 0.0001786237399017389,
      "loss": 0.4668,
      "step": 6718
    },
    {
      "epoch": 3.1618823529411766,
      "grad_norm": 0.3805696814323528,
      "learning_rate": 0.000178580708033322,
      "loss": 0.3613,
      "step": 6719
    },
    {
      "epoch": 3.1623529411764704,
      "grad_norm": 0.38608352466762275,
      "learning_rate": 0.00017853767372397243,
      "loss": 0.3359,
      "step": 6720
    },
    {
      "epoch": 3.1628235294117646,
      "grad_norm": 0.4342282286772452,
      "learning_rate": 0.00017849463697736554,
      "loss": 0.1309,
      "step": 6721
    },
    {
      "epoch": 3.163294117647059,
      "grad_norm": 0.7437657893556601,
      "learning_rate": 0.0001784515977971769,
      "loss": 0.8828,
      "step": 6722
    },
    {
      "epoch": 3.163764705882353,
      "grad_norm": 1.1466525974089348,
      "learning_rate": 0.00017840855618708227,
      "loss": 0.4492,
      "step": 6723
    },
    {
      "epoch": 3.1642352941176473,
      "grad_norm": 0.38626164009996317,
      "learning_rate": 0.00017836551215075758,
      "loss": 0.2637,
      "step": 6724
    },
    {
      "epoch": 3.164705882352941,
      "grad_norm": 0.3478035188098691,
      "learning_rate": 0.000178322465691879,
      "loss": 0.1216,
      "step": 6725
    },
    {
      "epoch": 3.1651764705882353,
      "grad_norm": 0.5704547508722358,
      "learning_rate": 0.00017827941681412298,
      "loss": 0.4453,
      "step": 6726
    },
    {
      "epoch": 3.1656470588235295,
      "grad_norm": 0.9383161034306192,
      "learning_rate": 0.00017823636552116603,
      "loss": 0.8242,
      "step": 6727
    },
    {
      "epoch": 3.1661176470588237,
      "grad_norm": 0.3190097788818233,
      "learning_rate": 0.00017819331181668496,
      "loss": 0.0835,
      "step": 6728
    },
    {
      "epoch": 3.1665882352941175,
      "grad_norm": 0.3621777665162696,
      "learning_rate": 0.00017815025570435677,
      "loss": 0.1582,
      "step": 6729
    },
    {
      "epoch": 3.1670588235294117,
      "grad_norm": 0.5452536641683926,
      "learning_rate": 0.0001781071971878587,
      "loss": 0.2373,
      "step": 6730
    },
    {
      "epoch": 3.167529411764706,
      "grad_norm": 0.4825065670172693,
      "learning_rate": 0.00017806413627086807,
      "loss": 0.4883,
      "step": 6731
    },
    {
      "epoch": 3.168,
      "grad_norm": 0.35012668102127487,
      "learning_rate": 0.0001780210729570626,
      "loss": 0.1846,
      "step": 6732
    },
    {
      "epoch": 3.168470588235294,
      "grad_norm": 0.6182329619214402,
      "learning_rate": 0.00017797800725012002,
      "loss": 0.4727,
      "step": 6733
    },
    {
      "epoch": 3.168941176470588,
      "grad_norm": 0.6981876243989316,
      "learning_rate": 0.00017793493915371842,
      "loss": 0.7812,
      "step": 6734
    },
    {
      "epoch": 3.1694117647058824,
      "grad_norm": 0.36543310370565313,
      "learning_rate": 0.00017789186867153594,
      "loss": 0.2402,
      "step": 6735
    },
    {
      "epoch": 3.1698823529411766,
      "grad_norm": 0.55050788033804,
      "learning_rate": 0.00017784879580725113,
      "loss": 0.2471,
      "step": 6736
    },
    {
      "epoch": 3.1703529411764704,
      "grad_norm": 0.48493977098962865,
      "learning_rate": 0.00017780572056454256,
      "loss": 0.3535,
      "step": 6737
    },
    {
      "epoch": 3.1708235294117646,
      "grad_norm": 0.4293667927192317,
      "learning_rate": 0.00017776264294708905,
      "loss": 0.3203,
      "step": 6738
    },
    {
      "epoch": 3.171294117647059,
      "grad_norm": 1.468278195656265,
      "learning_rate": 0.00017771956295856964,
      "loss": 0.1543,
      "step": 6739
    },
    {
      "epoch": 3.171764705882353,
      "grad_norm": 0.5625005391754008,
      "learning_rate": 0.0001776764806026636,
      "loss": 0.5117,
      "step": 6740
    },
    {
      "epoch": 3.1722352941176473,
      "grad_norm": 0.7705069134769388,
      "learning_rate": 0.00017763339588305037,
      "loss": 0.7891,
      "step": 6741
    },
    {
      "epoch": 3.172705882352941,
      "grad_norm": 0.3724801911083383,
      "learning_rate": 0.00017759030880340956,
      "loss": 0.0967,
      "step": 6742
    },
    {
      "epoch": 3.1731764705882353,
      "grad_norm": 0.466083161150135,
      "learning_rate": 0.00017754721936742108,
      "loss": 0.2598,
      "step": 6743
    },
    {
      "epoch": 3.1736470588235295,
      "grad_norm": 0.44024937042484485,
      "learning_rate": 0.00017750412757876492,
      "loss": 0.2168,
      "step": 6744
    },
    {
      "epoch": 3.1741176470588237,
      "grad_norm": 0.486163510780951,
      "learning_rate": 0.00017746103344112136,
      "loss": 0.3379,
      "step": 6745
    },
    {
      "epoch": 3.1745882352941175,
      "grad_norm": 0.3416733743601887,
      "learning_rate": 0.00017741793695817093,
      "loss": 0.1895,
      "step": 6746
    },
    {
      "epoch": 3.1750588235294117,
      "grad_norm": 0.3734864898338951,
      "learning_rate": 0.00017737483813359413,
      "loss": 0.1377,
      "step": 6747
    },
    {
      "epoch": 3.175529411764706,
      "grad_norm": 0.5190709734610356,
      "learning_rate": 0.00017733173697107187,
      "loss": 0.4375,
      "step": 6748
    },
    {
      "epoch": 3.176,
      "grad_norm": 2.785291547873798,
      "learning_rate": 0.00017728863347428524,
      "loss": 0.7578,
      "step": 6749
    },
    {
      "epoch": 3.176470588235294,
      "grad_norm": 0.5217017138205869,
      "learning_rate": 0.00017724552764691545,
      "loss": 0.5352,
      "step": 6750
    },
    {
      "epoch": 3.176941176470588,
      "grad_norm": 0.4587564947524449,
      "learning_rate": 0.000177202419492644,
      "loss": 0.1846,
      "step": 6751
    },
    {
      "epoch": 3.1774117647058824,
      "grad_norm": 0.4886889490171421,
      "learning_rate": 0.00017715930901515242,
      "loss": 0.5703,
      "step": 6752
    },
    {
      "epoch": 3.1778823529411766,
      "grad_norm": 0.24511797708343883,
      "learning_rate": 0.0001771161962181227,
      "loss": 0.1494,
      "step": 6753
    },
    {
      "epoch": 3.1783529411764704,
      "grad_norm": 0.41799552408602403,
      "learning_rate": 0.0001770730811052368,
      "loss": 0.4238,
      "step": 6754
    },
    {
      "epoch": 3.1788235294117646,
      "grad_norm": 0.2924451442684325,
      "learning_rate": 0.00017702996368017693,
      "loss": 0.1484,
      "step": 6755
    },
    {
      "epoch": 3.179294117647059,
      "grad_norm": 0.23232070841026994,
      "learning_rate": 0.00017698684394662563,
      "loss": 0.1855,
      "step": 6756
    },
    {
      "epoch": 3.179764705882353,
      "grad_norm": 0.6827064604355736,
      "learning_rate": 0.0001769437219082655,
      "loss": 0.543,
      "step": 6757
    },
    {
      "epoch": 3.1802352941176473,
      "grad_norm": 0.5663174424925341,
      "learning_rate": 0.00017690059756877933,
      "loss": 0.7617,
      "step": 6758
    },
    {
      "epoch": 3.180705882352941,
      "grad_norm": 0.4837601026472863,
      "learning_rate": 0.00017685747093185017,
      "loss": 0.3848,
      "step": 6759
    },
    {
      "epoch": 3.1811764705882353,
      "grad_norm": 0.6145635123396216,
      "learning_rate": 0.00017681434200116132,
      "loss": 0.7148,
      "step": 6760
    },
    {
      "epoch": 3.1816470588235295,
      "grad_norm": 0.39353744729275025,
      "learning_rate": 0.0001767712107803961,
      "loss": 0.4531,
      "step": 6761
    },
    {
      "epoch": 3.1821176470588237,
      "grad_norm": 0.451565202691096,
      "learning_rate": 0.00017672807727323818,
      "loss": 0.3516,
      "step": 6762
    },
    {
      "epoch": 3.1825882352941175,
      "grad_norm": 0.3928520839263105,
      "learning_rate": 0.00017668494148337133,
      "loss": 0.4102,
      "step": 6763
    },
    {
      "epoch": 3.1830588235294117,
      "grad_norm": 0.35805651857595167,
      "learning_rate": 0.00017664180341447964,
      "loss": 0.1299,
      "step": 6764
    },
    {
      "epoch": 3.183529411764706,
      "grad_norm": 0.5566072529907996,
      "learning_rate": 0.0001765986630702472,
      "loss": 0.208,
      "step": 6765
    },
    {
      "epoch": 3.184,
      "grad_norm": 0.3400245025547993,
      "learning_rate": 0.00017655552045435846,
      "loss": 0.3398,
      "step": 6766
    },
    {
      "epoch": 3.184470588235294,
      "grad_norm": 0.531249725443295,
      "learning_rate": 0.00017651237557049808,
      "loss": 0.5508,
      "step": 6767
    },
    {
      "epoch": 3.184941176470588,
      "grad_norm": 0.30943873358071783,
      "learning_rate": 0.00017646922842235076,
      "loss": 0.1167,
      "step": 6768
    },
    {
      "epoch": 3.1854117647058824,
      "grad_norm": 0.48439766301150305,
      "learning_rate": 0.00017642607901360147,
      "loss": 0.543,
      "step": 6769
    },
    {
      "epoch": 3.1858823529411766,
      "grad_norm": 0.40660407803009496,
      "learning_rate": 0.00017638292734793547,
      "loss": 0.2617,
      "step": 6770
    },
    {
      "epoch": 3.1863529411764704,
      "grad_norm": 0.48012750882931027,
      "learning_rate": 0.00017633977342903807,
      "loss": 0.375,
      "step": 6771
    },
    {
      "epoch": 3.1868235294117646,
      "grad_norm": 0.36533036121860574,
      "learning_rate": 0.0001762966172605948,
      "loss": 0.208,
      "step": 6772
    },
    {
      "epoch": 3.187294117647059,
      "grad_norm": 0.32648156210387436,
      "learning_rate": 0.00017625345884629141,
      "loss": 0.0972,
      "step": 6773
    },
    {
      "epoch": 3.187764705882353,
      "grad_norm": 0.459345117280975,
      "learning_rate": 0.00017621029818981387,
      "loss": 0.334,
      "step": 6774
    },
    {
      "epoch": 3.1882352941176473,
      "grad_norm": 0.48912714364308446,
      "learning_rate": 0.00017616713529484835,
      "loss": 0.6484,
      "step": 6775
    },
    {
      "epoch": 3.188705882352941,
      "grad_norm": 0.5169343413478008,
      "learning_rate": 0.00017612397016508104,
      "loss": 0.4902,
      "step": 6776
    },
    {
      "epoch": 3.1891764705882353,
      "grad_norm": 0.4783327042614918,
      "learning_rate": 0.0001760808028041986,
      "loss": 0.6367,
      "step": 6777
    },
    {
      "epoch": 3.1896470588235295,
      "grad_norm": 0.6764767601363868,
      "learning_rate": 0.0001760376332158877,
      "loss": 0.5195,
      "step": 6778
    },
    {
      "epoch": 3.1901176470588237,
      "grad_norm": 0.41290313195183587,
      "learning_rate": 0.00017599446140383512,
      "loss": 0.1846,
      "step": 6779
    },
    {
      "epoch": 3.1905882352941175,
      "grad_norm": 0.43918242743629804,
      "learning_rate": 0.00017595128737172812,
      "loss": 0.3027,
      "step": 6780
    },
    {
      "epoch": 3.1910588235294117,
      "grad_norm": 0.4552851048113956,
      "learning_rate": 0.00017590811112325384,
      "loss": 0.3438,
      "step": 6781
    },
    {
      "epoch": 3.191529411764706,
      "grad_norm": 0.46464463409950624,
      "learning_rate": 0.00017586493266209978,
      "loss": 0.5469,
      "step": 6782
    },
    {
      "epoch": 3.192,
      "grad_norm": 0.31946767495597833,
      "learning_rate": 0.00017582175199195365,
      "loss": 0.1328,
      "step": 6783
    },
    {
      "epoch": 3.192470588235294,
      "grad_norm": 0.31592292868288774,
      "learning_rate": 0.00017577856911650324,
      "loss": 0.1191,
      "step": 6784
    },
    {
      "epoch": 3.192941176470588,
      "grad_norm": 0.5600004092835918,
      "learning_rate": 0.00017573538403943656,
      "loss": 0.6641,
      "step": 6785
    },
    {
      "epoch": 3.1934117647058824,
      "grad_norm": 0.39939362432409775,
      "learning_rate": 0.00017569219676444187,
      "loss": 0.209,
      "step": 6786
    },
    {
      "epoch": 3.1938823529411766,
      "grad_norm": 0.23106221825030562,
      "learning_rate": 0.00017564900729520751,
      "loss": 0.0957,
      "step": 6787
    },
    {
      "epoch": 3.1943529411764704,
      "grad_norm": 0.5050654616638989,
      "learning_rate": 0.0001756058156354222,
      "loss": 0.4785,
      "step": 6788
    },
    {
      "epoch": 3.1948235294117646,
      "grad_norm": 0.6477941957558375,
      "learning_rate": 0.00017556262178877459,
      "loss": 0.5703,
      "step": 6789
    },
    {
      "epoch": 3.195294117647059,
      "grad_norm": 0.43992892710182596,
      "learning_rate": 0.00017551942575895368,
      "loss": 0.3359,
      "step": 6790
    },
    {
      "epoch": 3.195764705882353,
      "grad_norm": 0.6644808435893274,
      "learning_rate": 0.00017547622754964868,
      "loss": 0.7773,
      "step": 6791
    },
    {
      "epoch": 3.1962352941176473,
      "grad_norm": 0.5409131718643551,
      "learning_rate": 0.0001754330271645489,
      "loss": 0.3242,
      "step": 6792
    },
    {
      "epoch": 3.196705882352941,
      "grad_norm": 0.43623837795770215,
      "learning_rate": 0.00017538982460734375,
      "loss": 0.5234,
      "step": 6793
    },
    {
      "epoch": 3.1971764705882353,
      "grad_norm": 0.2744444822892545,
      "learning_rate": 0.00017534661988172315,
      "loss": 0.1445,
      "step": 6794
    },
    {
      "epoch": 3.1976470588235295,
      "grad_norm": 1.0471230068058288,
      "learning_rate": 0.00017530341299137685,
      "loss": 1.5391,
      "step": 6795
    },
    {
      "epoch": 3.1981176470588237,
      "grad_norm": 0.38161992819591917,
      "learning_rate": 0.00017526020393999496,
      "loss": 0.3789,
      "step": 6796
    },
    {
      "epoch": 3.1985882352941175,
      "grad_norm": 0.35718855678723044,
      "learning_rate": 0.00017521699273126778,
      "loss": 0.1299,
      "step": 6797
    },
    {
      "epoch": 3.1990588235294117,
      "grad_norm": 0.34402584045571394,
      "learning_rate": 0.00017517377936888568,
      "loss": 0.1621,
      "step": 6798
    },
    {
      "epoch": 3.199529411764706,
      "grad_norm": 0.35097776978951456,
      "learning_rate": 0.0001751305638565394,
      "loss": 0.2314,
      "step": 6799
    },
    {
      "epoch": 3.2,
      "grad_norm": 0.4387464855116542,
      "learning_rate": 0.00017508734619791966,
      "loss": 0.2578,
      "step": 6800
    },
    {
      "epoch": 3.200470588235294,
      "grad_norm": 0.27302780594002024,
      "learning_rate": 0.00017504412639671754,
      "loss": 0.21,
      "step": 6801
    },
    {
      "epoch": 3.200941176470588,
      "grad_norm": 0.3766486149689097,
      "learning_rate": 0.00017500090445662415,
      "loss": 0.293,
      "step": 6802
    },
    {
      "epoch": 3.2014117647058824,
      "grad_norm": 0.49203627865870087,
      "learning_rate": 0.0001749576803813309,
      "loss": 0.5469,
      "step": 6803
    },
    {
      "epoch": 3.2018823529411766,
      "grad_norm": 0.4782467827507355,
      "learning_rate": 0.00017491445417452934,
      "loss": 0.5117,
      "step": 6804
    },
    {
      "epoch": 3.2023529411764704,
      "grad_norm": 0.22693708713072,
      "learning_rate": 0.00017487122583991123,
      "loss": 0.1797,
      "step": 6805
    },
    {
      "epoch": 3.2028235294117646,
      "grad_norm": 0.37580314478516874,
      "learning_rate": 0.00017482799538116834,
      "loss": 0.1465,
      "step": 6806
    },
    {
      "epoch": 3.203294117647059,
      "grad_norm": 0.37932156993283256,
      "learning_rate": 0.00017478476280199294,
      "loss": 0.3633,
      "step": 6807
    },
    {
      "epoch": 3.203764705882353,
      "grad_norm": 0.3673266188483124,
      "learning_rate": 0.0001747415281060772,
      "loss": 0.25,
      "step": 6808
    },
    {
      "epoch": 3.2042352941176473,
      "grad_norm": 0.49348387544843014,
      "learning_rate": 0.00017469829129711361,
      "loss": 0.4355,
      "step": 6809
    },
    {
      "epoch": 3.204705882352941,
      "grad_norm": 0.6421564615999262,
      "learning_rate": 0.00017465505237879476,
      "loss": 0.4375,
      "step": 6810
    },
    {
      "epoch": 3.2051764705882353,
      "grad_norm": 0.3135325540226971,
      "learning_rate": 0.00017461181135481355,
      "loss": 0.127,
      "step": 6811
    },
    {
      "epoch": 3.2056470588235295,
      "grad_norm": 0.6222106118152041,
      "learning_rate": 0.00017456856822886297,
      "loss": 0.5781,
      "step": 6812
    },
    {
      "epoch": 3.2061176470588233,
      "grad_norm": 0.5299192420865345,
      "learning_rate": 0.00017452532300463609,
      "loss": 0.7734,
      "step": 6813
    },
    {
      "epoch": 3.2065882352941175,
      "grad_norm": 0.27464808878098707,
      "learning_rate": 0.00017448207568582634,
      "loss": 0.085,
      "step": 6814
    },
    {
      "epoch": 3.2070588235294117,
      "grad_norm": 0.44846469438298664,
      "learning_rate": 0.00017443882627612726,
      "loss": 0.4688,
      "step": 6815
    },
    {
      "epoch": 3.207529411764706,
      "grad_norm": 0.45203587594346895,
      "learning_rate": 0.00017439557477923255,
      "loss": 0.3008,
      "step": 6816
    },
    {
      "epoch": 3.208,
      "grad_norm": 0.2202339202238719,
      "learning_rate": 0.00017435232119883604,
      "loss": 0.053,
      "step": 6817
    },
    {
      "epoch": 3.208470588235294,
      "grad_norm": 0.37935557963044864,
      "learning_rate": 0.0001743090655386319,
      "loss": 0.2891,
      "step": 6818
    },
    {
      "epoch": 3.208941176470588,
      "grad_norm": 0.46422152167458836,
      "learning_rate": 0.00017426580780231435,
      "loss": 0.25,
      "step": 6819
    },
    {
      "epoch": 3.2094117647058824,
      "grad_norm": 0.28643547861165924,
      "learning_rate": 0.0001742225479935777,
      "loss": 0.1152,
      "step": 6820
    },
    {
      "epoch": 3.2098823529411766,
      "grad_norm": 0.39049398317795686,
      "learning_rate": 0.00017417928611611675,
      "loss": 0.3906,
      "step": 6821
    },
    {
      "epoch": 3.2103529411764704,
      "grad_norm": 0.5096277152687365,
      "learning_rate": 0.0001741360221736261,
      "loss": 0.5,
      "step": 6822
    },
    {
      "epoch": 3.2108235294117646,
      "grad_norm": 0.3506364649433773,
      "learning_rate": 0.0001740927561698008,
      "loss": 0.2051,
      "step": 6823
    },
    {
      "epoch": 3.211294117647059,
      "grad_norm": 0.48733118441200607,
      "learning_rate": 0.00017404948810833594,
      "loss": 0.3398,
      "step": 6824
    },
    {
      "epoch": 3.211764705882353,
      "grad_norm": 0.6526618515424497,
      "learning_rate": 0.0001740062179929268,
      "loss": 0.2715,
      "step": 6825
    },
    {
      "epoch": 3.212235294117647,
      "grad_norm": 0.5233258717397213,
      "learning_rate": 0.00017396294582726896,
      "loss": 0.4375,
      "step": 6826
    },
    {
      "epoch": 3.212705882352941,
      "grad_norm": 0.788416917602976,
      "learning_rate": 0.00017391967161505795,
      "loss": 0.5938,
      "step": 6827
    },
    {
      "epoch": 3.2131764705882353,
      "grad_norm": 0.28871158577920564,
      "learning_rate": 0.00017387639535998967,
      "loss": 0.2461,
      "step": 6828
    },
    {
      "epoch": 3.2136470588235295,
      "grad_norm": 0.2876921089285478,
      "learning_rate": 0.00017383311706576016,
      "loss": 0.2217,
      "step": 6829
    },
    {
      "epoch": 3.2141176470588233,
      "grad_norm": 0.617563857026378,
      "learning_rate": 0.00017378983673606548,
      "loss": 0.8203,
      "step": 6830
    },
    {
      "epoch": 3.2145882352941175,
      "grad_norm": 0.4959879846184653,
      "learning_rate": 0.00017374655437460206,
      "loss": 0.3496,
      "step": 6831
    },
    {
      "epoch": 3.2150588235294117,
      "grad_norm": 0.6305992196204733,
      "learning_rate": 0.00017370326998506646,
      "loss": 0.5586,
      "step": 6832
    },
    {
      "epoch": 3.215529411764706,
      "grad_norm": 0.30621844542282417,
      "learning_rate": 0.00017365998357115535,
      "loss": 0.104,
      "step": 6833
    },
    {
      "epoch": 3.216,
      "grad_norm": 0.3136225886105274,
      "learning_rate": 0.00017361669513656558,
      "loss": 0.167,
      "step": 6834
    },
    {
      "epoch": 3.216470588235294,
      "grad_norm": 0.6851203639075936,
      "learning_rate": 0.0001735734046849942,
      "loss": 0.3633,
      "step": 6835
    },
    {
      "epoch": 3.216941176470588,
      "grad_norm": 0.4541345047985539,
      "learning_rate": 0.0001735301122201384,
      "loss": 0.334,
      "step": 6836
    },
    {
      "epoch": 3.2174117647058824,
      "grad_norm": 0.4804719295715254,
      "learning_rate": 0.00017348681774569563,
      "loss": 0.5859,
      "step": 6837
    },
    {
      "epoch": 3.2178823529411766,
      "grad_norm": 0.4538129151432295,
      "learning_rate": 0.00017344352126536343,
      "loss": 0.2012,
      "step": 6838
    },
    {
      "epoch": 3.2183529411764704,
      "grad_norm": 0.6775502565637787,
      "learning_rate": 0.00017340022278283953,
      "loss": 0.6445,
      "step": 6839
    },
    {
      "epoch": 3.2188235294117646,
      "grad_norm": 0.4687473373351891,
      "learning_rate": 0.00017335692230182183,
      "loss": 0.2578,
      "step": 6840
    },
    {
      "epoch": 3.219294117647059,
      "grad_norm": 0.4521599807551494,
      "learning_rate": 0.00017331361982600836,
      "loss": 0.4121,
      "step": 6841
    },
    {
      "epoch": 3.219764705882353,
      "grad_norm": 0.608182126272122,
      "learning_rate": 0.00017327031535909745,
      "loss": 0.4316,
      "step": 6842
    },
    {
      "epoch": 3.220235294117647,
      "grad_norm": 0.6135674264301552,
      "learning_rate": 0.00017322700890478744,
      "loss": 0.668,
      "step": 6843
    },
    {
      "epoch": 3.220705882352941,
      "grad_norm": 0.41982347956085814,
      "learning_rate": 0.00017318370046677693,
      "loss": 0.5898,
      "step": 6844
    },
    {
      "epoch": 3.2211764705882353,
      "grad_norm": 0.6582008122819268,
      "learning_rate": 0.00017314039004876473,
      "loss": 0.5195,
      "step": 6845
    },
    {
      "epoch": 3.2216470588235295,
      "grad_norm": 0.3791525969951085,
      "learning_rate": 0.0001730970776544497,
      "loss": 0.1875,
      "step": 6846
    },
    {
      "epoch": 3.2221176470588233,
      "grad_norm": 0.5862509279550906,
      "learning_rate": 0.00017305376328753092,
      "loss": 0.3418,
      "step": 6847
    },
    {
      "epoch": 3.2225882352941175,
      "grad_norm": 0.4566942810628375,
      "learning_rate": 0.00017301044695170769,
      "loss": 0.4785,
      "step": 6848
    },
    {
      "epoch": 3.2230588235294118,
      "grad_norm": 0.2669321234493054,
      "learning_rate": 0.00017296712865067944,
      "loss": 0.1147,
      "step": 6849
    },
    {
      "epoch": 3.223529411764706,
      "grad_norm": 0.538311787847182,
      "learning_rate": 0.00017292380838814577,
      "loss": 0.3906,
      "step": 6850
    },
    {
      "epoch": 3.224,
      "grad_norm": 0.4985284616623147,
      "learning_rate": 0.00017288048616780636,
      "loss": 0.5547,
      "step": 6851
    },
    {
      "epoch": 3.224470588235294,
      "grad_norm": 0.4822397999785073,
      "learning_rate": 0.00017283716199336123,
      "loss": 0.3398,
      "step": 6852
    },
    {
      "epoch": 3.224941176470588,
      "grad_norm": 0.5329637061700669,
      "learning_rate": 0.0001727938358685105,
      "loss": 0.4922,
      "step": 6853
    },
    {
      "epoch": 3.2254117647058824,
      "grad_norm": 0.5706923107337121,
      "learning_rate": 0.0001727505077969543,
      "loss": 0.8359,
      "step": 6854
    },
    {
      "epoch": 3.2258823529411766,
      "grad_norm": 0.6821968675412869,
      "learning_rate": 0.00017270717778239316,
      "loss": 1.0703,
      "step": 6855
    },
    {
      "epoch": 3.2263529411764704,
      "grad_norm": 0.43520375866622485,
      "learning_rate": 0.0001726638458285277,
      "loss": 0.6836,
      "step": 6856
    },
    {
      "epoch": 3.2268235294117646,
      "grad_norm": 0.415417150899474,
      "learning_rate": 0.00017262051193905865,
      "loss": 0.1621,
      "step": 6857
    },
    {
      "epoch": 3.227294117647059,
      "grad_norm": 0.4182763988003298,
      "learning_rate": 0.00017257717611768685,
      "loss": 0.5234,
      "step": 6858
    },
    {
      "epoch": 3.227764705882353,
      "grad_norm": 0.6045052481410694,
      "learning_rate": 0.00017253383836811356,
      "loss": 0.4609,
      "step": 6859
    },
    {
      "epoch": 3.228235294117647,
      "grad_norm": 0.4851722496776122,
      "learning_rate": 0.0001724904986940399,
      "loss": 0.457,
      "step": 6860
    },
    {
      "epoch": 3.228705882352941,
      "grad_norm": 0.380221610785792,
      "learning_rate": 0.00017244715709916733,
      "loss": 0.2031,
      "step": 6861
    },
    {
      "epoch": 3.2291764705882353,
      "grad_norm": 0.3240077636764891,
      "learning_rate": 0.00017240381358719746,
      "loss": 0.3008,
      "step": 6862
    },
    {
      "epoch": 3.2296470588235295,
      "grad_norm": 0.24283403445406573,
      "learning_rate": 0.000172360468161832,
      "loss": 0.1426,
      "step": 6863
    },
    {
      "epoch": 3.2301176470588233,
      "grad_norm": 0.5472040678382756,
      "learning_rate": 0.00017231712082677291,
      "loss": 0.4316,
      "step": 6864
    },
    {
      "epoch": 3.2305882352941175,
      "grad_norm": 0.5474238902069883,
      "learning_rate": 0.0001722737715857222,
      "loss": 0.5039,
      "step": 6865
    },
    {
      "epoch": 3.2310588235294118,
      "grad_norm": 0.28567366125906923,
      "learning_rate": 0.00017223042044238219,
      "loss": 0.1797,
      "step": 6866
    },
    {
      "epoch": 3.231529411764706,
      "grad_norm": 0.5566926936403978,
      "learning_rate": 0.0001721870674004552,
      "loss": 0.334,
      "step": 6867
    },
    {
      "epoch": 3.232,
      "grad_norm": 0.6087689524809592,
      "learning_rate": 0.00017214371246364384,
      "loss": 0.4082,
      "step": 6868
    },
    {
      "epoch": 3.232470588235294,
      "grad_norm": 0.3777066853130039,
      "learning_rate": 0.00017210035563565082,
      "loss": 0.1406,
      "step": 6869
    },
    {
      "epoch": 3.232941176470588,
      "grad_norm": 0.21568953885235853,
      "learning_rate": 0.00017205699692017902,
      "loss": 0.085,
      "step": 6870
    },
    {
      "epoch": 3.2334117647058824,
      "grad_norm": 0.4772704468949652,
      "learning_rate": 0.0001720136363209315,
      "loss": 0.2207,
      "step": 6871
    },
    {
      "epoch": 3.2338823529411767,
      "grad_norm": 0.564479351983983,
      "learning_rate": 0.0001719702738416115,
      "loss": 0.4141,
      "step": 6872
    },
    {
      "epoch": 3.2343529411764704,
      "grad_norm": 0.5040897936768829,
      "learning_rate": 0.0001719269094859223,
      "loss": 0.5742,
      "step": 6873
    },
    {
      "epoch": 3.2348235294117647,
      "grad_norm": 0.6594847452751028,
      "learning_rate": 0.0001718835432575675,
      "loss": 0.2969,
      "step": 6874
    },
    {
      "epoch": 3.235294117647059,
      "grad_norm": 0.5483559025234184,
      "learning_rate": 0.00017184017516025076,
      "loss": 0.3379,
      "step": 6875
    },
    {
      "epoch": 3.235764705882353,
      "grad_norm": 0.3624472144943908,
      "learning_rate": 0.0001717968051976759,
      "loss": 0.1895,
      "step": 6876
    },
    {
      "epoch": 3.236235294117647,
      "grad_norm": 0.204323654647677,
      "learning_rate": 0.00017175343337354698,
      "loss": 0.1816,
      "step": 6877
    },
    {
      "epoch": 3.236705882352941,
      "grad_norm": 0.5474637867366029,
      "learning_rate": 0.00017171005969156817,
      "loss": 0.5391,
      "step": 6878
    },
    {
      "epoch": 3.2371764705882353,
      "grad_norm": 0.41990470831102467,
      "learning_rate": 0.00017166668415544376,
      "loss": 0.2969,
      "step": 6879
    },
    {
      "epoch": 3.2376470588235295,
      "grad_norm": 0.6005678288570269,
      "learning_rate": 0.0001716233067688783,
      "loss": 0.6133,
      "step": 6880
    },
    {
      "epoch": 3.2381176470588233,
      "grad_norm": 0.2661084247406313,
      "learning_rate": 0.00017157992753557633,
      "loss": 0.0991,
      "step": 6881
    },
    {
      "epoch": 3.2385882352941175,
      "grad_norm": 0.24671458576665728,
      "learning_rate": 0.0001715365464592427,
      "loss": 0.2715,
      "step": 6882
    },
    {
      "epoch": 3.2390588235294118,
      "grad_norm": 0.3501128487023379,
      "learning_rate": 0.00017149316354358237,
      "loss": 0.2832,
      "step": 6883
    },
    {
      "epoch": 3.239529411764706,
      "grad_norm": 0.4698646980677021,
      "learning_rate": 0.00017144977879230046,
      "loss": 0.5234,
      "step": 6884
    },
    {
      "epoch": 3.24,
      "grad_norm": 0.24762660724630708,
      "learning_rate": 0.00017140639220910222,
      "loss": 0.2656,
      "step": 6885
    },
    {
      "epoch": 3.240470588235294,
      "grad_norm": 0.44349822219432994,
      "learning_rate": 0.00017136300379769312,
      "loss": 0.4258,
      "step": 6886
    },
    {
      "epoch": 3.240941176470588,
      "grad_norm": 0.43266645321970537,
      "learning_rate": 0.00017131961356177869,
      "loss": 0.252,
      "step": 6887
    },
    {
      "epoch": 3.2414117647058824,
      "grad_norm": 0.6769940736090722,
      "learning_rate": 0.00017127622150506472,
      "loss": 0.7578,
      "step": 6888
    },
    {
      "epoch": 3.2418823529411767,
      "grad_norm": 0.409925503368348,
      "learning_rate": 0.000171232827631257,
      "loss": 0.25,
      "step": 6889
    },
    {
      "epoch": 3.2423529411764704,
      "grad_norm": 0.41705514720150694,
      "learning_rate": 0.00017118943194406169,
      "loss": 0.3262,
      "step": 6890
    },
    {
      "epoch": 3.2428235294117647,
      "grad_norm": 0.5178546396994328,
      "learning_rate": 0.00017114603444718496,
      "loss": 0.5664,
      "step": 6891
    },
    {
      "epoch": 3.243294117647059,
      "grad_norm": 0.4212098669353555,
      "learning_rate": 0.00017110263514433317,
      "loss": 0.1748,
      "step": 6892
    },
    {
      "epoch": 3.243764705882353,
      "grad_norm": 0.5053757834770506,
      "learning_rate": 0.00017105923403921283,
      "loss": 0.2539,
      "step": 6893
    },
    {
      "epoch": 3.244235294117647,
      "grad_norm": 0.374193564279367,
      "learning_rate": 0.00017101583113553063,
      "loss": 0.2539,
      "step": 6894
    },
    {
      "epoch": 3.244705882352941,
      "grad_norm": 0.3909129329019082,
      "learning_rate": 0.00017097242643699333,
      "loss": 0.1699,
      "step": 6895
    },
    {
      "epoch": 3.2451764705882353,
      "grad_norm": 0.2398387051955318,
      "learning_rate": 0.00017092901994730797,
      "loss": 0.1797,
      "step": 6896
    },
    {
      "epoch": 3.2456470588235296,
      "grad_norm": 0.28029241425556806,
      "learning_rate": 0.00017088561167018164,
      "loss": 0.2305,
      "step": 6897
    },
    {
      "epoch": 3.2461176470588233,
      "grad_norm": 0.4907130038244924,
      "learning_rate": 0.0001708422016093216,
      "loss": 0.5898,
      "step": 6898
    },
    {
      "epoch": 3.2465882352941176,
      "grad_norm": 0.7664899682201448,
      "learning_rate": 0.00017079878976843533,
      "loss": 0.7656,
      "step": 6899
    },
    {
      "epoch": 3.2470588235294118,
      "grad_norm": 0.3405019587731749,
      "learning_rate": 0.00017075537615123042,
      "loss": 0.2852,
      "step": 6900
    },
    {
      "epoch": 3.247529411764706,
      "grad_norm": 0.7736903728378353,
      "learning_rate": 0.0001707119607614146,
      "loss": 0.8945,
      "step": 6901
    },
    {
      "epoch": 3.248,
      "grad_norm": 0.5520543765584675,
      "learning_rate": 0.0001706685436026957,
      "loss": 0.4551,
      "step": 6902
    },
    {
      "epoch": 3.248470588235294,
      "grad_norm": 0.4428727505803324,
      "learning_rate": 0.0001706251246787818,
      "loss": 0.3066,
      "step": 6903
    },
    {
      "epoch": 3.248941176470588,
      "grad_norm": 0.44649812659395105,
      "learning_rate": 0.0001705817039933811,
      "loss": 0.5117,
      "step": 6904
    },
    {
      "epoch": 3.2494117647058824,
      "grad_norm": 0.248077436322293,
      "learning_rate": 0.00017053828155020193,
      "loss": 0.1533,
      "step": 6905
    },
    {
      "epoch": 3.2498823529411767,
      "grad_norm": 0.588290095209457,
      "learning_rate": 0.00017049485735295275,
      "loss": 0.6094,
      "step": 6906
    },
    {
      "epoch": 3.2503529411764704,
      "grad_norm": 1.9405269878393532,
      "learning_rate": 0.00017045143140534226,
      "loss": 0.1973,
      "step": 6907
    },
    {
      "epoch": 3.2508235294117647,
      "grad_norm": 0.5543412408039393,
      "learning_rate": 0.0001704080037110792,
      "loss": 0.5,
      "step": 6908
    },
    {
      "epoch": 3.251294117647059,
      "grad_norm": 0.2641806131623773,
      "learning_rate": 0.00017036457427387248,
      "loss": 0.0962,
      "step": 6909
    },
    {
      "epoch": 3.251764705882353,
      "grad_norm": 0.44899874356749175,
      "learning_rate": 0.00017032114309743128,
      "loss": 0.2539,
      "step": 6910
    },
    {
      "epoch": 3.252235294117647,
      "grad_norm": 0.6335950543650805,
      "learning_rate": 0.0001702777101854648,
      "loss": 0.4375,
      "step": 6911
    },
    {
      "epoch": 3.252705882352941,
      "grad_norm": 0.594918657611492,
      "learning_rate": 0.0001702342755416824,
      "loss": 0.5547,
      "step": 6912
    },
    {
      "epoch": 3.2531764705882353,
      "grad_norm": 0.35484945673451723,
      "learning_rate": 0.00017019083916979357,
      "loss": 0.2832,
      "step": 6913
    },
    {
      "epoch": 3.2536470588235296,
      "grad_norm": 0.2800780582254593,
      "learning_rate": 0.0001701474010735081,
      "loss": 0.1001,
      "step": 6914
    },
    {
      "epoch": 3.2541176470588233,
      "grad_norm": 0.43272021989209514,
      "learning_rate": 0.00017010396125653572,
      "loss": 0.5078,
      "step": 6915
    },
    {
      "epoch": 3.2545882352941176,
      "grad_norm": 0.4764831805829617,
      "learning_rate": 0.00017006051972258645,
      "loss": 0.4512,
      "step": 6916
    },
    {
      "epoch": 3.255058823529412,
      "grad_norm": 0.5178133260785143,
      "learning_rate": 0.0001700170764753704,
      "loss": 0.4688,
      "step": 6917
    },
    {
      "epoch": 3.255529411764706,
      "grad_norm": 0.6111032869031086,
      "learning_rate": 0.00016997363151859785,
      "loss": 0.3906,
      "step": 6918
    },
    {
      "epoch": 3.2560000000000002,
      "grad_norm": 0.5134768182352992,
      "learning_rate": 0.00016993018485597915,
      "loss": 0.5781,
      "step": 6919
    },
    {
      "epoch": 3.256470588235294,
      "grad_norm": 0.3277282181365161,
      "learning_rate": 0.00016988673649122492,
      "loss": 0.3203,
      "step": 6920
    },
    {
      "epoch": 3.2569411764705882,
      "grad_norm": 0.5194733786155757,
      "learning_rate": 0.00016984328642804587,
      "loss": 0.6797,
      "step": 6921
    },
    {
      "epoch": 3.2574117647058825,
      "grad_norm": 0.7186140805042182,
      "learning_rate": 0.00016979983467015284,
      "loss": 0.3281,
      "step": 6922
    },
    {
      "epoch": 3.2578823529411762,
      "grad_norm": 0.34527192425074005,
      "learning_rate": 0.00016975638122125678,
      "loss": 0.2109,
      "step": 6923
    },
    {
      "epoch": 3.2583529411764705,
      "grad_norm": 0.1674089397266576,
      "learning_rate": 0.00016971292608506888,
      "loss": 0.0732,
      "step": 6924
    },
    {
      "epoch": 3.2588235294117647,
      "grad_norm": 0.43229279141914007,
      "learning_rate": 0.0001696694692653004,
      "loss": 0.2754,
      "step": 6925
    },
    {
      "epoch": 3.259294117647059,
      "grad_norm": 0.6217450491018549,
      "learning_rate": 0.00016962601076566275,
      "loss": 0.543,
      "step": 6926
    },
    {
      "epoch": 3.259764705882353,
      "grad_norm": 0.3743181120155998,
      "learning_rate": 0.00016958255058986753,
      "loss": 0.2832,
      "step": 6927
    },
    {
      "epoch": 3.260235294117647,
      "grad_norm": 0.4612390033168644,
      "learning_rate": 0.00016953908874162645,
      "loss": 0.4492,
      "step": 6928
    },
    {
      "epoch": 3.260705882352941,
      "grad_norm": 0.4205309903655632,
      "learning_rate": 0.00016949562522465135,
      "loss": 0.4492,
      "step": 6929
    },
    {
      "epoch": 3.2611764705882353,
      "grad_norm": 0.45342682671065704,
      "learning_rate": 0.00016945216004265418,
      "loss": 0.4609,
      "step": 6930
    },
    {
      "epoch": 3.2616470588235296,
      "grad_norm": 0.27941538256916254,
      "learning_rate": 0.0001694086931993472,
      "loss": 0.1147,
      "step": 6931
    },
    {
      "epoch": 3.2621176470588233,
      "grad_norm": 0.7356264588845044,
      "learning_rate": 0.0001693652246984426,
      "loss": 0.9844,
      "step": 6932
    },
    {
      "epoch": 3.2625882352941176,
      "grad_norm": 0.23515836586268185,
      "learning_rate": 0.0001693217545436528,
      "loss": 0.0493,
      "step": 6933
    },
    {
      "epoch": 3.263058823529412,
      "grad_norm": 0.4688755899820909,
      "learning_rate": 0.0001692782827386904,
      "loss": 0.4883,
      "step": 6934
    },
    {
      "epoch": 3.263529411764706,
      "grad_norm": 0.417584283093714,
      "learning_rate": 0.00016923480928726813,
      "loss": 0.1963,
      "step": 6935
    },
    {
      "epoch": 3.2640000000000002,
      "grad_norm": 0.5214808361093327,
      "learning_rate": 0.00016919133419309875,
      "loss": 0.4883,
      "step": 6936
    },
    {
      "epoch": 3.264470588235294,
      "grad_norm": 0.45483401123276995,
      "learning_rate": 0.00016914785745989532,
      "loss": 0.3984,
      "step": 6937
    },
    {
      "epoch": 3.2649411764705882,
      "grad_norm": 0.5424366990921182,
      "learning_rate": 0.00016910437909137098,
      "loss": 0.3281,
      "step": 6938
    },
    {
      "epoch": 3.2654117647058825,
      "grad_norm": 0.4400702339328995,
      "learning_rate": 0.00016906089909123892,
      "loss": 0.4805,
      "step": 6939
    },
    {
      "epoch": 3.2658823529411762,
      "grad_norm": 0.43642801603616327,
      "learning_rate": 0.00016901741746321253,
      "loss": 0.5391,
      "step": 6940
    },
    {
      "epoch": 3.2663529411764705,
      "grad_norm": 0.4324212341646542,
      "learning_rate": 0.0001689739342110055,
      "loss": 0.1787,
      "step": 6941
    },
    {
      "epoch": 3.2668235294117647,
      "grad_norm": 0.5632333130588874,
      "learning_rate": 0.00016893044933833135,
      "loss": 0.418,
      "step": 6942
    },
    {
      "epoch": 3.267294117647059,
      "grad_norm": 0.6363882680702216,
      "learning_rate": 0.00016888696284890402,
      "loss": 0.5352,
      "step": 6943
    },
    {
      "epoch": 3.267764705882353,
      "grad_norm": 0.5809636421574336,
      "learning_rate": 0.0001688434747464374,
      "loss": 0.8008,
      "step": 6944
    },
    {
      "epoch": 3.268235294117647,
      "grad_norm": 0.49721907542897303,
      "learning_rate": 0.00016879998503464561,
      "loss": 0.332,
      "step": 6945
    },
    {
      "epoch": 3.268705882352941,
      "grad_norm": 0.3102564109099619,
      "learning_rate": 0.00016875649371724292,
      "loss": 0.2969,
      "step": 6946
    },
    {
      "epoch": 3.2691764705882354,
      "grad_norm": 0.19855103000717647,
      "learning_rate": 0.00016871300079794363,
      "loss": 0.1035,
      "step": 6947
    },
    {
      "epoch": 3.2696470588235296,
      "grad_norm": 0.4966210505187245,
      "learning_rate": 0.0001686695062804623,
      "loss": 0.668,
      "step": 6948
    },
    {
      "epoch": 3.2701176470588234,
      "grad_norm": 0.3358900059070325,
      "learning_rate": 0.00016862601016851358,
      "loss": 0.2314,
      "step": 6949
    },
    {
      "epoch": 3.2705882352941176,
      "grad_norm": 0.48746816215681027,
      "learning_rate": 0.00016858251246581216,
      "loss": 0.6172,
      "step": 6950
    },
    {
      "epoch": 3.271058823529412,
      "grad_norm": 0.3137882222848124,
      "learning_rate": 0.0001685390131760731,
      "loss": 0.2451,
      "step": 6951
    },
    {
      "epoch": 3.271529411764706,
      "grad_norm": 0.29661201477952015,
      "learning_rate": 0.00016849551230301135,
      "loss": 0.1108,
      "step": 6952
    },
    {
      "epoch": 3.2720000000000002,
      "grad_norm": 0.3615113938271359,
      "learning_rate": 0.00016845200985034215,
      "loss": 0.2148,
      "step": 6953
    },
    {
      "epoch": 3.272470588235294,
      "grad_norm": 0.2670158627943289,
      "learning_rate": 0.0001684085058217808,
      "loss": 0.0815,
      "step": 6954
    },
    {
      "epoch": 3.2729411764705882,
      "grad_norm": 0.4538331702744244,
      "learning_rate": 0.00016836500022104274,
      "loss": 0.4609,
      "step": 6955
    },
    {
      "epoch": 3.2734117647058825,
      "grad_norm": 0.6017344655952702,
      "learning_rate": 0.00016832149305184363,
      "loss": 0.6523,
      "step": 6956
    },
    {
      "epoch": 3.2738823529411762,
      "grad_norm": 0.24681687458879767,
      "learning_rate": 0.00016827798431789907,
      "loss": 0.1885,
      "step": 6957
    },
    {
      "epoch": 3.2743529411764705,
      "grad_norm": 0.5260709024221214,
      "learning_rate": 0.00016823447402292504,
      "loss": 0.252,
      "step": 6958
    },
    {
      "epoch": 3.2748235294117647,
      "grad_norm": 0.45689538434909693,
      "learning_rate": 0.00016819096217063752,
      "loss": 0.4434,
      "step": 6959
    },
    {
      "epoch": 3.275294117647059,
      "grad_norm": 0.4442015364757435,
      "learning_rate": 0.00016814744876475252,
      "loss": 0.625,
      "step": 6960
    },
    {
      "epoch": 3.275764705882353,
      "grad_norm": 0.4470133089625983,
      "learning_rate": 0.00016810393380898645,
      "loss": 0.5,
      "step": 6961
    },
    {
      "epoch": 3.276235294117647,
      "grad_norm": 0.5585968161087664,
      "learning_rate": 0.0001680604173070556,
      "loss": 0.7031,
      "step": 6962
    },
    {
      "epoch": 3.276705882352941,
      "grad_norm": 0.33962803890947696,
      "learning_rate": 0.00016801689926267653,
      "loss": 0.4434,
      "step": 6963
    },
    {
      "epoch": 3.2771764705882354,
      "grad_norm": 0.5073928868496597,
      "learning_rate": 0.00016797337967956592,
      "loss": 0.6211,
      "step": 6964
    },
    {
      "epoch": 3.2776470588235296,
      "grad_norm": 0.6394969056969533,
      "learning_rate": 0.00016792985856144049,
      "loss": 0.6719,
      "step": 6965
    },
    {
      "epoch": 3.2781176470588234,
      "grad_norm": 0.3260590211713184,
      "learning_rate": 0.00016788633591201723,
      "loss": 0.3984,
      "step": 6966
    },
    {
      "epoch": 3.2785882352941176,
      "grad_norm": 0.2758219526338519,
      "learning_rate": 0.0001678428117350131,
      "loss": 0.1436,
      "step": 6967
    },
    {
      "epoch": 3.279058823529412,
      "grad_norm": 0.4288902228493082,
      "learning_rate": 0.0001677992860341454,
      "loss": 0.293,
      "step": 6968
    },
    {
      "epoch": 3.279529411764706,
      "grad_norm": 0.44085645804744694,
      "learning_rate": 0.00016775575881313132,
      "loss": 0.4297,
      "step": 6969
    },
    {
      "epoch": 3.2800000000000002,
      "grad_norm": 0.24576211002651946,
      "learning_rate": 0.00016771223007568836,
      "loss": 0.2617,
      "step": 6970
    },
    {
      "epoch": 3.280470588235294,
      "grad_norm": 0.3464203390578683,
      "learning_rate": 0.00016766869982553405,
      "loss": 0.2012,
      "step": 6971
    },
    {
      "epoch": 3.2809411764705882,
      "grad_norm": 0.4058286625607499,
      "learning_rate": 0.0001676251680663861,
      "loss": 0.3398,
      "step": 6972
    },
    {
      "epoch": 3.2814117647058825,
      "grad_norm": 0.27554829910166284,
      "learning_rate": 0.0001675816348019624,
      "loss": 0.106,
      "step": 6973
    },
    {
      "epoch": 3.2818823529411763,
      "grad_norm": 0.5357281053017078,
      "learning_rate": 0.0001675381000359808,
      "loss": 0.543,
      "step": 6974
    },
    {
      "epoch": 3.2823529411764705,
      "grad_norm": 0.24562590245994959,
      "learning_rate": 0.0001674945637721595,
      "loss": 0.0879,
      "step": 6975
    },
    {
      "epoch": 3.2828235294117647,
      "grad_norm": 0.4886371326066231,
      "learning_rate": 0.00016745102601421663,
      "loss": 0.6328,
      "step": 6976
    },
    {
      "epoch": 3.283294117647059,
      "grad_norm": 1.1090851497690912,
      "learning_rate": 0.00016740748676587045,
      "loss": 0.5273,
      "step": 6977
    },
    {
      "epoch": 3.283764705882353,
      "grad_norm": 0.4602188596483834,
      "learning_rate": 0.0001673639460308396,
      "loss": 0.7383,
      "step": 6978
    },
    {
      "epoch": 3.284235294117647,
      "grad_norm": 0.45504270716592865,
      "learning_rate": 0.00016732040381284258,
      "loss": 0.7031,
      "step": 6979
    },
    {
      "epoch": 3.284705882352941,
      "grad_norm": 0.3422707840526675,
      "learning_rate": 0.00016727686011559813,
      "loss": 0.3164,
      "step": 6980
    },
    {
      "epoch": 3.2851764705882354,
      "grad_norm": 0.599122794677649,
      "learning_rate": 0.00016723331494282502,
      "loss": 0.5352,
      "step": 6981
    },
    {
      "epoch": 3.2856470588235296,
      "grad_norm": 0.5791806365611445,
      "learning_rate": 0.00016718976829824234,
      "loss": 0.9297,
      "step": 6982
    },
    {
      "epoch": 3.2861176470588234,
      "grad_norm": 0.4476834493235841,
      "learning_rate": 0.00016714622018556915,
      "loss": 0.3398,
      "step": 6983
    },
    {
      "epoch": 3.2865882352941176,
      "grad_norm": 0.3253190506876197,
      "learning_rate": 0.0001671026706085246,
      "loss": 0.1191,
      "step": 6984
    },
    {
      "epoch": 3.287058823529412,
      "grad_norm": 0.2204614230065123,
      "learning_rate": 0.0001670591195708281,
      "loss": 0.1758,
      "step": 6985
    },
    {
      "epoch": 3.287529411764706,
      "grad_norm": 0.22466727470054795,
      "learning_rate": 0.00016701556707619914,
      "loss": 0.0845,
      "step": 6986
    },
    {
      "epoch": 3.288,
      "grad_norm": 0.29444232841306606,
      "learning_rate": 0.00016697201312835727,
      "loss": 0.127,
      "step": 6987
    },
    {
      "epoch": 3.288470588235294,
      "grad_norm": 0.4067186532026847,
      "learning_rate": 0.00016692845773102222,
      "loss": 0.2148,
      "step": 6988
    },
    {
      "epoch": 3.2889411764705883,
      "grad_norm": 0.4336258651134875,
      "learning_rate": 0.0001668849008879139,
      "loss": 0.2002,
      "step": 6989
    },
    {
      "epoch": 3.2894117647058825,
      "grad_norm": 0.4452786130576474,
      "learning_rate": 0.00016684134260275217,
      "loss": 0.2832,
      "step": 6990
    },
    {
      "epoch": 3.2898823529411763,
      "grad_norm": 0.38549887952471296,
      "learning_rate": 0.0001667977828792572,
      "loss": 0.3867,
      "step": 6991
    },
    {
      "epoch": 3.2903529411764705,
      "grad_norm": 0.5123209784382067,
      "learning_rate": 0.00016675422172114917,
      "loss": 0.4336,
      "step": 6992
    },
    {
      "epoch": 3.2908235294117647,
      "grad_norm": 0.7094910620520212,
      "learning_rate": 0.00016671065913214845,
      "loss": 0.668,
      "step": 6993
    },
    {
      "epoch": 3.291294117647059,
      "grad_norm": 0.4236494187210899,
      "learning_rate": 0.00016666709511597546,
      "loss": 0.1943,
      "step": 6994
    },
    {
      "epoch": 3.291764705882353,
      "grad_norm": 0.4122205132957139,
      "learning_rate": 0.0001666235296763508,
      "loss": 0.4453,
      "step": 6995
    },
    {
      "epoch": 3.292235294117647,
      "grad_norm": 0.32175711231368376,
      "learning_rate": 0.00016657996281699514,
      "loss": 0.3516,
      "step": 6996
    },
    {
      "epoch": 3.292705882352941,
      "grad_norm": 0.40870206074569193,
      "learning_rate": 0.00016653639454162938,
      "loss": 0.3887,
      "step": 6997
    },
    {
      "epoch": 3.2931764705882354,
      "grad_norm": 0.5315359927858426,
      "learning_rate": 0.00016649282485397437,
      "loss": 0.5469,
      "step": 6998
    },
    {
      "epoch": 3.2936470588235296,
      "grad_norm": 0.408446159160695,
      "learning_rate": 0.00016644925375775124,
      "loss": 0.3574,
      "step": 6999
    },
    {
      "epoch": 3.2941176470588234,
      "grad_norm": 0.337226181233945,
      "learning_rate": 0.00016640568125668117,
      "loss": 0.1201,
      "step": 7000
    },
    {
      "epoch": 3.2945882352941176,
      "grad_norm": 0.2881163470930246,
      "learning_rate": 0.00016636210735448539,
      "loss": 0.166,
      "step": 7001
    },
    {
      "epoch": 3.295058823529412,
      "grad_norm": 0.32169913606924877,
      "learning_rate": 0.00016631853205488546,
      "loss": 0.1055,
      "step": 7002
    },
    {
      "epoch": 3.295529411764706,
      "grad_norm": 0.21091690752308853,
      "learning_rate": 0.00016627495536160284,
      "loss": 0.1416,
      "step": 7003
    },
    {
      "epoch": 3.296,
      "grad_norm": 0.47236286493405344,
      "learning_rate": 0.0001662313772783592,
      "loss": 0.4395,
      "step": 7004
    },
    {
      "epoch": 3.296470588235294,
      "grad_norm": 0.6322114290116899,
      "learning_rate": 0.00016618779780887627,
      "loss": 0.6875,
      "step": 7005
    },
    {
      "epoch": 3.2969411764705883,
      "grad_norm": 0.5394718025905954,
      "learning_rate": 0.00016614421695687606,
      "loss": 0.6289,
      "step": 7006
    },
    {
      "epoch": 3.2974117647058825,
      "grad_norm": 0.33718513345387807,
      "learning_rate": 0.00016610063472608053,
      "loss": 0.2891,
      "step": 7007
    },
    {
      "epoch": 3.2978823529411763,
      "grad_norm": 0.5047937480702099,
      "learning_rate": 0.0001660570511202118,
      "loss": 0.6055,
      "step": 7008
    },
    {
      "epoch": 3.2983529411764705,
      "grad_norm": 0.24192512835230606,
      "learning_rate": 0.00016601346614299214,
      "loss": 0.0718,
      "step": 7009
    },
    {
      "epoch": 3.2988235294117647,
      "grad_norm": 0.35680002456499094,
      "learning_rate": 0.00016596987979814395,
      "loss": 0.2363,
      "step": 7010
    },
    {
      "epoch": 3.299294117647059,
      "grad_norm": 0.406688879110809,
      "learning_rate": 0.0001659262920893897,
      "loss": 0.2422,
      "step": 7011
    },
    {
      "epoch": 3.299764705882353,
      "grad_norm": 0.4707836368422136,
      "learning_rate": 0.00016588270302045197,
      "loss": 0.459,
      "step": 7012
    },
    {
      "epoch": 3.300235294117647,
      "grad_norm": 0.5067549060430268,
      "learning_rate": 0.00016583911259505352,
      "loss": 0.249,
      "step": 7013
    },
    {
      "epoch": 3.300705882352941,
      "grad_norm": 1.860089862658878,
      "learning_rate": 0.00016579552081691715,
      "loss": 0.3691,
      "step": 7014
    },
    {
      "epoch": 3.3011764705882354,
      "grad_norm": 0.20882307500825822,
      "learning_rate": 0.0001657519276897658,
      "loss": 0.0649,
      "step": 7015
    },
    {
      "epoch": 3.3016470588235296,
      "grad_norm": 0.5963378793246612,
      "learning_rate": 0.0001657083332173226,
      "loss": 0.7773,
      "step": 7016
    },
    {
      "epoch": 3.3021176470588234,
      "grad_norm": 0.5000644458135518,
      "learning_rate": 0.0001656647374033107,
      "loss": 0.4043,
      "step": 7017
    },
    {
      "epoch": 3.3025882352941176,
      "grad_norm": 0.45786284139429706,
      "learning_rate": 0.00016562114025145332,
      "loss": 0.2363,
      "step": 7018
    },
    {
      "epoch": 3.303058823529412,
      "grad_norm": 0.9856583740284879,
      "learning_rate": 0.00016557754176547404,
      "loss": 0.416,
      "step": 7019
    },
    {
      "epoch": 3.303529411764706,
      "grad_norm": 1.538029550322813,
      "learning_rate": 0.0001655339419490962,
      "loss": 1.0781,
      "step": 7020
    },
    {
      "epoch": 3.304,
      "grad_norm": 0.4829531971406288,
      "learning_rate": 0.00016549034080604359,
      "loss": 0.2969,
      "step": 7021
    },
    {
      "epoch": 3.304470588235294,
      "grad_norm": 0.3953623392909417,
      "learning_rate": 0.00016544673834003985,
      "loss": 0.2656,
      "step": 7022
    },
    {
      "epoch": 3.3049411764705883,
      "grad_norm": 0.44222676376844466,
      "learning_rate": 0.00016540313455480895,
      "loss": 0.3477,
      "step": 7023
    },
    {
      "epoch": 3.3054117647058825,
      "grad_norm": 0.43163476114183363,
      "learning_rate": 0.00016535952945407473,
      "loss": 0.5078,
      "step": 7024
    },
    {
      "epoch": 3.3058823529411763,
      "grad_norm": 0.41888996686506497,
      "learning_rate": 0.00016531592304156137,
      "loss": 0.2598,
      "step": 7025
    },
    {
      "epoch": 3.3063529411764705,
      "grad_norm": 0.3870043607316303,
      "learning_rate": 0.0001652723153209931,
      "loss": 0.1963,
      "step": 7026
    },
    {
      "epoch": 3.3068235294117647,
      "grad_norm": 0.5416799224450184,
      "learning_rate": 0.0001652287062960942,
      "loss": 0.668,
      "step": 7027
    },
    {
      "epoch": 3.307294117647059,
      "grad_norm": 0.2191793231258709,
      "learning_rate": 0.00016518509597058902,
      "loss": 0.1094,
      "step": 7028
    },
    {
      "epoch": 3.307764705882353,
      "grad_norm": 0.059664469581630915,
      "learning_rate": 0.0001651414843482022,
      "loss": 0.0184,
      "step": 7029
    },
    {
      "epoch": 3.308235294117647,
      "grad_norm": 0.1505223212866609,
      "learning_rate": 0.00016509787143265836,
      "loss": 0.0459,
      "step": 7030
    },
    {
      "epoch": 3.308705882352941,
      "grad_norm": 0.5482169226772623,
      "learning_rate": 0.0001650542572276822,
      "loss": 0.4727,
      "step": 7031
    },
    {
      "epoch": 3.3091764705882354,
      "grad_norm": 0.35544955689336455,
      "learning_rate": 0.00016501064173699866,
      "loss": 0.1172,
      "step": 7032
    },
    {
      "epoch": 3.309647058823529,
      "grad_norm": 0.5308833987256052,
      "learning_rate": 0.0001649670249643327,
      "loss": 0.2988,
      "step": 7033
    },
    {
      "epoch": 3.3101176470588234,
      "grad_norm": 0.8257769177810069,
      "learning_rate": 0.0001649234069134094,
      "loss": 0.5312,
      "step": 7034
    },
    {
      "epoch": 3.3105882352941176,
      "grad_norm": 0.22151528445141094,
      "learning_rate": 0.00016487978758795397,
      "loss": 0.0649,
      "step": 7035
    },
    {
      "epoch": 3.311058823529412,
      "grad_norm": 0.5973013305154915,
      "learning_rate": 0.00016483616699169163,
      "loss": 0.5,
      "step": 7036
    },
    {
      "epoch": 3.311529411764706,
      "grad_norm": 0.4022452081646853,
      "learning_rate": 0.0001647925451283479,
      "loss": 0.3711,
      "step": 7037
    },
    {
      "epoch": 3.312,
      "grad_norm": 0.22429781899891443,
      "learning_rate": 0.00016474892200164825,
      "loss": 0.1069,
      "step": 7038
    },
    {
      "epoch": 3.312470588235294,
      "grad_norm": 0.526363535939289,
      "learning_rate": 0.0001647052976153183,
      "loss": 0.5859,
      "step": 7039
    },
    {
      "epoch": 3.3129411764705883,
      "grad_norm": 0.40487757692048015,
      "learning_rate": 0.00016466167197308384,
      "loss": 0.165,
      "step": 7040
    },
    {
      "epoch": 3.3134117647058825,
      "grad_norm": 0.6493087535117595,
      "learning_rate": 0.00016461804507867067,
      "loss": 0.5625,
      "step": 7041
    },
    {
      "epoch": 3.3138823529411763,
      "grad_norm": 0.7270109122802328,
      "learning_rate": 0.00016457441693580472,
      "loss": 0.9609,
      "step": 7042
    },
    {
      "epoch": 3.3143529411764705,
      "grad_norm": 0.6028141379943024,
      "learning_rate": 0.0001645307875482121,
      "loss": 0.4766,
      "step": 7043
    },
    {
      "epoch": 3.3148235294117647,
      "grad_norm": 1.8977007579579501,
      "learning_rate": 0.00016448715691961898,
      "loss": 0.8438,
      "step": 7044
    },
    {
      "epoch": 3.315294117647059,
      "grad_norm": 0.3747418941967982,
      "learning_rate": 0.00016444352505375155,
      "loss": 0.375,
      "step": 7045
    },
    {
      "epoch": 3.315764705882353,
      "grad_norm": 0.4696970701520477,
      "learning_rate": 0.00016439989195433626,
      "loss": 0.5391,
      "step": 7046
    },
    {
      "epoch": 3.316235294117647,
      "grad_norm": 0.32809969099052355,
      "learning_rate": 0.00016435625762509955,
      "loss": 0.1196,
      "step": 7047
    },
    {
      "epoch": 3.316705882352941,
      "grad_norm": 0.6783378850782571,
      "learning_rate": 0.00016431262206976805,
      "loss": 0.9141,
      "step": 7048
    },
    {
      "epoch": 3.3171764705882354,
      "grad_norm": 0.4191418441108275,
      "learning_rate": 0.00016426898529206836,
      "loss": 0.1582,
      "step": 7049
    },
    {
      "epoch": 3.317647058823529,
      "grad_norm": 0.39898217337144243,
      "learning_rate": 0.00016422534729572738,
      "loss": 0.1924,
      "step": 7050
    },
    {
      "epoch": 3.3181176470588234,
      "grad_norm": 0.33782692905783807,
      "learning_rate": 0.00016418170808447198,
      "loss": 0.1191,
      "step": 7051
    },
    {
      "epoch": 3.3185882352941176,
      "grad_norm": 0.5209851399603962,
      "learning_rate": 0.00016413806766202912,
      "loss": 0.2695,
      "step": 7052
    },
    {
      "epoch": 3.319058823529412,
      "grad_norm": 0.3581357564728952,
      "learning_rate": 0.00016409442603212594,
      "loss": 0.125,
      "step": 7053
    },
    {
      "epoch": 3.319529411764706,
      "grad_norm": 0.624927212147444,
      "learning_rate": 0.00016405078319848965,
      "loss": 0.6484,
      "step": 7054
    },
    {
      "epoch": 3.32,
      "grad_norm": 0.20914117866846624,
      "learning_rate": 0.0001640071391648476,
      "loss": 0.0483,
      "step": 7055
    },
    {
      "epoch": 3.320470588235294,
      "grad_norm": 0.6086003478776031,
      "learning_rate": 0.0001639634939349271,
      "loss": 0.4434,
      "step": 7056
    },
    {
      "epoch": 3.3209411764705883,
      "grad_norm": 0.2575413556024565,
      "learning_rate": 0.00016391984751245572,
      "loss": 0.2139,
      "step": 7057
    },
    {
      "epoch": 3.3214117647058825,
      "grad_norm": 0.9123577226457764,
      "learning_rate": 0.00016387619990116116,
      "loss": 0.7773,
      "step": 7058
    },
    {
      "epoch": 3.3218823529411763,
      "grad_norm": 0.40838596200387534,
      "learning_rate": 0.00016383255110477102,
      "loss": 0.4414,
      "step": 7059
    },
    {
      "epoch": 3.3223529411764705,
      "grad_norm": 0.5471748468132435,
      "learning_rate": 0.00016378890112701312,
      "loss": 0.6523,
      "step": 7060
    },
    {
      "epoch": 3.3228235294117647,
      "grad_norm": 0.5107279154414126,
      "learning_rate": 0.0001637452499716155,
      "loss": 0.4023,
      "step": 7061
    },
    {
      "epoch": 3.323294117647059,
      "grad_norm": 0.5595723308456575,
      "learning_rate": 0.0001637015976423061,
      "loss": 0.4609,
      "step": 7062
    },
    {
      "epoch": 3.323764705882353,
      "grad_norm": 0.41801207026901965,
      "learning_rate": 0.00016365794414281303,
      "loss": 0.2871,
      "step": 7063
    },
    {
      "epoch": 3.324235294117647,
      "grad_norm": 0.7165235614439284,
      "learning_rate": 0.00016361428947686457,
      "loss": 0.5508,
      "step": 7064
    },
    {
      "epoch": 3.324705882352941,
      "grad_norm": 0.36356355231581533,
      "learning_rate": 0.00016357063364818902,
      "loss": 0.2344,
      "step": 7065
    },
    {
      "epoch": 3.3251764705882354,
      "grad_norm": 0.24289436048734958,
      "learning_rate": 0.00016352697666051474,
      "loss": 0.0698,
      "step": 7066
    },
    {
      "epoch": 3.325647058823529,
      "grad_norm": 0.24537245156652968,
      "learning_rate": 0.0001634833185175703,
      "loss": 0.125,
      "step": 7067
    },
    {
      "epoch": 3.3261176470588234,
      "grad_norm": 0.38120730268329284,
      "learning_rate": 0.00016343965922308438,
      "loss": 0.3027,
      "step": 7068
    },
    {
      "epoch": 3.3265882352941176,
      "grad_norm": 0.42973555668803126,
      "learning_rate": 0.00016339599878078563,
      "loss": 0.6445,
      "step": 7069
    },
    {
      "epoch": 3.327058823529412,
      "grad_norm": 0.5355077716942908,
      "learning_rate": 0.00016335233719440282,
      "loss": 0.6406,
      "step": 7070
    },
    {
      "epoch": 3.327529411764706,
      "grad_norm": 0.49091129023012076,
      "learning_rate": 0.000163308674467665,
      "loss": 0.3145,
      "step": 7071
    },
    {
      "epoch": 3.328,
      "grad_norm": 0.4231104843153672,
      "learning_rate": 0.00016326501060430105,
      "loss": 0.4121,
      "step": 7072
    },
    {
      "epoch": 3.328470588235294,
      "grad_norm": 0.4762802150855757,
      "learning_rate": 0.00016322134560804017,
      "loss": 0.6484,
      "step": 7073
    },
    {
      "epoch": 3.3289411764705883,
      "grad_norm": 0.2880151934296432,
      "learning_rate": 0.00016317767948261148,
      "loss": 0.2012,
      "step": 7074
    },
    {
      "epoch": 3.3294117647058825,
      "grad_norm": 0.49886784826087943,
      "learning_rate": 0.00016313401223174434,
      "loss": 0.5156,
      "step": 7075
    },
    {
      "epoch": 3.3298823529411763,
      "grad_norm": 0.27879234841658573,
      "learning_rate": 0.00016309034385916815,
      "loss": 0.1309,
      "step": 7076
    },
    {
      "epoch": 3.3303529411764705,
      "grad_norm": 0.31603328473327497,
      "learning_rate": 0.00016304667436861233,
      "loss": 0.3613,
      "step": 7077
    },
    {
      "epoch": 3.3308235294117647,
      "grad_norm": 0.39810865757062247,
      "learning_rate": 0.00016300300376380657,
      "loss": 0.3594,
      "step": 7078
    },
    {
      "epoch": 3.331294117647059,
      "grad_norm": 0.25838552201412335,
      "learning_rate": 0.00016295933204848052,
      "loss": 0.166,
      "step": 7079
    },
    {
      "epoch": 3.331764705882353,
      "grad_norm": 0.6054074397483462,
      "learning_rate": 0.00016291565922636392,
      "loss": 0.625,
      "step": 7080
    },
    {
      "epoch": 3.332235294117647,
      "grad_norm": 0.37987452875463806,
      "learning_rate": 0.0001628719853011867,
      "loss": 0.5508,
      "step": 7081
    },
    {
      "epoch": 3.332705882352941,
      "grad_norm": 0.3787877948418088,
      "learning_rate": 0.00016282831027667881,
      "loss": 0.2949,
      "step": 7082
    },
    {
      "epoch": 3.3331764705882354,
      "grad_norm": 0.20772229275458728,
      "learning_rate": 0.00016278463415657027,
      "loss": 0.0603,
      "step": 7083
    },
    {
      "epoch": 3.333647058823529,
      "grad_norm": 0.4906190499838942,
      "learning_rate": 0.00016274095694459128,
      "loss": 0.4629,
      "step": 7084
    },
    {
      "epoch": 3.3341176470588234,
      "grad_norm": 0.35339775084786185,
      "learning_rate": 0.00016269727864447206,
      "loss": 0.1621,
      "step": 7085
    },
    {
      "epoch": 3.3345882352941176,
      "grad_norm": 0.5267539822623367,
      "learning_rate": 0.00016265359925994298,
      "loss": 0.4492,
      "step": 7086
    },
    {
      "epoch": 3.335058823529412,
      "grad_norm": 0.4379016611176938,
      "learning_rate": 0.00016260991879473441,
      "loss": 0.4668,
      "step": 7087
    },
    {
      "epoch": 3.335529411764706,
      "grad_norm": 0.4948823850095478,
      "learning_rate": 0.000162566237252577,
      "loss": 0.3359,
      "step": 7088
    },
    {
      "epoch": 3.336,
      "grad_norm": 0.3650417450443442,
      "learning_rate": 0.00016252255463720125,
      "loss": 0.2266,
      "step": 7089
    },
    {
      "epoch": 3.336470588235294,
      "grad_norm": 0.5373799752359832,
      "learning_rate": 0.0001624788709523379,
      "loss": 0.8008,
      "step": 7090
    },
    {
      "epoch": 3.3369411764705883,
      "grad_norm": 0.47629382359926903,
      "learning_rate": 0.00016243518620171778,
      "loss": 0.2988,
      "step": 7091
    },
    {
      "epoch": 3.3374117647058825,
      "grad_norm": 0.4269035519559883,
      "learning_rate": 0.00016239150038907177,
      "loss": 0.4199,
      "step": 7092
    },
    {
      "epoch": 3.3378823529411763,
      "grad_norm": 0.5105895499458738,
      "learning_rate": 0.00016234781351813079,
      "loss": 0.4238,
      "step": 7093
    },
    {
      "epoch": 3.3383529411764705,
      "grad_norm": 0.6504406316483539,
      "learning_rate": 0.00016230412559262606,
      "loss": 0.9648,
      "step": 7094
    },
    {
      "epoch": 3.3388235294117647,
      "grad_norm": 0.4517841982465247,
      "learning_rate": 0.00016226043661628862,
      "loss": 0.3984,
      "step": 7095
    },
    {
      "epoch": 3.339294117647059,
      "grad_norm": 0.43767133503518024,
      "learning_rate": 0.00016221674659284976,
      "loss": 0.2598,
      "step": 7096
    },
    {
      "epoch": 3.339764705882353,
      "grad_norm": 0.341563497868345,
      "learning_rate": 0.00016217305552604076,
      "loss": 0.1084,
      "step": 7097
    },
    {
      "epoch": 3.340235294117647,
      "grad_norm": 0.5096451838747961,
      "learning_rate": 0.00016212936341959312,
      "loss": 0.5117,
      "step": 7098
    },
    {
      "epoch": 3.340705882352941,
      "grad_norm": 0.543832974130045,
      "learning_rate": 0.0001620856702772384,
      "loss": 0.7383,
      "step": 7099
    },
    {
      "epoch": 3.3411764705882354,
      "grad_norm": 0.41328810340274946,
      "learning_rate": 0.00016204197610270816,
      "loss": 0.4258,
      "step": 7100
    },
    {
      "epoch": 3.341647058823529,
      "grad_norm": 0.3306859635318176,
      "learning_rate": 0.00016199828089973406,
      "loss": 0.3457,
      "step": 7101
    },
    {
      "epoch": 3.3421176470588234,
      "grad_norm": 0.47223102507484455,
      "learning_rate": 0.00016195458467204793,
      "loss": 0.3652,
      "step": 7102
    },
    {
      "epoch": 3.3425882352941176,
      "grad_norm": 0.17929737877491198,
      "learning_rate": 0.00016191088742338166,
      "loss": 0.1025,
      "step": 7103
    },
    {
      "epoch": 3.343058823529412,
      "grad_norm": 0.3507075242572101,
      "learning_rate": 0.00016186718915746714,
      "loss": 0.1582,
      "step": 7104
    },
    {
      "epoch": 3.343529411764706,
      "grad_norm": 0.7699702011857834,
      "learning_rate": 0.00016182348987803648,
      "loss": 1.4375,
      "step": 7105
    },
    {
      "epoch": 3.344,
      "grad_norm": 0.38817913748833877,
      "learning_rate": 0.0001617797895888218,
      "loss": 0.1699,
      "step": 7106
    },
    {
      "epoch": 3.344470588235294,
      "grad_norm": 0.5261708514990894,
      "learning_rate": 0.00016173608829355527,
      "loss": 0.2471,
      "step": 7107
    },
    {
      "epoch": 3.3449411764705883,
      "grad_norm": 0.49833812281758866,
      "learning_rate": 0.00016169238599596927,
      "loss": 0.5273,
      "step": 7108
    },
    {
      "epoch": 3.3454117647058825,
      "grad_norm": 0.3654569828123673,
      "learning_rate": 0.00016164868269979619,
      "loss": 0.3008,
      "step": 7109
    },
    {
      "epoch": 3.3458823529411763,
      "grad_norm": 0.44239325121502976,
      "learning_rate": 0.0001616049784087684,
      "loss": 0.6562,
      "step": 7110
    },
    {
      "epoch": 3.3463529411764705,
      "grad_norm": 0.41022741126069,
      "learning_rate": 0.00016156127312661855,
      "loss": 0.2812,
      "step": 7111
    },
    {
      "epoch": 3.3468235294117648,
      "grad_norm": 0.27874969169706815,
      "learning_rate": 0.0001615175668570793,
      "loss": 0.0977,
      "step": 7112
    },
    {
      "epoch": 3.347294117647059,
      "grad_norm": 0.39041671669110767,
      "learning_rate": 0.0001614738596038833,
      "loss": 0.2598,
      "step": 7113
    },
    {
      "epoch": 3.3477647058823528,
      "grad_norm": 0.3903222648404416,
      "learning_rate": 0.00016143015137076345,
      "loss": 0.3516,
      "step": 7114
    },
    {
      "epoch": 3.348235294117647,
      "grad_norm": 0.48764383263450967,
      "learning_rate": 0.0001613864421614526,
      "loss": 0.5273,
      "step": 7115
    },
    {
      "epoch": 3.348705882352941,
      "grad_norm": 0.2774893747808157,
      "learning_rate": 0.00016134273197968374,
      "loss": 0.2715,
      "step": 7116
    },
    {
      "epoch": 3.3491764705882354,
      "grad_norm": 0.2900632521800892,
      "learning_rate": 0.0001612990208291899,
      "loss": 0.2676,
      "step": 7117
    },
    {
      "epoch": 3.349647058823529,
      "grad_norm": 0.36524527615474683,
      "learning_rate": 0.0001612553087137043,
      "loss": 0.3203,
      "step": 7118
    },
    {
      "epoch": 3.3501176470588234,
      "grad_norm": 0.350487880878673,
      "learning_rate": 0.0001612115956369601,
      "loss": 0.2217,
      "step": 7119
    },
    {
      "epoch": 3.3505882352941176,
      "grad_norm": 0.21272608880565427,
      "learning_rate": 0.00016116788160269065,
      "loss": 0.2197,
      "step": 7120
    },
    {
      "epoch": 3.351058823529412,
      "grad_norm": 0.2693675377515333,
      "learning_rate": 0.0001611241666146293,
      "loss": 0.1001,
      "step": 7121
    },
    {
      "epoch": 3.351529411764706,
      "grad_norm": 0.28518608508121074,
      "learning_rate": 0.00016108045067650962,
      "loss": 0.1943,
      "step": 7122
    },
    {
      "epoch": 3.352,
      "grad_norm": 0.4203406571833206,
      "learning_rate": 0.0001610367337920651,
      "loss": 0.4062,
      "step": 7123
    },
    {
      "epoch": 3.352470588235294,
      "grad_norm": 0.21488600017235182,
      "learning_rate": 0.00016099301596502936,
      "loss": 0.0654,
      "step": 7124
    },
    {
      "epoch": 3.3529411764705883,
      "grad_norm": 0.46239224819770747,
      "learning_rate": 0.00016094929719913612,
      "loss": 0.5898,
      "step": 7125
    },
    {
      "epoch": 3.3534117647058825,
      "grad_norm": 0.5032390547686428,
      "learning_rate": 0.0001609055774981192,
      "loss": 0.4961,
      "step": 7126
    },
    {
      "epoch": 3.3538823529411763,
      "grad_norm": 0.3026846524189839,
      "learning_rate": 0.0001608618568657125,
      "loss": 0.1338,
      "step": 7127
    },
    {
      "epoch": 3.3543529411764705,
      "grad_norm": 0.31224741834668124,
      "learning_rate": 0.0001608181353056499,
      "loss": 0.2314,
      "step": 7128
    },
    {
      "epoch": 3.3548235294117648,
      "grad_norm": 0.6148491292615903,
      "learning_rate": 0.0001607744128216655,
      "loss": 0.543,
      "step": 7129
    },
    {
      "epoch": 3.355294117647059,
      "grad_norm": 0.41291234964313567,
      "learning_rate": 0.00016073068941749343,
      "loss": 0.1738,
      "step": 7130
    },
    {
      "epoch": 3.3557647058823528,
      "grad_norm": 0.42501661333655183,
      "learning_rate": 0.0001606869650968678,
      "loss": 0.5391,
      "step": 7131
    },
    {
      "epoch": 3.356235294117647,
      "grad_norm": 0.2721768642716766,
      "learning_rate": 0.00016064323986352295,
      "loss": 0.1328,
      "step": 7132
    },
    {
      "epoch": 3.356705882352941,
      "grad_norm": 0.5267063034963971,
      "learning_rate": 0.00016059951372119323,
      "loss": 0.5,
      "step": 7133
    },
    {
      "epoch": 3.3571764705882354,
      "grad_norm": 0.530971125028512,
      "learning_rate": 0.000160555786673613,
      "loss": 0.5977,
      "step": 7134
    },
    {
      "epoch": 3.357647058823529,
      "grad_norm": 0.34310905275690057,
      "learning_rate": 0.00016051205872451686,
      "loss": 0.2559,
      "step": 7135
    },
    {
      "epoch": 3.3581176470588234,
      "grad_norm": 0.4915710697962814,
      "learning_rate": 0.00016046832987763935,
      "loss": 0.5273,
      "step": 7136
    },
    {
      "epoch": 3.3585882352941177,
      "grad_norm": 0.20407180054905266,
      "learning_rate": 0.0001604246001367151,
      "loss": 0.0776,
      "step": 7137
    },
    {
      "epoch": 3.359058823529412,
      "grad_norm": 0.46389906573283396,
      "learning_rate": 0.00016038086950547885,
      "loss": 0.1963,
      "step": 7138
    },
    {
      "epoch": 3.359529411764706,
      "grad_norm": 0.35756999388508703,
      "learning_rate": 0.00016033713798766546,
      "loss": 0.5312,
      "step": 7139
    },
    {
      "epoch": 3.36,
      "grad_norm": 0.4885949871640689,
      "learning_rate": 0.00016029340558700977,
      "loss": 0.4141,
      "step": 7140
    },
    {
      "epoch": 3.360470588235294,
      "grad_norm": 0.3826946441431254,
      "learning_rate": 0.00016024967230724672,
      "loss": 0.1826,
      "step": 7141
    },
    {
      "epoch": 3.3609411764705883,
      "grad_norm": 0.3885604763614334,
      "learning_rate": 0.00016020593815211138,
      "loss": 0.2412,
      "step": 7142
    },
    {
      "epoch": 3.3614117647058825,
      "grad_norm": 0.3567776382484825,
      "learning_rate": 0.00016016220312533888,
      "loss": 0.3867,
      "step": 7143
    },
    {
      "epoch": 3.3618823529411763,
      "grad_norm": 0.28154050122805113,
      "learning_rate": 0.00016011846723066445,
      "loss": 0.1318,
      "step": 7144
    },
    {
      "epoch": 3.3623529411764705,
      "grad_norm": 0.46933124220894723,
      "learning_rate": 0.0001600747304718232,
      "loss": 0.6602,
      "step": 7145
    },
    {
      "epoch": 3.3628235294117648,
      "grad_norm": 0.399381698613404,
      "learning_rate": 0.00016003099285255058,
      "loss": 0.1787,
      "step": 7146
    },
    {
      "epoch": 3.363294117647059,
      "grad_norm": 0.3830812698018788,
      "learning_rate": 0.000159987254376582,
      "loss": 0.2871,
      "step": 7147
    },
    {
      "epoch": 3.3637647058823528,
      "grad_norm": 0.6783507072707292,
      "learning_rate": 0.00015994351504765287,
      "loss": 0.8594,
      "step": 7148
    },
    {
      "epoch": 3.364235294117647,
      "grad_norm": 0.45696116042003604,
      "learning_rate": 0.00015989977486949883,
      "loss": 0.5273,
      "step": 7149
    },
    {
      "epoch": 3.364705882352941,
      "grad_norm": 0.6052242810345191,
      "learning_rate": 0.00015985603384585542,
      "loss": 0.5312,
      "step": 7150
    },
    {
      "epoch": 3.3651764705882354,
      "grad_norm": 0.43807057367584223,
      "learning_rate": 0.0001598122919804584,
      "loss": 0.2471,
      "step": 7151
    },
    {
      "epoch": 3.365647058823529,
      "grad_norm": 0.2894835453596497,
      "learning_rate": 0.0001597685492770435,
      "loss": 0.1289,
      "step": 7152
    },
    {
      "epoch": 3.3661176470588234,
      "grad_norm": 0.3955545106121837,
      "learning_rate": 0.00015972480573934656,
      "loss": 0.2793,
      "step": 7153
    },
    {
      "epoch": 3.3665882352941177,
      "grad_norm": 0.3576538155730258,
      "learning_rate": 0.00015968106137110352,
      "loss": 0.4375,
      "step": 7154
    },
    {
      "epoch": 3.367058823529412,
      "grad_norm": 0.4500363957601794,
      "learning_rate": 0.00015963731617605034,
      "loss": 0.6016,
      "step": 7155
    },
    {
      "epoch": 3.367529411764706,
      "grad_norm": 0.5000526975208305,
      "learning_rate": 0.00015959357015792313,
      "loss": 0.6172,
      "step": 7156
    },
    {
      "epoch": 3.368,
      "grad_norm": 0.6460897470096464,
      "learning_rate": 0.00015954982332045794,
      "loss": 0.6992,
      "step": 7157
    },
    {
      "epoch": 3.368470588235294,
      "grad_norm": 0.6443037279994347,
      "learning_rate": 0.000159506075667391,
      "loss": 0.793,
      "step": 7158
    },
    {
      "epoch": 3.3689411764705883,
      "grad_norm": 0.5025809292782359,
      "learning_rate": 0.0001594623272024586,
      "loss": 0.6992,
      "step": 7159
    },
    {
      "epoch": 3.369411764705882,
      "grad_norm": 0.47058931000415355,
      "learning_rate": 0.000159418577929397,
      "loss": 0.4453,
      "step": 7160
    },
    {
      "epoch": 3.3698823529411763,
      "grad_norm": 0.3701959425809595,
      "learning_rate": 0.0001593748278519427,
      "loss": 0.1904,
      "step": 7161
    },
    {
      "epoch": 3.3703529411764706,
      "grad_norm": 0.5282664854709613,
      "learning_rate": 0.00015933107697383206,
      "loss": 0.7734,
      "step": 7162
    },
    {
      "epoch": 3.3708235294117648,
      "grad_norm": 0.39930004310849615,
      "learning_rate": 0.0001592873252988017,
      "loss": 0.293,
      "step": 7163
    },
    {
      "epoch": 3.371294117647059,
      "grad_norm": 0.3494348628959517,
      "learning_rate": 0.0001592435728305882,
      "loss": 0.1553,
      "step": 7164
    },
    {
      "epoch": 3.3717647058823528,
      "grad_norm": 0.3654399835515288,
      "learning_rate": 0.00015919981957292825,
      "loss": 0.3203,
      "step": 7165
    },
    {
      "epoch": 3.372235294117647,
      "grad_norm": 0.5440115243271069,
      "learning_rate": 0.00015915606552955856,
      "loss": 0.6992,
      "step": 7166
    },
    {
      "epoch": 3.372705882352941,
      "grad_norm": 0.4608178526914635,
      "learning_rate": 0.000159112310704216,
      "loss": 0.377,
      "step": 7167
    },
    {
      "epoch": 3.3731764705882354,
      "grad_norm": 0.5101763455296203,
      "learning_rate": 0.00015906855510063744,
      "loss": 0.5547,
      "step": 7168
    },
    {
      "epoch": 3.373647058823529,
      "grad_norm": 0.48541500280667427,
      "learning_rate": 0.0001590247987225597,
      "loss": 0.6055,
      "step": 7169
    },
    {
      "epoch": 3.3741176470588234,
      "grad_norm": 0.71184710792478,
      "learning_rate": 0.00015898104157371993,
      "loss": 0.7539,
      "step": 7170
    },
    {
      "epoch": 3.3745882352941177,
      "grad_norm": 0.28169752241121454,
      "learning_rate": 0.00015893728365785518,
      "loss": 0.0884,
      "step": 7171
    },
    {
      "epoch": 3.375058823529412,
      "grad_norm": 0.33012916680486776,
      "learning_rate": 0.00015889352497870253,
      "loss": 0.2051,
      "step": 7172
    },
    {
      "epoch": 3.375529411764706,
      "grad_norm": 0.44684109932105526,
      "learning_rate": 0.00015884976553999925,
      "loss": 0.3027,
      "step": 7173
    },
    {
      "epoch": 3.376,
      "grad_norm": 0.25149124487231866,
      "learning_rate": 0.0001588060053454826,
      "loss": 0.082,
      "step": 7174
    },
    {
      "epoch": 3.376470588235294,
      "grad_norm": 0.5514248055111305,
      "learning_rate": 0.0001587622443988899,
      "loss": 0.4629,
      "step": 7175
    },
    {
      "epoch": 3.3769411764705883,
      "grad_norm": 0.25274423519114153,
      "learning_rate": 0.0001587184827039585,
      "loss": 0.1445,
      "step": 7176
    },
    {
      "epoch": 3.377411764705882,
      "grad_norm": 0.4560946194963575,
      "learning_rate": 0.00015867472026442596,
      "loss": 0.5352,
      "step": 7177
    },
    {
      "epoch": 3.3778823529411763,
      "grad_norm": 0.8483161738782856,
      "learning_rate": 0.00015863095708402972,
      "loss": 0.6719,
      "step": 7178
    },
    {
      "epoch": 3.3783529411764706,
      "grad_norm": 0.44245051560520515,
      "learning_rate": 0.00015858719316650744,
      "loss": 0.2969,
      "step": 7179
    },
    {
      "epoch": 3.378823529411765,
      "grad_norm": 0.5159589984843559,
      "learning_rate": 0.00015854342851559673,
      "loss": 0.543,
      "step": 7180
    },
    {
      "epoch": 3.379294117647059,
      "grad_norm": 0.37469588679408405,
      "learning_rate": 0.00015849966313503537,
      "loss": 0.3691,
      "step": 7181
    },
    {
      "epoch": 3.379764705882353,
      "grad_norm": 0.23825491393805648,
      "learning_rate": 0.00015845589702856102,
      "loss": 0.0771,
      "step": 7182
    },
    {
      "epoch": 3.380235294117647,
      "grad_norm": 0.4487238058952765,
      "learning_rate": 0.0001584121301999116,
      "loss": 0.5781,
      "step": 7183
    },
    {
      "epoch": 3.3807058823529412,
      "grad_norm": 0.27416450964416617,
      "learning_rate": 0.00015836836265282502,
      "loss": 0.0598,
      "step": 7184
    },
    {
      "epoch": 3.3811764705882354,
      "grad_norm": 0.46177969474717034,
      "learning_rate": 0.0001583245943910392,
      "loss": 0.418,
      "step": 7185
    },
    {
      "epoch": 3.3816470588235292,
      "grad_norm": 0.6827955657704371,
      "learning_rate": 0.0001582808254182922,
      "loss": 0.5039,
      "step": 7186
    },
    {
      "epoch": 3.3821176470588235,
      "grad_norm": 0.39279856373104527,
      "learning_rate": 0.0001582370557383221,
      "loss": 0.3848,
      "step": 7187
    },
    {
      "epoch": 3.3825882352941177,
      "grad_norm": 0.43380249481795174,
      "learning_rate": 0.00015819328535486704,
      "loss": 0.5234,
      "step": 7188
    },
    {
      "epoch": 3.383058823529412,
      "grad_norm": 0.36206094290954094,
      "learning_rate": 0.0001581495142716652,
      "loss": 0.2773,
      "step": 7189
    },
    {
      "epoch": 3.383529411764706,
      "grad_norm": 0.24370166616326386,
      "learning_rate": 0.00015810574249245488,
      "loss": 0.0732,
      "step": 7190
    },
    {
      "epoch": 3.384,
      "grad_norm": 0.22510926989214297,
      "learning_rate": 0.00015806197002097438,
      "loss": 0.1621,
      "step": 7191
    },
    {
      "epoch": 3.384470588235294,
      "grad_norm": 0.48011462412091205,
      "learning_rate": 0.00015801819686096213,
      "loss": 0.5078,
      "step": 7192
    },
    {
      "epoch": 3.3849411764705883,
      "grad_norm": 0.5059201804608381,
      "learning_rate": 0.00015797442301615652,
      "loss": 0.4062,
      "step": 7193
    },
    {
      "epoch": 3.385411764705882,
      "grad_norm": 0.520730810087993,
      "learning_rate": 0.00015793064849029608,
      "loss": 0.5391,
      "step": 7194
    },
    {
      "epoch": 3.3858823529411763,
      "grad_norm": 0.39107911790179845,
      "learning_rate": 0.00015788687328711933,
      "loss": 0.4434,
      "step": 7195
    },
    {
      "epoch": 3.3863529411764706,
      "grad_norm": 0.4606298089592112,
      "learning_rate": 0.00015784309741036492,
      "loss": 0.3672,
      "step": 7196
    },
    {
      "epoch": 3.386823529411765,
      "grad_norm": 0.36510858299021176,
      "learning_rate": 0.00015779932086377153,
      "loss": 0.2178,
      "step": 7197
    },
    {
      "epoch": 3.387294117647059,
      "grad_norm": 0.3563771338014015,
      "learning_rate": 0.00015775554365107791,
      "loss": 0.3926,
      "step": 7198
    },
    {
      "epoch": 3.387764705882353,
      "grad_norm": 0.3282603985120203,
      "learning_rate": 0.0001577117657760228,
      "loss": 0.4824,
      "step": 7199
    },
    {
      "epoch": 3.388235294117647,
      "grad_norm": 0.5656101465532433,
      "learning_rate": 0.00015766798724234506,
      "loss": 0.3691,
      "step": 7200
    },
    {
      "epoch": 3.3887058823529412,
      "grad_norm": 0.29803302053962066,
      "learning_rate": 0.00015762420805378363,
      "loss": 0.1416,
      "step": 7201
    },
    {
      "epoch": 3.3891764705882355,
      "grad_norm": 0.8491189081583276,
      "learning_rate": 0.00015758042821407742,
      "loss": 0.8555,
      "step": 7202
    },
    {
      "epoch": 3.3896470588235292,
      "grad_norm": 0.4451132691948338,
      "learning_rate": 0.00015753664772696544,
      "loss": 0.4727,
      "step": 7203
    },
    {
      "epoch": 3.3901176470588235,
      "grad_norm": 0.5047776731330093,
      "learning_rate": 0.00015749286659618678,
      "loss": 0.4258,
      "step": 7204
    },
    {
      "epoch": 3.3905882352941177,
      "grad_norm": 0.4909601607091956,
      "learning_rate": 0.00015744908482548059,
      "loss": 0.3809,
      "step": 7205
    },
    {
      "epoch": 3.391058823529412,
      "grad_norm": 0.6788777471125146,
      "learning_rate": 0.00015740530241858597,
      "loss": 0.3398,
      "step": 7206
    },
    {
      "epoch": 3.391529411764706,
      "grad_norm": 0.39380520207819886,
      "learning_rate": 0.00015736151937924224,
      "loss": 0.4219,
      "step": 7207
    },
    {
      "epoch": 3.392,
      "grad_norm": 0.37637314433630287,
      "learning_rate": 0.00015731773571118864,
      "loss": 0.5078,
      "step": 7208
    },
    {
      "epoch": 3.392470588235294,
      "grad_norm": 0.585331933419635,
      "learning_rate": 0.00015727395141816448,
      "loss": 0.4258,
      "step": 7209
    },
    {
      "epoch": 3.3929411764705883,
      "grad_norm": 0.29287551065874456,
      "learning_rate": 0.00015723016650390922,
      "loss": 0.1279,
      "step": 7210
    },
    {
      "epoch": 3.393411764705882,
      "grad_norm": 0.4249057606410479,
      "learning_rate": 0.0001571863809721623,
      "loss": 0.2793,
      "step": 7211
    },
    {
      "epoch": 3.3938823529411764,
      "grad_norm": 0.4070092914135337,
      "learning_rate": 0.00015714259482666317,
      "loss": 0.3965,
      "step": 7212
    },
    {
      "epoch": 3.3943529411764706,
      "grad_norm": 0.4663313738824815,
      "learning_rate": 0.00015709880807115138,
      "loss": 0.4355,
      "step": 7213
    },
    {
      "epoch": 3.394823529411765,
      "grad_norm": 0.2343397268747706,
      "learning_rate": 0.0001570550207093666,
      "loss": 0.1709,
      "step": 7214
    },
    {
      "epoch": 3.395294117647059,
      "grad_norm": 0.32087817802685564,
      "learning_rate": 0.00015701123274504845,
      "loss": 0.1108,
      "step": 7215
    },
    {
      "epoch": 3.395764705882353,
      "grad_norm": 0.5474887199433364,
      "learning_rate": 0.00015696744418193664,
      "loss": 0.4961,
      "step": 7216
    },
    {
      "epoch": 3.396235294117647,
      "grad_norm": 0.2786332210036668,
      "learning_rate": 0.00015692365502377088,
      "loss": 0.1226,
      "step": 7217
    },
    {
      "epoch": 3.3967058823529412,
      "grad_norm": 0.4784972569519095,
      "learning_rate": 0.00015687986527429103,
      "loss": 0.5078,
      "step": 7218
    },
    {
      "epoch": 3.3971764705882355,
      "grad_norm": 0.42945798269143687,
      "learning_rate": 0.00015683607493723699,
      "loss": 0.2539,
      "step": 7219
    },
    {
      "epoch": 3.3976470588235292,
      "grad_norm": 0.45180490657037314,
      "learning_rate": 0.00015679228401634855,
      "loss": 0.457,
      "step": 7220
    },
    {
      "epoch": 3.3981176470588235,
      "grad_norm": 0.24881196159720922,
      "learning_rate": 0.0001567484925153658,
      "loss": 0.2324,
      "step": 7221
    },
    {
      "epoch": 3.3985882352941177,
      "grad_norm": 0.41069834725681853,
      "learning_rate": 0.00015670470043802867,
      "loss": 0.3613,
      "step": 7222
    },
    {
      "epoch": 3.399058823529412,
      "grad_norm": 0.2821638029907191,
      "learning_rate": 0.0001566609077880772,
      "loss": 0.1152,
      "step": 7223
    },
    {
      "epoch": 3.3995294117647057,
      "grad_norm": 0.791129398331098,
      "learning_rate": 0.00015661711456925158,
      "loss": 0.4727,
      "step": 7224
    },
    {
      "epoch": 3.4,
      "grad_norm": 0.31154042276116345,
      "learning_rate": 0.0001565733207852919,
      "loss": 0.1396,
      "step": 7225
    },
    {
      "epoch": 3.400470588235294,
      "grad_norm": 0.39004682101366844,
      "learning_rate": 0.00015652952643993841,
      "loss": 0.2178,
      "step": 7226
    },
    {
      "epoch": 3.4009411764705884,
      "grad_norm": 0.5415255530506662,
      "learning_rate": 0.00015648573153693128,
      "loss": 0.6562,
      "step": 7227
    },
    {
      "epoch": 3.401411764705882,
      "grad_norm": 0.368970954740116,
      "learning_rate": 0.0001564419360800109,
      "loss": 0.5625,
      "step": 7228
    },
    {
      "epoch": 3.4018823529411764,
      "grad_norm": 0.35393913027319485,
      "learning_rate": 0.0001563981400729176,
      "loss": 0.2012,
      "step": 7229
    },
    {
      "epoch": 3.4023529411764706,
      "grad_norm": 0.48511005776243277,
      "learning_rate": 0.0001563543435193917,
      "loss": 0.4121,
      "step": 7230
    },
    {
      "epoch": 3.402823529411765,
      "grad_norm": 0.4500225682106602,
      "learning_rate": 0.0001563105464231737,
      "loss": 0.2812,
      "step": 7231
    },
    {
      "epoch": 3.403294117647059,
      "grad_norm": 0.5317730187153811,
      "learning_rate": 0.0001562667487880041,
      "loss": 0.416,
      "step": 7232
    },
    {
      "epoch": 3.403764705882353,
      "grad_norm": 0.43817795560223566,
      "learning_rate": 0.00015622295061762346,
      "loss": 0.4746,
      "step": 7233
    },
    {
      "epoch": 3.404235294117647,
      "grad_norm": 0.24534812028475955,
      "learning_rate": 0.00015617915191577222,
      "loss": 0.2197,
      "step": 7234
    },
    {
      "epoch": 3.4047058823529412,
      "grad_norm": 0.3417703685398537,
      "learning_rate": 0.0001561353526861912,
      "loss": 0.3848,
      "step": 7235
    },
    {
      "epoch": 3.4051764705882355,
      "grad_norm": 0.23988985181149505,
      "learning_rate": 0.0001560915529326209,
      "loss": 0.208,
      "step": 7236
    },
    {
      "epoch": 3.4056470588235292,
      "grad_norm": 0.3950260336681269,
      "learning_rate": 0.00015604775265880212,
      "loss": 0.1836,
      "step": 7237
    },
    {
      "epoch": 3.4061176470588235,
      "grad_norm": 0.3762878757923616,
      "learning_rate": 0.0001560039518684756,
      "loss": 0.1709,
      "step": 7238
    },
    {
      "epoch": 3.4065882352941177,
      "grad_norm": 0.28697092480794,
      "learning_rate": 0.00015596015056538215,
      "loss": 0.1807,
      "step": 7239
    },
    {
      "epoch": 3.407058823529412,
      "grad_norm": 0.4236417182976252,
      "learning_rate": 0.0001559163487532626,
      "loss": 0.21,
      "step": 7240
    },
    {
      "epoch": 3.4075294117647057,
      "grad_norm": 0.287070837259474,
      "learning_rate": 0.00015587254643585784,
      "loss": 0.2227,
      "step": 7241
    },
    {
      "epoch": 3.408,
      "grad_norm": 0.32214388267819466,
      "learning_rate": 0.00015582874361690885,
      "loss": 0.1904,
      "step": 7242
    },
    {
      "epoch": 3.408470588235294,
      "grad_norm": 0.6050788346235622,
      "learning_rate": 0.00015578494030015652,
      "loss": 0.4746,
      "step": 7243
    },
    {
      "epoch": 3.4089411764705884,
      "grad_norm": 0.2604196346014749,
      "learning_rate": 0.00015574113648934192,
      "loss": 0.0967,
      "step": 7244
    },
    {
      "epoch": 3.409411764705882,
      "grad_norm": 0.3376884826671314,
      "learning_rate": 0.0001556973321882061,
      "loss": 0.0918,
      "step": 7245
    },
    {
      "epoch": 3.4098823529411764,
      "grad_norm": 0.6062732344070348,
      "learning_rate": 0.0001556535274004902,
      "loss": 0.5859,
      "step": 7246
    },
    {
      "epoch": 3.4103529411764706,
      "grad_norm": 0.5149931877962911,
      "learning_rate": 0.00015560972212993527,
      "loss": 0.375,
      "step": 7247
    },
    {
      "epoch": 3.410823529411765,
      "grad_norm": 0.49353464344618786,
      "learning_rate": 0.0001555659163802826,
      "loss": 0.2871,
      "step": 7248
    },
    {
      "epoch": 3.411294117647059,
      "grad_norm": 0.25605097008593053,
      "learning_rate": 0.00015552211015527336,
      "loss": 0.1001,
      "step": 7249
    },
    {
      "epoch": 3.411764705882353,
      "grad_norm": 0.4933488893093389,
      "learning_rate": 0.00015547830345864885,
      "loss": 0.4863,
      "step": 7250
    },
    {
      "epoch": 3.412235294117647,
      "grad_norm": 0.5439165450543678,
      "learning_rate": 0.00015543449629415028,
      "loss": 0.3848,
      "step": 7251
    },
    {
      "epoch": 3.4127058823529413,
      "grad_norm": 0.6271101865406451,
      "learning_rate": 0.00015539068866551916,
      "loss": 0.8047,
      "step": 7252
    },
    {
      "epoch": 3.4131764705882355,
      "grad_norm": 0.4221981092544511,
      "learning_rate": 0.00015534688057649673,
      "loss": 0.4805,
      "step": 7253
    },
    {
      "epoch": 3.4136470588235293,
      "grad_norm": 0.3936169873429984,
      "learning_rate": 0.0001553030720308245,
      "loss": 0.3281,
      "step": 7254
    },
    {
      "epoch": 3.4141176470588235,
      "grad_norm": 0.519235368000271,
      "learning_rate": 0.00015525926303224387,
      "loss": 0.3965,
      "step": 7255
    },
    {
      "epoch": 3.4145882352941177,
      "grad_norm": 0.11176874424889346,
      "learning_rate": 0.0001552154535844964,
      "loss": 0.0569,
      "step": 7256
    },
    {
      "epoch": 3.415058823529412,
      "grad_norm": 0.48487708662977674,
      "learning_rate": 0.00015517164369132364,
      "loss": 0.4199,
      "step": 7257
    },
    {
      "epoch": 3.4155294117647057,
      "grad_norm": 0.32020114364794017,
      "learning_rate": 0.0001551278333564671,
      "loss": 0.1836,
      "step": 7258
    },
    {
      "epoch": 3.416,
      "grad_norm": 0.5029115046952336,
      "learning_rate": 0.00015508402258366846,
      "loss": 0.8086,
      "step": 7259
    },
    {
      "epoch": 3.416470588235294,
      "grad_norm": 0.41985784503785567,
      "learning_rate": 0.00015504021137666938,
      "loss": 0.4766,
      "step": 7260
    },
    {
      "epoch": 3.4169411764705884,
      "grad_norm": 0.3660733954800565,
      "learning_rate": 0.00015499639973921148,
      "loss": 0.2119,
      "step": 7261
    },
    {
      "epoch": 3.417411764705882,
      "grad_norm": 0.6442418346302405,
      "learning_rate": 0.00015495258767503657,
      "loss": 0.625,
      "step": 7262
    },
    {
      "epoch": 3.4178823529411764,
      "grad_norm": 0.47539151393573365,
      "learning_rate": 0.0001549087751878864,
      "loss": 0.4141,
      "step": 7263
    },
    {
      "epoch": 3.4183529411764706,
      "grad_norm": 0.4476321370950057,
      "learning_rate": 0.00015486496228150266,
      "loss": 0.5586,
      "step": 7264
    },
    {
      "epoch": 3.418823529411765,
      "grad_norm": 0.33618364703121084,
      "learning_rate": 0.00015482114895962735,
      "loss": 0.1318,
      "step": 7265
    },
    {
      "epoch": 3.419294117647059,
      "grad_norm": 0.5121573718376757,
      "learning_rate": 0.00015477733522600227,
      "loss": 0.5195,
      "step": 7266
    },
    {
      "epoch": 3.419764705882353,
      "grad_norm": 0.5583826134336329,
      "learning_rate": 0.00015473352108436934,
      "loss": 0.707,
      "step": 7267
    },
    {
      "epoch": 3.420235294117647,
      "grad_norm": 0.4077235090194791,
      "learning_rate": 0.00015468970653847043,
      "loss": 0.3516,
      "step": 7268
    },
    {
      "epoch": 3.4207058823529413,
      "grad_norm": 0.4404179075931578,
      "learning_rate": 0.00015464589159204764,
      "loss": 0.5586,
      "step": 7269
    },
    {
      "epoch": 3.4211764705882355,
      "grad_norm": 0.6022292149599113,
      "learning_rate": 0.0001546020762488429,
      "loss": 0.5117,
      "step": 7270
    },
    {
      "epoch": 3.4216470588235293,
      "grad_norm": 0.523645725631281,
      "learning_rate": 0.00015455826051259826,
      "loss": 0.6406,
      "step": 7271
    },
    {
      "epoch": 3.4221176470588235,
      "grad_norm": 0.38714393292322075,
      "learning_rate": 0.00015451444438705584,
      "loss": 0.416,
      "step": 7272
    },
    {
      "epoch": 3.4225882352941177,
      "grad_norm": 0.3514607965018619,
      "learning_rate": 0.00015447062787595772,
      "loss": 0.3262,
      "step": 7273
    },
    {
      "epoch": 3.423058823529412,
      "grad_norm": 0.5218353074247992,
      "learning_rate": 0.00015442681098304603,
      "loss": 0.7148,
      "step": 7274
    },
    {
      "epoch": 3.4235294117647057,
      "grad_norm": 0.7639637745815525,
      "learning_rate": 0.000154382993712063,
      "loss": 0.7617,
      "step": 7275
    },
    {
      "epoch": 3.424,
      "grad_norm": 0.35827982620887705,
      "learning_rate": 0.00015433917606675075,
      "loss": 0.1963,
      "step": 7276
    },
    {
      "epoch": 3.424470588235294,
      "grad_norm": 0.5790224922767785,
      "learning_rate": 0.00015429535805085162,
      "loss": 0.9023,
      "step": 7277
    },
    {
      "epoch": 3.4249411764705884,
      "grad_norm": 0.2967030972814797,
      "learning_rate": 0.00015425153966810783,
      "loss": 0.2578,
      "step": 7278
    },
    {
      "epoch": 3.425411764705882,
      "grad_norm": 0.2217742064321871,
      "learning_rate": 0.00015420772092226173,
      "loss": 0.0884,
      "step": 7279
    },
    {
      "epoch": 3.4258823529411764,
      "grad_norm": 0.36458280521769537,
      "learning_rate": 0.00015416390181705562,
      "loss": 0.459,
      "step": 7280
    },
    {
      "epoch": 3.4263529411764706,
      "grad_norm": 0.4966251080678819,
      "learning_rate": 0.00015412008235623186,
      "loss": 0.6211,
      "step": 7281
    },
    {
      "epoch": 3.426823529411765,
      "grad_norm": 0.4535219158432179,
      "learning_rate": 0.00015407626254353282,
      "loss": 0.2773,
      "step": 7282
    },
    {
      "epoch": 3.427294117647059,
      "grad_norm": 0.32388482001031865,
      "learning_rate": 0.00015403244238270102,
      "loss": 0.1299,
      "step": 7283
    },
    {
      "epoch": 3.427764705882353,
      "grad_norm": 0.4253610950226975,
      "learning_rate": 0.00015398862187747888,
      "loss": 0.4668,
      "step": 7284
    },
    {
      "epoch": 3.428235294117647,
      "grad_norm": 0.5823885171802474,
      "learning_rate": 0.0001539448010316088,
      "loss": 0.5117,
      "step": 7285
    },
    {
      "epoch": 3.4287058823529413,
      "grad_norm": 0.5932648773092261,
      "learning_rate": 0.0001539009798488334,
      "loss": 0.3164,
      "step": 7286
    },
    {
      "epoch": 3.429176470588235,
      "grad_norm": 0.6086334752612785,
      "learning_rate": 0.0001538571583328952,
      "loss": 0.5664,
      "step": 7287
    },
    {
      "epoch": 3.4296470588235293,
      "grad_norm": 0.6541917323817473,
      "learning_rate": 0.0001538133364875367,
      "loss": 0.7266,
      "step": 7288
    },
    {
      "epoch": 3.4301176470588235,
      "grad_norm": 0.3693497643366708,
      "learning_rate": 0.00015376951431650063,
      "loss": 0.373,
      "step": 7289
    },
    {
      "epoch": 3.4305882352941177,
      "grad_norm": 0.4404526208562239,
      "learning_rate": 0.00015372569182352952,
      "loss": 0.4785,
      "step": 7290
    },
    {
      "epoch": 3.431058823529412,
      "grad_norm": 0.4525864466036143,
      "learning_rate": 0.00015368186901236605,
      "loss": 0.4316,
      "step": 7291
    },
    {
      "epoch": 3.4315294117647057,
      "grad_norm": 0.5162415474517298,
      "learning_rate": 0.00015363804588675292,
      "loss": 0.7852,
      "step": 7292
    },
    {
      "epoch": 3.432,
      "grad_norm": 0.2952546677030305,
      "learning_rate": 0.00015359422245043278,
      "loss": 0.1602,
      "step": 7293
    },
    {
      "epoch": 3.432470588235294,
      "grad_norm": 0.48476910602853535,
      "learning_rate": 0.00015355039870714848,
      "loss": 0.5,
      "step": 7294
    },
    {
      "epoch": 3.4329411764705884,
      "grad_norm": 0.24936670832924304,
      "learning_rate": 0.00015350657466064264,
      "loss": 0.1064,
      "step": 7295
    },
    {
      "epoch": 3.433411764705882,
      "grad_norm": 0.4817781649822802,
      "learning_rate": 0.00015346275031465818,
      "loss": 0.3828,
      "step": 7296
    },
    {
      "epoch": 3.4338823529411764,
      "grad_norm": 0.4509503859507429,
      "learning_rate": 0.00015341892567293785,
      "loss": 0.4824,
      "step": 7297
    },
    {
      "epoch": 3.4343529411764706,
      "grad_norm": 0.27532625493404456,
      "learning_rate": 0.0001533751007392245,
      "loss": 0.1055,
      "step": 7298
    },
    {
      "epoch": 3.434823529411765,
      "grad_norm": 0.5179755659941834,
      "learning_rate": 0.00015333127551726096,
      "loss": 0.6289,
      "step": 7299
    },
    {
      "epoch": 3.435294117647059,
      "grad_norm": 0.30863971234185816,
      "learning_rate": 0.0001532874500107902,
      "loss": 0.1074,
      "step": 7300
    },
    {
      "epoch": 3.435764705882353,
      "grad_norm": 0.2903541332428663,
      "learning_rate": 0.00015324362422355507,
      "loss": 0.2598,
      "step": 7301
    },
    {
      "epoch": 3.436235294117647,
      "grad_norm": 0.3041298663098362,
      "learning_rate": 0.00015319979815929847,
      "loss": 0.1641,
      "step": 7302
    },
    {
      "epoch": 3.4367058823529413,
      "grad_norm": 0.22303358600516734,
      "learning_rate": 0.00015315597182176348,
      "loss": 0.0737,
      "step": 7303
    },
    {
      "epoch": 3.437176470588235,
      "grad_norm": 0.4170184996998411,
      "learning_rate": 0.000153112145214693,
      "loss": 0.1953,
      "step": 7304
    },
    {
      "epoch": 3.4376470588235293,
      "grad_norm": 0.3583236780112225,
      "learning_rate": 0.00015306831834183008,
      "loss": 0.4141,
      "step": 7305
    },
    {
      "epoch": 3.4381176470588235,
      "grad_norm": 0.3919087724607981,
      "learning_rate": 0.00015302449120691765,
      "loss": 0.3535,
      "step": 7306
    },
    {
      "epoch": 3.4385882352941177,
      "grad_norm": 0.21479671459090693,
      "learning_rate": 0.00015298066381369888,
      "loss": 0.0574,
      "step": 7307
    },
    {
      "epoch": 3.439058823529412,
      "grad_norm": 0.6613898584897646,
      "learning_rate": 0.0001529368361659168,
      "loss": 0.9023,
      "step": 7308
    },
    {
      "epoch": 3.4395294117647057,
      "grad_norm": 0.499343846738671,
      "learning_rate": 0.0001528930082673145,
      "loss": 0.4297,
      "step": 7309
    },
    {
      "epoch": 3.44,
      "grad_norm": 0.40245165206222094,
      "learning_rate": 0.0001528491801216351,
      "loss": 0.3789,
      "step": 7310
    },
    {
      "epoch": 3.440470588235294,
      "grad_norm": 0.30158933711301644,
      "learning_rate": 0.00015280535173262176,
      "loss": 0.1807,
      "step": 7311
    },
    {
      "epoch": 3.4409411764705884,
      "grad_norm": 0.6308093769859704,
      "learning_rate": 0.0001527615231040176,
      "loss": 0.5859,
      "step": 7312
    },
    {
      "epoch": 3.441411764705882,
      "grad_norm": 0.4092062626598098,
      "learning_rate": 0.00015271769423956584,
      "loss": 0.4609,
      "step": 7313
    },
    {
      "epoch": 3.4418823529411764,
      "grad_norm": 0.45572761784519794,
      "learning_rate": 0.00015267386514300965,
      "loss": 0.5781,
      "step": 7314
    },
    {
      "epoch": 3.4423529411764706,
      "grad_norm": 0.47761058868237893,
      "learning_rate": 0.00015263003581809226,
      "loss": 0.5195,
      "step": 7315
    },
    {
      "epoch": 3.442823529411765,
      "grad_norm": 0.5295509888248575,
      "learning_rate": 0.00015258620626855691,
      "loss": 0.3867,
      "step": 7316
    },
    {
      "epoch": 3.443294117647059,
      "grad_norm": 0.45257894074217975,
      "learning_rate": 0.00015254237649814687,
      "loss": 0.5,
      "step": 7317
    },
    {
      "epoch": 3.443764705882353,
      "grad_norm": 0.4422900992909717,
      "learning_rate": 0.00015249854651060542,
      "loss": 0.3906,
      "step": 7318
    },
    {
      "epoch": 3.444235294117647,
      "grad_norm": 0.5117678401525869,
      "learning_rate": 0.00015245471630967583,
      "loss": 0.5039,
      "step": 7319
    },
    {
      "epoch": 3.4447058823529413,
      "grad_norm": 0.774848007904134,
      "learning_rate": 0.00015241088589910138,
      "loss": 0.6953,
      "step": 7320
    },
    {
      "epoch": 3.445176470588235,
      "grad_norm": 0.39527854520656275,
      "learning_rate": 0.0001523670552826255,
      "loss": 0.2656,
      "step": 7321
    },
    {
      "epoch": 3.4456470588235293,
      "grad_norm": 0.5096604211781963,
      "learning_rate": 0.00015232322446399147,
      "loss": 0.5,
      "step": 7322
    },
    {
      "epoch": 3.4461176470588235,
      "grad_norm": 0.4403680407451653,
      "learning_rate": 0.00015227939344694268,
      "loss": 0.2314,
      "step": 7323
    },
    {
      "epoch": 3.4465882352941177,
      "grad_norm": 0.3396192045000506,
      "learning_rate": 0.0001522355622352225,
      "loss": 0.3105,
      "step": 7324
    },
    {
      "epoch": 3.447058823529412,
      "grad_norm": 0.442067200303811,
      "learning_rate": 0.00015219173083257437,
      "loss": 0.2393,
      "step": 7325
    },
    {
      "epoch": 3.4475294117647057,
      "grad_norm": 0.43720800513884844,
      "learning_rate": 0.00015214789924274168,
      "loss": 0.3281,
      "step": 7326
    },
    {
      "epoch": 3.448,
      "grad_norm": 0.5986331890232326,
      "learning_rate": 0.00015210406746946783,
      "loss": 0.5156,
      "step": 7327
    },
    {
      "epoch": 3.448470588235294,
      "grad_norm": 0.5108138525362083,
      "learning_rate": 0.0001520602355164963,
      "loss": 0.3398,
      "step": 7328
    },
    {
      "epoch": 3.4489411764705884,
      "grad_norm": 0.5492385060869277,
      "learning_rate": 0.00015201640338757054,
      "loss": 0.4766,
      "step": 7329
    },
    {
      "epoch": 3.449411764705882,
      "grad_norm": 0.4439946117973668,
      "learning_rate": 0.00015197257108643406,
      "loss": 0.4395,
      "step": 7330
    },
    {
      "epoch": 3.4498823529411764,
      "grad_norm": 0.2503838764836207,
      "learning_rate": 0.00015192873861683036,
      "loss": 0.1768,
      "step": 7331
    },
    {
      "epoch": 3.4503529411764706,
      "grad_norm": 0.451706305878288,
      "learning_rate": 0.00015188490598250288,
      "loss": 0.7227,
      "step": 7332
    },
    {
      "epoch": 3.450823529411765,
      "grad_norm": 0.4405824667365426,
      "learning_rate": 0.0001518410731871952,
      "loss": 0.5508,
      "step": 7333
    },
    {
      "epoch": 3.451294117647059,
      "grad_norm": 0.32410397334101726,
      "learning_rate": 0.00015179724023465084,
      "loss": 0.3184,
      "step": 7334
    },
    {
      "epoch": 3.451764705882353,
      "grad_norm": 0.42817251132273093,
      "learning_rate": 0.00015175340712861337,
      "loss": 0.4102,
      "step": 7335
    },
    {
      "epoch": 3.452235294117647,
      "grad_norm": 0.6513853191432629,
      "learning_rate": 0.00015170957387282627,
      "loss": 0.5859,
      "step": 7336
    },
    {
      "epoch": 3.4527058823529413,
      "grad_norm": 0.5142805762302076,
      "learning_rate": 0.00015166574047103322,
      "loss": 0.3984,
      "step": 7337
    },
    {
      "epoch": 3.453176470588235,
      "grad_norm": 0.6555572977650156,
      "learning_rate": 0.00015162190692697778,
      "loss": 0.4238,
      "step": 7338
    },
    {
      "epoch": 3.4536470588235293,
      "grad_norm": 0.420743086121181,
      "learning_rate": 0.00015157807324440347,
      "loss": 0.293,
      "step": 7339
    },
    {
      "epoch": 3.4541176470588235,
      "grad_norm": 0.4380415597700594,
      "learning_rate": 0.00015153423942705402,
      "loss": 0.4473,
      "step": 7340
    },
    {
      "epoch": 3.4545882352941177,
      "grad_norm": 0.31770738324891057,
      "learning_rate": 0.000151490405478673,
      "loss": 0.2949,
      "step": 7341
    },
    {
      "epoch": 3.455058823529412,
      "grad_norm": 0.3900657696015838,
      "learning_rate": 0.000151446571403004,
      "loss": 0.4102,
      "step": 7342
    },
    {
      "epoch": 3.4555294117647057,
      "grad_norm": 0.4924118466974919,
      "learning_rate": 0.0001514027372037907,
      "loss": 0.2471,
      "step": 7343
    },
    {
      "epoch": 3.456,
      "grad_norm": 0.6412736005178875,
      "learning_rate": 0.0001513589028847768,
      "loss": 0.6523,
      "step": 7344
    },
    {
      "epoch": 3.456470588235294,
      "grad_norm": 0.39835800472372124,
      "learning_rate": 0.00015131506844970596,
      "loss": 0.6836,
      "step": 7345
    },
    {
      "epoch": 3.4569411764705884,
      "grad_norm": 0.33711123742871074,
      "learning_rate": 0.00015127123390232177,
      "loss": 0.1406,
      "step": 7346
    },
    {
      "epoch": 3.457411764705882,
      "grad_norm": 0.6138941719893573,
      "learning_rate": 0.00015122739924636796,
      "loss": 0.5898,
      "step": 7347
    },
    {
      "epoch": 3.4578823529411764,
      "grad_norm": 0.41986366460499336,
      "learning_rate": 0.00015118356448558828,
      "loss": 0.5586,
      "step": 7348
    },
    {
      "epoch": 3.4583529411764706,
      "grad_norm": 0.35882455910099526,
      "learning_rate": 0.00015113972962372635,
      "loss": 0.1904,
      "step": 7349
    },
    {
      "epoch": 3.458823529411765,
      "grad_norm": 1.2622238328667528,
      "learning_rate": 0.00015109589466452594,
      "loss": 0.7148,
      "step": 7350
    },
    {
      "epoch": 3.4592941176470586,
      "grad_norm": 0.6768960645220403,
      "learning_rate": 0.00015105205961173074,
      "loss": 0.9297,
      "step": 7351
    },
    {
      "epoch": 3.459764705882353,
      "grad_norm": 0.41483074053732794,
      "learning_rate": 0.00015100822446908452,
      "loss": 0.3105,
      "step": 7352
    },
    {
      "epoch": 3.460235294117647,
      "grad_norm": 0.38502521677889995,
      "learning_rate": 0.00015096438924033095,
      "loss": 0.1768,
      "step": 7353
    },
    {
      "epoch": 3.4607058823529413,
      "grad_norm": 0.480911454918458,
      "learning_rate": 0.00015092055392921383,
      "loss": 0.2988,
      "step": 7354
    },
    {
      "epoch": 3.461176470588235,
      "grad_norm": 0.3699595656711025,
      "learning_rate": 0.00015087671853947684,
      "loss": 0.3828,
      "step": 7355
    },
    {
      "epoch": 3.4616470588235293,
      "grad_norm": 0.43431357334838944,
      "learning_rate": 0.00015083288307486388,
      "loss": 0.3301,
      "step": 7356
    },
    {
      "epoch": 3.4621176470588235,
      "grad_norm": 0.4130704535677841,
      "learning_rate": 0.00015078904753911853,
      "loss": 0.5039,
      "step": 7357
    },
    {
      "epoch": 3.4625882352941177,
      "grad_norm": 0.5289731395362528,
      "learning_rate": 0.0001507452119359847,
      "loss": 0.457,
      "step": 7358
    },
    {
      "epoch": 3.463058823529412,
      "grad_norm": 0.49402878618235824,
      "learning_rate": 0.0001507013762692061,
      "loss": 0.4824,
      "step": 7359
    },
    {
      "epoch": 3.4635294117647057,
      "grad_norm": 0.5138898101551114,
      "learning_rate": 0.0001506575405425265,
      "loss": 0.5586,
      "step": 7360
    },
    {
      "epoch": 3.464,
      "grad_norm": 0.4187239778393084,
      "learning_rate": 0.00015061370475968975,
      "loss": 0.2148,
      "step": 7361
    },
    {
      "epoch": 3.464470588235294,
      "grad_norm": 0.3982600338581352,
      "learning_rate": 0.0001505698689244396,
      "loss": 0.4004,
      "step": 7362
    },
    {
      "epoch": 3.4649411764705884,
      "grad_norm": 0.32024760097278676,
      "learning_rate": 0.0001505260330405198,
      "loss": 0.168,
      "step": 7363
    },
    {
      "epoch": 3.465411764705882,
      "grad_norm": 0.8957384385716652,
      "learning_rate": 0.00015048219711167422,
      "loss": 0.6719,
      "step": 7364
    },
    {
      "epoch": 3.4658823529411764,
      "grad_norm": 0.3324982218941441,
      "learning_rate": 0.00015043836114164667,
      "loss": 0.1021,
      "step": 7365
    },
    {
      "epoch": 3.4663529411764706,
      "grad_norm": 0.5234669770085972,
      "learning_rate": 0.00015039452513418094,
      "loss": 0.6875,
      "step": 7366
    },
    {
      "epoch": 3.466823529411765,
      "grad_norm": 0.36369291286624134,
      "learning_rate": 0.00015035068909302079,
      "loss": 0.4395,
      "step": 7367
    },
    {
      "epoch": 3.4672941176470586,
      "grad_norm": 0.41835380630330793,
      "learning_rate": 0.00015030685302191011,
      "loss": 0.4609,
      "step": 7368
    },
    {
      "epoch": 3.467764705882353,
      "grad_norm": 0.3311926929412264,
      "learning_rate": 0.00015026301692459265,
      "loss": 0.1992,
      "step": 7369
    },
    {
      "epoch": 3.468235294117647,
      "grad_norm": 0.4750944528943279,
      "learning_rate": 0.00015021918080481225,
      "loss": 0.3359,
      "step": 7370
    },
    {
      "epoch": 3.4687058823529413,
      "grad_norm": 0.33229647987709127,
      "learning_rate": 0.00015017534466631272,
      "loss": 0.4434,
      "step": 7371
    },
    {
      "epoch": 3.469176470588235,
      "grad_norm": 0.3558453045773293,
      "learning_rate": 0.00015013150851283795,
      "loss": 0.2324,
      "step": 7372
    },
    {
      "epoch": 3.4696470588235293,
      "grad_norm": 0.562619768760433,
      "learning_rate": 0.0001500876723481317,
      "loss": 0.3105,
      "step": 7373
    },
    {
      "epoch": 3.4701176470588235,
      "grad_norm": 0.4607913513128973,
      "learning_rate": 0.00015004383617593772,
      "loss": 0.5508,
      "step": 7374
    },
    {
      "epoch": 3.4705882352941178,
      "grad_norm": 0.3491086806472399,
      "learning_rate": 0.00015,
      "loss": 0.3828,
      "step": 7375
    },
    {
      "epoch": 3.471058823529412,
      "grad_norm": 0.36773377666773327,
      "learning_rate": 0.00014995616382406225,
      "loss": 0.207,
      "step": 7376
    },
    {
      "epoch": 3.4715294117647058,
      "grad_norm": 0.4494937744731658,
      "learning_rate": 0.00014991232765186832,
      "loss": 0.4336,
      "step": 7377
    },
    {
      "epoch": 3.472,
      "grad_norm": 0.6564883768191019,
      "learning_rate": 0.00014986849148716205,
      "loss": 0.418,
      "step": 7378
    },
    {
      "epoch": 3.472470588235294,
      "grad_norm": 0.7112708344158183,
      "learning_rate": 0.00014982465533368725,
      "loss": 0.6641,
      "step": 7379
    },
    {
      "epoch": 3.4729411764705884,
      "grad_norm": 0.5431171315486135,
      "learning_rate": 0.00014978081919518772,
      "loss": 0.5781,
      "step": 7380
    },
    {
      "epoch": 3.473411764705882,
      "grad_norm": 0.6231565359528483,
      "learning_rate": 0.00014973698307540735,
      "loss": 0.8516,
      "step": 7381
    },
    {
      "epoch": 3.4738823529411764,
      "grad_norm": 0.3246845364425116,
      "learning_rate": 0.00014969314697808991,
      "loss": 0.3281,
      "step": 7382
    },
    {
      "epoch": 3.4743529411764706,
      "grad_norm": 0.5156638980300201,
      "learning_rate": 0.00014964931090697922,
      "loss": 0.2676,
      "step": 7383
    },
    {
      "epoch": 3.474823529411765,
      "grad_norm": 0.33795484071747084,
      "learning_rate": 0.00014960547486581906,
      "loss": 0.291,
      "step": 7384
    },
    {
      "epoch": 3.4752941176470586,
      "grad_norm": 0.1006584725039127,
      "learning_rate": 0.0001495616388583533,
      "loss": 0.0114,
      "step": 7385
    },
    {
      "epoch": 3.475764705882353,
      "grad_norm": 1.1205591152061642,
      "learning_rate": 0.00014951780288832575,
      "loss": 0.9023,
      "step": 7386
    },
    {
      "epoch": 3.476235294117647,
      "grad_norm": 0.3600470242796245,
      "learning_rate": 0.0001494739669594802,
      "loss": 0.3555,
      "step": 7387
    },
    {
      "epoch": 3.4767058823529413,
      "grad_norm": 0.6035812037977585,
      "learning_rate": 0.00014943013107556044,
      "loss": 0.6055,
      "step": 7388
    },
    {
      "epoch": 3.477176470588235,
      "grad_norm": 0.3659280806130849,
      "learning_rate": 0.00014938629524031028,
      "loss": 0.1602,
      "step": 7389
    },
    {
      "epoch": 3.4776470588235293,
      "grad_norm": 0.6223800352688676,
      "learning_rate": 0.00014934245945747349,
      "loss": 0.5312,
      "step": 7390
    },
    {
      "epoch": 3.4781176470588235,
      "grad_norm": 0.37994532276507614,
      "learning_rate": 0.00014929862373079392,
      "loss": 0.3164,
      "step": 7391
    },
    {
      "epoch": 3.4785882352941178,
      "grad_norm": 0.5549446844032917,
      "learning_rate": 0.00014925478806401533,
      "loss": 0.418,
      "step": 7392
    },
    {
      "epoch": 3.479058823529412,
      "grad_norm": 1.0404448009214957,
      "learning_rate": 0.0001492109524608815,
      "loss": 0.9297,
      "step": 7393
    },
    {
      "epoch": 3.4795294117647058,
      "grad_norm": 0.4800587801937296,
      "learning_rate": 0.00014916711692513618,
      "loss": 0.4824,
      "step": 7394
    },
    {
      "epoch": 3.48,
      "grad_norm": 0.31092953431430254,
      "learning_rate": 0.00014912328146052316,
      "loss": 0.1709,
      "step": 7395
    },
    {
      "epoch": 3.480470588235294,
      "grad_norm": 0.7983561155181911,
      "learning_rate": 0.00014907944607078623,
      "loss": 0.5977,
      "step": 7396
    },
    {
      "epoch": 3.4809411764705884,
      "grad_norm": 0.40157488546060577,
      "learning_rate": 0.00014903561075966908,
      "loss": 0.105,
      "step": 7397
    },
    {
      "epoch": 3.481411764705882,
      "grad_norm": 0.48263058905965583,
      "learning_rate": 0.00014899177553091543,
      "loss": 0.3301,
      "step": 7398
    },
    {
      "epoch": 3.4818823529411764,
      "grad_norm": 0.3000884632924536,
      "learning_rate": 0.0001489479403882692,
      "loss": 0.1543,
      "step": 7399
    },
    {
      "epoch": 3.4823529411764707,
      "grad_norm": 0.7890545783496971,
      "learning_rate": 0.00014890410533547404,
      "loss": 0.5742,
      "step": 7400
    },
    {
      "epoch": 3.482823529411765,
      "grad_norm": 0.29571979387105474,
      "learning_rate": 0.00014886027037627362,
      "loss": 0.1108,
      "step": 7401
    },
    {
      "epoch": 3.4832941176470587,
      "grad_norm": 0.4622503606265793,
      "learning_rate": 0.0001488164355144117,
      "loss": 0.2949,
      "step": 7402
    },
    {
      "epoch": 3.483764705882353,
      "grad_norm": 0.4879516940921415,
      "learning_rate": 0.00014877260075363199,
      "loss": 0.4238,
      "step": 7403
    },
    {
      "epoch": 3.484235294117647,
      "grad_norm": 0.33226875163780034,
      "learning_rate": 0.00014872876609767823,
      "loss": 0.3594,
      "step": 7404
    },
    {
      "epoch": 3.4847058823529413,
      "grad_norm": 0.6880616621408739,
      "learning_rate": 0.00014868493155029402,
      "loss": 0.4375,
      "step": 7405
    },
    {
      "epoch": 3.485176470588235,
      "grad_norm": 0.6724435610675331,
      "learning_rate": 0.00014864109711522315,
      "loss": 0.4648,
      "step": 7406
    },
    {
      "epoch": 3.4856470588235293,
      "grad_norm": 0.33606249459676896,
      "learning_rate": 0.00014859726279620926,
      "loss": 0.3652,
      "step": 7407
    },
    {
      "epoch": 3.4861176470588235,
      "grad_norm": 0.6261298102067391,
      "learning_rate": 0.00014855342859699597,
      "loss": 0.7812,
      "step": 7408
    },
    {
      "epoch": 3.4865882352941178,
      "grad_norm": 0.5288449821629994,
      "learning_rate": 0.000148509594521327,
      "loss": 0.5938,
      "step": 7409
    },
    {
      "epoch": 3.487058823529412,
      "grad_norm": 0.5443243934441745,
      "learning_rate": 0.00014846576057294596,
      "loss": 0.4609,
      "step": 7410
    },
    {
      "epoch": 3.4875294117647058,
      "grad_norm": 0.35582333825214507,
      "learning_rate": 0.0001484219267555965,
      "loss": 0.2598,
      "step": 7411
    },
    {
      "epoch": 3.488,
      "grad_norm": 0.5467128412755354,
      "learning_rate": 0.0001483780930730222,
      "loss": 0.3887,
      "step": 7412
    },
    {
      "epoch": 3.488470588235294,
      "grad_norm": 0.4356458393027335,
      "learning_rate": 0.00014833425952896675,
      "loss": 0.3613,
      "step": 7413
    },
    {
      "epoch": 3.488941176470588,
      "grad_norm": 0.45027411413126633,
      "learning_rate": 0.00014829042612717373,
      "loss": 0.3672,
      "step": 7414
    },
    {
      "epoch": 3.489411764705882,
      "grad_norm": 0.43104688848820233,
      "learning_rate": 0.00014824659287138663,
      "loss": 0.416,
      "step": 7415
    },
    {
      "epoch": 3.4898823529411764,
      "grad_norm": 0.5420383763912255,
      "learning_rate": 0.00014820275976534916,
      "loss": 0.2617,
      "step": 7416
    },
    {
      "epoch": 3.4903529411764707,
      "grad_norm": 0.34340239025727237,
      "learning_rate": 0.0001481589268128048,
      "loss": 0.1396,
      "step": 7417
    },
    {
      "epoch": 3.490823529411765,
      "grad_norm": 0.37623074918969784,
      "learning_rate": 0.0001481150940174971,
      "loss": 0.3789,
      "step": 7418
    },
    {
      "epoch": 3.4912941176470587,
      "grad_norm": 0.670970669672579,
      "learning_rate": 0.00014807126138316964,
      "loss": 0.793,
      "step": 7419
    },
    {
      "epoch": 3.491764705882353,
      "grad_norm": 0.37559089863855183,
      "learning_rate": 0.00014802742891356592,
      "loss": 0.4609,
      "step": 7420
    },
    {
      "epoch": 3.492235294117647,
      "grad_norm": 0.4320556396493454,
      "learning_rate": 0.00014798359661242943,
      "loss": 0.4102,
      "step": 7421
    },
    {
      "epoch": 3.4927058823529413,
      "grad_norm": 0.3488198492939669,
      "learning_rate": 0.00014793976448350366,
      "loss": 0.1484,
      "step": 7422
    },
    {
      "epoch": 3.493176470588235,
      "grad_norm": 0.5386333254431569,
      "learning_rate": 0.00014789593253053217,
      "loss": 0.5156,
      "step": 7423
    },
    {
      "epoch": 3.4936470588235293,
      "grad_norm": 0.3047513140911878,
      "learning_rate": 0.00014785210075725835,
      "loss": 0.0913,
      "step": 7424
    },
    {
      "epoch": 3.4941176470588236,
      "grad_norm": 0.5910392438565368,
      "learning_rate": 0.0001478082691674256,
      "loss": 0.5781,
      "step": 7425
    },
    {
      "epoch": 3.4945882352941178,
      "grad_norm": 0.47771440007054944,
      "learning_rate": 0.00014776443776477746,
      "loss": 0.4883,
      "step": 7426
    },
    {
      "epoch": 3.495058823529412,
      "grad_norm": 0.23933222126569453,
      "learning_rate": 0.00014772060655305732,
      "loss": 0.0815,
      "step": 7427
    },
    {
      "epoch": 3.4955294117647058,
      "grad_norm": 0.4101371174266656,
      "learning_rate": 0.0001476767755360085,
      "loss": 0.4297,
      "step": 7428
    },
    {
      "epoch": 3.496,
      "grad_norm": 0.5541662186372948,
      "learning_rate": 0.00014763294471737448,
      "loss": 0.4414,
      "step": 7429
    },
    {
      "epoch": 3.496470588235294,
      "grad_norm": 0.36137197556109446,
      "learning_rate": 0.00014758911410089862,
      "loss": 0.2021,
      "step": 7430
    },
    {
      "epoch": 3.496941176470588,
      "grad_norm": 0.6148622724033878,
      "learning_rate": 0.0001475452836903242,
      "loss": 0.7188,
      "step": 7431
    },
    {
      "epoch": 3.497411764705882,
      "grad_norm": 0.37943239804230505,
      "learning_rate": 0.0001475014534893946,
      "loss": 0.3262,
      "step": 7432
    },
    {
      "epoch": 3.4978823529411764,
      "grad_norm": 0.42518301647564183,
      "learning_rate": 0.00014745762350185316,
      "loss": 0.6055,
      "step": 7433
    },
    {
      "epoch": 3.4983529411764707,
      "grad_norm": 0.35291152754132626,
      "learning_rate": 0.0001474137937314431,
      "loss": 0.1738,
      "step": 7434
    },
    {
      "epoch": 3.498823529411765,
      "grad_norm": 0.3275118905314051,
      "learning_rate": 0.00014736996418190774,
      "loss": 0.0894,
      "step": 7435
    },
    {
      "epoch": 3.4992941176470587,
      "grad_norm": 0.3701109617557376,
      "learning_rate": 0.00014732613485699035,
      "loss": 0.3047,
      "step": 7436
    },
    {
      "epoch": 3.499764705882353,
      "grad_norm": 0.3490935682832823,
      "learning_rate": 0.0001472823057604342,
      "loss": 0.3457,
      "step": 7437
    },
    {
      "epoch": 3.500235294117647,
      "grad_norm": 0.25781903146330926,
      "learning_rate": 0.0001472384768959824,
      "loss": 0.1885,
      "step": 7438
    },
    {
      "epoch": 3.5007058823529413,
      "grad_norm": 0.3235969126348357,
      "learning_rate": 0.00014719464826737827,
      "loss": 0.3359,
      "step": 7439
    },
    {
      "epoch": 3.501176470588235,
      "grad_norm": 0.3129347418305164,
      "learning_rate": 0.0001471508198783649,
      "loss": 0.0967,
      "step": 7440
    },
    {
      "epoch": 3.5016470588235293,
      "grad_norm": 0.28185621111108966,
      "learning_rate": 0.0001471069917326855,
      "loss": 0.0962,
      "step": 7441
    },
    {
      "epoch": 3.5021176470588236,
      "grad_norm": 0.5717115435922742,
      "learning_rate": 0.0001470631638340832,
      "loss": 0.3945,
      "step": 7442
    },
    {
      "epoch": 3.502588235294118,
      "grad_norm": 0.5858127374231178,
      "learning_rate": 0.00014701933618630107,
      "loss": 0.4121,
      "step": 7443
    },
    {
      "epoch": 3.503058823529412,
      "grad_norm": 0.4037932520828935,
      "learning_rate": 0.00014697550879308233,
      "loss": 0.3652,
      "step": 7444
    },
    {
      "epoch": 3.503529411764706,
      "grad_norm": 0.35951838924755564,
      "learning_rate": 0.00014693168165816995,
      "loss": 0.1953,
      "step": 7445
    },
    {
      "epoch": 3.504,
      "grad_norm": 0.45003581513233964,
      "learning_rate": 0.00014688785478530697,
      "loss": 0.5195,
      "step": 7446
    },
    {
      "epoch": 3.5044705882352942,
      "grad_norm": 0.26592380890142475,
      "learning_rate": 0.00014684402817823647,
      "loss": 0.0977,
      "step": 7447
    },
    {
      "epoch": 3.504941176470588,
      "grad_norm": 0.4146967015287281,
      "learning_rate": 0.00014680020184070148,
      "loss": 0.4883,
      "step": 7448
    },
    {
      "epoch": 3.5054117647058822,
      "grad_norm": 0.45895619996376835,
      "learning_rate": 0.0001467563757764449,
      "loss": 0.4023,
      "step": 7449
    },
    {
      "epoch": 3.5058823529411764,
      "grad_norm": 0.3887737715791105,
      "learning_rate": 0.00014671254998920976,
      "loss": 0.1953,
      "step": 7450
    },
    {
      "epoch": 3.5063529411764707,
      "grad_norm": 0.8488151546741134,
      "learning_rate": 0.000146668724482739,
      "loss": 0.5391,
      "step": 7451
    },
    {
      "epoch": 3.506823529411765,
      "grad_norm": 0.3983775992090709,
      "learning_rate": 0.0001466248992607755,
      "loss": 0.1865,
      "step": 7452
    },
    {
      "epoch": 3.5072941176470587,
      "grad_norm": 0.2529252219031833,
      "learning_rate": 0.00014658107432706212,
      "loss": 0.1348,
      "step": 7453
    },
    {
      "epoch": 3.507764705882353,
      "grad_norm": 0.33942407627726284,
      "learning_rate": 0.0001465372496853418,
      "loss": 0.334,
      "step": 7454
    },
    {
      "epoch": 3.508235294117647,
      "grad_norm": 0.3644584585393134,
      "learning_rate": 0.00014649342533935733,
      "loss": 0.1108,
      "step": 7455
    },
    {
      "epoch": 3.5087058823529413,
      "grad_norm": 0.4090523803329226,
      "learning_rate": 0.00014644960129285152,
      "loss": 0.252,
      "step": 7456
    },
    {
      "epoch": 3.509176470588235,
      "grad_norm": 0.5439547871621176,
      "learning_rate": 0.0001464057775495672,
      "loss": 0.5547,
      "step": 7457
    },
    {
      "epoch": 3.5096470588235293,
      "grad_norm": 0.6192844333999606,
      "learning_rate": 0.0001463619541132471,
      "loss": 0.4863,
      "step": 7458
    },
    {
      "epoch": 3.5101176470588236,
      "grad_norm": 0.3722682712233397,
      "learning_rate": 0.00014631813098763392,
      "loss": 0.3809,
      "step": 7459
    },
    {
      "epoch": 3.510588235294118,
      "grad_norm": 0.5097530565041258,
      "learning_rate": 0.00014627430817647048,
      "loss": 0.3711,
      "step": 7460
    },
    {
      "epoch": 3.511058823529412,
      "grad_norm": 0.2699152239545047,
      "learning_rate": 0.00014623048568349937,
      "loss": 0.2832,
      "step": 7461
    },
    {
      "epoch": 3.511529411764706,
      "grad_norm": 0.5794810668849366,
      "learning_rate": 0.0001461866635124633,
      "loss": 0.3594,
      "step": 7462
    },
    {
      "epoch": 3.512,
      "grad_norm": 0.4105879199799546,
      "learning_rate": 0.0001461428416671048,
      "loss": 0.2773,
      "step": 7463
    },
    {
      "epoch": 3.5124705882352942,
      "grad_norm": 0.29410579188462177,
      "learning_rate": 0.0001460990201511666,
      "loss": 0.1455,
      "step": 7464
    },
    {
      "epoch": 3.512941176470588,
      "grad_norm": 0.32441352784112976,
      "learning_rate": 0.0001460551989683912,
      "loss": 0.2109,
      "step": 7465
    },
    {
      "epoch": 3.5134117647058822,
      "grad_norm": 0.4215237960740636,
      "learning_rate": 0.00014601137812252112,
      "loss": 0.2178,
      "step": 7466
    },
    {
      "epoch": 3.5138823529411765,
      "grad_norm": 0.2965247781048533,
      "learning_rate": 0.00014596755761729895,
      "loss": 0.2891,
      "step": 7467
    },
    {
      "epoch": 3.5143529411764707,
      "grad_norm": 0.4318070812718701,
      "learning_rate": 0.00014592373745646715,
      "loss": 0.459,
      "step": 7468
    },
    {
      "epoch": 3.514823529411765,
      "grad_norm": 0.7202584254307193,
      "learning_rate": 0.00014587991764376812,
      "loss": 0.668,
      "step": 7469
    },
    {
      "epoch": 3.5152941176470587,
      "grad_norm": 0.5677069057856268,
      "learning_rate": 0.00014583609818294438,
      "loss": 0.4121,
      "step": 7470
    },
    {
      "epoch": 3.515764705882353,
      "grad_norm": 0.6433333336316552,
      "learning_rate": 0.00014579227907773827,
      "loss": 0.4473,
      "step": 7471
    },
    {
      "epoch": 3.516235294117647,
      "grad_norm": 0.38542345429912755,
      "learning_rate": 0.00014574846033189217,
      "loss": 0.1758,
      "step": 7472
    },
    {
      "epoch": 3.516705882352941,
      "grad_norm": 0.39868596263267875,
      "learning_rate": 0.00014570464194914835,
      "loss": 0.3203,
      "step": 7473
    },
    {
      "epoch": 3.517176470588235,
      "grad_norm": 0.6641536978374661,
      "learning_rate": 0.00014566082393324925,
      "loss": 0.5156,
      "step": 7474
    },
    {
      "epoch": 3.5176470588235293,
      "grad_norm": 0.48864865329878954,
      "learning_rate": 0.00014561700628793706,
      "loss": 0.3613,
      "step": 7475
    },
    {
      "epoch": 3.5181176470588236,
      "grad_norm": 0.43421290152821057,
      "learning_rate": 0.000145573189016954,
      "loss": 0.1543,
      "step": 7476
    },
    {
      "epoch": 3.518588235294118,
      "grad_norm": 0.4819878223696003,
      "learning_rate": 0.0001455293721240423,
      "loss": 0.4121,
      "step": 7477
    },
    {
      "epoch": 3.519058823529412,
      "grad_norm": 0.6317898867993459,
      "learning_rate": 0.0001454855556129442,
      "loss": 0.5469,
      "step": 7478
    },
    {
      "epoch": 3.519529411764706,
      "grad_norm": 0.7952627403145693,
      "learning_rate": 0.00014544173948740174,
      "loss": 0.6836,
      "step": 7479
    },
    {
      "epoch": 3.52,
      "grad_norm": 0.6380373046659246,
      "learning_rate": 0.00014539792375115713,
      "loss": 0.8164,
      "step": 7480
    },
    {
      "epoch": 3.5204705882352942,
      "grad_norm": 0.618422898244654,
      "learning_rate": 0.0001453541084079524,
      "loss": 0.6719,
      "step": 7481
    },
    {
      "epoch": 3.520941176470588,
      "grad_norm": 1.1358351064982213,
      "learning_rate": 0.0001453102934615296,
      "loss": 0.5508,
      "step": 7482
    },
    {
      "epoch": 3.5214117647058822,
      "grad_norm": 0.3332086285027726,
      "learning_rate": 0.0001452664789156307,
      "loss": 0.3203,
      "step": 7483
    },
    {
      "epoch": 3.5218823529411765,
      "grad_norm": 0.30672269426012927,
      "learning_rate": 0.00014522266477399776,
      "loss": 0.1738,
      "step": 7484
    },
    {
      "epoch": 3.5223529411764707,
      "grad_norm": 0.5457359940642176,
      "learning_rate": 0.00014517885104037268,
      "loss": 0.2734,
      "step": 7485
    },
    {
      "epoch": 3.522823529411765,
      "grad_norm": 0.2356454562608774,
      "learning_rate": 0.00014513503771849734,
      "loss": 0.1504,
      "step": 7486
    },
    {
      "epoch": 3.5232941176470587,
      "grad_norm": 0.2994265751563665,
      "learning_rate": 0.00014509122481211358,
      "loss": 0.1079,
      "step": 7487
    },
    {
      "epoch": 3.523764705882353,
      "grad_norm": 0.5211641987716952,
      "learning_rate": 0.0001450474123249634,
      "loss": 0.3574,
      "step": 7488
    },
    {
      "epoch": 3.524235294117647,
      "grad_norm": 0.6327590567209339,
      "learning_rate": 0.00014500360026078847,
      "loss": 0.5508,
      "step": 7489
    },
    {
      "epoch": 3.524705882352941,
      "grad_norm": 0.46814478055746633,
      "learning_rate": 0.00014495978862333063,
      "loss": 0.3652,
      "step": 7490
    },
    {
      "epoch": 3.525176470588235,
      "grad_norm": 0.5072797379976173,
      "learning_rate": 0.00014491597741633148,
      "loss": 0.334,
      "step": 7491
    },
    {
      "epoch": 3.5256470588235294,
      "grad_norm": 0.18432541067014935,
      "learning_rate": 0.00014487216664353287,
      "loss": 0.0515,
      "step": 7492
    },
    {
      "epoch": 3.5261176470588236,
      "grad_norm": 0.38195699593127413,
      "learning_rate": 0.00014482835630867636,
      "loss": 0.1797,
      "step": 7493
    },
    {
      "epoch": 3.526588235294118,
      "grad_norm": 0.2068753492459608,
      "learning_rate": 0.00014478454641550354,
      "loss": 0.0562,
      "step": 7494
    },
    {
      "epoch": 3.527058823529412,
      "grad_norm": 0.3518004063944332,
      "learning_rate": 0.0001447407369677561,
      "loss": 0.2227,
      "step": 7495
    },
    {
      "epoch": 3.527529411764706,
      "grad_norm": 0.5178398214373707,
      "learning_rate": 0.0001446969279691755,
      "loss": 0.4902,
      "step": 7496
    },
    {
      "epoch": 3.528,
      "grad_norm": 0.14627094910598892,
      "learning_rate": 0.00014465311942350324,
      "loss": 0.0796,
      "step": 7497
    },
    {
      "epoch": 3.5284705882352942,
      "grad_norm": 0.45472300818985945,
      "learning_rate": 0.00014460931133448084,
      "loss": 0.2754,
      "step": 7498
    },
    {
      "epoch": 3.528941176470588,
      "grad_norm": 0.21780041944398015,
      "learning_rate": 0.0001445655037058497,
      "loss": 0.0742,
      "step": 7499
    },
    {
      "epoch": 3.5294117647058822,
      "grad_norm": 0.2099579756055235,
      "learning_rate": 0.00014452169654135115,
      "loss": 0.0635,
      "step": 7500
    },
    {
      "epoch": 3.5298823529411765,
      "grad_norm": 0.2352598341917752,
      "learning_rate": 0.0001444778898447266,
      "loss": 0.0806,
      "step": 7501
    },
    {
      "epoch": 3.5303529411764707,
      "grad_norm": 0.4208925305342409,
      "learning_rate": 0.00014443408361971739,
      "loss": 0.373,
      "step": 7502
    },
    {
      "epoch": 3.530823529411765,
      "grad_norm": 0.2914612793343351,
      "learning_rate": 0.0001443902778700647,
      "loss": 0.1309,
      "step": 7503
    },
    {
      "epoch": 3.5312941176470587,
      "grad_norm": 0.22276020698874596,
      "learning_rate": 0.0001443464725995098,
      "loss": 0.0664,
      "step": 7504
    },
    {
      "epoch": 3.531764705882353,
      "grad_norm": 0.48339382314842494,
      "learning_rate": 0.0001443026678117939,
      "loss": 0.543,
      "step": 7505
    },
    {
      "epoch": 3.532235294117647,
      "grad_norm": 0.427643402240134,
      "learning_rate": 0.00014425886351065808,
      "loss": 0.5898,
      "step": 7506
    },
    {
      "epoch": 3.532705882352941,
      "grad_norm": 0.46999874419718457,
      "learning_rate": 0.00014421505969984346,
      "loss": 0.6211,
      "step": 7507
    },
    {
      "epoch": 3.533176470588235,
      "grad_norm": 0.5175258593609168,
      "learning_rate": 0.00014417125638309115,
      "loss": 0.7734,
      "step": 7508
    },
    {
      "epoch": 3.5336470588235294,
      "grad_norm": 0.21479442691779224,
      "learning_rate": 0.00014412745356414213,
      "loss": 0.0693,
      "step": 7509
    },
    {
      "epoch": 3.5341176470588236,
      "grad_norm": 0.32857532939819445,
      "learning_rate": 0.00014408365124673737,
      "loss": 0.3496,
      "step": 7510
    },
    {
      "epoch": 3.534588235294118,
      "grad_norm": 0.36706782867696824,
      "learning_rate": 0.00014403984943461783,
      "loss": 0.208,
      "step": 7511
    },
    {
      "epoch": 3.535058823529412,
      "grad_norm": 0.4816568310560391,
      "learning_rate": 0.0001439960481315244,
      "loss": 0.3359,
      "step": 7512
    },
    {
      "epoch": 3.535529411764706,
      "grad_norm": 0.272228449896402,
      "learning_rate": 0.00014395224734119788,
      "loss": 0.1533,
      "step": 7513
    },
    {
      "epoch": 3.536,
      "grad_norm": 0.2077461698350222,
      "learning_rate": 0.00014390844706737907,
      "loss": 0.0299,
      "step": 7514
    },
    {
      "epoch": 3.5364705882352943,
      "grad_norm": 0.49946054794417366,
      "learning_rate": 0.00014386464731380881,
      "loss": 0.2441,
      "step": 7515
    },
    {
      "epoch": 3.536941176470588,
      "grad_norm": 0.5467237012596602,
      "learning_rate": 0.00014382084808422775,
      "loss": 0.375,
      "step": 7516
    },
    {
      "epoch": 3.5374117647058823,
      "grad_norm": 0.4225586852442954,
      "learning_rate": 0.00014377704938237654,
      "loss": 0.4082,
      "step": 7517
    },
    {
      "epoch": 3.5378823529411765,
      "grad_norm": 0.6437584714648072,
      "learning_rate": 0.00014373325121199587,
      "loss": 0.5898,
      "step": 7518
    },
    {
      "epoch": 3.5383529411764707,
      "grad_norm": 0.2593810515422007,
      "learning_rate": 0.00014368945357682628,
      "loss": 0.1133,
      "step": 7519
    },
    {
      "epoch": 3.538823529411765,
      "grad_norm": 0.27421830652968887,
      "learning_rate": 0.0001436456564806083,
      "loss": 0.1289,
      "step": 7520
    },
    {
      "epoch": 3.5392941176470587,
      "grad_norm": 0.24183818663747716,
      "learning_rate": 0.00014360185992708243,
      "loss": 0.2119,
      "step": 7521
    },
    {
      "epoch": 3.539764705882353,
      "grad_norm": 0.3897043247200762,
      "learning_rate": 0.00014355806391998913,
      "loss": 0.3086,
      "step": 7522
    },
    {
      "epoch": 3.540235294117647,
      "grad_norm": 0.9297020488271235,
      "learning_rate": 0.00014351426846306875,
      "loss": 1.2734,
      "step": 7523
    },
    {
      "epoch": 3.540705882352941,
      "grad_norm": 0.2156648312465196,
      "learning_rate": 0.00014347047356006161,
      "loss": 0.2275,
      "step": 7524
    },
    {
      "epoch": 3.541176470588235,
      "grad_norm": 0.444207677612058,
      "learning_rate": 0.00014342667921470812,
      "loss": 0.4531,
      "step": 7525
    },
    {
      "epoch": 3.5416470588235294,
      "grad_norm": 0.4565925518933455,
      "learning_rate": 0.00014338288543074845,
      "loss": 0.5508,
      "step": 7526
    },
    {
      "epoch": 3.5421176470588236,
      "grad_norm": 0.285350645112774,
      "learning_rate": 0.00014333909221192278,
      "loss": 0.3262,
      "step": 7527
    },
    {
      "epoch": 3.542588235294118,
      "grad_norm": 0.3154122570222938,
      "learning_rate": 0.00014329529956197135,
      "loss": 0.2617,
      "step": 7528
    },
    {
      "epoch": 3.543058823529412,
      "grad_norm": 0.40495219077160605,
      "learning_rate": 0.00014325150748463423,
      "loss": 0.4121,
      "step": 7529
    },
    {
      "epoch": 3.543529411764706,
      "grad_norm": 0.34220729465942834,
      "learning_rate": 0.00014320771598365142,
      "loss": 0.4727,
      "step": 7530
    },
    {
      "epoch": 3.544,
      "grad_norm": 0.5281639996081727,
      "learning_rate": 0.00014316392506276302,
      "loss": 0.6133,
      "step": 7531
    },
    {
      "epoch": 3.5444705882352943,
      "grad_norm": 0.3008518665047457,
      "learning_rate": 0.0001431201347257089,
      "loss": 0.2139,
      "step": 7532
    },
    {
      "epoch": 3.544941176470588,
      "grad_norm": 0.5269560130180105,
      "learning_rate": 0.00014307634497622907,
      "loss": 0.3691,
      "step": 7533
    },
    {
      "epoch": 3.5454117647058823,
      "grad_norm": 0.3966200125054375,
      "learning_rate": 0.00014303255581806336,
      "loss": 0.3027,
      "step": 7534
    },
    {
      "epoch": 3.5458823529411765,
      "grad_norm": 0.3916453175734926,
      "learning_rate": 0.0001429887672549515,
      "loss": 0.2539,
      "step": 7535
    },
    {
      "epoch": 3.5463529411764707,
      "grad_norm": 0.6159427431448935,
      "learning_rate": 0.00014294497929063336,
      "loss": 0.6602,
      "step": 7536
    },
    {
      "epoch": 3.546823529411765,
      "grad_norm": 0.31338939691521844,
      "learning_rate": 0.0001429011919288486,
      "loss": 0.3926,
      "step": 7537
    },
    {
      "epoch": 3.5472941176470587,
      "grad_norm": 0.3440839103720374,
      "learning_rate": 0.0001428574051733368,
      "loss": 0.1865,
      "step": 7538
    },
    {
      "epoch": 3.547764705882353,
      "grad_norm": 0.30759520795671097,
      "learning_rate": 0.0001428136190278377,
      "loss": 0.1104,
      "step": 7539
    },
    {
      "epoch": 3.548235294117647,
      "grad_norm": 0.4275174413655653,
      "learning_rate": 0.00014276983349609075,
      "loss": 0.6094,
      "step": 7540
    },
    {
      "epoch": 3.548705882352941,
      "grad_norm": 0.5093575740850355,
      "learning_rate": 0.0001427260485818355,
      "loss": 0.6758,
      "step": 7541
    },
    {
      "epoch": 3.549176470588235,
      "grad_norm": 0.32563255314635814,
      "learning_rate": 0.00014268226428881136,
      "loss": 0.1108,
      "step": 7542
    },
    {
      "epoch": 3.5496470588235294,
      "grad_norm": 0.4169760923919175,
      "learning_rate": 0.00014263848062075776,
      "loss": 0.3477,
      "step": 7543
    },
    {
      "epoch": 3.5501176470588236,
      "grad_norm": 0.3270835803626343,
      "learning_rate": 0.00014259469758141403,
      "loss": 0.1738,
      "step": 7544
    },
    {
      "epoch": 3.550588235294118,
      "grad_norm": 0.3175584633453237,
      "learning_rate": 0.00014255091517451941,
      "loss": 0.2891,
      "step": 7545
    },
    {
      "epoch": 3.5510588235294116,
      "grad_norm": 0.269522853282329,
      "learning_rate": 0.0001425071334038132,
      "loss": 0.1143,
      "step": 7546
    },
    {
      "epoch": 3.551529411764706,
      "grad_norm": 0.4591752107069677,
      "learning_rate": 0.00014246335227303456,
      "loss": 0.4082,
      "step": 7547
    },
    {
      "epoch": 3.552,
      "grad_norm": 0.3298210429263618,
      "learning_rate": 0.00014241957178592258,
      "loss": 0.1338,
      "step": 7548
    },
    {
      "epoch": 3.5524705882352943,
      "grad_norm": 0.18391464965184653,
      "learning_rate": 0.00014237579194621637,
      "loss": 0.1035,
      "step": 7549
    },
    {
      "epoch": 3.552941176470588,
      "grad_norm": 0.236772669769894,
      "learning_rate": 0.00014233201275765494,
      "loss": 0.0977,
      "step": 7550
    },
    {
      "epoch": 3.5534117647058823,
      "grad_norm": 0.3388085587597329,
      "learning_rate": 0.0001422882342239772,
      "loss": 0.3965,
      "step": 7551
    },
    {
      "epoch": 3.5538823529411765,
      "grad_norm": 0.3750466399490247,
      "learning_rate": 0.00014224445634892206,
      "loss": 0.582,
      "step": 7552
    },
    {
      "epoch": 3.5543529411764707,
      "grad_norm": 0.504296252667378,
      "learning_rate": 0.00014220067913622844,
      "loss": 0.6836,
      "step": 7553
    },
    {
      "epoch": 3.554823529411765,
      "grad_norm": 0.4480742926580794,
      "learning_rate": 0.00014215690258963508,
      "loss": 0.4375,
      "step": 7554
    },
    {
      "epoch": 3.5552941176470587,
      "grad_norm": 0.3292401439936051,
      "learning_rate": 0.00014211312671288064,
      "loss": 0.1611,
      "step": 7555
    },
    {
      "epoch": 3.555764705882353,
      "grad_norm": 0.4266126487506681,
      "learning_rate": 0.00014206935150970392,
      "loss": 0.4961,
      "step": 7556
    },
    {
      "epoch": 3.556235294117647,
      "grad_norm": 0.43779785120941356,
      "learning_rate": 0.00014202557698384348,
      "loss": 0.5586,
      "step": 7557
    },
    {
      "epoch": 3.556705882352941,
      "grad_norm": 0.3360809442259622,
      "learning_rate": 0.00014198180313903784,
      "loss": 0.3379,
      "step": 7558
    },
    {
      "epoch": 3.557176470588235,
      "grad_norm": 0.3587899551681594,
      "learning_rate": 0.0001419380299790256,
      "loss": 0.2871,
      "step": 7559
    },
    {
      "epoch": 3.5576470588235294,
      "grad_norm": 0.4560502369091105,
      "learning_rate": 0.00014189425750754512,
      "loss": 0.5312,
      "step": 7560
    },
    {
      "epoch": 3.5581176470588236,
      "grad_norm": 0.39941181565514505,
      "learning_rate": 0.00014185048572833478,
      "loss": 0.3789,
      "step": 7561
    },
    {
      "epoch": 3.558588235294118,
      "grad_norm": 0.40477523833774565,
      "learning_rate": 0.00014180671464513296,
      "loss": 0.1875,
      "step": 7562
    },
    {
      "epoch": 3.5590588235294116,
      "grad_norm": 0.3567796901170971,
      "learning_rate": 0.0001417629442616779,
      "loss": 0.3984,
      "step": 7563
    },
    {
      "epoch": 3.559529411764706,
      "grad_norm": 0.29608572105692477,
      "learning_rate": 0.00014171917458170782,
      "loss": 0.2354,
      "step": 7564
    },
    {
      "epoch": 3.56,
      "grad_norm": 0.3783823700026084,
      "learning_rate": 0.0001416754056089608,
      "loss": 0.2021,
      "step": 7565
    },
    {
      "epoch": 3.5604705882352943,
      "grad_norm": 0.43905443938589495,
      "learning_rate": 0.000141631637347175,
      "loss": 0.2432,
      "step": 7566
    },
    {
      "epoch": 3.560941176470588,
      "grad_norm": 0.3347161614228062,
      "learning_rate": 0.0001415878698000884,
      "loss": 0.1201,
      "step": 7567
    },
    {
      "epoch": 3.5614117647058823,
      "grad_norm": 0.5289622791106274,
      "learning_rate": 0.00014154410297143898,
      "loss": 0.2227,
      "step": 7568
    },
    {
      "epoch": 3.5618823529411765,
      "grad_norm": 0.2580308247126082,
      "learning_rate": 0.00014150033686496469,
      "loss": 0.0952,
      "step": 7569
    },
    {
      "epoch": 3.5623529411764707,
      "grad_norm": 0.6833951363343981,
      "learning_rate": 0.0001414565714844033,
      "loss": 0.4531,
      "step": 7570
    },
    {
      "epoch": 3.562823529411765,
      "grad_norm": 0.43529217163026357,
      "learning_rate": 0.00014141280683349253,
      "loss": 0.3535,
      "step": 7571
    },
    {
      "epoch": 3.5632941176470587,
      "grad_norm": 0.10036767027431608,
      "learning_rate": 0.00014136904291597028,
      "loss": 0.0204,
      "step": 7572
    },
    {
      "epoch": 3.563764705882353,
      "grad_norm": 0.6087948484303144,
      "learning_rate": 0.00014132527973557407,
      "loss": 0.6406,
      "step": 7573
    },
    {
      "epoch": 3.564235294117647,
      "grad_norm": 0.34394339140316876,
      "learning_rate": 0.00014128151729604152,
      "loss": 0.1689,
      "step": 7574
    },
    {
      "epoch": 3.564705882352941,
      "grad_norm": 0.3163793263118033,
      "learning_rate": 0.00014123775560111009,
      "loss": 0.3184,
      "step": 7575
    },
    {
      "epoch": 3.565176470588235,
      "grad_norm": 0.47694775398948047,
      "learning_rate": 0.00014119399465451738,
      "loss": 0.3809,
      "step": 7576
    },
    {
      "epoch": 3.5656470588235294,
      "grad_norm": 0.3402789345697005,
      "learning_rate": 0.0001411502344600007,
      "loss": 0.1147,
      "step": 7577
    },
    {
      "epoch": 3.5661176470588236,
      "grad_norm": 0.4761258187419106,
      "learning_rate": 0.00014110647502129742,
      "loss": 0.459,
      "step": 7578
    },
    {
      "epoch": 3.566588235294118,
      "grad_norm": 0.5331671903570891,
      "learning_rate": 0.00014106271634214477,
      "loss": 0.5039,
      "step": 7579
    },
    {
      "epoch": 3.5670588235294116,
      "grad_norm": 0.36305738857004016,
      "learning_rate": 0.00014101895842628001,
      "loss": 0.2754,
      "step": 7580
    },
    {
      "epoch": 3.567529411764706,
      "grad_norm": 0.20239328295015713,
      "learning_rate": 0.00014097520127744024,
      "loss": 0.1582,
      "step": 7581
    },
    {
      "epoch": 3.568,
      "grad_norm": 0.31589038729922686,
      "learning_rate": 0.00014093144489936259,
      "loss": 0.0737,
      "step": 7582
    },
    {
      "epoch": 3.5684705882352943,
      "grad_norm": 0.4695671022054687,
      "learning_rate": 0.00014088768929578397,
      "loss": 0.5625,
      "step": 7583
    },
    {
      "epoch": 3.568941176470588,
      "grad_norm": 0.31609903801757483,
      "learning_rate": 0.00014084393447044138,
      "loss": 0.2363,
      "step": 7584
    },
    {
      "epoch": 3.5694117647058823,
      "grad_norm": 0.36770878731773915,
      "learning_rate": 0.00014080018042707173,
      "loss": 0.1455,
      "step": 7585
    },
    {
      "epoch": 3.5698823529411765,
      "grad_norm": 0.5649489045622266,
      "learning_rate": 0.00014075642716941176,
      "loss": 0.3691,
      "step": 7586
    },
    {
      "epoch": 3.5703529411764707,
      "grad_norm": 0.24763721285986381,
      "learning_rate": 0.0001407126747011983,
      "loss": 0.1758,
      "step": 7587
    },
    {
      "epoch": 3.570823529411765,
      "grad_norm": 0.7053901618564549,
      "learning_rate": 0.00014066892302616794,
      "loss": 0.459,
      "step": 7588
    },
    {
      "epoch": 3.5712941176470587,
      "grad_norm": 0.5274973807267324,
      "learning_rate": 0.0001406251721480573,
      "loss": 0.4453,
      "step": 7589
    },
    {
      "epoch": 3.571764705882353,
      "grad_norm": 0.5461155342360305,
      "learning_rate": 0.000140581422070603,
      "loss": 0.6641,
      "step": 7590
    },
    {
      "epoch": 3.572235294117647,
      "grad_norm": 0.4669143824369082,
      "learning_rate": 0.0001405376727975414,
      "loss": 0.3164,
      "step": 7591
    },
    {
      "epoch": 3.572705882352941,
      "grad_norm": 0.7066566500582802,
      "learning_rate": 0.000140493924332609,
      "loss": 0.7344,
      "step": 7592
    },
    {
      "epoch": 3.573176470588235,
      "grad_norm": 0.47436910479202243,
      "learning_rate": 0.00014045017667954203,
      "loss": 0.2676,
      "step": 7593
    },
    {
      "epoch": 3.5736470588235294,
      "grad_norm": 0.41765335768723666,
      "learning_rate": 0.00014040642984207687,
      "loss": 0.2471,
      "step": 7594
    },
    {
      "epoch": 3.5741176470588236,
      "grad_norm": 0.23707711250424213,
      "learning_rate": 0.00014036268382394964,
      "loss": 0.105,
      "step": 7595
    },
    {
      "epoch": 3.574588235294118,
      "grad_norm": 0.37011705642469334,
      "learning_rate": 0.00014031893862889646,
      "loss": 0.3672,
      "step": 7596
    },
    {
      "epoch": 3.5750588235294116,
      "grad_norm": 0.4198372118317309,
      "learning_rate": 0.00014027519426065344,
      "loss": 0.373,
      "step": 7597
    },
    {
      "epoch": 3.575529411764706,
      "grad_norm": 0.40081614821441314,
      "learning_rate": 0.00014023145072295654,
      "loss": 0.2451,
      "step": 7598
    },
    {
      "epoch": 3.576,
      "grad_norm": 0.5788131850659761,
      "learning_rate": 0.0001401877080195416,
      "loss": 0.5781,
      "step": 7599
    },
    {
      "epoch": 3.576470588235294,
      "grad_norm": 0.32243919521623715,
      "learning_rate": 0.00014014396615414458,
      "loss": 0.1357,
      "step": 7600
    },
    {
      "epoch": 3.576941176470588,
      "grad_norm": 0.46352149305870755,
      "learning_rate": 0.0001401002251305012,
      "loss": 0.3242,
      "step": 7601
    },
    {
      "epoch": 3.5774117647058823,
      "grad_norm": 0.42016821147095806,
      "learning_rate": 0.0001400564849523471,
      "loss": 0.4609,
      "step": 7602
    },
    {
      "epoch": 3.5778823529411765,
      "grad_norm": 0.2501601644478736,
      "learning_rate": 0.00014001274562341798,
      "loss": 0.0762,
      "step": 7603
    },
    {
      "epoch": 3.5783529411764707,
      "grad_norm": 0.4427966152709513,
      "learning_rate": 0.0001399690071474494,
      "loss": 0.3672,
      "step": 7604
    },
    {
      "epoch": 3.578823529411765,
      "grad_norm": 0.7193762439881424,
      "learning_rate": 0.0001399252695281768,
      "loss": 0.9766,
      "step": 7605
    },
    {
      "epoch": 3.5792941176470587,
      "grad_norm": 0.40083096581477506,
      "learning_rate": 0.00013988153276933555,
      "loss": 0.4629,
      "step": 7606
    },
    {
      "epoch": 3.579764705882353,
      "grad_norm": 0.5575116772978006,
      "learning_rate": 0.0001398377968746611,
      "loss": 0.7773,
      "step": 7607
    },
    {
      "epoch": 3.580235294117647,
      "grad_norm": 0.4450328551020761,
      "learning_rate": 0.0001397940618478886,
      "loss": 0.4609,
      "step": 7608
    },
    {
      "epoch": 3.580705882352941,
      "grad_norm": 0.5790102368360439,
      "learning_rate": 0.00013975032769275326,
      "loss": 0.6562,
      "step": 7609
    },
    {
      "epoch": 3.581176470588235,
      "grad_norm": 0.27363078643805555,
      "learning_rate": 0.00013970659441299026,
      "loss": 0.3242,
      "step": 7610
    },
    {
      "epoch": 3.5816470588235294,
      "grad_norm": 0.2621229296123265,
      "learning_rate": 0.00013966286201233457,
      "loss": 0.0767,
      "step": 7611
    },
    {
      "epoch": 3.5821176470588236,
      "grad_norm": 0.6552720886131695,
      "learning_rate": 0.00013961913049452115,
      "loss": 0.5,
      "step": 7612
    },
    {
      "epoch": 3.582588235294118,
      "grad_norm": 0.47078652826953016,
      "learning_rate": 0.00013957539986328492,
      "loss": 0.3047,
      "step": 7613
    },
    {
      "epoch": 3.5830588235294116,
      "grad_norm": 0.25393666344949856,
      "learning_rate": 0.00013953167012236068,
      "loss": 0.0864,
      "step": 7614
    },
    {
      "epoch": 3.583529411764706,
      "grad_norm": 0.4220863080479269,
      "learning_rate": 0.00013948794127548316,
      "loss": 0.3223,
      "step": 7615
    },
    {
      "epoch": 3.584,
      "grad_norm": 0.29089390230267614,
      "learning_rate": 0.00013944421332638698,
      "loss": 0.104,
      "step": 7616
    },
    {
      "epoch": 3.584470588235294,
      "grad_norm": 0.8403107700796236,
      "learning_rate": 0.0001394004862788068,
      "loss": 0.4629,
      "step": 7617
    },
    {
      "epoch": 3.584941176470588,
      "grad_norm": 0.2546757678693174,
      "learning_rate": 0.00013935676013647708,
      "loss": 0.0796,
      "step": 7618
    },
    {
      "epoch": 3.5854117647058823,
      "grad_norm": 0.41055104585096036,
      "learning_rate": 0.0001393130349031322,
      "loss": 0.4355,
      "step": 7619
    },
    {
      "epoch": 3.5858823529411765,
      "grad_norm": 0.5699545256955073,
      "learning_rate": 0.00013926931058250657,
      "loss": 0.5859,
      "step": 7620
    },
    {
      "epoch": 3.5863529411764707,
      "grad_norm": 0.1781045856519784,
      "learning_rate": 0.00013922558717833445,
      "loss": 0.0432,
      "step": 7621
    },
    {
      "epoch": 3.586823529411765,
      "grad_norm": 0.31773603027895525,
      "learning_rate": 0.00013918186469435005,
      "loss": 0.3262,
      "step": 7622
    },
    {
      "epoch": 3.5872941176470587,
      "grad_norm": 0.3669475961748643,
      "learning_rate": 0.00013913814313428748,
      "loss": 0.2061,
      "step": 7623
    },
    {
      "epoch": 3.587764705882353,
      "grad_norm": 0.4160133502998321,
      "learning_rate": 0.00013909442250188074,
      "loss": 0.5625,
      "step": 7624
    },
    {
      "epoch": 3.588235294117647,
      "grad_norm": 0.3745253193118116,
      "learning_rate": 0.00013905070280086386,
      "loss": 0.1953,
      "step": 7625
    },
    {
      "epoch": 3.588705882352941,
      "grad_norm": 0.5413151832952641,
      "learning_rate": 0.00013900698403497062,
      "loss": 0.4551,
      "step": 7626
    },
    {
      "epoch": 3.589176470588235,
      "grad_norm": 0.3187421413187184,
      "learning_rate": 0.00013896326620793488,
      "loss": 0.1885,
      "step": 7627
    },
    {
      "epoch": 3.5896470588235294,
      "grad_norm": 0.3608085649509762,
      "learning_rate": 0.00013891954932349033,
      "loss": 0.2139,
      "step": 7628
    },
    {
      "epoch": 3.5901176470588236,
      "grad_norm": 0.25608154770121294,
      "learning_rate": 0.00013887583338537064,
      "loss": 0.0771,
      "step": 7629
    },
    {
      "epoch": 3.590588235294118,
      "grad_norm": 0.3558705672595744,
      "learning_rate": 0.0001388321183973093,
      "loss": 0.2461,
      "step": 7630
    },
    {
      "epoch": 3.5910588235294116,
      "grad_norm": 0.29467665619974664,
      "learning_rate": 0.0001387884043630399,
      "loss": 0.103,
      "step": 7631
    },
    {
      "epoch": 3.591529411764706,
      "grad_norm": 0.45784620253322117,
      "learning_rate": 0.0001387446912862957,
      "loss": 0.291,
      "step": 7632
    },
    {
      "epoch": 3.592,
      "grad_norm": 0.23726612699151972,
      "learning_rate": 0.0001387009791708101,
      "loss": 0.0884,
      "step": 7633
    },
    {
      "epoch": 3.592470588235294,
      "grad_norm": 0.2723171584261378,
      "learning_rate": 0.00013865726802031627,
      "loss": 0.1367,
      "step": 7634
    },
    {
      "epoch": 3.592941176470588,
      "grad_norm": 0.6360179452916471,
      "learning_rate": 0.0001386135578385474,
      "loss": 0.4766,
      "step": 7635
    },
    {
      "epoch": 3.5934117647058823,
      "grad_norm": 0.2645023025570751,
      "learning_rate": 0.00013856984862923655,
      "loss": 0.1084,
      "step": 7636
    },
    {
      "epoch": 3.5938823529411765,
      "grad_norm": 0.6105790403091322,
      "learning_rate": 0.00013852614039611667,
      "loss": 0.75,
      "step": 7637
    },
    {
      "epoch": 3.5943529411764708,
      "grad_norm": 0.27366474522669015,
      "learning_rate": 0.0001384824331429207,
      "loss": 0.0874,
      "step": 7638
    },
    {
      "epoch": 3.594823529411765,
      "grad_norm": 0.559981516295656,
      "learning_rate": 0.00013843872687338142,
      "loss": 0.6953,
      "step": 7639
    },
    {
      "epoch": 3.5952941176470588,
      "grad_norm": 0.7493162571973496,
      "learning_rate": 0.00013839502159123158,
      "loss": 0.5547,
      "step": 7640
    },
    {
      "epoch": 3.595764705882353,
      "grad_norm": 0.45802519065752845,
      "learning_rate": 0.00013835131730020382,
      "loss": 0.4453,
      "step": 7641
    },
    {
      "epoch": 3.596235294117647,
      "grad_norm": 0.4731401205581557,
      "learning_rate": 0.0001383076140040307,
      "loss": 0.2598,
      "step": 7642
    },
    {
      "epoch": 3.596705882352941,
      "grad_norm": 0.5834273233237566,
      "learning_rate": 0.0001382639117064447,
      "loss": 0.4199,
      "step": 7643
    },
    {
      "epoch": 3.597176470588235,
      "grad_norm": 0.5462259462442168,
      "learning_rate": 0.0001382202104111782,
      "loss": 0.4648,
      "step": 7644
    },
    {
      "epoch": 3.5976470588235294,
      "grad_norm": 0.38463567798914816,
      "learning_rate": 0.0001381765101219635,
      "loss": 0.4668,
      "step": 7645
    },
    {
      "epoch": 3.5981176470588236,
      "grad_norm": 0.3669191949568673,
      "learning_rate": 0.00013813281084253286,
      "loss": 0.1621,
      "step": 7646
    },
    {
      "epoch": 3.598588235294118,
      "grad_norm": 0.3752321125832559,
      "learning_rate": 0.00013808911257661834,
      "loss": 0.3496,
      "step": 7647
    },
    {
      "epoch": 3.5990588235294116,
      "grad_norm": 1.0824537312179157,
      "learning_rate": 0.00013804541532795204,
      "loss": 0.9766,
      "step": 7648
    },
    {
      "epoch": 3.599529411764706,
      "grad_norm": 0.5587965025908851,
      "learning_rate": 0.00013800171910026594,
      "loss": 0.543,
      "step": 7649
    },
    {
      "epoch": 3.6,
      "grad_norm": 0.3415722689529945,
      "learning_rate": 0.00013795802389729184,
      "loss": 0.1348,
      "step": 7650
    },
    {
      "epoch": 3.600470588235294,
      "grad_norm": 0.28331154665827174,
      "learning_rate": 0.0001379143297227616,
      "loss": 0.377,
      "step": 7651
    },
    {
      "epoch": 3.600941176470588,
      "grad_norm": 0.5170815792856491,
      "learning_rate": 0.00013787063658040686,
      "loss": 0.5547,
      "step": 7652
    },
    {
      "epoch": 3.6014117647058823,
      "grad_norm": 0.3086781502040747,
      "learning_rate": 0.00013782694447395926,
      "loss": 0.1504,
      "step": 7653
    },
    {
      "epoch": 3.6018823529411765,
      "grad_norm": 0.39554394162399403,
      "learning_rate": 0.0001377832534071503,
      "loss": 0.4629,
      "step": 7654
    },
    {
      "epoch": 3.6023529411764708,
      "grad_norm": 0.23992034319430197,
      "learning_rate": 0.0001377395633837114,
      "loss": 0.1416,
      "step": 7655
    },
    {
      "epoch": 3.602823529411765,
      "grad_norm": 0.38408039342192446,
      "learning_rate": 0.000137695874407374,
      "loss": 0.3145,
      "step": 7656
    },
    {
      "epoch": 3.6032941176470588,
      "grad_norm": 0.4961024564976536,
      "learning_rate": 0.0001376521864818692,
      "loss": 0.582,
      "step": 7657
    },
    {
      "epoch": 3.603764705882353,
      "grad_norm": 1.0515997935781496,
      "learning_rate": 0.00013760849961092826,
      "loss": 0.7383,
      "step": 7658
    },
    {
      "epoch": 3.604235294117647,
      "grad_norm": 0.46975501531901215,
      "learning_rate": 0.00013756481379828225,
      "loss": 0.3691,
      "step": 7659
    },
    {
      "epoch": 3.604705882352941,
      "grad_norm": 0.35278547997989557,
      "learning_rate": 0.00013752112904766208,
      "loss": 0.2988,
      "step": 7660
    },
    {
      "epoch": 3.605176470588235,
      "grad_norm": 0.3667303311547629,
      "learning_rate": 0.00013747744536279878,
      "loss": 0.4316,
      "step": 7661
    },
    {
      "epoch": 3.6056470588235294,
      "grad_norm": 0.260014973575814,
      "learning_rate": 0.00013743376274742304,
      "loss": 0.1216,
      "step": 7662
    },
    {
      "epoch": 3.6061176470588236,
      "grad_norm": 0.39563068672591756,
      "learning_rate": 0.00013739008120526559,
      "loss": 0.4414,
      "step": 7663
    },
    {
      "epoch": 3.606588235294118,
      "grad_norm": 0.42649060826644375,
      "learning_rate": 0.00013734640074005702,
      "loss": 0.2021,
      "step": 7664
    },
    {
      "epoch": 3.6070588235294117,
      "grad_norm": 0.39495951057230316,
      "learning_rate": 0.0001373027213555279,
      "loss": 0.2373,
      "step": 7665
    },
    {
      "epoch": 3.607529411764706,
      "grad_norm": 0.3055996910225849,
      "learning_rate": 0.0001372590430554087,
      "loss": 0.3965,
      "step": 7666
    },
    {
      "epoch": 3.608,
      "grad_norm": 0.23339795263088947,
      "learning_rate": 0.0001372153658434297,
      "loss": 0.0981,
      "step": 7667
    },
    {
      "epoch": 3.608470588235294,
      "grad_norm": 0.2949453096323085,
      "learning_rate": 0.00013717168972332116,
      "loss": 0.0928,
      "step": 7668
    },
    {
      "epoch": 3.608941176470588,
      "grad_norm": 0.4068393914168415,
      "learning_rate": 0.00013712801469881324,
      "loss": 0.4316,
      "step": 7669
    },
    {
      "epoch": 3.6094117647058823,
      "grad_norm": 0.48381958714947454,
      "learning_rate": 0.00013708434077363603,
      "loss": 0.8242,
      "step": 7670
    },
    {
      "epoch": 3.6098823529411765,
      "grad_norm": 0.5218900392625612,
      "learning_rate": 0.00013704066795151946,
      "loss": 0.3965,
      "step": 7671
    },
    {
      "epoch": 3.6103529411764708,
      "grad_norm": 0.309564856333786,
      "learning_rate": 0.00013699699623619338,
      "loss": 0.2598,
      "step": 7672
    },
    {
      "epoch": 3.6108235294117645,
      "grad_norm": 0.4480412610607069,
      "learning_rate": 0.00013695332563138762,
      "loss": 0.4141,
      "step": 7673
    },
    {
      "epoch": 3.6112941176470588,
      "grad_norm": 0.466231636608734,
      "learning_rate": 0.00013690965614083185,
      "loss": 0.498,
      "step": 7674
    },
    {
      "epoch": 3.611764705882353,
      "grad_norm": 0.21283335073935522,
      "learning_rate": 0.00013686598776825563,
      "loss": 0.1167,
      "step": 7675
    },
    {
      "epoch": 3.612235294117647,
      "grad_norm": 0.25066307327599924,
      "learning_rate": 0.00013682232051738852,
      "loss": 0.0659,
      "step": 7676
    },
    {
      "epoch": 3.612705882352941,
      "grad_norm": 0.3611808626809454,
      "learning_rate": 0.00013677865439195986,
      "loss": 0.2354,
      "step": 7677
    },
    {
      "epoch": 3.613176470588235,
      "grad_norm": 0.24292298696270648,
      "learning_rate": 0.00013673498939569892,
      "loss": 0.085,
      "step": 7678
    },
    {
      "epoch": 3.6136470588235294,
      "grad_norm": 0.2915645486309518,
      "learning_rate": 0.000136691325532335,
      "loss": 0.1196,
      "step": 7679
    },
    {
      "epoch": 3.6141176470588237,
      "grad_norm": 0.5997254713631232,
      "learning_rate": 0.00013664766280559716,
      "loss": 0.5078,
      "step": 7680
    },
    {
      "epoch": 3.614588235294118,
      "grad_norm": 0.45165974237966344,
      "learning_rate": 0.00013660400121921434,
      "loss": 0.4316,
      "step": 7681
    },
    {
      "epoch": 3.6150588235294117,
      "grad_norm": 0.3710038006000516,
      "learning_rate": 0.00013656034077691562,
      "loss": 0.2393,
      "step": 7682
    },
    {
      "epoch": 3.615529411764706,
      "grad_norm": 0.25210429557770486,
      "learning_rate": 0.00013651668148242967,
      "loss": 0.085,
      "step": 7683
    },
    {
      "epoch": 3.616,
      "grad_norm": 0.4654756290817228,
      "learning_rate": 0.00013647302333948528,
      "loss": 0.5352,
      "step": 7684
    },
    {
      "epoch": 3.616470588235294,
      "grad_norm": 0.3273912299660429,
      "learning_rate": 0.000136429366351811,
      "loss": 0.2344,
      "step": 7685
    },
    {
      "epoch": 3.616941176470588,
      "grad_norm": 0.7012601914470311,
      "learning_rate": 0.00013638571052313543,
      "loss": 0.3867,
      "step": 7686
    },
    {
      "epoch": 3.6174117647058823,
      "grad_norm": 0.38513959558921496,
      "learning_rate": 0.00013634205585718698,
      "loss": 0.3145,
      "step": 7687
    },
    {
      "epoch": 3.6178823529411765,
      "grad_norm": 0.4386497304599927,
      "learning_rate": 0.00013629840235769388,
      "loss": 0.2432,
      "step": 7688
    },
    {
      "epoch": 3.6183529411764708,
      "grad_norm": 0.506099971437693,
      "learning_rate": 0.0001362547500283845,
      "loss": 0.4199,
      "step": 7689
    },
    {
      "epoch": 3.6188235294117646,
      "grad_norm": 0.36246967311061656,
      "learning_rate": 0.00013621109887298685,
      "loss": 0.4961,
      "step": 7690
    },
    {
      "epoch": 3.6192941176470588,
      "grad_norm": 0.45650286273538754,
      "learning_rate": 0.00013616744889522898,
      "loss": 0.543,
      "step": 7691
    },
    {
      "epoch": 3.619764705882353,
      "grad_norm": 0.45887454764421637,
      "learning_rate": 0.00013612380009883886,
      "loss": 0.6172,
      "step": 7692
    },
    {
      "epoch": 3.620235294117647,
      "grad_norm": 0.3910002041832349,
      "learning_rate": 0.00013608015248754425,
      "loss": 0.4551,
      "step": 7693
    },
    {
      "epoch": 3.620705882352941,
      "grad_norm": 0.483709095910911,
      "learning_rate": 0.00013603650606507293,
      "loss": 0.7578,
      "step": 7694
    },
    {
      "epoch": 3.621176470588235,
      "grad_norm": 0.5346121243915616,
      "learning_rate": 0.0001359928608351524,
      "loss": 0.5078,
      "step": 7695
    },
    {
      "epoch": 3.6216470588235294,
      "grad_norm": 0.2294526527506489,
      "learning_rate": 0.00013594921680151032,
      "loss": 0.0786,
      "step": 7696
    },
    {
      "epoch": 3.6221176470588237,
      "grad_norm": 0.4014673871007641,
      "learning_rate": 0.00013590557396787406,
      "loss": 0.2539,
      "step": 7697
    },
    {
      "epoch": 3.622588235294118,
      "grad_norm": 0.4845749900100622,
      "learning_rate": 0.00013586193233797086,
      "loss": 0.2559,
      "step": 7698
    },
    {
      "epoch": 3.6230588235294117,
      "grad_norm": 0.4408019996025883,
      "learning_rate": 0.00013581829191552802,
      "loss": 0.3164,
      "step": 7699
    },
    {
      "epoch": 3.623529411764706,
      "grad_norm": 0.4887713837649732,
      "learning_rate": 0.00013577465270427262,
      "loss": 0.5547,
      "step": 7700
    },
    {
      "epoch": 3.624,
      "grad_norm": 0.4629289304699556,
      "learning_rate": 0.00013573101470793161,
      "loss": 0.3301,
      "step": 7701
    },
    {
      "epoch": 3.624470588235294,
      "grad_norm": 0.5717719827842018,
      "learning_rate": 0.00013568737793023198,
      "loss": 0.6719,
      "step": 7702
    },
    {
      "epoch": 3.624941176470588,
      "grad_norm": 0.42880946657221364,
      "learning_rate": 0.00013564374237490047,
      "loss": 0.3477,
      "step": 7703
    },
    {
      "epoch": 3.6254117647058823,
      "grad_norm": 0.3169881108163279,
      "learning_rate": 0.0001356001080456638,
      "loss": 0.2734,
      "step": 7704
    },
    {
      "epoch": 3.6258823529411766,
      "grad_norm": 1.1368837390369426,
      "learning_rate": 0.00013555647494624846,
      "loss": 0.2578,
      "step": 7705
    },
    {
      "epoch": 3.626352941176471,
      "grad_norm": 0.37928343402211084,
      "learning_rate": 0.00013551284308038105,
      "loss": 0.3008,
      "step": 7706
    },
    {
      "epoch": 3.6268235294117646,
      "grad_norm": 0.4860493733592022,
      "learning_rate": 0.00013546921245178792,
      "loss": 0.1582,
      "step": 7707
    },
    {
      "epoch": 3.627294117647059,
      "grad_norm": 0.4219301052810458,
      "learning_rate": 0.00013542558306419526,
      "loss": 0.4277,
      "step": 7708
    },
    {
      "epoch": 3.627764705882353,
      "grad_norm": 0.4071504204343509,
      "learning_rate": 0.00013538195492132928,
      "loss": 0.2432,
      "step": 7709
    },
    {
      "epoch": 3.6282352941176472,
      "grad_norm": 0.3343235105118829,
      "learning_rate": 0.00013533832802691613,
      "loss": 0.1904,
      "step": 7710
    },
    {
      "epoch": 3.628705882352941,
      "grad_norm": 0.4553526219653225,
      "learning_rate": 0.00013529470238468166,
      "loss": 0.2812,
      "step": 7711
    },
    {
      "epoch": 3.6291764705882352,
      "grad_norm": 0.6951609748650084,
      "learning_rate": 0.00013525107799835172,
      "loss": 0.5742,
      "step": 7712
    },
    {
      "epoch": 3.6296470588235294,
      "grad_norm": 0.3288899178450636,
      "learning_rate": 0.00013520745487165207,
      "loss": 0.3926,
      "step": 7713
    },
    {
      "epoch": 3.6301176470588237,
      "grad_norm": 0.32610065066894084,
      "learning_rate": 0.00013516383300830835,
      "loss": 0.252,
      "step": 7714
    },
    {
      "epoch": 3.630588235294118,
      "grad_norm": 0.3826611782964099,
      "learning_rate": 0.00013512021241204604,
      "loss": 0.3691,
      "step": 7715
    },
    {
      "epoch": 3.6310588235294117,
      "grad_norm": 0.5483801101527979,
      "learning_rate": 0.00013507659308659053,
      "loss": 0.6602,
      "step": 7716
    },
    {
      "epoch": 3.631529411764706,
      "grad_norm": 0.2869486834573731,
      "learning_rate": 0.00013503297503566727,
      "loss": 0.1338,
      "step": 7717
    },
    {
      "epoch": 3.632,
      "grad_norm": 0.25050171782875696,
      "learning_rate": 0.00013498935826300132,
      "loss": 0.0771,
      "step": 7718
    },
    {
      "epoch": 3.632470588235294,
      "grad_norm": 0.5050177468965809,
      "learning_rate": 0.00013494574277231772,
      "loss": 0.3223,
      "step": 7719
    },
    {
      "epoch": 3.632941176470588,
      "grad_norm": 0.2696761694311814,
      "learning_rate": 0.00013490212856734162,
      "loss": 0.0986,
      "step": 7720
    },
    {
      "epoch": 3.6334117647058823,
      "grad_norm": 0.4026961991485009,
      "learning_rate": 0.00013485851565179778,
      "loss": 0.1562,
      "step": 7721
    },
    {
      "epoch": 3.6338823529411766,
      "grad_norm": 0.24281832305932072,
      "learning_rate": 0.00013481490402941098,
      "loss": 0.2383,
      "step": 7722
    },
    {
      "epoch": 3.634352941176471,
      "grad_norm": 0.3266057196379362,
      "learning_rate": 0.00013477129370390582,
      "loss": 0.1855,
      "step": 7723
    },
    {
      "epoch": 3.6348235294117646,
      "grad_norm": 0.22554243653022243,
      "learning_rate": 0.00013472768467900686,
      "loss": 0.166,
      "step": 7724
    },
    {
      "epoch": 3.635294117647059,
      "grad_norm": 0.4636553613462237,
      "learning_rate": 0.0001346840769584386,
      "loss": 0.2021,
      "step": 7725
    },
    {
      "epoch": 3.635764705882353,
      "grad_norm": 0.4067338421553674,
      "learning_rate": 0.00013464047054592525,
      "loss": 0.3633,
      "step": 7726
    },
    {
      "epoch": 3.636235294117647,
      "grad_norm": 0.45048987202002616,
      "learning_rate": 0.00013459686544519108,
      "loss": 0.2227,
      "step": 7727
    },
    {
      "epoch": 3.636705882352941,
      "grad_norm": 0.5342566700775413,
      "learning_rate": 0.00013455326165996015,
      "loss": 0.7148,
      "step": 7728
    },
    {
      "epoch": 3.6371764705882352,
      "grad_norm": 0.7725924698315604,
      "learning_rate": 0.0001345096591939564,
      "loss": 0.8398,
      "step": 7729
    },
    {
      "epoch": 3.6376470588235295,
      "grad_norm": 0.4139760492067676,
      "learning_rate": 0.00013446605805090376,
      "loss": 0.4023,
      "step": 7730
    },
    {
      "epoch": 3.6381176470588237,
      "grad_norm": 0.4263484108945411,
      "learning_rate": 0.000134422458234526,
      "loss": 0.5977,
      "step": 7731
    },
    {
      "epoch": 3.638588235294118,
      "grad_norm": 0.41264895875280905,
      "learning_rate": 0.00013437885974854665,
      "loss": 0.4609,
      "step": 7732
    },
    {
      "epoch": 3.6390588235294117,
      "grad_norm": 0.45607677157709003,
      "learning_rate": 0.00013433526259668931,
      "loss": 0.6445,
      "step": 7733
    },
    {
      "epoch": 3.639529411764706,
      "grad_norm": 0.26804663716941285,
      "learning_rate": 0.0001342916667826774,
      "loss": 0.125,
      "step": 7734
    },
    {
      "epoch": 3.64,
      "grad_norm": 0.2055422034149379,
      "learning_rate": 0.0001342480723102342,
      "loss": 0.0591,
      "step": 7735
    },
    {
      "epoch": 3.640470588235294,
      "grad_norm": 0.3474663863277537,
      "learning_rate": 0.00013420447918308285,
      "loss": 0.2656,
      "step": 7736
    },
    {
      "epoch": 3.640941176470588,
      "grad_norm": 0.4314795229046978,
      "learning_rate": 0.00013416088740494649,
      "loss": 0.4336,
      "step": 7737
    },
    {
      "epoch": 3.6414117647058823,
      "grad_norm": 0.5031452570015281,
      "learning_rate": 0.00013411729697954803,
      "loss": 0.8047,
      "step": 7738
    },
    {
      "epoch": 3.6418823529411766,
      "grad_norm": 0.4586985654440458,
      "learning_rate": 0.00013407370791061027,
      "loss": 0.3125,
      "step": 7739
    },
    {
      "epoch": 3.642352941176471,
      "grad_norm": 0.4922470699022768,
      "learning_rate": 0.00013403012020185603,
      "loss": 0.5664,
      "step": 7740
    },
    {
      "epoch": 3.6428235294117646,
      "grad_norm": 0.3916793574315279,
      "learning_rate": 0.00013398653385700784,
      "loss": 0.5547,
      "step": 7741
    },
    {
      "epoch": 3.643294117647059,
      "grad_norm": 0.6040521579712839,
      "learning_rate": 0.00013394294887978818,
      "loss": 0.8086,
      "step": 7742
    },
    {
      "epoch": 3.643764705882353,
      "grad_norm": 0.3017154938772682,
      "learning_rate": 0.0001338993652739195,
      "loss": 0.0908,
      "step": 7743
    },
    {
      "epoch": 3.644235294117647,
      "grad_norm": 0.3930531513187005,
      "learning_rate": 0.00013385578304312397,
      "loss": 0.543,
      "step": 7744
    },
    {
      "epoch": 3.644705882352941,
      "grad_norm": 0.4992610497296803,
      "learning_rate": 0.00013381220219112373,
      "loss": 0.5352,
      "step": 7745
    },
    {
      "epoch": 3.6451764705882352,
      "grad_norm": 0.544415480699547,
      "learning_rate": 0.00013376862272164084,
      "loss": 0.6484,
      "step": 7746
    },
    {
      "epoch": 3.6456470588235295,
      "grad_norm": 0.42735537957791886,
      "learning_rate": 0.00013372504463839718,
      "loss": 0.1895,
      "step": 7747
    },
    {
      "epoch": 3.6461176470588237,
      "grad_norm": 0.3543053479162226,
      "learning_rate": 0.00013368146794511457,
      "loss": 0.0747,
      "step": 7748
    },
    {
      "epoch": 3.646588235294118,
      "grad_norm": 0.6889484614618265,
      "learning_rate": 0.0001336378926455146,
      "loss": 0.8672,
      "step": 7749
    },
    {
      "epoch": 3.6470588235294117,
      "grad_norm": 0.408513938898098,
      "learning_rate": 0.00013359431874331886,
      "loss": 0.4668,
      "step": 7750
    },
    {
      "epoch": 3.647529411764706,
      "grad_norm": 0.7820683124967769,
      "learning_rate": 0.0001335507462422488,
      "loss": 0.6016,
      "step": 7751
    },
    {
      "epoch": 3.648,
      "grad_norm": 0.4466056534455818,
      "learning_rate": 0.00013350717514602563,
      "loss": 0.5,
      "step": 7752
    },
    {
      "epoch": 3.648470588235294,
      "grad_norm": 0.33911417514529246,
      "learning_rate": 0.00013346360545837062,
      "loss": 0.1377,
      "step": 7753
    },
    {
      "epoch": 3.648941176470588,
      "grad_norm": 0.2899944737106889,
      "learning_rate": 0.0001334200371830048,
      "loss": 0.1006,
      "step": 7754
    },
    {
      "epoch": 3.6494117647058824,
      "grad_norm": 1.1620281372418193,
      "learning_rate": 0.00013337647032364919,
      "loss": 0.5039,
      "step": 7755
    },
    {
      "epoch": 3.6498823529411766,
      "grad_norm": 0.4130573290717896,
      "learning_rate": 0.0001333329048840245,
      "loss": 0.5977,
      "step": 7756
    },
    {
      "epoch": 3.650352941176471,
      "grad_norm": 0.4088611551257631,
      "learning_rate": 0.0001332893408678515,
      "loss": 0.332,
      "step": 7757
    },
    {
      "epoch": 3.6508235294117646,
      "grad_norm": 0.315932967361672,
      "learning_rate": 0.00013324577827885078,
      "loss": 0.1245,
      "step": 7758
    },
    {
      "epoch": 3.651294117647059,
      "grad_norm": 0.4172418742682314,
      "learning_rate": 0.00013320221712074276,
      "loss": 0.6523,
      "step": 7759
    },
    {
      "epoch": 3.651764705882353,
      "grad_norm": 0.24536298869222406,
      "learning_rate": 0.00013315865739724778,
      "loss": 0.2695,
      "step": 7760
    },
    {
      "epoch": 3.652235294117647,
      "grad_norm": 0.3607583454101557,
      "learning_rate": 0.00013311509911208608,
      "loss": 0.1709,
      "step": 7761
    },
    {
      "epoch": 3.652705882352941,
      "grad_norm": 2.2139050883506823,
      "learning_rate": 0.00013307154226897773,
      "loss": 0.6367,
      "step": 7762
    },
    {
      "epoch": 3.6531764705882352,
      "grad_norm": 0.5465299943983983,
      "learning_rate": 0.0001330279868716427,
      "loss": 0.4062,
      "step": 7763
    },
    {
      "epoch": 3.6536470588235295,
      "grad_norm": 0.5421701922763061,
      "learning_rate": 0.00013298443292380084,
      "loss": 0.5781,
      "step": 7764
    },
    {
      "epoch": 3.6541176470588237,
      "grad_norm": 0.36509694915457663,
      "learning_rate": 0.00013294088042917188,
      "loss": 0.2109,
      "step": 7765
    },
    {
      "epoch": 3.654588235294118,
      "grad_norm": 0.5366384381105758,
      "learning_rate": 0.0001328973293914754,
      "loss": 0.5078,
      "step": 7766
    },
    {
      "epoch": 3.6550588235294117,
      "grad_norm": 0.42943791078679455,
      "learning_rate": 0.00013285377981443085,
      "loss": 0.3984,
      "step": 7767
    },
    {
      "epoch": 3.655529411764706,
      "grad_norm": 0.3534700040755129,
      "learning_rate": 0.00013281023170175766,
      "loss": 0.3301,
      "step": 7768
    },
    {
      "epoch": 3.656,
      "grad_norm": 0.27220758248526467,
      "learning_rate": 0.00013276668505717498,
      "loss": 0.0933,
      "step": 7769
    },
    {
      "epoch": 3.656470588235294,
      "grad_norm": 0.38065130698426497,
      "learning_rate": 0.00013272313988440187,
      "loss": 0.2373,
      "step": 7770
    },
    {
      "epoch": 3.656941176470588,
      "grad_norm": 0.35698387483500976,
      "learning_rate": 0.0001326795961871574,
      "loss": 0.1094,
      "step": 7771
    },
    {
      "epoch": 3.6574117647058824,
      "grad_norm": 0.1936880879275678,
      "learning_rate": 0.00013263605396916038,
      "loss": 0.1118,
      "step": 7772
    },
    {
      "epoch": 3.6578823529411766,
      "grad_norm": 0.6452043581738993,
      "learning_rate": 0.00013259251323412952,
      "loss": 0.4414,
      "step": 7773
    },
    {
      "epoch": 3.658352941176471,
      "grad_norm": 0.3078474253095656,
      "learning_rate": 0.0001325489739857834,
      "loss": 0.1787,
      "step": 7774
    },
    {
      "epoch": 3.6588235294117646,
      "grad_norm": 0.25044741997965586,
      "learning_rate": 0.0001325054362278405,
      "loss": 0.0537,
      "step": 7775
    },
    {
      "epoch": 3.659294117647059,
      "grad_norm": 0.2090598083124891,
      "learning_rate": 0.00013246189996401916,
      "loss": 0.1484,
      "step": 7776
    },
    {
      "epoch": 3.659764705882353,
      "grad_norm": 0.42682464663417713,
      "learning_rate": 0.00013241836519803756,
      "loss": 0.5352,
      "step": 7777
    },
    {
      "epoch": 3.660235294117647,
      "grad_norm": 0.40922519550729947,
      "learning_rate": 0.00013237483193361387,
      "loss": 0.4648,
      "step": 7778
    },
    {
      "epoch": 3.660705882352941,
      "grad_norm": 0.5453545162520438,
      "learning_rate": 0.00013233130017446595,
      "loss": 0.498,
      "step": 7779
    },
    {
      "epoch": 3.6611764705882353,
      "grad_norm": 0.259918083944239,
      "learning_rate": 0.00013228776992431165,
      "loss": 0.1934,
      "step": 7780
    },
    {
      "epoch": 3.6616470588235295,
      "grad_norm": 0.203642420227595,
      "learning_rate": 0.00013224424118686868,
      "loss": 0.1592,
      "step": 7781
    },
    {
      "epoch": 3.6621176470588237,
      "grad_norm": 0.23241110191938114,
      "learning_rate": 0.00013220071396585463,
      "loss": 0.1021,
      "step": 7782
    },
    {
      "epoch": 3.662588235294118,
      "grad_norm": 0.5989647343630099,
      "learning_rate": 0.0001321571882649869,
      "loss": 0.6875,
      "step": 7783
    },
    {
      "epoch": 3.6630588235294117,
      "grad_norm": 0.2515233075067577,
      "learning_rate": 0.00013211366408798277,
      "loss": 0.0786,
      "step": 7784
    },
    {
      "epoch": 3.663529411764706,
      "grad_norm": 0.5124474794185103,
      "learning_rate": 0.00013207014143855951,
      "loss": 0.6797,
      "step": 7785
    },
    {
      "epoch": 3.664,
      "grad_norm": 0.3617911792948004,
      "learning_rate": 0.00013202662032043408,
      "loss": 0.2695,
      "step": 7786
    },
    {
      "epoch": 3.664470588235294,
      "grad_norm": 0.40672854972571754,
      "learning_rate": 0.00013198310073732344,
      "loss": 0.5,
      "step": 7787
    },
    {
      "epoch": 3.664941176470588,
      "grad_norm": 0.413478948331357,
      "learning_rate": 0.0001319395826929444,
      "loss": 0.2188,
      "step": 7788
    },
    {
      "epoch": 3.6654117647058824,
      "grad_norm": 0.4625245862478023,
      "learning_rate": 0.00013189606619101358,
      "loss": 0.2656,
      "step": 7789
    },
    {
      "epoch": 3.6658823529411766,
      "grad_norm": 0.4690333476139545,
      "learning_rate": 0.00013185255123524745,
      "loss": 0.2168,
      "step": 7790
    },
    {
      "epoch": 3.666352941176471,
      "grad_norm": 0.29051184755824894,
      "learning_rate": 0.00013180903782936254,
      "loss": 0.209,
      "step": 7791
    },
    {
      "epoch": 3.6668235294117646,
      "grad_norm": 0.5525404071055371,
      "learning_rate": 0.000131765525977075,
      "loss": 0.5195,
      "step": 7792
    },
    {
      "epoch": 3.667294117647059,
      "grad_norm": 0.3839243432027879,
      "learning_rate": 0.0001317220156821009,
      "loss": 0.3105,
      "step": 7793
    },
    {
      "epoch": 3.667764705882353,
      "grad_norm": 0.2334597019762303,
      "learning_rate": 0.00013167850694815642,
      "loss": 0.1328,
      "step": 7794
    },
    {
      "epoch": 3.668235294117647,
      "grad_norm": 0.3376107783096432,
      "learning_rate": 0.0001316349997789573,
      "loss": 0.1289,
      "step": 7795
    },
    {
      "epoch": 3.668705882352941,
      "grad_norm": 0.5622285845333408,
      "learning_rate": 0.00013159149417821923,
      "loss": 0.3398,
      "step": 7796
    },
    {
      "epoch": 3.6691764705882353,
      "grad_norm": 0.5295106260168971,
      "learning_rate": 0.00013154799014965785,
      "loss": 0.3438,
      "step": 7797
    },
    {
      "epoch": 3.6696470588235295,
      "grad_norm": 0.2931378767249845,
      "learning_rate": 0.00013150448769698857,
      "loss": 0.2734,
      "step": 7798
    },
    {
      "epoch": 3.6701176470588237,
      "grad_norm": 0.5037139185686614,
      "learning_rate": 0.00013146098682392685,
      "loss": 0.6953,
      "step": 7799
    },
    {
      "epoch": 3.6705882352941175,
      "grad_norm": 0.5562391336309144,
      "learning_rate": 0.0001314174875341878,
      "loss": 0.4531,
      "step": 7800
    },
    {
      "epoch": 3.6710588235294117,
      "grad_norm": 0.3351296059776108,
      "learning_rate": 0.00013137398983148642,
      "loss": 0.3457,
      "step": 7801
    },
    {
      "epoch": 3.671529411764706,
      "grad_norm": 0.4577068298088916,
      "learning_rate": 0.00013133049371953767,
      "loss": 0.6172,
      "step": 7802
    },
    {
      "epoch": 3.672,
      "grad_norm": 0.31453342036655485,
      "learning_rate": 0.00013128699920205634,
      "loss": 0.1147,
      "step": 7803
    },
    {
      "epoch": 3.672470588235294,
      "grad_norm": 0.5342283155091991,
      "learning_rate": 0.00013124350628275708,
      "loss": 0.4688,
      "step": 7804
    },
    {
      "epoch": 3.672941176470588,
      "grad_norm": 0.2933975013101691,
      "learning_rate": 0.00013120001496535433,
      "loss": 0.1177,
      "step": 7805
    },
    {
      "epoch": 3.6734117647058824,
      "grad_norm": 0.5617173533643947,
      "learning_rate": 0.00013115652525356257,
      "loss": 0.8086,
      "step": 7806
    },
    {
      "epoch": 3.6738823529411766,
      "grad_norm": 0.35816027709033604,
      "learning_rate": 0.00013111303715109598,
      "loss": 0.4766,
      "step": 7807
    },
    {
      "epoch": 3.674352941176471,
      "grad_norm": 0.6444448250994887,
      "learning_rate": 0.0001310695506616686,
      "loss": 0.5469,
      "step": 7808
    },
    {
      "epoch": 3.6748235294117646,
      "grad_norm": 0.38762260612766963,
      "learning_rate": 0.0001310260657889945,
      "loss": 0.3242,
      "step": 7809
    },
    {
      "epoch": 3.675294117647059,
      "grad_norm": 0.4728977369146051,
      "learning_rate": 0.00013098258253678744,
      "loss": 0.3066,
      "step": 7810
    },
    {
      "epoch": 3.675764705882353,
      "grad_norm": 0.45278306416977976,
      "learning_rate": 0.00013093910090876108,
      "loss": 0.5352,
      "step": 7811
    },
    {
      "epoch": 3.676235294117647,
      "grad_norm": 0.4862215815497054,
      "learning_rate": 0.00013089562090862905,
      "loss": 0.7812,
      "step": 7812
    },
    {
      "epoch": 3.676705882352941,
      "grad_norm": 1.6786341239140807,
      "learning_rate": 0.00013085214254010465,
      "loss": 1.4375,
      "step": 7813
    },
    {
      "epoch": 3.6771764705882353,
      "grad_norm": 0.280936030000288,
      "learning_rate": 0.00013080866580690126,
      "loss": 0.2617,
      "step": 7814
    },
    {
      "epoch": 3.6776470588235295,
      "grad_norm": 0.2882636795087926,
      "learning_rate": 0.00013076519071273187,
      "loss": 0.0864,
      "step": 7815
    },
    {
      "epoch": 3.6781176470588237,
      "grad_norm": 0.5723971295683251,
      "learning_rate": 0.00013072171726130957,
      "loss": 0.4785,
      "step": 7816
    },
    {
      "epoch": 3.6785882352941175,
      "grad_norm": 0.34882540725546846,
      "learning_rate": 0.00013067824545634718,
      "loss": 0.2871,
      "step": 7817
    },
    {
      "epoch": 3.6790588235294117,
      "grad_norm": 0.3766291479715739,
      "learning_rate": 0.00013063477530155738,
      "loss": 0.2539,
      "step": 7818
    },
    {
      "epoch": 3.679529411764706,
      "grad_norm": 0.25701807934806625,
      "learning_rate": 0.00013059130680065278,
      "loss": 0.1016,
      "step": 7819
    },
    {
      "epoch": 3.68,
      "grad_norm": 0.4055480040736627,
      "learning_rate": 0.0001305478399573458,
      "loss": 0.6211,
      "step": 7820
    },
    {
      "epoch": 3.680470588235294,
      "grad_norm": 0.4560391708568458,
      "learning_rate": 0.00013050437477534862,
      "loss": 0.5117,
      "step": 7821
    },
    {
      "epoch": 3.680941176470588,
      "grad_norm": 0.5614204167510389,
      "learning_rate": 0.00013046091125837352,
      "loss": 0.5742,
      "step": 7822
    },
    {
      "epoch": 3.6814117647058824,
      "grad_norm": 0.3934283983467202,
      "learning_rate": 0.00013041744941013245,
      "loss": 0.4316,
      "step": 7823
    },
    {
      "epoch": 3.6818823529411766,
      "grad_norm": 0.3667855888736707,
      "learning_rate": 0.00013037398923433725,
      "loss": 0.2236,
      "step": 7824
    },
    {
      "epoch": 3.682352941176471,
      "grad_norm": 0.3458615282933873,
      "learning_rate": 0.00013033053073469958,
      "loss": 0.1484,
      "step": 7825
    },
    {
      "epoch": 3.6828235294117646,
      "grad_norm": 0.49979725944379333,
      "learning_rate": 0.00013028707391493112,
      "loss": 0.2227,
      "step": 7826
    },
    {
      "epoch": 3.683294117647059,
      "grad_norm": 0.5455510388259939,
      "learning_rate": 0.00013024361877874324,
      "loss": 0.3926,
      "step": 7827
    },
    {
      "epoch": 3.683764705882353,
      "grad_norm": 0.5199707866656595,
      "learning_rate": 0.00013020016532984714,
      "loss": 0.4238,
      "step": 7828
    },
    {
      "epoch": 3.684235294117647,
      "grad_norm": 0.503663081611354,
      "learning_rate": 0.00013015671357195413,
      "loss": 0.5469,
      "step": 7829
    },
    {
      "epoch": 3.684705882352941,
      "grad_norm": 0.34700306103151507,
      "learning_rate": 0.00013011326350877508,
      "loss": 0.1758,
      "step": 7830
    },
    {
      "epoch": 3.6851764705882353,
      "grad_norm": 0.28014147361849895,
      "learning_rate": 0.00013006981514402085,
      "loss": 0.2119,
      "step": 7831
    },
    {
      "epoch": 3.6856470588235295,
      "grad_norm": 0.3216437947007878,
      "learning_rate": 0.00013002636848140217,
      "loss": 0.1699,
      "step": 7832
    },
    {
      "epoch": 3.6861176470588237,
      "grad_norm": 0.5667521630642107,
      "learning_rate": 0.00012998292352462963,
      "loss": 1.0,
      "step": 7833
    },
    {
      "epoch": 3.6865882352941175,
      "grad_norm": 0.45318364428234087,
      "learning_rate": 0.00012993948027741357,
      "loss": 0.4648,
      "step": 7834
    },
    {
      "epoch": 3.6870588235294117,
      "grad_norm": 0.3203152634640447,
      "learning_rate": 0.0001298960387434643,
      "loss": 0.2598,
      "step": 7835
    },
    {
      "epoch": 3.687529411764706,
      "grad_norm": 0.658317258669805,
      "learning_rate": 0.00012985259892649191,
      "loss": 0.9531,
      "step": 7836
    },
    {
      "epoch": 3.6879999999999997,
      "grad_norm": 0.26484385325830384,
      "learning_rate": 0.00012980916083020646,
      "loss": 0.0757,
      "step": 7837
    },
    {
      "epoch": 3.688470588235294,
      "grad_norm": 4.345256487929026,
      "learning_rate": 0.0001297657244583176,
      "loss": 2.2656,
      "step": 7838
    },
    {
      "epoch": 3.688941176470588,
      "grad_norm": 0.5415710126157777,
      "learning_rate": 0.00012972228981453522,
      "loss": 0.4004,
      "step": 7839
    },
    {
      "epoch": 3.6894117647058824,
      "grad_norm": 0.36276454880650316,
      "learning_rate": 0.00012967885690256872,
      "loss": 0.334,
      "step": 7840
    },
    {
      "epoch": 3.6898823529411766,
      "grad_norm": 0.3135232249427704,
      "learning_rate": 0.0001296354257261275,
      "loss": 0.1235,
      "step": 7841
    },
    {
      "epoch": 3.690352941176471,
      "grad_norm": 0.4296759685648792,
      "learning_rate": 0.0001295919962889208,
      "loss": 0.3555,
      "step": 7842
    },
    {
      "epoch": 3.6908235294117646,
      "grad_norm": 0.32976126364316355,
      "learning_rate": 0.0001295485685946577,
      "loss": 0.124,
      "step": 7843
    },
    {
      "epoch": 3.691294117647059,
      "grad_norm": 0.37627085559136847,
      "learning_rate": 0.0001295051426470472,
      "loss": 0.1846,
      "step": 7844
    },
    {
      "epoch": 3.691764705882353,
      "grad_norm": 0.5997257731383632,
      "learning_rate": 0.00012946171844979804,
      "loss": 0.4043,
      "step": 7845
    },
    {
      "epoch": 3.692235294117647,
      "grad_norm": 0.3267379936202903,
      "learning_rate": 0.00012941829600661886,
      "loss": 0.3281,
      "step": 7846
    },
    {
      "epoch": 3.692705882352941,
      "grad_norm": 0.4218013873648513,
      "learning_rate": 0.00012937487532121818,
      "loss": 0.4746,
      "step": 7847
    },
    {
      "epoch": 3.6931764705882353,
      "grad_norm": 0.3544889116299087,
      "learning_rate": 0.00012933145639730428,
      "loss": 0.25,
      "step": 7848
    },
    {
      "epoch": 3.6936470588235295,
      "grad_norm": 0.6784070347699799,
      "learning_rate": 0.00012928803923858536,
      "loss": 0.8359,
      "step": 7849
    },
    {
      "epoch": 3.6941176470588237,
      "grad_norm": 0.26934098027575315,
      "learning_rate": 0.00012924462384876953,
      "loss": 0.124,
      "step": 7850
    },
    {
      "epoch": 3.6945882352941175,
      "grad_norm": 0.46261162854021326,
      "learning_rate": 0.00012920121023156462,
      "loss": 0.5234,
      "step": 7851
    },
    {
      "epoch": 3.6950588235294117,
      "grad_norm": 0.37066676145329547,
      "learning_rate": 0.00012915779839067838,
      "loss": 0.373,
      "step": 7852
    },
    {
      "epoch": 3.695529411764706,
      "grad_norm": 0.2745487832366932,
      "learning_rate": 0.00012911438832981836,
      "loss": 0.1099,
      "step": 7853
    },
    {
      "epoch": 3.6959999999999997,
      "grad_norm": 0.2846497982209928,
      "learning_rate": 0.00012907098005269203,
      "loss": 0.0889,
      "step": 7854
    },
    {
      "epoch": 3.696470588235294,
      "grad_norm": 0.541653328728903,
      "learning_rate": 0.00012902757356300667,
      "loss": 0.3438,
      "step": 7855
    },
    {
      "epoch": 3.696941176470588,
      "grad_norm": 0.2973361961316328,
      "learning_rate": 0.00012898416886446937,
      "loss": 0.2852,
      "step": 7856
    },
    {
      "epoch": 3.6974117647058824,
      "grad_norm": 0.5852887015704363,
      "learning_rate": 0.00012894076596078717,
      "loss": 0.8047,
      "step": 7857
    },
    {
      "epoch": 3.6978823529411766,
      "grad_norm": 0.4500432813342,
      "learning_rate": 0.00012889736485566683,
      "loss": 0.2217,
      "step": 7858
    },
    {
      "epoch": 3.698352941176471,
      "grad_norm": 0.40672960485898724,
      "learning_rate": 0.000128853965552815,
      "loss": 0.3008,
      "step": 7859
    },
    {
      "epoch": 3.6988235294117646,
      "grad_norm": 0.5515247524946336,
      "learning_rate": 0.0001288105680559383,
      "loss": 0.6445,
      "step": 7860
    },
    {
      "epoch": 3.699294117647059,
      "grad_norm": 0.42641304384465256,
      "learning_rate": 0.000128767172368743,
      "loss": 0.2656,
      "step": 7861
    },
    {
      "epoch": 3.699764705882353,
      "grad_norm": 0.19337910221989685,
      "learning_rate": 0.0001287237784949353,
      "loss": 0.0603,
      "step": 7862
    },
    {
      "epoch": 3.700235294117647,
      "grad_norm": 0.4647704180245083,
      "learning_rate": 0.0001286803864382213,
      "loss": 0.2832,
      "step": 7863
    },
    {
      "epoch": 3.700705882352941,
      "grad_norm": 0.2709067201786274,
      "learning_rate": 0.00012863699620230688,
      "loss": 0.0869,
      "step": 7864
    },
    {
      "epoch": 3.7011764705882353,
      "grad_norm": 0.37572479670767,
      "learning_rate": 0.00012859360779089775,
      "loss": 0.3301,
      "step": 7865
    },
    {
      "epoch": 3.7016470588235295,
      "grad_norm": 0.46025076189249087,
      "learning_rate": 0.00012855022120769952,
      "loss": 0.5938,
      "step": 7866
    },
    {
      "epoch": 3.7021176470588237,
      "grad_norm": 0.5548536154648639,
      "learning_rate": 0.0001285068364564176,
      "loss": 0.4551,
      "step": 7867
    },
    {
      "epoch": 3.7025882352941175,
      "grad_norm": 0.30445342132842634,
      "learning_rate": 0.0001284634535407573,
      "loss": 0.1475,
      "step": 7868
    },
    {
      "epoch": 3.7030588235294117,
      "grad_norm": 0.33226577391397716,
      "learning_rate": 0.00012842007246442367,
      "loss": 0.4414,
      "step": 7869
    },
    {
      "epoch": 3.703529411764706,
      "grad_norm": 0.44224024767352466,
      "learning_rate": 0.0001283766932311217,
      "loss": 0.5117,
      "step": 7870
    },
    {
      "epoch": 3.7039999999999997,
      "grad_norm": 0.2649308263597501,
      "learning_rate": 0.00012833331584455622,
      "loss": 0.1357,
      "step": 7871
    },
    {
      "epoch": 3.704470588235294,
      "grad_norm": 0.3946420901320758,
      "learning_rate": 0.0001282899403084318,
      "loss": 0.5781,
      "step": 7872
    },
    {
      "epoch": 3.704941176470588,
      "grad_norm": 0.3707157763264672,
      "learning_rate": 0.000128246566626453,
      "loss": 0.3809,
      "step": 7873
    },
    {
      "epoch": 3.7054117647058824,
      "grad_norm": 0.46792933298865536,
      "learning_rate": 0.0001282031948023241,
      "loss": 0.6211,
      "step": 7874
    },
    {
      "epoch": 3.7058823529411766,
      "grad_norm": 0.3485546736437897,
      "learning_rate": 0.0001281598248397493,
      "loss": 0.209,
      "step": 7875
    },
    {
      "epoch": 3.706352941176471,
      "grad_norm": 0.3142180391384018,
      "learning_rate": 0.00012811645674243254,
      "loss": 0.1533,
      "step": 7876
    },
    {
      "epoch": 3.7068235294117646,
      "grad_norm": 0.2777509938957215,
      "learning_rate": 0.00012807309051407774,
      "loss": 0.104,
      "step": 7877
    },
    {
      "epoch": 3.707294117647059,
      "grad_norm": 0.4128011914305683,
      "learning_rate": 0.00012802972615838856,
      "loss": 0.2676,
      "step": 7878
    },
    {
      "epoch": 3.707764705882353,
      "grad_norm": 0.48391572448063047,
      "learning_rate": 0.00012798636367906849,
      "loss": 0.5391,
      "step": 7879
    },
    {
      "epoch": 3.708235294117647,
      "grad_norm": 0.485626123958541,
      "learning_rate": 0.00012794300307982098,
      "loss": 0.5508,
      "step": 7880
    },
    {
      "epoch": 3.708705882352941,
      "grad_norm": 0.3301114049597177,
      "learning_rate": 0.0001278996443643492,
      "loss": 0.377,
      "step": 7881
    },
    {
      "epoch": 3.7091764705882353,
      "grad_norm": 0.37342637791376376,
      "learning_rate": 0.00012785628753635617,
      "loss": 0.5625,
      "step": 7882
    },
    {
      "epoch": 3.7096470588235295,
      "grad_norm": 0.2453936368805864,
      "learning_rate": 0.0001278129325995448,
      "loss": 0.1318,
      "step": 7883
    },
    {
      "epoch": 3.7101176470588237,
      "grad_norm": 0.3763601174901215,
      "learning_rate": 0.00012776957955761784,
      "loss": 0.2773,
      "step": 7884
    },
    {
      "epoch": 3.7105882352941175,
      "grad_norm": 0.40128113973679125,
      "learning_rate": 0.0001277262284142778,
      "loss": 0.4609,
      "step": 7885
    },
    {
      "epoch": 3.7110588235294117,
      "grad_norm": 0.6592184811713228,
      "learning_rate": 0.00012768287917322706,
      "loss": 0.5508,
      "step": 7886
    },
    {
      "epoch": 3.711529411764706,
      "grad_norm": 0.3780303782761441,
      "learning_rate": 0.00012763953183816794,
      "loss": 0.1289,
      "step": 7887
    },
    {
      "epoch": 3.7119999999999997,
      "grad_norm": 0.35846869219724176,
      "learning_rate": 0.00012759618641280251,
      "loss": 0.3828,
      "step": 7888
    },
    {
      "epoch": 3.712470588235294,
      "grad_norm": 0.6360094761321555,
      "learning_rate": 0.00012755284290083264,
      "loss": 0.5977,
      "step": 7889
    },
    {
      "epoch": 3.712941176470588,
      "grad_norm": 0.326648059528271,
      "learning_rate": 0.00012750950130596006,
      "loss": 0.1533,
      "step": 7890
    },
    {
      "epoch": 3.7134117647058824,
      "grad_norm": 0.437400553004312,
      "learning_rate": 0.0001274661616318864,
      "loss": 0.3047,
      "step": 7891
    },
    {
      "epoch": 3.7138823529411766,
      "grad_norm": 0.19463200275671583,
      "learning_rate": 0.0001274228238823131,
      "loss": 0.0601,
      "step": 7892
    },
    {
      "epoch": 3.714352941176471,
      "grad_norm": 1.0236301717355873,
      "learning_rate": 0.00012737948806094135,
      "loss": 1.3359,
      "step": 7893
    },
    {
      "epoch": 3.7148235294117646,
      "grad_norm": 0.38989531080340467,
      "learning_rate": 0.00012733615417147228,
      "loss": 0.2314,
      "step": 7894
    },
    {
      "epoch": 3.715294117647059,
      "grad_norm": 0.37013221493521364,
      "learning_rate": 0.0001272928222176068,
      "loss": 0.4824,
      "step": 7895
    },
    {
      "epoch": 3.715764705882353,
      "grad_norm": 0.2637715151249056,
      "learning_rate": 0.0001272494922030457,
      "loss": 0.0923,
      "step": 7896
    },
    {
      "epoch": 3.716235294117647,
      "grad_norm": 0.45827896134260165,
      "learning_rate": 0.0001272061641314895,
      "loss": 0.2129,
      "step": 7897
    },
    {
      "epoch": 3.716705882352941,
      "grad_norm": 0.9871390871451708,
      "learning_rate": 0.00012716283800663874,
      "loss": 1.2031,
      "step": 7898
    },
    {
      "epoch": 3.7171764705882353,
      "grad_norm": 0.2217128045875849,
      "learning_rate": 0.00012711951383219362,
      "loss": 0.0713,
      "step": 7899
    },
    {
      "epoch": 3.7176470588235295,
      "grad_norm": 0.35428590847479846,
      "learning_rate": 0.00012707619161185423,
      "loss": 0.2256,
      "step": 7900
    },
    {
      "epoch": 3.7181176470588237,
      "grad_norm": 0.3811553757442436,
      "learning_rate": 0.00012703287134932054,
      "loss": 0.3047,
      "step": 7901
    },
    {
      "epoch": 3.7185882352941175,
      "grad_norm": 0.43013716827467235,
      "learning_rate": 0.0001269895530482923,
      "loss": 0.3535,
      "step": 7902
    },
    {
      "epoch": 3.7190588235294118,
      "grad_norm": 0.6603005059921664,
      "learning_rate": 0.00012694623671246908,
      "loss": 0.8398,
      "step": 7903
    },
    {
      "epoch": 3.719529411764706,
      "grad_norm": 0.409912619543451,
      "learning_rate": 0.0001269029223455503,
      "loss": 0.3535,
      "step": 7904
    },
    {
      "epoch": 3.7199999999999998,
      "grad_norm": 0.2869892634445096,
      "learning_rate": 0.00012685960995123527,
      "loss": 0.1279,
      "step": 7905
    },
    {
      "epoch": 3.720470588235294,
      "grad_norm": 1.0057187223421993,
      "learning_rate": 0.00012681629953322307,
      "loss": 0.8945,
      "step": 7906
    },
    {
      "epoch": 3.720941176470588,
      "grad_norm": 0.45502343574482906,
      "learning_rate": 0.00012677299109521256,
      "loss": 0.5391,
      "step": 7907
    },
    {
      "epoch": 3.7214117647058824,
      "grad_norm": 0.23105498699535434,
      "learning_rate": 0.00012672968464090255,
      "loss": 0.1289,
      "step": 7908
    },
    {
      "epoch": 3.7218823529411766,
      "grad_norm": 0.5082530392822547,
      "learning_rate": 0.00012668638017399164,
      "loss": 0.7266,
      "step": 7909
    },
    {
      "epoch": 3.722352941176471,
      "grad_norm": 0.38381224966055855,
      "learning_rate": 0.00012664307769817815,
      "loss": 0.4805,
      "step": 7910
    },
    {
      "epoch": 3.7228235294117646,
      "grad_norm": 0.4703400505753365,
      "learning_rate": 0.00012659977721716045,
      "loss": 0.2715,
      "step": 7911
    },
    {
      "epoch": 3.723294117647059,
      "grad_norm": 0.3294941841431986,
      "learning_rate": 0.00012655647873463657,
      "loss": 0.4258,
      "step": 7912
    },
    {
      "epoch": 3.723764705882353,
      "grad_norm": 0.3367535742315707,
      "learning_rate": 0.00012651318225430437,
      "loss": 0.3633,
      "step": 7913
    },
    {
      "epoch": 3.724235294117647,
      "grad_norm": 0.256071917692903,
      "learning_rate": 0.00012646988777986157,
      "loss": 0.0938,
      "step": 7914
    },
    {
      "epoch": 3.724705882352941,
      "grad_norm": 0.2769819242595864,
      "learning_rate": 0.00012642659531500583,
      "loss": 0.0776,
      "step": 7915
    },
    {
      "epoch": 3.7251764705882353,
      "grad_norm": 0.5079924948006003,
      "learning_rate": 0.00012638330486343445,
      "loss": 0.5391,
      "step": 7916
    },
    {
      "epoch": 3.7256470588235295,
      "grad_norm": 0.5110461455279003,
      "learning_rate": 0.00012634001642884465,
      "loss": 0.4043,
      "step": 7917
    },
    {
      "epoch": 3.7261176470588238,
      "grad_norm": 0.4554375021035566,
      "learning_rate": 0.0001262967300149335,
      "loss": 0.3047,
      "step": 7918
    },
    {
      "epoch": 3.7265882352941175,
      "grad_norm": 0.24382515320215176,
      "learning_rate": 0.0001262534456253979,
      "loss": 0.1602,
      "step": 7919
    },
    {
      "epoch": 3.7270588235294118,
      "grad_norm": 0.309284162421824,
      "learning_rate": 0.0001262101632639345,
      "loss": 0.3496,
      "step": 7920
    },
    {
      "epoch": 3.727529411764706,
      "grad_norm": 0.41768758146006524,
      "learning_rate": 0.00012616688293423987,
      "loss": 0.3672,
      "step": 7921
    },
    {
      "epoch": 3.7279999999999998,
      "grad_norm": 1.393995013576825,
      "learning_rate": 0.00012612360464001033,
      "loss": 0.6758,
      "step": 7922
    },
    {
      "epoch": 3.728470588235294,
      "grad_norm": 0.41869321853892716,
      "learning_rate": 0.00012608032838494205,
      "loss": 0.3848,
      "step": 7923
    },
    {
      "epoch": 3.728941176470588,
      "grad_norm": 0.29941536594477314,
      "learning_rate": 0.00012603705417273107,
      "loss": 0.0996,
      "step": 7924
    },
    {
      "epoch": 3.7294117647058824,
      "grad_norm": 0.46547340603444337,
      "learning_rate": 0.0001259937820070732,
      "loss": 0.3516,
      "step": 7925
    },
    {
      "epoch": 3.7298823529411766,
      "grad_norm": 0.5849266111925563,
      "learning_rate": 0.0001259505118916641,
      "loss": 0.2441,
      "step": 7926
    },
    {
      "epoch": 3.7303529411764704,
      "grad_norm": 0.22883499801776813,
      "learning_rate": 0.0001259072438301992,
      "loss": 0.106,
      "step": 7927
    },
    {
      "epoch": 3.7308235294117646,
      "grad_norm": 0.3204466112992475,
      "learning_rate": 0.00012586397782637392,
      "loss": 0.1924,
      "step": 7928
    },
    {
      "epoch": 3.731294117647059,
      "grad_norm": 0.1881136698257575,
      "learning_rate": 0.0001258207138838833,
      "loss": 0.04,
      "step": 7929
    },
    {
      "epoch": 3.731764705882353,
      "grad_norm": 0.49467426519775587,
      "learning_rate": 0.00012577745200642227,
      "loss": 0.5547,
      "step": 7930
    },
    {
      "epoch": 3.732235294117647,
      "grad_norm": 0.39675853767210845,
      "learning_rate": 0.00012573419219768562,
      "loss": 0.3613,
      "step": 7931
    },
    {
      "epoch": 3.732705882352941,
      "grad_norm": 0.3166578100629279,
      "learning_rate": 0.00012569093446136806,
      "loss": 0.2021,
      "step": 7932
    },
    {
      "epoch": 3.7331764705882353,
      "grad_norm": 0.6496835811941284,
      "learning_rate": 0.0001256476788011639,
      "loss": 0.6875,
      "step": 7933
    },
    {
      "epoch": 3.7336470588235295,
      "grad_norm": 0.32301278673355227,
      "learning_rate": 0.00012560442522076745,
      "loss": 0.2031,
      "step": 7934
    },
    {
      "epoch": 3.7341176470588238,
      "grad_norm": 0.426372034886427,
      "learning_rate": 0.00012556117372387272,
      "loss": 0.5742,
      "step": 7935
    },
    {
      "epoch": 3.7345882352941175,
      "grad_norm": 0.5566145311658377,
      "learning_rate": 0.00012551792431417363,
      "loss": 0.4746,
      "step": 7936
    },
    {
      "epoch": 3.7350588235294118,
      "grad_norm": 0.7478756735436787,
      "learning_rate": 0.0001254746769953639,
      "loss": 0.8555,
      "step": 7937
    },
    {
      "epoch": 3.735529411764706,
      "grad_norm": 0.19917883574936124,
      "learning_rate": 0.000125431431771137,
      "loss": 0.1562,
      "step": 7938
    },
    {
      "epoch": 3.7359999999999998,
      "grad_norm": 0.32421614595085885,
      "learning_rate": 0.0001253881886451864,
      "loss": 0.3281,
      "step": 7939
    },
    {
      "epoch": 3.736470588235294,
      "grad_norm": 0.2200920178140999,
      "learning_rate": 0.0001253449476212052,
      "loss": 0.085,
      "step": 7940
    },
    {
      "epoch": 3.736941176470588,
      "grad_norm": 0.3379632560709963,
      "learning_rate": 0.00012530170870288636,
      "loss": 0.4492,
      "step": 7941
    },
    {
      "epoch": 3.7374117647058824,
      "grad_norm": 0.40278388039130847,
      "learning_rate": 0.00012525847189392278,
      "loss": 0.3066,
      "step": 7942
    },
    {
      "epoch": 3.7378823529411767,
      "grad_norm": 0.6244588582324708,
      "learning_rate": 0.00012521523719800706,
      "loss": 0.6836,
      "step": 7943
    },
    {
      "epoch": 3.7383529411764704,
      "grad_norm": 0.41990759281464074,
      "learning_rate": 0.00012517200461883163,
      "loss": 0.1816,
      "step": 7944
    },
    {
      "epoch": 3.7388235294117647,
      "grad_norm": 0.35681958214676135,
      "learning_rate": 0.00012512877416008878,
      "loss": 0.2051,
      "step": 7945
    },
    {
      "epoch": 3.739294117647059,
      "grad_norm": 0.3813680967782122,
      "learning_rate": 0.00012508554582547064,
      "loss": 0.2637,
      "step": 7946
    },
    {
      "epoch": 3.739764705882353,
      "grad_norm": 0.3883480941666314,
      "learning_rate": 0.00012504231961866908,
      "loss": 0.4199,
      "step": 7947
    },
    {
      "epoch": 3.740235294117647,
      "grad_norm": 0.5028767031875602,
      "learning_rate": 0.00012499909554337582,
      "loss": 0.3281,
      "step": 7948
    },
    {
      "epoch": 3.740705882352941,
      "grad_norm": 0.5725993266631717,
      "learning_rate": 0.00012495587360328246,
      "loss": 0.6328,
      "step": 7949
    },
    {
      "epoch": 3.7411764705882353,
      "grad_norm": 0.5317274342178159,
      "learning_rate": 0.00012491265380208032,
      "loss": 0.3711,
      "step": 7950
    },
    {
      "epoch": 3.7416470588235295,
      "grad_norm": 0.3863518446754361,
      "learning_rate": 0.0001248694361434606,
      "loss": 0.4668,
      "step": 7951
    },
    {
      "epoch": 3.7421176470588238,
      "grad_norm": 0.6028503580031574,
      "learning_rate": 0.0001248262206311143,
      "loss": 0.2168,
      "step": 7952
    },
    {
      "epoch": 3.7425882352941175,
      "grad_norm": 0.46680754141991637,
      "learning_rate": 0.00012478300726873225,
      "loss": 0.5312,
      "step": 7953
    },
    {
      "epoch": 3.7430588235294118,
      "grad_norm": 0.4737681543062057,
      "learning_rate": 0.00012473979606000504,
      "loss": 0.2637,
      "step": 7954
    },
    {
      "epoch": 3.743529411764706,
      "grad_norm": 0.5031936476996388,
      "learning_rate": 0.00012469658700862315,
      "loss": 0.5039,
      "step": 7955
    },
    {
      "epoch": 3.7439999999999998,
      "grad_norm": 0.42332709110949784,
      "learning_rate": 0.00012465338011827685,
      "loss": 0.2148,
      "step": 7956
    },
    {
      "epoch": 3.744470588235294,
      "grad_norm": 0.41684916209210526,
      "learning_rate": 0.00012461017539265622,
      "loss": 0.3848,
      "step": 7957
    },
    {
      "epoch": 3.744941176470588,
      "grad_norm": 0.46560503181881707,
      "learning_rate": 0.00012456697283545114,
      "loss": 0.6445,
      "step": 7958
    },
    {
      "epoch": 3.7454117647058824,
      "grad_norm": 0.3912292786120369,
      "learning_rate": 0.00012452377245035132,
      "loss": 0.3203,
      "step": 7959
    },
    {
      "epoch": 3.7458823529411767,
      "grad_norm": 0.2864100718736933,
      "learning_rate": 0.00012448057424104632,
      "loss": 0.0737,
      "step": 7960
    },
    {
      "epoch": 3.7463529411764704,
      "grad_norm": 0.3192673531938695,
      "learning_rate": 0.00012443737821122541,
      "loss": 0.1846,
      "step": 7961
    },
    {
      "epoch": 3.7468235294117647,
      "grad_norm": 0.2513324816282688,
      "learning_rate": 0.0001243941843645778,
      "loss": 0.1182,
      "step": 7962
    },
    {
      "epoch": 3.747294117647059,
      "grad_norm": 0.3909209287069493,
      "learning_rate": 0.00012435099270479246,
      "loss": 0.375,
      "step": 7963
    },
    {
      "epoch": 3.7477647058823527,
      "grad_norm": 0.8140364637678906,
      "learning_rate": 0.00012430780323555816,
      "loss": 0.8594,
      "step": 7964
    },
    {
      "epoch": 3.748235294117647,
      "grad_norm": 0.5178363401028989,
      "learning_rate": 0.00012426461596056344,
      "loss": 0.3633,
      "step": 7965
    },
    {
      "epoch": 3.748705882352941,
      "grad_norm": 0.3807102721695786,
      "learning_rate": 0.00012422143088349679,
      "loss": 0.3887,
      "step": 7966
    },
    {
      "epoch": 3.7491764705882353,
      "grad_norm": 0.26310158871349626,
      "learning_rate": 0.00012417824800804638,
      "loss": 0.0918,
      "step": 7967
    },
    {
      "epoch": 3.7496470588235296,
      "grad_norm": 0.4810322305916618,
      "learning_rate": 0.0001241350673379002,
      "loss": 0.2285,
      "step": 7968
    },
    {
      "epoch": 3.7501176470588238,
      "grad_norm": 0.09411022709133424,
      "learning_rate": 0.0001240918888767462,
      "loss": 0.0508,
      "step": 7969
    },
    {
      "epoch": 3.7505882352941176,
      "grad_norm": 0.4993146965256245,
      "learning_rate": 0.00012404871262827193,
      "loss": 0.6523,
      "step": 7970
    },
    {
      "epoch": 3.751058823529412,
      "grad_norm": 0.8968781667796917,
      "learning_rate": 0.00012400553859616488,
      "loss": 1.0078,
      "step": 7971
    },
    {
      "epoch": 3.751529411764706,
      "grad_norm": 0.7732272860513053,
      "learning_rate": 0.00012396236678411236,
      "loss": 0.7344,
      "step": 7972
    },
    {
      "epoch": 3.752,
      "grad_norm": 0.3563572082868704,
      "learning_rate": 0.00012391919719580144,
      "loss": 0.2207,
      "step": 7973
    },
    {
      "epoch": 3.752470588235294,
      "grad_norm": 21.767323611059595,
      "learning_rate": 0.00012387602983491894,
      "loss": 7.5938,
      "step": 7974
    },
    {
      "epoch": 3.7529411764705882,
      "grad_norm": 0.5590551236579956,
      "learning_rate": 0.00012383286470515166,
      "loss": 0.4785,
      "step": 7975
    },
    {
      "epoch": 3.7534117647058824,
      "grad_norm": 9.571457761870548,
      "learning_rate": 0.0001237897018101861,
      "loss": 2.8125,
      "step": 7976
    },
    {
      "epoch": 3.7538823529411767,
      "grad_norm": 3.693981566242514,
      "learning_rate": 0.00012374654115370856,
      "loss": 0.7578,
      "step": 7977
    },
    {
      "epoch": 3.7543529411764704,
      "grad_norm": 6.64765989983362,
      "learning_rate": 0.0001237033827394052,
      "loss": 1.2188,
      "step": 7978
    },
    {
      "epoch": 3.7548235294117647,
      "grad_norm": 1.5337904932939355,
      "learning_rate": 0.0001236602265709619,
      "loss": 0.21,
      "step": 7979
    },
    {
      "epoch": 3.755294117647059,
      "grad_norm": 0.40510596041115415,
      "learning_rate": 0.00012361707265206448,
      "loss": 0.3281,
      "step": 7980
    },
    {
      "epoch": 3.7557647058823527,
      "grad_norm": 0.3243651502407423,
      "learning_rate": 0.00012357392098639848,
      "loss": 0.1514,
      "step": 7981
    },
    {
      "epoch": 3.756235294117647,
      "grad_norm": 8.21627055974191,
      "learning_rate": 0.00012353077157764922,
      "loss": 1.25,
      "step": 7982
    },
    {
      "epoch": 3.756705882352941,
      "grad_norm": 0.8613443490895267,
      "learning_rate": 0.00012348762442950187,
      "loss": 0.6211,
      "step": 7983
    },
    {
      "epoch": 3.7571764705882353,
      "grad_norm": 0.7922877038684021,
      "learning_rate": 0.00012344447954564148,
      "loss": 0.625,
      "step": 7984
    },
    {
      "epoch": 3.7576470588235296,
      "grad_norm": 0.4256523165249565,
      "learning_rate": 0.0001234013369297528,
      "loss": 0.2158,
      "step": 7985
    },
    {
      "epoch": 3.758117647058824,
      "grad_norm": 0.23178484782438463,
      "learning_rate": 0.00012335819658552036,
      "loss": 0.0391,
      "step": 7986
    },
    {
      "epoch": 3.7585882352941176,
      "grad_norm": 0.9195229063825178,
      "learning_rate": 0.00012331505851662864,
      "loss": 0.7773,
      "step": 7987
    },
    {
      "epoch": 3.759058823529412,
      "grad_norm": 0.6413588822187417,
      "learning_rate": 0.00012327192272676183,
      "loss": 0.4551,
      "step": 7988
    },
    {
      "epoch": 3.759529411764706,
      "grad_norm": 0.21592003188866996,
      "learning_rate": 0.0001232287892196039,
      "loss": 0.0645,
      "step": 7989
    },
    {
      "epoch": 3.76,
      "grad_norm": 0.37163991460537743,
      "learning_rate": 0.00012318565799883865,
      "loss": 0.165,
      "step": 7990
    },
    {
      "epoch": 3.760470588235294,
      "grad_norm": 0.5386627515994303,
      "learning_rate": 0.0001231425290681498,
      "loss": 0.4648,
      "step": 7991
    },
    {
      "epoch": 3.7609411764705882,
      "grad_norm": 0.39046020836441764,
      "learning_rate": 0.00012309940243122064,
      "loss": 0.2041,
      "step": 7992
    },
    {
      "epoch": 3.7614117647058825,
      "grad_norm": 0.35655080719373866,
      "learning_rate": 0.00012305627809173448,
      "loss": 0.1611,
      "step": 7993
    },
    {
      "epoch": 3.7618823529411767,
      "grad_norm": 0.3596236579868322,
      "learning_rate": 0.00012301315605337435,
      "loss": 0.291,
      "step": 7994
    },
    {
      "epoch": 3.7623529411764705,
      "grad_norm": 0.5698794694311602,
      "learning_rate": 0.00012297003631982304,
      "loss": 0.9883,
      "step": 7995
    },
    {
      "epoch": 3.7628235294117647,
      "grad_norm": 0.3624010359025818,
      "learning_rate": 0.0001229269188947632,
      "loss": 0.1514,
      "step": 7996
    },
    {
      "epoch": 3.763294117647059,
      "grad_norm": 0.3951099820608265,
      "learning_rate": 0.0001228838037818773,
      "loss": 0.2949,
      "step": 7997
    },
    {
      "epoch": 3.7637647058823527,
      "grad_norm": 0.7536329342621673,
      "learning_rate": 0.00012284069098484758,
      "loss": 0.7305,
      "step": 7998
    },
    {
      "epoch": 3.764235294117647,
      "grad_norm": 0.3939752504551441,
      "learning_rate": 0.000122797580507356,
      "loss": 0.418,
      "step": 7999
    },
    {
      "epoch": 3.764705882352941,
      "grad_norm": 1.0149256473442925,
      "learning_rate": 0.00012275447235308453,
      "loss": 1.4219,
      "step": 8000
    },
    {
      "epoch": 3.7651764705882353,
      "grad_norm": 0.3308432013786029,
      "learning_rate": 0.00012271136652571474,
      "loss": 0.3047,
      "step": 8001
    },
    {
      "epoch": 3.7656470588235296,
      "grad_norm": 0.48207953996869274,
      "learning_rate": 0.0001226682630289281,
      "loss": 0.5156,
      "step": 8002
    },
    {
      "epoch": 3.766117647058824,
      "grad_norm": 0.8874010922505415,
      "learning_rate": 0.00012262516186640587,
      "loss": 0.5,
      "step": 8003
    },
    {
      "epoch": 3.7665882352941176,
      "grad_norm": 0.4252073175639161,
      "learning_rate": 0.0001225820630418291,
      "loss": 0.1504,
      "step": 8004
    },
    {
      "epoch": 3.767058823529412,
      "grad_norm": 0.5289448643732024,
      "learning_rate": 0.00012253896655887861,
      "loss": 0.5625,
      "step": 8005
    },
    {
      "epoch": 3.767529411764706,
      "grad_norm": 0.4289635582660922,
      "learning_rate": 0.00012249587242123505,
      "loss": 0.6406,
      "step": 8006
    },
    {
      "epoch": 3.768,
      "grad_norm": 0.6657689962115335,
      "learning_rate": 0.00012245278063257892,
      "loss": 0.3691,
      "step": 8007
    },
    {
      "epoch": 3.768470588235294,
      "grad_norm": 0.5312416169298625,
      "learning_rate": 0.00012240969119659044,
      "loss": 0.6758,
      "step": 8008
    },
    {
      "epoch": 3.7689411764705882,
      "grad_norm": 0.3232963724966368,
      "learning_rate": 0.00012236660411694964,
      "loss": 0.1572,
      "step": 8009
    },
    {
      "epoch": 3.7694117647058825,
      "grad_norm": 0.3632377956268893,
      "learning_rate": 0.00012232351939733642,
      "loss": 0.2773,
      "step": 8010
    },
    {
      "epoch": 3.7698823529411767,
      "grad_norm": 0.42264409744292136,
      "learning_rate": 0.0001222804370414304,
      "loss": 0.2891,
      "step": 8011
    },
    {
      "epoch": 3.7703529411764705,
      "grad_norm": 0.5595181836890646,
      "learning_rate": 0.00012223735705291098,
      "loss": 0.3516,
      "step": 8012
    },
    {
      "epoch": 3.7708235294117647,
      "grad_norm": 0.6151410415466032,
      "learning_rate": 0.00012219427943545744,
      "loss": 0.707,
      "step": 8013
    },
    {
      "epoch": 3.771294117647059,
      "grad_norm": 0.4219803546011156,
      "learning_rate": 0.00012215120419274887,
      "loss": 0.1445,
      "step": 8014
    },
    {
      "epoch": 3.7717647058823527,
      "grad_norm": 0.45644260355292343,
      "learning_rate": 0.00012210813132846406,
      "loss": 0.4082,
      "step": 8015
    },
    {
      "epoch": 3.772235294117647,
      "grad_norm": 0.39586936111487925,
      "learning_rate": 0.0001220650608462816,
      "loss": 0.5742,
      "step": 8016
    },
    {
      "epoch": 3.772705882352941,
      "grad_norm": 0.5420083872797131,
      "learning_rate": 0.00012202199274987999,
      "loss": 0.6484,
      "step": 8017
    },
    {
      "epoch": 3.7731764705882354,
      "grad_norm": 0.6017432213417391,
      "learning_rate": 0.00012197892704293743,
      "loss": 0.543,
      "step": 8018
    },
    {
      "epoch": 3.7736470588235296,
      "grad_norm": 0.5349569815635172,
      "learning_rate": 0.00012193586372913192,
      "loss": 0.5469,
      "step": 8019
    },
    {
      "epoch": 3.774117647058824,
      "grad_norm": 0.4100762723059277,
      "learning_rate": 0.00012189280281214126,
      "loss": 0.3047,
      "step": 8020
    },
    {
      "epoch": 3.7745882352941176,
      "grad_norm": 0.44801303109103746,
      "learning_rate": 0.00012184974429564318,
      "loss": 0.2695,
      "step": 8021
    },
    {
      "epoch": 3.775058823529412,
      "grad_norm": 0.44417903674132714,
      "learning_rate": 0.000121806688183315,
      "loss": 0.2295,
      "step": 8022
    },
    {
      "epoch": 3.775529411764706,
      "grad_norm": 0.6016638922646429,
      "learning_rate": 0.00012176363447883397,
      "loss": 0.5742,
      "step": 8023
    },
    {
      "epoch": 3.776,
      "grad_norm": 0.46597304441533177,
      "learning_rate": 0.00012172058318587698,
      "loss": 0.3848,
      "step": 8024
    },
    {
      "epoch": 3.776470588235294,
      "grad_norm": 0.3740540741548275,
      "learning_rate": 0.00012167753430812093,
      "loss": 0.1973,
      "step": 8025
    },
    {
      "epoch": 3.7769411764705882,
      "grad_norm": 0.4667122821631666,
      "learning_rate": 0.0001216344878492424,
      "loss": 0.5469,
      "step": 8026
    },
    {
      "epoch": 3.7774117647058825,
      "grad_norm": 0.5056749040925743,
      "learning_rate": 0.0001215914438129177,
      "loss": 0.4961,
      "step": 8027
    },
    {
      "epoch": 3.7778823529411767,
      "grad_norm": 0.5581131793301822,
      "learning_rate": 0.00012154840220282306,
      "loss": 0.3438,
      "step": 8028
    },
    {
      "epoch": 3.7783529411764705,
      "grad_norm": 0.17589272461974217,
      "learning_rate": 0.00012150536302263444,
      "loss": 0.0306,
      "step": 8029
    },
    {
      "epoch": 3.7788235294117647,
      "grad_norm": 0.6623564887766078,
      "learning_rate": 0.00012146232627602753,
      "loss": 0.4473,
      "step": 8030
    },
    {
      "epoch": 3.779294117647059,
      "grad_norm": 0.4970541204722617,
      "learning_rate": 0.00012141929196667798,
      "loss": 0.6602,
      "step": 8031
    },
    {
      "epoch": 3.7797647058823527,
      "grad_norm": 0.6496399577656582,
      "learning_rate": 0.00012137626009826108,
      "loss": 0.4844,
      "step": 8032
    },
    {
      "epoch": 3.780235294117647,
      "grad_norm": 0.3654884352798282,
      "learning_rate": 0.00012133323067445198,
      "loss": 0.4062,
      "step": 8033
    },
    {
      "epoch": 3.780705882352941,
      "grad_norm": 0.45083997516661956,
      "learning_rate": 0.00012129020369892555,
      "loss": 0.4023,
      "step": 8034
    },
    {
      "epoch": 3.7811764705882354,
      "grad_norm": 0.44302998426358225,
      "learning_rate": 0.00012124717917535656,
      "loss": 0.1934,
      "step": 8035
    },
    {
      "epoch": 3.7816470588235296,
      "grad_norm": 0.609855879334466,
      "learning_rate": 0.00012120415710741951,
      "loss": 0.7617,
      "step": 8036
    },
    {
      "epoch": 3.782117647058824,
      "grad_norm": 0.1973297717689093,
      "learning_rate": 0.00012116113749878867,
      "loss": 0.041,
      "step": 8037
    },
    {
      "epoch": 3.7825882352941176,
      "grad_norm": 0.5215828482790978,
      "learning_rate": 0.00012111812035313814,
      "loss": 0.75,
      "step": 8038
    },
    {
      "epoch": 3.783058823529412,
      "grad_norm": 0.3209545287834077,
      "learning_rate": 0.00012107510567414183,
      "loss": 0.2412,
      "step": 8039
    },
    {
      "epoch": 3.783529411764706,
      "grad_norm": 0.43388424065076936,
      "learning_rate": 0.00012103209346547332,
      "loss": 0.5156,
      "step": 8040
    },
    {
      "epoch": 3.784,
      "grad_norm": 0.33195260546331684,
      "learning_rate": 0.00012098908373080617,
      "loss": 0.1689,
      "step": 8041
    },
    {
      "epoch": 3.784470588235294,
      "grad_norm": 0.40179443160252676,
      "learning_rate": 0.00012094607647381355,
      "loss": 0.4102,
      "step": 8042
    },
    {
      "epoch": 3.7849411764705883,
      "grad_norm": 0.3815383494830323,
      "learning_rate": 0.00012090307169816847,
      "loss": 0.3613,
      "step": 8043
    },
    {
      "epoch": 3.7854117647058825,
      "grad_norm": 0.38125563863314554,
      "learning_rate": 0.00012086006940754382,
      "loss": 0.4512,
      "step": 8044
    },
    {
      "epoch": 3.7858823529411767,
      "grad_norm": 0.3965607350232322,
      "learning_rate": 0.00012081706960561217,
      "loss": 0.625,
      "step": 8045
    },
    {
      "epoch": 3.7863529411764705,
      "grad_norm": 0.49298860396336547,
      "learning_rate": 0.00012077407229604593,
      "loss": 0.3203,
      "step": 8046
    },
    {
      "epoch": 3.7868235294117647,
      "grad_norm": 0.4062752482745224,
      "learning_rate": 0.00012073107748251724,
      "loss": 0.3496,
      "step": 8047
    },
    {
      "epoch": 3.787294117647059,
      "grad_norm": 0.4393234048703341,
      "learning_rate": 0.00012068808516869812,
      "loss": 0.2637,
      "step": 8048
    },
    {
      "epoch": 3.7877647058823527,
      "grad_norm": 0.26834167872418924,
      "learning_rate": 0.00012064509535826029,
      "loss": 0.0952,
      "step": 8049
    },
    {
      "epoch": 3.788235294117647,
      "grad_norm": 0.623626364592754,
      "learning_rate": 0.00012060210805487529,
      "loss": 0.6992,
      "step": 8050
    },
    {
      "epoch": 3.788705882352941,
      "grad_norm": 0.331690464715107,
      "learning_rate": 0.00012055912326221448,
      "loss": 0.2051,
      "step": 8051
    },
    {
      "epoch": 3.7891764705882354,
      "grad_norm": 0.49332790754921446,
      "learning_rate": 0.00012051614098394894,
      "loss": 0.543,
      "step": 8052
    },
    {
      "epoch": 3.7896470588235296,
      "grad_norm": 0.4178099123774064,
      "learning_rate": 0.00012047316122374957,
      "loss": 0.3555,
      "step": 8053
    },
    {
      "epoch": 3.7901176470588234,
      "grad_norm": 0.2914755160846013,
      "learning_rate": 0.00012043018398528707,
      "loss": 0.1172,
      "step": 8054
    },
    {
      "epoch": 3.7905882352941176,
      "grad_norm": 0.4020844886639282,
      "learning_rate": 0.0001203872092722319,
      "loss": 0.293,
      "step": 8055
    },
    {
      "epoch": 3.791058823529412,
      "grad_norm": 0.3503448862608428,
      "learning_rate": 0.00012034423708825431,
      "loss": 0.3145,
      "step": 8056
    },
    {
      "epoch": 3.791529411764706,
      "grad_norm": 0.3264233702422448,
      "learning_rate": 0.00012030126743702431,
      "loss": 0.3086,
      "step": 8057
    },
    {
      "epoch": 3.792,
      "grad_norm": 0.4434421685392727,
      "learning_rate": 0.00012025830032221177,
      "loss": 0.3926,
      "step": 8058
    },
    {
      "epoch": 3.792470588235294,
      "grad_norm": 0.5910658536595735,
      "learning_rate": 0.00012021533574748625,
      "loss": 0.5273,
      "step": 8059
    },
    {
      "epoch": 3.7929411764705883,
      "grad_norm": 0.4415962408652097,
      "learning_rate": 0.00012017237371651714,
      "loss": 0.3203,
      "step": 8060
    },
    {
      "epoch": 3.7934117647058825,
      "grad_norm": 0.7081693279244261,
      "learning_rate": 0.00012012941423297363,
      "loss": 1.1172,
      "step": 8061
    },
    {
      "epoch": 3.7938823529411767,
      "grad_norm": 0.4125427287167464,
      "learning_rate": 0.00012008645730052468,
      "loss": 0.5938,
      "step": 8062
    },
    {
      "epoch": 3.7943529411764705,
      "grad_norm": 0.23956508695167722,
      "learning_rate": 0.00012004350292283896,
      "loss": 0.2949,
      "step": 8063
    },
    {
      "epoch": 3.7948235294117647,
      "grad_norm": 0.3996057937145064,
      "learning_rate": 0.00012000055110358502,
      "loss": 0.5352,
      "step": 8064
    },
    {
      "epoch": 3.795294117647059,
      "grad_norm": 0.5776078154301562,
      "learning_rate": 0.00011995760184643118,
      "loss": 0.5352,
      "step": 8065
    },
    {
      "epoch": 3.7957647058823527,
      "grad_norm": 0.35806508979079676,
      "learning_rate": 0.00011991465515504551,
      "loss": 0.3418,
      "step": 8066
    },
    {
      "epoch": 3.796235294117647,
      "grad_norm": 0.26222683106495653,
      "learning_rate": 0.00011987171103309586,
      "loss": 0.2139,
      "step": 8067
    },
    {
      "epoch": 3.796705882352941,
      "grad_norm": 0.34115838926832764,
      "learning_rate": 0.00011982876948424986,
      "loss": 0.1934,
      "step": 8068
    },
    {
      "epoch": 3.7971764705882354,
      "grad_norm": 0.40832983518881316,
      "learning_rate": 0.00011978583051217493,
      "loss": 0.4727,
      "step": 8069
    },
    {
      "epoch": 3.7976470588235296,
      "grad_norm": 0.7262588466716242,
      "learning_rate": 0.0001197428941205383,
      "loss": 0.9609,
      "step": 8070
    },
    {
      "epoch": 3.7981176470588234,
      "grad_norm": 0.8634465495208856,
      "learning_rate": 0.00011969996031300687,
      "loss": 0.8672,
      "step": 8071
    },
    {
      "epoch": 3.7985882352941176,
      "grad_norm": 0.37540138670531714,
      "learning_rate": 0.00011965702909324751,
      "loss": 0.2422,
      "step": 8072
    },
    {
      "epoch": 3.799058823529412,
      "grad_norm": 0.4525920570162232,
      "learning_rate": 0.00011961410046492667,
      "loss": 0.2793,
      "step": 8073
    },
    {
      "epoch": 3.799529411764706,
      "grad_norm": 0.40723882424985774,
      "learning_rate": 0.0001195711744317107,
      "loss": 0.2168,
      "step": 8074
    },
    {
      "epoch": 3.8,
      "grad_norm": 0.3740194976575628,
      "learning_rate": 0.00011952825099726567,
      "loss": 0.2109,
      "step": 8075
    },
    {
      "epoch": 3.800470588235294,
      "grad_norm": 0.24080884044448433,
      "learning_rate": 0.00011948533016525749,
      "loss": 0.0654,
      "step": 8076
    },
    {
      "epoch": 3.8009411764705883,
      "grad_norm": 0.5077274059419772,
      "learning_rate": 0.00011944241193935178,
      "loss": 0.4629,
      "step": 8077
    },
    {
      "epoch": 3.8014117647058825,
      "grad_norm": 0.4038567994600899,
      "learning_rate": 0.00011939949632321395,
      "loss": 0.6055,
      "step": 8078
    },
    {
      "epoch": 3.8018823529411767,
      "grad_norm": 0.3796506207184613,
      "learning_rate": 0.00011935658332050925,
      "loss": 0.3379,
      "step": 8079
    },
    {
      "epoch": 3.8023529411764705,
      "grad_norm": 0.5192107876097287,
      "learning_rate": 0.00011931367293490265,
      "loss": 0.5352,
      "step": 8080
    },
    {
      "epoch": 3.8028235294117647,
      "grad_norm": 0.6143995460850387,
      "learning_rate": 0.00011927076517005885,
      "loss": 0.6172,
      "step": 8081
    },
    {
      "epoch": 3.803294117647059,
      "grad_norm": 0.4044639969040947,
      "learning_rate": 0.0001192278600296425,
      "loss": 0.2617,
      "step": 8082
    },
    {
      "epoch": 3.8037647058823527,
      "grad_norm": 0.26487653723296867,
      "learning_rate": 0.00011918495751731781,
      "loss": 0.1777,
      "step": 8083
    },
    {
      "epoch": 3.804235294117647,
      "grad_norm": 0.4213840972331138,
      "learning_rate": 0.0001191420576367489,
      "loss": 0.4512,
      "step": 8084
    },
    {
      "epoch": 3.804705882352941,
      "grad_norm": 0.07524146041652698,
      "learning_rate": 0.00011909916039159958,
      "loss": 0.0153,
      "step": 8085
    },
    {
      "epoch": 3.8051764705882354,
      "grad_norm": 0.21646467750264461,
      "learning_rate": 0.00011905626578553359,
      "loss": 0.051,
      "step": 8086
    },
    {
      "epoch": 3.8056470588235296,
      "grad_norm": 0.43945683481950387,
      "learning_rate": 0.00011901337382221425,
      "loss": 0.3184,
      "step": 8087
    },
    {
      "epoch": 3.8061176470588234,
      "grad_norm": 0.2181313012632466,
      "learning_rate": 0.00011897048450530477,
      "loss": 0.0552,
      "step": 8088
    },
    {
      "epoch": 3.8065882352941176,
      "grad_norm": 0.3774076814286899,
      "learning_rate": 0.00011892759783846812,
      "loss": 0.3652,
      "step": 8089
    },
    {
      "epoch": 3.807058823529412,
      "grad_norm": 0.4015184205908067,
      "learning_rate": 0.00011888471382536704,
      "loss": 0.3379,
      "step": 8090
    },
    {
      "epoch": 3.8075294117647056,
      "grad_norm": 0.44706831733121694,
      "learning_rate": 0.00011884183246966399,
      "loss": 0.3848,
      "step": 8091
    },
    {
      "epoch": 3.808,
      "grad_norm": 0.3781110551741557,
      "learning_rate": 0.00011879895377502129,
      "loss": 0.1611,
      "step": 8092
    },
    {
      "epoch": 3.808470588235294,
      "grad_norm": 0.5868570725105721,
      "learning_rate": 0.00011875607774510099,
      "loss": 0.4414,
      "step": 8093
    },
    {
      "epoch": 3.8089411764705883,
      "grad_norm": 0.523931959540853,
      "learning_rate": 0.00011871320438356491,
      "loss": 0.4941,
      "step": 8094
    },
    {
      "epoch": 3.8094117647058825,
      "grad_norm": 0.22214624297871785,
      "learning_rate": 0.0001186703336940746,
      "loss": 0.0674,
      "step": 8095
    },
    {
      "epoch": 3.8098823529411767,
      "grad_norm": 0.26798428084586584,
      "learning_rate": 0.00011862746568029149,
      "loss": 0.0977,
      "step": 8096
    },
    {
      "epoch": 3.8103529411764705,
      "grad_norm": 0.30084561629073314,
      "learning_rate": 0.00011858460034587669,
      "loss": 0.2031,
      "step": 8097
    },
    {
      "epoch": 3.8108235294117647,
      "grad_norm": 0.34251715114967946,
      "learning_rate": 0.0001185417376944911,
      "loss": 0.2344,
      "step": 8098
    },
    {
      "epoch": 3.811294117647059,
      "grad_norm": 0.33217620737179626,
      "learning_rate": 0.00011849887772979541,
      "loss": 0.1216,
      "step": 8099
    },
    {
      "epoch": 3.8117647058823527,
      "grad_norm": 0.23824518319473803,
      "learning_rate": 0.00011845602045545008,
      "loss": 0.106,
      "step": 8100
    },
    {
      "epoch": 3.812235294117647,
      "grad_norm": 0.40331350877336025,
      "learning_rate": 0.00011841316587511531,
      "loss": 0.3301,
      "step": 8101
    },
    {
      "epoch": 3.812705882352941,
      "grad_norm": 0.3777697186645394,
      "learning_rate": 0.00011837031399245115,
      "loss": 0.1914,
      "step": 8102
    },
    {
      "epoch": 3.8131764705882354,
      "grad_norm": 0.4530170570339483,
      "learning_rate": 0.00011832746481111728,
      "loss": 0.2344,
      "step": 8103
    },
    {
      "epoch": 3.8136470588235296,
      "grad_norm": 0.747665128511384,
      "learning_rate": 0.00011828461833477326,
      "loss": 0.6211,
      "step": 8104
    },
    {
      "epoch": 3.8141176470588234,
      "grad_norm": 0.3887999295166668,
      "learning_rate": 0.00011824177456707844,
      "loss": 0.4238,
      "step": 8105
    },
    {
      "epoch": 3.8145882352941176,
      "grad_norm": 0.38499704050756745,
      "learning_rate": 0.00011819893351169183,
      "loss": 0.1855,
      "step": 8106
    },
    {
      "epoch": 3.815058823529412,
      "grad_norm": 0.37514186698207397,
      "learning_rate": 0.00011815609517227228,
      "loss": 0.3887,
      "step": 8107
    },
    {
      "epoch": 3.8155294117647056,
      "grad_norm": 0.2511748219180187,
      "learning_rate": 0.00011811325955247833,
      "loss": 0.2637,
      "step": 8108
    },
    {
      "epoch": 3.816,
      "grad_norm": 0.8658785572068933,
      "learning_rate": 0.00011807042665596846,
      "loss": 1.0234,
      "step": 8109
    },
    {
      "epoch": 3.816470588235294,
      "grad_norm": 0.5502817356153235,
      "learning_rate": 0.0001180275964864008,
      "loss": 0.2656,
      "step": 8110
    },
    {
      "epoch": 3.8169411764705883,
      "grad_norm": 0.5020510674722364,
      "learning_rate": 0.00011798476904743321,
      "loss": 0.2598,
      "step": 8111
    },
    {
      "epoch": 3.8174117647058825,
      "grad_norm": 0.5558864576207835,
      "learning_rate": 0.00011794194434272336,
      "loss": 0.3086,
      "step": 8112
    },
    {
      "epoch": 3.8178823529411767,
      "grad_norm": 0.36874598569743067,
      "learning_rate": 0.00011789912237592874,
      "loss": 0.1592,
      "step": 8113
    },
    {
      "epoch": 3.8183529411764705,
      "grad_norm": 0.3460405144228885,
      "learning_rate": 0.00011785630315070651,
      "loss": 0.2363,
      "step": 8114
    },
    {
      "epoch": 3.8188235294117647,
      "grad_norm": 0.48612479749881937,
      "learning_rate": 0.00011781348667071365,
      "loss": 0.6133,
      "step": 8115
    },
    {
      "epoch": 3.819294117647059,
      "grad_norm": 0.29280630118122963,
      "learning_rate": 0.0001177706729396069,
      "loss": 0.1045,
      "step": 8116
    },
    {
      "epoch": 3.8197647058823527,
      "grad_norm": 0.30677765623957143,
      "learning_rate": 0.00011772786196104276,
      "loss": 0.1143,
      "step": 8117
    },
    {
      "epoch": 3.820235294117647,
      "grad_norm": 0.275174436816757,
      "learning_rate": 0.00011768505373867753,
      "loss": 0.1279,
      "step": 8118
    },
    {
      "epoch": 3.820705882352941,
      "grad_norm": 0.3323578113193582,
      "learning_rate": 0.00011764224827616717,
      "loss": 0.1602,
      "step": 8119
    },
    {
      "epoch": 3.8211764705882354,
      "grad_norm": 0.4852122237025616,
      "learning_rate": 0.00011759944557716755,
      "loss": 0.4551,
      "step": 8120
    },
    {
      "epoch": 3.8216470588235296,
      "grad_norm": 0.4633264368697154,
      "learning_rate": 0.00011755664564533422,
      "loss": 0.3613,
      "step": 8121
    },
    {
      "epoch": 3.8221176470588234,
      "grad_norm": 0.4443981198532098,
      "learning_rate": 0.00011751384848432245,
      "loss": 0.1738,
      "step": 8122
    },
    {
      "epoch": 3.8225882352941176,
      "grad_norm": 0.3069523118027628,
      "learning_rate": 0.00011747105409778738,
      "loss": 0.1377,
      "step": 8123
    },
    {
      "epoch": 3.823058823529412,
      "grad_norm": 0.26984454902379235,
      "learning_rate": 0.00011742826248938384,
      "loss": 0.1807,
      "step": 8124
    },
    {
      "epoch": 3.8235294117647056,
      "grad_norm": 0.3942153766993779,
      "learning_rate": 0.00011738547366276645,
      "loss": 0.4453,
      "step": 8125
    },
    {
      "epoch": 3.824,
      "grad_norm": 0.1925167642134635,
      "learning_rate": 0.00011734268762158957,
      "loss": 0.03,
      "step": 8126
    },
    {
      "epoch": 3.824470588235294,
      "grad_norm": 0.4577517198338308,
      "learning_rate": 0.00011729990436950737,
      "loss": 0.4492,
      "step": 8127
    },
    {
      "epoch": 3.8249411764705883,
      "grad_norm": 0.4193611003459112,
      "learning_rate": 0.00011725712391017374,
      "loss": 0.4336,
      "step": 8128
    },
    {
      "epoch": 3.8254117647058825,
      "grad_norm": 0.6348999427514601,
      "learning_rate": 0.00011721434624724227,
      "loss": 0.6055,
      "step": 8129
    },
    {
      "epoch": 3.8258823529411767,
      "grad_norm": 0.7674582933250852,
      "learning_rate": 0.0001171715713843665,
      "loss": 0.6094,
      "step": 8130
    },
    {
      "epoch": 3.8263529411764705,
      "grad_norm": 0.16749623998867608,
      "learning_rate": 0.00011712879932519954,
      "loss": 0.0437,
      "step": 8131
    },
    {
      "epoch": 3.8268235294117647,
      "grad_norm": 0.6098419848556274,
      "learning_rate": 0.00011708603007339433,
      "loss": 0.5195,
      "step": 8132
    },
    {
      "epoch": 3.827294117647059,
      "grad_norm": 0.41644791907254325,
      "learning_rate": 0.00011704326363260365,
      "loss": 0.418,
      "step": 8133
    },
    {
      "epoch": 3.8277647058823527,
      "grad_norm": 0.4702727615113567,
      "learning_rate": 0.00011700050000647986,
      "loss": 0.3379,
      "step": 8134
    },
    {
      "epoch": 3.828235294117647,
      "grad_norm": 0.24773252811702998,
      "learning_rate": 0.00011695773919867525,
      "loss": 0.127,
      "step": 8135
    },
    {
      "epoch": 3.828705882352941,
      "grad_norm": 0.5144532039657812,
      "learning_rate": 0.00011691498121284176,
      "loss": 0.4531,
      "step": 8136
    },
    {
      "epoch": 3.8291764705882354,
      "grad_norm": 0.46521666269627915,
      "learning_rate": 0.00011687222605263118,
      "loss": 0.6367,
      "step": 8137
    },
    {
      "epoch": 3.8296470588235296,
      "grad_norm": 0.4560546654737266,
      "learning_rate": 0.000116829473721695,
      "loss": 0.7578,
      "step": 8138
    },
    {
      "epoch": 3.8301176470588234,
      "grad_norm": 0.3623185811265341,
      "learning_rate": 0.00011678672422368442,
      "loss": 0.1299,
      "step": 8139
    },
    {
      "epoch": 3.8305882352941176,
      "grad_norm": 0.3999636999041987,
      "learning_rate": 0.00011674397756225053,
      "loss": 0.375,
      "step": 8140
    },
    {
      "epoch": 3.831058823529412,
      "grad_norm": 0.4978268257968887,
      "learning_rate": 0.00011670123374104409,
      "loss": 0.249,
      "step": 8141
    },
    {
      "epoch": 3.8315294117647056,
      "grad_norm": 0.452378782336109,
      "learning_rate": 0.00011665849276371557,
      "loss": 0.2617,
      "step": 8142
    },
    {
      "epoch": 3.832,
      "grad_norm": 0.35956051766671465,
      "learning_rate": 0.00011661575463391534,
      "loss": 0.3594,
      "step": 8143
    },
    {
      "epoch": 3.832470588235294,
      "grad_norm": 0.46005545591709646,
      "learning_rate": 0.0001165730193552934,
      "loss": 0.5664,
      "step": 8144
    },
    {
      "epoch": 3.8329411764705883,
      "grad_norm": 0.40041335429989466,
      "learning_rate": 0.00011653028693149961,
      "loss": 0.2559,
      "step": 8145
    },
    {
      "epoch": 3.8334117647058825,
      "grad_norm": 0.4347733687066976,
      "learning_rate": 0.00011648755736618342,
      "loss": 0.5977,
      "step": 8146
    },
    {
      "epoch": 3.8338823529411767,
      "grad_norm": 0.3242357965837816,
      "learning_rate": 0.00011644483066299424,
      "loss": 0.4512,
      "step": 8147
    },
    {
      "epoch": 3.8343529411764705,
      "grad_norm": 0.5134989910868065,
      "learning_rate": 0.00011640210682558112,
      "loss": 0.3066,
      "step": 8148
    },
    {
      "epoch": 3.8348235294117647,
      "grad_norm": 0.3207503202621292,
      "learning_rate": 0.00011635938585759284,
      "loss": 0.2734,
      "step": 8149
    },
    {
      "epoch": 3.835294117647059,
      "grad_norm": 0.30597028051344605,
      "learning_rate": 0.00011631666776267803,
      "loss": 0.1436,
      "step": 8150
    },
    {
      "epoch": 3.8357647058823527,
      "grad_norm": 0.33865790648218486,
      "learning_rate": 0.00011627395254448504,
      "loss": 0.2256,
      "step": 8151
    },
    {
      "epoch": 3.836235294117647,
      "grad_norm": 0.42055329055803814,
      "learning_rate": 0.00011623124020666186,
      "loss": 0.3418,
      "step": 8152
    },
    {
      "epoch": 3.836705882352941,
      "grad_norm": 0.30559249913135555,
      "learning_rate": 0.00011618853075285644,
      "loss": 0.3281,
      "step": 8153
    },
    {
      "epoch": 3.8371764705882354,
      "grad_norm": 0.36868222076210067,
      "learning_rate": 0.00011614582418671629,
      "loss": 0.2451,
      "step": 8154
    },
    {
      "epoch": 3.8376470588235296,
      "grad_norm": 0.43657950497488185,
      "learning_rate": 0.00011610312051188888,
      "loss": 0.5781,
      "step": 8155
    },
    {
      "epoch": 3.8381176470588234,
      "grad_norm": 0.3290395749282395,
      "learning_rate": 0.00011606041973202123,
      "loss": 0.2109,
      "step": 8156
    },
    {
      "epoch": 3.8385882352941176,
      "grad_norm": 0.2546950277521385,
      "learning_rate": 0.00011601772185076018,
      "loss": 0.1973,
      "step": 8157
    },
    {
      "epoch": 3.839058823529412,
      "grad_norm": 0.43826052518828523,
      "learning_rate": 0.00011597502687175238,
      "loss": 0.4219,
      "step": 8158
    },
    {
      "epoch": 3.8395294117647056,
      "grad_norm": 0.38763151388486167,
      "learning_rate": 0.0001159323347986442,
      "loss": 0.4941,
      "step": 8159
    },
    {
      "epoch": 3.84,
      "grad_norm": 0.5807567320681605,
      "learning_rate": 0.00011588964563508167,
      "loss": 0.6953,
      "step": 8160
    },
    {
      "epoch": 3.840470588235294,
      "grad_norm": 0.2708611280415929,
      "learning_rate": 0.00011584695938471077,
      "loss": 0.1387,
      "step": 8161
    },
    {
      "epoch": 3.8409411764705883,
      "grad_norm": 0.5016272724629197,
      "learning_rate": 0.00011580427605117705,
      "loss": 0.4062,
      "step": 8162
    },
    {
      "epoch": 3.8414117647058825,
      "grad_norm": 0.3832852657781565,
      "learning_rate": 0.00011576159563812589,
      "loss": 0.3887,
      "step": 8163
    },
    {
      "epoch": 3.8418823529411767,
      "grad_norm": 0.5000968802684111,
      "learning_rate": 0.00011571891814920236,
      "loss": 0.4004,
      "step": 8164
    },
    {
      "epoch": 3.8423529411764705,
      "grad_norm": 0.8940295121095897,
      "learning_rate": 0.00011567624358805139,
      "loss": 0.4824,
      "step": 8165
    },
    {
      "epoch": 3.8428235294117647,
      "grad_norm": 0.3580089667209326,
      "learning_rate": 0.0001156335719583176,
      "loss": 0.1416,
      "step": 8166
    },
    {
      "epoch": 3.843294117647059,
      "grad_norm": 0.6376902871684824,
      "learning_rate": 0.00011559090326364527,
      "loss": 0.4082,
      "step": 8167
    },
    {
      "epoch": 3.8437647058823528,
      "grad_norm": 0.49647640425180045,
      "learning_rate": 0.00011554823750767863,
      "loss": 0.2734,
      "step": 8168
    },
    {
      "epoch": 3.844235294117647,
      "grad_norm": 0.4490169140387851,
      "learning_rate": 0.0001155055746940615,
      "loss": 0.4805,
      "step": 8169
    },
    {
      "epoch": 3.844705882352941,
      "grad_norm": 0.2595262874030208,
      "learning_rate": 0.00011546291482643744,
      "loss": 0.0869,
      "step": 8170
    },
    {
      "epoch": 3.8451764705882354,
      "grad_norm": 0.396608508635724,
      "learning_rate": 0.00011542025790844988,
      "loss": 0.1572,
      "step": 8171
    },
    {
      "epoch": 3.8456470588235296,
      "grad_norm": 0.23221010286306054,
      "learning_rate": 0.00011537760394374192,
      "loss": 0.1377,
      "step": 8172
    },
    {
      "epoch": 3.8461176470588234,
      "grad_norm": 0.2884082770376292,
      "learning_rate": 0.00011533495293595637,
      "loss": 0.1104,
      "step": 8173
    },
    {
      "epoch": 3.8465882352941176,
      "grad_norm": 0.2703302591145711,
      "learning_rate": 0.0001152923048887359,
      "loss": 0.1904,
      "step": 8174
    },
    {
      "epoch": 3.847058823529412,
      "grad_norm": 0.5686969795568971,
      "learning_rate": 0.00011524965980572284,
      "loss": 0.377,
      "step": 8175
    },
    {
      "epoch": 3.8475294117647056,
      "grad_norm": 0.3853125047355778,
      "learning_rate": 0.00011520701769055927,
      "loss": 0.1299,
      "step": 8176
    },
    {
      "epoch": 3.848,
      "grad_norm": 0.3944749091339462,
      "learning_rate": 0.00011516437854688703,
      "loss": 0.334,
      "step": 8177
    },
    {
      "epoch": 3.848470588235294,
      "grad_norm": 0.4066124110003747,
      "learning_rate": 0.00011512174237834776,
      "loss": 0.5781,
      "step": 8178
    },
    {
      "epoch": 3.8489411764705883,
      "grad_norm": 0.290902111188045,
      "learning_rate": 0.00011507910918858276,
      "loss": 0.1475,
      "step": 8179
    },
    {
      "epoch": 3.8494117647058825,
      "grad_norm": 0.23155731573208993,
      "learning_rate": 0.0001150364789812331,
      "loss": 0.0732,
      "step": 8180
    },
    {
      "epoch": 3.8498823529411763,
      "grad_norm": 0.3461135952368999,
      "learning_rate": 0.00011499385175993965,
      "loss": 0.1797,
      "step": 8181
    },
    {
      "epoch": 3.8503529411764705,
      "grad_norm": 0.4738068867989005,
      "learning_rate": 0.00011495122752834296,
      "loss": 0.4844,
      "step": 8182
    },
    {
      "epoch": 3.8508235294117648,
      "grad_norm": 0.2667908527358947,
      "learning_rate": 0.00011490860629008333,
      "loss": 0.127,
      "step": 8183
    },
    {
      "epoch": 3.851294117647059,
      "grad_norm": 0.23728753455823048,
      "learning_rate": 0.00011486598804880088,
      "loss": 0.126,
      "step": 8184
    },
    {
      "epoch": 3.8517647058823528,
      "grad_norm": 0.33692845107159053,
      "learning_rate": 0.00011482337280813536,
      "loss": 0.1494,
      "step": 8185
    },
    {
      "epoch": 3.852235294117647,
      "grad_norm": 0.7566593044966872,
      "learning_rate": 0.00011478076057172638,
      "loss": 0.6758,
      "step": 8186
    },
    {
      "epoch": 3.852705882352941,
      "grad_norm": 0.596774241920822,
      "learning_rate": 0.00011473815134321315,
      "loss": 0.5352,
      "step": 8187
    },
    {
      "epoch": 3.8531764705882354,
      "grad_norm": 0.2463458054913577,
      "learning_rate": 0.00011469554512623477,
      "loss": 0.0786,
      "step": 8188
    },
    {
      "epoch": 3.8536470588235296,
      "grad_norm": 0.40338838480985806,
      "learning_rate": 0.00011465294192443001,
      "loss": 0.4219,
      "step": 8189
    },
    {
      "epoch": 3.8541176470588234,
      "grad_norm": 0.31669294875402887,
      "learning_rate": 0.00011461034174143736,
      "loss": 0.3242,
      "step": 8190
    },
    {
      "epoch": 3.8545882352941176,
      "grad_norm": 0.44835640375574276,
      "learning_rate": 0.00011456774458089515,
      "loss": 0.5703,
      "step": 8191
    },
    {
      "epoch": 3.855058823529412,
      "grad_norm": 0.42419828928381165,
      "learning_rate": 0.00011452515044644132,
      "loss": 0.3223,
      "step": 8192
    },
    {
      "epoch": 3.8555294117647056,
      "grad_norm": 0.41292815280490924,
      "learning_rate": 0.00011448255934171363,
      "loss": 0.1572,
      "step": 8193
    },
    {
      "epoch": 3.856,
      "grad_norm": 0.2791431192466845,
      "learning_rate": 0.00011443997127034962,
      "loss": 0.1079,
      "step": 8194
    },
    {
      "epoch": 3.856470588235294,
      "grad_norm": 0.36259763001532674,
      "learning_rate": 0.0001143973862359865,
      "loss": 0.1367,
      "step": 8195
    },
    {
      "epoch": 3.8569411764705883,
      "grad_norm": 0.2660027298745148,
      "learning_rate": 0.0001143548042422612,
      "loss": 0.125,
      "step": 8196
    },
    {
      "epoch": 3.8574117647058825,
      "grad_norm": 0.44778983071307576,
      "learning_rate": 0.0001143122252928104,
      "loss": 0.1426,
      "step": 8197
    },
    {
      "epoch": 3.8578823529411763,
      "grad_norm": 0.6526260248158829,
      "learning_rate": 0.00011426964939127063,
      "loss": 0.9609,
      "step": 8198
    },
    {
      "epoch": 3.8583529411764705,
      "grad_norm": 0.5803479530695461,
      "learning_rate": 0.00011422707654127809,
      "loss": 0.668,
      "step": 8199
    },
    {
      "epoch": 3.8588235294117648,
      "grad_norm": 0.4937413469380715,
      "learning_rate": 0.00011418450674646868,
      "loss": 0.7109,
      "step": 8200
    },
    {
      "epoch": 3.859294117647059,
      "grad_norm": 0.2524296030831058,
      "learning_rate": 0.00011414194001047806,
      "loss": 0.1138,
      "step": 8201
    },
    {
      "epoch": 3.8597647058823528,
      "grad_norm": 0.5383334102972459,
      "learning_rate": 0.00011409937633694165,
      "loss": 0.5625,
      "step": 8202
    },
    {
      "epoch": 3.860235294117647,
      "grad_norm": 0.42506920546070864,
      "learning_rate": 0.0001140568157294946,
      "loss": 0.457,
      "step": 8203
    },
    {
      "epoch": 3.860705882352941,
      "grad_norm": 0.31237210882267735,
      "learning_rate": 0.00011401425819177179,
      "loss": 0.4922,
      "step": 8204
    },
    {
      "epoch": 3.8611764705882354,
      "grad_norm": 0.4858929297073521,
      "learning_rate": 0.00011397170372740782,
      "loss": 0.3672,
      "step": 8205
    },
    {
      "epoch": 3.8616470588235297,
      "grad_norm": 0.27037154637183025,
      "learning_rate": 0.0001139291523400371,
      "loss": 0.0898,
      "step": 8206
    },
    {
      "epoch": 3.8621176470588234,
      "grad_norm": 0.448172131646266,
      "learning_rate": 0.00011388660403329368,
      "loss": 0.3398,
      "step": 8207
    },
    {
      "epoch": 3.8625882352941177,
      "grad_norm": 0.43733306083516016,
      "learning_rate": 0.0001138440588108114,
      "loss": 0.4824,
      "step": 8208
    },
    {
      "epoch": 3.863058823529412,
      "grad_norm": 0.44870697903842266,
      "learning_rate": 0.00011380151667622384,
      "loss": 0.3125,
      "step": 8209
    },
    {
      "epoch": 3.8635294117647057,
      "grad_norm": 0.4371393266727609,
      "learning_rate": 0.00011375897763316433,
      "loss": 0.3379,
      "step": 8210
    },
    {
      "epoch": 3.864,
      "grad_norm": 0.37304695220858813,
      "learning_rate": 0.00011371644168526586,
      "loss": 0.4453,
      "step": 8211
    },
    {
      "epoch": 3.864470588235294,
      "grad_norm": 0.6219505668631972,
      "learning_rate": 0.00011367390883616126,
      "loss": 0.5859,
      "step": 8212
    },
    {
      "epoch": 3.8649411764705883,
      "grad_norm": 0.33253916877107703,
      "learning_rate": 0.00011363137908948303,
      "loss": 0.332,
      "step": 8213
    },
    {
      "epoch": 3.8654117647058825,
      "grad_norm": 0.42616408945390966,
      "learning_rate": 0.00011358885244886342,
      "loss": 0.2852,
      "step": 8214
    },
    {
      "epoch": 3.8658823529411763,
      "grad_norm": 0.5248473301590127,
      "learning_rate": 0.00011354632891793437,
      "loss": 0.3926,
      "step": 8215
    },
    {
      "epoch": 3.8663529411764705,
      "grad_norm": 0.4512756338786955,
      "learning_rate": 0.00011350380850032766,
      "loss": 0.3379,
      "step": 8216
    },
    {
      "epoch": 3.8668235294117648,
      "grad_norm": 0.36263806285482536,
      "learning_rate": 0.00011346129119967473,
      "loss": 0.3809,
      "step": 8217
    },
    {
      "epoch": 3.8672941176470585,
      "grad_norm": 0.3448759285403491,
      "learning_rate": 0.0001134187770196067,
      "loss": 0.3496,
      "step": 8218
    },
    {
      "epoch": 3.8677647058823528,
      "grad_norm": 0.3079155084375058,
      "learning_rate": 0.00011337626596375458,
      "loss": 0.3828,
      "step": 8219
    },
    {
      "epoch": 3.868235294117647,
      "grad_norm": 0.24516245963824734,
      "learning_rate": 0.000113333758035749,
      "loss": 0.1846,
      "step": 8220
    },
    {
      "epoch": 3.868705882352941,
      "grad_norm": 0.5377761926434735,
      "learning_rate": 0.00011329125323922031,
      "loss": 0.4316,
      "step": 8221
    },
    {
      "epoch": 3.8691764705882354,
      "grad_norm": 0.24187920912302308,
      "learning_rate": 0.00011324875157779865,
      "loss": 0.2891,
      "step": 8222
    },
    {
      "epoch": 3.8696470588235297,
      "grad_norm": 0.6566514129742823,
      "learning_rate": 0.00011320625305511388,
      "loss": 0.5781,
      "step": 8223
    },
    {
      "epoch": 3.8701176470588234,
      "grad_norm": 0.4416814438392551,
      "learning_rate": 0.00011316375767479554,
      "loss": 0.5859,
      "step": 8224
    },
    {
      "epoch": 3.8705882352941177,
      "grad_norm": 0.41435035908897283,
      "learning_rate": 0.00011312126544047302,
      "loss": 0.3633,
      "step": 8225
    },
    {
      "epoch": 3.871058823529412,
      "grad_norm": 0.3154382810720067,
      "learning_rate": 0.0001130787763557753,
      "loss": 0.0977,
      "step": 8226
    },
    {
      "epoch": 3.8715294117647057,
      "grad_norm": 0.41108002090092177,
      "learning_rate": 0.00011303629042433118,
      "loss": 0.2734,
      "step": 8227
    },
    {
      "epoch": 3.872,
      "grad_norm": 0.44944336951295244,
      "learning_rate": 0.00011299380764976915,
      "loss": 0.3555,
      "step": 8228
    },
    {
      "epoch": 3.872470588235294,
      "grad_norm": 0.9057590234461148,
      "learning_rate": 0.00011295132803571747,
      "loss": 0.6016,
      "step": 8229
    },
    {
      "epoch": 3.8729411764705883,
      "grad_norm": 0.35214447015607964,
      "learning_rate": 0.00011290885158580411,
      "loss": 0.3164,
      "step": 8230
    },
    {
      "epoch": 3.8734117647058826,
      "grad_norm": 0.39734571525849666,
      "learning_rate": 0.00011286637830365672,
      "loss": 0.4121,
      "step": 8231
    },
    {
      "epoch": 3.8738823529411763,
      "grad_norm": 0.2794299944278845,
      "learning_rate": 0.0001128239081929028,
      "loss": 0.1299,
      "step": 8232
    },
    {
      "epoch": 3.8743529411764706,
      "grad_norm": 0.5338700241769542,
      "learning_rate": 0.00011278144125716946,
      "loss": 0.7148,
      "step": 8233
    },
    {
      "epoch": 3.8748235294117648,
      "grad_norm": 0.3525750306725601,
      "learning_rate": 0.00011273897750008358,
      "loss": 0.1572,
      "step": 8234
    },
    {
      "epoch": 3.8752941176470586,
      "grad_norm": 0.5615744691425646,
      "learning_rate": 0.0001126965169252718,
      "loss": 0.5039,
      "step": 8235
    },
    {
      "epoch": 3.875764705882353,
      "grad_norm": 0.32566040697613874,
      "learning_rate": 0.00011265405953636044,
      "loss": 0.167,
      "step": 8236
    },
    {
      "epoch": 3.876235294117647,
      "grad_norm": 0.5427413730068192,
      "learning_rate": 0.0001126116053369756,
      "loss": 0.6406,
      "step": 8237
    },
    {
      "epoch": 3.8767058823529412,
      "grad_norm": 0.3873421146269492,
      "learning_rate": 0.000112569154330743,
      "loss": 0.4785,
      "step": 8238
    },
    {
      "epoch": 3.8771764705882354,
      "grad_norm": 0.3870838581913052,
      "learning_rate": 0.00011252670652128825,
      "loss": 0.3008,
      "step": 8239
    },
    {
      "epoch": 3.8776470588235297,
      "grad_norm": 0.22314942695466217,
      "learning_rate": 0.00011248426191223658,
      "loss": 0.1865,
      "step": 8240
    },
    {
      "epoch": 3.8781176470588234,
      "grad_norm": 0.16252732596256694,
      "learning_rate": 0.00011244182050721293,
      "loss": 0.0576,
      "step": 8241
    },
    {
      "epoch": 3.8785882352941177,
      "grad_norm": 0.30271184427920905,
      "learning_rate": 0.00011239938230984198,
      "loss": 0.2119,
      "step": 8242
    },
    {
      "epoch": 3.879058823529412,
      "grad_norm": 0.2906927953142728,
      "learning_rate": 0.00011235694732374829,
      "loss": 0.0913,
      "step": 8243
    },
    {
      "epoch": 3.8795294117647057,
      "grad_norm": 0.4562269277133529,
      "learning_rate": 0.00011231451555255592,
      "loss": 0.5352,
      "step": 8244
    },
    {
      "epoch": 3.88,
      "grad_norm": 0.4473864835529579,
      "learning_rate": 0.00011227208699988879,
      "loss": 0.2539,
      "step": 8245
    },
    {
      "epoch": 3.880470588235294,
      "grad_norm": 0.38445627288435,
      "learning_rate": 0.00011222966166937044,
      "loss": 0.2773,
      "step": 8246
    },
    {
      "epoch": 3.8809411764705883,
      "grad_norm": 0.44959600253326687,
      "learning_rate": 0.00011218723956462427,
      "loss": 0.5195,
      "step": 8247
    },
    {
      "epoch": 3.8814117647058826,
      "grad_norm": 0.39885759567151335,
      "learning_rate": 0.00011214482068927332,
      "loss": 0.1895,
      "step": 8248
    },
    {
      "epoch": 3.8818823529411763,
      "grad_norm": 0.36353825730435807,
      "learning_rate": 0.00011210240504694033,
      "loss": 0.2871,
      "step": 8249
    },
    {
      "epoch": 3.8823529411764706,
      "grad_norm": 0.45433306915858457,
      "learning_rate": 0.00011205999264124786,
      "loss": 0.5781,
      "step": 8250
    },
    {
      "epoch": 3.882823529411765,
      "grad_norm": 0.3417529651041211,
      "learning_rate": 0.00011201758347581814,
      "loss": 0.1289,
      "step": 8251
    },
    {
      "epoch": 3.8832941176470586,
      "grad_norm": 0.3383967185886712,
      "learning_rate": 0.00011197517755427303,
      "loss": 0.1572,
      "step": 8252
    },
    {
      "epoch": 3.883764705882353,
      "grad_norm": 0.4239448707528862,
      "learning_rate": 0.00011193277488023432,
      "loss": 0.4512,
      "step": 8253
    },
    {
      "epoch": 3.884235294117647,
      "grad_norm": 0.35028799534971555,
      "learning_rate": 0.00011189037545732335,
      "loss": 0.2129,
      "step": 8254
    },
    {
      "epoch": 3.8847058823529412,
      "grad_norm": 0.49175210075514086,
      "learning_rate": 0.00011184797928916124,
      "loss": 0.2314,
      "step": 8255
    },
    {
      "epoch": 3.8851764705882355,
      "grad_norm": 0.627566938582205,
      "learning_rate": 0.0001118055863793688,
      "loss": 0.5898,
      "step": 8256
    },
    {
      "epoch": 3.8856470588235297,
      "grad_norm": 0.31933094406208995,
      "learning_rate": 0.00011176319673156666,
      "loss": 0.2949,
      "step": 8257
    },
    {
      "epoch": 3.8861176470588235,
      "grad_norm": 0.6720126901624621,
      "learning_rate": 0.00011172081034937507,
      "loss": 0.4473,
      "step": 8258
    },
    {
      "epoch": 3.8865882352941177,
      "grad_norm": 0.22827480804462752,
      "learning_rate": 0.00011167842723641398,
      "loss": 0.0684,
      "step": 8259
    },
    {
      "epoch": 3.887058823529412,
      "grad_norm": 0.5976241321858007,
      "learning_rate": 0.00011163604739630324,
      "loss": 0.334,
      "step": 8260
    },
    {
      "epoch": 3.8875294117647057,
      "grad_norm": 0.6161719019480573,
      "learning_rate": 0.00011159367083266219,
      "loss": 0.4336,
      "step": 8261
    },
    {
      "epoch": 3.888,
      "grad_norm": 0.4390477893249706,
      "learning_rate": 0.00011155129754911002,
      "loss": 0.377,
      "step": 8262
    },
    {
      "epoch": 3.888470588235294,
      "grad_norm": 0.19957403787045566,
      "learning_rate": 0.00011150892754926568,
      "loss": 0.1895,
      "step": 8263
    },
    {
      "epoch": 3.8889411764705883,
      "grad_norm": 0.39895916252206526,
      "learning_rate": 0.00011146656083674768,
      "loss": 0.2559,
      "step": 8264
    },
    {
      "epoch": 3.8894117647058826,
      "grad_norm": 0.3501357061087468,
      "learning_rate": 0.00011142419741517443,
      "loss": 0.3711,
      "step": 8265
    },
    {
      "epoch": 3.8898823529411763,
      "grad_norm": 0.3879650558207739,
      "learning_rate": 0.00011138183728816389,
      "loss": 0.4824,
      "step": 8266
    },
    {
      "epoch": 3.8903529411764706,
      "grad_norm": 0.5050724441429387,
      "learning_rate": 0.0001113394804593339,
      "loss": 0.5898,
      "step": 8267
    },
    {
      "epoch": 3.890823529411765,
      "grad_norm": 0.46516367428704725,
      "learning_rate": 0.00011129712693230191,
      "loss": 0.2793,
      "step": 8268
    },
    {
      "epoch": 3.8912941176470586,
      "grad_norm": 0.47144968611613863,
      "learning_rate": 0.00011125477671068506,
      "loss": 0.3379,
      "step": 8269
    },
    {
      "epoch": 3.891764705882353,
      "grad_norm": 0.5170958598183116,
      "learning_rate": 0.0001112124297981004,
      "loss": 0.332,
      "step": 8270
    },
    {
      "epoch": 3.892235294117647,
      "grad_norm": 0.2803394759687192,
      "learning_rate": 0.00011117008619816445,
      "loss": 0.2256,
      "step": 8271
    },
    {
      "epoch": 3.8927058823529412,
      "grad_norm": 0.18319423352301534,
      "learning_rate": 0.00011112774591449356,
      "loss": 0.0476,
      "step": 8272
    },
    {
      "epoch": 3.8931764705882355,
      "grad_norm": 0.71167996850897,
      "learning_rate": 0.00011108540895070388,
      "loss": 0.6914,
      "step": 8273
    },
    {
      "epoch": 3.8936470588235297,
      "grad_norm": 0.14142527980311795,
      "learning_rate": 0.00011104307531041117,
      "loss": 0.0322,
      "step": 8274
    },
    {
      "epoch": 3.8941176470588235,
      "grad_norm": 0.5539537634658405,
      "learning_rate": 0.0001110007449972309,
      "loss": 0.6797,
      "step": 8275
    },
    {
      "epoch": 3.8945882352941177,
      "grad_norm": 0.35335582365158696,
      "learning_rate": 0.00011095841801477826,
      "loss": 0.1816,
      "step": 8276
    },
    {
      "epoch": 3.895058823529412,
      "grad_norm": 0.43511553319294094,
      "learning_rate": 0.00011091609436666825,
      "loss": 0.3965,
      "step": 8277
    },
    {
      "epoch": 3.8955294117647057,
      "grad_norm": 0.42392836544626417,
      "learning_rate": 0.00011087377405651549,
      "loss": 0.375,
      "step": 8278
    },
    {
      "epoch": 3.896,
      "grad_norm": 0.45779951312869893,
      "learning_rate": 0.00011083145708793431,
      "loss": 0.4609,
      "step": 8279
    },
    {
      "epoch": 3.896470588235294,
      "grad_norm": 0.4485617024655421,
      "learning_rate": 0.00011078914346453883,
      "loss": 0.4941,
      "step": 8280
    },
    {
      "epoch": 3.8969411764705884,
      "grad_norm": 0.636482193712555,
      "learning_rate": 0.00011074683318994286,
      "loss": 0.1318,
      "step": 8281
    },
    {
      "epoch": 3.8974117647058826,
      "grad_norm": 0.3571544646686622,
      "learning_rate": 0.00011070452626775981,
      "loss": 0.2002,
      "step": 8282
    },
    {
      "epoch": 3.8978823529411764,
      "grad_norm": 0.36989852601186984,
      "learning_rate": 0.000110662222701603,
      "loss": 0.4219,
      "step": 8283
    },
    {
      "epoch": 3.8983529411764706,
      "grad_norm": 0.5883391268158749,
      "learning_rate": 0.00011061992249508533,
      "loss": 0.3809,
      "step": 8284
    },
    {
      "epoch": 3.898823529411765,
      "grad_norm": 0.6300400328470785,
      "learning_rate": 0.0001105776256518194,
      "loss": 0.5703,
      "step": 8285
    },
    {
      "epoch": 3.8992941176470586,
      "grad_norm": 0.4591412729567924,
      "learning_rate": 0.0001105353321754176,
      "loss": 0.2197,
      "step": 8286
    },
    {
      "epoch": 3.899764705882353,
      "grad_norm": 0.4088815330513288,
      "learning_rate": 0.000110493042069492,
      "loss": 0.3438,
      "step": 8287
    },
    {
      "epoch": 3.900235294117647,
      "grad_norm": 0.2868974790676431,
      "learning_rate": 0.00011045075533765444,
      "loss": 0.334,
      "step": 8288
    },
    {
      "epoch": 3.9007058823529412,
      "grad_norm": 0.4111426622090268,
      "learning_rate": 0.00011040847198351635,
      "loss": 0.3184,
      "step": 8289
    },
    {
      "epoch": 3.9011764705882355,
      "grad_norm": 0.4293055771319868,
      "learning_rate": 0.00011036619201068892,
      "loss": 0.3164,
      "step": 8290
    },
    {
      "epoch": 3.9016470588235292,
      "grad_norm": 0.533545520565058,
      "learning_rate": 0.00011032391542278313,
      "loss": 0.4316,
      "step": 8291
    },
    {
      "epoch": 3.9021176470588235,
      "grad_norm": 0.491634277344488,
      "learning_rate": 0.00011028164222340957,
      "loss": 0.5312,
      "step": 8292
    },
    {
      "epoch": 3.9025882352941177,
      "grad_norm": 0.2706623593910742,
      "learning_rate": 0.00011023937241617853,
      "loss": 0.0669,
      "step": 8293
    },
    {
      "epoch": 3.903058823529412,
      "grad_norm": 0.5208909614661651,
      "learning_rate": 0.00011019710600470016,
      "loss": 0.3203,
      "step": 8294
    },
    {
      "epoch": 3.9035294117647057,
      "grad_norm": 0.11630342051235362,
      "learning_rate": 0.00011015484299258417,
      "loss": 0.0525,
      "step": 8295
    },
    {
      "epoch": 3.904,
      "grad_norm": 0.5338622022208757,
      "learning_rate": 0.00011011258338344002,
      "loss": 0.5664,
      "step": 8296
    },
    {
      "epoch": 3.904470588235294,
      "grad_norm": 0.3040995206000187,
      "learning_rate": 0.00011007032718087688,
      "loss": 0.1553,
      "step": 8297
    },
    {
      "epoch": 3.9049411764705884,
      "grad_norm": 0.4387680812360006,
      "learning_rate": 0.00011002807438850367,
      "loss": 0.1982,
      "step": 8298
    },
    {
      "epoch": 3.9054117647058826,
      "grad_norm": 0.5678918279729027,
      "learning_rate": 0.00010998582500992897,
      "loss": 0.4883,
      "step": 8299
    },
    {
      "epoch": 3.9058823529411764,
      "grad_norm": 0.3778048662334209,
      "learning_rate": 0.00010994357904876106,
      "loss": 0.1846,
      "step": 8300
    },
    {
      "epoch": 3.9063529411764706,
      "grad_norm": 0.5402893207323003,
      "learning_rate": 0.000109901336508608,
      "loss": 0.6836,
      "step": 8301
    },
    {
      "epoch": 3.906823529411765,
      "grad_norm": 0.3515436868094116,
      "learning_rate": 0.00010985909739307748,
      "loss": 0.3945,
      "step": 8302
    },
    {
      "epoch": 3.9072941176470586,
      "grad_norm": 0.33335269292694947,
      "learning_rate": 0.0001098168617057769,
      "loss": 0.2002,
      "step": 8303
    },
    {
      "epoch": 3.907764705882353,
      "grad_norm": 0.3256391587820496,
      "learning_rate": 0.00010977462945031347,
      "loss": 0.1211,
      "step": 8304
    },
    {
      "epoch": 3.908235294117647,
      "grad_norm": 0.31650802035274817,
      "learning_rate": 0.00010973240063029398,
      "loss": 0.1709,
      "step": 8305
    },
    {
      "epoch": 3.9087058823529413,
      "grad_norm": 0.6407504775958877,
      "learning_rate": 0.000109690175249325,
      "loss": 0.6211,
      "step": 8306
    },
    {
      "epoch": 3.9091764705882355,
      "grad_norm": 0.5630969233705267,
      "learning_rate": 0.00010964795331101273,
      "loss": 0.3262,
      "step": 8307
    },
    {
      "epoch": 3.9096470588235293,
      "grad_norm": 0.4428392852483625,
      "learning_rate": 0.00010960573481896322,
      "loss": 0.4609,
      "step": 8308
    },
    {
      "epoch": 3.9101176470588235,
      "grad_norm": 0.39144829090675837,
      "learning_rate": 0.00010956351977678206,
      "loss": 0.2295,
      "step": 8309
    },
    {
      "epoch": 3.9105882352941177,
      "grad_norm": 0.4422917115108659,
      "learning_rate": 0.00010952130818807465,
      "loss": 0.1875,
      "step": 8310
    },
    {
      "epoch": 3.911058823529412,
      "grad_norm": 0.5199945502118786,
      "learning_rate": 0.00010947910005644608,
      "loss": 0.6133,
      "step": 8311
    },
    {
      "epoch": 3.9115294117647057,
      "grad_norm": 0.40872635272598123,
      "learning_rate": 0.00010943689538550111,
      "loss": 0.3145,
      "step": 8312
    },
    {
      "epoch": 3.912,
      "grad_norm": 0.4866170767036623,
      "learning_rate": 0.00010939469417884424,
      "loss": 0.1406,
      "step": 8313
    },
    {
      "epoch": 3.912470588235294,
      "grad_norm": 0.27056409046213503,
      "learning_rate": 0.00010935249644007966,
      "loss": 0.0889,
      "step": 8314
    },
    {
      "epoch": 3.9129411764705884,
      "grad_norm": 0.36623553431287903,
      "learning_rate": 0.00010931030217281125,
      "loss": 0.4375,
      "step": 8315
    },
    {
      "epoch": 3.9134117647058826,
      "grad_norm": 0.44880569168615814,
      "learning_rate": 0.00010926811138064264,
      "loss": 0.4434,
      "step": 8316
    },
    {
      "epoch": 3.9138823529411764,
      "grad_norm": 0.23700929780980645,
      "learning_rate": 0.00010922592406717706,
      "loss": 0.1533,
      "step": 8317
    },
    {
      "epoch": 3.9143529411764706,
      "grad_norm": 0.2945438011724205,
      "learning_rate": 0.0001091837402360176,
      "loss": 0.3418,
      "step": 8318
    },
    {
      "epoch": 3.914823529411765,
      "grad_norm": 0.2999817599916257,
      "learning_rate": 0.00010914155989076692,
      "loss": 0.2949,
      "step": 8319
    },
    {
      "epoch": 3.9152941176470586,
      "grad_norm": 0.37654077987284945,
      "learning_rate": 0.00010909938303502741,
      "loss": 0.3574,
      "step": 8320
    },
    {
      "epoch": 3.915764705882353,
      "grad_norm": 0.30854329904713546,
      "learning_rate": 0.00010905720967240122,
      "loss": 0.103,
      "step": 8321
    },
    {
      "epoch": 3.916235294117647,
      "grad_norm": 0.38172212341104994,
      "learning_rate": 0.00010901503980649017,
      "loss": 0.1934,
      "step": 8322
    },
    {
      "epoch": 3.9167058823529413,
      "grad_norm": 0.6232279455694623,
      "learning_rate": 0.0001089728734408957,
      "loss": 0.875,
      "step": 8323
    },
    {
      "epoch": 3.9171764705882355,
      "grad_norm": 0.26393785576358075,
      "learning_rate": 0.00010893071057921911,
      "loss": 0.2031,
      "step": 8324
    },
    {
      "epoch": 3.9176470588235293,
      "grad_norm": 0.824832151312964,
      "learning_rate": 0.0001088885512250613,
      "loss": 0.5391,
      "step": 8325
    },
    {
      "epoch": 3.9181176470588235,
      "grad_norm": 0.5924457268440843,
      "learning_rate": 0.00010884639538202285,
      "loss": 0.4941,
      "step": 8326
    },
    {
      "epoch": 3.9185882352941177,
      "grad_norm": 0.47576728413005487,
      "learning_rate": 0.00010880424305370406,
      "loss": 0.3438,
      "step": 8327
    },
    {
      "epoch": 3.919058823529412,
      "grad_norm": 0.5466950020555206,
      "learning_rate": 0.00010876209424370502,
      "loss": 0.3203,
      "step": 8328
    },
    {
      "epoch": 3.9195294117647057,
      "grad_norm": 0.4214208896672667,
      "learning_rate": 0.00010871994895562538,
      "loss": 0.3496,
      "step": 8329
    },
    {
      "epoch": 3.92,
      "grad_norm": 0.3265696404448003,
      "learning_rate": 0.00010867780719306458,
      "loss": 0.1147,
      "step": 8330
    },
    {
      "epoch": 3.920470588235294,
      "grad_norm": 0.5050217778119225,
      "learning_rate": 0.00010863566895962166,
      "loss": 0.3926,
      "step": 8331
    },
    {
      "epoch": 3.9209411764705884,
      "grad_norm": 0.13667404861235777,
      "learning_rate": 0.0001085935342588956,
      "loss": 0.0189,
      "step": 8332
    },
    {
      "epoch": 3.9214117647058826,
      "grad_norm": 0.43016105454921677,
      "learning_rate": 0.00010855140309448479,
      "loss": 0.3613,
      "step": 8333
    },
    {
      "epoch": 3.9218823529411764,
      "grad_norm": 0.47578264665386216,
      "learning_rate": 0.00010850927546998744,
      "loss": 0.4648,
      "step": 8334
    },
    {
      "epoch": 3.9223529411764706,
      "grad_norm": 0.26299056138911414,
      "learning_rate": 0.00010846715138900148,
      "loss": 0.0527,
      "step": 8335
    },
    {
      "epoch": 3.922823529411765,
      "grad_norm": 0.4435022115684526,
      "learning_rate": 0.00010842503085512453,
      "loss": 0.6055,
      "step": 8336
    },
    {
      "epoch": 3.9232941176470586,
      "grad_norm": 0.5598183472826896,
      "learning_rate": 0.00010838291387195387,
      "loss": 0.7773,
      "step": 8337
    },
    {
      "epoch": 3.923764705882353,
      "grad_norm": 0.5405880743957698,
      "learning_rate": 0.00010834080044308646,
      "loss": 0.416,
      "step": 8338
    },
    {
      "epoch": 3.924235294117647,
      "grad_norm": 0.3667131702323656,
      "learning_rate": 0.00010829869057211906,
      "loss": 0.2324,
      "step": 8339
    },
    {
      "epoch": 3.9247058823529413,
      "grad_norm": 0.2723786709392717,
      "learning_rate": 0.00010825658426264803,
      "loss": 0.0874,
      "step": 8340
    },
    {
      "epoch": 3.9251764705882355,
      "grad_norm": 0.34954286085911507,
      "learning_rate": 0.00010821448151826941,
      "loss": 0.3398,
      "step": 8341
    },
    {
      "epoch": 3.9256470588235293,
      "grad_norm": 0.601956751272047,
      "learning_rate": 0.00010817238234257907,
      "loss": 0.6953,
      "step": 8342
    },
    {
      "epoch": 3.9261176470588235,
      "grad_norm": 0.5002249370954869,
      "learning_rate": 0.00010813028673917245,
      "loss": 0.252,
      "step": 8343
    },
    {
      "epoch": 3.9265882352941177,
      "grad_norm": 0.32240589800578984,
      "learning_rate": 0.0001080881947116447,
      "loss": 0.127,
      "step": 8344
    },
    {
      "epoch": 3.9270588235294115,
      "grad_norm": 0.4438671586313677,
      "learning_rate": 0.00010804610626359068,
      "loss": 0.2432,
      "step": 8345
    },
    {
      "epoch": 3.9275294117647057,
      "grad_norm": 0.3161707614865821,
      "learning_rate": 0.000108004021398605,
      "loss": 0.4062,
      "step": 8346
    },
    {
      "epoch": 3.928,
      "grad_norm": 0.6613237572132705,
      "learning_rate": 0.00010796194012028185,
      "loss": 0.8438,
      "step": 8347
    },
    {
      "epoch": 3.928470588235294,
      "grad_norm": 0.2549003393092343,
      "learning_rate": 0.00010791986243221521,
      "loss": 0.0693,
      "step": 8348
    },
    {
      "epoch": 3.9289411764705884,
      "grad_norm": 0.8563238367005754,
      "learning_rate": 0.00010787778833799875,
      "loss": 0.6289,
      "step": 8349
    },
    {
      "epoch": 3.9294117647058826,
      "grad_norm": 0.4621788842979162,
      "learning_rate": 0.00010783571784122577,
      "loss": 0.1553,
      "step": 8350
    },
    {
      "epoch": 3.9298823529411764,
      "grad_norm": 0.6185545668529792,
      "learning_rate": 0.00010779365094548926,
      "loss": 0.6289,
      "step": 8351
    },
    {
      "epoch": 3.9303529411764706,
      "grad_norm": 0.5790177313734785,
      "learning_rate": 0.00010775158765438202,
      "loss": 0.7227,
      "step": 8352
    },
    {
      "epoch": 3.930823529411765,
      "grad_norm": 0.1524565485143802,
      "learning_rate": 0.0001077095279714964,
      "loss": 0.0273,
      "step": 8353
    },
    {
      "epoch": 3.9312941176470586,
      "grad_norm": 0.3692271189725015,
      "learning_rate": 0.00010766747190042452,
      "loss": 0.1738,
      "step": 8354
    },
    {
      "epoch": 3.931764705882353,
      "grad_norm": 0.29891373372762203,
      "learning_rate": 0.00010762541944475821,
      "loss": 0.1953,
      "step": 8355
    },
    {
      "epoch": 3.932235294117647,
      "grad_norm": 0.5349274598980471,
      "learning_rate": 0.0001075833706080889,
      "loss": 0.4004,
      "step": 8356
    },
    {
      "epoch": 3.9327058823529413,
      "grad_norm": 0.36813677602708605,
      "learning_rate": 0.00010754132539400782,
      "loss": 0.2266,
      "step": 8357
    },
    {
      "epoch": 3.9331764705882355,
      "grad_norm": 0.4238028108632255,
      "learning_rate": 0.00010749928380610577,
      "loss": 0.4375,
      "step": 8358
    },
    {
      "epoch": 3.9336470588235293,
      "grad_norm": 0.46058375085340925,
      "learning_rate": 0.00010745724584797338,
      "loss": 0.4297,
      "step": 8359
    },
    {
      "epoch": 3.9341176470588235,
      "grad_norm": 0.6809786971125719,
      "learning_rate": 0.00010741521152320088,
      "loss": 0.6602,
      "step": 8360
    },
    {
      "epoch": 3.9345882352941177,
      "grad_norm": 0.6814717737286262,
      "learning_rate": 0.00010737318083537813,
      "loss": 0.7539,
      "step": 8361
    },
    {
      "epoch": 3.9350588235294115,
      "grad_norm": 0.35501676612965305,
      "learning_rate": 0.00010733115378809485,
      "loss": 0.5391,
      "step": 8362
    },
    {
      "epoch": 3.9355294117647057,
      "grad_norm": 0.3899847590114751,
      "learning_rate": 0.00010728913038494038,
      "loss": 0.2002,
      "step": 8363
    },
    {
      "epoch": 3.936,
      "grad_norm": 0.7265594370057681,
      "learning_rate": 0.00010724711062950358,
      "loss": 0.6289,
      "step": 8364
    },
    {
      "epoch": 3.936470588235294,
      "grad_norm": 0.8388474331906222,
      "learning_rate": 0.00010720509452537329,
      "loss": 0.6328,
      "step": 8365
    },
    {
      "epoch": 3.9369411764705884,
      "grad_norm": 0.3021373763961164,
      "learning_rate": 0.00010716308207613784,
      "loss": 0.4121,
      "step": 8366
    },
    {
      "epoch": 3.9374117647058826,
      "grad_norm": 0.44583910704694524,
      "learning_rate": 0.00010712107328538528,
      "loss": 0.3965,
      "step": 8367
    },
    {
      "epoch": 3.9378823529411764,
      "grad_norm": 0.5608880547648579,
      "learning_rate": 0.00010707906815670337,
      "loss": 0.4883,
      "step": 8368
    },
    {
      "epoch": 3.9383529411764706,
      "grad_norm": 0.21252282086161722,
      "learning_rate": 0.00010703706669367956,
      "loss": 0.1128,
      "step": 8369
    },
    {
      "epoch": 3.938823529411765,
      "grad_norm": 1.4821714398034935,
      "learning_rate": 0.00010699506889990102,
      "loss": 1.1172,
      "step": 8370
    },
    {
      "epoch": 3.9392941176470586,
      "grad_norm": 0.3445870452063049,
      "learning_rate": 0.00010695307477895449,
      "loss": 0.2285,
      "step": 8371
    },
    {
      "epoch": 3.939764705882353,
      "grad_norm": 0.26622936397004293,
      "learning_rate": 0.00010691108433442656,
      "loss": 0.0967,
      "step": 8372
    },
    {
      "epoch": 3.940235294117647,
      "grad_norm": 0.7960415398071258,
      "learning_rate": 0.00010686909756990337,
      "loss": 0.8281,
      "step": 8373
    },
    {
      "epoch": 3.9407058823529413,
      "grad_norm": 0.5364076647265258,
      "learning_rate": 0.00010682711448897077,
      "loss": 0.498,
      "step": 8374
    },
    {
      "epoch": 3.9411764705882355,
      "grad_norm": 0.45025295729830594,
      "learning_rate": 0.00010678513509521435,
      "loss": 0.1875,
      "step": 8375
    },
    {
      "epoch": 3.9416470588235293,
      "grad_norm": 0.42449185417557644,
      "learning_rate": 0.00010674315939221937,
      "loss": 0.2949,
      "step": 8376
    },
    {
      "epoch": 3.9421176470588235,
      "grad_norm": 0.406859767976578,
      "learning_rate": 0.00010670118738357078,
      "loss": 0.3242,
      "step": 8377
    },
    {
      "epoch": 3.9425882352941177,
      "grad_norm": 0.3592007678114134,
      "learning_rate": 0.00010665921907285317,
      "loss": 0.168,
      "step": 8378
    },
    {
      "epoch": 3.9430588235294115,
      "grad_norm": 0.3856532882714457,
      "learning_rate": 0.0001066172544636508,
      "loss": 0.2402,
      "step": 8379
    },
    {
      "epoch": 3.9435294117647057,
      "grad_norm": 0.2874947212769203,
      "learning_rate": 0.00010657529355954771,
      "loss": 0.166,
      "step": 8380
    },
    {
      "epoch": 3.944,
      "grad_norm": 0.5448147445891437,
      "learning_rate": 0.00010653333636412758,
      "loss": 0.6836,
      "step": 8381
    },
    {
      "epoch": 3.944470588235294,
      "grad_norm": 0.4314290209591104,
      "learning_rate": 0.00010649138288097366,
      "loss": 0.4277,
      "step": 8382
    },
    {
      "epoch": 3.9449411764705884,
      "grad_norm": 0.44340279889641776,
      "learning_rate": 0.00010644943311366911,
      "loss": 0.3301,
      "step": 8383
    },
    {
      "epoch": 3.9454117647058826,
      "grad_norm": 0.3309358762950234,
      "learning_rate": 0.00010640748706579658,
      "loss": 0.3906,
      "step": 8384
    },
    {
      "epoch": 3.9458823529411764,
      "grad_norm": 0.561200140025329,
      "learning_rate": 0.00010636554474093846,
      "loss": 0.7148,
      "step": 8385
    },
    {
      "epoch": 3.9463529411764706,
      "grad_norm": 0.3806828635646742,
      "learning_rate": 0.00010632360614267683,
      "loss": 0.5938,
      "step": 8386
    },
    {
      "epoch": 3.946823529411765,
      "grad_norm": 0.2064799600613145,
      "learning_rate": 0.0001062816712745935,
      "loss": 0.1621,
      "step": 8387
    },
    {
      "epoch": 3.9472941176470586,
      "grad_norm": 0.3746705706907731,
      "learning_rate": 0.00010623974014026986,
      "loss": 0.2129,
      "step": 8388
    },
    {
      "epoch": 3.947764705882353,
      "grad_norm": 1.9482876796906357,
      "learning_rate": 0.00010619781274328703,
      "loss": 0.3164,
      "step": 8389
    },
    {
      "epoch": 3.948235294117647,
      "grad_norm": 1.0486127855325742,
      "learning_rate": 0.00010615588908722588,
      "loss": 0.7383,
      "step": 8390
    },
    {
      "epoch": 3.9487058823529413,
      "grad_norm": 0.686832417695235,
      "learning_rate": 0.00010611396917566685,
      "loss": 0.1826,
      "step": 8391
    },
    {
      "epoch": 3.9491764705882355,
      "grad_norm": 0.34115168617588265,
      "learning_rate": 0.00010607205301219008,
      "loss": 0.2773,
      "step": 8392
    },
    {
      "epoch": 3.9496470588235293,
      "grad_norm": 0.33286573210501624,
      "learning_rate": 0.00010603014060037549,
      "loss": 0.1533,
      "step": 8393
    },
    {
      "epoch": 3.9501176470588235,
      "grad_norm": 0.42956943016766563,
      "learning_rate": 0.00010598823194380256,
      "loss": 0.543,
      "step": 8394
    },
    {
      "epoch": 3.9505882352941177,
      "grad_norm": 0.5700246284153211,
      "learning_rate": 0.0001059463270460505,
      "loss": 0.7344,
      "step": 8395
    },
    {
      "epoch": 3.9510588235294115,
      "grad_norm": 0.3563036566460102,
      "learning_rate": 0.00010590442591069816,
      "loss": 0.2178,
      "step": 8396
    },
    {
      "epoch": 3.9515294117647057,
      "grad_norm": 0.4042260586658323,
      "learning_rate": 0.00010586252854132415,
      "loss": 0.207,
      "step": 8397
    },
    {
      "epoch": 3.952,
      "grad_norm": 0.26276214097696254,
      "learning_rate": 0.0001058206349415067,
      "loss": 0.0864,
      "step": 8398
    },
    {
      "epoch": 3.952470588235294,
      "grad_norm": 0.21302494345229656,
      "learning_rate": 0.00010577874511482371,
      "loss": 0.1631,
      "step": 8399
    },
    {
      "epoch": 3.9529411764705884,
      "grad_norm": 0.3286407225621567,
      "learning_rate": 0.00010573685906485282,
      "loss": 0.4062,
      "step": 8400
    },
    {
      "epoch": 3.9534117647058826,
      "grad_norm": 0.28604464924933193,
      "learning_rate": 0.00010569497679517126,
      "loss": 0.3359,
      "step": 8401
    },
    {
      "epoch": 3.9538823529411764,
      "grad_norm": 0.21827584764495428,
      "learning_rate": 0.00010565309830935599,
      "loss": 0.062,
      "step": 8402
    },
    {
      "epoch": 3.9543529411764706,
      "grad_norm": 0.2919439218960125,
      "learning_rate": 0.00010561122361098365,
      "loss": 0.3008,
      "step": 8403
    },
    {
      "epoch": 3.954823529411765,
      "grad_norm": 0.28554513921446695,
      "learning_rate": 0.00010556935270363055,
      "loss": 0.0825,
      "step": 8404
    },
    {
      "epoch": 3.9552941176470586,
      "grad_norm": 0.8574805864589974,
      "learning_rate": 0.00010552748559087261,
      "loss": 0.5781,
      "step": 8405
    },
    {
      "epoch": 3.955764705882353,
      "grad_norm": 0.3817832522318883,
      "learning_rate": 0.00010548562227628559,
      "loss": 0.2041,
      "step": 8406
    },
    {
      "epoch": 3.956235294117647,
      "grad_norm": 0.4010561092555717,
      "learning_rate": 0.00010544376276344477,
      "loss": 0.2715,
      "step": 8407
    },
    {
      "epoch": 3.9567058823529413,
      "grad_norm": 0.5500668045841626,
      "learning_rate": 0.00010540190705592516,
      "loss": 0.5977,
      "step": 8408
    },
    {
      "epoch": 3.9571764705882355,
      "grad_norm": 0.299508173906333,
      "learning_rate": 0.00010536005515730141,
      "loss": 0.1128,
      "step": 8409
    },
    {
      "epoch": 3.9576470588235293,
      "grad_norm": 0.46183332990194725,
      "learning_rate": 0.00010531820707114793,
      "loss": 0.459,
      "step": 8410
    },
    {
      "epoch": 3.9581176470588235,
      "grad_norm": 0.5439515427518021,
      "learning_rate": 0.00010527636280103872,
      "loss": 0.4023,
      "step": 8411
    },
    {
      "epoch": 3.9585882352941177,
      "grad_norm": 0.2841071532883588,
      "learning_rate": 0.00010523452235054746,
      "loss": 0.1182,
      "step": 8412
    },
    {
      "epoch": 3.9590588235294115,
      "grad_norm": 0.43911432269266143,
      "learning_rate": 0.00010519268572324763,
      "loss": 0.4238,
      "step": 8413
    },
    {
      "epoch": 3.9595294117647057,
      "grad_norm": 0.17624939736581718,
      "learning_rate": 0.00010515085292271219,
      "loss": 0.0312,
      "step": 8414
    },
    {
      "epoch": 3.96,
      "grad_norm": 0.3621796969188557,
      "learning_rate": 0.00010510902395251387,
      "loss": 0.1992,
      "step": 8415
    },
    {
      "epoch": 3.960470588235294,
      "grad_norm": 0.2694597212972377,
      "learning_rate": 0.00010506719881622514,
      "loss": 0.0586,
      "step": 8416
    },
    {
      "epoch": 3.9609411764705884,
      "grad_norm": 0.622026344366122,
      "learning_rate": 0.00010502537751741801,
      "loss": 0.4336,
      "step": 8417
    },
    {
      "epoch": 3.961411764705882,
      "grad_norm": 0.4381289172696951,
      "learning_rate": 0.00010498356005966422,
      "loss": 0.4902,
      "step": 8418
    },
    {
      "epoch": 3.9618823529411764,
      "grad_norm": 0.37745950092182107,
      "learning_rate": 0.00010494174644653518,
      "loss": 0.1089,
      "step": 8419
    },
    {
      "epoch": 3.9623529411764706,
      "grad_norm": 0.4714044741701307,
      "learning_rate": 0.000104899936681602,
      "loss": 0.4082,
      "step": 8420
    },
    {
      "epoch": 3.962823529411765,
      "grad_norm": 0.4682391298942902,
      "learning_rate": 0.0001048581307684355,
      "loss": 0.3965,
      "step": 8421
    },
    {
      "epoch": 3.9632941176470586,
      "grad_norm": 0.8957719137142348,
      "learning_rate": 0.00010481632871060605,
      "loss": 1.0078,
      "step": 8422
    },
    {
      "epoch": 3.963764705882353,
      "grad_norm": 0.47061494232373907,
      "learning_rate": 0.0001047745305116837,
      "loss": 0.5273,
      "step": 8423
    },
    {
      "epoch": 3.964235294117647,
      "grad_norm": 0.33969333941878593,
      "learning_rate": 0.00010473273617523831,
      "loss": 0.2637,
      "step": 8424
    },
    {
      "epoch": 3.9647058823529413,
      "grad_norm": 0.5275929355234972,
      "learning_rate": 0.00010469094570483928,
      "loss": 0.2773,
      "step": 8425
    },
    {
      "epoch": 3.9651764705882355,
      "grad_norm": 0.6751202544144489,
      "learning_rate": 0.00010464915910405576,
      "loss": 0.6406,
      "step": 8426
    },
    {
      "epoch": 3.9656470588235293,
      "grad_norm": 0.24215696497145056,
      "learning_rate": 0.00010460737637645645,
      "loss": 0.1963,
      "step": 8427
    },
    {
      "epoch": 3.9661176470588235,
      "grad_norm": 0.34844790819140253,
      "learning_rate": 0.00010456559752560987,
      "loss": 0.3535,
      "step": 8428
    },
    {
      "epoch": 3.9665882352941177,
      "grad_norm": 0.32144745017667753,
      "learning_rate": 0.00010452382255508412,
      "loss": 0.0942,
      "step": 8429
    },
    {
      "epoch": 3.9670588235294115,
      "grad_norm": 0.47391526589605937,
      "learning_rate": 0.00010448205146844696,
      "loss": 0.3535,
      "step": 8430
    },
    {
      "epoch": 3.9675294117647057,
      "grad_norm": 0.377053805626841,
      "learning_rate": 0.00010444028426926588,
      "loss": 0.2373,
      "step": 8431
    },
    {
      "epoch": 3.968,
      "grad_norm": 0.2678975048445155,
      "learning_rate": 0.000104398520961108,
      "loss": 0.127,
      "step": 8432
    },
    {
      "epoch": 3.968470588235294,
      "grad_norm": 0.3528323959936717,
      "learning_rate": 0.00010435676154754009,
      "loss": 0.25,
      "step": 8433
    },
    {
      "epoch": 3.9689411764705884,
      "grad_norm": 0.37351798192616464,
      "learning_rate": 0.00010431500603212863,
      "loss": 0.2178,
      "step": 8434
    },
    {
      "epoch": 3.969411764705882,
      "grad_norm": 0.36910283648090736,
      "learning_rate": 0.00010427325441843976,
      "loss": 0.4277,
      "step": 8435
    },
    {
      "epoch": 3.9698823529411764,
      "grad_norm": 0.41074453569401775,
      "learning_rate": 0.00010423150671003924,
      "loss": 0.2246,
      "step": 8436
    },
    {
      "epoch": 3.9703529411764706,
      "grad_norm": 0.3095182558076142,
      "learning_rate": 0.00010418976291049251,
      "loss": 0.3125,
      "step": 8437
    },
    {
      "epoch": 3.970823529411765,
      "grad_norm": 0.571886845305298,
      "learning_rate": 0.00010414802302336474,
      "loss": 0.3047,
      "step": 8438
    },
    {
      "epoch": 3.9712941176470586,
      "grad_norm": 0.49027921867503776,
      "learning_rate": 0.00010410628705222073,
      "loss": 0.4707,
      "step": 8439
    },
    {
      "epoch": 3.971764705882353,
      "grad_norm": 0.4378125756215407,
      "learning_rate": 0.00010406455500062487,
      "loss": 0.5078,
      "step": 8440
    },
    {
      "epoch": 3.972235294117647,
      "grad_norm": 0.49270887587839274,
      "learning_rate": 0.00010402282687214137,
      "loss": 0.4219,
      "step": 8441
    },
    {
      "epoch": 3.9727058823529413,
      "grad_norm": 0.4183176664639366,
      "learning_rate": 0.00010398110267033396,
      "loss": 0.4707,
      "step": 8442
    },
    {
      "epoch": 3.9731764705882355,
      "grad_norm": 0.4118303661266202,
      "learning_rate": 0.00010393938239876608,
      "loss": 0.5352,
      "step": 8443
    },
    {
      "epoch": 3.9736470588235293,
      "grad_norm": 0.5755517739700963,
      "learning_rate": 0.0001038976660610009,
      "loss": 0.3301,
      "step": 8444
    },
    {
      "epoch": 3.9741176470588235,
      "grad_norm": 0.5923259151144817,
      "learning_rate": 0.00010385595366060116,
      "loss": 0.7344,
      "step": 8445
    },
    {
      "epoch": 3.9745882352941178,
      "grad_norm": 0.31399547617429313,
      "learning_rate": 0.0001038142452011293,
      "loss": 0.2217,
      "step": 8446
    },
    {
      "epoch": 3.9750588235294115,
      "grad_norm": 0.2509852045794222,
      "learning_rate": 0.00010377254068614743,
      "loss": 0.1377,
      "step": 8447
    },
    {
      "epoch": 3.9755294117647058,
      "grad_norm": 0.36260658750161107,
      "learning_rate": 0.00010373084011921736,
      "loss": 0.5391,
      "step": 8448
    },
    {
      "epoch": 3.976,
      "grad_norm": 0.5319826069934311,
      "learning_rate": 0.00010368914350390048,
      "loss": 0.7266,
      "step": 8449
    },
    {
      "epoch": 3.976470588235294,
      "grad_norm": 0.6322594775105854,
      "learning_rate": 0.0001036474508437579,
      "loss": 0.6836,
      "step": 8450
    },
    {
      "epoch": 3.9769411764705884,
      "grad_norm": 0.5099485333406142,
      "learning_rate": 0.00010360576214235036,
      "loss": 0.334,
      "step": 8451
    },
    {
      "epoch": 3.977411764705882,
      "grad_norm": 0.40000532415195,
      "learning_rate": 0.00010356407740323832,
      "loss": 0.3809,
      "step": 8452
    },
    {
      "epoch": 3.9778823529411764,
      "grad_norm": 0.5903065335099826,
      "learning_rate": 0.00010352239662998182,
      "loss": 0.4199,
      "step": 8453
    },
    {
      "epoch": 3.9783529411764706,
      "grad_norm": 0.4284215339967409,
      "learning_rate": 0.00010348071982614063,
      "loss": 0.1973,
      "step": 8454
    },
    {
      "epoch": 3.978823529411765,
      "grad_norm": 0.4500306068773587,
      "learning_rate": 0.00010343904699527415,
      "loss": 0.4043,
      "step": 8455
    },
    {
      "epoch": 3.9792941176470586,
      "grad_norm": 0.3576214325551646,
      "learning_rate": 0.00010339737814094144,
      "loss": 0.1523,
      "step": 8456
    },
    {
      "epoch": 3.979764705882353,
      "grad_norm": 0.19197699270745489,
      "learning_rate": 0.00010335571326670117,
      "loss": 0.0586,
      "step": 8457
    },
    {
      "epoch": 3.980235294117647,
      "grad_norm": 0.33860102168207085,
      "learning_rate": 0.00010331405237611182,
      "loss": 0.2441,
      "step": 8458
    },
    {
      "epoch": 3.9807058823529413,
      "grad_norm": 0.511369345560786,
      "learning_rate": 0.00010327239547273138,
      "loss": 0.2139,
      "step": 8459
    },
    {
      "epoch": 3.9811764705882355,
      "grad_norm": 0.4163433585311672,
      "learning_rate": 0.00010323074256011756,
      "loss": 0.2676,
      "step": 8460
    },
    {
      "epoch": 3.9816470588235293,
      "grad_norm": 0.3585646440833093,
      "learning_rate": 0.00010318909364182774,
      "loss": 0.293,
      "step": 8461
    },
    {
      "epoch": 3.9821176470588235,
      "grad_norm": 0.3757966641728613,
      "learning_rate": 0.00010314744872141893,
      "loss": 0.1099,
      "step": 8462
    },
    {
      "epoch": 3.9825882352941178,
      "grad_norm": 0.6900137851671185,
      "learning_rate": 0.00010310580780244779,
      "loss": 0.7773,
      "step": 8463
    },
    {
      "epoch": 3.9830588235294115,
      "grad_norm": 0.4318887806720996,
      "learning_rate": 0.00010306417088847067,
      "loss": 0.5586,
      "step": 8464
    },
    {
      "epoch": 3.9835294117647058,
      "grad_norm": 0.09351710674534315,
      "learning_rate": 0.00010302253798304356,
      "loss": 0.0167,
      "step": 8465
    },
    {
      "epoch": 3.984,
      "grad_norm": 0.49276294068082466,
      "learning_rate": 0.00010298090908972217,
      "loss": 0.2432,
      "step": 8466
    },
    {
      "epoch": 3.984470588235294,
      "grad_norm": 0.1656959933429622,
      "learning_rate": 0.00010293928421206179,
      "loss": 0.0532,
      "step": 8467
    },
    {
      "epoch": 3.9849411764705884,
      "grad_norm": 0.463770971368155,
      "learning_rate": 0.00010289766335361733,
      "loss": 0.25,
      "step": 8468
    },
    {
      "epoch": 3.985411764705882,
      "grad_norm": 0.547398452840854,
      "learning_rate": 0.00010285604651794347,
      "loss": 0.7383,
      "step": 8469
    },
    {
      "epoch": 3.9858823529411764,
      "grad_norm": 0.39941687627585803,
      "learning_rate": 0.00010281443370859449,
      "loss": 0.209,
      "step": 8470
    },
    {
      "epoch": 3.9863529411764707,
      "grad_norm": 0.42495136163717084,
      "learning_rate": 0.00010277282492912429,
      "loss": 0.5156,
      "step": 8471
    },
    {
      "epoch": 3.9868235294117644,
      "grad_norm": 0.5237001401138018,
      "learning_rate": 0.0001027312201830865,
      "loss": 0.6484,
      "step": 8472
    },
    {
      "epoch": 3.9872941176470587,
      "grad_norm": 0.3395510260967464,
      "learning_rate": 0.00010268961947403437,
      "loss": 0.1855,
      "step": 8473
    },
    {
      "epoch": 3.987764705882353,
      "grad_norm": 0.5431106138244336,
      "learning_rate": 0.00010264802280552077,
      "loss": 0.3125,
      "step": 8474
    },
    {
      "epoch": 3.988235294117647,
      "grad_norm": 0.475874578495662,
      "learning_rate": 0.0001026064301810983,
      "loss": 0.2773,
      "step": 8475
    },
    {
      "epoch": 3.9887058823529413,
      "grad_norm": 0.40769687441100716,
      "learning_rate": 0.00010256484160431917,
      "loss": 0.3711,
      "step": 8476
    },
    {
      "epoch": 3.9891764705882355,
      "grad_norm": 0.5501599346554636,
      "learning_rate": 0.00010252325707873523,
      "loss": 0.3281,
      "step": 8477
    },
    {
      "epoch": 3.9896470588235293,
      "grad_norm": 0.31581431209445576,
      "learning_rate": 0.00010248167660789797,
      "loss": 0.1689,
      "step": 8478
    },
    {
      "epoch": 3.9901176470588235,
      "grad_norm": 0.35825081276798215,
      "learning_rate": 0.00010244010019535864,
      "loss": 0.3145,
      "step": 8479
    },
    {
      "epoch": 3.9905882352941178,
      "grad_norm": 0.4153798168548264,
      "learning_rate": 0.00010239852784466803,
      "loss": 0.2617,
      "step": 8480
    },
    {
      "epoch": 3.9910588235294115,
      "grad_norm": 0.4450982145267042,
      "learning_rate": 0.00010235695955937661,
      "loss": 0.3281,
      "step": 8481
    },
    {
      "epoch": 3.9915294117647058,
      "grad_norm": 0.4617583565340838,
      "learning_rate": 0.00010231539534303454,
      "loss": 0.5,
      "step": 8482
    },
    {
      "epoch": 3.992,
      "grad_norm": 0.41374944562843535,
      "learning_rate": 0.00010227383519919161,
      "loss": 0.4395,
      "step": 8483
    },
    {
      "epoch": 3.992470588235294,
      "grad_norm": 0.5730652645968093,
      "learning_rate": 0.00010223227913139722,
      "loss": 0.2793,
      "step": 8484
    },
    {
      "epoch": 3.9929411764705884,
      "grad_norm": 0.2940011170798255,
      "learning_rate": 0.0001021907271432005,
      "loss": 0.1816,
      "step": 8485
    },
    {
      "epoch": 3.993411764705882,
      "grad_norm": 0.36593274486503485,
      "learning_rate": 0.00010214917923815022,
      "loss": 0.4883,
      "step": 8486
    },
    {
      "epoch": 3.9938823529411764,
      "grad_norm": 0.4219664471391849,
      "learning_rate": 0.00010210763541979473,
      "loss": 0.2852,
      "step": 8487
    },
    {
      "epoch": 3.9943529411764707,
      "grad_norm": 0.27298535093036536,
      "learning_rate": 0.00010206609569168204,
      "loss": 0.0903,
      "step": 8488
    },
    {
      "epoch": 3.9948235294117644,
      "grad_norm": 0.32780949005282106,
      "learning_rate": 0.00010202456005735993,
      "loss": 0.1826,
      "step": 8489
    },
    {
      "epoch": 3.9952941176470587,
      "grad_norm": 0.38867299252277643,
      "learning_rate": 0.00010198302852037573,
      "loss": 0.543,
      "step": 8490
    },
    {
      "epoch": 3.995764705882353,
      "grad_norm": 0.36516125868245436,
      "learning_rate": 0.00010194150108427637,
      "loss": 0.4414,
      "step": 8491
    },
    {
      "epoch": 3.996235294117647,
      "grad_norm": 0.38497439138979217,
      "learning_rate": 0.00010189997775260857,
      "loss": 0.1689,
      "step": 8492
    },
    {
      "epoch": 3.9967058823529413,
      "grad_norm": 0.31311707291660446,
      "learning_rate": 0.00010185845852891858,
      "loss": 0.1396,
      "step": 8493
    },
    {
      "epoch": 3.9971764705882356,
      "grad_norm": 0.42461655506786966,
      "learning_rate": 0.00010181694341675235,
      "loss": 0.3145,
      "step": 8494
    },
    {
      "epoch": 3.9976470588235293,
      "grad_norm": 0.49298471766639607,
      "learning_rate": 0.00010177543241965555,
      "loss": 0.5117,
      "step": 8495
    },
    {
      "epoch": 3.9981176470588236,
      "grad_norm": 0.46055785177438185,
      "learning_rate": 0.00010173392554117333,
      "loss": 0.4746,
      "step": 8496
    },
    {
      "epoch": 3.9985882352941178,
      "grad_norm": 0.23931632360347746,
      "learning_rate": 0.00010169242278485063,
      "loss": 0.0762,
      "step": 8497
    },
    {
      "epoch": 3.9990588235294116,
      "grad_norm": 0.5881064863532779,
      "learning_rate": 0.00010165092415423193,
      "loss": 0.3086,
      "step": 8498
    },
    {
      "epoch": 3.9995294117647058,
      "grad_norm": 0.4360333305583777,
      "learning_rate": 0.00010160942965286148,
      "loss": 0.2451,
      "step": 8499
    },
    {
      "epoch": 4.0,
      "grad_norm": 0.17320151633911163,
      "learning_rate": 0.0001015679392842831,
      "loss": 0.1167,
      "step": 8500
    },
    {
      "epoch": 4.0,
      "eval_loss": 0.35721418261528015,
      "eval_runtime": 43.9887,
      "eval_samples_per_second": 17.05,
      "eval_steps_per_second": 8.525,
      "step": 8500
    },
    {
      "epoch": 4.000470588235294,
      "grad_norm": 0.27046335048558967,
      "learning_rate": 0.00010152645305204022,
      "loss": 0.1216,
      "step": 8501
    },
    {
      "epoch": 4.0009411764705884,
      "grad_norm": 0.26947951826138056,
      "learning_rate": 0.00010148497095967606,
      "loss": 0.1992,
      "step": 8502
    },
    {
      "epoch": 4.001411764705883,
      "grad_norm": 0.8477873926497677,
      "learning_rate": 0.00010144349301073334,
      "loss": 0.7812,
      "step": 8503
    },
    {
      "epoch": 4.001882352941177,
      "grad_norm": 0.5810944775220681,
      "learning_rate": 0.00010140201920875445,
      "loss": 0.668,
      "step": 8504
    },
    {
      "epoch": 4.00235294117647,
      "grad_norm": 0.40551107544619175,
      "learning_rate": 0.00010136054955728154,
      "loss": 0.291,
      "step": 8505
    },
    {
      "epoch": 4.0028235294117644,
      "grad_norm": 0.45874007546915174,
      "learning_rate": 0.00010131908405985625,
      "loss": 0.4883,
      "step": 8506
    },
    {
      "epoch": 4.003294117647059,
      "grad_norm": 0.3662173510185639,
      "learning_rate": 0.00010127762272001996,
      "loss": 0.1484,
      "step": 8507
    },
    {
      "epoch": 4.003764705882353,
      "grad_norm": 0.32719726965246604,
      "learning_rate": 0.00010123616554131363,
      "loss": 0.1406,
      "step": 8508
    },
    {
      "epoch": 4.004235294117647,
      "grad_norm": 0.3092521059609218,
      "learning_rate": 0.00010119471252727796,
      "loss": 0.3359,
      "step": 8509
    },
    {
      "epoch": 4.004705882352941,
      "grad_norm": 0.7380638171569529,
      "learning_rate": 0.00010115326368145327,
      "loss": 0.2852,
      "step": 8510
    },
    {
      "epoch": 4.005176470588236,
      "grad_norm": 0.4172569445422251,
      "learning_rate": 0.00010111181900737945,
      "loss": 0.375,
      "step": 8511
    },
    {
      "epoch": 4.00564705882353,
      "grad_norm": 0.3410979054616121,
      "learning_rate": 0.00010107037850859605,
      "loss": 0.1396,
      "step": 8512
    },
    {
      "epoch": 4.006117647058823,
      "grad_norm": 0.523931569433084,
      "learning_rate": 0.00010102894218864235,
      "loss": 0.3184,
      "step": 8513
    },
    {
      "epoch": 4.006588235294117,
      "grad_norm": 0.445185978204343,
      "learning_rate": 0.00010098751005105719,
      "loss": 0.2871,
      "step": 8514
    },
    {
      "epoch": 4.007058823529412,
      "grad_norm": 0.17228144308992802,
      "learning_rate": 0.0001009460820993791,
      "loss": 0.0376,
      "step": 8515
    },
    {
      "epoch": 4.007529411764706,
      "grad_norm": 0.4401424205138171,
      "learning_rate": 0.00010090465833714616,
      "loss": 0.4277,
      "step": 8516
    },
    {
      "epoch": 4.008,
      "grad_norm": 0.5200582664015833,
      "learning_rate": 0.00010086323876789621,
      "loss": 0.1963,
      "step": 8517
    },
    {
      "epoch": 4.008470588235294,
      "grad_norm": 0.36063115416479896,
      "learning_rate": 0.00010082182339516674,
      "loss": 0.2354,
      "step": 8518
    },
    {
      "epoch": 4.0089411764705885,
      "grad_norm": 0.09019182764646934,
      "learning_rate": 0.00010078041222249469,
      "loss": 0.0129,
      "step": 8519
    },
    {
      "epoch": 4.009411764705883,
      "grad_norm": 0.5213530658861832,
      "learning_rate": 0.00010073900525341692,
      "loss": 0.582,
      "step": 8520
    },
    {
      "epoch": 4.009882352941177,
      "grad_norm": 0.5470148322232776,
      "learning_rate": 0.00010069760249146972,
      "loss": 0.7344,
      "step": 8521
    },
    {
      "epoch": 4.01035294117647,
      "grad_norm": 0.3283352065659244,
      "learning_rate": 0.00010065620394018904,
      "loss": 0.1104,
      "step": 8522
    },
    {
      "epoch": 4.0108235294117645,
      "grad_norm": 0.5068259673524047,
      "learning_rate": 0.00010061480960311061,
      "loss": 0.3672,
      "step": 8523
    },
    {
      "epoch": 4.011294117647059,
      "grad_norm": 0.36437437162967345,
      "learning_rate": 0.0001005734194837697,
      "loss": 0.459,
      "step": 8524
    },
    {
      "epoch": 4.011764705882353,
      "grad_norm": 0.15188542377151434,
      "learning_rate": 0.00010053203358570115,
      "loss": 0.0728,
      "step": 8525
    },
    {
      "epoch": 4.012235294117647,
      "grad_norm": 0.48558572401717753,
      "learning_rate": 0.00010049065191243961,
      "loss": 0.4434,
      "step": 8526
    },
    {
      "epoch": 4.012705882352941,
      "grad_norm": 0.47319394075169324,
      "learning_rate": 0.00010044927446751923,
      "loss": 0.1436,
      "step": 8527
    },
    {
      "epoch": 4.013176470588236,
      "grad_norm": 0.38158203961864473,
      "learning_rate": 0.00010040790125447386,
      "loss": 0.2637,
      "step": 8528
    },
    {
      "epoch": 4.01364705882353,
      "grad_norm": 0.40807990954502843,
      "learning_rate": 0.00010036653227683694,
      "loss": 0.582,
      "step": 8529
    },
    {
      "epoch": 4.014117647058823,
      "grad_norm": 0.8582899994624148,
      "learning_rate": 0.00010032516753814166,
      "loss": 0.5781,
      "step": 8530
    },
    {
      "epoch": 4.014588235294117,
      "grad_norm": 0.21976450696384148,
      "learning_rate": 0.0001002838070419207,
      "loss": 0.0786,
      "step": 8531
    },
    {
      "epoch": 4.015058823529412,
      "grad_norm": 0.3199254388997735,
      "learning_rate": 0.00010024245079170645,
      "loss": 0.1543,
      "step": 8532
    },
    {
      "epoch": 4.015529411764706,
      "grad_norm": 0.45404272680084834,
      "learning_rate": 0.000100201098791031,
      "loss": 0.2969,
      "step": 8533
    },
    {
      "epoch": 4.016,
      "grad_norm": 0.36692056089679653,
      "learning_rate": 0.00010015975104342598,
      "loss": 0.2578,
      "step": 8534
    },
    {
      "epoch": 4.016470588235294,
      "grad_norm": 0.2370819205688813,
      "learning_rate": 0.00010011840755242266,
      "loss": 0.1436,
      "step": 8535
    },
    {
      "epoch": 4.0169411764705885,
      "grad_norm": 0.2636033621913804,
      "learning_rate": 0.00010007706832155201,
      "loss": 0.0781,
      "step": 8536
    },
    {
      "epoch": 4.017411764705883,
      "grad_norm": 0.6170984066701112,
      "learning_rate": 0.00010003573335434462,
      "loss": 0.7578,
      "step": 8537
    },
    {
      "epoch": 4.017882352941177,
      "grad_norm": 0.1470076303982632,
      "learning_rate": 9.999440265433065e-05,
      "loss": 0.0261,
      "step": 8538
    },
    {
      "epoch": 4.01835294117647,
      "grad_norm": 0.35036356695154836,
      "learning_rate": 9.995307622503993e-05,
      "loss": 0.3613,
      "step": 8539
    },
    {
      "epoch": 4.0188235294117645,
      "grad_norm": 0.3596128515700946,
      "learning_rate": 9.991175407000206e-05,
      "loss": 0.373,
      "step": 8540
    },
    {
      "epoch": 4.019294117647059,
      "grad_norm": 0.1947371268685809,
      "learning_rate": 9.987043619274604e-05,
      "loss": 0.0544,
      "step": 8541
    },
    {
      "epoch": 4.019764705882353,
      "grad_norm": 0.28626667358326996,
      "learning_rate": 9.982912259680063e-05,
      "loss": 0.3145,
      "step": 8542
    },
    {
      "epoch": 4.020235294117647,
      "grad_norm": 0.19734889161686697,
      "learning_rate": 9.978781328569427e-05,
      "loss": 0.0747,
      "step": 8543
    },
    {
      "epoch": 4.020705882352941,
      "grad_norm": 0.44209776018534347,
      "learning_rate": 9.974650826295493e-05,
      "loss": 0.3008,
      "step": 8544
    },
    {
      "epoch": 4.021176470588236,
      "grad_norm": 0.5279384876985913,
      "learning_rate": 9.970520753211027e-05,
      "loss": 0.3848,
      "step": 8545
    },
    {
      "epoch": 4.02164705882353,
      "grad_norm": 0.21750261193457485,
      "learning_rate": 9.966391109668758e-05,
      "loss": 0.0815,
      "step": 8546
    },
    {
      "epoch": 4.022117647058823,
      "grad_norm": 0.13800868890574308,
      "learning_rate": 9.96226189602138e-05,
      "loss": 0.0211,
      "step": 8547
    },
    {
      "epoch": 4.022588235294117,
      "grad_norm": 0.3635088394202899,
      "learning_rate": 9.958133112621547e-05,
      "loss": 0.3652,
      "step": 8548
    },
    {
      "epoch": 4.023058823529412,
      "grad_norm": 0.48412004968092404,
      "learning_rate": 9.954004759821871e-05,
      "loss": 0.5156,
      "step": 8549
    },
    {
      "epoch": 4.023529411764706,
      "grad_norm": 1.146745074232935,
      "learning_rate": 9.949876837974944e-05,
      "loss": 0.5469,
      "step": 8550
    },
    {
      "epoch": 4.024,
      "grad_norm": 0.4283196706207605,
      "learning_rate": 9.945749347433304e-05,
      "loss": 0.3301,
      "step": 8551
    },
    {
      "epoch": 4.024470588235294,
      "grad_norm": 0.21443734735079556,
      "learning_rate": 9.941622288549458e-05,
      "loss": 0.0708,
      "step": 8552
    },
    {
      "epoch": 4.0249411764705885,
      "grad_norm": 0.34016331625334617,
      "learning_rate": 9.937495661675878e-05,
      "loss": 0.1191,
      "step": 8553
    },
    {
      "epoch": 4.025411764705883,
      "grad_norm": 0.2562251172147395,
      "learning_rate": 9.933369467165004e-05,
      "loss": 0.0796,
      "step": 8554
    },
    {
      "epoch": 4.025882352941177,
      "grad_norm": 0.6811365377527158,
      "learning_rate": 9.929243705369232e-05,
      "loss": 0.7539,
      "step": 8555
    },
    {
      "epoch": 4.02635294117647,
      "grad_norm": 0.2781793360370128,
      "learning_rate": 9.925118376640918e-05,
      "loss": 0.0825,
      "step": 8556
    },
    {
      "epoch": 4.0268235294117645,
      "grad_norm": 0.3913942740607705,
      "learning_rate": 9.920993481332383e-05,
      "loss": 0.5078,
      "step": 8557
    },
    {
      "epoch": 4.027294117647059,
      "grad_norm": 0.2048058593961362,
      "learning_rate": 9.916869019795922e-05,
      "loss": 0.051,
      "step": 8558
    },
    {
      "epoch": 4.027764705882353,
      "grad_norm": 0.12842776509497586,
      "learning_rate": 9.912744992383777e-05,
      "loss": 0.0117,
      "step": 8559
    },
    {
      "epoch": 4.028235294117647,
      "grad_norm": 0.3941073888589094,
      "learning_rate": 9.90862139944816e-05,
      "loss": 0.2539,
      "step": 8560
    },
    {
      "epoch": 4.028705882352941,
      "grad_norm": 0.34940320024687294,
      "learning_rate": 9.904498241341252e-05,
      "loss": 0.3535,
      "step": 8561
    },
    {
      "epoch": 4.029176470588236,
      "grad_norm": 0.18393101373807294,
      "learning_rate": 9.900375518415188e-05,
      "loss": 0.0747,
      "step": 8562
    },
    {
      "epoch": 4.02964705882353,
      "grad_norm": 0.4039522482477332,
      "learning_rate": 9.896253231022066e-05,
      "loss": 0.1641,
      "step": 8563
    },
    {
      "epoch": 4.030117647058823,
      "grad_norm": 0.3783611200773936,
      "learning_rate": 9.892131379513953e-05,
      "loss": 0.1621,
      "step": 8564
    },
    {
      "epoch": 4.030588235294117,
      "grad_norm": 0.32720511237855215,
      "learning_rate": 9.888009964242876e-05,
      "loss": 0.2109,
      "step": 8565
    },
    {
      "epoch": 4.031058823529412,
      "grad_norm": 0.3859644228497096,
      "learning_rate": 9.883888985560822e-05,
      "loss": 0.2773,
      "step": 8566
    },
    {
      "epoch": 4.031529411764706,
      "grad_norm": 0.5683126200184717,
      "learning_rate": 9.879768443819737e-05,
      "loss": 0.6875,
      "step": 8567
    },
    {
      "epoch": 4.032,
      "grad_norm": 0.435031539028382,
      "learning_rate": 9.875648339371546e-05,
      "loss": 0.1992,
      "step": 8568
    },
    {
      "epoch": 4.032470588235294,
      "grad_norm": 0.39078662136118125,
      "learning_rate": 9.871528672568124e-05,
      "loss": 0.4238,
      "step": 8569
    },
    {
      "epoch": 4.0329411764705885,
      "grad_norm": 0.3587667835714099,
      "learning_rate": 9.867409443761304e-05,
      "loss": 0.2305,
      "step": 8570
    },
    {
      "epoch": 4.033411764705883,
      "grad_norm": 0.43303153697344804,
      "learning_rate": 9.863290653302894e-05,
      "loss": 0.2188,
      "step": 8571
    },
    {
      "epoch": 4.033882352941176,
      "grad_norm": 0.31677401884904705,
      "learning_rate": 9.859172301544659e-05,
      "loss": 0.0703,
      "step": 8572
    },
    {
      "epoch": 4.03435294117647,
      "grad_norm": 0.24594702262628357,
      "learning_rate": 9.855054388838322e-05,
      "loss": 0.2734,
      "step": 8573
    },
    {
      "epoch": 4.0348235294117645,
      "grad_norm": 0.23982880494915434,
      "learning_rate": 9.85093691553558e-05,
      "loss": 0.0688,
      "step": 8574
    },
    {
      "epoch": 4.035294117647059,
      "grad_norm": 0.44437125550412127,
      "learning_rate": 9.84681988198808e-05,
      "loss": 0.1934,
      "step": 8575
    },
    {
      "epoch": 4.035764705882353,
      "grad_norm": 0.4599742845899598,
      "learning_rate": 9.842703288547438e-05,
      "loss": 0.2363,
      "step": 8576
    },
    {
      "epoch": 4.036235294117647,
      "grad_norm": 0.47223166717348547,
      "learning_rate": 9.838587135565229e-05,
      "loss": 0.2656,
      "step": 8577
    },
    {
      "epoch": 4.036705882352941,
      "grad_norm": 0.5692635952887349,
      "learning_rate": 9.834471423393e-05,
      "loss": 0.4258,
      "step": 8578
    },
    {
      "epoch": 4.037176470588236,
      "grad_norm": 0.40999780367653393,
      "learning_rate": 9.830356152382245e-05,
      "loss": 0.1934,
      "step": 8579
    },
    {
      "epoch": 4.03764705882353,
      "grad_norm": 0.663216915257029,
      "learning_rate": 9.826241322884432e-05,
      "loss": 0.7656,
      "step": 8580
    },
    {
      "epoch": 4.038117647058823,
      "grad_norm": 0.5234064937037354,
      "learning_rate": 9.822126935250987e-05,
      "loss": 0.3965,
      "step": 8581
    },
    {
      "epoch": 4.038588235294117,
      "grad_norm": 0.3561544974419018,
      "learning_rate": 9.818012989833301e-05,
      "loss": 0.3164,
      "step": 8582
    },
    {
      "epoch": 4.039058823529412,
      "grad_norm": 0.43396491129217324,
      "learning_rate": 9.81389948698272e-05,
      "loss": 0.4688,
      "step": 8583
    },
    {
      "epoch": 4.039529411764706,
      "grad_norm": 0.3862723229682532,
      "learning_rate": 9.809786427050562e-05,
      "loss": 0.3164,
      "step": 8584
    },
    {
      "epoch": 4.04,
      "grad_norm": 0.5198906124165554,
      "learning_rate": 9.805673810388103e-05,
      "loss": 0.2754,
      "step": 8585
    },
    {
      "epoch": 4.040470588235294,
      "grad_norm": 0.28502991160050817,
      "learning_rate": 9.801561637346575e-05,
      "loss": 0.0962,
      "step": 8586
    },
    {
      "epoch": 4.0409411764705885,
      "grad_norm": 0.3017594184284444,
      "learning_rate": 9.797449908277184e-05,
      "loss": 0.1875,
      "step": 8587
    },
    {
      "epoch": 4.041411764705883,
      "grad_norm": 0.2674259048375895,
      "learning_rate": 9.793338623531089e-05,
      "loss": 0.2344,
      "step": 8588
    },
    {
      "epoch": 4.041882352941176,
      "grad_norm": 0.3491143628779799,
      "learning_rate": 9.789227783459415e-05,
      "loss": 0.168,
      "step": 8589
    },
    {
      "epoch": 4.04235294117647,
      "grad_norm": 0.2560009643180767,
      "learning_rate": 9.785117388413244e-05,
      "loss": 0.1885,
      "step": 8590
    },
    {
      "epoch": 4.0428235294117645,
      "grad_norm": 0.47981043735802203,
      "learning_rate": 9.781007438743629e-05,
      "loss": 0.5508,
      "step": 8591
    },
    {
      "epoch": 4.043294117647059,
      "grad_norm": 0.2141967441971164,
      "learning_rate": 9.776897934801577e-05,
      "loss": 0.0544,
      "step": 8592
    },
    {
      "epoch": 4.043764705882353,
      "grad_norm": 0.34927368551514926,
      "learning_rate": 9.77278887693806e-05,
      "loss": 0.1523,
      "step": 8593
    },
    {
      "epoch": 4.044235294117647,
      "grad_norm": 0.3853583214017607,
      "learning_rate": 9.768680265504013e-05,
      "loss": 0.418,
      "step": 8594
    },
    {
      "epoch": 4.044705882352941,
      "grad_norm": 0.4237443650575791,
      "learning_rate": 9.764572100850331e-05,
      "loss": 0.2021,
      "step": 8595
    },
    {
      "epoch": 4.045176470588236,
      "grad_norm": 0.3779101908020774,
      "learning_rate": 9.760464383327868e-05,
      "loss": 0.2773,
      "step": 8596
    },
    {
      "epoch": 4.04564705882353,
      "grad_norm": 0.4234290638421039,
      "learning_rate": 9.756357113287449e-05,
      "loss": 0.2793,
      "step": 8597
    },
    {
      "epoch": 4.046117647058823,
      "grad_norm": 0.479952205172976,
      "learning_rate": 9.75225029107985e-05,
      "loss": 0.375,
      "step": 8598
    },
    {
      "epoch": 4.046588235294117,
      "grad_norm": 0.36216161636163885,
      "learning_rate": 9.74814391705582e-05,
      "loss": 0.3516,
      "step": 8599
    },
    {
      "epoch": 4.047058823529412,
      "grad_norm": 0.42350417184059774,
      "learning_rate": 9.744037991566058e-05,
      "loss": 0.2715,
      "step": 8600
    },
    {
      "epoch": 4.047529411764706,
      "grad_norm": 0.4403041360639695,
      "learning_rate": 9.739932514961232e-05,
      "loss": 0.5312,
      "step": 8601
    },
    {
      "epoch": 4.048,
      "grad_norm": 0.41308791698689035,
      "learning_rate": 9.73582748759197e-05,
      "loss": 0.3828,
      "step": 8602
    },
    {
      "epoch": 4.048470588235294,
      "grad_norm": 0.32500485398422485,
      "learning_rate": 9.731722909808863e-05,
      "loss": 0.127,
      "step": 8603
    },
    {
      "epoch": 4.0489411764705885,
      "grad_norm": 0.3408208256479765,
      "learning_rate": 9.727618781962457e-05,
      "loss": 0.25,
      "step": 8604
    },
    {
      "epoch": 4.049411764705883,
      "grad_norm": 0.3773403032123593,
      "learning_rate": 9.723515104403268e-05,
      "loss": 0.373,
      "step": 8605
    },
    {
      "epoch": 4.049882352941176,
      "grad_norm": 0.41483171733993796,
      "learning_rate": 9.719411877481776e-05,
      "loss": 0.4492,
      "step": 8606
    },
    {
      "epoch": 4.05035294117647,
      "grad_norm": 0.3065537488236469,
      "learning_rate": 9.715309101548406e-05,
      "loss": 0.1055,
      "step": 8607
    },
    {
      "epoch": 4.0508235294117645,
      "grad_norm": 0.23288127264869618,
      "learning_rate": 9.711206776953559e-05,
      "loss": 0.0737,
      "step": 8608
    },
    {
      "epoch": 4.051294117647059,
      "grad_norm": 0.627113063655111,
      "learning_rate": 9.707104904047596e-05,
      "loss": 0.4473,
      "step": 8609
    },
    {
      "epoch": 4.051764705882353,
      "grad_norm": 0.29222857671279934,
      "learning_rate": 9.703003483180836e-05,
      "loss": 0.3027,
      "step": 8610
    },
    {
      "epoch": 4.052235294117647,
      "grad_norm": 0.4342329701735141,
      "learning_rate": 9.698902514703558e-05,
      "loss": 0.2852,
      "step": 8611
    },
    {
      "epoch": 4.052705882352941,
      "grad_norm": 0.3582434104283325,
      "learning_rate": 9.694801998966011e-05,
      "loss": 0.1963,
      "step": 8612
    },
    {
      "epoch": 4.053176470588236,
      "grad_norm": 0.34295578395874626,
      "learning_rate": 9.690701936318395e-05,
      "loss": 0.3555,
      "step": 8613
    },
    {
      "epoch": 4.05364705882353,
      "grad_norm": 0.32280946078963496,
      "learning_rate": 9.686602327110869e-05,
      "loss": 0.2246,
      "step": 8614
    },
    {
      "epoch": 4.054117647058823,
      "grad_norm": 0.30556835074624616,
      "learning_rate": 9.682503171693573e-05,
      "loss": 0.3418,
      "step": 8615
    },
    {
      "epoch": 4.054588235294117,
      "grad_norm": 0.2436277780977727,
      "learning_rate": 9.678404470416587e-05,
      "loss": 0.1074,
      "step": 8616
    },
    {
      "epoch": 4.055058823529412,
      "grad_norm": 0.3336000852724605,
      "learning_rate": 9.674306223629963e-05,
      "loss": 0.2715,
      "step": 8617
    },
    {
      "epoch": 4.055529411764706,
      "grad_norm": 0.4621883591114871,
      "learning_rate": 9.670208431683704e-05,
      "loss": 0.375,
      "step": 8618
    },
    {
      "epoch": 4.056,
      "grad_norm": 0.40871537654023116,
      "learning_rate": 9.666111094927792e-05,
      "loss": 0.5664,
      "step": 8619
    },
    {
      "epoch": 4.056470588235294,
      "grad_norm": 0.3676348463864894,
      "learning_rate": 9.662014213712155e-05,
      "loss": 0.2217,
      "step": 8620
    },
    {
      "epoch": 4.0569411764705885,
      "grad_norm": 0.3276356741280457,
      "learning_rate": 9.657917788386684e-05,
      "loss": 0.2461,
      "step": 8621
    },
    {
      "epoch": 4.057411764705883,
      "grad_norm": 0.6186539094068465,
      "learning_rate": 9.65382181930124e-05,
      "loss": 0.5508,
      "step": 8622
    },
    {
      "epoch": 4.057882352941176,
      "grad_norm": 1.2134979023514194,
      "learning_rate": 9.649726306805635e-05,
      "loss": 0.4277,
      "step": 8623
    },
    {
      "epoch": 4.05835294117647,
      "grad_norm": 0.7561782679607293,
      "learning_rate": 9.645631251249643e-05,
      "loss": 0.6172,
      "step": 8624
    },
    {
      "epoch": 4.0588235294117645,
      "grad_norm": 0.429776068699409,
      "learning_rate": 9.641536652983008e-05,
      "loss": 0.1807,
      "step": 8625
    },
    {
      "epoch": 4.059294117647059,
      "grad_norm": 0.2707029829287279,
      "learning_rate": 9.637442512355428e-05,
      "loss": 0.2266,
      "step": 8626
    },
    {
      "epoch": 4.059764705882353,
      "grad_norm": 0.5321872126389142,
      "learning_rate": 9.63334882971656e-05,
      "loss": 0.2852,
      "step": 8627
    },
    {
      "epoch": 4.060235294117647,
      "grad_norm": 0.17403163357567367,
      "learning_rate": 9.629255605416021e-05,
      "loss": 0.0427,
      "step": 8628
    },
    {
      "epoch": 4.060705882352941,
      "grad_norm": 0.4113862258168615,
      "learning_rate": 9.6251628398034e-05,
      "loss": 0.2949,
      "step": 8629
    },
    {
      "epoch": 4.061176470588236,
      "grad_norm": 0.39825362332481345,
      "learning_rate": 9.621070533228236e-05,
      "loss": 0.4668,
      "step": 8630
    },
    {
      "epoch": 4.06164705882353,
      "grad_norm": 0.42120463436587957,
      "learning_rate": 9.616978686040028e-05,
      "loss": 0.2422,
      "step": 8631
    },
    {
      "epoch": 4.062117647058823,
      "grad_norm": 0.4688487845137679,
      "learning_rate": 9.612887298588246e-05,
      "loss": 0.3281,
      "step": 8632
    },
    {
      "epoch": 4.062588235294117,
      "grad_norm": 0.37439264579325227,
      "learning_rate": 9.608796371222316e-05,
      "loss": 0.2139,
      "step": 8633
    },
    {
      "epoch": 4.063058823529412,
      "grad_norm": 0.263013894624454,
      "learning_rate": 9.604705904291614e-05,
      "loss": 0.1279,
      "step": 8634
    },
    {
      "epoch": 4.063529411764706,
      "grad_norm": 0.4055781224741934,
      "learning_rate": 9.600615898145496e-05,
      "loss": 0.3301,
      "step": 8635
    },
    {
      "epoch": 4.064,
      "grad_norm": 0.49782088107424854,
      "learning_rate": 9.596526353133261e-05,
      "loss": 0.0835,
      "step": 8636
    },
    {
      "epoch": 4.064470588235294,
      "grad_norm": 0.2707434908189511,
      "learning_rate": 9.59243726960418e-05,
      "loss": 0.2275,
      "step": 8637
    },
    {
      "epoch": 4.0649411764705885,
      "grad_norm": 0.4900222815911684,
      "learning_rate": 9.588348647907481e-05,
      "loss": 0.5469,
      "step": 8638
    },
    {
      "epoch": 4.065411764705883,
      "grad_norm": 0.5145192639624151,
      "learning_rate": 9.584260488392353e-05,
      "loss": 0.4629,
      "step": 8639
    },
    {
      "epoch": 4.065882352941176,
      "grad_norm": 0.3872809306882426,
      "learning_rate": 9.580172791407942e-05,
      "loss": 0.1797,
      "step": 8640
    },
    {
      "epoch": 4.06635294117647,
      "grad_norm": 0.5510870214959588,
      "learning_rate": 9.576085557303355e-05,
      "loss": 0.6836,
      "step": 8641
    },
    {
      "epoch": 4.0668235294117645,
      "grad_norm": 0.17374789120363743,
      "learning_rate": 9.571998786427665e-05,
      "loss": 0.0381,
      "step": 8642
    },
    {
      "epoch": 4.067294117647059,
      "grad_norm": 0.5664031205382724,
      "learning_rate": 9.567912479129907e-05,
      "loss": 0.2969,
      "step": 8643
    },
    {
      "epoch": 4.067764705882353,
      "grad_norm": 0.4729711235109889,
      "learning_rate": 9.563826635759068e-05,
      "loss": 0.6133,
      "step": 8644
    },
    {
      "epoch": 4.068235294117647,
      "grad_norm": 0.46768445280332743,
      "learning_rate": 9.559741256664098e-05,
      "loss": 0.3438,
      "step": 8645
    },
    {
      "epoch": 4.068705882352941,
      "grad_norm": 0.5601329452916523,
      "learning_rate": 9.555656342193907e-05,
      "loss": 0.416,
      "step": 8646
    },
    {
      "epoch": 4.069176470588236,
      "grad_norm": 0.47683846460858215,
      "learning_rate": 9.551571892697372e-05,
      "loss": 0.3828,
      "step": 8647
    },
    {
      "epoch": 4.06964705882353,
      "grad_norm": 0.2979588059131156,
      "learning_rate": 9.547487908523322e-05,
      "loss": 0.2314,
      "step": 8648
    },
    {
      "epoch": 4.070117647058823,
      "grad_norm": 0.2775073415022456,
      "learning_rate": 9.543404390020543e-05,
      "loss": 0.2129,
      "step": 8649
    },
    {
      "epoch": 4.070588235294117,
      "grad_norm": 0.47795991633631124,
      "learning_rate": 9.5393213375378e-05,
      "loss": 0.4512,
      "step": 8650
    },
    {
      "epoch": 4.071058823529412,
      "grad_norm": 0.7319789633294287,
      "learning_rate": 9.535238751423799e-05,
      "loss": 0.625,
      "step": 8651
    },
    {
      "epoch": 4.071529411764706,
      "grad_norm": 0.2575959311831457,
      "learning_rate": 9.531156632027209e-05,
      "loss": 0.0796,
      "step": 8652
    },
    {
      "epoch": 4.072,
      "grad_norm": 0.44072008768993126,
      "learning_rate": 9.52707497969667e-05,
      "loss": 0.3008,
      "step": 8653
    },
    {
      "epoch": 4.072470588235294,
      "grad_norm": 0.3647895559235517,
      "learning_rate": 9.522993794780773e-05,
      "loss": 0.2012,
      "step": 8654
    },
    {
      "epoch": 4.0729411764705885,
      "grad_norm": 0.5252005711716032,
      "learning_rate": 9.518913077628066e-05,
      "loss": 0.4492,
      "step": 8655
    },
    {
      "epoch": 4.073411764705883,
      "grad_norm": 0.322497724631164,
      "learning_rate": 9.514832828587071e-05,
      "loss": 0.126,
      "step": 8656
    },
    {
      "epoch": 4.073882352941176,
      "grad_norm": 0.4016564131308385,
      "learning_rate": 9.510753048006257e-05,
      "loss": 0.3652,
      "step": 8657
    },
    {
      "epoch": 4.07435294117647,
      "grad_norm": 0.18571656295920544,
      "learning_rate": 9.506673736234057e-05,
      "loss": 0.1348,
      "step": 8658
    },
    {
      "epoch": 4.0748235294117645,
      "grad_norm": 0.3164465209843287,
      "learning_rate": 9.502594893618861e-05,
      "loss": 0.2334,
      "step": 8659
    },
    {
      "epoch": 4.075294117647059,
      "grad_norm": 0.3963493355001973,
      "learning_rate": 9.49851652050903e-05,
      "loss": 0.2402,
      "step": 8660
    },
    {
      "epoch": 4.075764705882353,
      "grad_norm": 0.5062783618925644,
      "learning_rate": 9.494438617252875e-05,
      "loss": 0.4883,
      "step": 8661
    },
    {
      "epoch": 4.076235294117647,
      "grad_norm": 0.36498807059627825,
      "learning_rate": 9.490361184198659e-05,
      "loss": 0.2734,
      "step": 8662
    },
    {
      "epoch": 4.076705882352941,
      "grad_norm": 0.26207365980172065,
      "learning_rate": 9.486284221694626e-05,
      "loss": 0.1182,
      "step": 8663
    },
    {
      "epoch": 4.077176470588236,
      "grad_norm": 0.33950139248458683,
      "learning_rate": 9.482207730088968e-05,
      "loss": 0.1514,
      "step": 8664
    },
    {
      "epoch": 4.07764705882353,
      "grad_norm": 0.20328977504582033,
      "learning_rate": 9.47813170972983e-05,
      "loss": 0.0623,
      "step": 8665
    },
    {
      "epoch": 4.078117647058823,
      "grad_norm": 0.3553034428729852,
      "learning_rate": 9.474056160965332e-05,
      "loss": 0.2754,
      "step": 8666
    },
    {
      "epoch": 4.078588235294117,
      "grad_norm": 0.26596409742959864,
      "learning_rate": 9.469981084143541e-05,
      "loss": 0.0913,
      "step": 8667
    },
    {
      "epoch": 4.079058823529412,
      "grad_norm": 0.2598367739742527,
      "learning_rate": 9.465906479612492e-05,
      "loss": 0.2373,
      "step": 8668
    },
    {
      "epoch": 4.079529411764706,
      "grad_norm": 0.3108005115862394,
      "learning_rate": 9.461832347720168e-05,
      "loss": 0.3418,
      "step": 8669
    },
    {
      "epoch": 4.08,
      "grad_norm": 0.46198355684154935,
      "learning_rate": 9.457758688814533e-05,
      "loss": 0.4297,
      "step": 8670
    },
    {
      "epoch": 4.080470588235294,
      "grad_norm": 0.3052601210864011,
      "learning_rate": 9.453685503243489e-05,
      "loss": 0.1094,
      "step": 8671
    },
    {
      "epoch": 4.0809411764705885,
      "grad_norm": 0.4155878621671173,
      "learning_rate": 9.449612791354904e-05,
      "loss": 0.3711,
      "step": 8672
    },
    {
      "epoch": 4.081411764705883,
      "grad_norm": 0.4114702526136615,
      "learning_rate": 9.445540553496614e-05,
      "loss": 0.2656,
      "step": 8673
    },
    {
      "epoch": 4.081882352941176,
      "grad_norm": 0.4904574261981477,
      "learning_rate": 9.441468790016406e-05,
      "loss": 0.7188,
      "step": 8674
    },
    {
      "epoch": 4.08235294117647,
      "grad_norm": 0.4077311921243188,
      "learning_rate": 9.437397501262026e-05,
      "loss": 0.165,
      "step": 8675
    },
    {
      "epoch": 4.0828235294117645,
      "grad_norm": 0.5200933781011267,
      "learning_rate": 9.433326687581183e-05,
      "loss": 0.5352,
      "step": 8676
    },
    {
      "epoch": 4.083294117647059,
      "grad_norm": 0.13717962407680856,
      "learning_rate": 9.429256349321548e-05,
      "loss": 0.0432,
      "step": 8677
    },
    {
      "epoch": 4.083764705882353,
      "grad_norm": 0.45414698160586536,
      "learning_rate": 9.425186486830745e-05,
      "loss": 0.3594,
      "step": 8678
    },
    {
      "epoch": 4.084235294117647,
      "grad_norm": 0.22915441543703657,
      "learning_rate": 9.421117100456356e-05,
      "loss": 0.2578,
      "step": 8679
    },
    {
      "epoch": 4.084705882352941,
      "grad_norm": 0.34244498655864053,
      "learning_rate": 9.417048190545935e-05,
      "loss": 0.3184,
      "step": 8680
    },
    {
      "epoch": 4.085176470588236,
      "grad_norm": 0.6498913056044286,
      "learning_rate": 9.412979757446983e-05,
      "loss": 0.5117,
      "step": 8681
    },
    {
      "epoch": 4.085647058823529,
      "grad_norm": 0.590878125466877,
      "learning_rate": 9.40891180150696e-05,
      "loss": 0.3359,
      "step": 8682
    },
    {
      "epoch": 4.086117647058823,
      "grad_norm": 0.4068149880005259,
      "learning_rate": 9.404844323073296e-05,
      "loss": 0.3555,
      "step": 8683
    },
    {
      "epoch": 4.086588235294117,
      "grad_norm": 0.7034459075889273,
      "learning_rate": 9.400777322493373e-05,
      "loss": 0.6328,
      "step": 8684
    },
    {
      "epoch": 4.087058823529412,
      "grad_norm": 0.2737813448823211,
      "learning_rate": 9.396710800114526e-05,
      "loss": 0.1807,
      "step": 8685
    },
    {
      "epoch": 4.087529411764706,
      "grad_norm": 0.37163894170390943,
      "learning_rate": 9.39264475628406e-05,
      "loss": 0.4336,
      "step": 8686
    },
    {
      "epoch": 4.088,
      "grad_norm": 0.4534820551379952,
      "learning_rate": 9.388579191349237e-05,
      "loss": 0.4688,
      "step": 8687
    },
    {
      "epoch": 4.088470588235294,
      "grad_norm": 0.29991122131389686,
      "learning_rate": 9.384514105657277e-05,
      "loss": 0.1602,
      "step": 8688
    },
    {
      "epoch": 4.0889411764705885,
      "grad_norm": 0.46101047729195327,
      "learning_rate": 9.380449499555358e-05,
      "loss": 0.6875,
      "step": 8689
    },
    {
      "epoch": 4.089411764705883,
      "grad_norm": 0.38810789166136345,
      "learning_rate": 9.376385373390611e-05,
      "loss": 0.4062,
      "step": 8690
    },
    {
      "epoch": 4.089882352941176,
      "grad_norm": 0.36642845933015855,
      "learning_rate": 9.372321727510139e-05,
      "loss": 0.2168,
      "step": 8691
    },
    {
      "epoch": 4.09035294117647,
      "grad_norm": 0.4650834024169917,
      "learning_rate": 9.368258562260999e-05,
      "loss": 0.2578,
      "step": 8692
    },
    {
      "epoch": 4.0908235294117645,
      "grad_norm": 0.628185482894092,
      "learning_rate": 9.364195877990195e-05,
      "loss": 0.4609,
      "step": 8693
    },
    {
      "epoch": 4.091294117647059,
      "grad_norm": 0.28755947623047323,
      "learning_rate": 9.360133675044708e-05,
      "loss": 0.082,
      "step": 8694
    },
    {
      "epoch": 4.091764705882353,
      "grad_norm": 0.4197698636484122,
      "learning_rate": 9.356071953771472e-05,
      "loss": 0.2832,
      "step": 8695
    },
    {
      "epoch": 4.092235294117647,
      "grad_norm": 0.3653862179432398,
      "learning_rate": 9.352010714517371e-05,
      "loss": 0.498,
      "step": 8696
    },
    {
      "epoch": 4.092705882352941,
      "grad_norm": 0.3558508504406742,
      "learning_rate": 9.347949957629258e-05,
      "loss": 0.3672,
      "step": 8697
    },
    {
      "epoch": 4.093176470588236,
      "grad_norm": 0.683943273238338,
      "learning_rate": 9.343889683453945e-05,
      "loss": 0.5781,
      "step": 8698
    },
    {
      "epoch": 4.09364705882353,
      "grad_norm": 0.4454810370181389,
      "learning_rate": 9.339829892338193e-05,
      "loss": 0.4043,
      "step": 8699
    },
    {
      "epoch": 4.094117647058823,
      "grad_norm": 0.6391889987033492,
      "learning_rate": 9.33577058462873e-05,
      "loss": 0.5234,
      "step": 8700
    },
    {
      "epoch": 4.094588235294117,
      "grad_norm": 0.4622797024435839,
      "learning_rate": 9.331711760672246e-05,
      "loss": 0.1973,
      "step": 8701
    },
    {
      "epoch": 4.095058823529412,
      "grad_norm": 0.41662865831529894,
      "learning_rate": 9.32765342081538e-05,
      "loss": 0.3789,
      "step": 8702
    },
    {
      "epoch": 4.095529411764706,
      "grad_norm": 0.3662949579849293,
      "learning_rate": 9.323595565404731e-05,
      "loss": 0.1602,
      "step": 8703
    },
    {
      "epoch": 4.096,
      "grad_norm": 0.2686603803930179,
      "learning_rate": 9.319538194786867e-05,
      "loss": 0.0859,
      "step": 8704
    },
    {
      "epoch": 4.096470588235294,
      "grad_norm": 0.4192134117632906,
      "learning_rate": 9.315481309308303e-05,
      "loss": 0.1777,
      "step": 8705
    },
    {
      "epoch": 4.0969411764705885,
      "grad_norm": 0.3926691372862366,
      "learning_rate": 9.311424909315514e-05,
      "loss": 0.167,
      "step": 8706
    },
    {
      "epoch": 4.097411764705883,
      "grad_norm": 0.5428913996094251,
      "learning_rate": 9.307368995154945e-05,
      "loss": 0.5508,
      "step": 8707
    },
    {
      "epoch": 4.097882352941176,
      "grad_norm": 0.48882952953913705,
      "learning_rate": 9.303313567172985e-05,
      "loss": 0.5703,
      "step": 8708
    },
    {
      "epoch": 4.09835294117647,
      "grad_norm": 0.38434335813264914,
      "learning_rate": 9.299258625715988e-05,
      "loss": 0.334,
      "step": 8709
    },
    {
      "epoch": 4.0988235294117645,
      "grad_norm": 0.627921342625777,
      "learning_rate": 9.295204171130264e-05,
      "loss": 0.4648,
      "step": 8710
    },
    {
      "epoch": 4.099294117647059,
      "grad_norm": 0.3377978940217193,
      "learning_rate": 9.291150203762087e-05,
      "loss": 0.2109,
      "step": 8711
    },
    {
      "epoch": 4.099764705882353,
      "grad_norm": 0.4040532323782568,
      "learning_rate": 9.287096723957685e-05,
      "loss": 0.416,
      "step": 8712
    },
    {
      "epoch": 4.100235294117647,
      "grad_norm": 0.6392199784123734,
      "learning_rate": 9.283043732063243e-05,
      "loss": 0.6211,
      "step": 8713
    },
    {
      "epoch": 4.100705882352941,
      "grad_norm": 0.32407257612176776,
      "learning_rate": 9.27899122842491e-05,
      "loss": 0.3008,
      "step": 8714
    },
    {
      "epoch": 4.101176470588236,
      "grad_norm": 0.39699478144180145,
      "learning_rate": 9.274939213388785e-05,
      "loss": 0.4551,
      "step": 8715
    },
    {
      "epoch": 4.101647058823529,
      "grad_norm": 0.3087574262352862,
      "learning_rate": 9.270887687300932e-05,
      "loss": 0.0718,
      "step": 8716
    },
    {
      "epoch": 4.102117647058823,
      "grad_norm": 0.37693630287655183,
      "learning_rate": 9.266836650507375e-05,
      "loss": 0.2637,
      "step": 8717
    },
    {
      "epoch": 4.102588235294117,
      "grad_norm": 0.5588381045558337,
      "learning_rate": 9.26278610335409e-05,
      "loss": 0.6328,
      "step": 8718
    },
    {
      "epoch": 4.103058823529412,
      "grad_norm": 0.6134987897315012,
      "learning_rate": 9.25873604618701e-05,
      "loss": 0.4961,
      "step": 8719
    },
    {
      "epoch": 4.103529411764706,
      "grad_norm": 0.35931439778273333,
      "learning_rate": 9.254686479352029e-05,
      "loss": 0.5312,
      "step": 8720
    },
    {
      "epoch": 4.104,
      "grad_norm": 0.23911600744054734,
      "learning_rate": 9.250637403195008e-05,
      "loss": 0.2295,
      "step": 8721
    },
    {
      "epoch": 4.104470588235294,
      "grad_norm": 0.24703730034910784,
      "learning_rate": 9.246588818061753e-05,
      "loss": 0.293,
      "step": 8722
    },
    {
      "epoch": 4.1049411764705885,
      "grad_norm": 0.41309233964392433,
      "learning_rate": 9.242540724298032e-05,
      "loss": 0.1895,
      "step": 8723
    },
    {
      "epoch": 4.105411764705883,
      "grad_norm": 0.4955249643700257,
      "learning_rate": 9.238493122249574e-05,
      "loss": 0.5781,
      "step": 8724
    },
    {
      "epoch": 4.105882352941176,
      "grad_norm": 0.6046304897290358,
      "learning_rate": 9.234446012262065e-05,
      "loss": 0.6016,
      "step": 8725
    },
    {
      "epoch": 4.10635294117647,
      "grad_norm": 0.5053396922808785,
      "learning_rate": 9.230399394681141e-05,
      "loss": 0.793,
      "step": 8726
    },
    {
      "epoch": 4.1068235294117645,
      "grad_norm": 1.296189695664184,
      "learning_rate": 9.226353269852413e-05,
      "loss": 0.7148,
      "step": 8727
    },
    {
      "epoch": 4.107294117647059,
      "grad_norm": 0.4486157147929308,
      "learning_rate": 9.222307638121436e-05,
      "loss": 0.4668,
      "step": 8728
    },
    {
      "epoch": 4.107764705882353,
      "grad_norm": 0.3232967449597418,
      "learning_rate": 9.218262499833723e-05,
      "loss": 0.2539,
      "step": 8729
    },
    {
      "epoch": 4.108235294117647,
      "grad_norm": 0.19737909445804327,
      "learning_rate": 9.21421785533475e-05,
      "loss": 0.063,
      "step": 8730
    },
    {
      "epoch": 4.108705882352941,
      "grad_norm": 0.4547779335155782,
      "learning_rate": 9.210173704969952e-05,
      "loss": 0.3066,
      "step": 8731
    },
    {
      "epoch": 4.109176470588236,
      "grad_norm": 0.3229809134606865,
      "learning_rate": 9.206130049084721e-05,
      "loss": 0.1621,
      "step": 8732
    },
    {
      "epoch": 4.109647058823529,
      "grad_norm": 0.47993246886911306,
      "learning_rate": 9.202086888024404e-05,
      "loss": 0.4902,
      "step": 8733
    },
    {
      "epoch": 4.110117647058823,
      "grad_norm": 0.9151206560997882,
      "learning_rate": 9.1980442221343e-05,
      "loss": 0.8477,
      "step": 8734
    },
    {
      "epoch": 4.110588235294117,
      "grad_norm": 0.3084144397421904,
      "learning_rate": 9.194002051759685e-05,
      "loss": 0.1436,
      "step": 8735
    },
    {
      "epoch": 4.111058823529412,
      "grad_norm": 0.16782383801163933,
      "learning_rate": 9.189960377245768e-05,
      "loss": 0.0869,
      "step": 8736
    },
    {
      "epoch": 4.111529411764706,
      "grad_norm": 0.4338319163382772,
      "learning_rate": 9.185919198937736e-05,
      "loss": 0.2754,
      "step": 8737
    },
    {
      "epoch": 4.112,
      "grad_norm": 0.4044996523462007,
      "learning_rate": 9.181878517180719e-05,
      "loss": 0.4316,
      "step": 8738
    },
    {
      "epoch": 4.112470588235294,
      "grad_norm": 0.3362106753972768,
      "learning_rate": 9.177838332319818e-05,
      "loss": 0.2617,
      "step": 8739
    },
    {
      "epoch": 4.1129411764705885,
      "grad_norm": 0.3718729111034109,
      "learning_rate": 9.173798644700079e-05,
      "loss": 0.2275,
      "step": 8740
    },
    {
      "epoch": 4.113411764705883,
      "grad_norm": 0.4360778256497852,
      "learning_rate": 9.169759454666511e-05,
      "loss": 0.5039,
      "step": 8741
    },
    {
      "epoch": 4.113882352941176,
      "grad_norm": 0.518760469667981,
      "learning_rate": 9.165720762564086e-05,
      "loss": 0.2617,
      "step": 8742
    },
    {
      "epoch": 4.11435294117647,
      "grad_norm": 0.5711005779273737,
      "learning_rate": 9.161682568737726e-05,
      "loss": 0.3789,
      "step": 8743
    },
    {
      "epoch": 4.1148235294117645,
      "grad_norm": 0.5090936022550024,
      "learning_rate": 9.157644873532308e-05,
      "loss": 0.5508,
      "step": 8744
    },
    {
      "epoch": 4.115294117647059,
      "grad_norm": 0.27330509556550797,
      "learning_rate": 9.153607677292676e-05,
      "loss": 0.124,
      "step": 8745
    },
    {
      "epoch": 4.115764705882353,
      "grad_norm": 0.36211182661898855,
      "learning_rate": 9.149570980363624e-05,
      "loss": 0.2471,
      "step": 8746
    },
    {
      "epoch": 4.116235294117647,
      "grad_norm": 0.27777388954797955,
      "learning_rate": 9.145534783089909e-05,
      "loss": 0.2617,
      "step": 8747
    },
    {
      "epoch": 4.116705882352941,
      "grad_norm": 0.436220842064467,
      "learning_rate": 9.141499085816237e-05,
      "loss": 0.1816,
      "step": 8748
    },
    {
      "epoch": 4.117176470588236,
      "grad_norm": 0.2951968593177712,
      "learning_rate": 9.137463888887279e-05,
      "loss": 0.1064,
      "step": 8749
    },
    {
      "epoch": 4.117647058823529,
      "grad_norm": 0.35527458966787134,
      "learning_rate": 9.133429192647661e-05,
      "loss": 0.3613,
      "step": 8750
    },
    {
      "epoch": 4.118117647058823,
      "grad_norm": 0.3985915211196465,
      "learning_rate": 9.129394997441963e-05,
      "loss": 0.5,
      "step": 8751
    },
    {
      "epoch": 4.118588235294117,
      "grad_norm": 0.48737033096875887,
      "learning_rate": 9.12536130361473e-05,
      "loss": 0.3477,
      "step": 8752
    },
    {
      "epoch": 4.119058823529412,
      "grad_norm": 0.3907792027615432,
      "learning_rate": 9.121328111510457e-05,
      "loss": 0.1299,
      "step": 8753
    },
    {
      "epoch": 4.119529411764706,
      "grad_norm": 0.4618592017524028,
      "learning_rate": 9.117295421473595e-05,
      "loss": 0.4238,
      "step": 8754
    },
    {
      "epoch": 4.12,
      "grad_norm": 0.28994030884030814,
      "learning_rate": 9.113263233848562e-05,
      "loss": 0.1357,
      "step": 8755
    },
    {
      "epoch": 4.120470588235294,
      "grad_norm": 0.47242089273981797,
      "learning_rate": 9.109231548979722e-05,
      "loss": 0.4609,
      "step": 8756
    },
    {
      "epoch": 4.1209411764705886,
      "grad_norm": 0.497070098929861,
      "learning_rate": 9.105200367211401e-05,
      "loss": 0.3125,
      "step": 8757
    },
    {
      "epoch": 4.121411764705883,
      "grad_norm": 0.18003284788930146,
      "learning_rate": 9.101169688887882e-05,
      "loss": 0.0776,
      "step": 8758
    },
    {
      "epoch": 4.121882352941176,
      "grad_norm": 0.4361783074458374,
      "learning_rate": 9.097139514353407e-05,
      "loss": 0.2676,
      "step": 8759
    },
    {
      "epoch": 4.12235294117647,
      "grad_norm": 0.4770591811386983,
      "learning_rate": 9.093109843952171e-05,
      "loss": 0.4277,
      "step": 8760
    },
    {
      "epoch": 4.1228235294117646,
      "grad_norm": 0.45434039017310285,
      "learning_rate": 9.089080678028325e-05,
      "loss": 0.5352,
      "step": 8761
    },
    {
      "epoch": 4.123294117647059,
      "grad_norm": 0.3754205868925047,
      "learning_rate": 9.085052016925984e-05,
      "loss": 0.3223,
      "step": 8762
    },
    {
      "epoch": 4.123764705882353,
      "grad_norm": 0.2875534874241577,
      "learning_rate": 9.081023860989213e-05,
      "loss": 0.3633,
      "step": 8763
    },
    {
      "epoch": 4.124235294117647,
      "grad_norm": 0.7833564490030605,
      "learning_rate": 9.076996210562035e-05,
      "loss": 0.4531,
      "step": 8764
    },
    {
      "epoch": 4.124705882352941,
      "grad_norm": 0.27805057702759767,
      "learning_rate": 9.072969065988435e-05,
      "loss": 0.1177,
      "step": 8765
    },
    {
      "epoch": 4.125176470588236,
      "grad_norm": 0.48727311924968103,
      "learning_rate": 9.06894242761235e-05,
      "loss": 0.5352,
      "step": 8766
    },
    {
      "epoch": 4.125647058823529,
      "grad_norm": 0.4896887886475817,
      "learning_rate": 9.064916295777669e-05,
      "loss": 0.4805,
      "step": 8767
    },
    {
      "epoch": 4.126117647058823,
      "grad_norm": 0.42214122734282933,
      "learning_rate": 9.06089067082825e-05,
      "loss": 0.2061,
      "step": 8768
    },
    {
      "epoch": 4.1265882352941174,
      "grad_norm": 0.31103343523100396,
      "learning_rate": 9.0568655531079e-05,
      "loss": 0.1143,
      "step": 8769
    },
    {
      "epoch": 4.127058823529412,
      "grad_norm": 0.37775867272263003,
      "learning_rate": 9.052840942960383e-05,
      "loss": 0.2949,
      "step": 8770
    },
    {
      "epoch": 4.127529411764706,
      "grad_norm": 0.5023097421466645,
      "learning_rate": 9.048816840729415e-05,
      "loss": 0.3945,
      "step": 8771
    },
    {
      "epoch": 4.128,
      "grad_norm": 0.2723556790788145,
      "learning_rate": 9.044793246758685e-05,
      "loss": 0.2207,
      "step": 8772
    },
    {
      "epoch": 4.128470588235294,
      "grad_norm": 0.533349946652861,
      "learning_rate": 9.04077016139182e-05,
      "loss": 0.209,
      "step": 8773
    },
    {
      "epoch": 4.128941176470589,
      "grad_norm": 0.3687309328618917,
      "learning_rate": 9.03674758497241e-05,
      "loss": 0.2227,
      "step": 8774
    },
    {
      "epoch": 4.129411764705883,
      "grad_norm": 0.34984226949173075,
      "learning_rate": 9.032725517844004e-05,
      "loss": 0.3496,
      "step": 8775
    },
    {
      "epoch": 4.129882352941176,
      "grad_norm": 0.19539644901417838,
      "learning_rate": 9.028703960350113e-05,
      "loss": 0.0781,
      "step": 8776
    },
    {
      "epoch": 4.13035294117647,
      "grad_norm": 0.4005700172922034,
      "learning_rate": 9.024682912834196e-05,
      "loss": 0.3164,
      "step": 8777
    },
    {
      "epoch": 4.130823529411765,
      "grad_norm": 0.4310544763612339,
      "learning_rate": 9.020662375639663e-05,
      "loss": 0.1855,
      "step": 8778
    },
    {
      "epoch": 4.131294117647059,
      "grad_norm": 0.33658585887600617,
      "learning_rate": 9.016642349109892e-05,
      "loss": 0.2969,
      "step": 8779
    },
    {
      "epoch": 4.131764705882353,
      "grad_norm": 0.45066966705746775,
      "learning_rate": 9.012622833588213e-05,
      "loss": 0.3926,
      "step": 8780
    },
    {
      "epoch": 4.132235294117647,
      "grad_norm": 0.40259026150970534,
      "learning_rate": 9.008603829417913e-05,
      "loss": 0.3477,
      "step": 8781
    },
    {
      "epoch": 4.1327058823529415,
      "grad_norm": 0.309320155916106,
      "learning_rate": 9.00458533694223e-05,
      "loss": 0.1494,
      "step": 8782
    },
    {
      "epoch": 4.133176470588236,
      "grad_norm": 0.5004123687371766,
      "learning_rate": 9.000567356504368e-05,
      "loss": 0.7305,
      "step": 8783
    },
    {
      "epoch": 4.133647058823529,
      "grad_norm": 0.5972160138199382,
      "learning_rate": 8.996549888447484e-05,
      "loss": 0.832,
      "step": 8784
    },
    {
      "epoch": 4.134117647058823,
      "grad_norm": 0.5864809529772539,
      "learning_rate": 8.99253293311468e-05,
      "loss": 0.5742,
      "step": 8785
    },
    {
      "epoch": 4.1345882352941175,
      "grad_norm": 0.4303840680782401,
      "learning_rate": 8.988516490849037e-05,
      "loss": 0.1787,
      "step": 8786
    },
    {
      "epoch": 4.135058823529412,
      "grad_norm": 0.4168481883759779,
      "learning_rate": 8.984500561993569e-05,
      "loss": 0.3867,
      "step": 8787
    },
    {
      "epoch": 4.135529411764706,
      "grad_norm": 0.36539297948281685,
      "learning_rate": 8.98048514689126e-05,
      "loss": 0.1719,
      "step": 8788
    },
    {
      "epoch": 4.136,
      "grad_norm": 0.2749607775520393,
      "learning_rate": 8.976470245885041e-05,
      "loss": 0.1631,
      "step": 8789
    },
    {
      "epoch": 4.136470588235294,
      "grad_norm": 0.2734318419760376,
      "learning_rate": 8.972455859317811e-05,
      "loss": 0.0786,
      "step": 8790
    },
    {
      "epoch": 4.136941176470589,
      "grad_norm": 0.23616812301438492,
      "learning_rate": 8.968441987532417e-05,
      "loss": 0.0845,
      "step": 8791
    },
    {
      "epoch": 4.137411764705883,
      "grad_norm": 0.30090010211906054,
      "learning_rate": 8.964428630871657e-05,
      "loss": 0.2422,
      "step": 8792
    },
    {
      "epoch": 4.137882352941176,
      "grad_norm": 0.27901968083384804,
      "learning_rate": 8.960415789678303e-05,
      "loss": 0.1865,
      "step": 8793
    },
    {
      "epoch": 4.13835294117647,
      "grad_norm": 0.25440179605029845,
      "learning_rate": 8.95640346429506e-05,
      "loss": 0.2188,
      "step": 8794
    },
    {
      "epoch": 4.138823529411765,
      "grad_norm": 0.43415338817714466,
      "learning_rate": 8.952391655064605e-05,
      "loss": 0.3027,
      "step": 8795
    },
    {
      "epoch": 4.139294117647059,
      "grad_norm": 0.3835287498662376,
      "learning_rate": 8.948380362329567e-05,
      "loss": 0.1582,
      "step": 8796
    },
    {
      "epoch": 4.139764705882353,
      "grad_norm": 0.5148089766236587,
      "learning_rate": 8.944369586432531e-05,
      "loss": 0.4199,
      "step": 8797
    },
    {
      "epoch": 4.140235294117647,
      "grad_norm": 0.4414019018014546,
      "learning_rate": 8.940359327716032e-05,
      "loss": 0.3613,
      "step": 8798
    },
    {
      "epoch": 4.1407058823529415,
      "grad_norm": 0.5800437006601528,
      "learning_rate": 8.936349586522569e-05,
      "loss": 0.4531,
      "step": 8799
    },
    {
      "epoch": 4.141176470588236,
      "grad_norm": 0.40910341475376494,
      "learning_rate": 8.932340363194595e-05,
      "loss": 0.4062,
      "step": 8800
    },
    {
      "epoch": 4.141647058823529,
      "grad_norm": 0.3063437574191971,
      "learning_rate": 8.928331658074516e-05,
      "loss": 0.0659,
      "step": 8801
    },
    {
      "epoch": 4.142117647058823,
      "grad_norm": 0.28524909872985904,
      "learning_rate": 8.92432347150469e-05,
      "loss": 0.124,
      "step": 8802
    },
    {
      "epoch": 4.1425882352941175,
      "grad_norm": 0.40606081693568163,
      "learning_rate": 8.920315803827446e-05,
      "loss": 0.4258,
      "step": 8803
    },
    {
      "epoch": 4.143058823529412,
      "grad_norm": 0.6450697499800901,
      "learning_rate": 8.916308655385054e-05,
      "loss": 0.4199,
      "step": 8804
    },
    {
      "epoch": 4.143529411764706,
      "grad_norm": 0.21238743928129097,
      "learning_rate": 8.912302026519738e-05,
      "loss": 0.0693,
      "step": 8805
    },
    {
      "epoch": 4.144,
      "grad_norm": 0.4311538945241728,
      "learning_rate": 8.908295917573692e-05,
      "loss": 0.3242,
      "step": 8806
    },
    {
      "epoch": 4.144470588235294,
      "grad_norm": 0.32036854722483105,
      "learning_rate": 8.904290328889055e-05,
      "loss": 0.2598,
      "step": 8807
    },
    {
      "epoch": 4.144941176470589,
      "grad_norm": 0.3169468029060372,
      "learning_rate": 8.900285260807922e-05,
      "loss": 0.3262,
      "step": 8808
    },
    {
      "epoch": 4.145411764705882,
      "grad_norm": 0.28614882802737746,
      "learning_rate": 8.896280713672344e-05,
      "loss": 0.1455,
      "step": 8809
    },
    {
      "epoch": 4.145882352941176,
      "grad_norm": 0.732073198210832,
      "learning_rate": 8.892276687824335e-05,
      "loss": 0.5781,
      "step": 8810
    },
    {
      "epoch": 4.14635294117647,
      "grad_norm": 0.7236436267389147,
      "learning_rate": 8.888273183605856e-05,
      "loss": 0.8008,
      "step": 8811
    },
    {
      "epoch": 4.146823529411765,
      "grad_norm": 0.28122777726729653,
      "learning_rate": 8.884270201358821e-05,
      "loss": 0.0762,
      "step": 8812
    },
    {
      "epoch": 4.147294117647059,
      "grad_norm": 0.4669148821776752,
      "learning_rate": 8.880267741425111e-05,
      "loss": 0.2871,
      "step": 8813
    },
    {
      "epoch": 4.147764705882353,
      "grad_norm": 0.3068232473174053,
      "learning_rate": 8.876265804146554e-05,
      "loss": 0.1748,
      "step": 8814
    },
    {
      "epoch": 4.148235294117647,
      "grad_norm": 0.23946117968676905,
      "learning_rate": 8.872264389864929e-05,
      "loss": 0.0732,
      "step": 8815
    },
    {
      "epoch": 4.1487058823529415,
      "grad_norm": 0.28239139762854065,
      "learning_rate": 8.868263498921987e-05,
      "loss": 0.2373,
      "step": 8816
    },
    {
      "epoch": 4.149176470588236,
      "grad_norm": 0.5291752912487,
      "learning_rate": 8.864263131659417e-05,
      "loss": 0.5664,
      "step": 8817
    },
    {
      "epoch": 4.149647058823529,
      "grad_norm": 0.3972627590923069,
      "learning_rate": 8.860263288418866e-05,
      "loss": 0.3398,
      "step": 8818
    },
    {
      "epoch": 4.150117647058823,
      "grad_norm": 0.26963479735499113,
      "learning_rate": 8.856263969541948e-05,
      "loss": 0.1348,
      "step": 8819
    },
    {
      "epoch": 4.1505882352941175,
      "grad_norm": 0.4007002494579039,
      "learning_rate": 8.852265175370223e-05,
      "loss": 0.2598,
      "step": 8820
    },
    {
      "epoch": 4.151058823529412,
      "grad_norm": 0.2873692067377464,
      "learning_rate": 8.848266906245207e-05,
      "loss": 0.1982,
      "step": 8821
    },
    {
      "epoch": 4.151529411764706,
      "grad_norm": 0.2954275580922407,
      "learning_rate": 8.844269162508373e-05,
      "loss": 0.1572,
      "step": 8822
    },
    {
      "epoch": 4.152,
      "grad_norm": 0.30626729020159477,
      "learning_rate": 8.840271944501143e-05,
      "loss": 0.0957,
      "step": 8823
    },
    {
      "epoch": 4.152470588235294,
      "grad_norm": 0.44851557675081477,
      "learning_rate": 8.836275252564903e-05,
      "loss": 0.1504,
      "step": 8824
    },
    {
      "epoch": 4.152941176470589,
      "grad_norm": 0.23690887413887235,
      "learning_rate": 8.832279087040992e-05,
      "loss": 0.0776,
      "step": 8825
    },
    {
      "epoch": 4.153411764705883,
      "grad_norm": 0.6361773922630727,
      "learning_rate": 8.828283448270698e-05,
      "loss": 0.4512,
      "step": 8826
    },
    {
      "epoch": 4.153882352941176,
      "grad_norm": 0.45939710138052064,
      "learning_rate": 8.824288336595267e-05,
      "loss": 0.2344,
      "step": 8827
    },
    {
      "epoch": 4.15435294117647,
      "grad_norm": 0.4129048138218797,
      "learning_rate": 8.820293752355906e-05,
      "loss": 0.3027,
      "step": 8828
    },
    {
      "epoch": 4.154823529411765,
      "grad_norm": 0.2920765802622494,
      "learning_rate": 8.816299695893773e-05,
      "loss": 0.0977,
      "step": 8829
    },
    {
      "epoch": 4.155294117647059,
      "grad_norm": 0.4221650946465966,
      "learning_rate": 8.81230616754997e-05,
      "loss": 0.4062,
      "step": 8830
    },
    {
      "epoch": 4.155764705882353,
      "grad_norm": 0.5149429095415254,
      "learning_rate": 8.808313167665575e-05,
      "loss": 0.6875,
      "step": 8831
    },
    {
      "epoch": 4.156235294117647,
      "grad_norm": 0.36182424847837297,
      "learning_rate": 8.804320696581606e-05,
      "loss": 0.4629,
      "step": 8832
    },
    {
      "epoch": 4.1567058823529415,
      "grad_norm": 0.19464768461308682,
      "learning_rate": 8.800328754639038e-05,
      "loss": 0.0564,
      "step": 8833
    },
    {
      "epoch": 4.157176470588236,
      "grad_norm": 0.1556008325170953,
      "learning_rate": 8.796337342178804e-05,
      "loss": 0.0178,
      "step": 8834
    },
    {
      "epoch": 4.157647058823529,
      "grad_norm": 0.4834600053748053,
      "learning_rate": 8.792346459541789e-05,
      "loss": 0.2734,
      "step": 8835
    },
    {
      "epoch": 4.158117647058823,
      "grad_norm": 0.4215298959968829,
      "learning_rate": 8.788356107068833e-05,
      "loss": 0.4004,
      "step": 8836
    },
    {
      "epoch": 4.1585882352941175,
      "grad_norm": 0.4373922822114494,
      "learning_rate": 8.784366285100738e-05,
      "loss": 0.1455,
      "step": 8837
    },
    {
      "epoch": 4.159058823529412,
      "grad_norm": 0.48788820680032646,
      "learning_rate": 8.780376993978246e-05,
      "loss": 0.6445,
      "step": 8838
    },
    {
      "epoch": 4.159529411764706,
      "grad_norm": 0.3457974250524493,
      "learning_rate": 8.77638823404207e-05,
      "loss": 0.2432,
      "step": 8839
    },
    {
      "epoch": 4.16,
      "grad_norm": 0.4304249627621974,
      "learning_rate": 8.772400005632859e-05,
      "loss": 0.2598,
      "step": 8840
    },
    {
      "epoch": 4.160470588235294,
      "grad_norm": 0.4208865195898361,
      "learning_rate": 8.768412309091238e-05,
      "loss": 0.1904,
      "step": 8841
    },
    {
      "epoch": 4.160941176470589,
      "grad_norm": 0.6123827374159859,
      "learning_rate": 8.764425144757771e-05,
      "loss": 0.2598,
      "step": 8842
    },
    {
      "epoch": 4.161411764705882,
      "grad_norm": 0.5087042526239667,
      "learning_rate": 8.760438512972982e-05,
      "loss": 0.3965,
      "step": 8843
    },
    {
      "epoch": 4.161882352941176,
      "grad_norm": 0.34428750485641657,
      "learning_rate": 8.756452414077351e-05,
      "loss": 0.3535,
      "step": 8844
    },
    {
      "epoch": 4.16235294117647,
      "grad_norm": 0.46033445604831286,
      "learning_rate": 8.752466848411305e-05,
      "loss": 0.459,
      "step": 8845
    },
    {
      "epoch": 4.162823529411765,
      "grad_norm": 0.9542698466288834,
      "learning_rate": 8.748481816315233e-05,
      "loss": 1.0078,
      "step": 8846
    },
    {
      "epoch": 4.163294117647059,
      "grad_norm": 0.7346781642969686,
      "learning_rate": 8.744497318129483e-05,
      "loss": 0.5,
      "step": 8847
    },
    {
      "epoch": 4.163764705882353,
      "grad_norm": 0.32140523595142506,
      "learning_rate": 8.74051335419434e-05,
      "loss": 0.1108,
      "step": 8848
    },
    {
      "epoch": 4.164235294117647,
      "grad_norm": 0.560450967868218,
      "learning_rate": 8.736529924850057e-05,
      "loss": 0.7812,
      "step": 8849
    },
    {
      "epoch": 4.1647058823529415,
      "grad_norm": 0.21475416559907892,
      "learning_rate": 8.73254703043685e-05,
      "loss": 0.2402,
      "step": 8850
    },
    {
      "epoch": 4.165176470588236,
      "grad_norm": 0.40831810547917613,
      "learning_rate": 8.72856467129486e-05,
      "loss": 0.582,
      "step": 8851
    },
    {
      "epoch": 4.165647058823529,
      "grad_norm": 0.3112125080521919,
      "learning_rate": 8.72458284776421e-05,
      "loss": 0.1992,
      "step": 8852
    },
    {
      "epoch": 4.166117647058823,
      "grad_norm": 0.349953968682995,
      "learning_rate": 8.72060156018497e-05,
      "loss": 0.1104,
      "step": 8853
    },
    {
      "epoch": 4.1665882352941175,
      "grad_norm": 0.48927644255116404,
      "learning_rate": 8.716620808897151e-05,
      "loss": 0.4082,
      "step": 8854
    },
    {
      "epoch": 4.167058823529412,
      "grad_norm": 0.5340347670937584,
      "learning_rate": 8.712640594240735e-05,
      "loss": 0.5742,
      "step": 8855
    },
    {
      "epoch": 4.167529411764706,
      "grad_norm": 0.2216252163055671,
      "learning_rate": 8.708660916555657e-05,
      "loss": 0.1138,
      "step": 8856
    },
    {
      "epoch": 4.168,
      "grad_norm": 0.4250042795732206,
      "learning_rate": 8.70468177618179e-05,
      "loss": 0.543,
      "step": 8857
    },
    {
      "epoch": 4.168470588235294,
      "grad_norm": 0.4063642622953747,
      "learning_rate": 8.700703173458978e-05,
      "loss": 0.4434,
      "step": 8858
    },
    {
      "epoch": 4.168941176470589,
      "grad_norm": 0.4609962667099972,
      "learning_rate": 8.69672510872702e-05,
      "loss": 0.4766,
      "step": 8859
    },
    {
      "epoch": 4.169411764705882,
      "grad_norm": 0.3159268355672686,
      "learning_rate": 8.692747582325646e-05,
      "loss": 0.2617,
      "step": 8860
    },
    {
      "epoch": 4.169882352941176,
      "grad_norm": 0.23603237916302822,
      "learning_rate": 8.68877059459457e-05,
      "loss": 0.0659,
      "step": 8861
    },
    {
      "epoch": 4.17035294117647,
      "grad_norm": 0.34664477672949695,
      "learning_rate": 8.684794145873443e-05,
      "loss": 0.3047,
      "step": 8862
    },
    {
      "epoch": 4.170823529411765,
      "grad_norm": 0.47803458429538387,
      "learning_rate": 8.680818236501867e-05,
      "loss": 0.5703,
      "step": 8863
    },
    {
      "epoch": 4.171294117647059,
      "grad_norm": 0.3923847475498699,
      "learning_rate": 8.67684286681941e-05,
      "loss": 0.4062,
      "step": 8864
    },
    {
      "epoch": 4.171764705882353,
      "grad_norm": 0.33935343910856325,
      "learning_rate": 8.672868037165587e-05,
      "loss": 0.1309,
      "step": 8865
    },
    {
      "epoch": 4.172235294117647,
      "grad_norm": 0.49721076078266885,
      "learning_rate": 8.668893747879869e-05,
      "loss": 0.3125,
      "step": 8866
    },
    {
      "epoch": 4.1727058823529415,
      "grad_norm": 0.2796811325183777,
      "learning_rate": 8.664919999301683e-05,
      "loss": 0.0781,
      "step": 8867
    },
    {
      "epoch": 4.173176470588236,
      "grad_norm": 0.31300693661206985,
      "learning_rate": 8.660946791770397e-05,
      "loss": 0.1582,
      "step": 8868
    },
    {
      "epoch": 4.173647058823529,
      "grad_norm": 0.17179048490073828,
      "learning_rate": 8.656974125625349e-05,
      "loss": 0.0469,
      "step": 8869
    },
    {
      "epoch": 4.174117647058823,
      "grad_norm": 0.18945214010338177,
      "learning_rate": 8.653002001205827e-05,
      "loss": 0.1396,
      "step": 8870
    },
    {
      "epoch": 4.1745882352941175,
      "grad_norm": 0.513636080432244,
      "learning_rate": 8.649030418851061e-05,
      "loss": 0.293,
      "step": 8871
    },
    {
      "epoch": 4.175058823529412,
      "grad_norm": 0.3625879388885882,
      "learning_rate": 8.645059378900255e-05,
      "loss": 0.2432,
      "step": 8872
    },
    {
      "epoch": 4.175529411764706,
      "grad_norm": 0.4019396543027709,
      "learning_rate": 8.641088881692542e-05,
      "loss": 0.4668,
      "step": 8873
    },
    {
      "epoch": 4.176,
      "grad_norm": 0.4287556812134267,
      "learning_rate": 8.637118927567028e-05,
      "loss": 0.543,
      "step": 8874
    },
    {
      "epoch": 4.176470588235294,
      "grad_norm": 0.4213831074911713,
      "learning_rate": 8.633149516862775e-05,
      "loss": 0.2969,
      "step": 8875
    },
    {
      "epoch": 4.176941176470589,
      "grad_norm": 0.3434539753176729,
      "learning_rate": 8.629180649918775e-05,
      "loss": 0.3281,
      "step": 8876
    },
    {
      "epoch": 4.177411764705882,
      "grad_norm": 0.33534903407054323,
      "learning_rate": 8.625212327073998e-05,
      "loss": 0.1523,
      "step": 8877
    },
    {
      "epoch": 4.177882352941176,
      "grad_norm": 0.5756446688548049,
      "learning_rate": 8.621244548667363e-05,
      "loss": 0.5352,
      "step": 8878
    },
    {
      "epoch": 4.17835294117647,
      "grad_norm": 0.5068799916021853,
      "learning_rate": 8.617277315037723e-05,
      "loss": 0.3613,
      "step": 8879
    },
    {
      "epoch": 4.178823529411765,
      "grad_norm": 0.21888828030240004,
      "learning_rate": 8.613310626523909e-05,
      "loss": 0.0801,
      "step": 8880
    },
    {
      "epoch": 4.179294117647059,
      "grad_norm": 0.244835243990979,
      "learning_rate": 8.609344483464697e-05,
      "loss": 0.0732,
      "step": 8881
    },
    {
      "epoch": 4.179764705882353,
      "grad_norm": 0.25329616905629143,
      "learning_rate": 8.60537888619881e-05,
      "loss": 0.0747,
      "step": 8882
    },
    {
      "epoch": 4.180235294117647,
      "grad_norm": 0.553065891073644,
      "learning_rate": 8.601413835064929e-05,
      "loss": 0.416,
      "step": 8883
    },
    {
      "epoch": 4.1807058823529415,
      "grad_norm": 0.37753462414466366,
      "learning_rate": 8.597449330401698e-05,
      "loss": 0.375,
      "step": 8884
    },
    {
      "epoch": 4.181176470588236,
      "grad_norm": 0.3585200347788356,
      "learning_rate": 8.593485372547693e-05,
      "loss": 0.3438,
      "step": 8885
    },
    {
      "epoch": 4.181647058823529,
      "grad_norm": 0.4253260238642813,
      "learning_rate": 8.589521961841462e-05,
      "loss": 0.3691,
      "step": 8886
    },
    {
      "epoch": 4.182117647058823,
      "grad_norm": 0.4220476900272191,
      "learning_rate": 8.585559098621503e-05,
      "loss": 0.3203,
      "step": 8887
    },
    {
      "epoch": 4.1825882352941175,
      "grad_norm": 0.2566556716792655,
      "learning_rate": 8.581596783226254e-05,
      "loss": 0.0713,
      "step": 8888
    },
    {
      "epoch": 4.183058823529412,
      "grad_norm": 0.2826202764206246,
      "learning_rate": 8.577635015994122e-05,
      "loss": 0.1226,
      "step": 8889
    },
    {
      "epoch": 4.183529411764706,
      "grad_norm": 0.2678328356536219,
      "learning_rate": 8.573673797263466e-05,
      "loss": 0.0947,
      "step": 8890
    },
    {
      "epoch": 4.184,
      "grad_norm": 0.16966658076029326,
      "learning_rate": 8.569713127372583e-05,
      "loss": 0.0391,
      "step": 8891
    },
    {
      "epoch": 4.184470588235294,
      "grad_norm": 0.4618553341869846,
      "learning_rate": 8.565753006659741e-05,
      "loss": 0.3789,
      "step": 8892
    },
    {
      "epoch": 4.184941176470589,
      "grad_norm": 0.7617108166367972,
      "learning_rate": 8.561793435463157e-05,
      "loss": 0.7188,
      "step": 8893
    },
    {
      "epoch": 4.185411764705882,
      "grad_norm": 0.5109523083210262,
      "learning_rate": 8.557834414120986e-05,
      "loss": 0.2617,
      "step": 8894
    },
    {
      "epoch": 4.185882352941176,
      "grad_norm": 0.39975168080841483,
      "learning_rate": 8.553875942971362e-05,
      "loss": 0.3984,
      "step": 8895
    },
    {
      "epoch": 4.18635294117647,
      "grad_norm": 0.36535977783443474,
      "learning_rate": 8.549918022352343e-05,
      "loss": 0.2139,
      "step": 8896
    },
    {
      "epoch": 4.186823529411765,
      "grad_norm": 0.5336933605508185,
      "learning_rate": 8.545960652601964e-05,
      "loss": 0.5,
      "step": 8897
    },
    {
      "epoch": 4.187294117647059,
      "grad_norm": 0.3768327595412157,
      "learning_rate": 8.542003834058207e-05,
      "loss": 0.1875,
      "step": 8898
    },
    {
      "epoch": 4.187764705882353,
      "grad_norm": 0.41399419303281093,
      "learning_rate": 8.538047567058994e-05,
      "loss": 0.2188,
      "step": 8899
    },
    {
      "epoch": 4.188235294117647,
      "grad_norm": 0.3368356156705352,
      "learning_rate": 8.534091851942214e-05,
      "loss": 0.3438,
      "step": 8900
    },
    {
      "epoch": 4.1887058823529415,
      "grad_norm": 0.26067446517109033,
      "learning_rate": 8.530136689045712e-05,
      "loss": 0.1504,
      "step": 8901
    },
    {
      "epoch": 4.189176470588236,
      "grad_norm": 0.27204248811254217,
      "learning_rate": 8.526182078707261e-05,
      "loss": 0.2715,
      "step": 8902
    },
    {
      "epoch": 4.189647058823529,
      "grad_norm": 0.22608814872759064,
      "learning_rate": 8.522228021264618e-05,
      "loss": 0.1699,
      "step": 8903
    },
    {
      "epoch": 4.190117647058823,
      "grad_norm": 0.3813531572664007,
      "learning_rate": 8.518274517055482e-05,
      "loss": 0.3535,
      "step": 8904
    },
    {
      "epoch": 4.1905882352941175,
      "grad_norm": 0.3693123588690012,
      "learning_rate": 8.514321566417488e-05,
      "loss": 0.2695,
      "step": 8905
    },
    {
      "epoch": 4.191058823529412,
      "grad_norm": 0.421050226591291,
      "learning_rate": 8.510369169688243e-05,
      "loss": 0.5898,
      "step": 8906
    },
    {
      "epoch": 4.191529411764706,
      "grad_norm": 0.14335213515681058,
      "learning_rate": 8.506417327205309e-05,
      "loss": 0.0258,
      "step": 8907
    },
    {
      "epoch": 4.192,
      "grad_norm": 0.3869434988794583,
      "learning_rate": 8.502466039306175e-05,
      "loss": 0.2676,
      "step": 8908
    },
    {
      "epoch": 4.192470588235294,
      "grad_norm": 0.5471563788948965,
      "learning_rate": 8.498515306328323e-05,
      "loss": 0.5664,
      "step": 8909
    },
    {
      "epoch": 4.192941176470589,
      "grad_norm": 0.3906388909899967,
      "learning_rate": 8.494565128609146e-05,
      "loss": 0.3086,
      "step": 8910
    },
    {
      "epoch": 4.193411764705882,
      "grad_norm": 0.6296955583972457,
      "learning_rate": 8.490615506486018e-05,
      "loss": 0.3887,
      "step": 8911
    },
    {
      "epoch": 4.193882352941176,
      "grad_norm": 0.36887284862395664,
      "learning_rate": 8.486666440296259e-05,
      "loss": 0.4453,
      "step": 8912
    },
    {
      "epoch": 4.19435294117647,
      "grad_norm": 0.2913721667853938,
      "learning_rate": 8.482717930377129e-05,
      "loss": 0.1553,
      "step": 8913
    },
    {
      "epoch": 4.194823529411765,
      "grad_norm": 0.5981787282966929,
      "learning_rate": 8.478769977065853e-05,
      "loss": 0.4199,
      "step": 8914
    },
    {
      "epoch": 4.195294117647059,
      "grad_norm": 0.1547725374368141,
      "learning_rate": 8.474822580699614e-05,
      "loss": 0.0547,
      "step": 8915
    },
    {
      "epoch": 4.195764705882353,
      "grad_norm": 0.3222709029817559,
      "learning_rate": 8.470875741615527e-05,
      "loss": 0.3477,
      "step": 8916
    },
    {
      "epoch": 4.196235294117647,
      "grad_norm": 0.31617739535902845,
      "learning_rate": 8.466929460150678e-05,
      "loss": 0.4102,
      "step": 8917
    },
    {
      "epoch": 4.1967058823529415,
      "grad_norm": 0.6467260213916086,
      "learning_rate": 8.462983736642101e-05,
      "loss": 0.5156,
      "step": 8918
    },
    {
      "epoch": 4.197176470588236,
      "grad_norm": 0.4608698012902347,
      "learning_rate": 8.459038571426772e-05,
      "loss": 0.2559,
      "step": 8919
    },
    {
      "epoch": 4.197647058823529,
      "grad_norm": 0.5923649886829965,
      "learning_rate": 8.455093964841632e-05,
      "loss": 0.6055,
      "step": 8920
    },
    {
      "epoch": 4.198117647058823,
      "grad_norm": 0.49404340438547717,
      "learning_rate": 8.451149917223576e-05,
      "loss": 0.375,
      "step": 8921
    },
    {
      "epoch": 4.1985882352941175,
      "grad_norm": 0.5181911667949981,
      "learning_rate": 8.447206428909434e-05,
      "loss": 0.1875,
      "step": 8922
    },
    {
      "epoch": 4.199058823529412,
      "grad_norm": 0.33074846261058205,
      "learning_rate": 8.44326350023601e-05,
      "loss": 0.126,
      "step": 8923
    },
    {
      "epoch": 4.199529411764706,
      "grad_norm": 0.37421376920786364,
      "learning_rate": 8.439321131540036e-05,
      "loss": 0.1455,
      "step": 8924
    },
    {
      "epoch": 4.2,
      "grad_norm": 0.32153755531164996,
      "learning_rate": 8.435379323158218e-05,
      "loss": 0.1777,
      "step": 8925
    },
    {
      "epoch": 4.200470588235294,
      "grad_norm": 0.35841456484021156,
      "learning_rate": 8.431438075427213e-05,
      "loss": 0.3379,
      "step": 8926
    },
    {
      "epoch": 4.200941176470589,
      "grad_norm": 0.22772509445032035,
      "learning_rate": 8.427497388683607e-05,
      "loss": 0.0869,
      "step": 8927
    },
    {
      "epoch": 4.201411764705882,
      "grad_norm": 0.4221320227520223,
      "learning_rate": 8.423557263263961e-05,
      "loss": 0.1855,
      "step": 8928
    },
    {
      "epoch": 4.201882352941176,
      "grad_norm": 0.37903409813858474,
      "learning_rate": 8.419617699504791e-05,
      "loss": 0.1445,
      "step": 8929
    },
    {
      "epoch": 4.20235294117647,
      "grad_norm": 0.23866285195887801,
      "learning_rate": 8.415678697742538e-05,
      "loss": 0.1768,
      "step": 8930
    },
    {
      "epoch": 4.202823529411765,
      "grad_norm": 0.4494876725819669,
      "learning_rate": 8.41174025831362e-05,
      "loss": 0.3457,
      "step": 8931
    },
    {
      "epoch": 4.203294117647059,
      "grad_norm": 0.6425528989559663,
      "learning_rate": 8.407802381554406e-05,
      "loss": 0.7148,
      "step": 8932
    },
    {
      "epoch": 4.203764705882353,
      "grad_norm": 0.4818237775959585,
      "learning_rate": 8.403865067801197e-05,
      "loss": 0.4238,
      "step": 8933
    },
    {
      "epoch": 4.204235294117647,
      "grad_norm": 0.2096202729180309,
      "learning_rate": 8.399928317390265e-05,
      "loss": 0.0444,
      "step": 8934
    },
    {
      "epoch": 4.2047058823529415,
      "grad_norm": 0.31578844854283405,
      "learning_rate": 8.395992130657834e-05,
      "loss": 0.2578,
      "step": 8935
    },
    {
      "epoch": 4.205176470588235,
      "grad_norm": 0.3233336843478089,
      "learning_rate": 8.392056507940061e-05,
      "loss": 0.3672,
      "step": 8936
    },
    {
      "epoch": 4.205647058823529,
      "grad_norm": 0.402408789353438,
      "learning_rate": 8.388121449573075e-05,
      "loss": 0.1445,
      "step": 8937
    },
    {
      "epoch": 4.206117647058823,
      "grad_norm": 0.23835104677949534,
      "learning_rate": 8.384186955892954e-05,
      "loss": 0.0835,
      "step": 8938
    },
    {
      "epoch": 4.2065882352941175,
      "grad_norm": 0.26998734923016876,
      "learning_rate": 8.380253027235709e-05,
      "loss": 0.1533,
      "step": 8939
    },
    {
      "epoch": 4.207058823529412,
      "grad_norm": 0.27655378235343536,
      "learning_rate": 8.37631966393733e-05,
      "loss": 0.0708,
      "step": 8940
    },
    {
      "epoch": 4.207529411764706,
      "grad_norm": 0.3453185027706619,
      "learning_rate": 8.372386866333743e-05,
      "loss": 0.3594,
      "step": 8941
    },
    {
      "epoch": 4.208,
      "grad_norm": 0.26965871677184666,
      "learning_rate": 8.36845463476082e-05,
      "loss": 0.2217,
      "step": 8942
    },
    {
      "epoch": 4.208470588235294,
      "grad_norm": 0.595684281860108,
      "learning_rate": 8.364522969554399e-05,
      "loss": 0.3145,
      "step": 8943
    },
    {
      "epoch": 4.208941176470589,
      "grad_norm": 0.4550053736760946,
      "learning_rate": 8.36059187105027e-05,
      "loss": 0.3984,
      "step": 8944
    },
    {
      "epoch": 4.209411764705882,
      "grad_norm": 0.4125782057901564,
      "learning_rate": 8.356661339584156e-05,
      "loss": 0.3125,
      "step": 8945
    },
    {
      "epoch": 4.209882352941176,
      "grad_norm": 0.39022011692449415,
      "learning_rate": 8.352731375491754e-05,
      "loss": 0.4258,
      "step": 8946
    },
    {
      "epoch": 4.21035294117647,
      "grad_norm": 0.4399747281351482,
      "learning_rate": 8.348801979108691e-05,
      "loss": 0.416,
      "step": 8947
    },
    {
      "epoch": 4.210823529411765,
      "grad_norm": 0.4605410535596943,
      "learning_rate": 8.344873150770562e-05,
      "loss": 0.5547,
      "step": 8948
    },
    {
      "epoch": 4.211294117647059,
      "grad_norm": 0.3329047670505672,
      "learning_rate": 8.340944890812916e-05,
      "loss": 0.1611,
      "step": 8949
    },
    {
      "epoch": 4.211764705882353,
      "grad_norm": 0.39902218630143294,
      "learning_rate": 8.337017199571235e-05,
      "loss": 0.5586,
      "step": 8950
    },
    {
      "epoch": 4.212235294117647,
      "grad_norm": 0.5048986463108406,
      "learning_rate": 8.333090077380964e-05,
      "loss": 0.4531,
      "step": 8951
    },
    {
      "epoch": 4.2127058823529415,
      "grad_norm": 0.34058140322220654,
      "learning_rate": 8.329163524577504e-05,
      "loss": 0.1201,
      "step": 8952
    },
    {
      "epoch": 4.213176470588236,
      "grad_norm": 0.17429424159134643,
      "learning_rate": 8.325237541496197e-05,
      "loss": 0.033,
      "step": 8953
    },
    {
      "epoch": 4.213647058823529,
      "grad_norm": 0.34035632341906596,
      "learning_rate": 8.32131212847235e-05,
      "loss": 0.4102,
      "step": 8954
    },
    {
      "epoch": 4.214117647058823,
      "grad_norm": 0.16590553784806927,
      "learning_rate": 8.3173872858412e-05,
      "loss": 0.105,
      "step": 8955
    },
    {
      "epoch": 4.2145882352941175,
      "grad_norm": 0.5477086316987575,
      "learning_rate": 8.313463013937953e-05,
      "loss": 0.3535,
      "step": 8956
    },
    {
      "epoch": 4.215058823529412,
      "grad_norm": 0.3217264553188834,
      "learning_rate": 8.309539313097769e-05,
      "loss": 0.0908,
      "step": 8957
    },
    {
      "epoch": 4.215529411764706,
      "grad_norm": 0.10633991429785994,
      "learning_rate": 8.305616183655738e-05,
      "loss": 0.0102,
      "step": 8958
    },
    {
      "epoch": 4.216,
      "grad_norm": 0.6064673644886337,
      "learning_rate": 8.30169362594692e-05,
      "loss": 0.4121,
      "step": 8959
    },
    {
      "epoch": 4.216470588235294,
      "grad_norm": 0.48012608159376874,
      "learning_rate": 8.29777164030633e-05,
      "loss": 0.2002,
      "step": 8960
    },
    {
      "epoch": 4.216941176470589,
      "grad_norm": 0.49256175065715496,
      "learning_rate": 8.293850227068908e-05,
      "loss": 0.5312,
      "step": 8961
    },
    {
      "epoch": 4.217411764705882,
      "grad_norm": 0.2296691658916398,
      "learning_rate": 8.289929386569569e-05,
      "loss": 0.0562,
      "step": 8962
    },
    {
      "epoch": 4.217882352941176,
      "grad_norm": 0.3546131400098165,
      "learning_rate": 8.28600911914318e-05,
      "loss": 0.207,
      "step": 8963
    },
    {
      "epoch": 4.21835294117647,
      "grad_norm": 0.47247881351804866,
      "learning_rate": 8.282089425124538e-05,
      "loss": 0.2412,
      "step": 8964
    },
    {
      "epoch": 4.218823529411765,
      "grad_norm": 0.3609681704774687,
      "learning_rate": 8.278170304848413e-05,
      "loss": 0.1826,
      "step": 8965
    },
    {
      "epoch": 4.219294117647059,
      "grad_norm": 0.3112577707685446,
      "learning_rate": 8.274251758649518e-05,
      "loss": 0.1226,
      "step": 8966
    },
    {
      "epoch": 4.219764705882353,
      "grad_norm": 0.4245902647505047,
      "learning_rate": 8.270333786862506e-05,
      "loss": 0.2871,
      "step": 8967
    },
    {
      "epoch": 4.220235294117647,
      "grad_norm": 0.3447035302715882,
      "learning_rate": 8.266416389821997e-05,
      "loss": 0.2158,
      "step": 8968
    },
    {
      "epoch": 4.2207058823529415,
      "grad_norm": 0.5604035516665866,
      "learning_rate": 8.262499567862565e-05,
      "loss": 0.2168,
      "step": 8969
    },
    {
      "epoch": 4.221176470588235,
      "grad_norm": 0.45921870361485717,
      "learning_rate": 8.258583321318708e-05,
      "loss": 0.459,
      "step": 8970
    },
    {
      "epoch": 4.221647058823529,
      "grad_norm": 0.16383588912396452,
      "learning_rate": 8.254667650524904e-05,
      "loss": 0.0354,
      "step": 8971
    },
    {
      "epoch": 4.222117647058823,
      "grad_norm": 0.4851346897434837,
      "learning_rate": 8.250752555815574e-05,
      "loss": 0.5273,
      "step": 8972
    },
    {
      "epoch": 4.2225882352941175,
      "grad_norm": 0.6536019781926283,
      "learning_rate": 8.246838037525073e-05,
      "loss": 0.5898,
      "step": 8973
    },
    {
      "epoch": 4.223058823529412,
      "grad_norm": 0.27698617900895833,
      "learning_rate": 8.242924095987736e-05,
      "loss": 0.1797,
      "step": 8974
    },
    {
      "epoch": 4.223529411764706,
      "grad_norm": 0.5286799184432726,
      "learning_rate": 8.239010731537816e-05,
      "loss": 0.3262,
      "step": 8975
    },
    {
      "epoch": 4.224,
      "grad_norm": 0.5365577478706604,
      "learning_rate": 8.235097944509545e-05,
      "loss": 0.707,
      "step": 8976
    },
    {
      "epoch": 4.224470588235294,
      "grad_norm": 0.41100849659970456,
      "learning_rate": 8.231185735237096e-05,
      "loss": 0.3242,
      "step": 8977
    },
    {
      "epoch": 4.224941176470589,
      "grad_norm": 0.4272047869228037,
      "learning_rate": 8.22727410405458e-05,
      "loss": 0.5234,
      "step": 8978
    },
    {
      "epoch": 4.225411764705882,
      "grad_norm": 0.2914383657847591,
      "learning_rate": 8.223363051296074e-05,
      "loss": 0.2891,
      "step": 8979
    },
    {
      "epoch": 4.225882352941176,
      "grad_norm": 0.4780062140230397,
      "learning_rate": 8.21945257729561e-05,
      "loss": 0.2891,
      "step": 8980
    },
    {
      "epoch": 4.22635294117647,
      "grad_norm": 0.5521767742491873,
      "learning_rate": 8.215542682387148e-05,
      "loss": 0.4277,
      "step": 8981
    },
    {
      "epoch": 4.226823529411765,
      "grad_norm": 0.3512717431032179,
      "learning_rate": 8.211633366904618e-05,
      "loss": 0.3984,
      "step": 8982
    },
    {
      "epoch": 4.227294117647059,
      "grad_norm": 0.5450746271071106,
      "learning_rate": 8.207724631181904e-05,
      "loss": 0.3711,
      "step": 8983
    },
    {
      "epoch": 4.227764705882353,
      "grad_norm": 0.4456555517512262,
      "learning_rate": 8.203816475552813e-05,
      "loss": 0.5547,
      "step": 8984
    },
    {
      "epoch": 4.228235294117647,
      "grad_norm": 0.2912081591767698,
      "learning_rate": 8.19990890035113e-05,
      "loss": 0.3672,
      "step": 8985
    },
    {
      "epoch": 4.2287058823529415,
      "grad_norm": 0.4502130201326533,
      "learning_rate": 8.196001905910589e-05,
      "loss": 0.3105,
      "step": 8986
    },
    {
      "epoch": 4.229176470588235,
      "grad_norm": 0.6589745338143242,
      "learning_rate": 8.192095492564851e-05,
      "loss": 0.5703,
      "step": 8987
    },
    {
      "epoch": 4.229647058823529,
      "grad_norm": 0.5147757340026036,
      "learning_rate": 8.188189660647554e-05,
      "loss": 0.6562,
      "step": 8988
    },
    {
      "epoch": 4.230117647058823,
      "grad_norm": 0.15007465375028867,
      "learning_rate": 8.184284410492274e-05,
      "loss": 0.0214,
      "step": 8989
    },
    {
      "epoch": 4.2305882352941175,
      "grad_norm": 0.3877789583681976,
      "learning_rate": 8.180379742432533e-05,
      "loss": 0.3281,
      "step": 8990
    },
    {
      "epoch": 4.231058823529412,
      "grad_norm": 0.5320303579680093,
      "learning_rate": 8.176475656801811e-05,
      "loss": 0.2617,
      "step": 8991
    },
    {
      "epoch": 4.231529411764706,
      "grad_norm": 0.2918865958474913,
      "learning_rate": 8.172572153933541e-05,
      "loss": 0.0952,
      "step": 8992
    },
    {
      "epoch": 4.232,
      "grad_norm": 0.3982254008383666,
      "learning_rate": 8.168669234161094e-05,
      "loss": 0.373,
      "step": 8993
    },
    {
      "epoch": 4.232470588235294,
      "grad_norm": 0.45249882901591837,
      "learning_rate": 8.164766897817806e-05,
      "loss": 0.6055,
      "step": 8994
    },
    {
      "epoch": 4.232941176470589,
      "grad_norm": 0.18428198896191197,
      "learning_rate": 8.160865145236946e-05,
      "loss": 0.0579,
      "step": 8995
    },
    {
      "epoch": 4.233411764705882,
      "grad_norm": 0.3133935674198944,
      "learning_rate": 8.15696397675175e-05,
      "loss": 0.2402,
      "step": 8996
    },
    {
      "epoch": 4.233882352941176,
      "grad_norm": 0.43942316453640645,
      "learning_rate": 8.153063392695394e-05,
      "loss": 0.4043,
      "step": 8997
    },
    {
      "epoch": 4.23435294117647,
      "grad_norm": 0.621524113986238,
      "learning_rate": 8.14916339340101e-05,
      "loss": 0.3281,
      "step": 8998
    },
    {
      "epoch": 4.234823529411765,
      "grad_norm": 0.35238387454813264,
      "learning_rate": 8.145263979201672e-05,
      "loss": 0.291,
      "step": 8999
    },
    {
      "epoch": 4.235294117647059,
      "grad_norm": 0.38138846466157894,
      "learning_rate": 8.141365150430421e-05,
      "loss": 0.3496,
      "step": 9000
    },
    {
      "epoch": 4.235764705882353,
      "grad_norm": 0.3951122971831811,
      "learning_rate": 8.137466907420222e-05,
      "loss": 0.2695,
      "step": 9001
    },
    {
      "epoch": 4.236235294117647,
      "grad_norm": 0.2273920759369876,
      "learning_rate": 8.133569250504013e-05,
      "loss": 0.124,
      "step": 9002
    },
    {
      "epoch": 4.2367058823529415,
      "grad_norm": 0.28181126970501513,
      "learning_rate": 8.129672180014666e-05,
      "loss": 0.1387,
      "step": 9003
    },
    {
      "epoch": 4.237176470588235,
      "grad_norm": 0.1818079278964884,
      "learning_rate": 8.125775696285013e-05,
      "loss": 0.0437,
      "step": 9004
    },
    {
      "epoch": 4.237647058823529,
      "grad_norm": 0.426893824428911,
      "learning_rate": 8.121879799647839e-05,
      "loss": 0.3262,
      "step": 9005
    },
    {
      "epoch": 4.238117647058823,
      "grad_norm": 0.3185441936249299,
      "learning_rate": 8.117984490435861e-05,
      "loss": 0.1147,
      "step": 9006
    },
    {
      "epoch": 4.2385882352941175,
      "grad_norm": 0.3919883833053011,
      "learning_rate": 8.114089768981765e-05,
      "loss": 0.4238,
      "step": 9007
    },
    {
      "epoch": 4.239058823529412,
      "grad_norm": 0.3624663528600621,
      "learning_rate": 8.110195635618184e-05,
      "loss": 0.1357,
      "step": 9008
    },
    {
      "epoch": 4.239529411764706,
      "grad_norm": 0.1732414377967584,
      "learning_rate": 8.106302090677682e-05,
      "loss": 0.0349,
      "step": 9009
    },
    {
      "epoch": 4.24,
      "grad_norm": 0.15757137476587735,
      "learning_rate": 8.102409134492795e-05,
      "loss": 0.0562,
      "step": 9010
    },
    {
      "epoch": 4.240470588235294,
      "grad_norm": 0.37909316056778747,
      "learning_rate": 8.098516767396008e-05,
      "loss": 0.1865,
      "step": 9011
    },
    {
      "epoch": 4.240941176470589,
      "grad_norm": 0.2651117988287469,
      "learning_rate": 8.094624989719734e-05,
      "loss": 0.0874,
      "step": 9012
    },
    {
      "epoch": 4.241411764705882,
      "grad_norm": 0.2921014879989212,
      "learning_rate": 8.090733801796356e-05,
      "loss": 0.1299,
      "step": 9013
    },
    {
      "epoch": 4.241882352941176,
      "grad_norm": 0.9154408015602191,
      "learning_rate": 8.086843203958207e-05,
      "loss": 0.5234,
      "step": 9014
    },
    {
      "epoch": 4.24235294117647,
      "grad_norm": 0.3347957901765982,
      "learning_rate": 8.082953196537553e-05,
      "loss": 0.1533,
      "step": 9015
    },
    {
      "epoch": 4.242823529411765,
      "grad_norm": 0.6249953222851355,
      "learning_rate": 8.079063779866622e-05,
      "loss": 0.6289,
      "step": 9016
    },
    {
      "epoch": 4.243294117647059,
      "grad_norm": 0.624169980602315,
      "learning_rate": 8.075174954277598e-05,
      "loss": 0.5898,
      "step": 9017
    },
    {
      "epoch": 4.243764705882353,
      "grad_norm": 0.4955508284234061,
      "learning_rate": 8.071286720102594e-05,
      "loss": 0.668,
      "step": 9018
    },
    {
      "epoch": 4.244235294117647,
      "grad_norm": 0.412575580208075,
      "learning_rate": 8.067399077673688e-05,
      "loss": 0.1069,
      "step": 9019
    },
    {
      "epoch": 4.2447058823529416,
      "grad_norm": 0.5151502956418564,
      "learning_rate": 8.06351202732291e-05,
      "loss": 0.5859,
      "step": 9020
    },
    {
      "epoch": 4.245176470588235,
      "grad_norm": 0.31744867919526343,
      "learning_rate": 8.059625569382223e-05,
      "loss": 0.1582,
      "step": 9021
    },
    {
      "epoch": 4.245647058823529,
      "grad_norm": 0.427931883165069,
      "learning_rate": 8.055739704183553e-05,
      "loss": 0.2812,
      "step": 9022
    },
    {
      "epoch": 4.246117647058823,
      "grad_norm": 0.5492047334942846,
      "learning_rate": 8.051854432058785e-05,
      "loss": 0.832,
      "step": 9023
    },
    {
      "epoch": 4.2465882352941176,
      "grad_norm": 0.6023517268691015,
      "learning_rate": 8.047969753339717e-05,
      "loss": 0.3711,
      "step": 9024
    },
    {
      "epoch": 4.247058823529412,
      "grad_norm": 0.44219790766303707,
      "learning_rate": 8.04408566835814e-05,
      "loss": 0.1543,
      "step": 9025
    },
    {
      "epoch": 4.247529411764706,
      "grad_norm": 0.33457990270836546,
      "learning_rate": 8.040202177445761e-05,
      "loss": 0.252,
      "step": 9026
    },
    {
      "epoch": 4.248,
      "grad_norm": 0.3280999574485532,
      "learning_rate": 8.036319280934251e-05,
      "loss": 0.1797,
      "step": 9027
    },
    {
      "epoch": 4.248470588235294,
      "grad_norm": 0.511460831467718,
      "learning_rate": 8.032436979155237e-05,
      "loss": 0.291,
      "step": 9028
    },
    {
      "epoch": 4.248941176470589,
      "grad_norm": 0.32537057976768236,
      "learning_rate": 8.02855527244028e-05,
      "loss": 0.0991,
      "step": 9029
    },
    {
      "epoch": 4.249411764705882,
      "grad_norm": 0.34047023176600305,
      "learning_rate": 8.024674161120892e-05,
      "loss": 0.2812,
      "step": 9030
    },
    {
      "epoch": 4.249882352941176,
      "grad_norm": 0.30667138658093723,
      "learning_rate": 8.020793645528552e-05,
      "loss": 0.3125,
      "step": 9031
    },
    {
      "epoch": 4.2503529411764704,
      "grad_norm": 0.4764182670988687,
      "learning_rate": 8.016913725994661e-05,
      "loss": 0.2598,
      "step": 9032
    },
    {
      "epoch": 4.250823529411765,
      "grad_norm": 0.3798844951744885,
      "learning_rate": 8.01303440285059e-05,
      "loss": 0.377,
      "step": 9033
    },
    {
      "epoch": 4.251294117647059,
      "grad_norm": 0.9880161384856512,
      "learning_rate": 8.009155676427659e-05,
      "loss": 0.6719,
      "step": 9034
    },
    {
      "epoch": 4.251764705882353,
      "grad_norm": 0.7209061108287825,
      "learning_rate": 8.005277547057114e-05,
      "loss": 0.4316,
      "step": 9035
    },
    {
      "epoch": 4.252235294117647,
      "grad_norm": 0.31722245270000354,
      "learning_rate": 8.001400015070178e-05,
      "loss": 0.2227,
      "step": 9036
    },
    {
      "epoch": 4.252705882352942,
      "grad_norm": 0.2385171906817742,
      "learning_rate": 7.997523080798014e-05,
      "loss": 0.0708,
      "step": 9037
    },
    {
      "epoch": 4.253176470588235,
      "grad_norm": 0.45547344989648647,
      "learning_rate": 7.993646744571719e-05,
      "loss": 0.3809,
      "step": 9038
    },
    {
      "epoch": 4.253647058823529,
      "grad_norm": 0.5277431350098755,
      "learning_rate": 7.98977100672236e-05,
      "loss": 0.5938,
      "step": 9039
    },
    {
      "epoch": 4.254117647058823,
      "grad_norm": 0.2626976059738847,
      "learning_rate": 7.985895867580948e-05,
      "loss": 0.1436,
      "step": 9040
    },
    {
      "epoch": 4.254588235294118,
      "grad_norm": 0.3099630482951901,
      "learning_rate": 7.982021327478419e-05,
      "loss": 0.1445,
      "step": 9041
    },
    {
      "epoch": 4.255058823529412,
      "grad_norm": 0.3563555363495139,
      "learning_rate": 7.978147386745705e-05,
      "loss": 0.2148,
      "step": 9042
    },
    {
      "epoch": 4.255529411764706,
      "grad_norm": 0.1999500003898729,
      "learning_rate": 7.974274045713638e-05,
      "loss": 0.0576,
      "step": 9043
    },
    {
      "epoch": 4.256,
      "grad_norm": 0.23688350016525578,
      "learning_rate": 7.97040130471303e-05,
      "loss": 0.0547,
      "step": 9044
    },
    {
      "epoch": 4.2564705882352944,
      "grad_norm": 0.18589458061546332,
      "learning_rate": 7.966529164074636e-05,
      "loss": 0.0457,
      "step": 9045
    },
    {
      "epoch": 4.256941176470589,
      "grad_norm": 0.6164737729293894,
      "learning_rate": 7.962657624129147e-05,
      "loss": 0.3652,
      "step": 9046
    },
    {
      "epoch": 4.257411764705882,
      "grad_norm": 0.5880349945906803,
      "learning_rate": 7.958786685207212e-05,
      "loss": 0.4941,
      "step": 9047
    },
    {
      "epoch": 4.257882352941176,
      "grad_norm": 0.46293212057028876,
      "learning_rate": 7.954916347639438e-05,
      "loss": 0.4434,
      "step": 9048
    },
    {
      "epoch": 4.2583529411764705,
      "grad_norm": 0.13777477476430347,
      "learning_rate": 7.951046611756358e-05,
      "loss": 0.0349,
      "step": 9049
    },
    {
      "epoch": 4.258823529411765,
      "grad_norm": 0.46857081831244024,
      "learning_rate": 7.947177477888472e-05,
      "loss": 0.7539,
      "step": 9050
    },
    {
      "epoch": 4.259294117647059,
      "grad_norm": 0.30332990429597434,
      "learning_rate": 7.94330894636623e-05,
      "loss": 0.106,
      "step": 9051
    },
    {
      "epoch": 4.259764705882353,
      "grad_norm": 0.19762532922475537,
      "learning_rate": 7.939441017520011e-05,
      "loss": 0.0996,
      "step": 9052
    },
    {
      "epoch": 4.260235294117647,
      "grad_norm": 0.6654483854184184,
      "learning_rate": 7.935573691680166e-05,
      "loss": 0.3984,
      "step": 9053
    },
    {
      "epoch": 4.260705882352942,
      "grad_norm": 0.26158959681932814,
      "learning_rate": 7.931706969176972e-05,
      "loss": 0.0864,
      "step": 9054
    },
    {
      "epoch": 4.261176470588235,
      "grad_norm": 0.2816171604580659,
      "learning_rate": 7.927840850340673e-05,
      "loss": 0.1348,
      "step": 9055
    },
    {
      "epoch": 4.261647058823529,
      "grad_norm": 0.3735588916346811,
      "learning_rate": 7.92397533550146e-05,
      "loss": 0.3789,
      "step": 9056
    },
    {
      "epoch": 4.262117647058823,
      "grad_norm": 0.5417027791644073,
      "learning_rate": 7.920110424989455e-05,
      "loss": 0.3242,
      "step": 9057
    },
    {
      "epoch": 4.262588235294118,
      "grad_norm": 0.19117423884789345,
      "learning_rate": 7.916246119134748e-05,
      "loss": 0.1211,
      "step": 9058
    },
    {
      "epoch": 4.263058823529412,
      "grad_norm": 0.29834070069037333,
      "learning_rate": 7.912382418267374e-05,
      "loss": 0.0825,
      "step": 9059
    },
    {
      "epoch": 4.263529411764706,
      "grad_norm": 0.3502997045147925,
      "learning_rate": 7.908519322717297e-05,
      "loss": 0.2432,
      "step": 9060
    },
    {
      "epoch": 4.264,
      "grad_norm": 0.23882826552429653,
      "learning_rate": 7.904656832814457e-05,
      "loss": 0.2021,
      "step": 9061
    },
    {
      "epoch": 4.2644705882352945,
      "grad_norm": 0.3626007430896375,
      "learning_rate": 7.900794948888728e-05,
      "loss": 0.2012,
      "step": 9062
    },
    {
      "epoch": 4.264941176470588,
      "grad_norm": 0.19775690740772603,
      "learning_rate": 7.896933671269929e-05,
      "loss": 0.1357,
      "step": 9063
    },
    {
      "epoch": 4.265411764705882,
      "grad_norm": 0.5346824018533897,
      "learning_rate": 7.893073000287834e-05,
      "loss": 0.6562,
      "step": 9064
    },
    {
      "epoch": 4.265882352941176,
      "grad_norm": 0.4797631584286585,
      "learning_rate": 7.88921293627217e-05,
      "loss": 0.2832,
      "step": 9065
    },
    {
      "epoch": 4.2663529411764705,
      "grad_norm": 0.4901976552088112,
      "learning_rate": 7.885353479552592e-05,
      "loss": 0.3516,
      "step": 9066
    },
    {
      "epoch": 4.266823529411765,
      "grad_norm": 0.2913233443204986,
      "learning_rate": 7.881494630458726e-05,
      "loss": 0.3887,
      "step": 9067
    },
    {
      "epoch": 4.267294117647059,
      "grad_norm": 0.48250766537558415,
      "learning_rate": 7.87763638932014e-05,
      "loss": 0.334,
      "step": 9068
    },
    {
      "epoch": 4.267764705882353,
      "grad_norm": 0.7031188107097764,
      "learning_rate": 7.873778756466335e-05,
      "loss": 0.4316,
      "step": 9069
    },
    {
      "epoch": 4.268235294117647,
      "grad_norm": 0.40019821739763406,
      "learning_rate": 7.869921732226779e-05,
      "loss": 0.4473,
      "step": 9070
    },
    {
      "epoch": 4.268705882352942,
      "grad_norm": 0.4351193612879972,
      "learning_rate": 7.866065316930886e-05,
      "loss": 0.2471,
      "step": 9071
    },
    {
      "epoch": 4.269176470588235,
      "grad_norm": 0.5367589489606975,
      "learning_rate": 7.862209510907999e-05,
      "loss": 0.3633,
      "step": 9072
    },
    {
      "epoch": 4.269647058823529,
      "grad_norm": 0.5925755014103721,
      "learning_rate": 7.858354314487432e-05,
      "loss": 0.7852,
      "step": 9073
    },
    {
      "epoch": 4.270117647058823,
      "grad_norm": 0.8244132338564611,
      "learning_rate": 7.854499727998442e-05,
      "loss": 0.8945,
      "step": 9074
    },
    {
      "epoch": 4.270588235294118,
      "grad_norm": 0.4805246922598709,
      "learning_rate": 7.850645751770217e-05,
      "loss": 0.5391,
      "step": 9075
    },
    {
      "epoch": 4.271058823529412,
      "grad_norm": 0.41702521922165997,
      "learning_rate": 7.846792386131917e-05,
      "loss": 0.5078,
      "step": 9076
    },
    {
      "epoch": 4.271529411764706,
      "grad_norm": 0.2937267568493849,
      "learning_rate": 7.84293963141263e-05,
      "loss": 0.082,
      "step": 9077
    },
    {
      "epoch": 4.272,
      "grad_norm": 0.2625696147631516,
      "learning_rate": 7.839087487941403e-05,
      "loss": 0.1973,
      "step": 9078
    },
    {
      "epoch": 4.2724705882352945,
      "grad_norm": 0.5761167853088186,
      "learning_rate": 7.835235956047234e-05,
      "loss": 0.6016,
      "step": 9079
    },
    {
      "epoch": 4.272941176470589,
      "grad_norm": 0.30928421199855494,
      "learning_rate": 7.83138503605905e-05,
      "loss": 0.1768,
      "step": 9080
    },
    {
      "epoch": 4.273411764705882,
      "grad_norm": 0.33603140588383346,
      "learning_rate": 7.827534728305747e-05,
      "loss": 0.3301,
      "step": 9081
    },
    {
      "epoch": 4.273882352941176,
      "grad_norm": 0.6661794304327708,
      "learning_rate": 7.823685033116166e-05,
      "loss": 0.2988,
      "step": 9082
    },
    {
      "epoch": 4.2743529411764705,
      "grad_norm": 0.367757522010398,
      "learning_rate": 7.819835950819078e-05,
      "loss": 0.291,
      "step": 9083
    },
    {
      "epoch": 4.274823529411765,
      "grad_norm": 0.20114135846268025,
      "learning_rate": 7.815987481743215e-05,
      "loss": 0.0562,
      "step": 9084
    },
    {
      "epoch": 4.275294117647059,
      "grad_norm": 0.6728082132060992,
      "learning_rate": 7.812139626217266e-05,
      "loss": 0.7734,
      "step": 9085
    },
    {
      "epoch": 4.275764705882353,
      "grad_norm": 0.4153209768453291,
      "learning_rate": 7.808292384569838e-05,
      "loss": 0.166,
      "step": 9086
    },
    {
      "epoch": 4.276235294117647,
      "grad_norm": 0.5212184158299212,
      "learning_rate": 7.804445757129525e-05,
      "loss": 0.2754,
      "step": 9087
    },
    {
      "epoch": 4.276705882352942,
      "grad_norm": 0.6121157896292811,
      "learning_rate": 7.800599744224834e-05,
      "loss": 0.3906,
      "step": 9088
    },
    {
      "epoch": 4.277176470588235,
      "grad_norm": 0.35894743972851884,
      "learning_rate": 7.796754346184238e-05,
      "loss": 0.3945,
      "step": 9089
    },
    {
      "epoch": 4.277647058823529,
      "grad_norm": 0.4615405832727202,
      "learning_rate": 7.792909563336157e-05,
      "loss": 0.1963,
      "step": 9090
    },
    {
      "epoch": 4.278117647058823,
      "grad_norm": 0.3910245634753097,
      "learning_rate": 7.789065396008945e-05,
      "loss": 0.2393,
      "step": 9091
    },
    {
      "epoch": 4.278588235294118,
      "grad_norm": 0.3259753034292089,
      "learning_rate": 7.785221844530918e-05,
      "loss": 0.0422,
      "step": 9092
    },
    {
      "epoch": 4.279058823529412,
      "grad_norm": 0.26029125111738644,
      "learning_rate": 7.78137890923034e-05,
      "loss": 0.1196,
      "step": 9093
    },
    {
      "epoch": 4.279529411764706,
      "grad_norm": 0.37398890321015194,
      "learning_rate": 7.777536590435404e-05,
      "loss": 0.2578,
      "step": 9094
    },
    {
      "epoch": 4.28,
      "grad_norm": 0.25660413473625465,
      "learning_rate": 7.773694888474267e-05,
      "loss": 0.1299,
      "step": 9095
    },
    {
      "epoch": 4.2804705882352945,
      "grad_norm": 0.49705262436096476,
      "learning_rate": 7.769853803675041e-05,
      "loss": 0.2949,
      "step": 9096
    },
    {
      "epoch": 4.280941176470588,
      "grad_norm": 0.4024465916644599,
      "learning_rate": 7.766013336365755e-05,
      "loss": 0.6406,
      "step": 9097
    },
    {
      "epoch": 4.281411764705882,
      "grad_norm": 0.5280352850369476,
      "learning_rate": 7.762173486874414e-05,
      "loss": 0.252,
      "step": 9098
    },
    {
      "epoch": 4.281882352941176,
      "grad_norm": 0.5850563510605804,
      "learning_rate": 7.758334255528963e-05,
      "loss": 0.3301,
      "step": 9099
    },
    {
      "epoch": 4.2823529411764705,
      "grad_norm": 0.5424320111845631,
      "learning_rate": 7.754495642657282e-05,
      "loss": 0.4688,
      "step": 9100
    },
    {
      "epoch": 4.282823529411765,
      "grad_norm": 0.21664258599765937,
      "learning_rate": 7.750657648587213e-05,
      "loss": 0.053,
      "step": 9101
    },
    {
      "epoch": 4.283294117647059,
      "grad_norm": 0.4124524887787863,
      "learning_rate": 7.746820273646542e-05,
      "loss": 0.2793,
      "step": 9102
    },
    {
      "epoch": 4.283764705882353,
      "grad_norm": 0.152106108228922,
      "learning_rate": 7.742983518162993e-05,
      "loss": 0.0527,
      "step": 9103
    },
    {
      "epoch": 4.284235294117647,
      "grad_norm": 0.35638387951777756,
      "learning_rate": 7.73914738246425e-05,
      "loss": 0.2812,
      "step": 9104
    },
    {
      "epoch": 4.284705882352942,
      "grad_norm": 0.4477433846633845,
      "learning_rate": 7.735311866877933e-05,
      "loss": 0.498,
      "step": 9105
    },
    {
      "epoch": 4.285176470588235,
      "grad_norm": 0.3758362766827656,
      "learning_rate": 7.731476971731612e-05,
      "loss": 0.3867,
      "step": 9106
    },
    {
      "epoch": 4.285647058823529,
      "grad_norm": 0.3332353571672327,
      "learning_rate": 7.727642697352817e-05,
      "loss": 0.3496,
      "step": 9107
    },
    {
      "epoch": 4.286117647058823,
      "grad_norm": 0.3860512392563833,
      "learning_rate": 7.723809044069e-05,
      "loss": 0.4062,
      "step": 9108
    },
    {
      "epoch": 4.286588235294118,
      "grad_norm": 0.4563046688915244,
      "learning_rate": 7.719976012207579e-05,
      "loss": 0.3418,
      "step": 9109
    },
    {
      "epoch": 4.287058823529412,
      "grad_norm": 0.45042717311393404,
      "learning_rate": 7.716143602095924e-05,
      "loss": 0.1807,
      "step": 9110
    },
    {
      "epoch": 4.287529411764706,
      "grad_norm": 0.33878296255452794,
      "learning_rate": 7.712311814061323e-05,
      "loss": 0.291,
      "step": 9111
    },
    {
      "epoch": 4.288,
      "grad_norm": 0.358654723814292,
      "learning_rate": 7.708480648431043e-05,
      "loss": 0.2617,
      "step": 9112
    },
    {
      "epoch": 4.2884705882352945,
      "grad_norm": 0.39292202397732245,
      "learning_rate": 7.704650105532284e-05,
      "loss": 0.3496,
      "step": 9113
    },
    {
      "epoch": 4.288941176470589,
      "grad_norm": 0.3438802117115795,
      "learning_rate": 7.700820185692184e-05,
      "loss": 0.3105,
      "step": 9114
    },
    {
      "epoch": 4.289411764705882,
      "grad_norm": 0.3263131736732308,
      "learning_rate": 7.69699088923784e-05,
      "loss": 0.2002,
      "step": 9115
    },
    {
      "epoch": 4.289882352941176,
      "grad_norm": 0.545428331080901,
      "learning_rate": 7.693162216496303e-05,
      "loss": 0.2676,
      "step": 9116
    },
    {
      "epoch": 4.2903529411764705,
      "grad_norm": 0.5319333285698497,
      "learning_rate": 7.689334167794547e-05,
      "loss": 0.459,
      "step": 9117
    },
    {
      "epoch": 4.290823529411765,
      "grad_norm": 0.3106926968327515,
      "learning_rate": 7.685506743459511e-05,
      "loss": 0.1963,
      "step": 9118
    },
    {
      "epoch": 4.291294117647059,
      "grad_norm": 0.14930926983457057,
      "learning_rate": 7.681679943818082e-05,
      "loss": 0.105,
      "step": 9119
    },
    {
      "epoch": 4.291764705882353,
      "grad_norm": 0.7120558979892558,
      "learning_rate": 7.677853769197078e-05,
      "loss": 0.793,
      "step": 9120
    },
    {
      "epoch": 4.292235294117647,
      "grad_norm": 0.3611062077855006,
      "learning_rate": 7.674028219923275e-05,
      "loss": 0.4297,
      "step": 9121
    },
    {
      "epoch": 4.292705882352941,
      "grad_norm": 0.22764738144762814,
      "learning_rate": 7.670203296323403e-05,
      "loss": 0.0693,
      "step": 9122
    },
    {
      "epoch": 4.293176470588235,
      "grad_norm": 0.36878200705203606,
      "learning_rate": 7.666378998724116e-05,
      "loss": 0.2295,
      "step": 9123
    },
    {
      "epoch": 4.293647058823529,
      "grad_norm": 0.7056959532140495,
      "learning_rate": 7.662555327452031e-05,
      "loss": 1.0938,
      "step": 9124
    },
    {
      "epoch": 4.294117647058823,
      "grad_norm": 0.21472482404122814,
      "learning_rate": 7.65873228283372e-05,
      "loss": 0.1602,
      "step": 9125
    },
    {
      "epoch": 4.294588235294118,
      "grad_norm": 0.7785763386380876,
      "learning_rate": 7.654909865195675e-05,
      "loss": 0.5078,
      "step": 9126
    },
    {
      "epoch": 4.295058823529412,
      "grad_norm": 0.4340097819152313,
      "learning_rate": 7.65108807486436e-05,
      "loss": 0.4395,
      "step": 9127
    },
    {
      "epoch": 4.295529411764706,
      "grad_norm": 0.27489844888685433,
      "learning_rate": 7.647266912166162e-05,
      "loss": 0.0674,
      "step": 9128
    },
    {
      "epoch": 4.296,
      "grad_norm": 0.4281437028939516,
      "learning_rate": 7.643446377427438e-05,
      "loss": 0.4199,
      "step": 9129
    },
    {
      "epoch": 4.2964705882352945,
      "grad_norm": 0.4170883142448119,
      "learning_rate": 7.639626470974474e-05,
      "loss": 0.4961,
      "step": 9130
    },
    {
      "epoch": 4.296941176470588,
      "grad_norm": 0.32231974817634224,
      "learning_rate": 7.635807193133512e-05,
      "loss": 0.0952,
      "step": 9131
    },
    {
      "epoch": 4.297411764705882,
      "grad_norm": 0.24713204929161062,
      "learning_rate": 7.631988544230745e-05,
      "loss": 0.0503,
      "step": 9132
    },
    {
      "epoch": 4.297882352941176,
      "grad_norm": 0.21573282909378005,
      "learning_rate": 7.628170524592288e-05,
      "loss": 0.0388,
      "step": 9133
    },
    {
      "epoch": 4.2983529411764705,
      "grad_norm": 1.326128481803506,
      "learning_rate": 7.624353134544229e-05,
      "loss": 0.8125,
      "step": 9134
    },
    {
      "epoch": 4.298823529411765,
      "grad_norm": 0.3495604788866552,
      "learning_rate": 7.620536374412594e-05,
      "loss": 0.2988,
      "step": 9135
    },
    {
      "epoch": 4.299294117647059,
      "grad_norm": 0.36525655698011916,
      "learning_rate": 7.616720244523342e-05,
      "loss": 0.3652,
      "step": 9136
    },
    {
      "epoch": 4.299764705882353,
      "grad_norm": 0.44639783423965007,
      "learning_rate": 7.612904745202399e-05,
      "loss": 0.4219,
      "step": 9137
    },
    {
      "epoch": 4.300235294117647,
      "grad_norm": 0.4966173049408877,
      "learning_rate": 7.609089876775628e-05,
      "loss": 0.3926,
      "step": 9138
    },
    {
      "epoch": 4.300705882352942,
      "grad_norm": 0.24240145900049437,
      "learning_rate": 7.605275639568828e-05,
      "loss": 0.1855,
      "step": 9139
    },
    {
      "epoch": 4.301176470588235,
      "grad_norm": 0.3361180695884243,
      "learning_rate": 7.60146203390776e-05,
      "loss": 0.1758,
      "step": 9140
    },
    {
      "epoch": 4.301647058823529,
      "grad_norm": 0.21491733890987758,
      "learning_rate": 7.59764906011813e-05,
      "loss": 0.1758,
      "step": 9141
    },
    {
      "epoch": 4.302117647058823,
      "grad_norm": 0.7577593945433859,
      "learning_rate": 7.593836718525572e-05,
      "loss": 0.5312,
      "step": 9142
    },
    {
      "epoch": 4.302588235294118,
      "grad_norm": 0.3517670422463502,
      "learning_rate": 7.590025009455684e-05,
      "loss": 0.2285,
      "step": 9143
    },
    {
      "epoch": 4.303058823529412,
      "grad_norm": 0.4343735881631535,
      "learning_rate": 7.586213933234013e-05,
      "loss": 0.3418,
      "step": 9144
    },
    {
      "epoch": 4.303529411764706,
      "grad_norm": 0.32333137049290245,
      "learning_rate": 7.58240349018603e-05,
      "loss": 0.2773,
      "step": 9145
    },
    {
      "epoch": 4.304,
      "grad_norm": 0.41704058073655587,
      "learning_rate": 7.578593680637173e-05,
      "loss": 0.3613,
      "step": 9146
    },
    {
      "epoch": 4.3044705882352945,
      "grad_norm": 0.31461273762201214,
      "learning_rate": 7.574784504912823e-05,
      "loss": 0.1787,
      "step": 9147
    },
    {
      "epoch": 4.304941176470588,
      "grad_norm": 0.4565407324246066,
      "learning_rate": 7.57097596333829e-05,
      "loss": 0.3516,
      "step": 9148
    },
    {
      "epoch": 4.305411764705882,
      "grad_norm": 0.31923036471931365,
      "learning_rate": 7.567168056238851e-05,
      "loss": 0.2773,
      "step": 9149
    },
    {
      "epoch": 4.305882352941176,
      "grad_norm": 0.379886121457824,
      "learning_rate": 7.563360783939722e-05,
      "loss": 0.2002,
      "step": 9150
    },
    {
      "epoch": 4.3063529411764705,
      "grad_norm": 0.41395462967483543,
      "learning_rate": 7.559554146766054e-05,
      "loss": 0.3887,
      "step": 9151
    },
    {
      "epoch": 4.306823529411765,
      "grad_norm": 0.4594383228840234,
      "learning_rate": 7.555748145042957e-05,
      "loss": 0.4316,
      "step": 9152
    },
    {
      "epoch": 4.307294117647059,
      "grad_norm": 0.24627270228867473,
      "learning_rate": 7.551942779095486e-05,
      "loss": 0.0525,
      "step": 9153
    },
    {
      "epoch": 4.307764705882353,
      "grad_norm": 0.36946575068995924,
      "learning_rate": 7.54813804924863e-05,
      "loss": 0.5977,
      "step": 9154
    },
    {
      "epoch": 4.308235294117647,
      "grad_norm": 0.5334139888899598,
      "learning_rate": 7.544333955827342e-05,
      "loss": 0.3906,
      "step": 9155
    },
    {
      "epoch": 4.308705882352941,
      "grad_norm": 0.3282430360203823,
      "learning_rate": 7.540530499156496e-05,
      "loss": 0.1416,
      "step": 9156
    },
    {
      "epoch": 4.309176470588235,
      "grad_norm": 0.3887524763299952,
      "learning_rate": 7.536727679560937e-05,
      "loss": 0.3281,
      "step": 9157
    },
    {
      "epoch": 4.309647058823529,
      "grad_norm": 0.30855508161121786,
      "learning_rate": 7.532925497365444e-05,
      "loss": 0.209,
      "step": 9158
    },
    {
      "epoch": 4.310117647058823,
      "grad_norm": 0.39262564665102573,
      "learning_rate": 7.529123952894734e-05,
      "loss": 0.3301,
      "step": 9159
    },
    {
      "epoch": 4.310588235294118,
      "grad_norm": 0.26529584433548065,
      "learning_rate": 7.525323046473485e-05,
      "loss": 0.1108,
      "step": 9160
    },
    {
      "epoch": 4.311058823529412,
      "grad_norm": 0.7268647131179242,
      "learning_rate": 7.521522778426315e-05,
      "loss": 0.4004,
      "step": 9161
    },
    {
      "epoch": 4.311529411764706,
      "grad_norm": 0.4821204223452355,
      "learning_rate": 7.517723149077776e-05,
      "loss": 0.418,
      "step": 9162
    },
    {
      "epoch": 4.312,
      "grad_norm": 0.4939311587278696,
      "learning_rate": 7.513924158752383e-05,
      "loss": 0.4746,
      "step": 9163
    },
    {
      "epoch": 4.3124705882352945,
      "grad_norm": 0.5276336086412754,
      "learning_rate": 7.510125807774589e-05,
      "loss": 0.5859,
      "step": 9164
    },
    {
      "epoch": 4.312941176470588,
      "grad_norm": 0.37208838898751906,
      "learning_rate": 7.506328096468782e-05,
      "loss": 0.3848,
      "step": 9165
    },
    {
      "epoch": 4.313411764705882,
      "grad_norm": 0.3348985299022811,
      "learning_rate": 7.502531025159313e-05,
      "loss": 0.0786,
      "step": 9166
    },
    {
      "epoch": 4.313882352941176,
      "grad_norm": 0.36950153724178975,
      "learning_rate": 7.498734594170474e-05,
      "loss": 0.2432,
      "step": 9167
    },
    {
      "epoch": 4.3143529411764705,
      "grad_norm": 0.503849021197552,
      "learning_rate": 7.49493880382649e-05,
      "loss": 0.2891,
      "step": 9168
    },
    {
      "epoch": 4.314823529411765,
      "grad_norm": 0.25840150879084156,
      "learning_rate": 7.491143654451544e-05,
      "loss": 0.1416,
      "step": 9169
    },
    {
      "epoch": 4.315294117647059,
      "grad_norm": 0.40898900039981684,
      "learning_rate": 7.487349146369766e-05,
      "loss": 0.5273,
      "step": 9170
    },
    {
      "epoch": 4.315764705882353,
      "grad_norm": 0.3851186796407412,
      "learning_rate": 7.483555279905215e-05,
      "loss": 0.2539,
      "step": 9171
    },
    {
      "epoch": 4.316235294117647,
      "grad_norm": 0.29126353017991613,
      "learning_rate": 7.479762055381909e-05,
      "loss": 0.1562,
      "step": 9172
    },
    {
      "epoch": 4.316705882352942,
      "grad_norm": 0.3641785120573993,
      "learning_rate": 7.475969473123819e-05,
      "loss": 0.4121,
      "step": 9173
    },
    {
      "epoch": 4.317176470588235,
      "grad_norm": 0.36406003114477803,
      "learning_rate": 7.472177533454833e-05,
      "loss": 0.3125,
      "step": 9174
    },
    {
      "epoch": 4.317647058823529,
      "grad_norm": 0.30825868942971607,
      "learning_rate": 7.46838623669881e-05,
      "loss": 0.1924,
      "step": 9175
    },
    {
      "epoch": 4.318117647058823,
      "grad_norm": 0.44483260269616665,
      "learning_rate": 7.464595583179544e-05,
      "loss": 0.2988,
      "step": 9176
    },
    {
      "epoch": 4.318588235294118,
      "grad_norm": 0.30748668946788266,
      "learning_rate": 7.460805573220779e-05,
      "loss": 0.2324,
      "step": 9177
    },
    {
      "epoch": 4.319058823529412,
      "grad_norm": 0.6511464338286199,
      "learning_rate": 7.4570162071462e-05,
      "loss": 0.3164,
      "step": 9178
    },
    {
      "epoch": 4.319529411764706,
      "grad_norm": 0.2666139750124164,
      "learning_rate": 7.453227485279432e-05,
      "loss": 0.1113,
      "step": 9179
    },
    {
      "epoch": 4.32,
      "grad_norm": 0.31679653011038944,
      "learning_rate": 7.449439407944052e-05,
      "loss": 0.1108,
      "step": 9180
    },
    {
      "epoch": 4.3204705882352945,
      "grad_norm": 0.21717805855555286,
      "learning_rate": 7.445651975463588e-05,
      "loss": 0.1875,
      "step": 9181
    },
    {
      "epoch": 4.320941176470588,
      "grad_norm": 0.2715614731716132,
      "learning_rate": 7.441865188161494e-05,
      "loss": 0.1875,
      "step": 9182
    },
    {
      "epoch": 4.321411764705882,
      "grad_norm": 0.23286023332648634,
      "learning_rate": 7.43807904636119e-05,
      "loss": 0.0679,
      "step": 9183
    },
    {
      "epoch": 4.321882352941176,
      "grad_norm": 0.6682582728107496,
      "learning_rate": 7.434293550386021e-05,
      "loss": 0.7148,
      "step": 9184
    },
    {
      "epoch": 4.3223529411764705,
      "grad_norm": 0.3689493456694277,
      "learning_rate": 7.430508700559292e-05,
      "loss": 0.3477,
      "step": 9185
    },
    {
      "epoch": 4.322823529411765,
      "grad_norm": 0.3725155738678531,
      "learning_rate": 7.426724497204254e-05,
      "loss": 0.2168,
      "step": 9186
    },
    {
      "epoch": 4.323294117647059,
      "grad_norm": 0.2552989354553129,
      "learning_rate": 7.422940940644085e-05,
      "loss": 0.0742,
      "step": 9187
    },
    {
      "epoch": 4.323764705882353,
      "grad_norm": 0.47973380625368345,
      "learning_rate": 7.419158031201924e-05,
      "loss": 0.4961,
      "step": 9188
    },
    {
      "epoch": 4.324235294117647,
      "grad_norm": 0.46246670713314036,
      "learning_rate": 7.415375769200859e-05,
      "loss": 0.3691,
      "step": 9189
    },
    {
      "epoch": 4.324705882352941,
      "grad_norm": 0.1126183988303518,
      "learning_rate": 7.411594154963897e-05,
      "loss": 0.0222,
      "step": 9190
    },
    {
      "epoch": 4.325176470588235,
      "grad_norm": 0.5205459753717047,
      "learning_rate": 7.407813188814017e-05,
      "loss": 0.3203,
      "step": 9191
    },
    {
      "epoch": 4.325647058823529,
      "grad_norm": 0.5410340771109393,
      "learning_rate": 7.404032871074135e-05,
      "loss": 0.3691,
      "step": 9192
    },
    {
      "epoch": 4.326117647058823,
      "grad_norm": 0.525361931338769,
      "learning_rate": 7.400253202067101e-05,
      "loss": 0.5312,
      "step": 9193
    },
    {
      "epoch": 4.326588235294118,
      "grad_norm": 0.4454383291975423,
      "learning_rate": 7.396474182115717e-05,
      "loss": 0.4902,
      "step": 9194
    },
    {
      "epoch": 4.327058823529412,
      "grad_norm": 0.2499123325304417,
      "learning_rate": 7.392695811542739e-05,
      "loss": 0.063,
      "step": 9195
    },
    {
      "epoch": 4.327529411764706,
      "grad_norm": 0.16502435537781762,
      "learning_rate": 7.388918090670847e-05,
      "loss": 0.0197,
      "step": 9196
    },
    {
      "epoch": 4.328,
      "grad_norm": 0.3064182631235547,
      "learning_rate": 7.385141019822683e-05,
      "loss": 0.2402,
      "step": 9197
    },
    {
      "epoch": 4.3284705882352945,
      "grad_norm": 0.4155459706880878,
      "learning_rate": 7.381364599320831e-05,
      "loss": 0.4648,
      "step": 9198
    },
    {
      "epoch": 4.328941176470588,
      "grad_norm": 0.6423592750746244,
      "learning_rate": 7.377588829487806e-05,
      "loss": 0.5469,
      "step": 9199
    },
    {
      "epoch": 4.329411764705882,
      "grad_norm": 0.33761517040482236,
      "learning_rate": 7.373813710646083e-05,
      "loss": 0.2051,
      "step": 9200
    },
    {
      "epoch": 4.329882352941176,
      "grad_norm": 0.3422685501597231,
      "learning_rate": 7.370039243118082e-05,
      "loss": 0.165,
      "step": 9201
    },
    {
      "epoch": 4.3303529411764705,
      "grad_norm": 0.47890547411550294,
      "learning_rate": 7.366265427226148e-05,
      "loss": 0.3457,
      "step": 9202
    },
    {
      "epoch": 4.330823529411765,
      "grad_norm": 0.5363722926836639,
      "learning_rate": 7.36249226329259e-05,
      "loss": 0.6719,
      "step": 9203
    },
    {
      "epoch": 4.331294117647059,
      "grad_norm": 0.21236113729893893,
      "learning_rate": 7.358719751639659e-05,
      "loss": 0.0569,
      "step": 9204
    },
    {
      "epoch": 4.331764705882353,
      "grad_norm": 0.6217968349596771,
      "learning_rate": 7.354947892589536e-05,
      "loss": 0.373,
      "step": 9205
    },
    {
      "epoch": 4.332235294117647,
      "grad_norm": 0.3866587104667717,
      "learning_rate": 7.351176686464368e-05,
      "loss": 0.2734,
      "step": 9206
    },
    {
      "epoch": 4.332705882352942,
      "grad_norm": 0.4887224294448381,
      "learning_rate": 7.347406133586223e-05,
      "loss": 0.4141,
      "step": 9207
    },
    {
      "epoch": 4.333176470588235,
      "grad_norm": 0.4043513145164196,
      "learning_rate": 7.34363623427713e-05,
      "loss": 0.3281,
      "step": 9208
    },
    {
      "epoch": 4.333647058823529,
      "grad_norm": 0.3770255815789027,
      "learning_rate": 7.339866988859061e-05,
      "loss": 0.127,
      "step": 9209
    },
    {
      "epoch": 4.334117647058823,
      "grad_norm": 0.358331731178642,
      "learning_rate": 7.33609839765392e-05,
      "loss": 0.1572,
      "step": 9210
    },
    {
      "epoch": 4.334588235294118,
      "grad_norm": 0.381024536330139,
      "learning_rate": 7.332330460983565e-05,
      "loss": 0.3691,
      "step": 9211
    },
    {
      "epoch": 4.335058823529412,
      "grad_norm": 0.2884333640360362,
      "learning_rate": 7.328563179169806e-05,
      "loss": 0.2168,
      "step": 9212
    },
    {
      "epoch": 4.335529411764706,
      "grad_norm": 0.39465208688899867,
      "learning_rate": 7.324796552534373e-05,
      "loss": 0.1865,
      "step": 9213
    },
    {
      "epoch": 4.336,
      "grad_norm": 0.6948428477801513,
      "learning_rate": 7.321030581398961e-05,
      "loss": 0.7461,
      "step": 9214
    },
    {
      "epoch": 4.3364705882352945,
      "grad_norm": 0.04533624549537172,
      "learning_rate": 7.317265266085209e-05,
      "loss": 0.0106,
      "step": 9215
    },
    {
      "epoch": 4.336941176470588,
      "grad_norm": 0.37093005451456157,
      "learning_rate": 7.31350060691468e-05,
      "loss": 0.1602,
      "step": 9216
    },
    {
      "epoch": 4.337411764705882,
      "grad_norm": 0.31060822812333866,
      "learning_rate": 7.309736604208902e-05,
      "loss": 0.126,
      "step": 9217
    },
    {
      "epoch": 4.337882352941176,
      "grad_norm": 0.3690640696132945,
      "learning_rate": 7.305973258289343e-05,
      "loss": 0.1602,
      "step": 9218
    },
    {
      "epoch": 4.3383529411764705,
      "grad_norm": 0.4839717377206807,
      "learning_rate": 7.302210569477397e-05,
      "loss": 0.4512,
      "step": 9219
    },
    {
      "epoch": 4.338823529411765,
      "grad_norm": 0.4626366476319911,
      "learning_rate": 7.298448538094436e-05,
      "loss": 0.3574,
      "step": 9220
    },
    {
      "epoch": 4.339294117647059,
      "grad_norm": 0.45294140424712037,
      "learning_rate": 7.294687164461741e-05,
      "loss": 0.3184,
      "step": 9221
    },
    {
      "epoch": 4.339764705882353,
      "grad_norm": 0.33981998534251046,
      "learning_rate": 7.290926448900557e-05,
      "loss": 0.2109,
      "step": 9222
    },
    {
      "epoch": 4.340235294117647,
      "grad_norm": 0.2582546534845553,
      "learning_rate": 7.287166391732071e-05,
      "loss": 0.1021,
      "step": 9223
    },
    {
      "epoch": 4.340705882352941,
      "grad_norm": 0.41447177641840194,
      "learning_rate": 7.283406993277401e-05,
      "loss": 0.5312,
      "step": 9224
    },
    {
      "epoch": 4.341176470588235,
      "grad_norm": 0.5552290047551043,
      "learning_rate": 7.279648253857626e-05,
      "loss": 0.7227,
      "step": 9225
    },
    {
      "epoch": 4.341647058823529,
      "grad_norm": 0.44250229984547806,
      "learning_rate": 7.275890173793761e-05,
      "loss": 0.3906,
      "step": 9226
    },
    {
      "epoch": 4.342117647058823,
      "grad_norm": 0.35740281540932356,
      "learning_rate": 7.272132753406757e-05,
      "loss": 0.1934,
      "step": 9227
    },
    {
      "epoch": 4.342588235294118,
      "grad_norm": 0.30886042084776066,
      "learning_rate": 7.26837599301752e-05,
      "loss": 0.1689,
      "step": 9228
    },
    {
      "epoch": 4.343058823529412,
      "grad_norm": 0.4992702701918813,
      "learning_rate": 7.264619892946904e-05,
      "loss": 0.2812,
      "step": 9229
    },
    {
      "epoch": 4.343529411764706,
      "grad_norm": 0.2348829373835545,
      "learning_rate": 7.260864453515683e-05,
      "loss": 0.0466,
      "step": 9230
    },
    {
      "epoch": 4.344,
      "grad_norm": 0.17034305437097028,
      "learning_rate": 7.2571096750446e-05,
      "loss": 0.0894,
      "step": 9231
    },
    {
      "epoch": 4.3444705882352945,
      "grad_norm": 0.17256114741152642,
      "learning_rate": 7.253355557854334e-05,
      "loss": 0.1465,
      "step": 9232
    },
    {
      "epoch": 4.344941176470588,
      "grad_norm": 0.30590539197641853,
      "learning_rate": 7.249602102265496e-05,
      "loss": 0.3828,
      "step": 9233
    },
    {
      "epoch": 4.345411764705882,
      "grad_norm": 0.356808690105918,
      "learning_rate": 7.245849308598658e-05,
      "loss": 0.249,
      "step": 9234
    },
    {
      "epoch": 4.345882352941176,
      "grad_norm": 0.5928234946234663,
      "learning_rate": 7.242097177174318e-05,
      "loss": 0.7266,
      "step": 9235
    },
    {
      "epoch": 4.3463529411764705,
      "grad_norm": 0.4554268016271465,
      "learning_rate": 7.238345708312931e-05,
      "loss": 0.1299,
      "step": 9236
    },
    {
      "epoch": 4.346823529411765,
      "grad_norm": 0.5822919427901334,
      "learning_rate": 7.234594902334897e-05,
      "loss": 0.707,
      "step": 9237
    },
    {
      "epoch": 4.347294117647059,
      "grad_norm": 0.3964178277029443,
      "learning_rate": 7.230844759560544e-05,
      "loss": 0.1826,
      "step": 9238
    },
    {
      "epoch": 4.347764705882353,
      "grad_norm": 0.38463181277207703,
      "learning_rate": 7.227095280310152e-05,
      "loss": 0.3359,
      "step": 9239
    },
    {
      "epoch": 4.348235294117647,
      "grad_norm": 0.35754674146397736,
      "learning_rate": 7.223346464903958e-05,
      "loss": 0.291,
      "step": 9240
    },
    {
      "epoch": 4.348705882352942,
      "grad_norm": 0.39123962763009823,
      "learning_rate": 7.219598313662113e-05,
      "loss": 0.1484,
      "step": 9241
    },
    {
      "epoch": 4.349176470588235,
      "grad_norm": 0.3826766369684668,
      "learning_rate": 7.215850826904734e-05,
      "loss": 0.3496,
      "step": 9242
    },
    {
      "epoch": 4.349647058823529,
      "grad_norm": 0.39232120223440614,
      "learning_rate": 7.212104004951883e-05,
      "loss": 0.3535,
      "step": 9243
    },
    {
      "epoch": 4.350117647058823,
      "grad_norm": 0.5054599809110758,
      "learning_rate": 7.208357848123543e-05,
      "loss": 0.6328,
      "step": 9244
    },
    {
      "epoch": 4.350588235294118,
      "grad_norm": 0.3426915216801162,
      "learning_rate": 7.204612356739658e-05,
      "loss": 0.3008,
      "step": 9245
    },
    {
      "epoch": 4.351058823529412,
      "grad_norm": 0.3232823245456781,
      "learning_rate": 7.200867531120124e-05,
      "loss": 0.3281,
      "step": 9246
    },
    {
      "epoch": 4.351529411764706,
      "grad_norm": 0.48456650180695504,
      "learning_rate": 7.197123371584749e-05,
      "loss": 0.5508,
      "step": 9247
    },
    {
      "epoch": 4.352,
      "grad_norm": 0.3410733732231766,
      "learning_rate": 7.193379878453312e-05,
      "loss": 0.2217,
      "step": 9248
    },
    {
      "epoch": 4.352470588235294,
      "grad_norm": 0.4152032965078636,
      "learning_rate": 7.189637052045529e-05,
      "loss": 0.2539,
      "step": 9249
    },
    {
      "epoch": 4.352941176470588,
      "grad_norm": 0.27065830115269107,
      "learning_rate": 7.185894892681048e-05,
      "loss": 0.0698,
      "step": 9250
    },
    {
      "epoch": 4.353411764705882,
      "grad_norm": 0.2968143753343362,
      "learning_rate": 7.182153400679468e-05,
      "loss": 0.0972,
      "step": 9251
    },
    {
      "epoch": 4.353882352941176,
      "grad_norm": 0.4703044249787477,
      "learning_rate": 7.17841257636034e-05,
      "loss": 0.4102,
      "step": 9252
    },
    {
      "epoch": 4.3543529411764705,
      "grad_norm": 0.14001704946766652,
      "learning_rate": 7.17467242004314e-05,
      "loss": 0.05,
      "step": 9253
    },
    {
      "epoch": 4.354823529411765,
      "grad_norm": 0.5466145576253112,
      "learning_rate": 7.170932932047296e-05,
      "loss": 0.4453,
      "step": 9254
    },
    {
      "epoch": 4.355294117647059,
      "grad_norm": 0.5104064303983924,
      "learning_rate": 7.167194112692186e-05,
      "loss": 0.2988,
      "step": 9255
    },
    {
      "epoch": 4.355764705882353,
      "grad_norm": 0.4454574095761511,
      "learning_rate": 7.163455962297116e-05,
      "loss": 0.4922,
      "step": 9256
    },
    {
      "epoch": 4.356235294117647,
      "grad_norm": 0.2667659090226364,
      "learning_rate": 7.159718481181346e-05,
      "loss": 0.1045,
      "step": 9257
    },
    {
      "epoch": 4.356705882352941,
      "grad_norm": 0.5108383986133888,
      "learning_rate": 7.155981669664072e-05,
      "loss": 0.5352,
      "step": 9258
    },
    {
      "epoch": 4.357176470588235,
      "grad_norm": 0.4332563471076132,
      "learning_rate": 7.152245528064437e-05,
      "loss": 0.2773,
      "step": 9259
    },
    {
      "epoch": 4.357647058823529,
      "grad_norm": 0.40325024625603434,
      "learning_rate": 7.148510056701532e-05,
      "loss": 0.4434,
      "step": 9260
    },
    {
      "epoch": 4.358117647058823,
      "grad_norm": 0.45426887996913595,
      "learning_rate": 7.144775255894375e-05,
      "loss": 0.1992,
      "step": 9261
    },
    {
      "epoch": 4.358588235294118,
      "grad_norm": 0.49690598205573916,
      "learning_rate": 7.14104112596194e-05,
      "loss": 0.7031,
      "step": 9262
    },
    {
      "epoch": 4.359058823529412,
      "grad_norm": 0.37453866622161514,
      "learning_rate": 7.137307667223141e-05,
      "loss": 0.3516,
      "step": 9263
    },
    {
      "epoch": 4.359529411764706,
      "grad_norm": 0.41264637178741276,
      "learning_rate": 7.133574879996833e-05,
      "loss": 0.3379,
      "step": 9264
    },
    {
      "epoch": 4.36,
      "grad_norm": 0.28926354135730126,
      "learning_rate": 7.129842764601821e-05,
      "loss": 0.4258,
      "step": 9265
    },
    {
      "epoch": 4.3604705882352945,
      "grad_norm": 0.2810221177588965,
      "learning_rate": 7.126111321356834e-05,
      "loss": 0.1445,
      "step": 9266
    },
    {
      "epoch": 4.360941176470588,
      "grad_norm": 0.7137080748994474,
      "learning_rate": 7.122380550580562e-05,
      "loss": 0.4668,
      "step": 9267
    },
    {
      "epoch": 4.361411764705882,
      "grad_norm": 0.6576678067862288,
      "learning_rate": 7.118650452591634e-05,
      "loss": 0.4336,
      "step": 9268
    },
    {
      "epoch": 4.361882352941176,
      "grad_norm": 0.49352800548291254,
      "learning_rate": 7.114921027708613e-05,
      "loss": 0.5547,
      "step": 9269
    },
    {
      "epoch": 4.3623529411764705,
      "grad_norm": 0.5806750913980118,
      "learning_rate": 7.111192276250011e-05,
      "loss": 0.3223,
      "step": 9270
    },
    {
      "epoch": 4.362823529411765,
      "grad_norm": 0.3303548513051497,
      "learning_rate": 7.107464198534287e-05,
      "loss": 0.2295,
      "step": 9271
    },
    {
      "epoch": 4.363294117647059,
      "grad_norm": 0.5119694345597424,
      "learning_rate": 7.103736794879829e-05,
      "loss": 0.0845,
      "step": 9272
    },
    {
      "epoch": 4.363764705882353,
      "grad_norm": 0.3774850099876811,
      "learning_rate": 7.100010065604978e-05,
      "loss": 0.167,
      "step": 9273
    },
    {
      "epoch": 4.364235294117647,
      "grad_norm": 0.054978859250301014,
      "learning_rate": 7.096284011028024e-05,
      "loss": 0.0107,
      "step": 9274
    },
    {
      "epoch": 4.364705882352941,
      "grad_norm": 0.5666040478404804,
      "learning_rate": 7.092558631467177e-05,
      "loss": 0.7578,
      "step": 9275
    },
    {
      "epoch": 4.365176470588235,
      "grad_norm": 0.48216978164421054,
      "learning_rate": 7.088833927240608e-05,
      "loss": 0.5156,
      "step": 9276
    },
    {
      "epoch": 4.365647058823529,
      "grad_norm": 0.553017042641381,
      "learning_rate": 7.085109898666433e-05,
      "loss": 0.2812,
      "step": 9277
    },
    {
      "epoch": 4.366117647058823,
      "grad_norm": 0.5171965548090816,
      "learning_rate": 7.081386546062687e-05,
      "loss": 0.6602,
      "step": 9278
    },
    {
      "epoch": 4.366588235294118,
      "grad_norm": 0.16911113301065525,
      "learning_rate": 7.077663869747372e-05,
      "loss": 0.0403,
      "step": 9279
    },
    {
      "epoch": 4.367058823529412,
      "grad_norm": 0.2120519236740119,
      "learning_rate": 7.073941870038427e-05,
      "loss": 0.0444,
      "step": 9280
    },
    {
      "epoch": 4.367529411764706,
      "grad_norm": 0.3048381570250042,
      "learning_rate": 7.070220547253717e-05,
      "loss": 0.252,
      "step": 9281
    },
    {
      "epoch": 4.368,
      "grad_norm": 0.1939810590632759,
      "learning_rate": 7.066499901711068e-05,
      "loss": 0.0386,
      "step": 9282
    },
    {
      "epoch": 4.368470588235294,
      "grad_norm": 0.403293891511959,
      "learning_rate": 7.062779933728247e-05,
      "loss": 0.2852,
      "step": 9283
    },
    {
      "epoch": 4.368941176470588,
      "grad_norm": 0.4831561834694883,
      "learning_rate": 7.059060643622944e-05,
      "loss": 0.543,
      "step": 9284
    },
    {
      "epoch": 4.369411764705882,
      "grad_norm": 0.36447229298454564,
      "learning_rate": 7.05534203171282e-05,
      "loss": 0.5117,
      "step": 9285
    },
    {
      "epoch": 4.369882352941176,
      "grad_norm": 0.3837628161407342,
      "learning_rate": 7.051624098315449e-05,
      "loss": 0.1523,
      "step": 9286
    },
    {
      "epoch": 4.3703529411764706,
      "grad_norm": 0.7765534404360468,
      "learning_rate": 7.047906843748365e-05,
      "loss": 0.3125,
      "step": 9287
    },
    {
      "epoch": 4.370823529411765,
      "grad_norm": 0.29150968157044665,
      "learning_rate": 7.044190268329049e-05,
      "loss": 0.1216,
      "step": 9288
    },
    {
      "epoch": 4.371294117647059,
      "grad_norm": 0.20752859084472391,
      "learning_rate": 7.0404743723749e-05,
      "loss": 0.0403,
      "step": 9289
    },
    {
      "epoch": 4.371764705882353,
      "grad_norm": 0.39891766895549396,
      "learning_rate": 7.036759156203281e-05,
      "loss": 0.6445,
      "step": 9290
    },
    {
      "epoch": 4.372235294117647,
      "grad_norm": 0.2481628327944302,
      "learning_rate": 7.033044620131495e-05,
      "loss": 0.2168,
      "step": 9291
    },
    {
      "epoch": 4.372705882352941,
      "grad_norm": 0.20590774202286521,
      "learning_rate": 7.02933076447677e-05,
      "loss": 0.0791,
      "step": 9292
    },
    {
      "epoch": 4.373176470588235,
      "grad_norm": 0.358610065780248,
      "learning_rate": 7.025617589556297e-05,
      "loss": 0.4668,
      "step": 9293
    },
    {
      "epoch": 4.373647058823529,
      "grad_norm": 0.44664320596023915,
      "learning_rate": 7.021905095687198e-05,
      "loss": 0.4609,
      "step": 9294
    },
    {
      "epoch": 4.374117647058823,
      "grad_norm": 0.3474162282747327,
      "learning_rate": 7.018193283186531e-05,
      "loss": 0.1475,
      "step": 9295
    },
    {
      "epoch": 4.374588235294118,
      "grad_norm": 0.46753811125813205,
      "learning_rate": 7.01448215237131e-05,
      "loss": 0.334,
      "step": 9296
    },
    {
      "epoch": 4.375058823529412,
      "grad_norm": 0.2728355033434803,
      "learning_rate": 7.010771703558488e-05,
      "loss": 0.104,
      "step": 9297
    },
    {
      "epoch": 4.375529411764706,
      "grad_norm": 0.2448389367013109,
      "learning_rate": 7.007061937064941e-05,
      "loss": 0.2773,
      "step": 9298
    },
    {
      "epoch": 4.376,
      "grad_norm": 0.58676154862654,
      "learning_rate": 7.00335285320751e-05,
      "loss": 0.5781,
      "step": 9299
    },
    {
      "epoch": 4.376470588235295,
      "grad_norm": 0.3612792941383435,
      "learning_rate": 6.999644452302975e-05,
      "loss": 0.105,
      "step": 9300
    },
    {
      "epoch": 4.376941176470588,
      "grad_norm": 0.24289354392380594,
      "learning_rate": 6.995936734668042e-05,
      "loss": 0.0669,
      "step": 9301
    },
    {
      "epoch": 4.377411764705882,
      "grad_norm": 0.13992618795556588,
      "learning_rate": 6.992229700619369e-05,
      "loss": 0.0332,
      "step": 9302
    },
    {
      "epoch": 4.377882352941176,
      "grad_norm": 0.4086480686965801,
      "learning_rate": 6.988523350473563e-05,
      "loss": 0.2236,
      "step": 9303
    },
    {
      "epoch": 4.378352941176471,
      "grad_norm": 0.2829765491739512,
      "learning_rate": 6.984817684547155e-05,
      "loss": 0.0488,
      "step": 9304
    },
    {
      "epoch": 4.378823529411765,
      "grad_norm": 0.3181599226523215,
      "learning_rate": 6.981112703156628e-05,
      "loss": 0.103,
      "step": 9305
    },
    {
      "epoch": 4.379294117647059,
      "grad_norm": 0.25355695269055983,
      "learning_rate": 6.977408406618418e-05,
      "loss": 0.1436,
      "step": 9306
    },
    {
      "epoch": 4.379764705882353,
      "grad_norm": 0.6031340710296735,
      "learning_rate": 6.973704795248871e-05,
      "loss": 0.3906,
      "step": 9307
    },
    {
      "epoch": 4.3802352941176474,
      "grad_norm": 0.6632081046060453,
      "learning_rate": 6.970001869364306e-05,
      "loss": 0.4414,
      "step": 9308
    },
    {
      "epoch": 4.380705882352941,
      "grad_norm": 0.24775036478730886,
      "learning_rate": 6.966299629280968e-05,
      "loss": 0.0581,
      "step": 9309
    },
    {
      "epoch": 4.381176470588235,
      "grad_norm": 0.5625209792997714,
      "learning_rate": 6.962598075315046e-05,
      "loss": 0.2734,
      "step": 9310
    },
    {
      "epoch": 4.381647058823529,
      "grad_norm": 0.3492517625769059,
      "learning_rate": 6.958897207782677e-05,
      "loss": 0.1201,
      "step": 9311
    },
    {
      "epoch": 4.3821176470588235,
      "grad_norm": 0.3617036001178197,
      "learning_rate": 6.955197026999925e-05,
      "loss": 0.2305,
      "step": 9312
    },
    {
      "epoch": 4.382588235294118,
      "grad_norm": 0.7008182879838394,
      "learning_rate": 6.951497533282809e-05,
      "loss": 0.5547,
      "step": 9313
    },
    {
      "epoch": 4.383058823529412,
      "grad_norm": 0.42302541944199085,
      "learning_rate": 6.947798726947277e-05,
      "loss": 0.1885,
      "step": 9314
    },
    {
      "epoch": 4.383529411764706,
      "grad_norm": 0.40393668967284657,
      "learning_rate": 6.94410060830923e-05,
      "loss": 0.3438,
      "step": 9315
    },
    {
      "epoch": 4.384,
      "grad_norm": 0.43564422534266684,
      "learning_rate": 6.940403177684509e-05,
      "loss": 0.4082,
      "step": 9316
    },
    {
      "epoch": 4.384470588235294,
      "grad_norm": 0.3459098780006858,
      "learning_rate": 6.936706435388882e-05,
      "loss": 0.1729,
      "step": 9317
    },
    {
      "epoch": 4.384941176470588,
      "grad_norm": 0.3053985544720599,
      "learning_rate": 6.93301038173808e-05,
      "loss": 0.1846,
      "step": 9318
    },
    {
      "epoch": 4.385411764705882,
      "grad_norm": 0.5634507904303442,
      "learning_rate": 6.929315017047762e-05,
      "loss": 0.4707,
      "step": 9319
    },
    {
      "epoch": 4.385882352941176,
      "grad_norm": 0.3607312284896959,
      "learning_rate": 6.925620341633522e-05,
      "loss": 0.2578,
      "step": 9320
    },
    {
      "epoch": 4.386352941176471,
      "grad_norm": 0.5240708682933163,
      "learning_rate": 6.921926355810911e-05,
      "loss": 0.4238,
      "step": 9321
    },
    {
      "epoch": 4.386823529411765,
      "grad_norm": 0.2680096577254863,
      "learning_rate": 6.918233059895418e-05,
      "loss": 0.0698,
      "step": 9322
    },
    {
      "epoch": 4.387294117647059,
      "grad_norm": 0.35151419605585943,
      "learning_rate": 6.914540454202457e-05,
      "loss": 0.249,
      "step": 9323
    },
    {
      "epoch": 4.387764705882353,
      "grad_norm": 0.4161595894962681,
      "learning_rate": 6.910848539047399e-05,
      "loss": 0.4238,
      "step": 9324
    },
    {
      "epoch": 4.3882352941176475,
      "grad_norm": 0.17635748449347335,
      "learning_rate": 6.907157314745558e-05,
      "loss": 0.0325,
      "step": 9325
    },
    {
      "epoch": 4.388705882352941,
      "grad_norm": 0.3506574049598607,
      "learning_rate": 6.903466781612173e-05,
      "loss": 0.1426,
      "step": 9326
    },
    {
      "epoch": 4.389176470588235,
      "grad_norm": 0.21895802738216166,
      "learning_rate": 6.899776939962438e-05,
      "loss": 0.0835,
      "step": 9327
    },
    {
      "epoch": 4.389647058823529,
      "grad_norm": 0.458664919909177,
      "learning_rate": 6.89608779011149e-05,
      "loss": 0.332,
      "step": 9328
    },
    {
      "epoch": 4.3901176470588235,
      "grad_norm": 0.3981603189080298,
      "learning_rate": 6.892399332374387e-05,
      "loss": 0.3945,
      "step": 9329
    },
    {
      "epoch": 4.390588235294118,
      "grad_norm": 0.41857180980075215,
      "learning_rate": 6.888711567066149e-05,
      "loss": 0.1426,
      "step": 9330
    },
    {
      "epoch": 4.391058823529412,
      "grad_norm": 0.426955465805134,
      "learning_rate": 6.885024494501735e-05,
      "loss": 0.1699,
      "step": 9331
    },
    {
      "epoch": 4.391529411764706,
      "grad_norm": 0.5526457673504717,
      "learning_rate": 6.881338114996027e-05,
      "loss": 0.6719,
      "step": 9332
    },
    {
      "epoch": 4.392,
      "grad_norm": 0.34140463841034235,
      "learning_rate": 6.877652428863866e-05,
      "loss": 0.1514,
      "step": 9333
    },
    {
      "epoch": 4.392470588235295,
      "grad_norm": 0.41818159428182355,
      "learning_rate": 6.873967436420033e-05,
      "loss": 0.2236,
      "step": 9334
    },
    {
      "epoch": 4.392941176470588,
      "grad_norm": 0.39171488576263164,
      "learning_rate": 6.870283137979233e-05,
      "loss": 0.249,
      "step": 9335
    },
    {
      "epoch": 4.393411764705882,
      "grad_norm": 0.41720213666437833,
      "learning_rate": 6.866599533856135e-05,
      "loss": 0.2656,
      "step": 9336
    },
    {
      "epoch": 4.393882352941176,
      "grad_norm": 0.474779482700993,
      "learning_rate": 6.862916624365325e-05,
      "loss": 0.2168,
      "step": 9337
    },
    {
      "epoch": 4.394352941176471,
      "grad_norm": 0.5747636786543048,
      "learning_rate": 6.859234409821347e-05,
      "loss": 0.6562,
      "step": 9338
    },
    {
      "epoch": 4.394823529411765,
      "grad_norm": 0.29528429275220797,
      "learning_rate": 6.855552890538688e-05,
      "loss": 0.1157,
      "step": 9339
    },
    {
      "epoch": 4.395294117647059,
      "grad_norm": 0.35273891590700074,
      "learning_rate": 6.851872066831751e-05,
      "loss": 0.2451,
      "step": 9340
    },
    {
      "epoch": 4.395764705882353,
      "grad_norm": 0.4345101940879654,
      "learning_rate": 6.848191939014909e-05,
      "loss": 0.1689,
      "step": 9341
    },
    {
      "epoch": 4.3962352941176475,
      "grad_norm": 0.4097194584480348,
      "learning_rate": 6.844512507402465e-05,
      "loss": 0.3711,
      "step": 9342
    },
    {
      "epoch": 4.396705882352941,
      "grad_norm": 0.33665850345143267,
      "learning_rate": 6.84083377230865e-05,
      "loss": 0.293,
      "step": 9343
    },
    {
      "epoch": 4.397176470588235,
      "grad_norm": 0.2663443816986656,
      "learning_rate": 6.837155734047652e-05,
      "loss": 0.1113,
      "step": 9344
    },
    {
      "epoch": 4.397647058823529,
      "grad_norm": 0.36956199595803335,
      "learning_rate": 6.833478392933597e-05,
      "loss": 0.0576,
      "step": 9345
    },
    {
      "epoch": 4.3981176470588235,
      "grad_norm": 0.3419829760707544,
      "learning_rate": 6.829801749280537e-05,
      "loss": 0.1069,
      "step": 9346
    },
    {
      "epoch": 4.398588235294118,
      "grad_norm": 0.33264457751545023,
      "learning_rate": 6.826125803402484e-05,
      "loss": 0.2441,
      "step": 9347
    },
    {
      "epoch": 4.399058823529412,
      "grad_norm": 1.8381178702801977,
      "learning_rate": 6.822450555613385e-05,
      "loss": 1.8828,
      "step": 9348
    },
    {
      "epoch": 4.399529411764706,
      "grad_norm": 0.4122482935377509,
      "learning_rate": 6.818776006227114e-05,
      "loss": 0.2383,
      "step": 9349
    },
    {
      "epoch": 4.4,
      "grad_norm": 0.44074589609244136,
      "learning_rate": 6.815102155557501e-05,
      "loss": 0.3008,
      "step": 9350
    },
    {
      "epoch": 4.400470588235294,
      "grad_norm": 0.4666264262773634,
      "learning_rate": 6.811429003918317e-05,
      "loss": 0.334,
      "step": 9351
    },
    {
      "epoch": 4.400941176470588,
      "grad_norm": 0.5275073895672688,
      "learning_rate": 6.807756551623248e-05,
      "loss": 0.5586,
      "step": 9352
    },
    {
      "epoch": 4.401411764705882,
      "grad_norm": 0.2695433170254594,
      "learning_rate": 6.804084798985964e-05,
      "loss": 0.1396,
      "step": 9353
    },
    {
      "epoch": 4.401882352941176,
      "grad_norm": 0.37554184813198943,
      "learning_rate": 6.800413746320034e-05,
      "loss": 0.103,
      "step": 9354
    },
    {
      "epoch": 4.402352941176471,
      "grad_norm": 0.28670810921521356,
      "learning_rate": 6.796743393938987e-05,
      "loss": 0.1719,
      "step": 9355
    },
    {
      "epoch": 4.402823529411765,
      "grad_norm": 0.28857021039768005,
      "learning_rate": 6.793073742156298e-05,
      "loss": 0.2227,
      "step": 9356
    },
    {
      "epoch": 4.403294117647059,
      "grad_norm": 0.3975280336045467,
      "learning_rate": 6.78940479128536e-05,
      "loss": 0.2656,
      "step": 9357
    },
    {
      "epoch": 4.403764705882353,
      "grad_norm": 0.6177575156996603,
      "learning_rate": 6.785736541639526e-05,
      "loss": 0.5469,
      "step": 9358
    },
    {
      "epoch": 4.4042352941176475,
      "grad_norm": 0.4194811448105218,
      "learning_rate": 6.782068993532085e-05,
      "loss": 0.4082,
      "step": 9359
    },
    {
      "epoch": 4.404705882352941,
      "grad_norm": 0.2887767738633156,
      "learning_rate": 6.778402147276259e-05,
      "loss": 0.1177,
      "step": 9360
    },
    {
      "epoch": 4.405176470588235,
      "grad_norm": 0.47756583897985694,
      "learning_rate": 6.774736003185215e-05,
      "loss": 0.5742,
      "step": 9361
    },
    {
      "epoch": 4.405647058823529,
      "grad_norm": 1.0665319183917756,
      "learning_rate": 6.771070561572067e-05,
      "loss": 0.9453,
      "step": 9362
    },
    {
      "epoch": 4.4061176470588235,
      "grad_norm": 0.9610296366928064,
      "learning_rate": 6.767405822749848e-05,
      "loss": 0.7773,
      "step": 9363
    },
    {
      "epoch": 4.406588235294118,
      "grad_norm": 0.5674458517759072,
      "learning_rate": 6.76374178703156e-05,
      "loss": 0.7266,
      "step": 9364
    },
    {
      "epoch": 4.407058823529412,
      "grad_norm": 0.92927952117236,
      "learning_rate": 6.760078454730115e-05,
      "loss": 0.5391,
      "step": 9365
    },
    {
      "epoch": 4.407529411764706,
      "grad_norm": 0.37465859641019533,
      "learning_rate": 6.756415826158388e-05,
      "loss": 0.1494,
      "step": 9366
    },
    {
      "epoch": 4.408,
      "grad_norm": 0.5453589838930183,
      "learning_rate": 6.752753901629189e-05,
      "loss": 0.625,
      "step": 9367
    },
    {
      "epoch": 4.408470588235295,
      "grad_norm": 0.3558640833620377,
      "learning_rate": 6.749092681455255e-05,
      "loss": 0.1797,
      "step": 9368
    },
    {
      "epoch": 4.408941176470588,
      "grad_norm": 0.4686299022523818,
      "learning_rate": 6.745432165949274e-05,
      "loss": 0.6758,
      "step": 9369
    },
    {
      "epoch": 4.409411764705882,
      "grad_norm": 0.32678891189043097,
      "learning_rate": 6.741772355423882e-05,
      "loss": 0.3105,
      "step": 9370
    },
    {
      "epoch": 4.409882352941176,
      "grad_norm": 0.46305416168893093,
      "learning_rate": 6.73811325019163e-05,
      "loss": 0.2637,
      "step": 9371
    },
    {
      "epoch": 4.410352941176471,
      "grad_norm": 0.34373781300131506,
      "learning_rate": 6.734454850565031e-05,
      "loss": 0.3574,
      "step": 9372
    },
    {
      "epoch": 4.410823529411765,
      "grad_norm": 0.48425454814013236,
      "learning_rate": 6.730797156856537e-05,
      "loss": 0.1982,
      "step": 9373
    },
    {
      "epoch": 4.411294117647059,
      "grad_norm": 0.19550373462516527,
      "learning_rate": 6.727140169378521e-05,
      "loss": 0.1108,
      "step": 9374
    },
    {
      "epoch": 4.411764705882353,
      "grad_norm": 0.34110568721185597,
      "learning_rate": 6.72348388844331e-05,
      "loss": 0.457,
      "step": 9375
    },
    {
      "epoch": 4.412235294117647,
      "grad_norm": 0.40856344383859045,
      "learning_rate": 6.719828314363178e-05,
      "loss": 0.2305,
      "step": 9376
    },
    {
      "epoch": 4.412705882352941,
      "grad_norm": 0.36975686236148364,
      "learning_rate": 6.716173447450317e-05,
      "loss": 0.2256,
      "step": 9377
    },
    {
      "epoch": 4.413176470588235,
      "grad_norm": 0.3591967608640659,
      "learning_rate": 6.712519288016875e-05,
      "loss": 0.3184,
      "step": 9378
    },
    {
      "epoch": 4.413647058823529,
      "grad_norm": 0.3413487882946796,
      "learning_rate": 6.70886583637494e-05,
      "loss": 0.1885,
      "step": 9379
    },
    {
      "epoch": 4.4141176470588235,
      "grad_norm": 0.4949099588681511,
      "learning_rate": 6.705213092836527e-05,
      "loss": 0.4941,
      "step": 9380
    },
    {
      "epoch": 4.414588235294118,
      "grad_norm": 0.33078590572006134,
      "learning_rate": 6.701561057713599e-05,
      "loss": 0.1807,
      "step": 9381
    },
    {
      "epoch": 4.415058823529412,
      "grad_norm": 0.5149234256053193,
      "learning_rate": 6.697909731318069e-05,
      "loss": 0.5,
      "step": 9382
    },
    {
      "epoch": 4.415529411764706,
      "grad_norm": 0.5543069075560281,
      "learning_rate": 6.694259113961761e-05,
      "loss": 0.4004,
      "step": 9383
    },
    {
      "epoch": 4.416,
      "grad_norm": 0.6364349363930295,
      "learning_rate": 6.690609205956464e-05,
      "loss": 0.4551,
      "step": 9384
    },
    {
      "epoch": 4.416470588235294,
      "grad_norm": 0.35724629720055623,
      "learning_rate": 6.686960007613906e-05,
      "loss": 0.3711,
      "step": 9385
    },
    {
      "epoch": 4.416941176470588,
      "grad_norm": 0.6819482304754703,
      "learning_rate": 6.683311519245733e-05,
      "loss": 0.3242,
      "step": 9386
    },
    {
      "epoch": 4.417411764705882,
      "grad_norm": 0.48230754164785206,
      "learning_rate": 6.679663741163553e-05,
      "loss": 0.4785,
      "step": 9387
    },
    {
      "epoch": 4.417882352941176,
      "grad_norm": 0.4737147243208682,
      "learning_rate": 6.676016673678896e-05,
      "loss": 0.3555,
      "step": 9388
    },
    {
      "epoch": 4.418352941176471,
      "grad_norm": 0.4464886656855152,
      "learning_rate": 6.672370317103245e-05,
      "loss": 0.4551,
      "step": 9389
    },
    {
      "epoch": 4.418823529411765,
      "grad_norm": 0.3860774311907743,
      "learning_rate": 6.668724671748018e-05,
      "loss": 0.3887,
      "step": 9390
    },
    {
      "epoch": 4.419294117647059,
      "grad_norm": 0.31736552590919287,
      "learning_rate": 6.665079737924566e-05,
      "loss": 0.0879,
      "step": 9391
    },
    {
      "epoch": 4.419764705882353,
      "grad_norm": 0.3929797185658096,
      "learning_rate": 6.661435515944186e-05,
      "loss": 0.4375,
      "step": 9392
    },
    {
      "epoch": 4.4202352941176475,
      "grad_norm": 0.5214019255804232,
      "learning_rate": 6.65779200611812e-05,
      "loss": 0.5859,
      "step": 9393
    },
    {
      "epoch": 4.420705882352941,
      "grad_norm": 0.36237166802778636,
      "learning_rate": 6.654149208757529e-05,
      "loss": 0.2266,
      "step": 9394
    },
    {
      "epoch": 4.421176470588235,
      "grad_norm": 0.26020558615185985,
      "learning_rate": 6.65050712417353e-05,
      "loss": 0.0942,
      "step": 9395
    },
    {
      "epoch": 4.421647058823529,
      "grad_norm": 0.4488768397569073,
      "learning_rate": 6.646865752677186e-05,
      "loss": 0.5234,
      "step": 9396
    },
    {
      "epoch": 4.4221176470588235,
      "grad_norm": 0.37678656700561386,
      "learning_rate": 6.643225094579462e-05,
      "loss": 0.105,
      "step": 9397
    },
    {
      "epoch": 4.422588235294118,
      "grad_norm": 0.44982998465089985,
      "learning_rate": 6.63958515019132e-05,
      "loss": 0.5586,
      "step": 9398
    },
    {
      "epoch": 4.423058823529412,
      "grad_norm": 0.5863193373868075,
      "learning_rate": 6.635945919823607e-05,
      "loss": 0.3867,
      "step": 9399
    },
    {
      "epoch": 4.423529411764706,
      "grad_norm": 0.43139993254382597,
      "learning_rate": 6.632307403787138e-05,
      "loss": 0.5312,
      "step": 9400
    },
    {
      "epoch": 4.424,
      "grad_norm": 0.7362986709486334,
      "learning_rate": 6.628669602392665e-05,
      "loss": 0.6992,
      "step": 9401
    },
    {
      "epoch": 4.424470588235294,
      "grad_norm": 0.4266839122771184,
      "learning_rate": 6.625032515950865e-05,
      "loss": 0.1777,
      "step": 9402
    },
    {
      "epoch": 4.424941176470588,
      "grad_norm": 0.22691549540731412,
      "learning_rate": 6.621396144772365e-05,
      "loss": 0.1367,
      "step": 9403
    },
    {
      "epoch": 4.425411764705882,
      "grad_norm": 0.3500242078385897,
      "learning_rate": 6.617760489167738e-05,
      "loss": 0.3984,
      "step": 9404
    },
    {
      "epoch": 4.425882352941176,
      "grad_norm": 0.4839527442223689,
      "learning_rate": 6.614125549447475e-05,
      "loss": 0.4355,
      "step": 9405
    },
    {
      "epoch": 4.426352941176471,
      "grad_norm": 0.22075679287743336,
      "learning_rate": 6.610491325922022e-05,
      "loss": 0.043,
      "step": 9406
    },
    {
      "epoch": 4.426823529411765,
      "grad_norm": 0.4321895631650349,
      "learning_rate": 6.606857818901765e-05,
      "loss": 0.5703,
      "step": 9407
    },
    {
      "epoch": 4.427294117647059,
      "grad_norm": 0.25096499802136996,
      "learning_rate": 6.603225028697013e-05,
      "loss": 0.0461,
      "step": 9408
    },
    {
      "epoch": 4.427764705882353,
      "grad_norm": 0.4160266775746808,
      "learning_rate": 6.59959295561803e-05,
      "loss": 0.2354,
      "step": 9409
    },
    {
      "epoch": 4.428235294117647,
      "grad_norm": 0.21591509450139787,
      "learning_rate": 6.595961599975016e-05,
      "loss": 0.0688,
      "step": 9410
    },
    {
      "epoch": 4.428705882352941,
      "grad_norm": 0.3365397120399856,
      "learning_rate": 6.5923309620781e-05,
      "loss": 0.4746,
      "step": 9411
    },
    {
      "epoch": 4.429176470588235,
      "grad_norm": 0.4975046761969541,
      "learning_rate": 6.588701042237358e-05,
      "loss": 0.4609,
      "step": 9412
    },
    {
      "epoch": 4.429647058823529,
      "grad_norm": 0.5839630234973389,
      "learning_rate": 6.585071840762809e-05,
      "loss": 0.416,
      "step": 9413
    },
    {
      "epoch": 4.4301176470588235,
      "grad_norm": 0.40931674547281366,
      "learning_rate": 6.581443357964395e-05,
      "loss": 0.2354,
      "step": 9414
    },
    {
      "epoch": 4.430588235294118,
      "grad_norm": 0.2767915506377036,
      "learning_rate": 6.577815594152017e-05,
      "loss": 0.1191,
      "step": 9415
    },
    {
      "epoch": 4.431058823529412,
      "grad_norm": 0.6005944075410902,
      "learning_rate": 6.574188549635493e-05,
      "loss": 0.5234,
      "step": 9416
    },
    {
      "epoch": 4.431529411764706,
      "grad_norm": 0.27912524005793554,
      "learning_rate": 6.570562224724593e-05,
      "loss": 0.1758,
      "step": 9417
    },
    {
      "epoch": 4.432,
      "grad_norm": 0.34400742460274236,
      "learning_rate": 6.566936619729033e-05,
      "loss": 0.2041,
      "step": 9418
    },
    {
      "epoch": 4.432470588235294,
      "grad_norm": 0.34463095376742,
      "learning_rate": 6.563311734958444e-05,
      "loss": 0.2949,
      "step": 9419
    },
    {
      "epoch": 4.432941176470588,
      "grad_norm": 0.3448901505406867,
      "learning_rate": 6.559687570722413e-05,
      "loss": 0.3379,
      "step": 9420
    },
    {
      "epoch": 4.433411764705882,
      "grad_norm": 0.2970485868630446,
      "learning_rate": 6.556064127330467e-05,
      "loss": 0.1416,
      "step": 9421
    },
    {
      "epoch": 4.433882352941176,
      "grad_norm": 0.4715746185417592,
      "learning_rate": 6.552441405092058e-05,
      "loss": 0.4727,
      "step": 9422
    },
    {
      "epoch": 4.434352941176471,
      "grad_norm": 0.5531336223498571,
      "learning_rate": 6.54881940431659e-05,
      "loss": 0.4277,
      "step": 9423
    },
    {
      "epoch": 4.434823529411765,
      "grad_norm": 0.3248447561373077,
      "learning_rate": 6.5451981253134e-05,
      "loss": 0.252,
      "step": 9424
    },
    {
      "epoch": 4.435294117647059,
      "grad_norm": 0.2612562166925839,
      "learning_rate": 6.541577568391758e-05,
      "loss": 0.0903,
      "step": 9425
    },
    {
      "epoch": 4.435764705882353,
      "grad_norm": 0.212193535502076,
      "learning_rate": 6.537957733860879e-05,
      "loss": 0.0962,
      "step": 9426
    },
    {
      "epoch": 4.4362352941176475,
      "grad_norm": 0.7559994532132218,
      "learning_rate": 6.53433862202992e-05,
      "loss": 1.1016,
      "step": 9427
    },
    {
      "epoch": 4.436705882352941,
      "grad_norm": 0.31846236099680864,
      "learning_rate": 6.530720233207962e-05,
      "loss": 0.1436,
      "step": 9428
    },
    {
      "epoch": 4.437176470588235,
      "grad_norm": 0.3663695152153624,
      "learning_rate": 6.527102567704036e-05,
      "loss": 0.4199,
      "step": 9429
    },
    {
      "epoch": 4.437647058823529,
      "grad_norm": 0.16881429722143898,
      "learning_rate": 6.523485625827119e-05,
      "loss": 0.0649,
      "step": 9430
    },
    {
      "epoch": 4.4381176470588235,
      "grad_norm": 0.398215872418745,
      "learning_rate": 6.519869407886098e-05,
      "loss": 0.4902,
      "step": 9431
    },
    {
      "epoch": 4.438588235294118,
      "grad_norm": 0.4320558107173902,
      "learning_rate": 6.516253914189825e-05,
      "loss": 0.5,
      "step": 9432
    },
    {
      "epoch": 4.439058823529412,
      "grad_norm": 0.3151019045428687,
      "learning_rate": 6.512639145047086e-05,
      "loss": 0.1143,
      "step": 9433
    },
    {
      "epoch": 4.439529411764706,
      "grad_norm": 0.22104177947968826,
      "learning_rate": 6.509025100766591e-05,
      "loss": 0.1504,
      "step": 9434
    },
    {
      "epoch": 4.44,
      "grad_norm": 0.43063857239077324,
      "learning_rate": 6.505411781656997e-05,
      "loss": 0.5234,
      "step": 9435
    },
    {
      "epoch": 4.440470588235294,
      "grad_norm": 0.5110333981607902,
      "learning_rate": 6.50179918802691e-05,
      "loss": 0.6172,
      "step": 9436
    },
    {
      "epoch": 4.440941176470588,
      "grad_norm": 0.28800730985935513,
      "learning_rate": 6.49818732018485e-05,
      "loss": 0.1982,
      "step": 9437
    },
    {
      "epoch": 4.441411764705882,
      "grad_norm": 0.40315214255778886,
      "learning_rate": 6.494576178439298e-05,
      "loss": 0.3301,
      "step": 9438
    },
    {
      "epoch": 4.441882352941176,
      "grad_norm": 0.35863221441598714,
      "learning_rate": 6.490965763098654e-05,
      "loss": 0.4746,
      "step": 9439
    },
    {
      "epoch": 4.442352941176471,
      "grad_norm": 0.43108257710252534,
      "learning_rate": 6.487356074471272e-05,
      "loss": 0.4766,
      "step": 9440
    },
    {
      "epoch": 4.442823529411765,
      "grad_norm": 0.4170606929378129,
      "learning_rate": 6.483747112865436e-05,
      "loss": 0.3633,
      "step": 9441
    },
    {
      "epoch": 4.443294117647059,
      "grad_norm": 0.3743152218201505,
      "learning_rate": 6.480138878589368e-05,
      "loss": 0.1367,
      "step": 9442
    },
    {
      "epoch": 4.443764705882353,
      "grad_norm": 0.4111081724427319,
      "learning_rate": 6.476531371951229e-05,
      "loss": 0.6367,
      "step": 9443
    },
    {
      "epoch": 4.444235294117647,
      "grad_norm": 0.18922270685859593,
      "learning_rate": 6.472924593259122e-05,
      "loss": 0.0265,
      "step": 9444
    },
    {
      "epoch": 4.444705882352941,
      "grad_norm": 0.574539452241794,
      "learning_rate": 6.469318542821078e-05,
      "loss": 0.543,
      "step": 9445
    },
    {
      "epoch": 4.445176470588235,
      "grad_norm": 0.28250969781099255,
      "learning_rate": 6.465713220945075e-05,
      "loss": 0.104,
      "step": 9446
    },
    {
      "epoch": 4.445647058823529,
      "grad_norm": 0.26240749045701833,
      "learning_rate": 6.46210862793902e-05,
      "loss": 0.1982,
      "step": 9447
    },
    {
      "epoch": 4.4461176470588235,
      "grad_norm": 0.16165437529406768,
      "learning_rate": 6.458504764110764e-05,
      "loss": 0.0496,
      "step": 9448
    },
    {
      "epoch": 4.446588235294118,
      "grad_norm": 0.35878865069378624,
      "learning_rate": 6.454901629768105e-05,
      "loss": 0.1787,
      "step": 9449
    },
    {
      "epoch": 4.447058823529412,
      "grad_norm": 0.42113648442770296,
      "learning_rate": 6.451299225218754e-05,
      "loss": 0.5039,
      "step": 9450
    },
    {
      "epoch": 4.447529411764706,
      "grad_norm": 0.26091454748165105,
      "learning_rate": 6.447697550770381e-05,
      "loss": 0.0913,
      "step": 9451
    },
    {
      "epoch": 4.448,
      "grad_norm": 0.306664282733324,
      "learning_rate": 6.44409660673059e-05,
      "loss": 0.1592,
      "step": 9452
    },
    {
      "epoch": 4.448470588235294,
      "grad_norm": 0.46837396164322564,
      "learning_rate": 6.440496393406912e-05,
      "loss": 0.3535,
      "step": 9453
    },
    {
      "epoch": 4.448941176470588,
      "grad_norm": 0.5292832944868016,
      "learning_rate": 6.436896911106824e-05,
      "loss": 0.6406,
      "step": 9454
    },
    {
      "epoch": 4.449411764705882,
      "grad_norm": 0.39641155545462664,
      "learning_rate": 6.433298160137749e-05,
      "loss": 0.2773,
      "step": 9455
    },
    {
      "epoch": 4.449882352941176,
      "grad_norm": 0.37901327706316046,
      "learning_rate": 6.429700140807023e-05,
      "loss": 0.123,
      "step": 9456
    },
    {
      "epoch": 4.450352941176471,
      "grad_norm": 0.46603152971566364,
      "learning_rate": 6.426102853421942e-05,
      "loss": 0.2412,
      "step": 9457
    },
    {
      "epoch": 4.450823529411765,
      "grad_norm": 0.41317158937076265,
      "learning_rate": 6.42250629828974e-05,
      "loss": 0.5078,
      "step": 9458
    },
    {
      "epoch": 4.451294117647059,
      "grad_norm": 0.3703114831863103,
      "learning_rate": 6.418910475717565e-05,
      "loss": 0.3594,
      "step": 9459
    },
    {
      "epoch": 4.451764705882353,
      "grad_norm": 0.39156664407530245,
      "learning_rate": 6.415315386012525e-05,
      "loss": 0.3242,
      "step": 9460
    },
    {
      "epoch": 4.4522352941176475,
      "grad_norm": 0.7695764254158001,
      "learning_rate": 6.411721029481665e-05,
      "loss": 0.5078,
      "step": 9461
    },
    {
      "epoch": 4.452705882352941,
      "grad_norm": 0.5462943625388326,
      "learning_rate": 6.40812740643195e-05,
      "loss": 0.5117,
      "step": 9462
    },
    {
      "epoch": 4.453176470588235,
      "grad_norm": 0.509526822720963,
      "learning_rate": 6.404534517170296e-05,
      "loss": 0.5508,
      "step": 9463
    },
    {
      "epoch": 4.453647058823529,
      "grad_norm": 0.458523953728498,
      "learning_rate": 6.40094236200356e-05,
      "loss": 0.498,
      "step": 9464
    },
    {
      "epoch": 4.4541176470588235,
      "grad_norm": 0.32289993987245585,
      "learning_rate": 6.397350941238519e-05,
      "loss": 0.1953,
      "step": 9465
    },
    {
      "epoch": 4.454588235294118,
      "grad_norm": 0.34756510067339735,
      "learning_rate": 6.39376025518191e-05,
      "loss": 0.3906,
      "step": 9466
    },
    {
      "epoch": 4.455058823529412,
      "grad_norm": 0.3158097273104786,
      "learning_rate": 6.390170304140382e-05,
      "loss": 0.0825,
      "step": 9467
    },
    {
      "epoch": 4.455529411764706,
      "grad_norm": 0.23942085832273724,
      "learning_rate": 6.386581088420542e-05,
      "loss": 0.0645,
      "step": 9468
    },
    {
      "epoch": 4.456,
      "grad_norm": 0.2882826100704934,
      "learning_rate": 6.382992608328929e-05,
      "loss": 0.0977,
      "step": 9469
    },
    {
      "epoch": 4.456470588235294,
      "grad_norm": 0.356148739426535,
      "learning_rate": 6.37940486417201e-05,
      "loss": 0.248,
      "step": 9470
    },
    {
      "epoch": 4.456941176470588,
      "grad_norm": 0.38594494629391013,
      "learning_rate": 6.375817856256199e-05,
      "loss": 0.4707,
      "step": 9471
    },
    {
      "epoch": 4.457411764705882,
      "grad_norm": 0.6776915621572412,
      "learning_rate": 6.37223158488785e-05,
      "loss": 0.5352,
      "step": 9472
    },
    {
      "epoch": 4.457882352941176,
      "grad_norm": 0.12333847683312815,
      "learning_rate": 6.368646050373236e-05,
      "loss": 0.0172,
      "step": 9473
    },
    {
      "epoch": 4.458352941176471,
      "grad_norm": 0.5611326791468352,
      "learning_rate": 6.365061253018586e-05,
      "loss": 0.3848,
      "step": 9474
    },
    {
      "epoch": 4.458823529411765,
      "grad_norm": 0.2936080080586423,
      "learning_rate": 6.361477193130065e-05,
      "loss": 0.0732,
      "step": 9475
    },
    {
      "epoch": 4.459294117647059,
      "grad_norm": 0.5586618309940437,
      "learning_rate": 6.35789387101376e-05,
      "loss": 0.707,
      "step": 9476
    },
    {
      "epoch": 4.459764705882353,
      "grad_norm": 0.48672538742557664,
      "learning_rate": 6.354311286975705e-05,
      "loss": 0.3887,
      "step": 9477
    },
    {
      "epoch": 4.460235294117647,
      "grad_norm": 0.38311930183608645,
      "learning_rate": 6.35072944132188e-05,
      "loss": 0.3555,
      "step": 9478
    },
    {
      "epoch": 4.460705882352941,
      "grad_norm": 0.2789728420965133,
      "learning_rate": 6.34714833435818e-05,
      "loss": 0.1328,
      "step": 9479
    },
    {
      "epoch": 4.461176470588235,
      "grad_norm": 0.3387269981754175,
      "learning_rate": 6.343567966390452e-05,
      "loss": 0.3418,
      "step": 9480
    },
    {
      "epoch": 4.461647058823529,
      "grad_norm": 0.39268159809974246,
      "learning_rate": 6.339988337724487e-05,
      "loss": 0.5234,
      "step": 9481
    },
    {
      "epoch": 4.4621176470588235,
      "grad_norm": 0.2913521575027766,
      "learning_rate": 6.336409448665988e-05,
      "loss": 0.3027,
      "step": 9482
    },
    {
      "epoch": 4.462588235294118,
      "grad_norm": 0.3520690455135898,
      "learning_rate": 6.332831299520617e-05,
      "loss": 0.2393,
      "step": 9483
    },
    {
      "epoch": 4.463058823529412,
      "grad_norm": 0.3005043277355957,
      "learning_rate": 6.32925389059397e-05,
      "loss": 0.1309,
      "step": 9484
    },
    {
      "epoch": 4.463529411764706,
      "grad_norm": 0.3507331412581702,
      "learning_rate": 6.325677222191568e-05,
      "loss": 0.2539,
      "step": 9485
    },
    {
      "epoch": 4.464,
      "grad_norm": 0.2539788000265187,
      "learning_rate": 6.322101294618875e-05,
      "loss": 0.2461,
      "step": 9486
    },
    {
      "epoch": 4.464470588235294,
      "grad_norm": 0.3852352740046382,
      "learning_rate": 6.318526108181299e-05,
      "loss": 0.5508,
      "step": 9487
    },
    {
      "epoch": 4.464941176470588,
      "grad_norm": 0.6822020612200583,
      "learning_rate": 6.314951663184175e-05,
      "loss": 0.5352,
      "step": 9488
    },
    {
      "epoch": 4.465411764705882,
      "grad_norm": 0.8084805960943722,
      "learning_rate": 6.311377959932783e-05,
      "loss": 0.7305,
      "step": 9489
    },
    {
      "epoch": 4.465882352941176,
      "grad_norm": 0.22664663661476622,
      "learning_rate": 6.307804998732328e-05,
      "loss": 0.1426,
      "step": 9490
    },
    {
      "epoch": 4.466352941176471,
      "grad_norm": 0.5185589820350232,
      "learning_rate": 6.304232779887958e-05,
      "loss": 0.3809,
      "step": 9491
    },
    {
      "epoch": 4.466823529411765,
      "grad_norm": 0.3061451943902828,
      "learning_rate": 6.300661303704768e-05,
      "loss": 0.21,
      "step": 9492
    },
    {
      "epoch": 4.467294117647059,
      "grad_norm": 0.3759656195137285,
      "learning_rate": 6.297090570487766e-05,
      "loss": 0.418,
      "step": 9493
    },
    {
      "epoch": 4.467764705882353,
      "grad_norm": 0.2800039037076723,
      "learning_rate": 6.293520580541925e-05,
      "loss": 0.2354,
      "step": 9494
    },
    {
      "epoch": 4.4682352941176475,
      "grad_norm": 0.38226706023064916,
      "learning_rate": 6.289951334172123e-05,
      "loss": 0.5039,
      "step": 9495
    },
    {
      "epoch": 4.468705882352941,
      "grad_norm": 0.3075201980941335,
      "learning_rate": 6.2863828316832e-05,
      "loss": 0.168,
      "step": 9496
    },
    {
      "epoch": 4.469176470588235,
      "grad_norm": 0.35958913605304776,
      "learning_rate": 6.282815073379932e-05,
      "loss": 0.2969,
      "step": 9497
    },
    {
      "epoch": 4.469647058823529,
      "grad_norm": 0.2542541104778692,
      "learning_rate": 6.279248059567006e-05,
      "loss": 0.0928,
      "step": 9498
    },
    {
      "epoch": 4.4701176470588235,
      "grad_norm": 0.3663427590768311,
      "learning_rate": 6.275681790549072e-05,
      "loss": 0.5078,
      "step": 9499
    },
    {
      "epoch": 4.470588235294118,
      "grad_norm": 0.32880467429124344,
      "learning_rate": 6.27211626663071e-05,
      "loss": 0.1396,
      "step": 9500
    },
    {
      "epoch": 4.471058823529412,
      "grad_norm": 0.3485520786086976,
      "learning_rate": 6.268551488116423e-05,
      "loss": 0.123,
      "step": 9501
    },
    {
      "epoch": 4.471529411764706,
      "grad_norm": 0.3547583813440564,
      "learning_rate": 6.264987455310667e-05,
      "loss": 0.2949,
      "step": 9502
    },
    {
      "epoch": 4.4719999999999995,
      "grad_norm": 0.423122112024618,
      "learning_rate": 6.261424168517832e-05,
      "loss": 0.4277,
      "step": 9503
    },
    {
      "epoch": 4.472470588235294,
      "grad_norm": 0.3484871197591451,
      "learning_rate": 6.257861628042233e-05,
      "loss": 0.582,
      "step": 9504
    },
    {
      "epoch": 4.472941176470588,
      "grad_norm": 0.30320131696741637,
      "learning_rate": 6.254299834188128e-05,
      "loss": 0.3418,
      "step": 9505
    },
    {
      "epoch": 4.473411764705882,
      "grad_norm": 0.40026298633639457,
      "learning_rate": 6.25073878725972e-05,
      "loss": 0.2559,
      "step": 9506
    },
    {
      "epoch": 4.473882352941176,
      "grad_norm": 0.3655891626146175,
      "learning_rate": 6.24717848756113e-05,
      "loss": 0.3477,
      "step": 9507
    },
    {
      "epoch": 4.474352941176471,
      "grad_norm": 0.3156509496626286,
      "learning_rate": 6.24361893539643e-05,
      "loss": 0.375,
      "step": 9508
    },
    {
      "epoch": 4.474823529411765,
      "grad_norm": 0.4436682196231618,
      "learning_rate": 6.240060131069628e-05,
      "loss": 0.3828,
      "step": 9509
    },
    {
      "epoch": 4.475294117647059,
      "grad_norm": 0.4392667432076447,
      "learning_rate": 6.23650207488465e-05,
      "loss": 0.3691,
      "step": 9510
    },
    {
      "epoch": 4.475764705882353,
      "grad_norm": 0.6211692807903398,
      "learning_rate": 6.232944767145384e-05,
      "loss": 0.4668,
      "step": 9511
    },
    {
      "epoch": 4.476235294117647,
      "grad_norm": 0.3443921547885283,
      "learning_rate": 6.229388208155639e-05,
      "loss": 0.3145,
      "step": 9512
    },
    {
      "epoch": 4.476705882352941,
      "grad_norm": 0.4879594602274584,
      "learning_rate": 6.225832398219158e-05,
      "loss": 0.707,
      "step": 9513
    },
    {
      "epoch": 4.477176470588235,
      "grad_norm": 0.33751208987768017,
      "learning_rate": 6.222277337639624e-05,
      "loss": 0.4375,
      "step": 9514
    },
    {
      "epoch": 4.477647058823529,
      "grad_norm": 0.7662170375010561,
      "learning_rate": 6.218723026720668e-05,
      "loss": 0.7734,
      "step": 9515
    },
    {
      "epoch": 4.4781176470588235,
      "grad_norm": 0.4315712485626141,
      "learning_rate": 6.21516946576583e-05,
      "loss": 0.5547,
      "step": 9516
    },
    {
      "epoch": 4.478588235294118,
      "grad_norm": 0.6066751643420278,
      "learning_rate": 6.211616655078616e-05,
      "loss": 0.3828,
      "step": 9517
    },
    {
      "epoch": 4.479058823529412,
      "grad_norm": 0.42352080788225843,
      "learning_rate": 6.208064594962441e-05,
      "loss": 0.4766,
      "step": 9518
    },
    {
      "epoch": 4.479529411764706,
      "grad_norm": 0.156718293031962,
      "learning_rate": 6.204513285720673e-05,
      "loss": 0.04,
      "step": 9519
    },
    {
      "epoch": 4.48,
      "grad_norm": 0.4499147458032399,
      "learning_rate": 6.200962727656616e-05,
      "loss": 0.4316,
      "step": 9520
    },
    {
      "epoch": 4.480470588235294,
      "grad_norm": 0.3798970123810873,
      "learning_rate": 6.197412921073498e-05,
      "loss": 0.2061,
      "step": 9521
    },
    {
      "epoch": 4.480941176470588,
      "grad_norm": 0.7468014327680358,
      "learning_rate": 6.193863866274492e-05,
      "loss": 0.2871,
      "step": 9522
    },
    {
      "epoch": 4.481411764705882,
      "grad_norm": 0.23374360193181354,
      "learning_rate": 6.190315563562711e-05,
      "loss": 0.2188,
      "step": 9523
    },
    {
      "epoch": 4.481882352941176,
      "grad_norm": 0.3108130759110661,
      "learning_rate": 6.186768013241188e-05,
      "loss": 0.1924,
      "step": 9524
    },
    {
      "epoch": 4.482352941176471,
      "grad_norm": 0.3182031199287919,
      "learning_rate": 6.183221215612904e-05,
      "loss": 0.2021,
      "step": 9525
    },
    {
      "epoch": 4.482823529411765,
      "grad_norm": 0.4184813093511476,
      "learning_rate": 6.179675170980779e-05,
      "loss": 0.3711,
      "step": 9526
    },
    {
      "epoch": 4.483294117647059,
      "grad_norm": 0.3064909161304625,
      "learning_rate": 6.176129879647654e-05,
      "loss": 0.3184,
      "step": 9527
    },
    {
      "epoch": 4.483764705882353,
      "grad_norm": 0.3086242381488112,
      "learning_rate": 6.172585341916317e-05,
      "loss": 0.1445,
      "step": 9528
    },
    {
      "epoch": 4.484235294117647,
      "grad_norm": 0.30229691019921556,
      "learning_rate": 6.169041558089496e-05,
      "loss": 0.2812,
      "step": 9529
    },
    {
      "epoch": 4.484705882352941,
      "grad_norm": 0.4547370862342093,
      "learning_rate": 6.16549852846983e-05,
      "loss": 0.3809,
      "step": 9530
    },
    {
      "epoch": 4.485176470588235,
      "grad_norm": 0.3741064373279513,
      "learning_rate": 6.161956253359935e-05,
      "loss": 0.1484,
      "step": 9531
    },
    {
      "epoch": 4.485647058823529,
      "grad_norm": 0.5527972119484892,
      "learning_rate": 6.15841473306232e-05,
      "loss": 0.4863,
      "step": 9532
    },
    {
      "epoch": 4.4861176470588235,
      "grad_norm": 0.23537244661307705,
      "learning_rate": 6.154873967879455e-05,
      "loss": 0.0869,
      "step": 9533
    },
    {
      "epoch": 4.486588235294118,
      "grad_norm": 0.3148681977642647,
      "learning_rate": 6.151333958113743e-05,
      "loss": 0.2314,
      "step": 9534
    },
    {
      "epoch": 4.487058823529412,
      "grad_norm": 0.38594990344871105,
      "learning_rate": 6.147794704067508e-05,
      "loss": 0.1982,
      "step": 9535
    },
    {
      "epoch": 4.487529411764706,
      "grad_norm": 0.34450038583631226,
      "learning_rate": 6.144256206043025e-05,
      "loss": 0.1846,
      "step": 9536
    },
    {
      "epoch": 4.4879999999999995,
      "grad_norm": 0.3601662650021701,
      "learning_rate": 6.140718464342505e-05,
      "loss": 0.1475,
      "step": 9537
    },
    {
      "epoch": 4.488470588235294,
      "grad_norm": 0.2878659694847214,
      "learning_rate": 6.137181479268077e-05,
      "loss": 0.1914,
      "step": 9538
    },
    {
      "epoch": 4.488941176470588,
      "grad_norm": 0.33405852661196533,
      "learning_rate": 6.133645251121823e-05,
      "loss": 0.1206,
      "step": 9539
    },
    {
      "epoch": 4.489411764705882,
      "grad_norm": 0.22998364225999282,
      "learning_rate": 6.130109780205759e-05,
      "loss": 0.0613,
      "step": 9540
    },
    {
      "epoch": 4.489882352941176,
      "grad_norm": 0.14946772967793373,
      "learning_rate": 6.126575066821821e-05,
      "loss": 0.0359,
      "step": 9541
    },
    {
      "epoch": 4.490352941176471,
      "grad_norm": 0.3966162995902042,
      "learning_rate": 6.123041111271896e-05,
      "loss": 0.1348,
      "step": 9542
    },
    {
      "epoch": 4.490823529411765,
      "grad_norm": 0.2428952875312396,
      "learning_rate": 6.119507913857807e-05,
      "loss": 0.0693,
      "step": 9543
    },
    {
      "epoch": 4.491294117647059,
      "grad_norm": 0.3801515892464311,
      "learning_rate": 6.115975474881296e-05,
      "loss": 0.2334,
      "step": 9544
    },
    {
      "epoch": 4.491764705882353,
      "grad_norm": 0.22857331776720835,
      "learning_rate": 6.11244379464406e-05,
      "loss": 0.0703,
      "step": 9545
    },
    {
      "epoch": 4.492235294117647,
      "grad_norm": 0.4668678240772989,
      "learning_rate": 6.108912873447712e-05,
      "loss": 0.2139,
      "step": 9546
    },
    {
      "epoch": 4.492705882352941,
      "grad_norm": 0.2574207230294784,
      "learning_rate": 6.105382711593814e-05,
      "loss": 0.1982,
      "step": 9547
    },
    {
      "epoch": 4.493176470588235,
      "grad_norm": 0.5877182953286925,
      "learning_rate": 6.101853309383866e-05,
      "loss": 0.5898,
      "step": 9548
    },
    {
      "epoch": 4.493647058823529,
      "grad_norm": 0.4398688388480697,
      "learning_rate": 6.0983246671192866e-05,
      "loss": 0.1963,
      "step": 9549
    },
    {
      "epoch": 4.4941176470588236,
      "grad_norm": 0.5193998931145282,
      "learning_rate": 6.0947967851014405e-05,
      "loss": 0.3594,
      "step": 9550
    },
    {
      "epoch": 4.494588235294118,
      "grad_norm": 0.40595367692150114,
      "learning_rate": 6.091269663631636e-05,
      "loss": 0.1904,
      "step": 9551
    },
    {
      "epoch": 4.495058823529412,
      "grad_norm": 0.3425153735134863,
      "learning_rate": 6.0877433030110935e-05,
      "loss": 0.3789,
      "step": 9552
    },
    {
      "epoch": 4.495529411764706,
      "grad_norm": 0.37958804410931507,
      "learning_rate": 6.084217703540988e-05,
      "loss": 0.1445,
      "step": 9553
    },
    {
      "epoch": 4.496,
      "grad_norm": 0.3489019751797683,
      "learning_rate": 6.0806928655224266e-05,
      "loss": 0.1455,
      "step": 9554
    },
    {
      "epoch": 4.496470588235294,
      "grad_norm": 0.6552296200977171,
      "learning_rate": 6.0771687892564394e-05,
      "loss": 0.4746,
      "step": 9555
    },
    {
      "epoch": 4.496941176470588,
      "grad_norm": 0.41543487751141817,
      "learning_rate": 6.0736454750440025e-05,
      "loss": 0.3125,
      "step": 9556
    },
    {
      "epoch": 4.497411764705882,
      "grad_norm": 0.7326426842698759,
      "learning_rate": 6.070122923186032e-05,
      "loss": 0.4922,
      "step": 9557
    },
    {
      "epoch": 4.497882352941176,
      "grad_norm": 0.26932977481489,
      "learning_rate": 6.0666011339833584e-05,
      "loss": 0.105,
      "step": 9558
    },
    {
      "epoch": 4.498352941176471,
      "grad_norm": 0.6332367336350668,
      "learning_rate": 6.063080107736767e-05,
      "loss": 0.8438,
      "step": 9559
    },
    {
      "epoch": 4.498823529411765,
      "grad_norm": 0.27430050408205925,
      "learning_rate": 6.0595598447469726e-05,
      "loss": 0.1016,
      "step": 9560
    },
    {
      "epoch": 4.499294117647059,
      "grad_norm": 0.2724993767193247,
      "learning_rate": 6.056040345314618e-05,
      "loss": 0.0771,
      "step": 9561
    },
    {
      "epoch": 4.499764705882353,
      "grad_norm": 0.3861617976333475,
      "learning_rate": 6.0525216097402865e-05,
      "loss": 0.2188,
      "step": 9562
    },
    {
      "epoch": 4.5002352941176476,
      "grad_norm": 0.43574260949107635,
      "learning_rate": 6.049003638324501e-05,
      "loss": 0.3242,
      "step": 9563
    },
    {
      "epoch": 4.500705882352941,
      "grad_norm": 0.29046132792008816,
      "learning_rate": 6.0454864313677054e-05,
      "loss": 0.3203,
      "step": 9564
    },
    {
      "epoch": 4.501176470588235,
      "grad_norm": 0.28302411852402665,
      "learning_rate": 6.041969989170289e-05,
      "loss": 0.1279,
      "step": 9565
    },
    {
      "epoch": 4.501647058823529,
      "grad_norm": 0.41182582315422095,
      "learning_rate": 6.0384543120325796e-05,
      "loss": 0.4258,
      "step": 9566
    },
    {
      "epoch": 4.502117647058824,
      "grad_norm": 0.46224919899361194,
      "learning_rate": 6.034939400254824e-05,
      "loss": 0.707,
      "step": 9567
    },
    {
      "epoch": 4.502588235294118,
      "grad_norm": 0.632784177810545,
      "learning_rate": 6.031425254137222e-05,
      "loss": 0.6094,
      "step": 9568
    },
    {
      "epoch": 4.503058823529412,
      "grad_norm": 0.3879983598938473,
      "learning_rate": 6.0279118739798874e-05,
      "loss": 0.167,
      "step": 9569
    },
    {
      "epoch": 4.503529411764706,
      "grad_norm": 0.3604770376807236,
      "learning_rate": 6.0243992600828875e-05,
      "loss": 0.1475,
      "step": 9570
    },
    {
      "epoch": 4.504,
      "grad_norm": 0.38145209603442537,
      "learning_rate": 6.020887412746221e-05,
      "loss": 0.5469,
      "step": 9571
    },
    {
      "epoch": 4.504470588235294,
      "grad_norm": 0.508228881507737,
      "learning_rate": 6.017376332269807e-05,
      "loss": 0.3184,
      "step": 9572
    },
    {
      "epoch": 4.504941176470588,
      "grad_norm": 0.28691805086967,
      "learning_rate": 6.0138660189535126e-05,
      "loss": 0.2305,
      "step": 9573
    },
    {
      "epoch": 4.505411764705882,
      "grad_norm": 0.45204571456161013,
      "learning_rate": 6.0103564730971375e-05,
      "loss": 0.2578,
      "step": 9574
    },
    {
      "epoch": 4.5058823529411764,
      "grad_norm": 0.38180748545954546,
      "learning_rate": 6.006847695000415e-05,
      "loss": 0.1416,
      "step": 9575
    },
    {
      "epoch": 4.506352941176471,
      "grad_norm": 0.9688364511985808,
      "learning_rate": 6.0033396849630136e-05,
      "loss": 0.4395,
      "step": 9576
    },
    {
      "epoch": 4.506823529411765,
      "grad_norm": 1.1545975943170719,
      "learning_rate": 5.999832443284527e-05,
      "loss": 0.3398,
      "step": 9577
    },
    {
      "epoch": 4.507294117647059,
      "grad_norm": 0.4583988484957539,
      "learning_rate": 5.9963259702644963e-05,
      "loss": 0.2871,
      "step": 9578
    },
    {
      "epoch": 4.507764705882353,
      "grad_norm": 0.31738695249834403,
      "learning_rate": 5.992820266202394e-05,
      "loss": 0.2852,
      "step": 9579
    },
    {
      "epoch": 4.508235294117647,
      "grad_norm": 0.3299397684084363,
      "learning_rate": 5.989315331397616e-05,
      "loss": 0.3262,
      "step": 9580
    },
    {
      "epoch": 4.508705882352941,
      "grad_norm": 0.3694659832261821,
      "learning_rate": 5.985811166149507e-05,
      "loss": 0.2305,
      "step": 9581
    },
    {
      "epoch": 4.509176470588235,
      "grad_norm": 0.3655067513215012,
      "learning_rate": 5.982307770757343e-05,
      "loss": 0.1426,
      "step": 9582
    },
    {
      "epoch": 4.509647058823529,
      "grad_norm": 0.45512438481789835,
      "learning_rate": 5.978805145520322e-05,
      "loss": 0.21,
      "step": 9583
    },
    {
      "epoch": 4.510117647058824,
      "grad_norm": 0.4707993146652324,
      "learning_rate": 5.97530329073759e-05,
      "loss": 0.6172,
      "step": 9584
    },
    {
      "epoch": 4.510588235294118,
      "grad_norm": 0.4677474009031081,
      "learning_rate": 5.971802206708229e-05,
      "loss": 0.416,
      "step": 9585
    },
    {
      "epoch": 4.511058823529412,
      "grad_norm": 0.30531740517928513,
      "learning_rate": 5.968301893731237e-05,
      "loss": 0.1187,
      "step": 9586
    },
    {
      "epoch": 4.511529411764706,
      "grad_norm": 0.35902155919868206,
      "learning_rate": 5.964802352105564e-05,
      "loss": 0.1426,
      "step": 9587
    },
    {
      "epoch": 4.5120000000000005,
      "grad_norm": 0.28066949723283985,
      "learning_rate": 5.961303582130092e-05,
      "loss": 0.2539,
      "step": 9588
    },
    {
      "epoch": 4.512470588235294,
      "grad_norm": 0.3547777544679758,
      "learning_rate": 5.957805584103623e-05,
      "loss": 0.2422,
      "step": 9589
    },
    {
      "epoch": 4.512941176470588,
      "grad_norm": 0.5366383311141081,
      "learning_rate": 5.954308358324909e-05,
      "loss": 0.8867,
      "step": 9590
    },
    {
      "epoch": 4.513411764705882,
      "grad_norm": 0.3879388439709665,
      "learning_rate": 5.950811905092636e-05,
      "loss": 0.4648,
      "step": 9591
    },
    {
      "epoch": 4.5138823529411765,
      "grad_norm": 0.4228807653813869,
      "learning_rate": 5.947316224705407e-05,
      "loss": 0.3438,
      "step": 9592
    },
    {
      "epoch": 4.514352941176471,
      "grad_norm": 0.35080292579840194,
      "learning_rate": 5.943821317461775e-05,
      "loss": 0.0947,
      "step": 9593
    },
    {
      "epoch": 4.514823529411765,
      "grad_norm": 0.357317458350918,
      "learning_rate": 5.940327183660228e-05,
      "loss": 0.4199,
      "step": 9594
    },
    {
      "epoch": 4.515294117647059,
      "grad_norm": 0.5248162446795899,
      "learning_rate": 5.936833823599172e-05,
      "loss": 0.2373,
      "step": 9595
    },
    {
      "epoch": 4.5157647058823525,
      "grad_norm": 0.3448085117544116,
      "learning_rate": 5.933341237576968e-05,
      "loss": 0.3262,
      "step": 9596
    },
    {
      "epoch": 4.516235294117647,
      "grad_norm": 0.3691126337648728,
      "learning_rate": 5.9298494258918875e-05,
      "loss": 0.104,
      "step": 9597
    },
    {
      "epoch": 4.516705882352941,
      "grad_norm": 0.6266062450992919,
      "learning_rate": 5.926358388842156e-05,
      "loss": 0.4238,
      "step": 9598
    },
    {
      "epoch": 4.517176470588235,
      "grad_norm": 0.3670365415933932,
      "learning_rate": 5.922868126725929e-05,
      "loss": 0.1689,
      "step": 9599
    },
    {
      "epoch": 4.517647058823529,
      "grad_norm": 0.3140122904505853,
      "learning_rate": 5.919378639841281e-05,
      "loss": 0.373,
      "step": 9600
    },
    {
      "epoch": 4.518117647058824,
      "grad_norm": 0.47913181281449485,
      "learning_rate": 5.915889928486239e-05,
      "loss": 0.5078,
      "step": 9601
    },
    {
      "epoch": 4.518588235294118,
      "grad_norm": 0.28838959256605157,
      "learning_rate": 5.912401992958757e-05,
      "loss": 0.1689,
      "step": 9602
    },
    {
      "epoch": 4.519058823529412,
      "grad_norm": 0.14413998942806855,
      "learning_rate": 5.908914833556715e-05,
      "loss": 0.0276,
      "step": 9603
    },
    {
      "epoch": 4.519529411764706,
      "grad_norm": 0.33411263335561203,
      "learning_rate": 5.9054284505779396e-05,
      "loss": 0.2754,
      "step": 9604
    },
    {
      "epoch": 4.52,
      "grad_norm": 0.37422902618386417,
      "learning_rate": 5.901942844320187e-05,
      "loss": 0.1177,
      "step": 9605
    },
    {
      "epoch": 4.520470588235294,
      "grad_norm": 0.3857608363128043,
      "learning_rate": 5.898458015081136e-05,
      "loss": 0.1768,
      "step": 9606
    },
    {
      "epoch": 4.520941176470588,
      "grad_norm": 0.40190839112825344,
      "learning_rate": 5.8949739631584136e-05,
      "loss": 0.5938,
      "step": 9607
    },
    {
      "epoch": 4.521411764705882,
      "grad_norm": 0.2040714005198363,
      "learning_rate": 5.8914906888495797e-05,
      "loss": 0.0684,
      "step": 9608
    },
    {
      "epoch": 4.5218823529411765,
      "grad_norm": 0.3364267343335748,
      "learning_rate": 5.888008192452112e-05,
      "loss": 0.1206,
      "step": 9609
    },
    {
      "epoch": 4.522352941176471,
      "grad_norm": 0.5261159843999879,
      "learning_rate": 5.8845264742634394e-05,
      "loss": 0.3828,
      "step": 9610
    },
    {
      "epoch": 4.522823529411765,
      "grad_norm": 0.4712691144131226,
      "learning_rate": 5.881045534580922e-05,
      "loss": 0.2188,
      "step": 9611
    },
    {
      "epoch": 4.523294117647059,
      "grad_norm": 0.47685090842919714,
      "learning_rate": 5.87756537370184e-05,
      "loss": 0.5117,
      "step": 9612
    },
    {
      "epoch": 4.523764705882353,
      "grad_norm": 0.2996457815592705,
      "learning_rate": 5.874085991923419e-05,
      "loss": 0.1455,
      "step": 9613
    },
    {
      "epoch": 4.524235294117647,
      "grad_norm": 0.405986468138809,
      "learning_rate": 5.8706073895428225e-05,
      "loss": 0.3398,
      "step": 9614
    },
    {
      "epoch": 4.524705882352941,
      "grad_norm": 0.8305363251172548,
      "learning_rate": 5.867129566857129e-05,
      "loss": 0.543,
      "step": 9615
    },
    {
      "epoch": 4.525176470588235,
      "grad_norm": 0.2791127835303083,
      "learning_rate": 5.8636525241633673e-05,
      "loss": 0.0942,
      "step": 9616
    },
    {
      "epoch": 4.525647058823529,
      "grad_norm": 0.45009708703245055,
      "learning_rate": 5.860176261758499e-05,
      "loss": 0.4805,
      "step": 9617
    },
    {
      "epoch": 4.526117647058824,
      "grad_norm": 0.3949134890808598,
      "learning_rate": 5.8567007799394015e-05,
      "loss": 0.4238,
      "step": 9618
    },
    {
      "epoch": 4.526588235294118,
      "grad_norm": 0.8243009595129891,
      "learning_rate": 5.853226079002906e-05,
      "loss": 0.6328,
      "step": 9619
    },
    {
      "epoch": 4.527058823529412,
      "grad_norm": 0.2876921587716547,
      "learning_rate": 5.849752159245769e-05,
      "loss": 0.1187,
      "step": 9620
    },
    {
      "epoch": 4.527529411764706,
      "grad_norm": 0.6705875367843939,
      "learning_rate": 5.84627902096468e-05,
      "loss": 0.2275,
      "step": 9621
    },
    {
      "epoch": 4.5280000000000005,
      "grad_norm": 0.20845544080564318,
      "learning_rate": 5.842806664456264e-05,
      "loss": 0.0781,
      "step": 9622
    },
    {
      "epoch": 4.528470588235294,
      "grad_norm": 0.250532208117697,
      "learning_rate": 5.839335090017072e-05,
      "loss": 0.083,
      "step": 9623
    },
    {
      "epoch": 4.528941176470588,
      "grad_norm": 0.39062051410298254,
      "learning_rate": 5.835864297943594e-05,
      "loss": 0.4141,
      "step": 9624
    },
    {
      "epoch": 4.529411764705882,
      "grad_norm": 0.5078410492027269,
      "learning_rate": 5.8323942885322605e-05,
      "loss": 0.3223,
      "step": 9625
    },
    {
      "epoch": 4.5298823529411765,
      "grad_norm": 0.35345235195331914,
      "learning_rate": 5.828925062079416e-05,
      "loss": 0.1055,
      "step": 9626
    },
    {
      "epoch": 4.530352941176471,
      "grad_norm": 0.38874765555076735,
      "learning_rate": 5.8254566188813615e-05,
      "loss": 0.5625,
      "step": 9627
    },
    {
      "epoch": 4.530823529411765,
      "grad_norm": 0.2217632759668071,
      "learning_rate": 5.821988959234308e-05,
      "loss": 0.0283,
      "step": 9628
    },
    {
      "epoch": 4.531294117647059,
      "grad_norm": 0.33999258090621937,
      "learning_rate": 5.8185220834344124e-05,
      "loss": 0.1973,
      "step": 9629
    },
    {
      "epoch": 4.5317647058823525,
      "grad_norm": 0.23625148825264594,
      "learning_rate": 5.815055991777773e-05,
      "loss": 0.064,
      "step": 9630
    },
    {
      "epoch": 4.532235294117647,
      "grad_norm": 0.5168981765084752,
      "learning_rate": 5.811590684560398e-05,
      "loss": 0.4453,
      "step": 9631
    },
    {
      "epoch": 4.532705882352941,
      "grad_norm": 0.3652534832513843,
      "learning_rate": 5.808126162078249e-05,
      "loss": 0.0825,
      "step": 9632
    },
    {
      "epoch": 4.533176470588235,
      "grad_norm": 0.38944083742153157,
      "learning_rate": 5.8046624246272146e-05,
      "loss": 0.3711,
      "step": 9633
    },
    {
      "epoch": 4.533647058823529,
      "grad_norm": 0.40238891178211117,
      "learning_rate": 5.801199472503109e-05,
      "loss": 0.4258,
      "step": 9634
    },
    {
      "epoch": 4.534117647058824,
      "grad_norm": 0.487008382233798,
      "learning_rate": 5.797737306001686e-05,
      "loss": 0.5156,
      "step": 9635
    },
    {
      "epoch": 4.534588235294118,
      "grad_norm": 0.5037167529588126,
      "learning_rate": 5.794275925418641e-05,
      "loss": 0.5273,
      "step": 9636
    },
    {
      "epoch": 4.535058823529412,
      "grad_norm": 0.39882470507744205,
      "learning_rate": 5.79081533104958e-05,
      "loss": 0.3535,
      "step": 9637
    },
    {
      "epoch": 4.535529411764706,
      "grad_norm": 0.28242986183687613,
      "learning_rate": 5.787355523190059e-05,
      "loss": 0.0605,
      "step": 9638
    },
    {
      "epoch": 4.536,
      "grad_norm": 0.6135845426158566,
      "learning_rate": 5.783896502135572e-05,
      "loss": 0.9609,
      "step": 9639
    },
    {
      "epoch": 4.536470588235294,
      "grad_norm": 0.3808864866548888,
      "learning_rate": 5.780438268181522e-05,
      "loss": 0.4316,
      "step": 9640
    },
    {
      "epoch": 4.536941176470588,
      "grad_norm": 0.5400743129021608,
      "learning_rate": 5.776980821623264e-05,
      "loss": 0.6367,
      "step": 9641
    },
    {
      "epoch": 4.537411764705882,
      "grad_norm": 0.5844289381591434,
      "learning_rate": 5.7735241627560905e-05,
      "loss": 0.4531,
      "step": 9642
    },
    {
      "epoch": 4.5378823529411765,
      "grad_norm": 0.20458094927490128,
      "learning_rate": 5.770068291875202e-05,
      "loss": 0.0972,
      "step": 9643
    },
    {
      "epoch": 4.538352941176471,
      "grad_norm": 0.2671239313144705,
      "learning_rate": 5.766613209275754e-05,
      "loss": 0.0918,
      "step": 9644
    },
    {
      "epoch": 4.538823529411765,
      "grad_norm": 0.4544514776286042,
      "learning_rate": 5.763158915252833e-05,
      "loss": 0.3594,
      "step": 9645
    },
    {
      "epoch": 4.539294117647059,
      "grad_norm": 0.331713667517555,
      "learning_rate": 5.759705410101443e-05,
      "loss": 0.3125,
      "step": 9646
    },
    {
      "epoch": 4.539764705882353,
      "grad_norm": 0.36934938202000134,
      "learning_rate": 5.756252694116538e-05,
      "loss": 0.3418,
      "step": 9647
    },
    {
      "epoch": 4.540235294117647,
      "grad_norm": 0.5783175523823624,
      "learning_rate": 5.752800767592988e-05,
      "loss": 0.4258,
      "step": 9648
    },
    {
      "epoch": 4.540705882352941,
      "grad_norm": 0.3947264317517217,
      "learning_rate": 5.74934963082561e-05,
      "loss": 0.4023,
      "step": 9649
    },
    {
      "epoch": 4.541176470588235,
      "grad_norm": 0.3563037256507478,
      "learning_rate": 5.745899284109154e-05,
      "loss": 0.207,
      "step": 9650
    },
    {
      "epoch": 4.541647058823529,
      "grad_norm": 0.411663922382121,
      "learning_rate": 5.742449727738285e-05,
      "loss": 0.3574,
      "step": 9651
    },
    {
      "epoch": 4.542117647058824,
      "grad_norm": 0.2885182557534745,
      "learning_rate": 5.739000962007616e-05,
      "loss": 0.2158,
      "step": 9652
    },
    {
      "epoch": 4.542588235294118,
      "grad_norm": 0.5876885281240865,
      "learning_rate": 5.7355529872116957e-05,
      "loss": 0.5156,
      "step": 9653
    },
    {
      "epoch": 4.543058823529412,
      "grad_norm": 0.512005080214034,
      "learning_rate": 5.732105803644986e-05,
      "loss": 0.457,
      "step": 9654
    },
    {
      "epoch": 4.543529411764706,
      "grad_norm": 0.24129572091726004,
      "learning_rate": 5.7286594116019e-05,
      "loss": 0.2158,
      "step": 9655
    },
    {
      "epoch": 4.5440000000000005,
      "grad_norm": 0.4208799861441959,
      "learning_rate": 5.725213811376782e-05,
      "loss": 0.3047,
      "step": 9656
    },
    {
      "epoch": 4.544470588235294,
      "grad_norm": 0.4452303151197179,
      "learning_rate": 5.7217690032638916e-05,
      "loss": 0.2852,
      "step": 9657
    },
    {
      "epoch": 4.544941176470588,
      "grad_norm": 0.301978382347478,
      "learning_rate": 5.718324987557438e-05,
      "loss": 0.3945,
      "step": 9658
    },
    {
      "epoch": 4.545411764705882,
      "grad_norm": 0.3459684963646186,
      "learning_rate": 5.714881764551561e-05,
      "loss": 0.1318,
      "step": 9659
    },
    {
      "epoch": 4.5458823529411765,
      "grad_norm": 0.42616513617134144,
      "learning_rate": 5.7114393345403194e-05,
      "loss": 0.373,
      "step": 9660
    },
    {
      "epoch": 4.546352941176471,
      "grad_norm": 0.43413735009648285,
      "learning_rate": 5.707997697817719e-05,
      "loss": 0.4863,
      "step": 9661
    },
    {
      "epoch": 4.546823529411765,
      "grad_norm": 0.286991859228179,
      "learning_rate": 5.704556854677698e-05,
      "loss": 0.2383,
      "step": 9662
    },
    {
      "epoch": 4.547294117647059,
      "grad_norm": 0.40478327049794605,
      "learning_rate": 5.701116805414104e-05,
      "loss": 0.3828,
      "step": 9663
    },
    {
      "epoch": 4.5477647058823525,
      "grad_norm": 0.38352149371337335,
      "learning_rate": 5.6976775503207574e-05,
      "loss": 0.166,
      "step": 9664
    },
    {
      "epoch": 4.548235294117647,
      "grad_norm": 0.41842085015104347,
      "learning_rate": 5.6942390896913686e-05,
      "loss": 0.4551,
      "step": 9665
    },
    {
      "epoch": 4.548705882352941,
      "grad_norm": 0.28379253118286707,
      "learning_rate": 5.690801423819608e-05,
      "loss": 0.1396,
      "step": 9666
    },
    {
      "epoch": 4.549176470588235,
      "grad_norm": 0.4455334444879779,
      "learning_rate": 5.687364552999071e-05,
      "loss": 0.4062,
      "step": 9667
    },
    {
      "epoch": 4.549647058823529,
      "grad_norm": 0.37338967680197277,
      "learning_rate": 5.683928477523275e-05,
      "loss": 0.2969,
      "step": 9668
    },
    {
      "epoch": 4.550117647058824,
      "grad_norm": 0.4635704442820427,
      "learning_rate": 5.680493197685681e-05,
      "loss": 0.5703,
      "step": 9669
    },
    {
      "epoch": 4.550588235294118,
      "grad_norm": 0.39641400212997485,
      "learning_rate": 5.677058713779687e-05,
      "loss": 0.1387,
      "step": 9670
    },
    {
      "epoch": 4.551058823529412,
      "grad_norm": 0.5533730081061994,
      "learning_rate": 5.6736250260986e-05,
      "loss": 0.6523,
      "step": 9671
    },
    {
      "epoch": 4.551529411764706,
      "grad_norm": 0.7525388044901183,
      "learning_rate": 5.6701921349356826e-05,
      "loss": 0.5312,
      "step": 9672
    },
    {
      "epoch": 4.552,
      "grad_norm": 0.5580673458778758,
      "learning_rate": 5.666760040584126e-05,
      "loss": 0.4297,
      "step": 9673
    },
    {
      "epoch": 4.552470588235294,
      "grad_norm": 0.37875038511040215,
      "learning_rate": 5.663328743337034e-05,
      "loss": 0.167,
      "step": 9674
    },
    {
      "epoch": 4.552941176470588,
      "grad_norm": 0.44334264035224846,
      "learning_rate": 5.659898243487463e-05,
      "loss": 0.2188,
      "step": 9675
    },
    {
      "epoch": 4.553411764705882,
      "grad_norm": 0.22404862506225798,
      "learning_rate": 5.6564685413284014e-05,
      "loss": 0.1797,
      "step": 9676
    },
    {
      "epoch": 4.5538823529411765,
      "grad_norm": 0.4375489794583197,
      "learning_rate": 5.6530396371527515e-05,
      "loss": 0.1318,
      "step": 9677
    },
    {
      "epoch": 4.554352941176471,
      "grad_norm": 0.7025894260011712,
      "learning_rate": 5.649611531253366e-05,
      "loss": 0.4375,
      "step": 9678
    },
    {
      "epoch": 4.554823529411765,
      "grad_norm": 0.33872577045222085,
      "learning_rate": 5.646184223923017e-05,
      "loss": 0.291,
      "step": 9679
    },
    {
      "epoch": 4.555294117647059,
      "grad_norm": 0.5384133258676036,
      "learning_rate": 5.642757715454413e-05,
      "loss": 0.5586,
      "step": 9680
    },
    {
      "epoch": 4.555764705882353,
      "grad_norm": 0.2832856054440504,
      "learning_rate": 5.639332006140203e-05,
      "loss": 0.2676,
      "step": 9681
    },
    {
      "epoch": 4.556235294117647,
      "grad_norm": 0.4931518685877791,
      "learning_rate": 5.635907096272949e-05,
      "loss": 0.3613,
      "step": 9682
    },
    {
      "epoch": 4.556705882352941,
      "grad_norm": 0.4125433330108535,
      "learning_rate": 5.632482986145159e-05,
      "loss": 0.4727,
      "step": 9683
    },
    {
      "epoch": 4.557176470588235,
      "grad_norm": 0.5191138000105854,
      "learning_rate": 5.6290596760492745e-05,
      "loss": 0.4766,
      "step": 9684
    },
    {
      "epoch": 4.557647058823529,
      "grad_norm": 0.20740687662571983,
      "learning_rate": 5.625637166277654e-05,
      "loss": 0.125,
      "step": 9685
    },
    {
      "epoch": 4.558117647058824,
      "grad_norm": 0.3668633201121336,
      "learning_rate": 5.622215457122598e-05,
      "loss": 0.3047,
      "step": 9686
    },
    {
      "epoch": 4.558588235294118,
      "grad_norm": 0.19878929584104962,
      "learning_rate": 5.618794548876346e-05,
      "loss": 0.0623,
      "step": 9687
    },
    {
      "epoch": 4.559058823529412,
      "grad_norm": 0.31452085508860456,
      "learning_rate": 5.6153744418310505e-05,
      "loss": 0.3906,
      "step": 9688
    },
    {
      "epoch": 4.559529411764705,
      "grad_norm": 0.6453020747102983,
      "learning_rate": 5.611955136278808e-05,
      "loss": 0.7383,
      "step": 9689
    },
    {
      "epoch": 4.5600000000000005,
      "grad_norm": 0.38791942037849164,
      "learning_rate": 5.608536632511649e-05,
      "loss": 0.2793,
      "step": 9690
    },
    {
      "epoch": 4.560470588235294,
      "grad_norm": 0.3518237396326775,
      "learning_rate": 5.6051189308215225e-05,
      "loss": 0.3789,
      "step": 9691
    },
    {
      "epoch": 4.560941176470588,
      "grad_norm": 0.33614532921108886,
      "learning_rate": 5.6017020315003195e-05,
      "loss": 0.1309,
      "step": 9692
    },
    {
      "epoch": 4.561411764705882,
      "grad_norm": 0.3323114211606641,
      "learning_rate": 5.598285934839868e-05,
      "loss": 0.0835,
      "step": 9693
    },
    {
      "epoch": 4.5618823529411765,
      "grad_norm": 0.4405441652584119,
      "learning_rate": 5.594870641131907e-05,
      "loss": 0.2471,
      "step": 9694
    },
    {
      "epoch": 4.562352941176471,
      "grad_norm": 0.2582037576818558,
      "learning_rate": 5.591456150668126e-05,
      "loss": 0.0972,
      "step": 9695
    },
    {
      "epoch": 4.562823529411765,
      "grad_norm": 0.48316369163416495,
      "learning_rate": 5.588042463740143e-05,
      "loss": 0.3301,
      "step": 9696
    },
    {
      "epoch": 4.563294117647059,
      "grad_norm": 0.41216971538740976,
      "learning_rate": 5.584629580639494e-05,
      "loss": 0.165,
      "step": 9697
    },
    {
      "epoch": 4.5637647058823525,
      "grad_norm": 0.42801742114268426,
      "learning_rate": 5.581217501657666e-05,
      "loss": 0.4375,
      "step": 9698
    },
    {
      "epoch": 4.564235294117647,
      "grad_norm": 0.5054368703896726,
      "learning_rate": 5.577806227086057e-05,
      "loss": 0.7148,
      "step": 9699
    },
    {
      "epoch": 4.564705882352941,
      "grad_norm": 0.224885559229696,
      "learning_rate": 5.57439575721601e-05,
      "loss": 0.043,
      "step": 9700
    },
    {
      "epoch": 4.565176470588235,
      "grad_norm": 0.3458101520117074,
      "learning_rate": 5.570986092338806e-05,
      "loss": 0.3594,
      "step": 9701
    },
    {
      "epoch": 4.565647058823529,
      "grad_norm": 0.6067283544266793,
      "learning_rate": 5.5675772327456313e-05,
      "loss": 0.9336,
      "step": 9702
    },
    {
      "epoch": 4.566117647058824,
      "grad_norm": 0.46960364436727403,
      "learning_rate": 5.5641691787276286e-05,
      "loss": 0.3633,
      "step": 9703
    },
    {
      "epoch": 4.566588235294118,
      "grad_norm": 0.4433445665692513,
      "learning_rate": 5.560761930575864e-05,
      "loss": 0.4238,
      "step": 9704
    },
    {
      "epoch": 4.567058823529412,
      "grad_norm": 0.35863498292404905,
      "learning_rate": 5.557355488581325e-05,
      "loss": 0.3711,
      "step": 9705
    },
    {
      "epoch": 4.567529411764706,
      "grad_norm": 0.5821149977018073,
      "learning_rate": 5.553949853034943e-05,
      "loss": 0.3691,
      "step": 9706
    },
    {
      "epoch": 4.568,
      "grad_norm": 0.29625532902478424,
      "learning_rate": 5.550545024227582e-05,
      "loss": 0.0918,
      "step": 9707
    },
    {
      "epoch": 4.568470588235294,
      "grad_norm": 0.28100965403593625,
      "learning_rate": 5.5471410024500156e-05,
      "loss": 0.1172,
      "step": 9708
    },
    {
      "epoch": 4.568941176470588,
      "grad_norm": 0.26619723139930573,
      "learning_rate": 5.5437377879929826e-05,
      "loss": 0.3125,
      "step": 9709
    },
    {
      "epoch": 4.569411764705882,
      "grad_norm": 0.5215342327756389,
      "learning_rate": 5.5403353811471194e-05,
      "loss": 0.2715,
      "step": 9710
    },
    {
      "epoch": 4.5698823529411765,
      "grad_norm": 0.20552148487820993,
      "learning_rate": 5.5369337822030154e-05,
      "loss": 0.0928,
      "step": 9711
    },
    {
      "epoch": 4.570352941176471,
      "grad_norm": 0.3442511354700939,
      "learning_rate": 5.533532991451187e-05,
      "loss": 0.084,
      "step": 9712
    },
    {
      "epoch": 4.570823529411765,
      "grad_norm": 0.3725239562989111,
      "learning_rate": 5.53013300918207e-05,
      "loss": 0.1396,
      "step": 9713
    },
    {
      "epoch": 4.571294117647059,
      "grad_norm": 0.4804349505945273,
      "learning_rate": 5.526733835686041e-05,
      "loss": 0.3574,
      "step": 9714
    },
    {
      "epoch": 4.571764705882353,
      "grad_norm": 0.3055548830116313,
      "learning_rate": 5.523335471253416e-05,
      "loss": 0.1211,
      "step": 9715
    },
    {
      "epoch": 4.572235294117647,
      "grad_norm": 0.31975254707433487,
      "learning_rate": 5.519937916174419e-05,
      "loss": 0.1143,
      "step": 9716
    },
    {
      "epoch": 4.572705882352941,
      "grad_norm": 0.34962980042080427,
      "learning_rate": 5.516541170739223e-05,
      "loss": 0.1157,
      "step": 9717
    },
    {
      "epoch": 4.573176470588235,
      "grad_norm": 0.4809394306784108,
      "learning_rate": 5.5131452352379314e-05,
      "loss": 0.6016,
      "step": 9718
    },
    {
      "epoch": 4.573647058823529,
      "grad_norm": 0.4102206115434395,
      "learning_rate": 5.509750109960566e-05,
      "loss": 0.416,
      "step": 9719
    },
    {
      "epoch": 4.574117647058824,
      "grad_norm": 0.39600983486151226,
      "learning_rate": 5.5063557951970885e-05,
      "loss": 0.1504,
      "step": 9720
    },
    {
      "epoch": 4.574588235294118,
      "grad_norm": 0.41470804069209166,
      "learning_rate": 5.502962291237399e-05,
      "loss": 0.2295,
      "step": 9721
    },
    {
      "epoch": 4.575058823529412,
      "grad_norm": 0.4519044934718447,
      "learning_rate": 5.4995695983713076e-05,
      "loss": 0.4766,
      "step": 9722
    },
    {
      "epoch": 4.575529411764705,
      "grad_norm": 0.39004258984049994,
      "learning_rate": 5.496177716888571e-05,
      "loss": 0.4551,
      "step": 9723
    },
    {
      "epoch": 4.576,
      "grad_norm": 0.18514505073485474,
      "learning_rate": 5.4927866470788786e-05,
      "loss": 0.0586,
      "step": 9724
    },
    {
      "epoch": 4.576470588235294,
      "grad_norm": 0.42213619875388164,
      "learning_rate": 5.489396389231834e-05,
      "loss": 0.3379,
      "step": 9725
    },
    {
      "epoch": 4.576941176470588,
      "grad_norm": 0.5411444520839763,
      "learning_rate": 5.4860069436369916e-05,
      "loss": 0.3301,
      "step": 9726
    },
    {
      "epoch": 4.577411764705882,
      "grad_norm": 0.42239253916666714,
      "learning_rate": 5.4826183105838184e-05,
      "loss": 0.3379,
      "step": 9727
    },
    {
      "epoch": 4.5778823529411765,
      "grad_norm": 0.2861489352083796,
      "learning_rate": 5.4792304903617226e-05,
      "loss": 0.2432,
      "step": 9728
    },
    {
      "epoch": 4.578352941176471,
      "grad_norm": 0.5224414357520559,
      "learning_rate": 5.4758434832600475e-05,
      "loss": 0.3691,
      "step": 9729
    },
    {
      "epoch": 4.578823529411765,
      "grad_norm": 0.45097031541149324,
      "learning_rate": 5.47245728956805e-05,
      "loss": 0.0315,
      "step": 9730
    },
    {
      "epoch": 4.579294117647059,
      "grad_norm": 0.3977640587094502,
      "learning_rate": 5.469071909574933e-05,
      "loss": 0.3145,
      "step": 9731
    },
    {
      "epoch": 4.5797647058823525,
      "grad_norm": 0.37578147017766667,
      "learning_rate": 5.465687343569828e-05,
      "loss": 0.2373,
      "step": 9732
    },
    {
      "epoch": 4.580235294117647,
      "grad_norm": 0.27724018931451433,
      "learning_rate": 5.462303591841783e-05,
      "loss": 0.0894,
      "step": 9733
    },
    {
      "epoch": 4.580705882352941,
      "grad_norm": 0.5368739955376366,
      "learning_rate": 5.458920654679795e-05,
      "loss": 0.5391,
      "step": 9734
    },
    {
      "epoch": 4.581176470588235,
      "grad_norm": 0.6412864436055337,
      "learning_rate": 5.4555385323727876e-05,
      "loss": 0.6133,
      "step": 9735
    },
    {
      "epoch": 4.581647058823529,
      "grad_norm": 0.35967896843586233,
      "learning_rate": 5.4521572252095985e-05,
      "loss": 0.3652,
      "step": 9736
    },
    {
      "epoch": 4.582117647058824,
      "grad_norm": 0.4024385132628453,
      "learning_rate": 5.4487767334790145e-05,
      "loss": 0.1943,
      "step": 9737
    },
    {
      "epoch": 4.582588235294118,
      "grad_norm": 0.30893478234531974,
      "learning_rate": 5.445397057469751e-05,
      "loss": 0.2412,
      "step": 9738
    },
    {
      "epoch": 4.583058823529412,
      "grad_norm": 0.4912417191999147,
      "learning_rate": 5.4420181974704386e-05,
      "loss": 0.4434,
      "step": 9739
    },
    {
      "epoch": 4.583529411764706,
      "grad_norm": 0.6133941429999756,
      "learning_rate": 5.4386401537696536e-05,
      "loss": 0.5625,
      "step": 9740
    },
    {
      "epoch": 4.584,
      "grad_norm": 0.5415753064779403,
      "learning_rate": 5.435262926655903e-05,
      "loss": 0.5234,
      "step": 9741
    },
    {
      "epoch": 4.584470588235294,
      "grad_norm": 0.4470184936304554,
      "learning_rate": 5.43188651641761e-05,
      "loss": 0.5977,
      "step": 9742
    },
    {
      "epoch": 4.584941176470588,
      "grad_norm": 0.4698783154291033,
      "learning_rate": 5.428510923343136e-05,
      "loss": 0.4004,
      "step": 9743
    },
    {
      "epoch": 4.585411764705882,
      "grad_norm": 0.3701138622555473,
      "learning_rate": 5.425136147720785e-05,
      "loss": 0.332,
      "step": 9744
    },
    {
      "epoch": 4.5858823529411765,
      "grad_norm": 0.5633061255301418,
      "learning_rate": 5.421762189838765e-05,
      "loss": 0.2598,
      "step": 9745
    },
    {
      "epoch": 4.586352941176471,
      "grad_norm": 0.5167311950587624,
      "learning_rate": 5.4183890499852364e-05,
      "loss": 0.6445,
      "step": 9746
    },
    {
      "epoch": 4.586823529411765,
      "grad_norm": 0.3572098097099825,
      "learning_rate": 5.4150167284482844e-05,
      "loss": 0.1377,
      "step": 9747
    },
    {
      "epoch": 4.587294117647059,
      "grad_norm": 0.35056223870901665,
      "learning_rate": 5.4116452255159126e-05,
      "loss": 0.1396,
      "step": 9748
    },
    {
      "epoch": 4.587764705882353,
      "grad_norm": 0.1246196175572148,
      "learning_rate": 5.408274541476075e-05,
      "loss": 0.0247,
      "step": 9749
    },
    {
      "epoch": 4.588235294117647,
      "grad_norm": 0.5015711954084626,
      "learning_rate": 5.4049046766166335e-05,
      "loss": 0.6641,
      "step": 9750
    },
    {
      "epoch": 4.588705882352941,
      "grad_norm": 0.5245632210373585,
      "learning_rate": 5.401535631225396e-05,
      "loss": 0.3848,
      "step": 9751
    },
    {
      "epoch": 4.589176470588235,
      "grad_norm": 0.4168088167100816,
      "learning_rate": 5.398167405590096e-05,
      "loss": 0.25,
      "step": 9752
    },
    {
      "epoch": 4.589647058823529,
      "grad_norm": 0.47372684965827333,
      "learning_rate": 5.3947999999983967e-05,
      "loss": 0.2002,
      "step": 9753
    },
    {
      "epoch": 4.590117647058824,
      "grad_norm": 0.4340126602678218,
      "learning_rate": 5.3914334147378894e-05,
      "loss": 0.4395,
      "step": 9754
    },
    {
      "epoch": 4.590588235294118,
      "grad_norm": 0.8370685897722102,
      "learning_rate": 5.388067650096106e-05,
      "loss": 0.5625,
      "step": 9755
    },
    {
      "epoch": 4.591058823529412,
      "grad_norm": 0.22703682814129275,
      "learning_rate": 5.384702706360483e-05,
      "loss": 0.1133,
      "step": 9756
    },
    {
      "epoch": 4.591529411764705,
      "grad_norm": 0.4372820055866906,
      "learning_rate": 5.3813385838184185e-05,
      "loss": 0.4961,
      "step": 9757
    },
    {
      "epoch": 4.592,
      "grad_norm": 0.20938440333881245,
      "learning_rate": 5.3779752827572126e-05,
      "loss": 0.0586,
      "step": 9758
    },
    {
      "epoch": 4.592470588235294,
      "grad_norm": 0.3227864834005108,
      "learning_rate": 5.374612803464115e-05,
      "loss": 0.208,
      "step": 9759
    },
    {
      "epoch": 4.592941176470588,
      "grad_norm": 0.4036762970295732,
      "learning_rate": 5.3712511462262994e-05,
      "loss": 0.168,
      "step": 9760
    },
    {
      "epoch": 4.593411764705882,
      "grad_norm": 0.3336069995229597,
      "learning_rate": 5.36789031133086e-05,
      "loss": 0.1172,
      "step": 9761
    },
    {
      "epoch": 4.5938823529411765,
      "grad_norm": 0.3711879282809874,
      "learning_rate": 5.364530299064834e-05,
      "loss": 0.5,
      "step": 9762
    },
    {
      "epoch": 4.594352941176471,
      "grad_norm": 0.47850038463421457,
      "learning_rate": 5.361171109715186e-05,
      "loss": 0.2598,
      "step": 9763
    },
    {
      "epoch": 4.594823529411765,
      "grad_norm": 0.21431050155059272,
      "learning_rate": 5.357812743568799e-05,
      "loss": 0.1338,
      "step": 9764
    },
    {
      "epoch": 4.595294117647059,
      "grad_norm": 0.392803599653901,
      "learning_rate": 5.354455200912498e-05,
      "loss": 0.1299,
      "step": 9765
    },
    {
      "epoch": 4.5957647058823525,
      "grad_norm": 0.317445555594523,
      "learning_rate": 5.35109848203304e-05,
      "loss": 0.1396,
      "step": 9766
    },
    {
      "epoch": 4.596235294117647,
      "grad_norm": 0.44700823433055054,
      "learning_rate": 5.3477425872170926e-05,
      "loss": 0.6211,
      "step": 9767
    },
    {
      "epoch": 4.596705882352941,
      "grad_norm": 0.5653936509103842,
      "learning_rate": 5.3443875167512736e-05,
      "loss": 0.6562,
      "step": 9768
    },
    {
      "epoch": 4.597176470588235,
      "grad_norm": 0.39806729839937294,
      "learning_rate": 5.341033270922126e-05,
      "loss": 0.3926,
      "step": 9769
    },
    {
      "epoch": 4.597647058823529,
      "grad_norm": 0.24498174945399806,
      "learning_rate": 5.337679850016109e-05,
      "loss": 0.0562,
      "step": 9770
    },
    {
      "epoch": 4.598117647058824,
      "grad_norm": 0.5797861425639192,
      "learning_rate": 5.3343272543196264e-05,
      "loss": 0.2988,
      "step": 9771
    },
    {
      "epoch": 4.598588235294118,
      "grad_norm": 0.47034361750531695,
      "learning_rate": 5.33097548411901e-05,
      "loss": 0.5469,
      "step": 9772
    },
    {
      "epoch": 4.599058823529412,
      "grad_norm": 0.3365452013002257,
      "learning_rate": 5.3276245397005094e-05,
      "loss": 0.418,
      "step": 9773
    },
    {
      "epoch": 4.599529411764706,
      "grad_norm": 0.4120963353346327,
      "learning_rate": 5.324274421350316e-05,
      "loss": 0.1895,
      "step": 9774
    },
    {
      "epoch": 4.6,
      "grad_norm": 0.1334058488553875,
      "learning_rate": 5.3209251293545514e-05,
      "loss": 0.0396,
      "step": 9775
    },
    {
      "epoch": 4.600470588235294,
      "grad_norm": 0.40695858616228914,
      "learning_rate": 5.317576663999251e-05,
      "loss": 0.1689,
      "step": 9776
    },
    {
      "epoch": 4.600941176470588,
      "grad_norm": 0.3343105195891496,
      "learning_rate": 5.314229025570401e-05,
      "loss": 0.1719,
      "step": 9777
    },
    {
      "epoch": 4.601411764705882,
      "grad_norm": 0.3247611619226119,
      "learning_rate": 5.3108822143538944e-05,
      "loss": 0.2314,
      "step": 9778
    },
    {
      "epoch": 4.6018823529411765,
      "grad_norm": 0.49543072327022586,
      "learning_rate": 5.307536230635571e-05,
      "loss": 0.6719,
      "step": 9779
    },
    {
      "epoch": 4.602352941176471,
      "grad_norm": 0.3695074007430424,
      "learning_rate": 5.304191074701201e-05,
      "loss": 0.3711,
      "step": 9780
    },
    {
      "epoch": 4.602823529411765,
      "grad_norm": 0.5814435909026452,
      "learning_rate": 5.3008467468364675e-05,
      "loss": 0.3711,
      "step": 9781
    },
    {
      "epoch": 4.603294117647059,
      "grad_norm": 0.21629827224601636,
      "learning_rate": 5.297503247326993e-05,
      "loss": 0.1436,
      "step": 9782
    },
    {
      "epoch": 4.603764705882353,
      "grad_norm": 0.3239994639821513,
      "learning_rate": 5.294160576458339e-05,
      "loss": 0.2373,
      "step": 9783
    },
    {
      "epoch": 4.604235294117647,
      "grad_norm": 0.31555053839398023,
      "learning_rate": 5.290818734515972e-05,
      "loss": 0.2637,
      "step": 9784
    },
    {
      "epoch": 4.604705882352941,
      "grad_norm": 0.47112912694394554,
      "learning_rate": 5.287477721785309e-05,
      "loss": 0.6602,
      "step": 9785
    },
    {
      "epoch": 4.605176470588235,
      "grad_norm": 0.1630083234600993,
      "learning_rate": 5.284137538551693e-05,
      "loss": 0.0297,
      "step": 9786
    },
    {
      "epoch": 4.605647058823529,
      "grad_norm": 0.5556621913330794,
      "learning_rate": 5.2807981851003825e-05,
      "loss": 0.5312,
      "step": 9787
    },
    {
      "epoch": 4.606117647058824,
      "grad_norm": 0.4768441144970694,
      "learning_rate": 5.2774596617165793e-05,
      "loss": 0.6172,
      "step": 9788
    },
    {
      "epoch": 4.606588235294118,
      "grad_norm": 0.32786035882555586,
      "learning_rate": 5.2741219686854146e-05,
      "loss": 0.085,
      "step": 9789
    },
    {
      "epoch": 4.607058823529412,
      "grad_norm": 0.33259711093244876,
      "learning_rate": 5.270785106291934e-05,
      "loss": 0.2793,
      "step": 9790
    },
    {
      "epoch": 4.607529411764705,
      "grad_norm": 0.4708364896506531,
      "learning_rate": 5.2674490748211265e-05,
      "loss": 0.418,
      "step": 9791
    },
    {
      "epoch": 4.608,
      "grad_norm": 0.3493461030418327,
      "learning_rate": 5.2641138745579126e-05,
      "loss": 0.1089,
      "step": 9792
    },
    {
      "epoch": 4.608470588235294,
      "grad_norm": 0.7777910589535498,
      "learning_rate": 5.2607795057871216e-05,
      "loss": 0.6406,
      "step": 9793
    },
    {
      "epoch": 4.608941176470588,
      "grad_norm": 0.3288740443391295,
      "learning_rate": 5.257445968793531e-05,
      "loss": 0.3809,
      "step": 9794
    },
    {
      "epoch": 4.609411764705882,
      "grad_norm": 0.5895352111713619,
      "learning_rate": 5.254113263861848e-05,
      "loss": 0.3281,
      "step": 9795
    },
    {
      "epoch": 4.6098823529411765,
      "grad_norm": 0.07858883725014154,
      "learning_rate": 5.250781391276684e-05,
      "loss": 0.0156,
      "step": 9796
    },
    {
      "epoch": 4.610352941176471,
      "grad_norm": 0.36565227022718133,
      "learning_rate": 5.24745035132262e-05,
      "loss": 0.3281,
      "step": 9797
    },
    {
      "epoch": 4.610823529411765,
      "grad_norm": 0.3782426335219559,
      "learning_rate": 5.2441201442841264e-05,
      "loss": 0.4609,
      "step": 9798
    },
    {
      "epoch": 4.611294117647059,
      "grad_norm": 0.3664213145392853,
      "learning_rate": 5.240790770445625e-05,
      "loss": 0.2217,
      "step": 9799
    },
    {
      "epoch": 4.6117647058823525,
      "grad_norm": 0.27069586942588036,
      "learning_rate": 5.237462230091467e-05,
      "loss": 0.1748,
      "step": 9800
    },
    {
      "epoch": 4.612235294117647,
      "grad_norm": 0.31975216454310434,
      "learning_rate": 5.234134523505912e-05,
      "loss": 0.1396,
      "step": 9801
    },
    {
      "epoch": 4.612705882352941,
      "grad_norm": 0.37363706105910677,
      "learning_rate": 5.230807650973172e-05,
      "loss": 0.4629,
      "step": 9802
    },
    {
      "epoch": 4.613176470588235,
      "grad_norm": 0.187901081013872,
      "learning_rate": 5.227481612777379e-05,
      "loss": 0.1216,
      "step": 9803
    },
    {
      "epoch": 4.613647058823529,
      "grad_norm": 0.24009570017643644,
      "learning_rate": 5.224156409202587e-05,
      "loss": 0.084,
      "step": 9804
    },
    {
      "epoch": 4.614117647058824,
      "grad_norm": 0.3760651335348669,
      "learning_rate": 5.2208320405327865e-05,
      "loss": 0.2832,
      "step": 9805
    },
    {
      "epoch": 4.614588235294118,
      "grad_norm": 0.5628205066509706,
      "learning_rate": 5.2175085070519024e-05,
      "loss": 0.4922,
      "step": 9806
    },
    {
      "epoch": 4.615058823529412,
      "grad_norm": 0.3861142566205307,
      "learning_rate": 5.2141858090437696e-05,
      "loss": 0.2148,
      "step": 9807
    },
    {
      "epoch": 4.615529411764706,
      "grad_norm": 0.4058733817999431,
      "learning_rate": 5.210863946792172e-05,
      "loss": 0.2656,
      "step": 9808
    },
    {
      "epoch": 4.616,
      "grad_norm": 0.5676881965208151,
      "learning_rate": 5.207542920580805e-05,
      "loss": 0.2617,
      "step": 9809
    },
    {
      "epoch": 4.616470588235294,
      "grad_norm": 0.2852270750963356,
      "learning_rate": 5.204222730693303e-05,
      "loss": 0.0718,
      "step": 9810
    },
    {
      "epoch": 4.616941176470588,
      "grad_norm": 0.6218708299485554,
      "learning_rate": 5.2009033774132345e-05,
      "loss": 0.5117,
      "step": 9811
    },
    {
      "epoch": 4.617411764705882,
      "grad_norm": 0.40572523754405515,
      "learning_rate": 5.197584861024076e-05,
      "loss": 0.1543,
      "step": 9812
    },
    {
      "epoch": 4.6178823529411765,
      "grad_norm": 0.455038007841748,
      "learning_rate": 5.194267181809251e-05,
      "loss": 0.5625,
      "step": 9813
    },
    {
      "epoch": 4.618352941176471,
      "grad_norm": 0.2623811370705975,
      "learning_rate": 5.1909503400521104e-05,
      "loss": 0.0613,
      "step": 9814
    },
    {
      "epoch": 4.618823529411765,
      "grad_norm": 0.2257645534961828,
      "learning_rate": 5.187634336035918e-05,
      "loss": 0.2432,
      "step": 9815
    },
    {
      "epoch": 4.619294117647058,
      "grad_norm": 0.4093956689138033,
      "learning_rate": 5.184319170043884e-05,
      "loss": 0.3398,
      "step": 9816
    },
    {
      "epoch": 4.619764705882353,
      "grad_norm": 1.2300252343351292,
      "learning_rate": 5.1810048423591424e-05,
      "loss": 0.5156,
      "step": 9817
    },
    {
      "epoch": 4.620235294117647,
      "grad_norm": 0.4010269785817353,
      "learning_rate": 5.177691353264746e-05,
      "loss": 0.4629,
      "step": 9818
    },
    {
      "epoch": 4.620705882352941,
      "grad_norm": 0.33311889673293466,
      "learning_rate": 5.174378703043684e-05,
      "loss": 0.126,
      "step": 9819
    },
    {
      "epoch": 4.621176470588235,
      "grad_norm": 0.36265208875452715,
      "learning_rate": 5.171066891978881e-05,
      "loss": 0.4395,
      "step": 9820
    },
    {
      "epoch": 4.621647058823529,
      "grad_norm": 0.430542412629088,
      "learning_rate": 5.167755920353171e-05,
      "loss": 0.5156,
      "step": 9821
    },
    {
      "epoch": 4.622117647058824,
      "grad_norm": 0.46428274656623275,
      "learning_rate": 5.164445788449334e-05,
      "loss": 0.2168,
      "step": 9822
    },
    {
      "epoch": 4.622588235294118,
      "grad_norm": 0.40432722684692024,
      "learning_rate": 5.161136496550072e-05,
      "loss": 0.6133,
      "step": 9823
    },
    {
      "epoch": 4.623058823529412,
      "grad_norm": 0.27001176908974583,
      "learning_rate": 5.15782804493801e-05,
      "loss": 0.0938,
      "step": 9824
    },
    {
      "epoch": 4.623529411764705,
      "grad_norm": 0.3622559518450012,
      "learning_rate": 5.1545204338957066e-05,
      "loss": 0.2285,
      "step": 9825
    },
    {
      "epoch": 4.624,
      "grad_norm": 0.46694890036762576,
      "learning_rate": 5.151213663705655e-05,
      "loss": 0.1348,
      "step": 9826
    },
    {
      "epoch": 4.624470588235294,
      "grad_norm": 0.3376982754308565,
      "learning_rate": 5.14790773465026e-05,
      "loss": 0.1982,
      "step": 9827
    },
    {
      "epoch": 4.624941176470588,
      "grad_norm": 0.4137338046317103,
      "learning_rate": 5.144602647011874e-05,
      "loss": 0.5234,
      "step": 9828
    },
    {
      "epoch": 4.625411764705882,
      "grad_norm": 0.4176459863660065,
      "learning_rate": 5.141298401072758e-05,
      "loss": 0.4336,
      "step": 9829
    },
    {
      "epoch": 4.625882352941177,
      "grad_norm": 0.43830322004852085,
      "learning_rate": 5.137994997115115e-05,
      "loss": 0.3574,
      "step": 9830
    },
    {
      "epoch": 4.626352941176471,
      "grad_norm": 0.3121854046758398,
      "learning_rate": 5.134692435421077e-05,
      "loss": 0.3164,
      "step": 9831
    },
    {
      "epoch": 4.626823529411765,
      "grad_norm": 0.4830147239442297,
      "learning_rate": 5.1313907162726883e-05,
      "loss": 0.4082,
      "step": 9832
    },
    {
      "epoch": 4.627294117647059,
      "grad_norm": 0.3916090198527471,
      "learning_rate": 5.128089839951937e-05,
      "loss": 0.3613,
      "step": 9833
    },
    {
      "epoch": 4.627764705882353,
      "grad_norm": 0.3764950640211214,
      "learning_rate": 5.124789806740742e-05,
      "loss": 0.1143,
      "step": 9834
    },
    {
      "epoch": 4.628235294117647,
      "grad_norm": 0.38854025000765235,
      "learning_rate": 5.12149061692093e-05,
      "loss": 0.1309,
      "step": 9835
    },
    {
      "epoch": 4.628705882352941,
      "grad_norm": 0.36147785083003087,
      "learning_rate": 5.118192270774271e-05,
      "loss": 0.5195,
      "step": 9836
    },
    {
      "epoch": 4.629176470588235,
      "grad_norm": 0.47935122556608556,
      "learning_rate": 5.114894768582468e-05,
      "loss": 0.4219,
      "step": 9837
    },
    {
      "epoch": 4.6296470588235294,
      "grad_norm": 0.3933548615663773,
      "learning_rate": 5.111598110627133e-05,
      "loss": 0.4062,
      "step": 9838
    },
    {
      "epoch": 4.630117647058824,
      "grad_norm": 0.5987398963332885,
      "learning_rate": 5.108302297189822e-05,
      "loss": 0.5391,
      "step": 9839
    },
    {
      "epoch": 4.630588235294118,
      "grad_norm": 0.40617887736137015,
      "learning_rate": 5.105007328552019e-05,
      "loss": 0.2871,
      "step": 9840
    },
    {
      "epoch": 4.631058823529412,
      "grad_norm": 0.23804195039586915,
      "learning_rate": 5.101713204995114e-05,
      "loss": 0.0864,
      "step": 9841
    },
    {
      "epoch": 4.631529411764706,
      "grad_norm": 0.264999662566438,
      "learning_rate": 5.0984199268004615e-05,
      "loss": 0.0903,
      "step": 9842
    },
    {
      "epoch": 4.632,
      "grad_norm": 0.21970705923297343,
      "learning_rate": 5.095127494249309e-05,
      "loss": 0.1216,
      "step": 9843
    },
    {
      "epoch": 4.632470588235294,
      "grad_norm": 0.4482877997919298,
      "learning_rate": 5.091835907622854e-05,
      "loss": 0.373,
      "step": 9844
    },
    {
      "epoch": 4.632941176470588,
      "grad_norm": 0.28798289417240647,
      "learning_rate": 5.0885451672022156e-05,
      "loss": 0.0679,
      "step": 9845
    },
    {
      "epoch": 4.633411764705882,
      "grad_norm": 0.18514394191457037,
      "learning_rate": 5.0852552732684296e-05,
      "loss": 0.0576,
      "step": 9846
    },
    {
      "epoch": 4.633882352941177,
      "grad_norm": 0.33865844502347736,
      "learning_rate": 5.081966226102474e-05,
      "loss": 0.2734,
      "step": 9847
    },
    {
      "epoch": 4.634352941176471,
      "grad_norm": 0.2683768367484477,
      "learning_rate": 5.078678025985257e-05,
      "loss": 0.1641,
      "step": 9848
    },
    {
      "epoch": 4.634823529411765,
      "grad_norm": 0.43871598740515544,
      "learning_rate": 5.0753906731975954e-05,
      "loss": 0.5742,
      "step": 9849
    },
    {
      "epoch": 4.635294117647058,
      "grad_norm": 0.4302208569562573,
      "learning_rate": 5.07210416802025e-05,
      "loss": 0.5703,
      "step": 9850
    },
    {
      "epoch": 4.635764705882353,
      "grad_norm": 0.3350816760854083,
      "learning_rate": 5.0688185107339084e-05,
      "loss": 0.1309,
      "step": 9851
    },
    {
      "epoch": 4.636235294117647,
      "grad_norm": 0.291564902684781,
      "learning_rate": 5.065533701619175e-05,
      "loss": 0.1069,
      "step": 9852
    },
    {
      "epoch": 4.636705882352941,
      "grad_norm": 0.2097678372641014,
      "learning_rate": 5.06224974095659e-05,
      "loss": 0.0645,
      "step": 9853
    },
    {
      "epoch": 4.637176470588235,
      "grad_norm": 0.4581415720819578,
      "learning_rate": 5.058966629026629e-05,
      "loss": 0.2676,
      "step": 9854
    },
    {
      "epoch": 4.6376470588235295,
      "grad_norm": 0.2535721187705191,
      "learning_rate": 5.0556843661096714e-05,
      "loss": 0.0649,
      "step": 9855
    },
    {
      "epoch": 4.638117647058824,
      "grad_norm": 0.4220720769594987,
      "learning_rate": 5.052402952486046e-05,
      "loss": 0.4355,
      "step": 9856
    },
    {
      "epoch": 4.638588235294118,
      "grad_norm": 0.36738346116746284,
      "learning_rate": 5.0491223884360054e-05,
      "loss": 0.1436,
      "step": 9857
    },
    {
      "epoch": 4.639058823529412,
      "grad_norm": 0.3526581991507026,
      "learning_rate": 5.0458426742397174e-05,
      "loss": 0.4141,
      "step": 9858
    },
    {
      "epoch": 4.6395294117647055,
      "grad_norm": 0.4018724907837993,
      "learning_rate": 5.042563810177296e-05,
      "loss": 0.3125,
      "step": 9859
    },
    {
      "epoch": 4.64,
      "grad_norm": 0.2576938444435635,
      "learning_rate": 5.039285796528759e-05,
      "loss": 0.1807,
      "step": 9860
    },
    {
      "epoch": 4.640470588235294,
      "grad_norm": 0.4523980276345541,
      "learning_rate": 5.0360086335740734e-05,
      "loss": 0.5664,
      "step": 9861
    },
    {
      "epoch": 4.640941176470588,
      "grad_norm": 0.4408990167703995,
      "learning_rate": 5.032732321593128e-05,
      "loss": 0.3789,
      "step": 9862
    },
    {
      "epoch": 4.641411764705882,
      "grad_norm": 0.4608020720923697,
      "learning_rate": 5.029456860865727e-05,
      "loss": 0.3027,
      "step": 9863
    },
    {
      "epoch": 4.641882352941177,
      "grad_norm": 0.2319831635653895,
      "learning_rate": 5.0261822516716146e-05,
      "loss": 0.0728,
      "step": 9864
    },
    {
      "epoch": 4.642352941176471,
      "grad_norm": 0.2616297529262699,
      "learning_rate": 5.022908494290465e-05,
      "loss": 0.0693,
      "step": 9865
    },
    {
      "epoch": 4.642823529411765,
      "grad_norm": 0.27500134223585576,
      "learning_rate": 5.0196355890018605e-05,
      "loss": 0.1152,
      "step": 9866
    },
    {
      "epoch": 4.643294117647059,
      "grad_norm": 0.5635119937318827,
      "learning_rate": 5.016363536085333e-05,
      "loss": 0.3496,
      "step": 9867
    },
    {
      "epoch": 4.643764705882353,
      "grad_norm": 0.3445376609801171,
      "learning_rate": 5.0130923358203316e-05,
      "loss": 0.3711,
      "step": 9868
    },
    {
      "epoch": 4.644235294117647,
      "grad_norm": 0.21652381981513494,
      "learning_rate": 5.0098219884862265e-05,
      "loss": 0.165,
      "step": 9869
    },
    {
      "epoch": 4.644705882352941,
      "grad_norm": 0.2176700382936537,
      "learning_rate": 5.006552494362325e-05,
      "loss": 0.0427,
      "step": 9870
    },
    {
      "epoch": 4.645176470588235,
      "grad_norm": 0.41465136115959456,
      "learning_rate": 5.0032838537278627e-05,
      "loss": 0.1133,
      "step": 9871
    },
    {
      "epoch": 4.6456470588235295,
      "grad_norm": 0.49897022399352503,
      "learning_rate": 5.000016066861988e-05,
      "loss": 0.6719,
      "step": 9872
    },
    {
      "epoch": 4.646117647058824,
      "grad_norm": 0.4237082624430968,
      "learning_rate": 4.9967491340437904e-05,
      "loss": 0.4258,
      "step": 9873
    },
    {
      "epoch": 4.646588235294118,
      "grad_norm": 0.38322209150250586,
      "learning_rate": 4.993483055552289e-05,
      "loss": 0.3164,
      "step": 9874
    },
    {
      "epoch": 4.647058823529412,
      "grad_norm": 0.36302398802939506,
      "learning_rate": 4.990217831666409e-05,
      "loss": 0.4375,
      "step": 9875
    },
    {
      "epoch": 4.647529411764706,
      "grad_norm": 0.38549874889581337,
      "learning_rate": 4.986953462665026e-05,
      "loss": 0.4355,
      "step": 9876
    },
    {
      "epoch": 4.648,
      "grad_norm": 0.36761053492872536,
      "learning_rate": 4.9836899488269345e-05,
      "loss": 0.2578,
      "step": 9877
    },
    {
      "epoch": 4.648470588235294,
      "grad_norm": 0.503841405177122,
      "learning_rate": 4.9804272904308465e-05,
      "loss": 0.3262,
      "step": 9878
    },
    {
      "epoch": 4.648941176470588,
      "grad_norm": 0.3259710736980811,
      "learning_rate": 4.97716548775542e-05,
      "loss": 0.1406,
      "step": 9879
    },
    {
      "epoch": 4.649411764705882,
      "grad_norm": 0.26268806104509634,
      "learning_rate": 4.9739045410792164e-05,
      "loss": 0.2578,
      "step": 9880
    },
    {
      "epoch": 4.649882352941177,
      "grad_norm": 0.29610401311544,
      "learning_rate": 4.970644450680742e-05,
      "loss": 0.0645,
      "step": 9881
    },
    {
      "epoch": 4.650352941176471,
      "grad_norm": 0.7167396813979294,
      "learning_rate": 4.9673852168384316e-05,
      "loss": 0.5664,
      "step": 9882
    },
    {
      "epoch": 4.650823529411765,
      "grad_norm": 0.27839108599400875,
      "learning_rate": 4.964126839830628e-05,
      "loss": 0.1328,
      "step": 9883
    },
    {
      "epoch": 4.651294117647058,
      "grad_norm": 0.8192667298767449,
      "learning_rate": 4.960869319935616e-05,
      "loss": 0.5156,
      "step": 9884
    },
    {
      "epoch": 4.651764705882353,
      "grad_norm": 0.43977365668861873,
      "learning_rate": 4.9576126574316086e-05,
      "loss": 0.3066,
      "step": 9885
    },
    {
      "epoch": 4.652235294117647,
      "grad_norm": 0.2386856620317903,
      "learning_rate": 4.9543568525967376e-05,
      "loss": 0.0889,
      "step": 9886
    },
    {
      "epoch": 4.652705882352941,
      "grad_norm": 0.2887815727513067,
      "learning_rate": 4.9511019057090684e-05,
      "loss": 0.0796,
      "step": 9887
    },
    {
      "epoch": 4.653176470588235,
      "grad_norm": 0.43077928154311207,
      "learning_rate": 4.947847817046582e-05,
      "loss": 0.3555,
      "step": 9888
    },
    {
      "epoch": 4.6536470588235295,
      "grad_norm": 0.630687959505518,
      "learning_rate": 4.9445945868871974e-05,
      "loss": 0.3164,
      "step": 9889
    },
    {
      "epoch": 4.654117647058824,
      "grad_norm": 0.6725525212697695,
      "learning_rate": 4.941342215508761e-05,
      "loss": 0.3223,
      "step": 9890
    },
    {
      "epoch": 4.654588235294118,
      "grad_norm": 0.39884764867708494,
      "learning_rate": 4.9380907031890315e-05,
      "loss": 0.165,
      "step": 9891
    },
    {
      "epoch": 4.655058823529412,
      "grad_norm": 0.5405355651371614,
      "learning_rate": 4.934840050205711e-05,
      "loss": 0.2734,
      "step": 9892
    },
    {
      "epoch": 4.6555294117647055,
      "grad_norm": 0.993734917633351,
      "learning_rate": 4.931590256836424e-05,
      "loss": 0.5664,
      "step": 9893
    },
    {
      "epoch": 4.656,
      "grad_norm": 0.28047965551647136,
      "learning_rate": 4.928341323358708e-05,
      "loss": 0.0864,
      "step": 9894
    },
    {
      "epoch": 4.656470588235294,
      "grad_norm": 0.15034725237880803,
      "learning_rate": 4.925093250050045e-05,
      "loss": 0.0176,
      "step": 9895
    },
    {
      "epoch": 4.656941176470588,
      "grad_norm": 0.4822774208109881,
      "learning_rate": 4.921846037187841e-05,
      "loss": 0.4434,
      "step": 9896
    },
    {
      "epoch": 4.657411764705882,
      "grad_norm": 0.545551383275364,
      "learning_rate": 4.918599685049414e-05,
      "loss": 0.3223,
      "step": 9897
    },
    {
      "epoch": 4.657882352941177,
      "grad_norm": 0.3006611256027537,
      "learning_rate": 4.915354193912021e-05,
      "loss": 0.1318,
      "step": 9898
    },
    {
      "epoch": 4.658352941176471,
      "grad_norm": 0.3484719736868654,
      "learning_rate": 4.9121095640528516e-05,
      "loss": 0.1709,
      "step": 9899
    },
    {
      "epoch": 4.658823529411765,
      "grad_norm": 0.3304081997714294,
      "learning_rate": 4.908865795748999e-05,
      "loss": 0.1553,
      "step": 9900
    },
    {
      "epoch": 4.659294117647059,
      "grad_norm": 0.5808203187532589,
      "learning_rate": 4.905622889277506e-05,
      "loss": 0.3711,
      "step": 9901
    },
    {
      "epoch": 4.659764705882353,
      "grad_norm": 0.4482634686558569,
      "learning_rate": 4.902380844915337e-05,
      "loss": 0.4297,
      "step": 9902
    },
    {
      "epoch": 4.660235294117647,
      "grad_norm": 0.29664949636162913,
      "learning_rate": 4.899139662939366e-05,
      "loss": 0.103,
      "step": 9903
    },
    {
      "epoch": 4.660705882352941,
      "grad_norm": 0.4722153191275,
      "learning_rate": 4.895899343626412e-05,
      "loss": 0.167,
      "step": 9904
    },
    {
      "epoch": 4.661176470588235,
      "grad_norm": 0.35657077685662714,
      "learning_rate": 4.892659887253219e-05,
      "loss": 0.2773,
      "step": 9905
    },
    {
      "epoch": 4.6616470588235295,
      "grad_norm": 0.6309667222685292,
      "learning_rate": 4.8894212940964445e-05,
      "loss": 0.7383,
      "step": 9906
    },
    {
      "epoch": 4.662117647058824,
      "grad_norm": 0.412945418278434,
      "learning_rate": 4.8861835644326866e-05,
      "loss": 0.332,
      "step": 9907
    },
    {
      "epoch": 4.662588235294118,
      "grad_norm": 0.38197335783494674,
      "learning_rate": 4.882946698538457e-05,
      "loss": 0.4492,
      "step": 9908
    },
    {
      "epoch": 4.663058823529412,
      "grad_norm": 0.4713818280902328,
      "learning_rate": 4.879710696690202e-05,
      "loss": 0.2197,
      "step": 9909
    },
    {
      "epoch": 4.663529411764706,
      "grad_norm": 0.29957130221710904,
      "learning_rate": 4.876475559164299e-05,
      "loss": 0.0757,
      "step": 9910
    },
    {
      "epoch": 4.664,
      "grad_norm": 0.37302284303889255,
      "learning_rate": 4.873241286237033e-05,
      "loss": 0.2656,
      "step": 9911
    },
    {
      "epoch": 4.664470588235294,
      "grad_norm": 0.33036542808266156,
      "learning_rate": 4.8700078781846326e-05,
      "loss": 0.293,
      "step": 9912
    },
    {
      "epoch": 4.664941176470588,
      "grad_norm": 0.2837110341969905,
      "learning_rate": 4.866775335283253e-05,
      "loss": 0.1543,
      "step": 9913
    },
    {
      "epoch": 4.665411764705882,
      "grad_norm": 0.3510040142385114,
      "learning_rate": 4.863543657808956e-05,
      "loss": 0.1494,
      "step": 9914
    },
    {
      "epoch": 4.665882352941177,
      "grad_norm": 0.5753122127517735,
      "learning_rate": 4.860312846037753e-05,
      "loss": 0.7305,
      "step": 9915
    },
    {
      "epoch": 4.666352941176471,
      "grad_norm": 0.3873804402063182,
      "learning_rate": 4.857082900245569e-05,
      "loss": 0.3984,
      "step": 9916
    },
    {
      "epoch": 4.666823529411765,
      "grad_norm": 0.19502274215451096,
      "learning_rate": 4.8538538207082534e-05,
      "loss": 0.0571,
      "step": 9917
    },
    {
      "epoch": 4.667294117647058,
      "grad_norm": 0.5228220393223588,
      "learning_rate": 4.850625607701588e-05,
      "loss": 0.5391,
      "step": 9918
    },
    {
      "epoch": 4.667764705882353,
      "grad_norm": 0.41375969112980776,
      "learning_rate": 4.8473982615012816e-05,
      "loss": 0.4297,
      "step": 9919
    },
    {
      "epoch": 4.668235294117647,
      "grad_norm": 0.4995081886013442,
      "learning_rate": 4.844171782382959e-05,
      "loss": 0.3203,
      "step": 9920
    },
    {
      "epoch": 4.668705882352941,
      "grad_norm": 0.4114521461460991,
      "learning_rate": 4.840946170622179e-05,
      "loss": 0.1875,
      "step": 9921
    },
    {
      "epoch": 4.669176470588235,
      "grad_norm": 0.47408747325351547,
      "learning_rate": 4.8377214264944296e-05,
      "loss": 0.1602,
      "step": 9922
    },
    {
      "epoch": 4.6696470588235295,
      "grad_norm": 0.3067900513239502,
      "learning_rate": 4.834497550275114e-05,
      "loss": 0.0679,
      "step": 9923
    },
    {
      "epoch": 4.670117647058824,
      "grad_norm": 0.6491618367074455,
      "learning_rate": 4.831274542239566e-05,
      "loss": 0.4473,
      "step": 9924
    },
    {
      "epoch": 4.670588235294118,
      "grad_norm": 0.6217213123202959,
      "learning_rate": 4.8280524026630565e-05,
      "loss": 0.291,
      "step": 9925
    },
    {
      "epoch": 4.671058823529412,
      "grad_norm": 0.24387826280550184,
      "learning_rate": 4.824831131820759e-05,
      "loss": 0.0977,
      "step": 9926
    },
    {
      "epoch": 4.6715294117647055,
      "grad_norm": 0.31892238560909625,
      "learning_rate": 4.821610729987792e-05,
      "loss": 0.3555,
      "step": 9927
    },
    {
      "epoch": 4.672,
      "grad_norm": 0.35977485066626436,
      "learning_rate": 4.818391197439199e-05,
      "loss": 0.3379,
      "step": 9928
    },
    {
      "epoch": 4.672470588235294,
      "grad_norm": 0.3605074768106404,
      "learning_rate": 4.815172534449933e-05,
      "loss": 0.1875,
      "step": 9929
    },
    {
      "epoch": 4.672941176470588,
      "grad_norm": 0.23838818203911907,
      "learning_rate": 4.811954741294889e-05,
      "loss": 0.1855,
      "step": 9930
    },
    {
      "epoch": 4.673411764705882,
      "grad_norm": 0.4307878945546924,
      "learning_rate": 4.808737818248882e-05,
      "loss": 0.168,
      "step": 9931
    },
    {
      "epoch": 4.673882352941177,
      "grad_norm": 0.3992931636107897,
      "learning_rate": 4.805521765586653e-05,
      "loss": 0.1533,
      "step": 9932
    },
    {
      "epoch": 4.674352941176471,
      "grad_norm": 0.3627672359870057,
      "learning_rate": 4.802306583582873e-05,
      "loss": 0.4219,
      "step": 9933
    },
    {
      "epoch": 4.674823529411765,
      "grad_norm": 0.524054863294572,
      "learning_rate": 4.7990922725121266e-05,
      "loss": 0.5508,
      "step": 9934
    },
    {
      "epoch": 4.675294117647059,
      "grad_norm": 0.25334435233723634,
      "learning_rate": 4.795878832648935e-05,
      "loss": 0.0674,
      "step": 9935
    },
    {
      "epoch": 4.675764705882353,
      "grad_norm": 0.39453629284379965,
      "learning_rate": 4.7926662642677466e-05,
      "loss": 0.3438,
      "step": 9936
    },
    {
      "epoch": 4.676235294117647,
      "grad_norm": 0.21404887764151842,
      "learning_rate": 4.7894545676429206e-05,
      "loss": 0.0452,
      "step": 9937
    },
    {
      "epoch": 4.676705882352941,
      "grad_norm": 0.32697097492327354,
      "learning_rate": 4.7862437430487635e-05,
      "loss": 0.1079,
      "step": 9938
    },
    {
      "epoch": 4.677176470588235,
      "grad_norm": 0.6629932275230813,
      "learning_rate": 4.783033790759481e-05,
      "loss": 0.4062,
      "step": 9939
    },
    {
      "epoch": 4.6776470588235295,
      "grad_norm": 0.4135649325218426,
      "learning_rate": 4.77982471104923e-05,
      "loss": 0.4805,
      "step": 9940
    },
    {
      "epoch": 4.678117647058824,
      "grad_norm": 0.45726460735236996,
      "learning_rate": 4.776616504192082e-05,
      "loss": 0.5664,
      "step": 9941
    },
    {
      "epoch": 4.678588235294118,
      "grad_norm": 0.3636260233927445,
      "learning_rate": 4.7734091704620244e-05,
      "loss": 0.3379,
      "step": 9942
    },
    {
      "epoch": 4.679058823529411,
      "grad_norm": 0.5267185856962179,
      "learning_rate": 4.7702027101329846e-05,
      "loss": 0.4238,
      "step": 9943
    },
    {
      "epoch": 4.679529411764706,
      "grad_norm": 0.3380851551198526,
      "learning_rate": 4.7669971234788165e-05,
      "loss": 0.1641,
      "step": 9944
    },
    {
      "epoch": 4.68,
      "grad_norm": 0.6029002812170646,
      "learning_rate": 4.763792410773281e-05,
      "loss": 0.7734,
      "step": 9945
    },
    {
      "epoch": 4.680470588235294,
      "grad_norm": 0.4249492652162394,
      "learning_rate": 4.760588572290082e-05,
      "loss": 0.3105,
      "step": 9946
    },
    {
      "epoch": 4.680941176470588,
      "grad_norm": 0.42609456259068434,
      "learning_rate": 4.7573856083028475e-05,
      "loss": 0.2949,
      "step": 9947
    },
    {
      "epoch": 4.681411764705882,
      "grad_norm": 0.2675428311419731,
      "learning_rate": 4.7541835190851175e-05,
      "loss": 0.1592,
      "step": 9948
    },
    {
      "epoch": 4.681882352941177,
      "grad_norm": 0.20485416551177854,
      "learning_rate": 4.7509823049103706e-05,
      "loss": 0.0474,
      "step": 9949
    },
    {
      "epoch": 4.682352941176471,
      "grad_norm": 0.5796127041022714,
      "learning_rate": 4.74778196605201e-05,
      "loss": 0.4512,
      "step": 9950
    },
    {
      "epoch": 4.682823529411765,
      "grad_norm": 0.5040774995840641,
      "learning_rate": 4.744582502783352e-05,
      "loss": 0.4297,
      "step": 9951
    },
    {
      "epoch": 4.683294117647058,
      "grad_norm": 0.37696078237990943,
      "learning_rate": 4.7413839153776504e-05,
      "loss": 0.3672,
      "step": 9952
    },
    {
      "epoch": 4.683764705882353,
      "grad_norm": 0.40635307467331333,
      "learning_rate": 4.7381862041080865e-05,
      "loss": 0.5391,
      "step": 9953
    },
    {
      "epoch": 4.684235294117647,
      "grad_norm": 0.8971691612317179,
      "learning_rate": 4.7349893692477495e-05,
      "loss": 0.543,
      "step": 9954
    },
    {
      "epoch": 4.684705882352941,
      "grad_norm": 0.479507786930503,
      "learning_rate": 4.7317934110696685e-05,
      "loss": 0.4316,
      "step": 9955
    },
    {
      "epoch": 4.685176470588235,
      "grad_norm": 0.41065578778125866,
      "learning_rate": 4.728598329846802e-05,
      "loss": 0.373,
      "step": 9956
    },
    {
      "epoch": 4.6856470588235295,
      "grad_norm": 0.28714814727496946,
      "learning_rate": 4.725404125852013e-05,
      "loss": 0.1602,
      "step": 9957
    },
    {
      "epoch": 4.686117647058824,
      "grad_norm": 0.653983171373012,
      "learning_rate": 4.7222107993581135e-05,
      "loss": 0.5352,
      "step": 9958
    },
    {
      "epoch": 4.686588235294118,
      "grad_norm": 0.26595500991288756,
      "learning_rate": 4.7190183506378205e-05,
      "loss": 0.1094,
      "step": 9959
    },
    {
      "epoch": 4.687058823529412,
      "grad_norm": 0.4245114409492281,
      "learning_rate": 4.7158267799637864e-05,
      "loss": 0.4336,
      "step": 9960
    },
    {
      "epoch": 4.6875294117647055,
      "grad_norm": 0.5225037861465197,
      "learning_rate": 4.712636087608594e-05,
      "loss": 0.3477,
      "step": 9961
    },
    {
      "epoch": 4.688,
      "grad_norm": 0.9813393930191554,
      "learning_rate": 4.709446273844736e-05,
      "loss": 0.5117,
      "step": 9962
    },
    {
      "epoch": 4.688470588235294,
      "grad_norm": 0.41352406200193886,
      "learning_rate": 4.7062573389446395e-05,
      "loss": 0.1934,
      "step": 9963
    },
    {
      "epoch": 4.688941176470588,
      "grad_norm": 0.4294327309237382,
      "learning_rate": 4.70306928318066e-05,
      "loss": 0.4395,
      "step": 9964
    },
    {
      "epoch": 4.689411764705882,
      "grad_norm": 0.42530242478697655,
      "learning_rate": 4.6998821068250675e-05,
      "loss": 0.5586,
      "step": 9965
    },
    {
      "epoch": 4.689882352941177,
      "grad_norm": 0.5088220643964471,
      "learning_rate": 4.696695810150065e-05,
      "loss": 0.4922,
      "step": 9966
    },
    {
      "epoch": 4.690352941176471,
      "grad_norm": 0.34238051732230174,
      "learning_rate": 4.693510393427781e-05,
      "loss": 0.3477,
      "step": 9967
    },
    {
      "epoch": 4.690823529411765,
      "grad_norm": 0.37262254277027157,
      "learning_rate": 4.690325856930258e-05,
      "loss": 0.3164,
      "step": 9968
    },
    {
      "epoch": 4.691294117647059,
      "grad_norm": 0.5928365394377254,
      "learning_rate": 4.6871422009294764e-05,
      "loss": 0.8555,
      "step": 9969
    },
    {
      "epoch": 4.691764705882353,
      "grad_norm": 0.3840725885860039,
      "learning_rate": 4.6839594256973406e-05,
      "loss": 0.3301,
      "step": 9970
    },
    {
      "epoch": 4.692235294117647,
      "grad_norm": 0.38559544370273596,
      "learning_rate": 4.680777531505662e-05,
      "loss": 0.3184,
      "step": 9971
    },
    {
      "epoch": 4.692705882352941,
      "grad_norm": 0.2652239483545428,
      "learning_rate": 4.6775965186262e-05,
      "loss": 0.377,
      "step": 9972
    },
    {
      "epoch": 4.693176470588235,
      "grad_norm": 0.3880442214382258,
      "learning_rate": 4.6744163873306314e-05,
      "loss": 0.293,
      "step": 9973
    },
    {
      "epoch": 4.6936470588235295,
      "grad_norm": 0.3542940627362567,
      "learning_rate": 4.671237137890539e-05,
      "loss": 0.1123,
      "step": 9974
    },
    {
      "epoch": 4.694117647058824,
      "grad_norm": 0.41794600799143994,
      "learning_rate": 4.668058770577468e-05,
      "loss": 0.3926,
      "step": 9975
    },
    {
      "epoch": 4.694588235294118,
      "grad_norm": 0.3606087800278437,
      "learning_rate": 4.6648812856628515e-05,
      "loss": 0.2188,
      "step": 9976
    },
    {
      "epoch": 4.695058823529411,
      "grad_norm": 0.32012863715558115,
      "learning_rate": 4.6617046834180656e-05,
      "loss": 0.3496,
      "step": 9977
    },
    {
      "epoch": 4.6955294117647055,
      "grad_norm": 1.0183479934710744,
      "learning_rate": 4.658528964114413e-05,
      "loss": 0.7227,
      "step": 9978
    },
    {
      "epoch": 4.696,
      "grad_norm": 0.4084292016154489,
      "learning_rate": 4.655354128023107e-05,
      "loss": 0.5273,
      "step": 9979
    },
    {
      "epoch": 4.696470588235294,
      "grad_norm": 0.27506170140782427,
      "learning_rate": 4.6521801754152995e-05,
      "loss": 0.0786,
      "step": 9980
    },
    {
      "epoch": 4.696941176470588,
      "grad_norm": 0.1784540900503676,
      "learning_rate": 4.649007106562066e-05,
      "loss": 0.0962,
      "step": 9981
    },
    {
      "epoch": 4.697411764705882,
      "grad_norm": 0.24971948070297148,
      "learning_rate": 4.645834921734392e-05,
      "loss": 0.0806,
      "step": 9982
    },
    {
      "epoch": 4.697882352941177,
      "grad_norm": 0.38907601914876616,
      "learning_rate": 4.642663621203201e-05,
      "loss": 0.1982,
      "step": 9983
    },
    {
      "epoch": 4.698352941176471,
      "grad_norm": 0.23903727736896513,
      "learning_rate": 4.639493205239346e-05,
      "loss": 0.1689,
      "step": 9984
    },
    {
      "epoch": 4.698823529411765,
      "grad_norm": 0.3308478858163479,
      "learning_rate": 4.636323674113583e-05,
      "loss": 0.3359,
      "step": 9985
    },
    {
      "epoch": 4.699294117647058,
      "grad_norm": 0.3509162757247081,
      "learning_rate": 4.6331550280966126e-05,
      "loss": 0.1748,
      "step": 9986
    },
    {
      "epoch": 4.699764705882353,
      "grad_norm": 0.29873929288289297,
      "learning_rate": 4.629987267459056e-05,
      "loss": 0.0947,
      "step": 9987
    },
    {
      "epoch": 4.700235294117647,
      "grad_norm": 0.215006743825852,
      "learning_rate": 4.626820392471446e-05,
      "loss": 0.043,
      "step": 9988
    },
    {
      "epoch": 4.700705882352941,
      "grad_norm": 0.31437511542886465,
      "learning_rate": 4.6236544034042604e-05,
      "loss": 0.2129,
      "step": 9989
    },
    {
      "epoch": 4.701176470588235,
      "grad_norm": 0.40014858436915546,
      "learning_rate": 4.62048930052788e-05,
      "loss": 0.2227,
      "step": 9990
    },
    {
      "epoch": 4.7016470588235295,
      "grad_norm": 1.005770343807084,
      "learning_rate": 4.6173250841126245e-05,
      "loss": 1.0781,
      "step": 9991
    },
    {
      "epoch": 4.702117647058824,
      "grad_norm": 0.5659713715866459,
      "learning_rate": 4.614161754428737e-05,
      "loss": 0.4609,
      "step": 9992
    },
    {
      "epoch": 4.702588235294118,
      "grad_norm": 0.411309035994991,
      "learning_rate": 4.6109993117463756e-05,
      "loss": 0.3789,
      "step": 9993
    },
    {
      "epoch": 4.703058823529412,
      "grad_norm": 0.2613212512907351,
      "learning_rate": 4.607837756335629e-05,
      "loss": 0.0649,
      "step": 9994
    },
    {
      "epoch": 4.7035294117647055,
      "grad_norm": 0.4027524917614379,
      "learning_rate": 4.604677088466516e-05,
      "loss": 0.3262,
      "step": 9995
    },
    {
      "epoch": 4.704,
      "grad_norm": 0.5086867561222372,
      "learning_rate": 4.6015173084089635e-05,
      "loss": 0.582,
      "step": 9996
    },
    {
      "epoch": 4.704470588235294,
      "grad_norm": 0.39120991981779124,
      "learning_rate": 4.598358416432838e-05,
      "loss": 0.4844,
      "step": 9997
    },
    {
      "epoch": 4.704941176470588,
      "grad_norm": 0.7000327649561482,
      "learning_rate": 4.595200412807927e-05,
      "loss": 0.4336,
      "step": 9998
    },
    {
      "epoch": 4.705411764705882,
      "grad_norm": 0.3872039483368851,
      "learning_rate": 4.592043297803932e-05,
      "loss": 0.1836,
      "step": 9999
    },
    {
      "epoch": 4.705882352941177,
      "grad_norm": 0.4403769582815634,
      "learning_rate": 4.58888707169049e-05,
      "loss": 0.5,
      "step": 10000
    },
    {
      "epoch": 4.706352941176471,
      "grad_norm": 0.522814885283815,
      "learning_rate": 4.585731734737164e-05,
      "loss": 0.3691,
      "step": 10001
    },
    {
      "epoch": 4.706823529411765,
      "grad_norm": 0.297304440397142,
      "learning_rate": 4.5825772872134234e-05,
      "loss": 0.1318,
      "step": 10002
    },
    {
      "epoch": 4.707294117647059,
      "grad_norm": 0.2079243635432003,
      "learning_rate": 4.579423729388681e-05,
      "loss": 0.1143,
      "step": 10003
    },
    {
      "epoch": 4.707764705882353,
      "grad_norm": 0.3563597515001611,
      "learning_rate": 4.576271061532268e-05,
      "loss": 0.1504,
      "step": 10004
    },
    {
      "epoch": 4.708235294117647,
      "grad_norm": 0.3893979487883005,
      "learning_rate": 4.573119283913431e-05,
      "loss": 0.1846,
      "step": 10005
    },
    {
      "epoch": 4.708705882352941,
      "grad_norm": 0.38116249192175666,
      "learning_rate": 4.569968396801351e-05,
      "loss": 0.2676,
      "step": 10006
    },
    {
      "epoch": 4.709176470588235,
      "grad_norm": 0.12713762538119988,
      "learning_rate": 4.566818400465133e-05,
      "loss": 0.0171,
      "step": 10007
    },
    {
      "epoch": 4.7096470588235295,
      "grad_norm": 0.27257858483211955,
      "learning_rate": 4.563669295173793e-05,
      "loss": 0.1079,
      "step": 10008
    },
    {
      "epoch": 4.710117647058824,
      "grad_norm": 0.35549900970284853,
      "learning_rate": 4.560521081196289e-05,
      "loss": 0.2539,
      "step": 10009
    },
    {
      "epoch": 4.710588235294118,
      "grad_norm": 0.3782545922710049,
      "learning_rate": 4.5573737588014875e-05,
      "loss": 0.103,
      "step": 10010
    },
    {
      "epoch": 4.711058823529411,
      "grad_norm": 0.714399536689937,
      "learning_rate": 4.554227328258185e-05,
      "loss": 0.4902,
      "step": 10011
    },
    {
      "epoch": 4.7115294117647055,
      "grad_norm": 0.4746037743536439,
      "learning_rate": 4.551081789835113e-05,
      "loss": 0.5469,
      "step": 10012
    },
    {
      "epoch": 4.712,
      "grad_norm": 0.45749628609965304,
      "learning_rate": 4.547937143800901e-05,
      "loss": 0.1143,
      "step": 10013
    },
    {
      "epoch": 4.712470588235294,
      "grad_norm": 0.38031992094731587,
      "learning_rate": 4.544793390424124e-05,
      "loss": 0.1719,
      "step": 10014
    },
    {
      "epoch": 4.712941176470588,
      "grad_norm": 0.4043574670906792,
      "learning_rate": 4.541650529973279e-05,
      "loss": 0.3359,
      "step": 10015
    },
    {
      "epoch": 4.713411764705882,
      "grad_norm": 0.38177138420671447,
      "learning_rate": 4.5385085627167717e-05,
      "loss": 0.5742,
      "step": 10016
    },
    {
      "epoch": 4.713882352941177,
      "grad_norm": 0.24254417759160127,
      "learning_rate": 4.535367488922947e-05,
      "loss": 0.0825,
      "step": 10017
    },
    {
      "epoch": 4.714352941176471,
      "grad_norm": 0.41085285082614387,
      "learning_rate": 4.5322273088600714e-05,
      "loss": 0.5156,
      "step": 10018
    },
    {
      "epoch": 4.714823529411765,
      "grad_norm": 0.54818588891114,
      "learning_rate": 4.5290880227963186e-05,
      "loss": 0.2207,
      "step": 10019
    },
    {
      "epoch": 4.715294117647058,
      "grad_norm": 0.47302755664875246,
      "learning_rate": 4.525949630999818e-05,
      "loss": 0.3438,
      "step": 10020
    },
    {
      "epoch": 4.715764705882353,
      "grad_norm": 0.426995066496936,
      "learning_rate": 4.522812133738589e-05,
      "loss": 0.1758,
      "step": 10021
    },
    {
      "epoch": 4.716235294117647,
      "grad_norm": 0.37409137202911563,
      "learning_rate": 4.519675531280594e-05,
      "loss": 0.25,
      "step": 10022
    },
    {
      "epoch": 4.716705882352941,
      "grad_norm": 0.2716548464283152,
      "learning_rate": 4.5165398238937194e-05,
      "loss": 0.1016,
      "step": 10023
    },
    {
      "epoch": 4.717176470588235,
      "grad_norm": 0.4257315422967523,
      "learning_rate": 4.51340501184576e-05,
      "loss": 0.4043,
      "step": 10024
    },
    {
      "epoch": 4.7176470588235295,
      "grad_norm": 0.188955983858007,
      "learning_rate": 4.510271095404449e-05,
      "loss": 0.0444,
      "step": 10025
    },
    {
      "epoch": 4.718117647058824,
      "grad_norm": 0.5607981066828972,
      "learning_rate": 4.507138074837443e-05,
      "loss": 0.3496,
      "step": 10026
    },
    {
      "epoch": 4.718588235294118,
      "grad_norm": 0.4307915454006439,
      "learning_rate": 4.504005950412308e-05,
      "loss": 0.3535,
      "step": 10027
    },
    {
      "epoch": 4.719058823529412,
      "grad_norm": 0.43446392836746744,
      "learning_rate": 4.5008747223965456e-05,
      "loss": 0.3008,
      "step": 10028
    },
    {
      "epoch": 4.7195294117647055,
      "grad_norm": 0.2691306276373155,
      "learning_rate": 4.4977443910575866e-05,
      "loss": 0.0542,
      "step": 10029
    },
    {
      "epoch": 4.72,
      "grad_norm": 0.2292658964569921,
      "learning_rate": 4.494614956662765e-05,
      "loss": 0.123,
      "step": 10030
    },
    {
      "epoch": 4.720470588235294,
      "grad_norm": 0.48607038547244574,
      "learning_rate": 4.491486419479355e-05,
      "loss": 0.334,
      "step": 10031
    },
    {
      "epoch": 4.720941176470588,
      "grad_norm": 0.38549807835261996,
      "learning_rate": 4.488358779774551e-05,
      "loss": 0.4648,
      "step": 10032
    },
    {
      "epoch": 4.721411764705882,
      "grad_norm": 0.44982849940936154,
      "learning_rate": 4.485232037815465e-05,
      "loss": 0.4336,
      "step": 10033
    },
    {
      "epoch": 4.721882352941177,
      "grad_norm": 0.3170553249580862,
      "learning_rate": 4.4821061938691355e-05,
      "loss": 0.3535,
      "step": 10034
    },
    {
      "epoch": 4.722352941176471,
      "grad_norm": 0.5784108413389388,
      "learning_rate": 4.478981248202533e-05,
      "loss": 0.5117,
      "step": 10035
    },
    {
      "epoch": 4.722823529411765,
      "grad_norm": 0.27984848616582053,
      "learning_rate": 4.4758572010825324e-05,
      "loss": 0.2041,
      "step": 10036
    },
    {
      "epoch": 4.723294117647059,
      "grad_norm": 0.22166355649798783,
      "learning_rate": 4.472734052775947e-05,
      "loss": 0.1699,
      "step": 10037
    },
    {
      "epoch": 4.723764705882353,
      "grad_norm": 0.25716920556683165,
      "learning_rate": 4.4696118035495136e-05,
      "loss": 0.0962,
      "step": 10038
    },
    {
      "epoch": 4.724235294117647,
      "grad_norm": 0.41253426157700396,
      "learning_rate": 4.4664904536698796e-05,
      "loss": 0.5195,
      "step": 10039
    },
    {
      "epoch": 4.724705882352941,
      "grad_norm": 0.19327725467804305,
      "learning_rate": 4.463370003403632e-05,
      "loss": 0.0439,
      "step": 10040
    },
    {
      "epoch": 4.725176470588235,
      "grad_norm": 0.350580460220105,
      "learning_rate": 4.4602504530172636e-05,
      "loss": 0.2246,
      "step": 10041
    },
    {
      "epoch": 4.7256470588235295,
      "grad_norm": 0.5073638398792979,
      "learning_rate": 4.457131802777202e-05,
      "loss": 0.4277,
      "step": 10042
    },
    {
      "epoch": 4.726117647058824,
      "grad_norm": 0.24190347087349034,
      "learning_rate": 4.454014052949804e-05,
      "loss": 0.1475,
      "step": 10043
    },
    {
      "epoch": 4.726588235294118,
      "grad_norm": 0.424228940514529,
      "learning_rate": 4.4508972038013274e-05,
      "loss": 0.3906,
      "step": 10044
    },
    {
      "epoch": 4.727058823529411,
      "grad_norm": 0.1634448644928483,
      "learning_rate": 4.4477812555979724e-05,
      "loss": 0.0251,
      "step": 10045
    },
    {
      "epoch": 4.7275294117647055,
      "grad_norm": 0.290753509504713,
      "learning_rate": 4.4446662086058634e-05,
      "loss": 0.3184,
      "step": 10046
    },
    {
      "epoch": 4.728,
      "grad_norm": 0.4189344870400787,
      "learning_rate": 4.4415520630910265e-05,
      "loss": 0.293,
      "step": 10047
    },
    {
      "epoch": 4.728470588235294,
      "grad_norm": 0.3912551299425931,
      "learning_rate": 4.438438819319434e-05,
      "loss": 0.1943,
      "step": 10048
    },
    {
      "epoch": 4.728941176470588,
      "grad_norm": 0.29608154185875324,
      "learning_rate": 4.435326477556975e-05,
      "loss": 0.1069,
      "step": 10049
    },
    {
      "epoch": 4.729411764705882,
      "grad_norm": 0.33903097255959913,
      "learning_rate": 4.432215038069449e-05,
      "loss": 0.1201,
      "step": 10050
    },
    {
      "epoch": 4.729882352941177,
      "grad_norm": 0.4144768895064162,
      "learning_rate": 4.4291045011225934e-05,
      "loss": 0.543,
      "step": 10051
    },
    {
      "epoch": 4.730352941176471,
      "grad_norm": 0.3291247274469476,
      "learning_rate": 4.4259948669820676e-05,
      "loss": 0.4199,
      "step": 10052
    },
    {
      "epoch": 4.730823529411765,
      "grad_norm": 0.5792980113976095,
      "learning_rate": 4.422886135913441e-05,
      "loss": 0.4727,
      "step": 10053
    },
    {
      "epoch": 4.731294117647058,
      "grad_norm": 0.4112744087960155,
      "learning_rate": 4.419778308182219e-05,
      "loss": 0.2041,
      "step": 10054
    },
    {
      "epoch": 4.731764705882353,
      "grad_norm": 0.2170305164613273,
      "learning_rate": 4.41667138405383e-05,
      "loss": 0.0544,
      "step": 10055
    },
    {
      "epoch": 4.732235294117647,
      "grad_norm": 0.8631812738484961,
      "learning_rate": 4.413565363793611e-05,
      "loss": 0.7266,
      "step": 10056
    },
    {
      "epoch": 4.732705882352941,
      "grad_norm": 0.9612842906507093,
      "learning_rate": 4.4104602476668364e-05,
      "loss": 0.6602,
      "step": 10057
    },
    {
      "epoch": 4.733176470588235,
      "grad_norm": 0.17748962634143842,
      "learning_rate": 4.4073560359387035e-05,
      "loss": 0.0259,
      "step": 10058
    },
    {
      "epoch": 4.7336470588235295,
      "grad_norm": 0.2506404703235974,
      "learning_rate": 4.4042527288743155e-05,
      "loss": 0.1089,
      "step": 10059
    },
    {
      "epoch": 4.734117647058824,
      "grad_norm": 0.32719858524488193,
      "learning_rate": 4.401150326738722e-05,
      "loss": 0.1025,
      "step": 10060
    },
    {
      "epoch": 4.734588235294118,
      "grad_norm": 0.38501207833896334,
      "learning_rate": 4.3980488297968726e-05,
      "loss": 0.2949,
      "step": 10061
    },
    {
      "epoch": 4.735058823529412,
      "grad_norm": 0.4574856615246932,
      "learning_rate": 4.394948238313655e-05,
      "loss": 0.4844,
      "step": 10062
    },
    {
      "epoch": 4.7355294117647055,
      "grad_norm": 0.416266185213785,
      "learning_rate": 4.391848552553875e-05,
      "loss": 0.4121,
      "step": 10063
    },
    {
      "epoch": 4.736,
      "grad_norm": 0.40843519070164963,
      "learning_rate": 4.3887497727822616e-05,
      "loss": 0.2021,
      "step": 10064
    },
    {
      "epoch": 4.736470588235294,
      "grad_norm": 0.3188375289581663,
      "learning_rate": 4.385651899263467e-05,
      "loss": 0.1982,
      "step": 10065
    },
    {
      "epoch": 4.736941176470588,
      "grad_norm": 0.44476068929123247,
      "learning_rate": 4.3825549322620657e-05,
      "loss": 0.4863,
      "step": 10066
    },
    {
      "epoch": 4.737411764705882,
      "grad_norm": 1.3649332426959169,
      "learning_rate": 4.3794588720425486e-05,
      "loss": 1.0234,
      "step": 10067
    },
    {
      "epoch": 4.737882352941177,
      "grad_norm": 0.5588080473567062,
      "learning_rate": 4.3763637188693404e-05,
      "loss": 0.6484,
      "step": 10068
    },
    {
      "epoch": 4.738352941176471,
      "grad_norm": 0.33403472772317827,
      "learning_rate": 4.3732694730067766e-05,
      "loss": 0.1953,
      "step": 10069
    },
    {
      "epoch": 4.738823529411764,
      "grad_norm": 0.2400473608748517,
      "learning_rate": 4.370176134719121e-05,
      "loss": 0.0623,
      "step": 10070
    },
    {
      "epoch": 4.739294117647058,
      "grad_norm": 0.5256770516339645,
      "learning_rate": 4.36708370427057e-05,
      "loss": 0.7422,
      "step": 10071
    },
    {
      "epoch": 4.739764705882353,
      "grad_norm": 0.31721028569020726,
      "learning_rate": 4.363992181925218e-05,
      "loss": 0.4414,
      "step": 10072
    },
    {
      "epoch": 4.740235294117647,
      "grad_norm": 0.5447086471714366,
      "learning_rate": 4.3609015679471036e-05,
      "loss": 0.2832,
      "step": 10073
    },
    {
      "epoch": 4.740705882352941,
      "grad_norm": 0.7709945966992444,
      "learning_rate": 4.357811862600185e-05,
      "loss": 0.543,
      "step": 10074
    },
    {
      "epoch": 4.741176470588235,
      "grad_norm": 0.40530497168151663,
      "learning_rate": 4.3547230661483276e-05,
      "loss": 0.4727,
      "step": 10075
    },
    {
      "epoch": 4.7416470588235295,
      "grad_norm": 0.4026718016591765,
      "learning_rate": 4.351635178855336e-05,
      "loss": 0.166,
      "step": 10076
    },
    {
      "epoch": 4.742117647058824,
      "grad_norm": 0.44881510786816525,
      "learning_rate": 4.348548200984934e-05,
      "loss": 0.5547,
      "step": 10077
    },
    {
      "epoch": 4.742588235294118,
      "grad_norm": 0.3791854458518429,
      "learning_rate": 4.345462132800756e-05,
      "loss": 0.4141,
      "step": 10078
    },
    {
      "epoch": 4.743058823529411,
      "grad_norm": 0.29930414225775265,
      "learning_rate": 4.342376974566372e-05,
      "loss": 0.1196,
      "step": 10079
    },
    {
      "epoch": 4.7435294117647056,
      "grad_norm": 0.28731756704839284,
      "learning_rate": 4.339292726545275e-05,
      "loss": 0.1108,
      "step": 10080
    },
    {
      "epoch": 4.744,
      "grad_norm": 0.3714379136663129,
      "learning_rate": 4.3362093890008645e-05,
      "loss": 0.4375,
      "step": 10081
    },
    {
      "epoch": 4.744470588235294,
      "grad_norm": 0.7414983145308799,
      "learning_rate": 4.333126962196478e-05,
      "loss": 0.9297,
      "step": 10082
    },
    {
      "epoch": 4.744941176470588,
      "grad_norm": 0.29895589353148955,
      "learning_rate": 4.330045446395374e-05,
      "loss": 0.3652,
      "step": 10083
    },
    {
      "epoch": 4.745411764705882,
      "grad_norm": 0.9048698527036397,
      "learning_rate": 4.3269648418607194e-05,
      "loss": 0.3945,
      "step": 10084
    },
    {
      "epoch": 4.745882352941177,
      "grad_norm": 0.41748897195259443,
      "learning_rate": 4.32388514885562e-05,
      "loss": 0.2402,
      "step": 10085
    },
    {
      "epoch": 4.746352941176471,
      "grad_norm": 0.5020206007197912,
      "learning_rate": 4.320806367643098e-05,
      "loss": 0.3066,
      "step": 10086
    },
    {
      "epoch": 4.746823529411765,
      "grad_norm": 0.6165180820392225,
      "learning_rate": 4.3177284984860895e-05,
      "loss": 0.3066,
      "step": 10087
    },
    {
      "epoch": 4.747294117647058,
      "grad_norm": 0.01619097259633723,
      "learning_rate": 4.314651541647468e-05,
      "loss": 0.0013,
      "step": 10088
    },
    {
      "epoch": 4.747764705882353,
      "grad_norm": 0.30326808233092384,
      "learning_rate": 4.3115754973900104e-05,
      "loss": 0.1006,
      "step": 10089
    },
    {
      "epoch": 4.748235294117647,
      "grad_norm": 0.5371053704596098,
      "learning_rate": 4.308500365976434e-05,
      "loss": 0.8125,
      "step": 10090
    },
    {
      "epoch": 4.748705882352941,
      "grad_norm": 0.19920087341353745,
      "learning_rate": 4.305426147669371e-05,
      "loss": 0.1533,
      "step": 10091
    },
    {
      "epoch": 4.749176470588235,
      "grad_norm": 0.43899943391460994,
      "learning_rate": 4.302352842731368e-05,
      "loss": 0.249,
      "step": 10092
    },
    {
      "epoch": 4.7496470588235296,
      "grad_norm": 0.3292217468818548,
      "learning_rate": 4.299280451424903e-05,
      "loss": 0.127,
      "step": 10093
    },
    {
      "epoch": 4.750117647058824,
      "grad_norm": 0.31771263389831794,
      "learning_rate": 4.296208974012379e-05,
      "loss": 0.0967,
      "step": 10094
    },
    {
      "epoch": 4.750588235294118,
      "grad_norm": 0.3932324300680281,
      "learning_rate": 4.2931384107561065e-05,
      "loss": 0.2305,
      "step": 10095
    },
    {
      "epoch": 4.751058823529412,
      "grad_norm": 0.401419745282336,
      "learning_rate": 4.290068761918329e-05,
      "loss": 0.3105,
      "step": 10096
    },
    {
      "epoch": 4.751529411764706,
      "grad_norm": 0.3953616412224378,
      "learning_rate": 4.287000027761217e-05,
      "loss": 0.3984,
      "step": 10097
    },
    {
      "epoch": 4.752,
      "grad_norm": 0.39063800507172847,
      "learning_rate": 4.2839322085468445e-05,
      "loss": 0.4766,
      "step": 10098
    },
    {
      "epoch": 4.752470588235294,
      "grad_norm": 0.5618159393941978,
      "learning_rate": 4.280865304537223e-05,
      "loss": 0.418,
      "step": 10099
    },
    {
      "epoch": 4.752941176470588,
      "grad_norm": 0.43525707126556185,
      "learning_rate": 4.277799315994286e-05,
      "loss": 0.373,
      "step": 10100
    },
    {
      "epoch": 4.7534117647058824,
      "grad_norm": 0.5771714951999294,
      "learning_rate": 4.2747342431798767e-05,
      "loss": 0.4609,
      "step": 10101
    },
    {
      "epoch": 4.753882352941177,
      "grad_norm": 0.5344079443681421,
      "learning_rate": 4.271670086355768e-05,
      "loss": 0.2637,
      "step": 10102
    },
    {
      "epoch": 4.754352941176471,
      "grad_norm": 0.4862554774875983,
      "learning_rate": 4.268606845783663e-05,
      "loss": 0.4883,
      "step": 10103
    },
    {
      "epoch": 4.754823529411764,
      "grad_norm": 0.30695795426578365,
      "learning_rate": 4.265544521725165e-05,
      "loss": 0.1016,
      "step": 10104
    },
    {
      "epoch": 4.7552941176470584,
      "grad_norm": 0.7310726850558175,
      "learning_rate": 4.2624831144418166e-05,
      "loss": 0.3008,
      "step": 10105
    },
    {
      "epoch": 4.755764705882353,
      "grad_norm": 0.3977082997312183,
      "learning_rate": 4.259422624195082e-05,
      "loss": 0.4512,
      "step": 10106
    },
    {
      "epoch": 4.756235294117647,
      "grad_norm": 0.40867742605044705,
      "learning_rate": 4.2563630512463284e-05,
      "loss": 0.3945,
      "step": 10107
    },
    {
      "epoch": 4.756705882352941,
      "grad_norm": 0.48103486910665255,
      "learning_rate": 4.253304395856876e-05,
      "loss": 0.3867,
      "step": 10108
    },
    {
      "epoch": 4.757176470588235,
      "grad_norm": 0.33163403834610494,
      "learning_rate": 4.2502466582879375e-05,
      "loss": 0.1484,
      "step": 10109
    },
    {
      "epoch": 4.75764705882353,
      "grad_norm": 0.4415541225227546,
      "learning_rate": 4.2471898388006606e-05,
      "loss": 0.3691,
      "step": 10110
    },
    {
      "epoch": 4.758117647058824,
      "grad_norm": 0.5422162328239825,
      "learning_rate": 4.244133937656118e-05,
      "loss": 0.6875,
      "step": 10111
    },
    {
      "epoch": 4.758588235294118,
      "grad_norm": 0.4015618916782442,
      "learning_rate": 4.2410789551152894e-05,
      "loss": 0.2266,
      "step": 10112
    },
    {
      "epoch": 4.759058823529411,
      "grad_norm": 0.8243063807139683,
      "learning_rate": 4.238024891439092e-05,
      "loss": 0.6992,
      "step": 10113
    },
    {
      "epoch": 4.759529411764706,
      "grad_norm": 0.34499014622133184,
      "learning_rate": 4.234971746888359e-05,
      "loss": 0.3633,
      "step": 10114
    },
    {
      "epoch": 4.76,
      "grad_norm": 0.34425795692978456,
      "learning_rate": 4.231919521723836e-05,
      "loss": 0.2871,
      "step": 10115
    },
    {
      "epoch": 4.760470588235294,
      "grad_norm": 0.3905698684592081,
      "learning_rate": 4.228868216206206e-05,
      "loss": 0.125,
      "step": 10116
    },
    {
      "epoch": 4.760941176470588,
      "grad_norm": 0.38795663132140085,
      "learning_rate": 4.225817830596064e-05,
      "loss": 0.2422,
      "step": 10117
    },
    {
      "epoch": 4.7614117647058825,
      "grad_norm": 0.43719606466659044,
      "learning_rate": 4.2227683651539236e-05,
      "loss": 0.373,
      "step": 10118
    },
    {
      "epoch": 4.761882352941177,
      "grad_norm": 0.19574635888565084,
      "learning_rate": 4.21971982014023e-05,
      "loss": 0.0532,
      "step": 10119
    },
    {
      "epoch": 4.762352941176471,
      "grad_norm": 0.27354699675779864,
      "learning_rate": 4.2166721958153365e-05,
      "loss": 0.25,
      "step": 10120
    },
    {
      "epoch": 4.762823529411765,
      "grad_norm": 0.4164483296718392,
      "learning_rate": 4.2136254924395305e-05,
      "loss": 0.1543,
      "step": 10121
    },
    {
      "epoch": 4.7632941176470585,
      "grad_norm": 0.2657113838253951,
      "learning_rate": 4.210579710273018e-05,
      "loss": 0.084,
      "step": 10122
    },
    {
      "epoch": 4.763764705882353,
      "grad_norm": 0.20422728006160493,
      "learning_rate": 4.207534849575916e-05,
      "loss": 0.1211,
      "step": 10123
    },
    {
      "epoch": 4.764235294117647,
      "grad_norm": 0.11609139549310439,
      "learning_rate": 4.204490910608274e-05,
      "loss": 0.0038,
      "step": 10124
    },
    {
      "epoch": 4.764705882352941,
      "grad_norm": 0.3183908655113132,
      "learning_rate": 4.201447893630065e-05,
      "loss": 0.2988,
      "step": 10125
    },
    {
      "epoch": 4.765176470588235,
      "grad_norm": 0.1662949464611959,
      "learning_rate": 4.1984057989011674e-05,
      "loss": 0.0417,
      "step": 10126
    },
    {
      "epoch": 4.76564705882353,
      "grad_norm": 0.4313982737809837,
      "learning_rate": 4.195364626681396e-05,
      "loss": 0.4297,
      "step": 10127
    },
    {
      "epoch": 4.766117647058824,
      "grad_norm": 0.40378919972930327,
      "learning_rate": 4.192324377230487e-05,
      "loss": 0.4902,
      "step": 10128
    },
    {
      "epoch": 4.766588235294118,
      "grad_norm": 0.3755654858000037,
      "learning_rate": 4.1892850508080824e-05,
      "loss": 0.3164,
      "step": 10129
    },
    {
      "epoch": 4.767058823529412,
      "grad_norm": 0.4110009965777808,
      "learning_rate": 4.18624664767376e-05,
      "loss": 0.166,
      "step": 10130
    },
    {
      "epoch": 4.767529411764706,
      "grad_norm": 0.41401666554191957,
      "learning_rate": 4.1832091680870214e-05,
      "loss": 0.3438,
      "step": 10131
    },
    {
      "epoch": 4.768,
      "grad_norm": 0.698832684464465,
      "learning_rate": 4.1801726123072696e-05,
      "loss": 0.5625,
      "step": 10132
    },
    {
      "epoch": 4.768470588235294,
      "grad_norm": 0.27372499545168644,
      "learning_rate": 4.1771369805938486e-05,
      "loss": 0.0854,
      "step": 10133
    },
    {
      "epoch": 4.768941176470588,
      "grad_norm": 0.27907615787955553,
      "learning_rate": 4.1741022732060204e-05,
      "loss": 0.2715,
      "step": 10134
    },
    {
      "epoch": 4.7694117647058825,
      "grad_norm": 0.3658413533691605,
      "learning_rate": 4.171068490402954e-05,
      "loss": 0.2988,
      "step": 10135
    },
    {
      "epoch": 4.769882352941177,
      "grad_norm": 0.20116019818511302,
      "learning_rate": 4.168035632443754e-05,
      "loss": 0.0306,
      "step": 10136
    },
    {
      "epoch": 4.770352941176471,
      "grad_norm": 0.3437519483695752,
      "learning_rate": 4.165003699587445e-05,
      "loss": 0.4082,
      "step": 10137
    },
    {
      "epoch": 4.770823529411764,
      "grad_norm": 0.3616207890848114,
      "learning_rate": 4.161972692092962e-05,
      "loss": 0.4512,
      "step": 10138
    },
    {
      "epoch": 4.7712941176470585,
      "grad_norm": 0.49739523159493937,
      "learning_rate": 4.158942610219175e-05,
      "loss": 0.7383,
      "step": 10139
    },
    {
      "epoch": 4.771764705882353,
      "grad_norm": 0.5719689496526058,
      "learning_rate": 4.155913454224862e-05,
      "loss": 0.2734,
      "step": 10140
    },
    {
      "epoch": 4.772235294117647,
      "grad_norm": 0.424117925123123,
      "learning_rate": 4.1528852243687276e-05,
      "loss": 0.2988,
      "step": 10141
    },
    {
      "epoch": 4.772705882352941,
      "grad_norm": 0.41579589346393014,
      "learning_rate": 4.1498579209094046e-05,
      "loss": 0.2471,
      "step": 10142
    },
    {
      "epoch": 4.773176470588235,
      "grad_norm": 0.4186730231096994,
      "learning_rate": 4.146831544105431e-05,
      "loss": 0.4219,
      "step": 10143
    },
    {
      "epoch": 4.77364705882353,
      "grad_norm": 0.42330322702185913,
      "learning_rate": 4.143806094215277e-05,
      "loss": 0.3574,
      "step": 10144
    },
    {
      "epoch": 4.774117647058824,
      "grad_norm": 0.37843061319407595,
      "learning_rate": 4.140781571497336e-05,
      "loss": 0.208,
      "step": 10145
    },
    {
      "epoch": 4.774588235294118,
      "grad_norm": 0.29391726363067366,
      "learning_rate": 4.137757976209911e-05,
      "loss": 0.1689,
      "step": 10146
    },
    {
      "epoch": 4.775058823529411,
      "grad_norm": 0.33986191156567586,
      "learning_rate": 4.134735308611232e-05,
      "loss": 0.2676,
      "step": 10147
    },
    {
      "epoch": 4.775529411764706,
      "grad_norm": 0.267309248123084,
      "learning_rate": 4.131713568959457e-05,
      "loss": 0.0571,
      "step": 10148
    },
    {
      "epoch": 4.776,
      "grad_norm": 0.4690676082048576,
      "learning_rate": 4.128692757512646e-05,
      "loss": 0.3926,
      "step": 10149
    },
    {
      "epoch": 4.776470588235294,
      "grad_norm": 0.39579359486700777,
      "learning_rate": 4.125672874528797e-05,
      "loss": 0.3008,
      "step": 10150
    },
    {
      "epoch": 4.776941176470588,
      "grad_norm": 0.47793033239621346,
      "learning_rate": 4.122653920265828e-05,
      "loss": 0.293,
      "step": 10151
    },
    {
      "epoch": 4.7774117647058825,
      "grad_norm": 0.4618159456364034,
      "learning_rate": 4.119635894981558e-05,
      "loss": 0.3789,
      "step": 10152
    },
    {
      "epoch": 4.777882352941177,
      "grad_norm": 0.4400280542647608,
      "learning_rate": 4.1166187989337596e-05,
      "loss": 0.416,
      "step": 10153
    },
    {
      "epoch": 4.778352941176471,
      "grad_norm": 0.3720404083783957,
      "learning_rate": 4.113602632380094e-05,
      "loss": 0.2119,
      "step": 10154
    },
    {
      "epoch": 4.778823529411765,
      "grad_norm": 0.5273414187780805,
      "learning_rate": 4.110587395578161e-05,
      "loss": 0.9258,
      "step": 10155
    },
    {
      "epoch": 4.7792941176470585,
      "grad_norm": 0.4316154560203269,
      "learning_rate": 4.1075730887854805e-05,
      "loss": 0.4766,
      "step": 10156
    },
    {
      "epoch": 4.779764705882353,
      "grad_norm": 0.2887068208912801,
      "learning_rate": 4.104559712259482e-05,
      "loss": 0.1025,
      "step": 10157
    },
    {
      "epoch": 4.780235294117647,
      "grad_norm": 0.19544681093596664,
      "learning_rate": 4.101547266257526e-05,
      "loss": 0.0762,
      "step": 10158
    },
    {
      "epoch": 4.780705882352941,
      "grad_norm": 0.272469952908167,
      "learning_rate": 4.098535751036894e-05,
      "loss": 0.207,
      "step": 10159
    },
    {
      "epoch": 4.781176470588235,
      "grad_norm": 0.36023946784826905,
      "learning_rate": 4.0955251668547754e-05,
      "loss": 0.3672,
      "step": 10160
    },
    {
      "epoch": 4.78164705882353,
      "grad_norm": 0.41192572017499257,
      "learning_rate": 4.092515513968294e-05,
      "loss": 0.4453,
      "step": 10161
    },
    {
      "epoch": 4.782117647058824,
      "grad_norm": 0.32427548316084814,
      "learning_rate": 4.0895067926344935e-05,
      "loss": 0.3379,
      "step": 10162
    },
    {
      "epoch": 4.782588235294118,
      "grad_norm": 0.33097377052996174,
      "learning_rate": 4.0864990031103235e-05,
      "loss": 0.3184,
      "step": 10163
    },
    {
      "epoch": 4.783058823529412,
      "grad_norm": 0.21159616565856687,
      "learning_rate": 4.0834921456526684e-05,
      "loss": 0.0957,
      "step": 10164
    },
    {
      "epoch": 4.783529411764706,
      "grad_norm": 0.730899396947326,
      "learning_rate": 4.080486220518336e-05,
      "loss": 0.5469,
      "step": 10165
    },
    {
      "epoch": 4.784,
      "grad_norm": 0.3046152924613556,
      "learning_rate": 4.0774812279640325e-05,
      "loss": 0.3105,
      "step": 10166
    },
    {
      "epoch": 4.784470588235294,
      "grad_norm": 0.48185286222232515,
      "learning_rate": 4.074477168246409e-05,
      "loss": 0.2617,
      "step": 10167
    },
    {
      "epoch": 4.784941176470588,
      "grad_norm": 0.5059433427171223,
      "learning_rate": 4.071474041622029e-05,
      "loss": 0.3652,
      "step": 10168
    },
    {
      "epoch": 4.7854117647058825,
      "grad_norm": 0.23793990235492324,
      "learning_rate": 4.0684718483473656e-05,
      "loss": 0.051,
      "step": 10169
    },
    {
      "epoch": 4.785882352941177,
      "grad_norm": 0.5178033841776363,
      "learning_rate": 4.06547058867883e-05,
      "loss": 0.6016,
      "step": 10170
    },
    {
      "epoch": 4.786352941176471,
      "grad_norm": 0.31989957801219737,
      "learning_rate": 4.0624702628727345e-05,
      "loss": 0.2324,
      "step": 10171
    },
    {
      "epoch": 4.786823529411764,
      "grad_norm": 0.3761416708481422,
      "learning_rate": 4.0594708711853267e-05,
      "loss": 0.1914,
      "step": 10172
    },
    {
      "epoch": 4.7872941176470585,
      "grad_norm": 0.2684822136090926,
      "learning_rate": 4.0564724138727734e-05,
      "loss": 0.125,
      "step": 10173
    },
    {
      "epoch": 4.787764705882353,
      "grad_norm": 0.533623486760203,
      "learning_rate": 4.0534748911911505e-05,
      "loss": 0.7383,
      "step": 10174
    },
    {
      "epoch": 4.788235294117647,
      "grad_norm": 0.3254780034837881,
      "learning_rate": 4.0504783033964645e-05,
      "loss": 0.1777,
      "step": 10175
    },
    {
      "epoch": 4.788705882352941,
      "grad_norm": 0.4260014483538682,
      "learning_rate": 4.047482650744641e-05,
      "loss": 0.3359,
      "step": 10176
    },
    {
      "epoch": 4.789176470588235,
      "grad_norm": 0.2961132353602182,
      "learning_rate": 4.044487933491518e-05,
      "loss": 0.2031,
      "step": 10177
    },
    {
      "epoch": 4.78964705882353,
      "grad_norm": 0.4214945462695894,
      "learning_rate": 4.041494151892862e-05,
      "loss": 0.3145,
      "step": 10178
    },
    {
      "epoch": 4.790117647058824,
      "grad_norm": 0.5145233901991327,
      "learning_rate": 4.038501306204361e-05,
      "loss": 0.4492,
      "step": 10179
    },
    {
      "epoch": 4.790588235294118,
      "grad_norm": 0.3723004469826672,
      "learning_rate": 4.0355093966816095e-05,
      "loss": 0.3418,
      "step": 10180
    },
    {
      "epoch": 4.791058823529411,
      "grad_norm": 0.4124618239994643,
      "learning_rate": 4.032518423580135e-05,
      "loss": 0.4297,
      "step": 10181
    },
    {
      "epoch": 4.791529411764706,
      "grad_norm": 0.2547303740031099,
      "learning_rate": 4.029528387155389e-05,
      "loss": 0.0913,
      "step": 10182
    },
    {
      "epoch": 4.792,
      "grad_norm": 0.5203317666900487,
      "learning_rate": 4.026539287662722e-05,
      "loss": 0.5898,
      "step": 10183
    },
    {
      "epoch": 4.792470588235294,
      "grad_norm": 0.41153756767904276,
      "learning_rate": 4.0235511253574256e-05,
      "loss": 0.3887,
      "step": 10184
    },
    {
      "epoch": 4.792941176470588,
      "grad_norm": 0.20241136990477337,
      "learning_rate": 4.020563900494707e-05,
      "loss": 0.027,
      "step": 10185
    },
    {
      "epoch": 4.7934117647058825,
      "grad_norm": 0.5855354873436825,
      "learning_rate": 4.01757761332968e-05,
      "loss": 0.5312,
      "step": 10186
    },
    {
      "epoch": 4.793882352941177,
      "grad_norm": 0.7760075141914563,
      "learning_rate": 4.0145922641173934e-05,
      "loss": 0.5312,
      "step": 10187
    },
    {
      "epoch": 4.794352941176471,
      "grad_norm": 0.4403139743410383,
      "learning_rate": 4.011607853112815e-05,
      "loss": 0.2256,
      "step": 10188
    },
    {
      "epoch": 4.794823529411765,
      "grad_norm": 9.186981310363372,
      "learning_rate": 4.008624380570819e-05,
      "loss": 2.7188,
      "step": 10189
    },
    {
      "epoch": 4.7952941176470585,
      "grad_norm": 0.3192529799937618,
      "learning_rate": 4.0056418467462184e-05,
      "loss": 0.3223,
      "step": 10190
    },
    {
      "epoch": 4.795764705882353,
      "grad_norm": 0.4531680612778442,
      "learning_rate": 4.002660251893727e-05,
      "loss": 0.2734,
      "step": 10191
    },
    {
      "epoch": 4.796235294117647,
      "grad_norm": 0.36239881479107067,
      "learning_rate": 3.9996795962679896e-05,
      "loss": 0.1167,
      "step": 10192
    },
    {
      "epoch": 4.796705882352941,
      "grad_norm": 0.4183866459031348,
      "learning_rate": 3.9966998801235766e-05,
      "loss": 0.2305,
      "step": 10193
    },
    {
      "epoch": 4.797176470588235,
      "grad_norm": 0.35799211977656714,
      "learning_rate": 3.9937211037149616e-05,
      "loss": 0.2471,
      "step": 10194
    },
    {
      "epoch": 4.79764705882353,
      "grad_norm": 0.514151971889005,
      "learning_rate": 3.9907432672965485e-05,
      "loss": 0.3711,
      "step": 10195
    },
    {
      "epoch": 4.798117647058824,
      "grad_norm": 0.3981826131629596,
      "learning_rate": 3.987766371122661e-05,
      "loss": 0.3301,
      "step": 10196
    },
    {
      "epoch": 4.798588235294117,
      "grad_norm": 4.12852257015085,
      "learning_rate": 3.984790415447539e-05,
      "loss": 0.6719,
      "step": 10197
    },
    {
      "epoch": 4.799058823529411,
      "grad_norm": 0.43214928167340516,
      "learning_rate": 3.9818154005253514e-05,
      "loss": 0.2793,
      "step": 10198
    },
    {
      "epoch": 4.799529411764706,
      "grad_norm": 0.39980190216719014,
      "learning_rate": 3.978841326610166e-05,
      "loss": 0.1504,
      "step": 10199
    },
    {
      "epoch": 4.8,
      "grad_norm": 0.482469654574777,
      "learning_rate": 3.97586819395599e-05,
      "loss": 0.3125,
      "step": 10200
    },
    {
      "epoch": 4.800470588235294,
      "grad_norm": 0.3391789361992673,
      "learning_rate": 3.972896002816749e-05,
      "loss": 0.0996,
      "step": 10201
    },
    {
      "epoch": 4.800941176470588,
      "grad_norm": 0.5452932400696332,
      "learning_rate": 3.969924753446271e-05,
      "loss": 0.2119,
      "step": 10202
    },
    {
      "epoch": 4.8014117647058825,
      "grad_norm": 0.579368441006046,
      "learning_rate": 3.966954446098322e-05,
      "loss": 0.7383,
      "step": 10203
    },
    {
      "epoch": 4.801882352941177,
      "grad_norm": 0.5710694972045934,
      "learning_rate": 3.9639850810265845e-05,
      "loss": 0.6719,
      "step": 10204
    },
    {
      "epoch": 4.802352941176471,
      "grad_norm": 0.43992550705191225,
      "learning_rate": 3.961016658484648e-05,
      "loss": 0.4121,
      "step": 10205
    },
    {
      "epoch": 4.802823529411764,
      "grad_norm": 0.3735133852983404,
      "learning_rate": 3.9580491787260346e-05,
      "loss": 0.1133,
      "step": 10206
    },
    {
      "epoch": 4.8032941176470585,
      "grad_norm": 0.32519572802957536,
      "learning_rate": 3.955082642004187e-05,
      "loss": 0.4375,
      "step": 10207
    },
    {
      "epoch": 4.803764705882353,
      "grad_norm": 0.16279337841878508,
      "learning_rate": 3.952117048572452e-05,
      "loss": 0.0461,
      "step": 10208
    },
    {
      "epoch": 4.804235294117647,
      "grad_norm": 0.5345636897046075,
      "learning_rate": 3.94915239868411e-05,
      "loss": 0.2773,
      "step": 10209
    },
    {
      "epoch": 4.804705882352941,
      "grad_norm": 0.2890485198261588,
      "learning_rate": 3.946188692592362e-05,
      "loss": 0.2471,
      "step": 10210
    },
    {
      "epoch": 4.805176470588235,
      "grad_norm": 0.415288995045094,
      "learning_rate": 3.943225930550315e-05,
      "loss": 0.2754,
      "step": 10211
    },
    {
      "epoch": 4.80564705882353,
      "grad_norm": 0.4617272467607213,
      "learning_rate": 3.9402641128110054e-05,
      "loss": 0.4688,
      "step": 10212
    },
    {
      "epoch": 4.806117647058824,
      "grad_norm": 0.3137741412935111,
      "learning_rate": 3.937303239627392e-05,
      "loss": 0.1338,
      "step": 10213
    },
    {
      "epoch": 4.806588235294118,
      "grad_norm": 0.40716626033130393,
      "learning_rate": 3.934343311252341e-05,
      "loss": 0.2324,
      "step": 10214
    },
    {
      "epoch": 4.807058823529411,
      "grad_norm": 0.40788266981290144,
      "learning_rate": 3.9313843279386486e-05,
      "loss": 0.2246,
      "step": 10215
    },
    {
      "epoch": 4.807529411764706,
      "grad_norm": 0.3387306135925566,
      "learning_rate": 3.9284262899390285e-05,
      "loss": 0.1885,
      "step": 10216
    },
    {
      "epoch": 4.808,
      "grad_norm": 0.41011619446723185,
      "learning_rate": 3.925469197506107e-05,
      "loss": 0.582,
      "step": 10217
    },
    {
      "epoch": 4.808470588235294,
      "grad_norm": 0.44233195580371504,
      "learning_rate": 3.922513050892434e-05,
      "loss": 0.3652,
      "step": 10218
    },
    {
      "epoch": 4.808941176470588,
      "grad_norm": 0.8577057143112184,
      "learning_rate": 3.919557850350486e-05,
      "loss": 0.4434,
      "step": 10219
    },
    {
      "epoch": 4.8094117647058825,
      "grad_norm": 0.4995438348880415,
      "learning_rate": 3.916603596132642e-05,
      "loss": 0.7266,
      "step": 10220
    },
    {
      "epoch": 4.809882352941177,
      "grad_norm": 0.37441409963510797,
      "learning_rate": 3.91365028849122e-05,
      "loss": 0.3008,
      "step": 10221
    },
    {
      "epoch": 4.810352941176471,
      "grad_norm": 0.3788796249130768,
      "learning_rate": 3.910697927678437e-05,
      "loss": 0.1416,
      "step": 10222
    },
    {
      "epoch": 4.810823529411765,
      "grad_norm": 0.1927985301718677,
      "learning_rate": 3.9077465139464446e-05,
      "loss": 0.0742,
      "step": 10223
    },
    {
      "epoch": 4.8112941176470585,
      "grad_norm": 0.35174380687461115,
      "learning_rate": 3.9047960475473095e-05,
      "loss": 0.1758,
      "step": 10224
    },
    {
      "epoch": 4.811764705882353,
      "grad_norm": 0.5786535659749811,
      "learning_rate": 3.901846528733011e-05,
      "loss": 0.4375,
      "step": 10225
    },
    {
      "epoch": 4.812235294117647,
      "grad_norm": 0.5154547146331261,
      "learning_rate": 3.898897957755454e-05,
      "loss": 0.252,
      "step": 10226
    },
    {
      "epoch": 4.812705882352941,
      "grad_norm": 0.355409437207723,
      "learning_rate": 3.895950334866468e-05,
      "loss": 0.2158,
      "step": 10227
    },
    {
      "epoch": 4.813176470588235,
      "grad_norm": 0.22493149557579392,
      "learning_rate": 3.893003660317783e-05,
      "loss": 0.0742,
      "step": 10228
    },
    {
      "epoch": 4.81364705882353,
      "grad_norm": 0.5186548782768176,
      "learning_rate": 3.8900579343610646e-05,
      "loss": 0.7227,
      "step": 10229
    },
    {
      "epoch": 4.814117647058824,
      "grad_norm": 0.4788141651511925,
      "learning_rate": 3.887113157247897e-05,
      "loss": 0.3711,
      "step": 10230
    },
    {
      "epoch": 4.814588235294117,
      "grad_norm": 0.4237022866915779,
      "learning_rate": 3.884169329229769e-05,
      "loss": 0.3496,
      "step": 10231
    },
    {
      "epoch": 4.815058823529411,
      "grad_norm": 0.5438830797215634,
      "learning_rate": 3.881226450558104e-05,
      "loss": 0.2451,
      "step": 10232
    },
    {
      "epoch": 4.815529411764706,
      "grad_norm": 0.39621233576054693,
      "learning_rate": 3.878284521484242e-05,
      "loss": 0.252,
      "step": 10233
    },
    {
      "epoch": 4.816,
      "grad_norm": 0.28754836024279173,
      "learning_rate": 3.8753435422594295e-05,
      "loss": 0.062,
      "step": 10234
    },
    {
      "epoch": 4.816470588235294,
      "grad_norm": 0.3397951043307862,
      "learning_rate": 3.8724035131348437e-05,
      "loss": 0.3066,
      "step": 10235
    },
    {
      "epoch": 4.816941176470588,
      "grad_norm": 0.37355943723588875,
      "learning_rate": 3.8694644343615836e-05,
      "loss": 0.127,
      "step": 10236
    },
    {
      "epoch": 4.8174117647058825,
      "grad_norm": 0.4893485121475639,
      "learning_rate": 3.866526306190651e-05,
      "loss": 0.457,
      "step": 10237
    },
    {
      "epoch": 4.817882352941177,
      "grad_norm": 0.46409255681358685,
      "learning_rate": 3.863589128872981e-05,
      "loss": 0.3984,
      "step": 10238
    },
    {
      "epoch": 4.818352941176471,
      "grad_norm": 0.22282560037010352,
      "learning_rate": 3.8606529026594276e-05,
      "loss": 0.042,
      "step": 10239
    },
    {
      "epoch": 4.818823529411764,
      "grad_norm": 0.3044409830811277,
      "learning_rate": 3.85771762780075e-05,
      "loss": 0.252,
      "step": 10240
    },
    {
      "epoch": 4.8192941176470585,
      "grad_norm": 0.48874631705642424,
      "learning_rate": 3.854783304547639e-05,
      "loss": 0.2441,
      "step": 10241
    },
    {
      "epoch": 4.819764705882353,
      "grad_norm": 0.4312841154894323,
      "learning_rate": 3.851849933150703e-05,
      "loss": 0.4785,
      "step": 10242
    },
    {
      "epoch": 4.820235294117647,
      "grad_norm": 0.46793235343564915,
      "learning_rate": 3.848917513860463e-05,
      "loss": 0.3047,
      "step": 10243
    },
    {
      "epoch": 4.820705882352941,
      "grad_norm": 0.4128757671369025,
      "learning_rate": 3.8459860469273666e-05,
      "loss": 0.3301,
      "step": 10244
    },
    {
      "epoch": 4.821176470588235,
      "grad_norm": 0.4171933488224796,
      "learning_rate": 3.8430555326017684e-05,
      "loss": 0.1982,
      "step": 10245
    },
    {
      "epoch": 4.82164705882353,
      "grad_norm": 0.574680086085898,
      "learning_rate": 3.840125971133952e-05,
      "loss": 0.3379,
      "step": 10246
    },
    {
      "epoch": 4.822117647058824,
      "grad_norm": 0.4590908198486815,
      "learning_rate": 3.8371973627741204e-05,
      "loss": 0.2148,
      "step": 10247
    },
    {
      "epoch": 4.822588235294118,
      "grad_norm": 0.2182002509133239,
      "learning_rate": 3.8342697077723845e-05,
      "loss": 0.0752,
      "step": 10248
    },
    {
      "epoch": 4.823058823529411,
      "grad_norm": 0.2929879820816739,
      "learning_rate": 3.8313430063787856e-05,
      "loss": 0.0991,
      "step": 10249
    },
    {
      "epoch": 4.823529411764706,
      "grad_norm": 0.08526546841341792,
      "learning_rate": 3.8284172588432716e-05,
      "loss": 0.0062,
      "step": 10250
    },
    {
      "epoch": 4.824,
      "grad_norm": 0.3975287633748878,
      "learning_rate": 3.8254924654157214e-05,
      "loss": 0.373,
      "step": 10251
    },
    {
      "epoch": 4.824470588235294,
      "grad_norm": 0.43641888170955345,
      "learning_rate": 3.8225686263459296e-05,
      "loss": 0.4023,
      "step": 10252
    },
    {
      "epoch": 4.824941176470588,
      "grad_norm": 0.24570289982389235,
      "learning_rate": 3.819645741883597e-05,
      "loss": 0.1416,
      "step": 10253
    },
    {
      "epoch": 4.8254117647058825,
      "grad_norm": 0.21190729516395082,
      "learning_rate": 3.816723812278358e-05,
      "loss": 0.0496,
      "step": 10254
    },
    {
      "epoch": 4.825882352941177,
      "grad_norm": 0.5838365964890732,
      "learning_rate": 3.813802837779762e-05,
      "loss": 0.5078,
      "step": 10255
    },
    {
      "epoch": 4.826352941176471,
      "grad_norm": 0.3809178761321305,
      "learning_rate": 3.810882818637268e-05,
      "loss": 0.249,
      "step": 10256
    },
    {
      "epoch": 4.826823529411765,
      "grad_norm": 0.847624475536907,
      "learning_rate": 3.807963755100263e-05,
      "loss": 0.582,
      "step": 10257
    },
    {
      "epoch": 4.8272941176470585,
      "grad_norm": 0.4110511217001908,
      "learning_rate": 3.8050456474180554e-05,
      "loss": 0.1777,
      "step": 10258
    },
    {
      "epoch": 4.827764705882353,
      "grad_norm": 0.43496324281505333,
      "learning_rate": 3.802128495839856e-05,
      "loss": 0.293,
      "step": 10259
    },
    {
      "epoch": 4.828235294117647,
      "grad_norm": 0.24104935764068183,
      "learning_rate": 3.799212300614809e-05,
      "loss": 0.0618,
      "step": 10260
    },
    {
      "epoch": 4.828705882352941,
      "grad_norm": 0.529635096569068,
      "learning_rate": 3.7962970619919745e-05,
      "loss": 0.4023,
      "step": 10261
    },
    {
      "epoch": 4.829176470588235,
      "grad_norm": 0.6307547130366797,
      "learning_rate": 3.7933827802203207e-05,
      "loss": 0.1514,
      "step": 10262
    },
    {
      "epoch": 4.82964705882353,
      "grad_norm": 0.4381142434628669,
      "learning_rate": 3.790469455548745e-05,
      "loss": 0.7227,
      "step": 10263
    },
    {
      "epoch": 4.830117647058824,
      "grad_norm": 0.4533044879829673,
      "learning_rate": 3.787557088226065e-05,
      "loss": 0.252,
      "step": 10264
    },
    {
      "epoch": 4.830588235294117,
      "grad_norm": 0.5228314511511168,
      "learning_rate": 3.7846456785010014e-05,
      "loss": 0.3809,
      "step": 10265
    },
    {
      "epoch": 4.831058823529411,
      "grad_norm": 0.30729166668680646,
      "learning_rate": 3.7817352266222086e-05,
      "loss": 0.1094,
      "step": 10266
    },
    {
      "epoch": 4.831529411764706,
      "grad_norm": 0.3224992292697546,
      "learning_rate": 3.778825732838257e-05,
      "loss": 0.2734,
      "step": 10267
    },
    {
      "epoch": 4.832,
      "grad_norm": 0.4029616101902434,
      "learning_rate": 3.775917197397622e-05,
      "loss": 0.1338,
      "step": 10268
    },
    {
      "epoch": 4.832470588235294,
      "grad_norm": 0.8175151902412915,
      "learning_rate": 3.773009620548718e-05,
      "loss": 0.6016,
      "step": 10269
    },
    {
      "epoch": 4.832941176470588,
      "grad_norm": 0.4197251376671563,
      "learning_rate": 3.770103002539857e-05,
      "loss": 0.1924,
      "step": 10270
    },
    {
      "epoch": 4.8334117647058825,
      "grad_norm": 0.4099841859328283,
      "learning_rate": 3.767197343619279e-05,
      "loss": 0.4629,
      "step": 10271
    },
    {
      "epoch": 4.833882352941177,
      "grad_norm": 0.39903774054712565,
      "learning_rate": 3.7642926440351514e-05,
      "loss": 0.1523,
      "step": 10272
    },
    {
      "epoch": 4.834352941176471,
      "grad_norm": 0.7444937138023686,
      "learning_rate": 3.761388904035537e-05,
      "loss": 0.5586,
      "step": 10273
    },
    {
      "epoch": 4.834823529411764,
      "grad_norm": 0.5209670873774115,
      "learning_rate": 3.758486123868435e-05,
      "loss": 0.3105,
      "step": 10274
    },
    {
      "epoch": 4.8352941176470585,
      "grad_norm": 0.3034068116114575,
      "learning_rate": 3.755584303781763e-05,
      "loss": 0.3184,
      "step": 10275
    },
    {
      "epoch": 4.835764705882353,
      "grad_norm": 0.468146399953602,
      "learning_rate": 3.7526834440233395e-05,
      "loss": 0.3672,
      "step": 10276
    },
    {
      "epoch": 4.836235294117647,
      "grad_norm": 0.7447211379328381,
      "learning_rate": 3.749783544840918e-05,
      "loss": 0.6484,
      "step": 10277
    },
    {
      "epoch": 4.836705882352941,
      "grad_norm": 0.23191577847443107,
      "learning_rate": 3.746884606482169e-05,
      "loss": 0.0613,
      "step": 10278
    },
    {
      "epoch": 4.837176470588235,
      "grad_norm": 1.2435735974646553,
      "learning_rate": 3.743986629194666e-05,
      "loss": 0.2773,
      "step": 10279
    },
    {
      "epoch": 4.83764705882353,
      "grad_norm": 0.7069084638727526,
      "learning_rate": 3.741089613225914e-05,
      "loss": 0.7773,
      "step": 10280
    },
    {
      "epoch": 4.838117647058824,
      "grad_norm": 0.32035251917067287,
      "learning_rate": 3.7381935588233385e-05,
      "loss": 0.1523,
      "step": 10281
    },
    {
      "epoch": 4.838588235294118,
      "grad_norm": 0.3309958807536925,
      "learning_rate": 3.735298466234269e-05,
      "loss": 0.5117,
      "step": 10282
    },
    {
      "epoch": 4.839058823529411,
      "grad_norm": 0.7400490778879562,
      "learning_rate": 3.732404335705962e-05,
      "loss": 0.6836,
      "step": 10283
    },
    {
      "epoch": 4.839529411764706,
      "grad_norm": 0.457385660054626,
      "learning_rate": 3.729511167485597e-05,
      "loss": 0.6797,
      "step": 10284
    },
    {
      "epoch": 4.84,
      "grad_norm": 0.21533193669663175,
      "learning_rate": 3.72661896182025e-05,
      "loss": 0.0371,
      "step": 10285
    },
    {
      "epoch": 4.840470588235294,
      "grad_norm": 0.29807807179625995,
      "learning_rate": 3.7237277189569505e-05,
      "loss": 0.1074,
      "step": 10286
    },
    {
      "epoch": 4.840941176470588,
      "grad_norm": 0.34764176094591975,
      "learning_rate": 3.720837439142608e-05,
      "loss": 0.1006,
      "step": 10287
    },
    {
      "epoch": 4.8414117647058825,
      "grad_norm": 0.6708349664969264,
      "learning_rate": 3.717948122624072e-05,
      "loss": 0.4141,
      "step": 10288
    },
    {
      "epoch": 4.841882352941177,
      "grad_norm": 0.28810628559299495,
      "learning_rate": 3.7150597696481076e-05,
      "loss": 0.1895,
      "step": 10289
    },
    {
      "epoch": 4.842352941176471,
      "grad_norm": 0.37193767739689493,
      "learning_rate": 3.712172380461387e-05,
      "loss": 0.3301,
      "step": 10290
    },
    {
      "epoch": 4.842823529411765,
      "grad_norm": 0.2108686644784852,
      "learning_rate": 3.709285955310513e-05,
      "loss": 0.062,
      "step": 10291
    },
    {
      "epoch": 4.8432941176470585,
      "grad_norm": 0.6513058623567664,
      "learning_rate": 3.7064004944420026e-05,
      "loss": 0.3867,
      "step": 10292
    },
    {
      "epoch": 4.843764705882353,
      "grad_norm": 0.36993505693347284,
      "learning_rate": 3.703515998102281e-05,
      "loss": 0.4492,
      "step": 10293
    },
    {
      "epoch": 4.844235294117647,
      "grad_norm": 0.2319464101933329,
      "learning_rate": 3.7006324665377025e-05,
      "loss": 0.0693,
      "step": 10294
    },
    {
      "epoch": 4.844705882352941,
      "grad_norm": 0.38425383675688185,
      "learning_rate": 3.6977498999945384e-05,
      "loss": 0.1885,
      "step": 10295
    },
    {
      "epoch": 4.845176470588235,
      "grad_norm": 0.6141954305351097,
      "learning_rate": 3.694868298718965e-05,
      "loss": 0.2832,
      "step": 10296
    },
    {
      "epoch": 4.84564705882353,
      "grad_norm": 0.19356595577261404,
      "learning_rate": 3.6919876629570905e-05,
      "loss": 0.0493,
      "step": 10297
    },
    {
      "epoch": 4.846117647058824,
      "grad_norm": 0.5150528004972327,
      "learning_rate": 3.68910799295494e-05,
      "loss": 0.3652,
      "step": 10298
    },
    {
      "epoch": 4.846588235294117,
      "grad_norm": 0.3278445540000696,
      "learning_rate": 3.686229288958442e-05,
      "loss": 0.2266,
      "step": 10299
    },
    {
      "epoch": 4.847058823529411,
      "grad_norm": 0.3083200618594993,
      "learning_rate": 3.6833515512134606e-05,
      "loss": 0.0977,
      "step": 10300
    },
    {
      "epoch": 4.847529411764706,
      "grad_norm": 0.17271657579846536,
      "learning_rate": 3.6804747799657594e-05,
      "loss": 0.0571,
      "step": 10301
    },
    {
      "epoch": 4.848,
      "grad_norm": 0.41983932567882787,
      "learning_rate": 3.677598975461036e-05,
      "loss": 0.1826,
      "step": 10302
    },
    {
      "epoch": 4.848470588235294,
      "grad_norm": 0.3067565511315871,
      "learning_rate": 3.6747241379448986e-05,
      "loss": 0.3438,
      "step": 10303
    },
    {
      "epoch": 4.848941176470588,
      "grad_norm": 0.37435464661892626,
      "learning_rate": 3.671850267662866e-05,
      "loss": 0.3008,
      "step": 10304
    },
    {
      "epoch": 4.8494117647058825,
      "grad_norm": 0.34260890937881366,
      "learning_rate": 3.668977364860386e-05,
      "loss": 0.543,
      "step": 10305
    },
    {
      "epoch": 4.849882352941177,
      "grad_norm": 0.33564537678610545,
      "learning_rate": 3.666105429782821e-05,
      "loss": 0.2715,
      "step": 10306
    },
    {
      "epoch": 4.850352941176471,
      "grad_norm": 0.29617395934665536,
      "learning_rate": 3.6632344626754405e-05,
      "loss": 0.2598,
      "step": 10307
    },
    {
      "epoch": 4.850823529411764,
      "grad_norm": 0.6673164039386025,
      "learning_rate": 3.660364463783443e-05,
      "loss": 0.4434,
      "step": 10308
    },
    {
      "epoch": 4.8512941176470585,
      "grad_norm": 0.3796318181333085,
      "learning_rate": 3.657495433351945e-05,
      "loss": 0.4277,
      "step": 10309
    },
    {
      "epoch": 4.851764705882353,
      "grad_norm": 0.5525262335207453,
      "learning_rate": 3.6546273716259685e-05,
      "loss": 0.7344,
      "step": 10310
    },
    {
      "epoch": 4.852235294117647,
      "grad_norm": 0.42697960388503503,
      "learning_rate": 3.6517602788504613e-05,
      "loss": 0.5156,
      "step": 10311
    },
    {
      "epoch": 4.852705882352941,
      "grad_norm": 0.4325968783489871,
      "learning_rate": 3.648894155270295e-05,
      "loss": 0.3047,
      "step": 10312
    },
    {
      "epoch": 4.853176470588235,
      "grad_norm": 0.3573125591275935,
      "learning_rate": 3.64602900113024e-05,
      "loss": 0.3516,
      "step": 10313
    },
    {
      "epoch": 4.85364705882353,
      "grad_norm": 0.44869340256047935,
      "learning_rate": 3.643164816674999e-05,
      "loss": 0.5078,
      "step": 10314
    },
    {
      "epoch": 4.854117647058824,
      "grad_norm": 0.31284239845602574,
      "learning_rate": 3.6403016021491924e-05,
      "loss": 0.2051,
      "step": 10315
    },
    {
      "epoch": 4.854588235294118,
      "grad_norm": 0.3098473200331777,
      "learning_rate": 3.637439357797343e-05,
      "loss": 0.0649,
      "step": 10316
    },
    {
      "epoch": 4.855058823529411,
      "grad_norm": 0.47803254294709857,
      "learning_rate": 3.634578083863905e-05,
      "loss": 0.4141,
      "step": 10317
    },
    {
      "epoch": 4.855529411764706,
      "grad_norm": 0.2598219390470127,
      "learning_rate": 3.631717780593252e-05,
      "loss": 0.0557,
      "step": 10318
    },
    {
      "epoch": 4.856,
      "grad_norm": 0.2894106005968788,
      "learning_rate": 3.628858448229655e-05,
      "loss": 0.2539,
      "step": 10319
    },
    {
      "epoch": 4.856470588235294,
      "grad_norm": 0.31260246145242765,
      "learning_rate": 3.6260000870173266e-05,
      "loss": 0.1533,
      "step": 10320
    },
    {
      "epoch": 4.856941176470588,
      "grad_norm": 0.25951599495536753,
      "learning_rate": 3.6231426972003766e-05,
      "loss": 0.1973,
      "step": 10321
    },
    {
      "epoch": 4.8574117647058825,
      "grad_norm": 0.499697895135527,
      "learning_rate": 3.620286279022843e-05,
      "loss": 0.4785,
      "step": 10322
    },
    {
      "epoch": 4.857882352941177,
      "grad_norm": 0.43592839047281556,
      "learning_rate": 3.617430832728682e-05,
      "loss": 0.3438,
      "step": 10323
    },
    {
      "epoch": 4.85835294117647,
      "grad_norm": 0.21956723231280378,
      "learning_rate": 3.614576358561757e-05,
      "loss": 0.0664,
      "step": 10324
    },
    {
      "epoch": 4.858823529411764,
      "grad_norm": 0.38058049088070084,
      "learning_rate": 3.611722856765855e-05,
      "loss": 0.3008,
      "step": 10325
    },
    {
      "epoch": 4.8592941176470585,
      "grad_norm": 0.49794443096369295,
      "learning_rate": 3.608870327584683e-05,
      "loss": 0.4688,
      "step": 10326
    },
    {
      "epoch": 4.859764705882353,
      "grad_norm": 0.4125022727258155,
      "learning_rate": 3.606018771261856e-05,
      "loss": 0.3242,
      "step": 10327
    },
    {
      "epoch": 4.860235294117647,
      "grad_norm": 0.15330950742729754,
      "learning_rate": 3.6031681880409106e-05,
      "loss": 0.0245,
      "step": 10328
    },
    {
      "epoch": 4.860705882352941,
      "grad_norm": 0.5481500771586074,
      "learning_rate": 3.6003185781653097e-05,
      "loss": 0.3379,
      "step": 10329
    },
    {
      "epoch": 4.861176470588235,
      "grad_norm": 0.4448441810043439,
      "learning_rate": 3.5974699418784084e-05,
      "loss": 0.3066,
      "step": 10330
    },
    {
      "epoch": 4.86164705882353,
      "grad_norm": 0.3481337350058846,
      "learning_rate": 3.5946222794235105e-05,
      "loss": 0.3047,
      "step": 10331
    },
    {
      "epoch": 4.862117647058824,
      "grad_norm": 0.531477236527567,
      "learning_rate": 3.5917755910438094e-05,
      "loss": 0.543,
      "step": 10332
    },
    {
      "epoch": 4.862588235294117,
      "grad_norm": 0.44810462632398607,
      "learning_rate": 3.58892987698243e-05,
      "loss": 0.3281,
      "step": 10333
    },
    {
      "epoch": 4.863058823529411,
      "grad_norm": 0.6323492254169368,
      "learning_rate": 3.5860851374824134e-05,
      "loss": 0.6055,
      "step": 10334
    },
    {
      "epoch": 4.863529411764706,
      "grad_norm": 0.3736588885529651,
      "learning_rate": 3.5832413727867075e-05,
      "loss": 0.293,
      "step": 10335
    },
    {
      "epoch": 4.864,
      "grad_norm": 0.5543331620868417,
      "learning_rate": 3.580398583138188e-05,
      "loss": 0.4609,
      "step": 10336
    },
    {
      "epoch": 4.864470588235294,
      "grad_norm": 0.3246592112276442,
      "learning_rate": 3.577556768779646e-05,
      "loss": 0.0903,
      "step": 10337
    },
    {
      "epoch": 4.864941176470588,
      "grad_norm": 0.33650651768673734,
      "learning_rate": 3.574715929953778e-05,
      "loss": 0.2969,
      "step": 10338
    },
    {
      "epoch": 4.8654117647058825,
      "grad_norm": 0.5103079618795893,
      "learning_rate": 3.571876066903211e-05,
      "loss": 0.5547,
      "step": 10339
    },
    {
      "epoch": 4.865882352941177,
      "grad_norm": 0.7337725327567186,
      "learning_rate": 3.569037179870487e-05,
      "loss": 0.4629,
      "step": 10340
    },
    {
      "epoch": 4.866352941176471,
      "grad_norm": 0.5261750943229802,
      "learning_rate": 3.5661992690980506e-05,
      "loss": 0.5859,
      "step": 10341
    },
    {
      "epoch": 4.866823529411764,
      "grad_norm": 0.5230940885499928,
      "learning_rate": 3.56336233482828e-05,
      "loss": 0.3203,
      "step": 10342
    },
    {
      "epoch": 4.8672941176470585,
      "grad_norm": 0.28646535382947574,
      "learning_rate": 3.560526377303467e-05,
      "loss": 0.0894,
      "step": 10343
    },
    {
      "epoch": 4.867764705882353,
      "grad_norm": 0.36590521719511254,
      "learning_rate": 3.557691396765807e-05,
      "loss": 0.2676,
      "step": 10344
    },
    {
      "epoch": 4.868235294117647,
      "grad_norm": 0.35085857590722247,
      "learning_rate": 3.554857393457426e-05,
      "loss": 0.3652,
      "step": 10345
    },
    {
      "epoch": 4.868705882352941,
      "grad_norm": 0.6719898430512952,
      "learning_rate": 3.5520243676203664e-05,
      "loss": 0.5742,
      "step": 10346
    },
    {
      "epoch": 4.869176470588235,
      "grad_norm": 0.30584964526344377,
      "learning_rate": 3.5491923194965735e-05,
      "loss": 0.1016,
      "step": 10347
    },
    {
      "epoch": 4.86964705882353,
      "grad_norm": 0.4041502289504116,
      "learning_rate": 3.546361249327922e-05,
      "loss": 0.3086,
      "step": 10348
    },
    {
      "epoch": 4.870117647058824,
      "grad_norm": 0.2755655756707276,
      "learning_rate": 3.5435311573562024e-05,
      "loss": 0.1226,
      "step": 10349
    },
    {
      "epoch": 4.870588235294118,
      "grad_norm": 0.38240964259253524,
      "learning_rate": 3.540702043823113e-05,
      "loss": 0.3691,
      "step": 10350
    },
    {
      "epoch": 4.871058823529411,
      "grad_norm": 0.31005950023086726,
      "learning_rate": 3.537873908970282e-05,
      "loss": 0.127,
      "step": 10351
    },
    {
      "epoch": 4.871529411764706,
      "grad_norm": 0.2950018344496935,
      "learning_rate": 3.535046753039235e-05,
      "loss": 0.209,
      "step": 10352
    },
    {
      "epoch": 4.872,
      "grad_norm": 0.34426447141740635,
      "learning_rate": 3.53222057627143e-05,
      "loss": 0.0747,
      "step": 10353
    },
    {
      "epoch": 4.872470588235294,
      "grad_norm": 0.385887494817862,
      "learning_rate": 3.5293953789082425e-05,
      "loss": 0.2129,
      "step": 10354
    },
    {
      "epoch": 4.872941176470588,
      "grad_norm": 0.21647343927506899,
      "learning_rate": 3.5265711611909476e-05,
      "loss": 0.0544,
      "step": 10355
    },
    {
      "epoch": 4.8734117647058826,
      "grad_norm": 0.4817218964750303,
      "learning_rate": 3.5237479233607525e-05,
      "loss": 0.6211,
      "step": 10356
    },
    {
      "epoch": 4.873882352941177,
      "grad_norm": 0.7504061614700912,
      "learning_rate": 3.520925665658779e-05,
      "loss": 0.4453,
      "step": 10357
    },
    {
      "epoch": 4.87435294117647,
      "grad_norm": 0.34265930493227703,
      "learning_rate": 3.518104388326055e-05,
      "loss": 0.4551,
      "step": 10358
    },
    {
      "epoch": 4.874823529411764,
      "grad_norm": 0.20551565236150746,
      "learning_rate": 3.5152840916035344e-05,
      "loss": 0.0442,
      "step": 10359
    },
    {
      "epoch": 4.875294117647059,
      "grad_norm": 0.44084739590058825,
      "learning_rate": 3.5124647757320883e-05,
      "loss": 0.3242,
      "step": 10360
    },
    {
      "epoch": 4.875764705882353,
      "grad_norm": 0.3176260136541746,
      "learning_rate": 3.509646440952492e-05,
      "loss": 0.1357,
      "step": 10361
    },
    {
      "epoch": 4.876235294117647,
      "grad_norm": 0.41002140224765665,
      "learning_rate": 3.506829087505451e-05,
      "loss": 0.3008,
      "step": 10362
    },
    {
      "epoch": 4.876705882352941,
      "grad_norm": 0.3821484929974657,
      "learning_rate": 3.504012715631583e-05,
      "loss": 0.2441,
      "step": 10363
    },
    {
      "epoch": 4.8771764705882354,
      "grad_norm": 0.47981833423332687,
      "learning_rate": 3.501197325571411e-05,
      "loss": 0.6914,
      "step": 10364
    },
    {
      "epoch": 4.87764705882353,
      "grad_norm": 0.38484324431742134,
      "learning_rate": 3.49838291756539e-05,
      "loss": 0.2871,
      "step": 10365
    },
    {
      "epoch": 4.878117647058824,
      "grad_norm": 0.36026593946968816,
      "learning_rate": 3.495569491853887e-05,
      "loss": 0.252,
      "step": 10366
    },
    {
      "epoch": 4.878588235294117,
      "grad_norm": 0.4407857181382575,
      "learning_rate": 3.492757048677175e-05,
      "loss": 0.1904,
      "step": 10367
    },
    {
      "epoch": 4.8790588235294114,
      "grad_norm": 0.47622590992819314,
      "learning_rate": 3.4899455882754526e-05,
      "loss": 0.1689,
      "step": 10368
    },
    {
      "epoch": 4.879529411764706,
      "grad_norm": 0.7349373543523117,
      "learning_rate": 3.4871351108888375e-05,
      "loss": 0.668,
      "step": 10369
    },
    {
      "epoch": 4.88,
      "grad_norm": 0.2756916026910254,
      "learning_rate": 3.4843256167573514e-05,
      "loss": 0.3086,
      "step": 10370
    },
    {
      "epoch": 4.880470588235294,
      "grad_norm": 0.44694193980344304,
      "learning_rate": 3.481517106120944e-05,
      "loss": 0.3164,
      "step": 10371
    },
    {
      "epoch": 4.880941176470588,
      "grad_norm": 0.24887060359147645,
      "learning_rate": 3.478709579219472e-05,
      "loss": 0.0815,
      "step": 10372
    },
    {
      "epoch": 4.881411764705883,
      "grad_norm": 0.389475036692713,
      "learning_rate": 3.4759030362927115e-05,
      "loss": 0.3145,
      "step": 10373
    },
    {
      "epoch": 4.881882352941177,
      "grad_norm": 0.20174207057960739,
      "learning_rate": 3.4730974775803574e-05,
      "loss": 0.0293,
      "step": 10374
    },
    {
      "epoch": 4.882352941176471,
      "grad_norm": 0.37284552946557403,
      "learning_rate": 3.4702929033220174e-05,
      "loss": 0.1592,
      "step": 10375
    },
    {
      "epoch": 4.882823529411764,
      "grad_norm": 0.5499686295484199,
      "learning_rate": 3.467489313757217e-05,
      "loss": 0.2988,
      "step": 10376
    },
    {
      "epoch": 4.883294117647059,
      "grad_norm": 0.34914180196730626,
      "learning_rate": 3.4646867091254e-05,
      "loss": 0.373,
      "step": 10377
    },
    {
      "epoch": 4.883764705882353,
      "grad_norm": 0.3059706244031491,
      "learning_rate": 3.4618850896659135e-05,
      "loss": 0.3457,
      "step": 10378
    },
    {
      "epoch": 4.884235294117647,
      "grad_norm": 0.3249994167863996,
      "learning_rate": 3.4590844556180395e-05,
      "loss": 0.3398,
      "step": 10379
    },
    {
      "epoch": 4.884705882352941,
      "grad_norm": 0.8933732062092652,
      "learning_rate": 3.456284807220958e-05,
      "loss": 1.2891,
      "step": 10380
    },
    {
      "epoch": 4.8851764705882355,
      "grad_norm": 0.31525185510694653,
      "learning_rate": 3.453486144713772e-05,
      "loss": 0.0933,
      "step": 10381
    },
    {
      "epoch": 4.88564705882353,
      "grad_norm": 0.4322478751993242,
      "learning_rate": 3.450688468335512e-05,
      "loss": 0.3105,
      "step": 10382
    },
    {
      "epoch": 4.886117647058824,
      "grad_norm": 0.5054479666522824,
      "learning_rate": 3.4478917783251e-05,
      "loss": 0.3105,
      "step": 10383
    },
    {
      "epoch": 4.886588235294118,
      "grad_norm": 0.23171892694197557,
      "learning_rate": 3.445096074921393e-05,
      "loss": 0.1025,
      "step": 10384
    },
    {
      "epoch": 4.8870588235294115,
      "grad_norm": 0.6039245827342317,
      "learning_rate": 3.442301358363163e-05,
      "loss": 0.3906,
      "step": 10385
    },
    {
      "epoch": 4.887529411764706,
      "grad_norm": 0.30798626374620247,
      "learning_rate": 3.4395076288890825e-05,
      "loss": 0.1201,
      "step": 10386
    },
    {
      "epoch": 4.888,
      "grad_norm": 0.37171778925239973,
      "learning_rate": 3.436714886737753e-05,
      "loss": 0.2812,
      "step": 10387
    },
    {
      "epoch": 4.888470588235294,
      "grad_norm": 0.3872107881119503,
      "learning_rate": 3.4339231321476965e-05,
      "loss": 0.2773,
      "step": 10388
    },
    {
      "epoch": 4.888941176470588,
      "grad_norm": 0.27561050778713736,
      "learning_rate": 3.43113236535733e-05,
      "loss": 0.0801,
      "step": 10389
    },
    {
      "epoch": 4.889411764705883,
      "grad_norm": 0.34771386522024944,
      "learning_rate": 3.428342586605005e-05,
      "loss": 0.3672,
      "step": 10390
    },
    {
      "epoch": 4.889882352941177,
      "grad_norm": 0.4956512890560088,
      "learning_rate": 3.425553796128985e-05,
      "loss": 0.3887,
      "step": 10391
    },
    {
      "epoch": 4.89035294117647,
      "grad_norm": 0.22649981355944743,
      "learning_rate": 3.42276599416744e-05,
      "loss": 0.0576,
      "step": 10392
    },
    {
      "epoch": 4.890823529411764,
      "grad_norm": 0.403399195650498,
      "learning_rate": 3.419979180958464e-05,
      "loss": 0.2363,
      "step": 10393
    },
    {
      "epoch": 4.891294117647059,
      "grad_norm": 0.3268671433715739,
      "learning_rate": 3.417193356740071e-05,
      "loss": 0.0918,
      "step": 10394
    },
    {
      "epoch": 4.891764705882353,
      "grad_norm": 0.3241897853980765,
      "learning_rate": 3.414408521750173e-05,
      "loss": 0.2383,
      "step": 10395
    },
    {
      "epoch": 4.892235294117647,
      "grad_norm": 0.3835580863834809,
      "learning_rate": 3.411624676226616e-05,
      "loss": 0.25,
      "step": 10396
    },
    {
      "epoch": 4.892705882352941,
      "grad_norm": 0.30375527876830416,
      "learning_rate": 3.408841820407157e-05,
      "loss": 0.293,
      "step": 10397
    },
    {
      "epoch": 4.8931764705882355,
      "grad_norm": 0.4337464866095333,
      "learning_rate": 3.406059954529454e-05,
      "loss": 0.3535,
      "step": 10398
    },
    {
      "epoch": 4.89364705882353,
      "grad_norm": 0.31821738826740276,
      "learning_rate": 3.403279078831101e-05,
      "loss": 0.291,
      "step": 10399
    },
    {
      "epoch": 4.894117647058824,
      "grad_norm": 0.4967256957296153,
      "learning_rate": 3.4004991935496004e-05,
      "loss": 0.4551,
      "step": 10400
    },
    {
      "epoch": 4.894588235294117,
      "grad_norm": 0.4489370470069187,
      "learning_rate": 3.397720298922358e-05,
      "loss": 0.4961,
      "step": 10401
    },
    {
      "epoch": 4.8950588235294115,
      "grad_norm": 0.6042477380165354,
      "learning_rate": 3.394942395186716e-05,
      "loss": 0.4922,
      "step": 10402
    },
    {
      "epoch": 4.895529411764706,
      "grad_norm": 0.3312965833046754,
      "learning_rate": 3.392165482579912e-05,
      "loss": 0.4512,
      "step": 10403
    },
    {
      "epoch": 4.896,
      "grad_norm": 0.2944899422918263,
      "learning_rate": 3.389389561339112e-05,
      "loss": 0.0815,
      "step": 10404
    },
    {
      "epoch": 4.896470588235294,
      "grad_norm": 0.4040220887289951,
      "learning_rate": 3.3866146317013976e-05,
      "loss": 0.209,
      "step": 10405
    },
    {
      "epoch": 4.896941176470588,
      "grad_norm": 0.5311687315963006,
      "learning_rate": 3.3838406939037525e-05,
      "loss": 0.918,
      "step": 10406
    },
    {
      "epoch": 4.897411764705883,
      "grad_norm": 0.2720169836777457,
      "learning_rate": 3.381067748183087e-05,
      "loss": 0.166,
      "step": 10407
    },
    {
      "epoch": 4.897882352941177,
      "grad_norm": 0.4171200828411228,
      "learning_rate": 3.3782957947762336e-05,
      "loss": 0.1445,
      "step": 10408
    },
    {
      "epoch": 4.898352941176471,
      "grad_norm": 0.4633729218246374,
      "learning_rate": 3.375524833919918e-05,
      "loss": 0.2266,
      "step": 10409
    },
    {
      "epoch": 4.898823529411764,
      "grad_norm": 0.49568961658361,
      "learning_rate": 3.372754865850799e-05,
      "loss": 0.4355,
      "step": 10410
    },
    {
      "epoch": 4.899294117647059,
      "grad_norm": 0.263487541557602,
      "learning_rate": 3.3699858908054506e-05,
      "loss": 0.1816,
      "step": 10411
    },
    {
      "epoch": 4.899764705882353,
      "grad_norm": 0.23872221774103325,
      "learning_rate": 3.367217909020348e-05,
      "loss": 0.105,
      "step": 10412
    },
    {
      "epoch": 4.900235294117647,
      "grad_norm": 0.3556636055109967,
      "learning_rate": 3.364450920731894e-05,
      "loss": 0.1855,
      "step": 10413
    },
    {
      "epoch": 4.900705882352941,
      "grad_norm": 0.31241518810000357,
      "learning_rate": 3.361684926176408e-05,
      "loss": 0.248,
      "step": 10414
    },
    {
      "epoch": 4.9011764705882355,
      "grad_norm": 0.3788076370475698,
      "learning_rate": 3.35891992559011e-05,
      "loss": 0.1465,
      "step": 10415
    },
    {
      "epoch": 4.90164705882353,
      "grad_norm": 0.35364186466071634,
      "learning_rate": 3.35615591920915e-05,
      "loss": 0.4316,
      "step": 10416
    },
    {
      "epoch": 4.902117647058824,
      "grad_norm": 0.29528424447088153,
      "learning_rate": 3.353392907269591e-05,
      "loss": 0.1025,
      "step": 10417
    },
    {
      "epoch": 4.902588235294118,
      "grad_norm": 0.46699625118867505,
      "learning_rate": 3.3506308900073964e-05,
      "loss": 0.4668,
      "step": 10418
    },
    {
      "epoch": 4.9030588235294115,
      "grad_norm": 0.32203080780263904,
      "learning_rate": 3.3478698676584714e-05,
      "loss": 0.0742,
      "step": 10419
    },
    {
      "epoch": 4.903529411764706,
      "grad_norm": 0.5303074023548132,
      "learning_rate": 3.34510984045861e-05,
      "loss": 0.5625,
      "step": 10420
    },
    {
      "epoch": 4.904,
      "grad_norm": 0.5802607584937796,
      "learning_rate": 3.3423508086435326e-05,
      "loss": 0.1826,
      "step": 10421
    },
    {
      "epoch": 4.904470588235294,
      "grad_norm": 0.3891140412871733,
      "learning_rate": 3.339592772448883e-05,
      "loss": 0.2812,
      "step": 10422
    },
    {
      "epoch": 4.904941176470588,
      "grad_norm": 0.7170705908343284,
      "learning_rate": 3.3368357321102e-05,
      "loss": 0.332,
      "step": 10423
    },
    {
      "epoch": 4.905411764705883,
      "grad_norm": 0.7673466184306974,
      "learning_rate": 3.334079687862952e-05,
      "loss": 0.2891,
      "step": 10424
    },
    {
      "epoch": 4.905882352941177,
      "grad_norm": 0.41619632756154024,
      "learning_rate": 3.331324639942526e-05,
      "loss": 0.4043,
      "step": 10425
    },
    {
      "epoch": 4.90635294117647,
      "grad_norm": 0.1907351996357228,
      "learning_rate": 3.328570588584204e-05,
      "loss": 0.0356,
      "step": 10426
    },
    {
      "epoch": 4.906823529411764,
      "grad_norm": 0.38858590520048425,
      "learning_rate": 3.325817534023202e-05,
      "loss": 0.2949,
      "step": 10427
    },
    {
      "epoch": 4.907294117647059,
      "grad_norm": 0.5240083645349513,
      "learning_rate": 3.32306547649465e-05,
      "loss": 0.582,
      "step": 10428
    },
    {
      "epoch": 4.907764705882353,
      "grad_norm": 0.48441991799059414,
      "learning_rate": 3.320314416233574e-05,
      "loss": 0.5156,
      "step": 10429
    },
    {
      "epoch": 4.908235294117647,
      "grad_norm": 0.38475288329101465,
      "learning_rate": 3.317564353474943e-05,
      "loss": 0.1777,
      "step": 10430
    },
    {
      "epoch": 4.908705882352941,
      "grad_norm": 0.4433722359324539,
      "learning_rate": 3.314815288453613e-05,
      "loss": 0.1709,
      "step": 10431
    },
    {
      "epoch": 4.9091764705882355,
      "grad_norm": 0.31985304951766824,
      "learning_rate": 3.312067221404371e-05,
      "loss": 0.1147,
      "step": 10432
    },
    {
      "epoch": 4.90964705882353,
      "grad_norm": 0.42303828111862135,
      "learning_rate": 3.309320152561923e-05,
      "loss": 0.4727,
      "step": 10433
    },
    {
      "epoch": 4.910117647058824,
      "grad_norm": 0.5375134344127896,
      "learning_rate": 3.3065740821608725e-05,
      "loss": 0.3066,
      "step": 10434
    },
    {
      "epoch": 4.910588235294117,
      "grad_norm": 0.4926262423230143,
      "learning_rate": 3.303829010435751e-05,
      "loss": 0.4082,
      "step": 10435
    },
    {
      "epoch": 4.9110588235294115,
      "grad_norm": 0.2761135117981655,
      "learning_rate": 3.3010849376210056e-05,
      "loss": 0.1011,
      "step": 10436
    },
    {
      "epoch": 4.911529411764706,
      "grad_norm": 0.563543195017394,
      "learning_rate": 3.298341863950986e-05,
      "loss": 0.6758,
      "step": 10437
    },
    {
      "epoch": 4.912,
      "grad_norm": 0.41233790579742496,
      "learning_rate": 3.295599789659967e-05,
      "loss": 0.2197,
      "step": 10438
    },
    {
      "epoch": 4.912470588235294,
      "grad_norm": 0.35329170083454314,
      "learning_rate": 3.2928587149821385e-05,
      "loss": 0.2695,
      "step": 10439
    },
    {
      "epoch": 4.912941176470588,
      "grad_norm": 0.24450015784545623,
      "learning_rate": 3.290118640151597e-05,
      "loss": 0.2207,
      "step": 10440
    },
    {
      "epoch": 4.913411764705883,
      "grad_norm": 0.4172469377658168,
      "learning_rate": 3.287379565402358e-05,
      "loss": 0.2314,
      "step": 10441
    },
    {
      "epoch": 4.913882352941177,
      "grad_norm": 0.3003733594859479,
      "learning_rate": 3.284641490968359e-05,
      "loss": 0.126,
      "step": 10442
    },
    {
      "epoch": 4.914352941176471,
      "grad_norm": 0.38918393396053486,
      "learning_rate": 3.281904417083436e-05,
      "loss": 0.3047,
      "step": 10443
    },
    {
      "epoch": 4.914823529411764,
      "grad_norm": 0.13768420910789106,
      "learning_rate": 3.279168343981355e-05,
      "loss": 0.0229,
      "step": 10444
    },
    {
      "epoch": 4.915294117647059,
      "grad_norm": 0.6205493975659881,
      "learning_rate": 3.2764332718957894e-05,
      "loss": 0.5586,
      "step": 10445
    },
    {
      "epoch": 4.915764705882353,
      "grad_norm": 0.35916520367086485,
      "learning_rate": 3.2736992010603226e-05,
      "loss": 0.2539,
      "step": 10446
    },
    {
      "epoch": 4.916235294117647,
      "grad_norm": 0.5474132088834559,
      "learning_rate": 3.270966131708461e-05,
      "loss": 0.4707,
      "step": 10447
    },
    {
      "epoch": 4.916705882352941,
      "grad_norm": 1.030197742670083,
      "learning_rate": 3.268234064073628e-05,
      "loss": 0.8359,
      "step": 10448
    },
    {
      "epoch": 4.9171764705882355,
      "grad_norm": 0.5981936684819527,
      "learning_rate": 3.265502998389143e-05,
      "loss": 0.4473,
      "step": 10449
    },
    {
      "epoch": 4.91764705882353,
      "grad_norm": 0.4533713465223028,
      "learning_rate": 3.262772934888265e-05,
      "loss": 0.4258,
      "step": 10450
    },
    {
      "epoch": 4.918117647058823,
      "grad_norm": 0.2578101375382084,
      "learning_rate": 3.260043873804147e-05,
      "loss": 0.1338,
      "step": 10451
    },
    {
      "epoch": 4.918588235294117,
      "grad_norm": 0.44596345794891357,
      "learning_rate": 3.257315815369863e-05,
      "loss": 0.1982,
      "step": 10452
    },
    {
      "epoch": 4.9190588235294115,
      "grad_norm": 0.6167897875348802,
      "learning_rate": 3.2545887598184114e-05,
      "loss": 0.3789,
      "step": 10453
    },
    {
      "epoch": 4.919529411764706,
      "grad_norm": 0.30406988441684574,
      "learning_rate": 3.251862707382686e-05,
      "loss": 0.3125,
      "step": 10454
    },
    {
      "epoch": 4.92,
      "grad_norm": 0.3334441426062555,
      "learning_rate": 3.249137658295511e-05,
      "loss": 0.0894,
      "step": 10455
    },
    {
      "epoch": 4.920470588235294,
      "grad_norm": 0.4306005304620657,
      "learning_rate": 3.246413612789622e-05,
      "loss": 0.4316,
      "step": 10456
    },
    {
      "epoch": 4.920941176470588,
      "grad_norm": 0.2616667518741126,
      "learning_rate": 3.243690571097656e-05,
      "loss": 0.1953,
      "step": 10457
    },
    {
      "epoch": 4.921411764705883,
      "grad_norm": 0.46809365876066855,
      "learning_rate": 3.24096853345218e-05,
      "loss": 0.2314,
      "step": 10458
    },
    {
      "epoch": 4.921882352941177,
      "grad_norm": 0.4748670942302647,
      "learning_rate": 3.2382475000856725e-05,
      "loss": 0.3691,
      "step": 10459
    },
    {
      "epoch": 4.92235294117647,
      "grad_norm": 0.25779508912708315,
      "learning_rate": 3.2355274712305154e-05,
      "loss": 0.0703,
      "step": 10460
    },
    {
      "epoch": 4.922823529411764,
      "grad_norm": 0.3957713200920124,
      "learning_rate": 3.232808447119018e-05,
      "loss": 0.1475,
      "step": 10461
    },
    {
      "epoch": 4.923294117647059,
      "grad_norm": 0.1967768523786477,
      "learning_rate": 3.2300904279834e-05,
      "loss": 0.0413,
      "step": 10462
    },
    {
      "epoch": 4.923764705882353,
      "grad_norm": 0.34723909002613107,
      "learning_rate": 3.227373414055782e-05,
      "loss": 0.2617,
      "step": 10463
    },
    {
      "epoch": 4.924235294117647,
      "grad_norm": 0.5678714432850863,
      "learning_rate": 3.224657405568229e-05,
      "loss": 0.6914,
      "step": 10464
    },
    {
      "epoch": 4.924705882352941,
      "grad_norm": 0.7176496016294088,
      "learning_rate": 3.221942402752685e-05,
      "loss": 0.582,
      "step": 10465
    },
    {
      "epoch": 4.9251764705882355,
      "grad_norm": 0.39859103395476303,
      "learning_rate": 3.2192284058410324e-05,
      "loss": 0.4766,
      "step": 10466
    },
    {
      "epoch": 4.92564705882353,
      "grad_norm": 0.3740601454227675,
      "learning_rate": 3.216515415065061e-05,
      "loss": 0.3672,
      "step": 10467
    },
    {
      "epoch": 4.926117647058824,
      "grad_norm": 0.44554641983595555,
      "learning_rate": 3.213803430656468e-05,
      "loss": 0.6133,
      "step": 10468
    },
    {
      "epoch": 4.926588235294117,
      "grad_norm": 0.43902339744458185,
      "learning_rate": 3.2110924528468725e-05,
      "loss": 0.4805,
      "step": 10469
    },
    {
      "epoch": 4.9270588235294115,
      "grad_norm": 0.33813670932424156,
      "learning_rate": 3.208382481867809e-05,
      "loss": 0.0713,
      "step": 10470
    },
    {
      "epoch": 4.927529411764706,
      "grad_norm": 0.3447355924922257,
      "learning_rate": 3.205673517950716e-05,
      "loss": 0.3164,
      "step": 10471
    },
    {
      "epoch": 4.928,
      "grad_norm": 0.46980629171518296,
      "learning_rate": 3.202965561326955e-05,
      "loss": 0.5078,
      "step": 10472
    },
    {
      "epoch": 4.928470588235294,
      "grad_norm": 0.41188422407338765,
      "learning_rate": 3.200258612227803e-05,
      "loss": 0.1777,
      "step": 10473
    },
    {
      "epoch": 4.928941176470588,
      "grad_norm": 0.40063758412962897,
      "learning_rate": 3.19755267088444e-05,
      "loss": 0.1523,
      "step": 10474
    },
    {
      "epoch": 4.929411764705883,
      "grad_norm": 0.325943960358485,
      "learning_rate": 3.194847737527969e-05,
      "loss": 0.2891,
      "step": 10475
    },
    {
      "epoch": 4.929882352941177,
      "grad_norm": 0.36926845265578584,
      "learning_rate": 3.1921438123894086e-05,
      "loss": 0.5039,
      "step": 10476
    },
    {
      "epoch": 4.930352941176471,
      "grad_norm": 0.4475282816127152,
      "learning_rate": 3.189440895699681e-05,
      "loss": 0.7148,
      "step": 10477
    },
    {
      "epoch": 4.930823529411764,
      "grad_norm": 0.5040583848773678,
      "learning_rate": 3.1867389876896296e-05,
      "loss": 0.375,
      "step": 10478
    },
    {
      "epoch": 4.931294117647059,
      "grad_norm": 0.3978985625933918,
      "learning_rate": 3.184038088590019e-05,
      "loss": 0.248,
      "step": 10479
    },
    {
      "epoch": 4.931764705882353,
      "grad_norm": 0.5553065240399548,
      "learning_rate": 3.181338198631506e-05,
      "loss": 0.7383,
      "step": 10480
    },
    {
      "epoch": 4.932235294117647,
      "grad_norm": 0.4102831727311949,
      "learning_rate": 3.178639318044687e-05,
      "loss": 0.1475,
      "step": 10481
    },
    {
      "epoch": 4.932705882352941,
      "grad_norm": 0.37873522403442716,
      "learning_rate": 3.175941447060048e-05,
      "loss": 0.168,
      "step": 10482
    },
    {
      "epoch": 4.9331764705882355,
      "grad_norm": 0.446792851060286,
      "learning_rate": 3.173244585908007e-05,
      "loss": 0.375,
      "step": 10483
    },
    {
      "epoch": 4.93364705882353,
      "grad_norm": 0.4041443652606441,
      "learning_rate": 3.170548734818893e-05,
      "loss": 0.4609,
      "step": 10484
    },
    {
      "epoch": 4.934117647058823,
      "grad_norm": 0.36775011910707683,
      "learning_rate": 3.1678538940229356e-05,
      "loss": 0.1289,
      "step": 10485
    },
    {
      "epoch": 4.934588235294117,
      "grad_norm": 0.4756067796665757,
      "learning_rate": 3.16516006375029e-05,
      "loss": 0.1533,
      "step": 10486
    },
    {
      "epoch": 4.9350588235294115,
      "grad_norm": 0.32083856218564066,
      "learning_rate": 3.1624672442310304e-05,
      "loss": 0.0977,
      "step": 10487
    },
    {
      "epoch": 4.935529411764706,
      "grad_norm": 0.3455646319872594,
      "learning_rate": 3.159775435695126e-05,
      "loss": 0.1396,
      "step": 10488
    },
    {
      "epoch": 4.936,
      "grad_norm": 0.31028919674666033,
      "learning_rate": 3.1570846383724775e-05,
      "loss": 0.2617,
      "step": 10489
    },
    {
      "epoch": 4.936470588235294,
      "grad_norm": 0.3386517287307065,
      "learning_rate": 3.154394852492892e-05,
      "loss": 0.3457,
      "step": 10490
    },
    {
      "epoch": 4.936941176470588,
      "grad_norm": 0.22061790327208966,
      "learning_rate": 3.1517060782860865e-05,
      "loss": 0.0498,
      "step": 10491
    },
    {
      "epoch": 4.937411764705883,
      "grad_norm": 0.6151554089503438,
      "learning_rate": 3.1490183159816956e-05,
      "loss": 0.7148,
      "step": 10492
    },
    {
      "epoch": 4.937882352941177,
      "grad_norm": 0.25396447151271967,
      "learning_rate": 3.146331565809273e-05,
      "loss": 0.0359,
      "step": 10493
    },
    {
      "epoch": 4.93835294117647,
      "grad_norm": 0.2793186161887173,
      "learning_rate": 3.143645827998274e-05,
      "loss": 0.085,
      "step": 10494
    },
    {
      "epoch": 4.938823529411764,
      "grad_norm": 0.31255859032166183,
      "learning_rate": 3.140961102778075e-05,
      "loss": 0.2676,
      "step": 10495
    },
    {
      "epoch": 4.939294117647059,
      "grad_norm": 0.23305172079071032,
      "learning_rate": 3.138277390377971e-05,
      "loss": 0.1396,
      "step": 10496
    },
    {
      "epoch": 4.939764705882353,
      "grad_norm": 0.4444097140435479,
      "learning_rate": 3.135594691027155e-05,
      "loss": 0.2188,
      "step": 10497
    },
    {
      "epoch": 4.940235294117647,
      "grad_norm": 0.4439050259101525,
      "learning_rate": 3.132913004954747e-05,
      "loss": 0.3555,
      "step": 10498
    },
    {
      "epoch": 4.940705882352941,
      "grad_norm": 0.45300294422713,
      "learning_rate": 3.130232332389781e-05,
      "loss": 0.3984,
      "step": 10499
    },
    {
      "epoch": 4.9411764705882355,
      "grad_norm": 0.4414923141571146,
      "learning_rate": 3.1275526735611896e-05,
      "loss": 0.6289,
      "step": 10500
    },
    {
      "epoch": 4.94164705882353,
      "grad_norm": 0.341826012245524,
      "learning_rate": 3.1248740286978375e-05,
      "loss": 0.1982,
      "step": 10501
    },
    {
      "epoch": 4.942117647058824,
      "grad_norm": 0.3744526533247029,
      "learning_rate": 3.1221963980284876e-05,
      "loss": 0.4141,
      "step": 10502
    },
    {
      "epoch": 4.942588235294117,
      "grad_norm": 1.1379287167197685,
      "learning_rate": 3.119519781781825e-05,
      "loss": 1.5859,
      "step": 10503
    },
    {
      "epoch": 4.9430588235294115,
      "grad_norm": 0.835801393994894,
      "learning_rate": 3.1168441801864505e-05,
      "loss": 0.6992,
      "step": 10504
    },
    {
      "epoch": 4.943529411764706,
      "grad_norm": 0.5894179708551639,
      "learning_rate": 3.1141695934708664e-05,
      "loss": 0.3965,
      "step": 10505
    },
    {
      "epoch": 4.944,
      "grad_norm": 0.3028720881283701,
      "learning_rate": 3.1114960218634986e-05,
      "loss": 0.3555,
      "step": 10506
    },
    {
      "epoch": 4.944470588235294,
      "grad_norm": 0.406023243403315,
      "learning_rate": 3.1088234655926815e-05,
      "loss": 0.1611,
      "step": 10507
    },
    {
      "epoch": 4.944941176470588,
      "grad_norm": 0.38184791305184485,
      "learning_rate": 3.106151924886668e-05,
      "loss": 0.2383,
      "step": 10508
    },
    {
      "epoch": 4.945411764705883,
      "grad_norm": 0.4364395186301691,
      "learning_rate": 3.103481399973622e-05,
      "loss": 0.332,
      "step": 10509
    },
    {
      "epoch": 4.945882352941177,
      "grad_norm": 0.4316962050109894,
      "learning_rate": 3.100811891081614e-05,
      "loss": 0.4434,
      "step": 10510
    },
    {
      "epoch": 4.946352941176471,
      "grad_norm": 0.3368280413658059,
      "learning_rate": 3.098143398438635e-05,
      "loss": 0.1836,
      "step": 10511
    },
    {
      "epoch": 4.946823529411764,
      "grad_norm": 0.8187271674853294,
      "learning_rate": 3.0954759222725916e-05,
      "loss": 0.21,
      "step": 10512
    },
    {
      "epoch": 4.947294117647059,
      "grad_norm": 0.6460567233349492,
      "learning_rate": 3.092809462811292e-05,
      "loss": 0.3828,
      "step": 10513
    },
    {
      "epoch": 4.947764705882353,
      "grad_norm": 0.5530037009010028,
      "learning_rate": 3.090144020282469e-05,
      "loss": 0.3867,
      "step": 10514
    },
    {
      "epoch": 4.948235294117647,
      "grad_norm": 0.23659586257675097,
      "learning_rate": 3.087479594913767e-05,
      "loss": 0.0562,
      "step": 10515
    },
    {
      "epoch": 4.948705882352941,
      "grad_norm": 0.6815771622948691,
      "learning_rate": 3.0848161869327356e-05,
      "loss": 0.416,
      "step": 10516
    },
    {
      "epoch": 4.9491764705882355,
      "grad_norm": 0.4095950308121157,
      "learning_rate": 3.0821537965668454e-05,
      "loss": 0.1318,
      "step": 10517
    },
    {
      "epoch": 4.94964705882353,
      "grad_norm": 0.48557471580604444,
      "learning_rate": 3.079492424043481e-05,
      "loss": 0.2832,
      "step": 10518
    },
    {
      "epoch": 4.950117647058823,
      "grad_norm": 1.4782415303470042,
      "learning_rate": 3.0768320695899294e-05,
      "loss": 1.75,
      "step": 10519
    },
    {
      "epoch": 4.950588235294117,
      "grad_norm": 0.41130108064654325,
      "learning_rate": 3.0741727334334026e-05,
      "loss": 0.4531,
      "step": 10520
    },
    {
      "epoch": 4.9510588235294115,
      "grad_norm": 0.53785103083939,
      "learning_rate": 3.071514415801024e-05,
      "loss": 0.6484,
      "step": 10521
    },
    {
      "epoch": 4.951529411764706,
      "grad_norm": 0.1515389385061578,
      "learning_rate": 3.068857116919819e-05,
      "loss": 0.0151,
      "step": 10522
    },
    {
      "epoch": 4.952,
      "grad_norm": 0.29491125885248903,
      "learning_rate": 3.066200837016737e-05,
      "loss": 0.2969,
      "step": 10523
    },
    {
      "epoch": 4.952470588235294,
      "grad_norm": 0.3406193520085104,
      "learning_rate": 3.063545576318644e-05,
      "loss": 0.3828,
      "step": 10524
    },
    {
      "epoch": 4.952941176470588,
      "grad_norm": 0.2371967667642872,
      "learning_rate": 3.060891335052302e-05,
      "loss": 0.0654,
      "step": 10525
    },
    {
      "epoch": 4.953411764705883,
      "grad_norm": 0.42722336962096064,
      "learning_rate": 3.058238113444401e-05,
      "loss": 0.3984,
      "step": 10526
    },
    {
      "epoch": 4.953882352941177,
      "grad_norm": 0.3806364926763182,
      "learning_rate": 3.0555859117215446e-05,
      "loss": 0.2988,
      "step": 10527
    },
    {
      "epoch": 4.95435294117647,
      "grad_norm": 0.32222782949614753,
      "learning_rate": 3.052934730110232e-05,
      "loss": 0.3477,
      "step": 10528
    },
    {
      "epoch": 4.954823529411764,
      "grad_norm": 0.3430516983294559,
      "learning_rate": 3.050284568836895e-05,
      "loss": 0.1846,
      "step": 10529
    },
    {
      "epoch": 4.955294117647059,
      "grad_norm": 0.10741049768065554,
      "learning_rate": 3.0476354281278726e-05,
      "loss": 0.0225,
      "step": 10530
    },
    {
      "epoch": 4.955764705882353,
      "grad_norm": 0.370691829949242,
      "learning_rate": 3.044987308209407e-05,
      "loss": 0.3809,
      "step": 10531
    },
    {
      "epoch": 4.956235294117647,
      "grad_norm": 0.4092542221654714,
      "learning_rate": 3.0423402093076683e-05,
      "loss": 0.2949,
      "step": 10532
    },
    {
      "epoch": 4.956705882352941,
      "grad_norm": 0.26465813133794813,
      "learning_rate": 3.039694131648723e-05,
      "loss": 0.1279,
      "step": 10533
    },
    {
      "epoch": 4.9571764705882355,
      "grad_norm": 0.3888862630059226,
      "learning_rate": 3.0370490754585636e-05,
      "loss": 0.168,
      "step": 10534
    },
    {
      "epoch": 4.95764705882353,
      "grad_norm": 0.4936174663582434,
      "learning_rate": 3.034405040963095e-05,
      "loss": 0.5,
      "step": 10535
    },
    {
      "epoch": 4.958117647058824,
      "grad_norm": 0.37197023502778165,
      "learning_rate": 3.0317620283881243e-05,
      "loss": 0.1523,
      "step": 10536
    },
    {
      "epoch": 4.958588235294117,
      "grad_norm": 0.12946233507016675,
      "learning_rate": 3.0291200379593782e-05,
      "loss": 0.0182,
      "step": 10537
    },
    {
      "epoch": 4.9590588235294115,
      "grad_norm": 0.35740469807473807,
      "learning_rate": 3.0264790699025027e-05,
      "loss": 0.4414,
      "step": 10538
    },
    {
      "epoch": 4.959529411764706,
      "grad_norm": 0.5918438076589752,
      "learning_rate": 3.0238391244430405e-05,
      "loss": 0.3105,
      "step": 10539
    },
    {
      "epoch": 4.96,
      "grad_norm": 0.31298449134384915,
      "learning_rate": 3.0212002018064614e-05,
      "loss": 0.0952,
      "step": 10540
    },
    {
      "epoch": 4.960470588235294,
      "grad_norm": 0.48742405960989293,
      "learning_rate": 3.0185623022181415e-05,
      "loss": 0.4121,
      "step": 10541
    },
    {
      "epoch": 4.960941176470588,
      "grad_norm": 0.3933441528694656,
      "learning_rate": 3.015925425903368e-05,
      "loss": 0.1914,
      "step": 10542
    },
    {
      "epoch": 4.961411764705883,
      "grad_norm": 0.3462455965001865,
      "learning_rate": 3.0132895730873448e-05,
      "loss": 0.334,
      "step": 10543
    },
    {
      "epoch": 4.961882352941177,
      "grad_norm": 0.23426953523852578,
      "learning_rate": 3.010654743995189e-05,
      "loss": 0.165,
      "step": 10544
    },
    {
      "epoch": 4.962352941176471,
      "grad_norm": 0.3243464081310165,
      "learning_rate": 3.0080209388519222e-05,
      "loss": 0.0854,
      "step": 10545
    },
    {
      "epoch": 4.962823529411764,
      "grad_norm": 0.24364357510869253,
      "learning_rate": 3.005388157882488e-05,
      "loss": 0.0747,
      "step": 10546
    },
    {
      "epoch": 4.963294117647059,
      "grad_norm": 0.4035991395072143,
      "learning_rate": 3.002756401311741e-05,
      "loss": 0.3359,
      "step": 10547
    },
    {
      "epoch": 4.963764705882353,
      "grad_norm": 0.45781586728094487,
      "learning_rate": 3.0001256693644405e-05,
      "loss": 0.625,
      "step": 10548
    },
    {
      "epoch": 4.964235294117647,
      "grad_norm": 0.5586680947822787,
      "learning_rate": 2.997495962265266e-05,
      "loss": 0.7578,
      "step": 10549
    },
    {
      "epoch": 4.964705882352941,
      "grad_norm": 0.5090618944753675,
      "learning_rate": 2.9948672802388135e-05,
      "loss": 0.4668,
      "step": 10550
    },
    {
      "epoch": 4.9651764705882355,
      "grad_norm": 0.34761820239999625,
      "learning_rate": 2.9922396235095762e-05,
      "loss": 0.209,
      "step": 10551
    },
    {
      "epoch": 4.96564705882353,
      "grad_norm": 0.2836397669280646,
      "learning_rate": 2.9896129923019717e-05,
      "loss": 0.0859,
      "step": 10552
    },
    {
      "epoch": 4.966117647058823,
      "grad_norm": 0.3090549745251846,
      "learning_rate": 2.9869873868403287e-05,
      "loss": 0.2188,
      "step": 10553
    },
    {
      "epoch": 4.966588235294117,
      "grad_norm": 0.23100223664017522,
      "learning_rate": 2.9843628073488863e-05,
      "loss": 0.2188,
      "step": 10554
    },
    {
      "epoch": 4.9670588235294115,
      "grad_norm": 0.34435789559232255,
      "learning_rate": 2.9817392540518014e-05,
      "loss": 0.1108,
      "step": 10555
    },
    {
      "epoch": 4.967529411764706,
      "grad_norm": 0.4510754882613896,
      "learning_rate": 2.9791167271731287e-05,
      "loss": 0.2637,
      "step": 10556
    },
    {
      "epoch": 4.968,
      "grad_norm": 0.3359059503344406,
      "learning_rate": 2.9764952269368487e-05,
      "loss": 0.1396,
      "step": 10557
    },
    {
      "epoch": 4.968470588235294,
      "grad_norm": 0.4754753326380249,
      "learning_rate": 2.973874753566857e-05,
      "loss": 0.543,
      "step": 10558
    },
    {
      "epoch": 4.968941176470588,
      "grad_norm": 0.38625580186206016,
      "learning_rate": 2.9712553072869437e-05,
      "loss": 0.377,
      "step": 10559
    },
    {
      "epoch": 4.969411764705883,
      "grad_norm": 0.42890985605830984,
      "learning_rate": 2.968636888320833e-05,
      "loss": 0.5625,
      "step": 10560
    },
    {
      "epoch": 4.969882352941177,
      "grad_norm": 0.4391089255396445,
      "learning_rate": 2.9660194968921403e-05,
      "loss": 0.4922,
      "step": 10561
    },
    {
      "epoch": 4.97035294117647,
      "grad_norm": 0.44974159835027494,
      "learning_rate": 2.963403133224409e-05,
      "loss": 0.2832,
      "step": 10562
    },
    {
      "epoch": 4.970823529411764,
      "grad_norm": 0.4775950671082729,
      "learning_rate": 2.9607877975410936e-05,
      "loss": 0.6758,
      "step": 10563
    },
    {
      "epoch": 4.971294117647059,
      "grad_norm": 0.37181693677675126,
      "learning_rate": 2.958173490065549e-05,
      "loss": 0.3242,
      "step": 10564
    },
    {
      "epoch": 4.971764705882353,
      "grad_norm": 0.37993039372315696,
      "learning_rate": 2.955560211021053e-05,
      "loss": 0.165,
      "step": 10565
    },
    {
      "epoch": 4.972235294117647,
      "grad_norm": 0.39768207776200126,
      "learning_rate": 2.952947960630797e-05,
      "loss": 0.252,
      "step": 10566
    },
    {
      "epoch": 4.972705882352941,
      "grad_norm": 0.2149265825529703,
      "learning_rate": 2.9503367391178713e-05,
      "loss": 0.0854,
      "step": 10567
    },
    {
      "epoch": 4.9731764705882355,
      "grad_norm": 0.27044623461861106,
      "learning_rate": 2.94772654670529e-05,
      "loss": 0.1748,
      "step": 10568
    },
    {
      "epoch": 4.97364705882353,
      "grad_norm": 0.36989475630096885,
      "learning_rate": 2.9451173836159826e-05,
      "loss": 0.2266,
      "step": 10569
    },
    {
      "epoch": 4.974117647058824,
      "grad_norm": 0.2973623276696074,
      "learning_rate": 2.9425092500727764e-05,
      "loss": 0.2852,
      "step": 10570
    },
    {
      "epoch": 4.974588235294117,
      "grad_norm": 0.2155336461226652,
      "learning_rate": 2.9399021462984206e-05,
      "loss": 0.0549,
      "step": 10571
    },
    {
      "epoch": 4.9750588235294115,
      "grad_norm": 0.3585260851812566,
      "learning_rate": 2.9372960725155804e-05,
      "loss": 0.418,
      "step": 10572
    },
    {
      "epoch": 4.975529411764706,
      "grad_norm": 0.1782542443477925,
      "learning_rate": 2.934691028946819e-05,
      "loss": 0.0277,
      "step": 10573
    },
    {
      "epoch": 4.976,
      "grad_norm": 0.4558315748819703,
      "learning_rate": 2.9320870158146235e-05,
      "loss": 0.2988,
      "step": 10574
    },
    {
      "epoch": 4.976470588235294,
      "grad_norm": 0.5120873715754682,
      "learning_rate": 2.9294840333413932e-05,
      "loss": 0.3652,
      "step": 10575
    },
    {
      "epoch": 4.976941176470588,
      "grad_norm": 0.6479977009357665,
      "learning_rate": 2.926882081749428e-05,
      "loss": 0.543,
      "step": 10576
    },
    {
      "epoch": 4.977411764705883,
      "grad_norm": 0.4074127942066533,
      "learning_rate": 2.9242811612609522e-05,
      "loss": 0.1924,
      "step": 10577
    },
    {
      "epoch": 4.977882352941176,
      "grad_norm": 0.6506135176079952,
      "learning_rate": 2.9216812720980998e-05,
      "loss": 0.6094,
      "step": 10578
    },
    {
      "epoch": 4.97835294117647,
      "grad_norm": 0.6266039442041016,
      "learning_rate": 2.919082414482908e-05,
      "loss": 0.4863,
      "step": 10579
    },
    {
      "epoch": 4.978823529411764,
      "grad_norm": 0.7967651790716739,
      "learning_rate": 2.9164845886373323e-05,
      "loss": 0.6445,
      "step": 10580
    },
    {
      "epoch": 4.979294117647059,
      "grad_norm": 0.7196739672916843,
      "learning_rate": 2.9138877947832474e-05,
      "loss": 1.0469,
      "step": 10581
    },
    {
      "epoch": 4.979764705882353,
      "grad_norm": 0.5283402017472195,
      "learning_rate": 2.911292033142422e-05,
      "loss": 0.3438,
      "step": 10582
    },
    {
      "epoch": 4.980235294117647,
      "grad_norm": 0.2544915468376065,
      "learning_rate": 2.9086973039365562e-05,
      "loss": 0.0845,
      "step": 10583
    },
    {
      "epoch": 4.980705882352941,
      "grad_norm": 0.4430010829135719,
      "learning_rate": 2.906103607387244e-05,
      "loss": 0.3418,
      "step": 10584
    },
    {
      "epoch": 4.9811764705882355,
      "grad_norm": 0.42194594162385646,
      "learning_rate": 2.9035109437160047e-05,
      "loss": 0.4688,
      "step": 10585
    },
    {
      "epoch": 4.98164705882353,
      "grad_norm": 0.2612773802062996,
      "learning_rate": 2.9009193131442665e-05,
      "loss": 0.3066,
      "step": 10586
    },
    {
      "epoch": 4.982117647058823,
      "grad_norm": 0.572004260974795,
      "learning_rate": 2.8983287158933622e-05,
      "loss": 0.4121,
      "step": 10587
    },
    {
      "epoch": 4.982588235294117,
      "grad_norm": 0.528500434149093,
      "learning_rate": 2.895739152184543e-05,
      "loss": 0.4375,
      "step": 10588
    },
    {
      "epoch": 4.9830588235294115,
      "grad_norm": 0.3613962339375791,
      "learning_rate": 2.8931506222389756e-05,
      "loss": 0.4727,
      "step": 10589
    },
    {
      "epoch": 4.983529411764706,
      "grad_norm": 0.32666083967754583,
      "learning_rate": 2.8905631262777245e-05,
      "loss": 0.5391,
      "step": 10590
    },
    {
      "epoch": 4.984,
      "grad_norm": 0.31892857511833345,
      "learning_rate": 2.8879766645217784e-05,
      "loss": 0.127,
      "step": 10591
    },
    {
      "epoch": 4.984470588235294,
      "grad_norm": 0.35898216159394725,
      "learning_rate": 2.8853912371920384e-05,
      "loss": 0.4512,
      "step": 10592
    },
    {
      "epoch": 4.984941176470588,
      "grad_norm": 0.42031556606950965,
      "learning_rate": 2.882806844509305e-05,
      "loss": 0.1436,
      "step": 10593
    },
    {
      "epoch": 4.985411764705883,
      "grad_norm": 0.16393268579491888,
      "learning_rate": 2.8802234866943012e-05,
      "loss": 0.03,
      "step": 10594
    },
    {
      "epoch": 4.985882352941177,
      "grad_norm": 0.42555411257048764,
      "learning_rate": 2.8776411639676644e-05,
      "loss": 0.3555,
      "step": 10595
    },
    {
      "epoch": 4.98635294117647,
      "grad_norm": 0.15655265991173223,
      "learning_rate": 2.8750598765499226e-05,
      "loss": 0.0669,
      "step": 10596
    },
    {
      "epoch": 4.986823529411764,
      "grad_norm": 0.2818906829913154,
      "learning_rate": 2.8724796246615485e-05,
      "loss": 0.1025,
      "step": 10597
    },
    {
      "epoch": 4.987294117647059,
      "grad_norm": 0.31760217575522526,
      "learning_rate": 2.8699004085228948e-05,
      "loss": 0.084,
      "step": 10598
    },
    {
      "epoch": 4.987764705882353,
      "grad_norm": 0.5111855156751416,
      "learning_rate": 2.867322228354243e-05,
      "loss": 0.3184,
      "step": 10599
    },
    {
      "epoch": 4.988235294117647,
      "grad_norm": 0.16998113258281022,
      "learning_rate": 2.8647450843757897e-05,
      "loss": 0.0236,
      "step": 10600
    },
    {
      "epoch": 4.988705882352941,
      "grad_norm": 0.8269959704598967,
      "learning_rate": 2.862168976807624e-05,
      "loss": 0.5156,
      "step": 10601
    },
    {
      "epoch": 4.9891764705882355,
      "grad_norm": 0.46333910380218163,
      "learning_rate": 2.859593905869763e-05,
      "loss": 0.2002,
      "step": 10602
    },
    {
      "epoch": 4.98964705882353,
      "grad_norm": 0.32057622717883605,
      "learning_rate": 2.8570198717821362e-05,
      "loss": 0.125,
      "step": 10603
    },
    {
      "epoch": 4.990117647058824,
      "grad_norm": 0.34997241981979166,
      "learning_rate": 2.8544468747645676e-05,
      "loss": 0.1089,
      "step": 10604
    },
    {
      "epoch": 4.990588235294117,
      "grad_norm": 0.47578854564517703,
      "learning_rate": 2.8518749150368114e-05,
      "loss": 0.4668,
      "step": 10605
    },
    {
      "epoch": 4.9910588235294115,
      "grad_norm": 0.5018699908481596,
      "learning_rate": 2.849303992818527e-05,
      "loss": 0.4863,
      "step": 10606
    },
    {
      "epoch": 4.991529411764706,
      "grad_norm": 0.30640467041006103,
      "learning_rate": 2.846734108329275e-05,
      "loss": 0.3066,
      "step": 10607
    },
    {
      "epoch": 4.992,
      "grad_norm": 0.3635940218644451,
      "learning_rate": 2.8441652617885437e-05,
      "loss": 0.3555,
      "step": 10608
    },
    {
      "epoch": 4.992470588235294,
      "grad_norm": 0.19425431585214123,
      "learning_rate": 2.8415974534157255e-05,
      "loss": 0.0703,
      "step": 10609
    },
    {
      "epoch": 4.992941176470588,
      "grad_norm": 0.4092928305916217,
      "learning_rate": 2.8390306834301186e-05,
      "loss": 0.25,
      "step": 10610
    },
    {
      "epoch": 4.993411764705883,
      "grad_norm": 0.19253819249562146,
      "learning_rate": 2.836464952050944e-05,
      "loss": 0.1426,
      "step": 10611
    },
    {
      "epoch": 4.993882352941176,
      "grad_norm": 0.270914745505626,
      "learning_rate": 2.8339002594973204e-05,
      "loss": 0.1167,
      "step": 10612
    },
    {
      "epoch": 4.99435294117647,
      "grad_norm": 0.4000925039841037,
      "learning_rate": 2.8313366059882887e-05,
      "loss": 0.3633,
      "step": 10613
    },
    {
      "epoch": 4.994823529411764,
      "grad_norm": 0.36870537593250474,
      "learning_rate": 2.828773991742802e-05,
      "loss": 0.1641,
      "step": 10614
    },
    {
      "epoch": 4.995294117647059,
      "grad_norm": 0.365100473512378,
      "learning_rate": 2.8262124169797106e-05,
      "loss": 0.3223,
      "step": 10615
    },
    {
      "epoch": 4.995764705882353,
      "grad_norm": 0.4082327309502926,
      "learning_rate": 2.823651881917793e-05,
      "loss": 0.5273,
      "step": 10616
    },
    {
      "epoch": 4.996235294117647,
      "grad_norm": 0.22310053198583946,
      "learning_rate": 2.8210923867757317e-05,
      "loss": 0.0491,
      "step": 10617
    },
    {
      "epoch": 4.996705882352941,
      "grad_norm": 0.30942654437105244,
      "learning_rate": 2.8185339317721146e-05,
      "loss": 0.2217,
      "step": 10618
    },
    {
      "epoch": 4.9971764705882356,
      "grad_norm": 0.33153003849426377,
      "learning_rate": 2.8159765171254495e-05,
      "loss": 0.334,
      "step": 10619
    },
    {
      "epoch": 4.99764705882353,
      "grad_norm": 0.24513118146457438,
      "learning_rate": 2.813420143054157e-05,
      "loss": 0.0593,
      "step": 10620
    },
    {
      "epoch": 4.998117647058823,
      "grad_norm": 0.43818583448454396,
      "learning_rate": 2.810864809776554e-05,
      "loss": 0.5234,
      "step": 10621
    },
    {
      "epoch": 4.998588235294117,
      "grad_norm": 0.26434643222990983,
      "learning_rate": 2.8083105175108845e-05,
      "loss": 0.0977,
      "step": 10622
    },
    {
      "epoch": 4.9990588235294116,
      "grad_norm": 0.6642201014174791,
      "learning_rate": 2.8057572664752986e-05,
      "loss": 0.5195,
      "step": 10623
    },
    {
      "epoch": 4.999529411764706,
      "grad_norm": 0.39702343405698476,
      "learning_rate": 2.8032050568878524e-05,
      "loss": 0.2314,
      "step": 10624
    },
    {
      "epoch": 5.0,
      "grad_norm": 0.38341267255282296,
      "learning_rate": 2.800653888966519e-05,
      "loss": 0.2373,
      "step": 10625
    },
    {
      "epoch": 5.0,
      "eval_loss": 0.33440348505973816,
      "eval_runtime": 47.177,
      "eval_samples_per_second": 15.898,
      "eval_steps_per_second": 7.949,
      "step": 10625
    },
    {
      "epoch": 5.000470588235294,
      "grad_norm": 0.33887371599405225,
      "learning_rate": 2.7981037629291852e-05,
      "loss": 0.1895,
      "step": 10626
    },
    {
      "epoch": 5.0009411764705884,
      "grad_norm": 0.3300116167703941,
      "learning_rate": 2.7955546789936363e-05,
      "loss": 0.1089,
      "step": 10627
    },
    {
      "epoch": 5.001411764705883,
      "grad_norm": 0.38311984355286516,
      "learning_rate": 2.7930066373775794e-05,
      "loss": 0.2734,
      "step": 10628
    },
    {
      "epoch": 5.001882352941177,
      "grad_norm": 0.2922404510520201,
      "learning_rate": 2.7904596382986343e-05,
      "loss": 0.1797,
      "step": 10629
    },
    {
      "epoch": 5.00235294117647,
      "grad_norm": 0.21614480197926944,
      "learning_rate": 2.7879136819743203e-05,
      "loss": 0.082,
      "step": 10630
    },
    {
      "epoch": 5.0028235294117644,
      "grad_norm": 0.2920717822074737,
      "learning_rate": 2.7853687686220766e-05,
      "loss": 0.1377,
      "step": 10631
    },
    {
      "epoch": 5.003294117647059,
      "grad_norm": 0.453275146997676,
      "learning_rate": 2.782824898459258e-05,
      "loss": 0.4355,
      "step": 10632
    },
    {
      "epoch": 5.003764705882353,
      "grad_norm": 0.2905409588425226,
      "learning_rate": 2.7802820717031132e-05,
      "loss": 0.0559,
      "step": 10633
    },
    {
      "epoch": 5.004235294117647,
      "grad_norm": 0.3992105352221691,
      "learning_rate": 2.7777402885708212e-05,
      "loss": 0.4453,
      "step": 10634
    },
    {
      "epoch": 5.004705882352941,
      "grad_norm": 0.2577216508991211,
      "learning_rate": 2.7751995492794544e-05,
      "loss": 0.1309,
      "step": 10635
    },
    {
      "epoch": 5.005176470588236,
      "grad_norm": 0.38830912182322114,
      "learning_rate": 2.7726598540460076e-05,
      "loss": 0.3262,
      "step": 10636
    },
    {
      "epoch": 5.00564705882353,
      "grad_norm": 0.323357499870607,
      "learning_rate": 2.770121203087389e-05,
      "loss": 0.1099,
      "step": 10637
    },
    {
      "epoch": 5.006117647058823,
      "grad_norm": 0.1826696289247411,
      "learning_rate": 2.7675835966204024e-05,
      "loss": 0.0193,
      "step": 10638
    },
    {
      "epoch": 5.006588235294117,
      "grad_norm": 0.32282864718507714,
      "learning_rate": 2.7650470348617772e-05,
      "loss": 0.1689,
      "step": 10639
    },
    {
      "epoch": 5.007058823529412,
      "grad_norm": 0.31814374954711544,
      "learning_rate": 2.7625115180281464e-05,
      "loss": 0.0991,
      "step": 10640
    },
    {
      "epoch": 5.007529411764706,
      "grad_norm": 0.2589482975069025,
      "learning_rate": 2.7599770463360576e-05,
      "loss": 0.1299,
      "step": 10641
    },
    {
      "epoch": 5.008,
      "grad_norm": 0.3342559669873211,
      "learning_rate": 2.7574436200019688e-05,
      "loss": 0.3301,
      "step": 10642
    },
    {
      "epoch": 5.008470588235294,
      "grad_norm": 0.16896258825422633,
      "learning_rate": 2.7549112392422405e-05,
      "loss": 0.0723,
      "step": 10643
    },
    {
      "epoch": 5.0089411764705885,
      "grad_norm": 0.23345764626745322,
      "learning_rate": 2.7523799042731548e-05,
      "loss": 0.1216,
      "step": 10644
    },
    {
      "epoch": 5.009411764705883,
      "grad_norm": 0.5452240135443354,
      "learning_rate": 2.749849615310902e-05,
      "loss": 0.248,
      "step": 10645
    },
    {
      "epoch": 5.009882352941177,
      "grad_norm": 0.30140476054603554,
      "learning_rate": 2.7473203725715754e-05,
      "loss": 0.0972,
      "step": 10646
    },
    {
      "epoch": 5.01035294117647,
      "grad_norm": 0.4215565041376949,
      "learning_rate": 2.7447921762711894e-05,
      "loss": 0.4219,
      "step": 10647
    },
    {
      "epoch": 5.0108235294117645,
      "grad_norm": 0.4801857175227623,
      "learning_rate": 2.7422650266256663e-05,
      "loss": 0.373,
      "step": 10648
    },
    {
      "epoch": 5.011294117647059,
      "grad_norm": 0.25842759167921175,
      "learning_rate": 2.7397389238508282e-05,
      "loss": 0.0903,
      "step": 10649
    },
    {
      "epoch": 5.011764705882353,
      "grad_norm": 0.26126030725915667,
      "learning_rate": 2.7372138681624244e-05,
      "loss": 0.126,
      "step": 10650
    },
    {
      "epoch": 5.012235294117647,
      "grad_norm": 0.5768001243028088,
      "learning_rate": 2.7346898597761073e-05,
      "loss": 0.3301,
      "step": 10651
    },
    {
      "epoch": 5.012705882352941,
      "grad_norm": 0.4385658950052787,
      "learning_rate": 2.7321668989074347e-05,
      "loss": 0.2637,
      "step": 10652
    },
    {
      "epoch": 5.013176470588236,
      "grad_norm": 0.3582715812165666,
      "learning_rate": 2.7296449857718806e-05,
      "loss": 0.2695,
      "step": 10653
    },
    {
      "epoch": 5.01364705882353,
      "grad_norm": 0.40642015097096307,
      "learning_rate": 2.727124120584836e-05,
      "loss": 0.4902,
      "step": 10654
    },
    {
      "epoch": 5.014117647058823,
      "grad_norm": 0.7192273165984804,
      "learning_rate": 2.724604303561584e-05,
      "loss": 0.4727,
      "step": 10655
    },
    {
      "epoch": 5.014588235294117,
      "grad_norm": 0.4490943873580337,
      "learning_rate": 2.7220855349173353e-05,
      "loss": 0.5625,
      "step": 10656
    },
    {
      "epoch": 5.015058823529412,
      "grad_norm": 0.4692815715282006,
      "learning_rate": 2.71956781486721e-05,
      "loss": 0.4512,
      "step": 10657
    },
    {
      "epoch": 5.015529411764706,
      "grad_norm": 0.31158052572676853,
      "learning_rate": 2.7170511436262243e-05,
      "loss": 0.1436,
      "step": 10658
    },
    {
      "epoch": 5.016,
      "grad_norm": 0.28564946895320326,
      "learning_rate": 2.7145355214093178e-05,
      "loss": 0.2695,
      "step": 10659
    },
    {
      "epoch": 5.016470588235294,
      "grad_norm": 0.18618844481347366,
      "learning_rate": 2.7120209484313422e-05,
      "loss": 0.0476,
      "step": 10660
    },
    {
      "epoch": 5.0169411764705885,
      "grad_norm": 0.2789762634147063,
      "learning_rate": 2.7095074249070455e-05,
      "loss": 0.1172,
      "step": 10661
    },
    {
      "epoch": 5.017411764705883,
      "grad_norm": 0.5310587668082489,
      "learning_rate": 2.7069949510511023e-05,
      "loss": 0.457,
      "step": 10662
    },
    {
      "epoch": 5.017882352941177,
      "grad_norm": 0.5101426419297518,
      "learning_rate": 2.7044835270780836e-05,
      "loss": 0.5742,
      "step": 10663
    },
    {
      "epoch": 5.01835294117647,
      "grad_norm": 0.4570863463529672,
      "learning_rate": 2.701973153202482e-05,
      "loss": 0.334,
      "step": 10664
    },
    {
      "epoch": 5.0188235294117645,
      "grad_norm": 0.3159700485867312,
      "learning_rate": 2.6994638296386977e-05,
      "loss": 0.1738,
      "step": 10665
    },
    {
      "epoch": 5.019294117647059,
      "grad_norm": 0.5454041806558476,
      "learning_rate": 2.6969555566010316e-05,
      "loss": 0.4355,
      "step": 10666
    },
    {
      "epoch": 5.019764705882353,
      "grad_norm": 0.5457086095417278,
      "learning_rate": 2.694448334303708e-05,
      "loss": 0.2773,
      "step": 10667
    },
    {
      "epoch": 5.020235294117647,
      "grad_norm": 0.4789342923932575,
      "learning_rate": 2.6919421629608584e-05,
      "loss": 0.3359,
      "step": 10668
    },
    {
      "epoch": 5.020705882352941,
      "grad_norm": 0.4328634482494007,
      "learning_rate": 2.6894370427865157e-05,
      "loss": 0.5078,
      "step": 10669
    },
    {
      "epoch": 5.021176470588236,
      "grad_norm": 0.3587595385412372,
      "learning_rate": 2.6869329739946312e-05,
      "loss": 0.248,
      "step": 10670
    },
    {
      "epoch": 5.02164705882353,
      "grad_norm": 0.2818657154493176,
      "learning_rate": 2.6844299567990713e-05,
      "loss": 0.0718,
      "step": 10671
    },
    {
      "epoch": 5.022117647058823,
      "grad_norm": 0.3881068030146368,
      "learning_rate": 2.6819279914135956e-05,
      "loss": 0.1465,
      "step": 10672
    },
    {
      "epoch": 5.022588235294117,
      "grad_norm": 0.20570577578139293,
      "learning_rate": 2.6794270780518895e-05,
      "loss": 0.0481,
      "step": 10673
    },
    {
      "epoch": 5.023058823529412,
      "grad_norm": 0.39262846460678025,
      "learning_rate": 2.6769272169275474e-05,
      "loss": 0.2285,
      "step": 10674
    },
    {
      "epoch": 5.023529411764706,
      "grad_norm": 0.4627857933064332,
      "learning_rate": 2.67442840825406e-05,
      "loss": 0.2109,
      "step": 10675
    },
    {
      "epoch": 5.024,
      "grad_norm": 0.3573568311086129,
      "learning_rate": 2.671930652244845e-05,
      "loss": 0.1084,
      "step": 10676
    },
    {
      "epoch": 5.024470588235294,
      "grad_norm": 0.24770988226979254,
      "learning_rate": 2.6694339491132237e-05,
      "loss": 0.0703,
      "step": 10677
    },
    {
      "epoch": 5.0249411764705885,
      "grad_norm": 0.2185897528896203,
      "learning_rate": 2.666938299072422e-05,
      "loss": 0.1035,
      "step": 10678
    },
    {
      "epoch": 5.025411764705883,
      "grad_norm": 0.4910390033783015,
      "learning_rate": 2.664443702335581e-05,
      "loss": 0.4102,
      "step": 10679
    },
    {
      "epoch": 5.025882352941177,
      "grad_norm": 0.4585841629149214,
      "learning_rate": 2.6619501591157583e-05,
      "loss": 0.4727,
      "step": 10680
    },
    {
      "epoch": 5.02635294117647,
      "grad_norm": 0.3821212610942609,
      "learning_rate": 2.6594576696259055e-05,
      "loss": 0.4453,
      "step": 10681
    },
    {
      "epoch": 5.0268235294117645,
      "grad_norm": 0.3988433842083488,
      "learning_rate": 2.6569662340789012e-05,
      "loss": 0.1797,
      "step": 10682
    },
    {
      "epoch": 5.027294117647059,
      "grad_norm": 0.2482660385753323,
      "learning_rate": 2.6544758526875193e-05,
      "loss": 0.1064,
      "step": 10683
    },
    {
      "epoch": 5.027764705882353,
      "grad_norm": 0.5220524601761711,
      "learning_rate": 2.651986525664455e-05,
      "loss": 0.2578,
      "step": 10684
    },
    {
      "epoch": 5.028235294117647,
      "grad_norm": 0.1697651654077359,
      "learning_rate": 2.649498253222307e-05,
      "loss": 0.0383,
      "step": 10685
    },
    {
      "epoch": 5.028705882352941,
      "grad_norm": 0.24607723425139644,
      "learning_rate": 2.647011035573588e-05,
      "loss": 0.1494,
      "step": 10686
    },
    {
      "epoch": 5.029176470588236,
      "grad_norm": 0.3940430277645166,
      "learning_rate": 2.6445248729307165e-05,
      "loss": 0.3262,
      "step": 10687
    },
    {
      "epoch": 5.02964705882353,
      "grad_norm": 0.24859073443698057,
      "learning_rate": 2.6420397655060282e-05,
      "loss": 0.1689,
      "step": 10688
    },
    {
      "epoch": 5.030117647058823,
      "grad_norm": 0.2956874034519343,
      "learning_rate": 2.639555713511756e-05,
      "loss": 0.2773,
      "step": 10689
    },
    {
      "epoch": 5.030588235294117,
      "grad_norm": 0.42810184924336686,
      "learning_rate": 2.637072717160056e-05,
      "loss": 0.2852,
      "step": 10690
    },
    {
      "epoch": 5.031058823529412,
      "grad_norm": 0.4235949976322634,
      "learning_rate": 2.6345907766629843e-05,
      "loss": 0.4277,
      "step": 10691
    },
    {
      "epoch": 5.031529411764706,
      "grad_norm": 0.4689052025388171,
      "learning_rate": 2.6321098922325102e-05,
      "loss": 0.3496,
      "step": 10692
    },
    {
      "epoch": 5.032,
      "grad_norm": 0.44602155855634484,
      "learning_rate": 2.6296300640805206e-05,
      "loss": 0.5977,
      "step": 10693
    },
    {
      "epoch": 5.032470588235294,
      "grad_norm": 0.5479614285365033,
      "learning_rate": 2.627151292418795e-05,
      "loss": 0.3496,
      "step": 10694
    },
    {
      "epoch": 5.0329411764705885,
      "grad_norm": 0.39689973146316354,
      "learning_rate": 2.6246735774590383e-05,
      "loss": 0.5703,
      "step": 10695
    },
    {
      "epoch": 5.033411764705883,
      "grad_norm": 0.29465599332962844,
      "learning_rate": 2.6221969194128617e-05,
      "loss": 0.2734,
      "step": 10696
    },
    {
      "epoch": 5.033882352941176,
      "grad_norm": 0.4288762382101276,
      "learning_rate": 2.6197213184917775e-05,
      "loss": 0.2158,
      "step": 10697
    },
    {
      "epoch": 5.03435294117647,
      "grad_norm": 0.44245140123741394,
      "learning_rate": 2.6172467749072185e-05,
      "loss": 0.4766,
      "step": 10698
    },
    {
      "epoch": 5.0348235294117645,
      "grad_norm": 0.6235639589054224,
      "learning_rate": 2.6147732888705235e-05,
      "loss": 0.5898,
      "step": 10699
    },
    {
      "epoch": 5.035294117647059,
      "grad_norm": 0.1815716501359462,
      "learning_rate": 2.6123008605929375e-05,
      "loss": 0.0591,
      "step": 10700
    },
    {
      "epoch": 5.035764705882353,
      "grad_norm": 0.28309213588052545,
      "learning_rate": 2.6098294902856177e-05,
      "loss": 0.0815,
      "step": 10701
    },
    {
      "epoch": 5.036235294117647,
      "grad_norm": 0.17943050198206756,
      "learning_rate": 2.607359178159637e-05,
      "loss": 0.0359,
      "step": 10702
    },
    {
      "epoch": 5.036705882352941,
      "grad_norm": 0.4415855867830408,
      "learning_rate": 2.604889924425965e-05,
      "loss": 0.1494,
      "step": 10703
    },
    {
      "epoch": 5.037176470588236,
      "grad_norm": 0.29880141683106953,
      "learning_rate": 2.6024217292954917e-05,
      "loss": 0.3398,
      "step": 10704
    },
    {
      "epoch": 5.03764705882353,
      "grad_norm": 0.5804656040195666,
      "learning_rate": 2.599954592979015e-05,
      "loss": 0.3965,
      "step": 10705
    },
    {
      "epoch": 5.038117647058823,
      "grad_norm": 0.21048441539992954,
      "learning_rate": 2.5974885156872345e-05,
      "loss": 0.0437,
      "step": 10706
    },
    {
      "epoch": 5.038588235294117,
      "grad_norm": 0.33738849340224,
      "learning_rate": 2.5950234976307682e-05,
      "loss": 0.2891,
      "step": 10707
    },
    {
      "epoch": 5.039058823529412,
      "grad_norm": 0.3434858667217957,
      "learning_rate": 2.5925595390201466e-05,
      "loss": 0.209,
      "step": 10708
    },
    {
      "epoch": 5.039529411764706,
      "grad_norm": 0.5055280502871228,
      "learning_rate": 2.5900966400657923e-05,
      "loss": 0.2256,
      "step": 10709
    },
    {
      "epoch": 5.04,
      "grad_norm": 0.3356685342750128,
      "learning_rate": 2.5876348009780564e-05,
      "loss": 0.2891,
      "step": 10710
    },
    {
      "epoch": 5.040470588235294,
      "grad_norm": 0.9000996122420575,
      "learning_rate": 2.5851740219671945e-05,
      "loss": 0.5234,
      "step": 10711
    },
    {
      "epoch": 5.0409411764705885,
      "grad_norm": 0.2244257532311148,
      "learning_rate": 2.5827143032433627e-05,
      "loss": 0.1094,
      "step": 10712
    },
    {
      "epoch": 5.041411764705883,
      "grad_norm": 0.392025665584791,
      "learning_rate": 2.5802556450166396e-05,
      "loss": 0.1523,
      "step": 10713
    },
    {
      "epoch": 5.041882352941176,
      "grad_norm": 0.39341700302197363,
      "learning_rate": 2.5777980474969983e-05,
      "loss": 0.249,
      "step": 10714
    },
    {
      "epoch": 5.04235294117647,
      "grad_norm": 0.326394274909371,
      "learning_rate": 2.5753415108943332e-05,
      "loss": 0.2207,
      "step": 10715
    },
    {
      "epoch": 5.0428235294117645,
      "grad_norm": 0.27883388659120867,
      "learning_rate": 2.572886035418451e-05,
      "loss": 0.1846,
      "step": 10716
    },
    {
      "epoch": 5.043294117647059,
      "grad_norm": 0.49584473193367534,
      "learning_rate": 2.5704316212790527e-05,
      "loss": 0.5078,
      "step": 10717
    },
    {
      "epoch": 5.043764705882353,
      "grad_norm": 0.4481127380620061,
      "learning_rate": 2.5679782686857586e-05,
      "loss": 0.1758,
      "step": 10718
    },
    {
      "epoch": 5.044235294117647,
      "grad_norm": 0.5135517943968984,
      "learning_rate": 2.5655259778481045e-05,
      "loss": 0.5156,
      "step": 10719
    },
    {
      "epoch": 5.044705882352941,
      "grad_norm": 0.3420737919939533,
      "learning_rate": 2.5630747489755182e-05,
      "loss": 0.0967,
      "step": 10720
    },
    {
      "epoch": 5.045176470588236,
      "grad_norm": 0.34638148895092924,
      "learning_rate": 2.560624582277352e-05,
      "loss": 0.3535,
      "step": 10721
    },
    {
      "epoch": 5.04564705882353,
      "grad_norm": 0.6998041299186729,
      "learning_rate": 2.558175477962865e-05,
      "loss": 0.6445,
      "step": 10722
    },
    {
      "epoch": 5.046117647058823,
      "grad_norm": 0.27991601525690574,
      "learning_rate": 2.5557274362412132e-05,
      "loss": 0.1787,
      "step": 10723
    },
    {
      "epoch": 5.046588235294117,
      "grad_norm": 0.38290976344103395,
      "learning_rate": 2.55328045732148e-05,
      "loss": 0.3301,
      "step": 10724
    },
    {
      "epoch": 5.047058823529412,
      "grad_norm": 0.48462341343848353,
      "learning_rate": 2.5508345414126474e-05,
      "loss": 0.418,
      "step": 10725
    },
    {
      "epoch": 5.047529411764706,
      "grad_norm": 0.3068756115103885,
      "learning_rate": 2.5483896887236056e-05,
      "loss": 0.2188,
      "step": 10726
    },
    {
      "epoch": 5.048,
      "grad_norm": 0.28623367720684584,
      "learning_rate": 2.5459458994631604e-05,
      "loss": 0.0737,
      "step": 10727
    },
    {
      "epoch": 5.048470588235294,
      "grad_norm": 0.4518855449179929,
      "learning_rate": 2.5435031738400234e-05,
      "loss": 0.2754,
      "step": 10728
    },
    {
      "epoch": 5.0489411764705885,
      "grad_norm": 0.32393867408256505,
      "learning_rate": 2.541061512062808e-05,
      "loss": 0.3574,
      "step": 10729
    },
    {
      "epoch": 5.049411764705883,
      "grad_norm": 0.2852135473865517,
      "learning_rate": 2.538620914340057e-05,
      "loss": 0.0962,
      "step": 10730
    },
    {
      "epoch": 5.049882352941176,
      "grad_norm": 0.3295636070271566,
      "learning_rate": 2.5361813808802007e-05,
      "loss": 0.1885,
      "step": 10731
    },
    {
      "epoch": 5.05035294117647,
      "grad_norm": 0.319597169974809,
      "learning_rate": 2.533742911891587e-05,
      "loss": 0.2285,
      "step": 10732
    },
    {
      "epoch": 5.0508235294117645,
      "grad_norm": 0.5176214735152884,
      "learning_rate": 2.531305507582481e-05,
      "loss": 0.5273,
      "step": 10733
    },
    {
      "epoch": 5.051294117647059,
      "grad_norm": 0.27853089561769406,
      "learning_rate": 2.5288691681610383e-05,
      "loss": 0.2754,
      "step": 10734
    },
    {
      "epoch": 5.051764705882353,
      "grad_norm": 0.24246005734361978,
      "learning_rate": 2.526433893835339e-05,
      "loss": 0.0815,
      "step": 10735
    },
    {
      "epoch": 5.052235294117647,
      "grad_norm": 0.37486268549662854,
      "learning_rate": 2.5239996848133715e-05,
      "loss": 0.3652,
      "step": 10736
    },
    {
      "epoch": 5.052705882352941,
      "grad_norm": 0.45758114220037627,
      "learning_rate": 2.5215665413030213e-05,
      "loss": 0.1904,
      "step": 10737
    },
    {
      "epoch": 5.053176470588236,
      "grad_norm": 0.1350312935883144,
      "learning_rate": 2.519134463512095e-05,
      "loss": 0.0352,
      "step": 10738
    },
    {
      "epoch": 5.05364705882353,
      "grad_norm": 0.4021605075618699,
      "learning_rate": 2.516703451648309e-05,
      "loss": 0.4004,
      "step": 10739
    },
    {
      "epoch": 5.054117647058823,
      "grad_norm": 0.27425717734301525,
      "learning_rate": 2.5142735059192737e-05,
      "loss": 0.1465,
      "step": 10740
    },
    {
      "epoch": 5.054588235294117,
      "grad_norm": 0.34370133487945176,
      "learning_rate": 2.511844626532526e-05,
      "loss": 0.4648,
      "step": 10741
    },
    {
      "epoch": 5.055058823529412,
      "grad_norm": 0.25220398492680457,
      "learning_rate": 2.5094168136954985e-05,
      "loss": 0.0693,
      "step": 10742
    },
    {
      "epoch": 5.055529411764706,
      "grad_norm": 0.3561879892839412,
      "learning_rate": 2.506990067615542e-05,
      "loss": 0.0747,
      "step": 10743
    },
    {
      "epoch": 5.056,
      "grad_norm": 1.0410591313139506,
      "learning_rate": 2.504564388499914e-05,
      "loss": 0.5195,
      "step": 10744
    },
    {
      "epoch": 5.056470588235294,
      "grad_norm": 0.3676534335867735,
      "learning_rate": 2.502139776555772e-05,
      "loss": 0.2354,
      "step": 10745
    },
    {
      "epoch": 5.0569411764705885,
      "grad_norm": 0.316606864821027,
      "learning_rate": 2.4997162319901977e-05,
      "loss": 0.1846,
      "step": 10746
    },
    {
      "epoch": 5.057411764705883,
      "grad_norm": 0.36773014844486646,
      "learning_rate": 2.497293755010173e-05,
      "loss": 0.4551,
      "step": 10747
    },
    {
      "epoch": 5.057882352941176,
      "grad_norm": 0.21244962112034654,
      "learning_rate": 2.4948723458225827e-05,
      "loss": 0.0933,
      "step": 10748
    },
    {
      "epoch": 5.05835294117647,
      "grad_norm": 0.4110200806335249,
      "learning_rate": 2.4924520046342317e-05,
      "loss": 0.3359,
      "step": 10749
    },
    {
      "epoch": 5.0588235294117645,
      "grad_norm": 0.46246398602653965,
      "learning_rate": 2.4900327316518326e-05,
      "loss": 0.4609,
      "step": 10750
    },
    {
      "epoch": 5.059294117647059,
      "grad_norm": 0.5221807501012902,
      "learning_rate": 2.4876145270819954e-05,
      "loss": 0.3496,
      "step": 10751
    },
    {
      "epoch": 5.059764705882353,
      "grad_norm": 0.15468943742948923,
      "learning_rate": 2.485197391131251e-05,
      "loss": 0.0189,
      "step": 10752
    },
    {
      "epoch": 5.060235294117647,
      "grad_norm": 0.45285838190937133,
      "learning_rate": 2.4827813240060367e-05,
      "loss": 0.3848,
      "step": 10753
    },
    {
      "epoch": 5.060705882352941,
      "grad_norm": 0.2802210830109263,
      "learning_rate": 2.480366325912692e-05,
      "loss": 0.1768,
      "step": 10754
    },
    {
      "epoch": 5.061176470588236,
      "grad_norm": 0.7906322612634646,
      "learning_rate": 2.4779523970574703e-05,
      "loss": 0.3789,
      "step": 10755
    },
    {
      "epoch": 5.06164705882353,
      "grad_norm": 0.5129201444163347,
      "learning_rate": 2.4755395376465386e-05,
      "loss": 0.3203,
      "step": 10756
    },
    {
      "epoch": 5.062117647058823,
      "grad_norm": 0.5771413816848462,
      "learning_rate": 2.4731277478859586e-05,
      "loss": 0.6758,
      "step": 10757
    },
    {
      "epoch": 5.062588235294117,
      "grad_norm": 0.2351286722545061,
      "learning_rate": 2.4707170279817118e-05,
      "loss": 0.1523,
      "step": 10758
    },
    {
      "epoch": 5.063058823529412,
      "grad_norm": 0.4275422317719175,
      "learning_rate": 2.4683073781396924e-05,
      "loss": 0.3164,
      "step": 10759
    },
    {
      "epoch": 5.063529411764706,
      "grad_norm": 0.32969501655205924,
      "learning_rate": 2.465898798565685e-05,
      "loss": 0.2285,
      "step": 10760
    },
    {
      "epoch": 5.064,
      "grad_norm": 0.4110738155392335,
      "learning_rate": 2.4634912894653996e-05,
      "loss": 0.5273,
      "step": 10761
    },
    {
      "epoch": 5.064470588235294,
      "grad_norm": 0.78862011643728,
      "learning_rate": 2.461084851044454e-05,
      "loss": 0.7383,
      "step": 10762
    },
    {
      "epoch": 5.0649411764705885,
      "grad_norm": 1.5272618413946146,
      "learning_rate": 2.4586794835083613e-05,
      "loss": 0.8906,
      "step": 10763
    },
    {
      "epoch": 5.065411764705883,
      "grad_norm": 0.22156104244253916,
      "learning_rate": 2.4562751870625597e-05,
      "loss": 0.1133,
      "step": 10764
    },
    {
      "epoch": 5.065882352941176,
      "grad_norm": 0.4372078604298546,
      "learning_rate": 2.4538719619123786e-05,
      "loss": 0.4414,
      "step": 10765
    },
    {
      "epoch": 5.06635294117647,
      "grad_norm": 0.32930023232393063,
      "learning_rate": 2.4514698082630706e-05,
      "loss": 0.0898,
      "step": 10766
    },
    {
      "epoch": 5.0668235294117645,
      "grad_norm": 0.3665143116094524,
      "learning_rate": 2.449068726319795e-05,
      "loss": 0.2197,
      "step": 10767
    },
    {
      "epoch": 5.067294117647059,
      "grad_norm": 0.09672259149133675,
      "learning_rate": 2.4466687162876093e-05,
      "loss": 0.0121,
      "step": 10768
    },
    {
      "epoch": 5.067764705882353,
      "grad_norm": 0.326195206292974,
      "learning_rate": 2.4442697783714894e-05,
      "loss": 0.0908,
      "step": 10769
    },
    {
      "epoch": 5.068235294117647,
      "grad_norm": 0.48322610765775087,
      "learning_rate": 2.4418719127763183e-05,
      "loss": 0.3535,
      "step": 10770
    },
    {
      "epoch": 5.068705882352941,
      "grad_norm": 0.4032073822967864,
      "learning_rate": 2.43947511970688e-05,
      "loss": 0.4316,
      "step": 10771
    },
    {
      "epoch": 5.069176470588236,
      "grad_norm": 0.5098149652638331,
      "learning_rate": 2.437079399367875e-05,
      "loss": 0.1631,
      "step": 10772
    },
    {
      "epoch": 5.06964705882353,
      "grad_norm": 0.49534909964784096,
      "learning_rate": 2.434684751963914e-05,
      "loss": 0.4805,
      "step": 10773
    },
    {
      "epoch": 5.070117647058823,
      "grad_norm": 0.49153018776222934,
      "learning_rate": 2.4322911776995002e-05,
      "loss": 0.1787,
      "step": 10774
    },
    {
      "epoch": 5.070588235294117,
      "grad_norm": 0.31608658178282945,
      "learning_rate": 2.429898676779072e-05,
      "loss": 0.1709,
      "step": 10775
    },
    {
      "epoch": 5.071058823529412,
      "grad_norm": 0.3208980890083939,
      "learning_rate": 2.4275072494069487e-05,
      "loss": 0.1157,
      "step": 10776
    },
    {
      "epoch": 5.071529411764706,
      "grad_norm": 0.3364055768237257,
      "learning_rate": 2.425116895787374e-05,
      "loss": 0.1196,
      "step": 10777
    },
    {
      "epoch": 5.072,
      "grad_norm": 0.7893562005507696,
      "learning_rate": 2.4227276161244995e-05,
      "loss": 0.3945,
      "step": 10778
    },
    {
      "epoch": 5.072470588235294,
      "grad_norm": 0.4634231835838743,
      "learning_rate": 2.4203394106223755e-05,
      "loss": 0.1748,
      "step": 10779
    },
    {
      "epoch": 5.0729411764705885,
      "grad_norm": 0.34974336495098823,
      "learning_rate": 2.417952279484967e-05,
      "loss": 0.4141,
      "step": 10780
    },
    {
      "epoch": 5.073411764705883,
      "grad_norm": 0.45077216556095046,
      "learning_rate": 2.4155662229161527e-05,
      "loss": 0.4043,
      "step": 10781
    },
    {
      "epoch": 5.073882352941176,
      "grad_norm": 0.44360726320685034,
      "learning_rate": 2.4131812411197045e-05,
      "loss": 0.459,
      "step": 10782
    },
    {
      "epoch": 5.07435294117647,
      "grad_norm": 0.30761218476883956,
      "learning_rate": 2.4107973342993174e-05,
      "loss": 0.2676,
      "step": 10783
    },
    {
      "epoch": 5.0748235294117645,
      "grad_norm": 0.38019052776744083,
      "learning_rate": 2.408414502658591e-05,
      "loss": 0.2129,
      "step": 10784
    },
    {
      "epoch": 5.075294117647059,
      "grad_norm": 0.20214732848957473,
      "learning_rate": 2.4060327464010254e-05,
      "loss": 0.0791,
      "step": 10785
    },
    {
      "epoch": 5.075764705882353,
      "grad_norm": 0.773475418301166,
      "learning_rate": 2.4036520657300346e-05,
      "loss": 0.8945,
      "step": 10786
    },
    {
      "epoch": 5.076235294117647,
      "grad_norm": 0.326466701260215,
      "learning_rate": 2.401272460848946e-05,
      "loss": 0.0889,
      "step": 10787
    },
    {
      "epoch": 5.076705882352941,
      "grad_norm": 0.2420821100758766,
      "learning_rate": 2.3988939319609817e-05,
      "loss": 0.0386,
      "step": 10788
    },
    {
      "epoch": 5.077176470588236,
      "grad_norm": 0.4520306414182426,
      "learning_rate": 2.3965164792692833e-05,
      "loss": 0.5117,
      "step": 10789
    },
    {
      "epoch": 5.07764705882353,
      "grad_norm": 0.3599827091362326,
      "learning_rate": 2.3941401029769024e-05,
      "loss": 0.3555,
      "step": 10790
    },
    {
      "epoch": 5.078117647058823,
      "grad_norm": 0.1898105134838104,
      "learning_rate": 2.3917648032867844e-05,
      "loss": 0.0347,
      "step": 10791
    },
    {
      "epoch": 5.078588235294117,
      "grad_norm": 0.4695616641568177,
      "learning_rate": 2.3893905804017977e-05,
      "loss": 0.5078,
      "step": 10792
    },
    {
      "epoch": 5.079058823529412,
      "grad_norm": 0.7698857977531243,
      "learning_rate": 2.387017434524707e-05,
      "loss": 0.5039,
      "step": 10793
    },
    {
      "epoch": 5.079529411764706,
      "grad_norm": 0.29708885490311954,
      "learning_rate": 2.3846453658581927e-05,
      "loss": 0.1846,
      "step": 10794
    },
    {
      "epoch": 5.08,
      "grad_norm": 0.406342141368498,
      "learning_rate": 2.382274374604847e-05,
      "loss": 0.2285,
      "step": 10795
    },
    {
      "epoch": 5.080470588235294,
      "grad_norm": 0.2566755163259756,
      "learning_rate": 2.3799044609671558e-05,
      "loss": 0.0688,
      "step": 10796
    },
    {
      "epoch": 5.0809411764705885,
      "grad_norm": 0.3609181001979962,
      "learning_rate": 2.3775356251475248e-05,
      "loss": 0.4414,
      "step": 10797
    },
    {
      "epoch": 5.081411764705883,
      "grad_norm": 0.3809707529440873,
      "learning_rate": 2.375167867348268e-05,
      "loss": 0.1934,
      "step": 10798
    },
    {
      "epoch": 5.081882352941176,
      "grad_norm": 0.45953677132910864,
      "learning_rate": 2.3728011877715947e-05,
      "loss": 0.2988,
      "step": 10799
    },
    {
      "epoch": 5.08235294117647,
      "grad_norm": 0.34695121050884214,
      "learning_rate": 2.3704355866196373e-05,
      "loss": 0.3965,
      "step": 10800
    },
    {
      "epoch": 5.0828235294117645,
      "grad_norm": 0.3190836043696282,
      "learning_rate": 2.3680710640944328e-05,
      "loss": 0.2393,
      "step": 10801
    },
    {
      "epoch": 5.083294117647059,
      "grad_norm": 0.3800209764965195,
      "learning_rate": 2.3657076203979132e-05,
      "loss": 0.4121,
      "step": 10802
    },
    {
      "epoch": 5.083764705882353,
      "grad_norm": 0.33993236073110233,
      "learning_rate": 2.363345255731935e-05,
      "loss": 0.2422,
      "step": 10803
    },
    {
      "epoch": 5.084235294117647,
      "grad_norm": 0.46349103634477246,
      "learning_rate": 2.360983970298257e-05,
      "loss": 0.5195,
      "step": 10804
    },
    {
      "epoch": 5.084705882352941,
      "grad_norm": 0.32710515764957315,
      "learning_rate": 2.3586237642985396e-05,
      "loss": 0.3887,
      "step": 10805
    },
    {
      "epoch": 5.085176470588236,
      "grad_norm": 0.4316950520208455,
      "learning_rate": 2.356264637934358e-05,
      "loss": 0.4688,
      "step": 10806
    },
    {
      "epoch": 5.085647058823529,
      "grad_norm": 0.31767163829792133,
      "learning_rate": 2.3539065914071963e-05,
      "loss": 0.3184,
      "step": 10807
    },
    {
      "epoch": 5.086117647058823,
      "grad_norm": 0.3506913519530369,
      "learning_rate": 2.351549624918437e-05,
      "loss": 0.1592,
      "step": 10808
    },
    {
      "epoch": 5.086588235294117,
      "grad_norm": 0.3650806924451155,
      "learning_rate": 2.34919373866938e-05,
      "loss": 0.2148,
      "step": 10809
    },
    {
      "epoch": 5.087058823529412,
      "grad_norm": 0.517406762860533,
      "learning_rate": 2.3468389328612348e-05,
      "loss": 0.7734,
      "step": 10810
    },
    {
      "epoch": 5.087529411764706,
      "grad_norm": 0.42088244939958064,
      "learning_rate": 2.344485207695102e-05,
      "loss": 0.4707,
      "step": 10811
    },
    {
      "epoch": 5.088,
      "grad_norm": 0.2824081882724975,
      "learning_rate": 2.3421325633720094e-05,
      "loss": 0.0859,
      "step": 10812
    },
    {
      "epoch": 5.088470588235294,
      "grad_norm": 0.19241202789995024,
      "learning_rate": 2.3397810000928845e-05,
      "loss": 0.0564,
      "step": 10813
    },
    {
      "epoch": 5.0889411764705885,
      "grad_norm": 0.4377016131119603,
      "learning_rate": 2.337430518058557e-05,
      "loss": 0.2949,
      "step": 10814
    },
    {
      "epoch": 5.089411764705883,
      "grad_norm": 0.43973200347256236,
      "learning_rate": 2.335081117469777e-05,
      "loss": 0.2002,
      "step": 10815
    },
    {
      "epoch": 5.089882352941176,
      "grad_norm": 0.42477694598315,
      "learning_rate": 2.3327327985271865e-05,
      "loss": 0.1621,
      "step": 10816
    },
    {
      "epoch": 5.09035294117647,
      "grad_norm": 0.3125832040786432,
      "learning_rate": 2.3303855614313483e-05,
      "loss": 0.0601,
      "step": 10817
    },
    {
      "epoch": 5.0908235294117645,
      "grad_norm": 0.30315903865463295,
      "learning_rate": 2.328039406382727e-05,
      "loss": 0.1582,
      "step": 10818
    },
    {
      "epoch": 5.091294117647059,
      "grad_norm": 0.22845526467675345,
      "learning_rate": 2.325694333581695e-05,
      "loss": 0.0566,
      "step": 10819
    },
    {
      "epoch": 5.091764705882353,
      "grad_norm": 0.33766394673071126,
      "learning_rate": 2.32335034322854e-05,
      "loss": 0.2275,
      "step": 10820
    },
    {
      "epoch": 5.092235294117647,
      "grad_norm": 0.3373428119902627,
      "learning_rate": 2.3210074355234392e-05,
      "loss": 0.3301,
      "step": 10821
    },
    {
      "epoch": 5.092705882352941,
      "grad_norm": 0.4006083269792818,
      "learning_rate": 2.318665610666493e-05,
      "loss": 0.3047,
      "step": 10822
    },
    {
      "epoch": 5.093176470588236,
      "grad_norm": 0.4082294358069784,
      "learning_rate": 2.3163248688577098e-05,
      "loss": 0.1348,
      "step": 10823
    },
    {
      "epoch": 5.09364705882353,
      "grad_norm": 0.4711747591170635,
      "learning_rate": 2.3139852102969906e-05,
      "loss": 0.3965,
      "step": 10824
    },
    {
      "epoch": 5.094117647058823,
      "grad_norm": 0.3855206261011857,
      "learning_rate": 2.31164663518416e-05,
      "loss": 0.1387,
      "step": 10825
    },
    {
      "epoch": 5.094588235294117,
      "grad_norm": 0.25661198364893484,
      "learning_rate": 2.3093091437189476e-05,
      "loss": 0.0913,
      "step": 10826
    },
    {
      "epoch": 5.095058823529412,
      "grad_norm": 0.487592668644679,
      "learning_rate": 2.306972736100977e-05,
      "loss": 0.5039,
      "step": 10827
    },
    {
      "epoch": 5.095529411764706,
      "grad_norm": 0.30487319244065647,
      "learning_rate": 2.304637412529794e-05,
      "loss": 0.2617,
      "step": 10828
    },
    {
      "epoch": 5.096,
      "grad_norm": 0.3957945565416029,
      "learning_rate": 2.3023031732048504e-05,
      "loss": 0.5469,
      "step": 10829
    },
    {
      "epoch": 5.096470588235294,
      "grad_norm": 0.33975954016090293,
      "learning_rate": 2.2999700183254934e-05,
      "loss": 0.1484,
      "step": 10830
    },
    {
      "epoch": 5.0969411764705885,
      "grad_norm": 0.1429448087143421,
      "learning_rate": 2.29763794809099e-05,
      "loss": 0.0219,
      "step": 10831
    },
    {
      "epoch": 5.097411764705883,
      "grad_norm": 0.37521983372284823,
      "learning_rate": 2.2953069627005132e-05,
      "loss": 0.2656,
      "step": 10832
    },
    {
      "epoch": 5.097882352941176,
      "grad_norm": 0.35830507833994685,
      "learning_rate": 2.2929770623531368e-05,
      "loss": 0.1826,
      "step": 10833
    },
    {
      "epoch": 5.09835294117647,
      "grad_norm": 0.45396396388518906,
      "learning_rate": 2.290648247247845e-05,
      "loss": 0.4043,
      "step": 10834
    },
    {
      "epoch": 5.0988235294117645,
      "grad_norm": 0.18786831072042112,
      "learning_rate": 2.2883205175835368e-05,
      "loss": 0.1157,
      "step": 10835
    },
    {
      "epoch": 5.099294117647059,
      "grad_norm": 0.4188641401750368,
      "learning_rate": 2.2859938735590017e-05,
      "loss": 0.2969,
      "step": 10836
    },
    {
      "epoch": 5.099764705882353,
      "grad_norm": 0.43405865794538456,
      "learning_rate": 2.2836683153729522e-05,
      "loss": 0.5234,
      "step": 10837
    },
    {
      "epoch": 5.100235294117647,
      "grad_norm": 0.45822755252556535,
      "learning_rate": 2.2813438432240067e-05,
      "loss": 0.2363,
      "step": 10838
    },
    {
      "epoch": 5.100705882352941,
      "grad_norm": 0.47664918189311617,
      "learning_rate": 2.2790204573106756e-05,
      "loss": 0.5586,
      "step": 10839
    },
    {
      "epoch": 5.101176470588236,
      "grad_norm": 0.35136030598345447,
      "learning_rate": 2.276698157831396e-05,
      "loss": 0.3027,
      "step": 10840
    },
    {
      "epoch": 5.101647058823529,
      "grad_norm": 0.4625631749189069,
      "learning_rate": 2.2743769449845035e-05,
      "loss": 0.3086,
      "step": 10841
    },
    {
      "epoch": 5.102117647058823,
      "grad_norm": 0.4146868822415654,
      "learning_rate": 2.272056818968235e-05,
      "loss": 0.1582,
      "step": 10842
    },
    {
      "epoch": 5.102588235294117,
      "grad_norm": 0.35473943038670874,
      "learning_rate": 2.269737779980748e-05,
      "loss": 0.1748,
      "step": 10843
    },
    {
      "epoch": 5.103058823529412,
      "grad_norm": 0.4810458286825521,
      "learning_rate": 2.2674198282200934e-05,
      "loss": 0.6367,
      "step": 10844
    },
    {
      "epoch": 5.103529411764706,
      "grad_norm": 0.31548421307109875,
      "learning_rate": 2.2651029638842372e-05,
      "loss": 0.209,
      "step": 10845
    },
    {
      "epoch": 5.104,
      "grad_norm": 0.3099725389144918,
      "learning_rate": 2.262787187171058e-05,
      "loss": 0.0747,
      "step": 10846
    },
    {
      "epoch": 5.104470588235294,
      "grad_norm": 0.22134358618299727,
      "learning_rate": 2.2604724982783224e-05,
      "loss": 0.0532,
      "step": 10847
    },
    {
      "epoch": 5.1049411764705885,
      "grad_norm": 0.4536253361929319,
      "learning_rate": 2.258158897403724e-05,
      "loss": 0.2949,
      "step": 10848
    },
    {
      "epoch": 5.105411764705883,
      "grad_norm": 0.35162132605160773,
      "learning_rate": 2.2558463847448566e-05,
      "loss": 0.416,
      "step": 10849
    },
    {
      "epoch": 5.105882352941176,
      "grad_norm": 0.4277665669656716,
      "learning_rate": 2.2535349604992153e-05,
      "loss": 0.2236,
      "step": 10850
    },
    {
      "epoch": 5.10635294117647,
      "grad_norm": 0.3925182944023937,
      "learning_rate": 2.251224624864209e-05,
      "loss": 0.2422,
      "step": 10851
    },
    {
      "epoch": 5.1068235294117645,
      "grad_norm": 0.34482282583656454,
      "learning_rate": 2.2489153780371566e-05,
      "loss": 0.3184,
      "step": 10852
    },
    {
      "epoch": 5.107294117647059,
      "grad_norm": 0.3480897952979181,
      "learning_rate": 2.2466072202152702e-05,
      "loss": 0.2812,
      "step": 10853
    },
    {
      "epoch": 5.107764705882353,
      "grad_norm": 0.47803761508054915,
      "learning_rate": 2.2443001515956823e-05,
      "loss": 0.4336,
      "step": 10854
    },
    {
      "epoch": 5.108235294117647,
      "grad_norm": 0.34318973651435325,
      "learning_rate": 2.24199417237543e-05,
      "loss": 0.418,
      "step": 10855
    },
    {
      "epoch": 5.108705882352941,
      "grad_norm": 0.27315313773970257,
      "learning_rate": 2.2396892827514507e-05,
      "loss": 0.2598,
      "step": 10856
    },
    {
      "epoch": 5.109176470588236,
      "grad_norm": 0.511548353976875,
      "learning_rate": 2.2373854829205957e-05,
      "loss": 0.4258,
      "step": 10857
    },
    {
      "epoch": 5.109647058823529,
      "grad_norm": 0.35853907429627757,
      "learning_rate": 2.235082773079624e-05,
      "loss": 0.2119,
      "step": 10858
    },
    {
      "epoch": 5.110117647058823,
      "grad_norm": 0.4856411366836384,
      "learning_rate": 2.2327811534251933e-05,
      "loss": 0.3652,
      "step": 10859
    },
    {
      "epoch": 5.110588235294117,
      "grad_norm": 0.42798481454011034,
      "learning_rate": 2.230480624153873e-05,
      "loss": 0.4648,
      "step": 10860
    },
    {
      "epoch": 5.111058823529412,
      "grad_norm": 0.37158219178290397,
      "learning_rate": 2.2281811854621445e-05,
      "loss": 0.291,
      "step": 10861
    },
    {
      "epoch": 5.111529411764706,
      "grad_norm": 0.19161525059932244,
      "learning_rate": 2.2258828375463855e-05,
      "loss": 0.0369,
      "step": 10862
    },
    {
      "epoch": 5.112,
      "grad_norm": 0.34625919554713513,
      "learning_rate": 2.223585580602889e-05,
      "loss": 0.1787,
      "step": 10863
    },
    {
      "epoch": 5.112470588235294,
      "grad_norm": 0.20437605344668847,
      "learning_rate": 2.221289414827853e-05,
      "loss": 0.1289,
      "step": 10864
    },
    {
      "epoch": 5.1129411764705885,
      "grad_norm": 0.8887325994543117,
      "learning_rate": 2.218994340417378e-05,
      "loss": 0.9258,
      "step": 10865
    },
    {
      "epoch": 5.113411764705883,
      "grad_norm": 0.36777474650536507,
      "learning_rate": 2.2167003575674824e-05,
      "loss": 0.1777,
      "step": 10866
    },
    {
      "epoch": 5.113882352941176,
      "grad_norm": 0.343725214375303,
      "learning_rate": 2.2144074664740723e-05,
      "loss": 0.0898,
      "step": 10867
    },
    {
      "epoch": 5.11435294117647,
      "grad_norm": 0.2496435422834644,
      "learning_rate": 2.2121156673329794e-05,
      "loss": 0.0603,
      "step": 10868
    },
    {
      "epoch": 5.1148235294117645,
      "grad_norm": 0.2837293841074581,
      "learning_rate": 2.2098249603399348e-05,
      "loss": 0.1162,
      "step": 10869
    },
    {
      "epoch": 5.115294117647059,
      "grad_norm": 1.058312715443899,
      "learning_rate": 2.2075353456905727e-05,
      "loss": 1.0391,
      "step": 10870
    },
    {
      "epoch": 5.115764705882353,
      "grad_norm": 0.3881511865827986,
      "learning_rate": 2.2052468235804393e-05,
      "loss": 0.3438,
      "step": 10871
    },
    {
      "epoch": 5.116235294117647,
      "grad_norm": 0.3405738861597628,
      "learning_rate": 2.2029593942049838e-05,
      "loss": 0.1797,
      "step": 10872
    },
    {
      "epoch": 5.116705882352941,
      "grad_norm": 0.6779733105818943,
      "learning_rate": 2.200673057759565e-05,
      "loss": 0.3066,
      "step": 10873
    },
    {
      "epoch": 5.117176470588236,
      "grad_norm": 0.3056397169434053,
      "learning_rate": 2.1983878144394497e-05,
      "loss": 0.1699,
      "step": 10874
    },
    {
      "epoch": 5.117647058823529,
      "grad_norm": 0.5953907730502153,
      "learning_rate": 2.1961036644398035e-05,
      "loss": 0.5234,
      "step": 10875
    },
    {
      "epoch": 5.118117647058823,
      "grad_norm": 0.6479914956406189,
      "learning_rate": 2.193820607955707e-05,
      "loss": 0.7031,
      "step": 10876
    },
    {
      "epoch": 5.118588235294117,
      "grad_norm": 0.4227354817124151,
      "learning_rate": 2.1915386451821494e-05,
      "loss": 0.1348,
      "step": 10877
    },
    {
      "epoch": 5.119058823529412,
      "grad_norm": 0.33576139111951536,
      "learning_rate": 2.1892577763140112e-05,
      "loss": 0.2129,
      "step": 10878
    },
    {
      "epoch": 5.119529411764706,
      "grad_norm": 0.47938108509155486,
      "learning_rate": 2.186978001546097e-05,
      "loss": 0.4648,
      "step": 10879
    },
    {
      "epoch": 5.12,
      "grad_norm": 0.518896089458212,
      "learning_rate": 2.1846993210731108e-05,
      "loss": 0.5508,
      "step": 10880
    },
    {
      "epoch": 5.120470588235294,
      "grad_norm": 0.36731780330386604,
      "learning_rate": 2.1824217350896585e-05,
      "loss": 0.1069,
      "step": 10881
    },
    {
      "epoch": 5.1209411764705886,
      "grad_norm": 0.41672252370320373,
      "learning_rate": 2.180145243790261e-05,
      "loss": 0.2852,
      "step": 10882
    },
    {
      "epoch": 5.121411764705883,
      "grad_norm": 0.22700535255001375,
      "learning_rate": 2.1778698473693435e-05,
      "loss": 0.0427,
      "step": 10883
    },
    {
      "epoch": 5.121882352941176,
      "grad_norm": 0.30558350000731505,
      "learning_rate": 2.1755955460212282e-05,
      "loss": 0.1094,
      "step": 10884
    },
    {
      "epoch": 5.12235294117647,
      "grad_norm": 0.4223603153853135,
      "learning_rate": 2.1733223399401585e-05,
      "loss": 0.3379,
      "step": 10885
    },
    {
      "epoch": 5.1228235294117646,
      "grad_norm": 0.317074906546524,
      "learning_rate": 2.1710502293202786e-05,
      "loss": 0.1299,
      "step": 10886
    },
    {
      "epoch": 5.123294117647059,
      "grad_norm": 0.285226429085904,
      "learning_rate": 2.1687792143556303e-05,
      "loss": 0.124,
      "step": 10887
    },
    {
      "epoch": 5.123764705882353,
      "grad_norm": 0.4345631407809011,
      "learning_rate": 2.1665092952401754e-05,
      "loss": 0.5781,
      "step": 10888
    },
    {
      "epoch": 5.124235294117647,
      "grad_norm": 0.36584380386946586,
      "learning_rate": 2.1642404721677765e-05,
      "loss": 0.2354,
      "step": 10889
    },
    {
      "epoch": 5.124705882352941,
      "grad_norm": 0.4248632509995961,
      "learning_rate": 2.1619727453321974e-05,
      "loss": 0.2832,
      "step": 10890
    },
    {
      "epoch": 5.125176470588236,
      "grad_norm": 0.3775723699011433,
      "learning_rate": 2.159706114927116e-05,
      "loss": 0.1504,
      "step": 10891
    },
    {
      "epoch": 5.125647058823529,
      "grad_norm": 0.42806232324537613,
      "learning_rate": 2.157440581146118e-05,
      "loss": 0.5039,
      "step": 10892
    },
    {
      "epoch": 5.126117647058823,
      "grad_norm": 0.18565577940315633,
      "learning_rate": 2.155176144182682e-05,
      "loss": 0.0532,
      "step": 10893
    },
    {
      "epoch": 5.1265882352941174,
      "grad_norm": 0.39507314197333954,
      "learning_rate": 2.1529128042302113e-05,
      "loss": 0.1992,
      "step": 10894
    },
    {
      "epoch": 5.127058823529412,
      "grad_norm": 0.4943599919288338,
      "learning_rate": 2.150650561481998e-05,
      "loss": 0.3594,
      "step": 10895
    },
    {
      "epoch": 5.127529411764706,
      "grad_norm": 0.3737355985757874,
      "learning_rate": 2.1483894161312525e-05,
      "loss": 0.4121,
      "step": 10896
    },
    {
      "epoch": 5.128,
      "grad_norm": 0.32727594341873295,
      "learning_rate": 2.1461293683710905e-05,
      "loss": 0.2412,
      "step": 10897
    },
    {
      "epoch": 5.128470588235294,
      "grad_norm": 0.25736305759324446,
      "learning_rate": 2.143870418394526e-05,
      "loss": 0.168,
      "step": 10898
    },
    {
      "epoch": 5.128941176470589,
      "grad_norm": 0.4271439210554061,
      "learning_rate": 2.141612566394486e-05,
      "loss": 0.3145,
      "step": 10899
    },
    {
      "epoch": 5.129411764705883,
      "grad_norm": 0.16353489117461575,
      "learning_rate": 2.1393558125638066e-05,
      "loss": 0.0479,
      "step": 10900
    },
    {
      "epoch": 5.129882352941176,
      "grad_norm": 0.5389530019171385,
      "learning_rate": 2.1371001570952186e-05,
      "loss": 0.6289,
      "step": 10901
    },
    {
      "epoch": 5.13035294117647,
      "grad_norm": 0.3033909152811337,
      "learning_rate": 2.1348456001813674e-05,
      "loss": 0.0981,
      "step": 10902
    },
    {
      "epoch": 5.130823529411765,
      "grad_norm": 0.5067174379835678,
      "learning_rate": 2.132592142014811e-05,
      "loss": 0.4355,
      "step": 10903
    },
    {
      "epoch": 5.131294117647059,
      "grad_norm": 0.46657683351188917,
      "learning_rate": 2.130339782787996e-05,
      "loss": 0.3105,
      "step": 10904
    },
    {
      "epoch": 5.131764705882353,
      "grad_norm": 0.5781729191695523,
      "learning_rate": 2.1280885226932877e-05,
      "loss": 0.2188,
      "step": 10905
    },
    {
      "epoch": 5.132235294117647,
      "grad_norm": 0.3921741533514012,
      "learning_rate": 2.125838361922961e-05,
      "loss": 0.3301,
      "step": 10906
    },
    {
      "epoch": 5.1327058823529415,
      "grad_norm": 0.38411708208499334,
      "learning_rate": 2.1235893006691786e-05,
      "loss": 0.2969,
      "step": 10907
    },
    {
      "epoch": 5.133176470588236,
      "grad_norm": 0.09238044123673622,
      "learning_rate": 2.1213413391240335e-05,
      "loss": 0.0114,
      "step": 10908
    },
    {
      "epoch": 5.133647058823529,
      "grad_norm": 0.36810633557390715,
      "learning_rate": 2.119094477479506e-05,
      "loss": 0.2871,
      "step": 10909
    },
    {
      "epoch": 5.134117647058823,
      "grad_norm": 0.3875651526684002,
      "learning_rate": 2.116848715927489e-05,
      "loss": 0.125,
      "step": 10910
    },
    {
      "epoch": 5.1345882352941175,
      "grad_norm": 0.4482505627803363,
      "learning_rate": 2.114604054659787e-05,
      "loss": 0.5547,
      "step": 10911
    },
    {
      "epoch": 5.135058823529412,
      "grad_norm": 0.5521957975082556,
      "learning_rate": 2.1123604938680987e-05,
      "loss": 0.7461,
      "step": 10912
    },
    {
      "epoch": 5.135529411764706,
      "grad_norm": 0.6331283293011701,
      "learning_rate": 2.1101180337440378e-05,
      "loss": 0.3711,
      "step": 10913
    },
    {
      "epoch": 5.136,
      "grad_norm": 0.4106680331024209,
      "learning_rate": 2.1078766744791238e-05,
      "loss": 0.3047,
      "step": 10914
    },
    {
      "epoch": 5.136470588235294,
      "grad_norm": 0.25330399430596695,
      "learning_rate": 2.1056364162647736e-05,
      "loss": 0.1226,
      "step": 10915
    },
    {
      "epoch": 5.136941176470589,
      "grad_norm": 0.44660344970896226,
      "learning_rate": 2.1033972592923214e-05,
      "loss": 0.3906,
      "step": 10916
    },
    {
      "epoch": 5.137411764705883,
      "grad_norm": 0.3532986935219733,
      "learning_rate": 2.1011592037530023e-05,
      "loss": 0.2217,
      "step": 10917
    },
    {
      "epoch": 5.137882352941176,
      "grad_norm": 0.5017350455509066,
      "learning_rate": 2.098922249837953e-05,
      "loss": 0.4824,
      "step": 10918
    },
    {
      "epoch": 5.13835294117647,
      "grad_norm": 0.3527465717785937,
      "learning_rate": 2.096686397738223e-05,
      "loss": 0.2578,
      "step": 10919
    },
    {
      "epoch": 5.138823529411765,
      "grad_norm": 0.7483748848026234,
      "learning_rate": 2.094451647644769e-05,
      "loss": 0.4023,
      "step": 10920
    },
    {
      "epoch": 5.139294117647059,
      "grad_norm": 0.3476879067135464,
      "learning_rate": 2.0922179997484424e-05,
      "loss": 0.1167,
      "step": 10921
    },
    {
      "epoch": 5.139764705882353,
      "grad_norm": 0.4566259042711356,
      "learning_rate": 2.0899854542400147e-05,
      "loss": 0.3828,
      "step": 10922
    },
    {
      "epoch": 5.140235294117647,
      "grad_norm": 0.25734655576595633,
      "learning_rate": 2.087754011310147e-05,
      "loss": 0.0486,
      "step": 10923
    },
    {
      "epoch": 5.1407058823529415,
      "grad_norm": 0.48339109430458543,
      "learning_rate": 2.0855236711494223e-05,
      "loss": 0.332,
      "step": 10924
    },
    {
      "epoch": 5.141176470588236,
      "grad_norm": 0.38224058234783653,
      "learning_rate": 2.083294433948324e-05,
      "loss": 0.1172,
      "step": 10925
    },
    {
      "epoch": 5.141647058823529,
      "grad_norm": 0.4775986785388747,
      "learning_rate": 2.0810662998972345e-05,
      "loss": 0.4238,
      "step": 10926
    },
    {
      "epoch": 5.142117647058823,
      "grad_norm": 0.41599378894421424,
      "learning_rate": 2.0788392691864504e-05,
      "loss": 0.165,
      "step": 10927
    },
    {
      "epoch": 5.1425882352941175,
      "grad_norm": 0.4301686496927024,
      "learning_rate": 2.0766133420061732e-05,
      "loss": 0.1689,
      "step": 10928
    },
    {
      "epoch": 5.143058823529412,
      "grad_norm": 0.3914605124580477,
      "learning_rate": 2.0743885185465025e-05,
      "loss": 0.3145,
      "step": 10929
    },
    {
      "epoch": 5.143529411764706,
      "grad_norm": 0.3808257013636342,
      "learning_rate": 2.0721647989974514e-05,
      "loss": 0.1953,
      "step": 10930
    },
    {
      "epoch": 5.144,
      "grad_norm": 0.24650397826153939,
      "learning_rate": 2.0699421835489417e-05,
      "loss": 0.0659,
      "step": 10931
    },
    {
      "epoch": 5.144470588235294,
      "grad_norm": 0.40894969396466524,
      "learning_rate": 2.0677206723907857e-05,
      "loss": 0.1138,
      "step": 10932
    },
    {
      "epoch": 5.144941176470589,
      "grad_norm": 0.4400581426933124,
      "learning_rate": 2.0655002657127188e-05,
      "loss": 0.2695,
      "step": 10933
    },
    {
      "epoch": 5.145411764705882,
      "grad_norm": 0.7890252957123995,
      "learning_rate": 2.0632809637043747e-05,
      "loss": 0.9141,
      "step": 10934
    },
    {
      "epoch": 5.145882352941176,
      "grad_norm": 0.39709340597880427,
      "learning_rate": 2.0610627665552875e-05,
      "loss": 0.1963,
      "step": 10935
    },
    {
      "epoch": 5.14635294117647,
      "grad_norm": 0.3385434096763562,
      "learning_rate": 2.058845674454905e-05,
      "loss": 0.0815,
      "step": 10936
    },
    {
      "epoch": 5.146823529411765,
      "grad_norm": 0.27793957161325855,
      "learning_rate": 2.0566296875925813e-05,
      "loss": 0.2061,
      "step": 10937
    },
    {
      "epoch": 5.147294117647059,
      "grad_norm": 0.47180267238865425,
      "learning_rate": 2.054414806157566e-05,
      "loss": 0.2539,
      "step": 10938
    },
    {
      "epoch": 5.147764705882353,
      "grad_norm": 0.39674609966484053,
      "learning_rate": 2.052201030339024e-05,
      "loss": 0.4648,
      "step": 10939
    },
    {
      "epoch": 5.148235294117647,
      "grad_norm": 0.35486128658771804,
      "learning_rate": 2.0499883603260274e-05,
      "loss": 0.1592,
      "step": 10940
    },
    {
      "epoch": 5.1487058823529415,
      "grad_norm": 0.25518108338382467,
      "learning_rate": 2.0477767963075386e-05,
      "loss": 0.0542,
      "step": 10941
    },
    {
      "epoch": 5.149176470588236,
      "grad_norm": 0.3765946611446134,
      "learning_rate": 2.045566338472443e-05,
      "loss": 0.1289,
      "step": 10942
    },
    {
      "epoch": 5.149647058823529,
      "grad_norm": 0.37464970453932206,
      "learning_rate": 2.043356987009526e-05,
      "loss": 0.3574,
      "step": 10943
    },
    {
      "epoch": 5.150117647058823,
      "grad_norm": 0.16242298705568756,
      "learning_rate": 2.041148742107471e-05,
      "loss": 0.0245,
      "step": 10944
    },
    {
      "epoch": 5.1505882352941175,
      "grad_norm": 0.5421452306401199,
      "learning_rate": 2.038941603954879e-05,
      "loss": 0.2832,
      "step": 10945
    },
    {
      "epoch": 5.151058823529412,
      "grad_norm": 0.42390003590548986,
      "learning_rate": 2.036735572740245e-05,
      "loss": 0.1602,
      "step": 10946
    },
    {
      "epoch": 5.151529411764706,
      "grad_norm": 0.40571371740060486,
      "learning_rate": 2.0345306486519787e-05,
      "loss": 0.4238,
      "step": 10947
    },
    {
      "epoch": 5.152,
      "grad_norm": 0.07676481235786317,
      "learning_rate": 2.032326831878392e-05,
      "loss": 0.011,
      "step": 10948
    },
    {
      "epoch": 5.152470588235294,
      "grad_norm": 0.43964211590162483,
      "learning_rate": 2.0301241226076963e-05,
      "loss": 0.5938,
      "step": 10949
    },
    {
      "epoch": 5.152941176470589,
      "grad_norm": 0.4043074217019887,
      "learning_rate": 2.027922521028018e-05,
      "loss": 0.3027,
      "step": 10950
    },
    {
      "epoch": 5.153411764705883,
      "grad_norm": 0.5021161102178239,
      "learning_rate": 2.025722027327385e-05,
      "loss": 0.5898,
      "step": 10951
    },
    {
      "epoch": 5.153882352941176,
      "grad_norm": 0.2891298513249822,
      "learning_rate": 2.0235226416937284e-05,
      "loss": 0.0884,
      "step": 10952
    },
    {
      "epoch": 5.15435294117647,
      "grad_norm": 0.5555850958461166,
      "learning_rate": 2.0213243643148914e-05,
      "loss": 0.5898,
      "step": 10953
    },
    {
      "epoch": 5.154823529411765,
      "grad_norm": 0.48292831998332325,
      "learning_rate": 2.0191271953786104e-05,
      "loss": 0.4102,
      "step": 10954
    },
    {
      "epoch": 5.155294117647059,
      "grad_norm": 0.2270889375001231,
      "learning_rate": 2.016931135072537e-05,
      "loss": 0.0698,
      "step": 10955
    },
    {
      "epoch": 5.155764705882353,
      "grad_norm": 0.9499231826685408,
      "learning_rate": 2.0147361835842296e-05,
      "loss": 0.9336,
      "step": 10956
    },
    {
      "epoch": 5.156235294117647,
      "grad_norm": 0.5181199652324462,
      "learning_rate": 2.0125423411011416e-05,
      "loss": 0.625,
      "step": 10957
    },
    {
      "epoch": 5.1567058823529415,
      "grad_norm": 0.3378258685209499,
      "learning_rate": 2.0103496078106396e-05,
      "loss": 0.1196,
      "step": 10958
    },
    {
      "epoch": 5.157176470588236,
      "grad_norm": 0.3510778522748811,
      "learning_rate": 2.0081579839000005e-05,
      "loss": 0.3184,
      "step": 10959
    },
    {
      "epoch": 5.157647058823529,
      "grad_norm": 0.2895219333232852,
      "learning_rate": 2.00596746955639e-05,
      "loss": 0.1167,
      "step": 10960
    },
    {
      "epoch": 5.158117647058823,
      "grad_norm": 0.3695941656187407,
      "learning_rate": 2.0037780649668912e-05,
      "loss": 0.1641,
      "step": 10961
    },
    {
      "epoch": 5.1585882352941175,
      "grad_norm": 0.31657843681780007,
      "learning_rate": 2.001589770318498e-05,
      "loss": 0.0889,
      "step": 10962
    },
    {
      "epoch": 5.159058823529412,
      "grad_norm": 0.3267063855388944,
      "learning_rate": 1.9994025857980896e-05,
      "loss": 0.1089,
      "step": 10963
    },
    {
      "epoch": 5.159529411764706,
      "grad_norm": 0.46192922559412536,
      "learning_rate": 1.9972165115924695e-05,
      "loss": 0.2158,
      "step": 10964
    },
    {
      "epoch": 5.16,
      "grad_norm": 0.46277494609736747,
      "learning_rate": 1.99503154788834e-05,
      "loss": 0.5312,
      "step": 10965
    },
    {
      "epoch": 5.160470588235294,
      "grad_norm": 0.45181119159493444,
      "learning_rate": 1.9928476948723022e-05,
      "loss": 0.1367,
      "step": 10966
    },
    {
      "epoch": 5.160941176470589,
      "grad_norm": 0.5382040424882057,
      "learning_rate": 1.99066495273087e-05,
      "loss": 0.5273,
      "step": 10967
    },
    {
      "epoch": 5.161411764705882,
      "grad_norm": 0.30948367110833347,
      "learning_rate": 1.9884833216504657e-05,
      "loss": 0.2578,
      "step": 10968
    },
    {
      "epoch": 5.161882352941176,
      "grad_norm": 0.3762078805958822,
      "learning_rate": 1.9863028018174043e-05,
      "loss": 0.2988,
      "step": 10969
    },
    {
      "epoch": 5.16235294117647,
      "grad_norm": 0.4821750559009902,
      "learning_rate": 1.9841233934179145e-05,
      "loss": 0.5977,
      "step": 10970
    },
    {
      "epoch": 5.162823529411765,
      "grad_norm": 0.5423052988005482,
      "learning_rate": 1.9819450966381328e-05,
      "loss": 0.4258,
      "step": 10971
    },
    {
      "epoch": 5.163294117647059,
      "grad_norm": 0.3847311029336597,
      "learning_rate": 1.9797679116640903e-05,
      "loss": 0.4082,
      "step": 10972
    },
    {
      "epoch": 5.163764705882353,
      "grad_norm": 0.3819967284210202,
      "learning_rate": 1.9775918386817346e-05,
      "loss": 0.1699,
      "step": 10973
    },
    {
      "epoch": 5.164235294117647,
      "grad_norm": 0.3726442902183563,
      "learning_rate": 1.9754168778769058e-05,
      "loss": 0.1455,
      "step": 10974
    },
    {
      "epoch": 5.1647058823529415,
      "grad_norm": 0.5709326495699802,
      "learning_rate": 1.9732430294353636e-05,
      "loss": 0.5469,
      "step": 10975
    },
    {
      "epoch": 5.165176470588236,
      "grad_norm": 0.4928398115725743,
      "learning_rate": 1.9710702935427642e-05,
      "loss": 0.6289,
      "step": 10976
    },
    {
      "epoch": 5.165647058823529,
      "grad_norm": 0.17135485615939544,
      "learning_rate": 1.9688986703846658e-05,
      "loss": 0.0273,
      "step": 10977
    },
    {
      "epoch": 5.166117647058823,
      "grad_norm": 0.5936034197957695,
      "learning_rate": 1.9667281601465372e-05,
      "loss": 0.3652,
      "step": 10978
    },
    {
      "epoch": 5.1665882352941175,
      "grad_norm": 0.4272078214908844,
      "learning_rate": 1.964558763013755e-05,
      "loss": 0.6328,
      "step": 10979
    },
    {
      "epoch": 5.167058823529412,
      "grad_norm": 0.40861509550910025,
      "learning_rate": 1.9623904791715907e-05,
      "loss": 0.4609,
      "step": 10980
    },
    {
      "epoch": 5.167529411764706,
      "grad_norm": 0.39422973399199807,
      "learning_rate": 1.9602233088052277e-05,
      "loss": 0.3496,
      "step": 10981
    },
    {
      "epoch": 5.168,
      "grad_norm": 0.3542817170549665,
      "learning_rate": 1.9580572520997566e-05,
      "loss": 0.1562,
      "step": 10982
    },
    {
      "epoch": 5.168470588235294,
      "grad_norm": 0.35240951787943064,
      "learning_rate": 1.9558923092401645e-05,
      "loss": 0.167,
      "step": 10983
    },
    {
      "epoch": 5.168941176470589,
      "grad_norm": 0.39760898399611627,
      "learning_rate": 1.95372848041135e-05,
      "loss": 0.3867,
      "step": 10984
    },
    {
      "epoch": 5.169411764705882,
      "grad_norm": 0.21139147418358717,
      "learning_rate": 1.951565765798117e-05,
      "loss": 0.0366,
      "step": 10985
    },
    {
      "epoch": 5.169882352941176,
      "grad_norm": 0.3814506482374552,
      "learning_rate": 1.9494041655851677e-05,
      "loss": 0.2969,
      "step": 10986
    },
    {
      "epoch": 5.17035294117647,
      "grad_norm": 0.43837498270336517,
      "learning_rate": 1.9472436799571145e-05,
      "loss": 0.5078,
      "step": 10987
    },
    {
      "epoch": 5.170823529411765,
      "grad_norm": 0.28907000536592187,
      "learning_rate": 1.9450843090984797e-05,
      "loss": 0.1182,
      "step": 10988
    },
    {
      "epoch": 5.171294117647059,
      "grad_norm": 0.20293081624944018,
      "learning_rate": 1.9429260531936743e-05,
      "loss": 0.1416,
      "step": 10989
    },
    {
      "epoch": 5.171764705882353,
      "grad_norm": 0.7733621918644589,
      "learning_rate": 1.940768912427027e-05,
      "loss": 0.4766,
      "step": 10990
    },
    {
      "epoch": 5.172235294117647,
      "grad_norm": 0.40870863126833773,
      "learning_rate": 1.9386128869827743e-05,
      "loss": 0.2891,
      "step": 10991
    },
    {
      "epoch": 5.1727058823529415,
      "grad_norm": 0.3322080402634779,
      "learning_rate": 1.936457977045044e-05,
      "loss": 0.2041,
      "step": 10992
    },
    {
      "epoch": 5.173176470588236,
      "grad_norm": 0.25899088956718425,
      "learning_rate": 1.9343041827978785e-05,
      "loss": 0.0825,
      "step": 10993
    },
    {
      "epoch": 5.173647058823529,
      "grad_norm": 0.33746747629503454,
      "learning_rate": 1.9321515044252273e-05,
      "loss": 0.2832,
      "step": 10994
    },
    {
      "epoch": 5.174117647058823,
      "grad_norm": 0.9158035211590401,
      "learning_rate": 1.9299999421109303e-05,
      "loss": 0.6484,
      "step": 10995
    },
    {
      "epoch": 5.1745882352941175,
      "grad_norm": 0.10793740360800018,
      "learning_rate": 1.9278494960387452e-05,
      "loss": 0.0115,
      "step": 10996
    },
    {
      "epoch": 5.175058823529412,
      "grad_norm": 0.3761238742639311,
      "learning_rate": 1.9257001663923338e-05,
      "loss": 0.1396,
      "step": 10997
    },
    {
      "epoch": 5.175529411764706,
      "grad_norm": 0.21940627255626485,
      "learning_rate": 1.9235519533552556e-05,
      "loss": 0.0569,
      "step": 10998
    },
    {
      "epoch": 5.176,
      "grad_norm": 0.15947582477041838,
      "learning_rate": 1.921404857110984e-05,
      "loss": 0.0374,
      "step": 10999
    },
    {
      "epoch": 5.176470588235294,
      "grad_norm": 0.3278119580331948,
      "learning_rate": 1.9192588778428842e-05,
      "loss": 0.1123,
      "step": 11000
    },
    {
      "epoch": 5.176941176470589,
      "grad_norm": 0.30425951946347923,
      "learning_rate": 1.9171140157342407e-05,
      "loss": 0.0757,
      "step": 11001
    },
    {
      "epoch": 5.177411764705882,
      "grad_norm": 0.4057245463546872,
      "learning_rate": 1.9149702709682292e-05,
      "loss": 0.3652,
      "step": 11002
    },
    {
      "epoch": 5.177882352941176,
      "grad_norm": 0.5159957461832355,
      "learning_rate": 1.9128276437279362e-05,
      "loss": 0.3105,
      "step": 11003
    },
    {
      "epoch": 5.17835294117647,
      "grad_norm": 0.2909293333719154,
      "learning_rate": 1.9106861341963587e-05,
      "loss": 0.0986,
      "step": 11004
    },
    {
      "epoch": 5.178823529411765,
      "grad_norm": 0.3601160153354235,
      "learning_rate": 1.9085457425563855e-05,
      "loss": 0.2305,
      "step": 11005
    },
    {
      "epoch": 5.179294117647059,
      "grad_norm": 0.3572322679568525,
      "learning_rate": 1.9064064689908187e-05,
      "loss": 0.2949,
      "step": 11006
    },
    {
      "epoch": 5.179764705882353,
      "grad_norm": 0.36230824511770304,
      "learning_rate": 1.9042683136823654e-05,
      "loss": 0.3652,
      "step": 11007
    },
    {
      "epoch": 5.180235294117647,
      "grad_norm": 0.4250792930190004,
      "learning_rate": 1.9021312768136314e-05,
      "loss": 0.2314,
      "step": 11008
    },
    {
      "epoch": 5.1807058823529415,
      "grad_norm": 0.2927655854292466,
      "learning_rate": 1.899995358567129e-05,
      "loss": 0.1729,
      "step": 11009
    },
    {
      "epoch": 5.181176470588236,
      "grad_norm": 0.3241778672155132,
      "learning_rate": 1.897860559125282e-05,
      "loss": 0.1338,
      "step": 11010
    },
    {
      "epoch": 5.181647058823529,
      "grad_norm": 0.561719195339048,
      "learning_rate": 1.895726878670405e-05,
      "loss": 0.7656,
      "step": 11011
    },
    {
      "epoch": 5.182117647058823,
      "grad_norm": 0.44258115344534726,
      "learning_rate": 1.8935943173847302e-05,
      "loss": 0.2578,
      "step": 11012
    },
    {
      "epoch": 5.1825882352941175,
      "grad_norm": 0.342161697399961,
      "learning_rate": 1.8914628754503885e-05,
      "loss": 0.2637,
      "step": 11013
    },
    {
      "epoch": 5.183058823529412,
      "grad_norm": 0.42507689945251054,
      "learning_rate": 1.889332553049413e-05,
      "loss": 0.3809,
      "step": 11014
    },
    {
      "epoch": 5.183529411764706,
      "grad_norm": 0.3814678814805855,
      "learning_rate": 1.887203350363743e-05,
      "loss": 0.2891,
      "step": 11015
    },
    {
      "epoch": 5.184,
      "grad_norm": 0.2697009325450263,
      "learning_rate": 1.8850752675752295e-05,
      "loss": 0.0547,
      "step": 11016
    },
    {
      "epoch": 5.184470588235294,
      "grad_norm": 0.5447607919355523,
      "learning_rate": 1.8829483048656124e-05,
      "loss": 0.4023,
      "step": 11017
    },
    {
      "epoch": 5.184941176470589,
      "grad_norm": 0.3364354209097221,
      "learning_rate": 1.8808224624165497e-05,
      "loss": 0.1138,
      "step": 11018
    },
    {
      "epoch": 5.185411764705882,
      "grad_norm": 0.41490797115120753,
      "learning_rate": 1.878697740409601e-05,
      "loss": 0.2793,
      "step": 11019
    },
    {
      "epoch": 5.185882352941176,
      "grad_norm": 0.4656388126739465,
      "learning_rate": 1.8765741390262212e-05,
      "loss": 0.291,
      "step": 11020
    },
    {
      "epoch": 5.18635294117647,
      "grad_norm": 0.4877935073043651,
      "learning_rate": 1.8744516584477803e-05,
      "loss": 0.5938,
      "step": 11021
    },
    {
      "epoch": 5.186823529411765,
      "grad_norm": 0.2480020353736314,
      "learning_rate": 1.87233029885555e-05,
      "loss": 0.0403,
      "step": 11022
    },
    {
      "epoch": 5.187294117647059,
      "grad_norm": 0.34903488227196344,
      "learning_rate": 1.8702100604307018e-05,
      "loss": 0.2695,
      "step": 11023
    },
    {
      "epoch": 5.187764705882353,
      "grad_norm": 0.32350697530705474,
      "learning_rate": 1.868090943354318e-05,
      "loss": 0.3398,
      "step": 11024
    },
    {
      "epoch": 5.188235294117647,
      "grad_norm": 0.4731314240800775,
      "learning_rate": 1.8659729478073764e-05,
      "loss": 0.1777,
      "step": 11025
    },
    {
      "epoch": 5.1887058823529415,
      "grad_norm": 0.4021433042338483,
      "learning_rate": 1.8638560739707664e-05,
      "loss": 0.2217,
      "step": 11026
    },
    {
      "epoch": 5.189176470588236,
      "grad_norm": 0.10093955184190519,
      "learning_rate": 1.8617403220252846e-05,
      "loss": 0.0059,
      "step": 11027
    },
    {
      "epoch": 5.189647058823529,
      "grad_norm": 0.24472872672164112,
      "learning_rate": 1.8596256921516166e-05,
      "loss": 0.0327,
      "step": 11028
    },
    {
      "epoch": 5.190117647058823,
      "grad_norm": 0.3169587504636523,
      "learning_rate": 1.8575121845303698e-05,
      "loss": 0.3047,
      "step": 11029
    },
    {
      "epoch": 5.1905882352941175,
      "grad_norm": 0.42808281731937475,
      "learning_rate": 1.8553997993420495e-05,
      "loss": 0.332,
      "step": 11030
    },
    {
      "epoch": 5.191058823529412,
      "grad_norm": 0.33440469038108694,
      "learning_rate": 1.8532885367670547e-05,
      "loss": 0.1611,
      "step": 11031
    },
    {
      "epoch": 5.191529411764706,
      "grad_norm": 0.4130288997202829,
      "learning_rate": 1.8511783969857046e-05,
      "loss": 0.1611,
      "step": 11032
    },
    {
      "epoch": 5.192,
      "grad_norm": 0.3448861624019751,
      "learning_rate": 1.849069380178217e-05,
      "loss": 0.0547,
      "step": 11033
    },
    {
      "epoch": 5.192470588235294,
      "grad_norm": 0.4140571160266421,
      "learning_rate": 1.8469614865247075e-05,
      "loss": 0.1641,
      "step": 11034
    },
    {
      "epoch": 5.192941176470589,
      "grad_norm": 0.3438996710251541,
      "learning_rate": 1.8448547162052002e-05,
      "loss": 0.1963,
      "step": 11035
    },
    {
      "epoch": 5.193411764705882,
      "grad_norm": 0.36025474210263486,
      "learning_rate": 1.8427490693996298e-05,
      "loss": 0.0879,
      "step": 11036
    },
    {
      "epoch": 5.193882352941176,
      "grad_norm": 0.326107132309151,
      "learning_rate": 1.840644546287821e-05,
      "loss": 0.1641,
      "step": 11037
    },
    {
      "epoch": 5.19435294117647,
      "grad_norm": 0.6909183870340792,
      "learning_rate": 1.838541147049515e-05,
      "loss": 0.332,
      "step": 11038
    },
    {
      "epoch": 5.194823529411765,
      "grad_norm": 0.30194181391294395,
      "learning_rate": 1.8364388718643542e-05,
      "loss": 0.2324,
      "step": 11039
    },
    {
      "epoch": 5.195294117647059,
      "grad_norm": 0.5735216706228551,
      "learning_rate": 1.8343377209118755e-05,
      "loss": 0.5469,
      "step": 11040
    },
    {
      "epoch": 5.195764705882353,
      "grad_norm": 0.44280431299654266,
      "learning_rate": 1.8322376943715382e-05,
      "loss": 0.3301,
      "step": 11041
    },
    {
      "epoch": 5.196235294117647,
      "grad_norm": 0.5225767250122546,
      "learning_rate": 1.8301387924226863e-05,
      "loss": 0.3984,
      "step": 11042
    },
    {
      "epoch": 5.1967058823529415,
      "grad_norm": 0.49834655042821535,
      "learning_rate": 1.828041015244579e-05,
      "loss": 0.2402,
      "step": 11043
    },
    {
      "epoch": 5.197176470588236,
      "grad_norm": 0.32695648330353805,
      "learning_rate": 1.82594436301638e-05,
      "loss": 0.0723,
      "step": 11044
    },
    {
      "epoch": 5.197647058823529,
      "grad_norm": 0.5285649703102959,
      "learning_rate": 1.8238488359171476e-05,
      "loss": 0.7109,
      "step": 11045
    },
    {
      "epoch": 5.198117647058823,
      "grad_norm": 0.43865939984645896,
      "learning_rate": 1.821754434125854e-05,
      "loss": 0.3359,
      "step": 11046
    },
    {
      "epoch": 5.1985882352941175,
      "grad_norm": 0.4506562570600854,
      "learning_rate": 1.8196611578213738e-05,
      "loss": 0.2266,
      "step": 11047
    },
    {
      "epoch": 5.199058823529412,
      "grad_norm": 0.34277935788301295,
      "learning_rate": 1.817569007182476e-05,
      "loss": 0.2197,
      "step": 11048
    },
    {
      "epoch": 5.199529411764706,
      "grad_norm": 0.421056922483593,
      "learning_rate": 1.815477982387846e-05,
      "loss": 0.0996,
      "step": 11049
    },
    {
      "epoch": 5.2,
      "grad_norm": 0.3572110213437531,
      "learning_rate": 1.813388083616068e-05,
      "loss": 0.1035,
      "step": 11050
    },
    {
      "epoch": 5.200470588235294,
      "grad_norm": 0.309427759594951,
      "learning_rate": 1.8112993110456235e-05,
      "loss": 0.1865,
      "step": 11051
    },
    {
      "epoch": 5.200941176470589,
      "grad_norm": 0.6106355171550911,
      "learning_rate": 1.8092116648549136e-05,
      "loss": 0.5039,
      "step": 11052
    },
    {
      "epoch": 5.201411764705882,
      "grad_norm": 0.4366970105055448,
      "learning_rate": 1.8071251452222224e-05,
      "loss": 0.2246,
      "step": 11053
    },
    {
      "epoch": 5.201882352941176,
      "grad_norm": 0.39062833870987845,
      "learning_rate": 1.8050397523257555e-05,
      "loss": 0.1172,
      "step": 11054
    },
    {
      "epoch": 5.20235294117647,
      "grad_norm": 0.3340336128665545,
      "learning_rate": 1.8029554863436186e-05,
      "loss": 0.1484,
      "step": 11055
    },
    {
      "epoch": 5.202823529411765,
      "grad_norm": 0.6583361395306012,
      "learning_rate": 1.8008723474538095e-05,
      "loss": 0.4219,
      "step": 11056
    },
    {
      "epoch": 5.203294117647059,
      "grad_norm": 0.8918662778686935,
      "learning_rate": 1.798790335834244e-05,
      "loss": 0.5703,
      "step": 11057
    },
    {
      "epoch": 5.203764705882353,
      "grad_norm": 0.3204515318404005,
      "learning_rate": 1.796709451662739e-05,
      "loss": 0.2676,
      "step": 11058
    },
    {
      "epoch": 5.204235294117647,
      "grad_norm": 0.6331832504497599,
      "learning_rate": 1.7946296951170065e-05,
      "loss": 0.4824,
      "step": 11059
    },
    {
      "epoch": 5.2047058823529415,
      "grad_norm": 0.345100918716395,
      "learning_rate": 1.792551066374668e-05,
      "loss": 0.2422,
      "step": 11060
    },
    {
      "epoch": 5.205176470588235,
      "grad_norm": 0.8031419122794448,
      "learning_rate": 1.7904735656132545e-05,
      "loss": 0.3867,
      "step": 11061
    },
    {
      "epoch": 5.205647058823529,
      "grad_norm": 0.31585317157957743,
      "learning_rate": 1.7883971930101883e-05,
      "loss": 0.0928,
      "step": 11062
    },
    {
      "epoch": 5.206117647058823,
      "grad_norm": 0.330456284839654,
      "learning_rate": 1.786321948742803e-05,
      "loss": 0.2715,
      "step": 11063
    },
    {
      "epoch": 5.2065882352941175,
      "grad_norm": 0.5273687324938677,
      "learning_rate": 1.78424783298834e-05,
      "loss": 0.4062,
      "step": 11064
    },
    {
      "epoch": 5.207058823529412,
      "grad_norm": 0.24431985996557648,
      "learning_rate": 1.782174845923931e-05,
      "loss": 0.0388,
      "step": 11065
    },
    {
      "epoch": 5.207529411764706,
      "grad_norm": 0.19350737405359975,
      "learning_rate": 1.7801029877266237e-05,
      "loss": 0.0747,
      "step": 11066
    },
    {
      "epoch": 5.208,
      "grad_norm": 0.3323918589719757,
      "learning_rate": 1.778032258573368e-05,
      "loss": 0.1797,
      "step": 11067
    },
    {
      "epoch": 5.208470588235294,
      "grad_norm": 0.34922007885131195,
      "learning_rate": 1.7759626586410074e-05,
      "loss": 0.4336,
      "step": 11068
    },
    {
      "epoch": 5.208941176470589,
      "grad_norm": 0.3556989432640917,
      "learning_rate": 1.7738941881063007e-05,
      "loss": 0.3164,
      "step": 11069
    },
    {
      "epoch": 5.209411764705882,
      "grad_norm": 0.32244860643183604,
      "learning_rate": 1.7718268471459045e-05,
      "loss": 0.0977,
      "step": 11070
    },
    {
      "epoch": 5.209882352941176,
      "grad_norm": 0.3960957587887785,
      "learning_rate": 1.769760635936378e-05,
      "loss": 0.3555,
      "step": 11071
    },
    {
      "epoch": 5.21035294117647,
      "grad_norm": 0.3024849359202911,
      "learning_rate": 1.7676955546541867e-05,
      "loss": 0.1582,
      "step": 11072
    },
    {
      "epoch": 5.210823529411765,
      "grad_norm": 0.3477944371747678,
      "learning_rate": 1.7656316034757024e-05,
      "loss": 0.1904,
      "step": 11073
    },
    {
      "epoch": 5.211294117647059,
      "grad_norm": 0.3408043732396899,
      "learning_rate": 1.7635687825771894e-05,
      "loss": 0.0874,
      "step": 11074
    },
    {
      "epoch": 5.211764705882353,
      "grad_norm": 0.48149485741777576,
      "learning_rate": 1.7615070921348316e-05,
      "loss": 0.3945,
      "step": 11075
    },
    {
      "epoch": 5.212235294117647,
      "grad_norm": 0.5672334253446016,
      "learning_rate": 1.7594465323246987e-05,
      "loss": 0.4785,
      "step": 11076
    },
    {
      "epoch": 5.2127058823529415,
      "grad_norm": 0.3796409475114746,
      "learning_rate": 1.757387103322776e-05,
      "loss": 0.2559,
      "step": 11077
    },
    {
      "epoch": 5.213176470588236,
      "grad_norm": 0.2641323287927041,
      "learning_rate": 1.7553288053049515e-05,
      "loss": 0.1377,
      "step": 11078
    },
    {
      "epoch": 5.213647058823529,
      "grad_norm": 0.4055892826234095,
      "learning_rate": 1.7532716384470108e-05,
      "loss": 0.2139,
      "step": 11079
    },
    {
      "epoch": 5.214117647058823,
      "grad_norm": 0.1141212318592223,
      "learning_rate": 1.7512156029246453e-05,
      "loss": 0.0167,
      "step": 11080
    },
    {
      "epoch": 5.2145882352941175,
      "grad_norm": 0.4144151176760045,
      "learning_rate": 1.7491606989134566e-05,
      "loss": 0.1895,
      "step": 11081
    },
    {
      "epoch": 5.215058823529412,
      "grad_norm": 0.8695226547975345,
      "learning_rate": 1.7471069265889353e-05,
      "loss": 0.7852,
      "step": 11082
    },
    {
      "epoch": 5.215529411764706,
      "grad_norm": 0.4121840060676502,
      "learning_rate": 1.7450542861264883e-05,
      "loss": 0.166,
      "step": 11083
    },
    {
      "epoch": 5.216,
      "grad_norm": 0.33142072244199106,
      "learning_rate": 1.7430027777014217e-05,
      "loss": 0.083,
      "step": 11084
    },
    {
      "epoch": 5.216470588235294,
      "grad_norm": 0.9440837446281947,
      "learning_rate": 1.7409524014889388e-05,
      "loss": 0.4355,
      "step": 11085
    },
    {
      "epoch": 5.216941176470589,
      "grad_norm": 0.2587811922746792,
      "learning_rate": 1.738903157664161e-05,
      "loss": 0.1572,
      "step": 11086
    },
    {
      "epoch": 5.217411764705882,
      "grad_norm": 0.28177397406835725,
      "learning_rate": 1.736855046402097e-05,
      "loss": 0.0942,
      "step": 11087
    },
    {
      "epoch": 5.217882352941176,
      "grad_norm": 0.34539740523676976,
      "learning_rate": 1.7348080678776665e-05,
      "loss": 0.1079,
      "step": 11088
    },
    {
      "epoch": 5.21835294117647,
      "grad_norm": 0.5147099141725945,
      "learning_rate": 1.732762222265695e-05,
      "loss": 0.2539,
      "step": 11089
    },
    {
      "epoch": 5.218823529411765,
      "grad_norm": 0.33400524796253783,
      "learning_rate": 1.7307175097409044e-05,
      "loss": 0.1309,
      "step": 11090
    },
    {
      "epoch": 5.219294117647059,
      "grad_norm": 0.5074715028948964,
      "learning_rate": 1.7286739304779214e-05,
      "loss": 0.3574,
      "step": 11091
    },
    {
      "epoch": 5.219764705882353,
      "grad_norm": 0.3561262954117963,
      "learning_rate": 1.7266314846512848e-05,
      "loss": 0.2402,
      "step": 11092
    },
    {
      "epoch": 5.220235294117647,
      "grad_norm": 0.5215214124299496,
      "learning_rate": 1.7245901724354218e-05,
      "loss": 0.6211,
      "step": 11093
    },
    {
      "epoch": 5.2207058823529415,
      "grad_norm": 0.3155448347196214,
      "learning_rate": 1.7225499940046726e-05,
      "loss": 0.3086,
      "step": 11094
    },
    {
      "epoch": 5.221176470588235,
      "grad_norm": 0.47010090002036026,
      "learning_rate": 1.720510949533282e-05,
      "loss": 0.2832,
      "step": 11095
    },
    {
      "epoch": 5.221647058823529,
      "grad_norm": 0.40893610046442774,
      "learning_rate": 1.71847303919539e-05,
      "loss": 0.2178,
      "step": 11096
    },
    {
      "epoch": 5.222117647058823,
      "grad_norm": 0.4115057006211163,
      "learning_rate": 1.7164362631650457e-05,
      "loss": 0.4004,
      "step": 11097
    },
    {
      "epoch": 5.2225882352941175,
      "grad_norm": 0.4948465935476294,
      "learning_rate": 1.7144006216162036e-05,
      "loss": 0.6367,
      "step": 11098
    },
    {
      "epoch": 5.223058823529412,
      "grad_norm": 0.4780061421525178,
      "learning_rate": 1.7123661147227096e-05,
      "loss": 0.3691,
      "step": 11099
    },
    {
      "epoch": 5.223529411764706,
      "grad_norm": 0.23980424128932512,
      "learning_rate": 1.7103327426583265e-05,
      "loss": 0.106,
      "step": 11100
    },
    {
      "epoch": 5.224,
      "grad_norm": 0.5684274163677544,
      "learning_rate": 1.7083005055967146e-05,
      "loss": 0.3535,
      "step": 11101
    },
    {
      "epoch": 5.224470588235294,
      "grad_norm": 0.3076055377416615,
      "learning_rate": 1.7062694037114326e-05,
      "loss": 0.0576,
      "step": 11102
    },
    {
      "epoch": 5.224941176470589,
      "grad_norm": 0.5539229239940775,
      "learning_rate": 1.7042394371759516e-05,
      "loss": 0.3574,
      "step": 11103
    },
    {
      "epoch": 5.225411764705882,
      "grad_norm": 0.39947679314482964,
      "learning_rate": 1.7022106061636342e-05,
      "loss": 0.1582,
      "step": 11104
    },
    {
      "epoch": 5.225882352941176,
      "grad_norm": 0.4778528752130699,
      "learning_rate": 1.700182910847757e-05,
      "loss": 0.1982,
      "step": 11105
    },
    {
      "epoch": 5.22635294117647,
      "grad_norm": 0.19030578625072098,
      "learning_rate": 1.6981563514014967e-05,
      "loss": 0.0447,
      "step": 11106
    },
    {
      "epoch": 5.226823529411765,
      "grad_norm": 0.5342942439637267,
      "learning_rate": 1.6961309279979278e-05,
      "loss": 0.5156,
      "step": 11107
    },
    {
      "epoch": 5.227294117647059,
      "grad_norm": 0.47587531366523,
      "learning_rate": 1.6941066408100308e-05,
      "loss": 0.6211,
      "step": 11108
    },
    {
      "epoch": 5.227764705882353,
      "grad_norm": 0.20891748184050368,
      "learning_rate": 1.6920834900106967e-05,
      "loss": 0.0374,
      "step": 11109
    },
    {
      "epoch": 5.228235294117647,
      "grad_norm": 0.5043469807307505,
      "learning_rate": 1.690061475772703e-05,
      "loss": 0.5547,
      "step": 11110
    },
    {
      "epoch": 5.2287058823529415,
      "grad_norm": 0.3443973495653574,
      "learning_rate": 1.6880405982687455e-05,
      "loss": 0.2793,
      "step": 11111
    },
    {
      "epoch": 5.229176470588235,
      "grad_norm": 0.4292578803576571,
      "learning_rate": 1.6860208576714173e-05,
      "loss": 0.3867,
      "step": 11112
    },
    {
      "epoch": 5.229647058823529,
      "grad_norm": 0.19706136763768067,
      "learning_rate": 1.68400225415321e-05,
      "loss": 0.0471,
      "step": 11113
    },
    {
      "epoch": 5.230117647058823,
      "grad_norm": 0.4486384829765295,
      "learning_rate": 1.6819847878865255e-05,
      "loss": 0.3535,
      "step": 11114
    },
    {
      "epoch": 5.2305882352941175,
      "grad_norm": 0.39394251934240976,
      "learning_rate": 1.6799684590436663e-05,
      "loss": 0.2412,
      "step": 11115
    },
    {
      "epoch": 5.231058823529412,
      "grad_norm": 0.3255335663103779,
      "learning_rate": 1.6779532677968327e-05,
      "loss": 0.1104,
      "step": 11116
    },
    {
      "epoch": 5.231529411764706,
      "grad_norm": 0.4700977883266481,
      "learning_rate": 1.6759392143181354e-05,
      "loss": 0.377,
      "step": 11117
    },
    {
      "epoch": 5.232,
      "grad_norm": 0.3242856142055197,
      "learning_rate": 1.6739262987795854e-05,
      "loss": 0.0693,
      "step": 11118
    },
    {
      "epoch": 5.232470588235294,
      "grad_norm": 0.3895306236094225,
      "learning_rate": 1.6719145213530898e-05,
      "loss": 0.2832,
      "step": 11119
    },
    {
      "epoch": 5.232941176470589,
      "grad_norm": 0.2630859474544982,
      "learning_rate": 1.669903882210468e-05,
      "loss": 0.2734,
      "step": 11120
    },
    {
      "epoch": 5.233411764705882,
      "grad_norm": 0.46035975308199567,
      "learning_rate": 1.667894381523443e-05,
      "loss": 0.3398,
      "step": 11121
    },
    {
      "epoch": 5.233882352941176,
      "grad_norm": 0.39254752401813986,
      "learning_rate": 1.665886019463627e-05,
      "loss": 0.2969,
      "step": 11122
    },
    {
      "epoch": 5.23435294117647,
      "grad_norm": 0.44516553319177815,
      "learning_rate": 1.6638787962025495e-05,
      "loss": 0.165,
      "step": 11123
    },
    {
      "epoch": 5.234823529411765,
      "grad_norm": 0.35374791294208263,
      "learning_rate": 1.6618727119116388e-05,
      "loss": 0.3848,
      "step": 11124
    },
    {
      "epoch": 5.235294117647059,
      "grad_norm": 0.4772396910772002,
      "learning_rate": 1.6598677667622175e-05,
      "loss": 0.416,
      "step": 11125
    },
    {
      "epoch": 5.235764705882353,
      "grad_norm": 0.22293972981186586,
      "learning_rate": 1.657863960925525e-05,
      "loss": 0.0383,
      "step": 11126
    },
    {
      "epoch": 5.236235294117647,
      "grad_norm": 0.5217155775850898,
      "learning_rate": 1.655861294572692e-05,
      "loss": 0.5,
      "step": 11127
    },
    {
      "epoch": 5.2367058823529415,
      "grad_norm": 0.4393931907145245,
      "learning_rate": 1.6538597678747555e-05,
      "loss": 0.4355,
      "step": 11128
    },
    {
      "epoch": 5.237176470588235,
      "grad_norm": 0.36745642506965,
      "learning_rate": 1.651859381002658e-05,
      "loss": 0.1279,
      "step": 11129
    },
    {
      "epoch": 5.237647058823529,
      "grad_norm": 0.3738812005610708,
      "learning_rate": 1.649860134127241e-05,
      "loss": 0.1118,
      "step": 11130
    },
    {
      "epoch": 5.238117647058823,
      "grad_norm": 0.3221945300990321,
      "learning_rate": 1.647862027419251e-05,
      "loss": 0.208,
      "step": 11131
    },
    {
      "epoch": 5.2385882352941175,
      "grad_norm": 0.4419869869065061,
      "learning_rate": 1.6458650610493385e-05,
      "loss": 0.2988,
      "step": 11132
    },
    {
      "epoch": 5.239058823529412,
      "grad_norm": 0.47458446275147503,
      "learning_rate": 1.6438692351880482e-05,
      "loss": 0.2363,
      "step": 11133
    },
    {
      "epoch": 5.239529411764706,
      "grad_norm": 0.4014149984563704,
      "learning_rate": 1.6418745500058395e-05,
      "loss": 0.3379,
      "step": 11134
    },
    {
      "epoch": 5.24,
      "grad_norm": 0.5533693506909879,
      "learning_rate": 1.639881005673064e-05,
      "loss": 0.5859,
      "step": 11135
    },
    {
      "epoch": 5.240470588235294,
      "grad_norm": 0.5069984109727871,
      "learning_rate": 1.6378886023599803e-05,
      "loss": 0.6172,
      "step": 11136
    },
    {
      "epoch": 5.240941176470589,
      "grad_norm": 0.538064673627957,
      "learning_rate": 1.6358973402367524e-05,
      "loss": 0.6172,
      "step": 11137
    },
    {
      "epoch": 5.241411764705882,
      "grad_norm": 0.4410089534218287,
      "learning_rate": 1.6339072194734414e-05,
      "loss": 0.3164,
      "step": 11138
    },
    {
      "epoch": 5.241882352941176,
      "grad_norm": 0.3404946507666312,
      "learning_rate": 1.6319182402400144e-05,
      "loss": 0.123,
      "step": 11139
    },
    {
      "epoch": 5.24235294117647,
      "grad_norm": 0.39949736948533177,
      "learning_rate": 1.629930402706342e-05,
      "loss": 0.1367,
      "step": 11140
    },
    {
      "epoch": 5.242823529411765,
      "grad_norm": 0.355308414432251,
      "learning_rate": 1.6279437070421885e-05,
      "loss": 0.1494,
      "step": 11141
    },
    {
      "epoch": 5.243294117647059,
      "grad_norm": 0.27900618902584895,
      "learning_rate": 1.625958153417235e-05,
      "loss": 0.334,
      "step": 11142
    },
    {
      "epoch": 5.243764705882353,
      "grad_norm": 0.4439455475108176,
      "learning_rate": 1.6239737420010558e-05,
      "loss": 0.2441,
      "step": 11143
    },
    {
      "epoch": 5.244235294117647,
      "grad_norm": 0.23874436059548207,
      "learning_rate": 1.621990472963125e-05,
      "loss": 0.1387,
      "step": 11144
    },
    {
      "epoch": 5.2447058823529416,
      "grad_norm": 0.44286087958071574,
      "learning_rate": 1.620008346472826e-05,
      "loss": 0.3574,
      "step": 11145
    },
    {
      "epoch": 5.245176470588235,
      "grad_norm": 0.3754660499363886,
      "learning_rate": 1.618027362699446e-05,
      "loss": 0.4473,
      "step": 11146
    },
    {
      "epoch": 5.245647058823529,
      "grad_norm": 0.34978762289947235,
      "learning_rate": 1.6160475218121654e-05,
      "loss": 0.2773,
      "step": 11147
    },
    {
      "epoch": 5.246117647058823,
      "grad_norm": 0.3626468108330472,
      "learning_rate": 1.6140688239800736e-05,
      "loss": 0.2197,
      "step": 11148
    },
    {
      "epoch": 5.2465882352941176,
      "grad_norm": 0.6196327746562178,
      "learning_rate": 1.6120912693721654e-05,
      "loss": 0.2051,
      "step": 11149
    },
    {
      "epoch": 5.247058823529412,
      "grad_norm": 0.3174013640080981,
      "learning_rate": 1.6101148581573274e-05,
      "loss": 0.2441,
      "step": 11150
    },
    {
      "epoch": 5.247529411764706,
      "grad_norm": 0.40049591879014923,
      "learning_rate": 1.6081395905043565e-05,
      "loss": 0.3379,
      "step": 11151
    },
    {
      "epoch": 5.248,
      "grad_norm": 0.5330460827166561,
      "learning_rate": 1.606165466581954e-05,
      "loss": 0.4395,
      "step": 11152
    },
    {
      "epoch": 5.248470588235294,
      "grad_norm": 0.3968476874091854,
      "learning_rate": 1.6041924865587152e-05,
      "loss": 0.1279,
      "step": 11153
    },
    {
      "epoch": 5.248941176470589,
      "grad_norm": 0.26675654074584704,
      "learning_rate": 1.602220650603147e-05,
      "loss": 0.1807,
      "step": 11154
    },
    {
      "epoch": 5.249411764705882,
      "grad_norm": 0.3992593783504691,
      "learning_rate": 1.600249958883648e-05,
      "loss": 0.2871,
      "step": 11155
    },
    {
      "epoch": 5.249882352941176,
      "grad_norm": 0.4375573201445125,
      "learning_rate": 1.5982804115685273e-05,
      "loss": 0.3438,
      "step": 11156
    },
    {
      "epoch": 5.2503529411764704,
      "grad_norm": 0.5663727394676625,
      "learning_rate": 1.5963120088259977e-05,
      "loss": 0.3984,
      "step": 11157
    },
    {
      "epoch": 5.250823529411765,
      "grad_norm": 0.4312049582586236,
      "learning_rate": 1.594344750824164e-05,
      "loss": 0.4707,
      "step": 11158
    },
    {
      "epoch": 5.251294117647059,
      "grad_norm": 0.2723849989504028,
      "learning_rate": 1.5923786377310433e-05,
      "loss": 0.0728,
      "step": 11159
    },
    {
      "epoch": 5.251764705882353,
      "grad_norm": 0.42797443479026087,
      "learning_rate": 1.5904136697145542e-05,
      "loss": 0.2598,
      "step": 11160
    },
    {
      "epoch": 5.252235294117647,
      "grad_norm": 0.610975493853293,
      "learning_rate": 1.5884498469425078e-05,
      "loss": 0.3672,
      "step": 11161
    },
    {
      "epoch": 5.252705882352942,
      "grad_norm": 0.3997521954978246,
      "learning_rate": 1.5864871695826282e-05,
      "loss": 0.1494,
      "step": 11162
    },
    {
      "epoch": 5.253176470588235,
      "grad_norm": 0.640963619960294,
      "learning_rate": 1.5845256378025395e-05,
      "loss": 0.4629,
      "step": 11163
    },
    {
      "epoch": 5.253647058823529,
      "grad_norm": 0.38658045823797615,
      "learning_rate": 1.582565251769761e-05,
      "loss": 0.3691,
      "step": 11164
    },
    {
      "epoch": 5.254117647058823,
      "grad_norm": 0.47092988362903765,
      "learning_rate": 1.5806060116517227e-05,
      "loss": 0.4648,
      "step": 11165
    },
    {
      "epoch": 5.254588235294118,
      "grad_norm": 0.3077878029256483,
      "learning_rate": 1.578647917615757e-05,
      "loss": 0.3398,
      "step": 11166
    },
    {
      "epoch": 5.255058823529412,
      "grad_norm": 0.47880072192916456,
      "learning_rate": 1.5766909698290863e-05,
      "loss": 0.5195,
      "step": 11167
    },
    {
      "epoch": 5.255529411764706,
      "grad_norm": 0.21455025574003705,
      "learning_rate": 1.574735168458846e-05,
      "loss": 0.0618,
      "step": 11168
    },
    {
      "epoch": 5.256,
      "grad_norm": 0.3252372209099783,
      "learning_rate": 1.572780513672079e-05,
      "loss": 0.1367,
      "step": 11169
    },
    {
      "epoch": 5.2564705882352944,
      "grad_norm": 0.43565774973525273,
      "learning_rate": 1.5708270056357115e-05,
      "loss": 0.1367,
      "step": 11170
    },
    {
      "epoch": 5.256941176470589,
      "grad_norm": 0.3890436800192596,
      "learning_rate": 1.5688746445165867e-05,
      "loss": 0.25,
      "step": 11171
    },
    {
      "epoch": 5.257411764705882,
      "grad_norm": 0.17137836378963472,
      "learning_rate": 1.5669234304814515e-05,
      "loss": 0.0311,
      "step": 11172
    },
    {
      "epoch": 5.257882352941176,
      "grad_norm": 0.4249897800772692,
      "learning_rate": 1.564973363696939e-05,
      "loss": 0.3086,
      "step": 11173
    },
    {
      "epoch": 5.2583529411764705,
      "grad_norm": 0.47269024604222964,
      "learning_rate": 1.5630244443296013e-05,
      "loss": 0.1768,
      "step": 11174
    },
    {
      "epoch": 5.258823529411765,
      "grad_norm": 0.36213775288431566,
      "learning_rate": 1.5610766725458834e-05,
      "loss": 0.3574,
      "step": 11175
    },
    {
      "epoch": 5.259294117647059,
      "grad_norm": 0.37916019526658823,
      "learning_rate": 1.5591300485121354e-05,
      "loss": 0.0972,
      "step": 11176
    },
    {
      "epoch": 5.259764705882353,
      "grad_norm": 0.28861748845304813,
      "learning_rate": 1.5571845723946113e-05,
      "loss": 0.2812,
      "step": 11177
    },
    {
      "epoch": 5.260235294117647,
      "grad_norm": 0.713375541621404,
      "learning_rate": 1.555240244359458e-05,
      "loss": 0.2314,
      "step": 11178
    },
    {
      "epoch": 5.260705882352942,
      "grad_norm": 0.5086443172865155,
      "learning_rate": 1.5532970645727344e-05,
      "loss": 0.2891,
      "step": 11179
    },
    {
      "epoch": 5.261176470588235,
      "grad_norm": 0.4379697374997071,
      "learning_rate": 1.551355033200401e-05,
      "loss": 0.5586,
      "step": 11180
    },
    {
      "epoch": 5.261647058823529,
      "grad_norm": 0.48636523532506953,
      "learning_rate": 1.5494141504083087e-05,
      "loss": 0.1426,
      "step": 11181
    },
    {
      "epoch": 5.262117647058823,
      "grad_norm": 0.4610387744568557,
      "learning_rate": 1.547474416362226e-05,
      "loss": 0.4043,
      "step": 11182
    },
    {
      "epoch": 5.262588235294118,
      "grad_norm": 0.5576969704028184,
      "learning_rate": 1.5455358312278093e-05,
      "loss": 0.5078,
      "step": 11183
    },
    {
      "epoch": 5.263058823529412,
      "grad_norm": 0.5093505528793817,
      "learning_rate": 1.5435983951706276e-05,
      "loss": 0.5625,
      "step": 11184
    },
    {
      "epoch": 5.263529411764706,
      "grad_norm": 0.4890551767286767,
      "learning_rate": 1.5416621083561486e-05,
      "loss": 0.1729,
      "step": 11185
    },
    {
      "epoch": 5.264,
      "grad_norm": 0.3056595757116989,
      "learning_rate": 1.5397269709497352e-05,
      "loss": 0.1465,
      "step": 11186
    },
    {
      "epoch": 5.2644705882352945,
      "grad_norm": 0.6042811202832084,
      "learning_rate": 1.5377929831166597e-05,
      "loss": 0.5508,
      "step": 11187
    },
    {
      "epoch": 5.264941176470588,
      "grad_norm": 0.2782581801565407,
      "learning_rate": 1.5358601450220988e-05,
      "loss": 0.0918,
      "step": 11188
    },
    {
      "epoch": 5.265411764705882,
      "grad_norm": 0.4528333052529647,
      "learning_rate": 1.53392845683112e-05,
      "loss": 0.2021,
      "step": 11189
    },
    {
      "epoch": 5.265882352941176,
      "grad_norm": 0.2719964298311609,
      "learning_rate": 1.5319979187087018e-05,
      "loss": 0.0996,
      "step": 11190
    },
    {
      "epoch": 5.2663529411764705,
      "grad_norm": 0.2210212930967223,
      "learning_rate": 1.5300685308197254e-05,
      "loss": 0.0605,
      "step": 11191
    },
    {
      "epoch": 5.266823529411765,
      "grad_norm": 0.23391526597714618,
      "learning_rate": 1.528140293328962e-05,
      "loss": 0.0435,
      "step": 11192
    },
    {
      "epoch": 5.267294117647059,
      "grad_norm": 0.4308050155679827,
      "learning_rate": 1.5262132064010956e-05,
      "loss": 0.249,
      "step": 11193
    },
    {
      "epoch": 5.267764705882353,
      "grad_norm": 0.3593920856320375,
      "learning_rate": 1.5242872702007142e-05,
      "loss": 0.1455,
      "step": 11194
    },
    {
      "epoch": 5.268235294117647,
      "grad_norm": 0.36333527437425894,
      "learning_rate": 1.522362484892296e-05,
      "loss": 0.2451,
      "step": 11195
    },
    {
      "epoch": 5.268705882352942,
      "grad_norm": 0.5255127321322136,
      "learning_rate": 1.5204388506402281e-05,
      "loss": 0.4102,
      "step": 11196
    },
    {
      "epoch": 5.269176470588235,
      "grad_norm": 0.36822448830914556,
      "learning_rate": 1.5185163676088007e-05,
      "loss": 0.1533,
      "step": 11197
    },
    {
      "epoch": 5.269647058823529,
      "grad_norm": 0.41951440690582126,
      "learning_rate": 1.5165950359622009e-05,
      "loss": 0.3242,
      "step": 11198
    },
    {
      "epoch": 5.270117647058823,
      "grad_norm": 0.5027641924099714,
      "learning_rate": 1.5146748558645205e-05,
      "loss": 0.6094,
      "step": 11199
    },
    {
      "epoch": 5.270588235294118,
      "grad_norm": 0.42490513901988664,
      "learning_rate": 1.5127558274797535e-05,
      "loss": 0.291,
      "step": 11200
    },
    {
      "epoch": 5.271058823529412,
      "grad_norm": 0.4067479666808853,
      "learning_rate": 1.5108379509717922e-05,
      "loss": 0.1484,
      "step": 11201
    },
    {
      "epoch": 5.271529411764706,
      "grad_norm": 0.25604439617582786,
      "learning_rate": 1.5089212265044336e-05,
      "loss": 0.0586,
      "step": 11202
    },
    {
      "epoch": 5.272,
      "grad_norm": 0.39976766681156856,
      "learning_rate": 1.5070056542413788e-05,
      "loss": 0.1455,
      "step": 11203
    },
    {
      "epoch": 5.2724705882352945,
      "grad_norm": 0.4700387150673559,
      "learning_rate": 1.5050912343462202e-05,
      "loss": 0.4277,
      "step": 11204
    },
    {
      "epoch": 5.272941176470589,
      "grad_norm": 0.3819513525986992,
      "learning_rate": 1.5031779669824668e-05,
      "loss": 0.1582,
      "step": 11205
    },
    {
      "epoch": 5.273411764705882,
      "grad_norm": 0.4339598564641775,
      "learning_rate": 1.5012658523135113e-05,
      "loss": 0.1797,
      "step": 11206
    },
    {
      "epoch": 5.273882352941176,
      "grad_norm": 0.384337767088757,
      "learning_rate": 1.4993548905026648e-05,
      "loss": 0.3496,
      "step": 11207
    },
    {
      "epoch": 5.2743529411764705,
      "grad_norm": 0.2633198884668039,
      "learning_rate": 1.4974450817131334e-05,
      "loss": 0.0767,
      "step": 11208
    },
    {
      "epoch": 5.274823529411765,
      "grad_norm": 0.6408781206444333,
      "learning_rate": 1.4955364261080183e-05,
      "loss": 0.5195,
      "step": 11209
    },
    {
      "epoch": 5.275294117647059,
      "grad_norm": 0.5945261234386703,
      "learning_rate": 1.4936289238503324e-05,
      "loss": 0.2236,
      "step": 11210
    },
    {
      "epoch": 5.275764705882353,
      "grad_norm": 0.25894594453005315,
      "learning_rate": 1.491722575102987e-05,
      "loss": 0.1504,
      "step": 11211
    },
    {
      "epoch": 5.276235294117647,
      "grad_norm": 0.47721569065413033,
      "learning_rate": 1.4898173800287888e-05,
      "loss": 0.2061,
      "step": 11212
    },
    {
      "epoch": 5.276705882352942,
      "grad_norm": 0.3851713784240159,
      "learning_rate": 1.4879133387904523e-05,
      "loss": 0.2285,
      "step": 11213
    },
    {
      "epoch": 5.277176470588235,
      "grad_norm": 0.3197315750320874,
      "learning_rate": 1.4860104515505977e-05,
      "loss": 0.165,
      "step": 11214
    },
    {
      "epoch": 5.277647058823529,
      "grad_norm": 0.2942898582834124,
      "learning_rate": 1.4841087184717315e-05,
      "loss": 0.0786,
      "step": 11215
    },
    {
      "epoch": 5.278117647058823,
      "grad_norm": 0.526689437312365,
      "learning_rate": 1.4822081397162771e-05,
      "loss": 0.5859,
      "step": 11216
    },
    {
      "epoch": 5.278588235294118,
      "grad_norm": 0.48557904531938756,
      "learning_rate": 1.4803087154465548e-05,
      "loss": 0.4766,
      "step": 11217
    },
    {
      "epoch": 5.279058823529412,
      "grad_norm": 0.44268801705528643,
      "learning_rate": 1.4784104458247764e-05,
      "loss": 0.1816,
      "step": 11218
    },
    {
      "epoch": 5.279529411764706,
      "grad_norm": 0.29245499464864283,
      "learning_rate": 1.476513331013074e-05,
      "loss": 0.2021,
      "step": 11219
    },
    {
      "epoch": 5.28,
      "grad_norm": 0.4779638382638363,
      "learning_rate": 1.4746173711734627e-05,
      "loss": 0.25,
      "step": 11220
    },
    {
      "epoch": 5.2804705882352945,
      "grad_norm": 0.33925369884382783,
      "learning_rate": 1.4727225664678699e-05,
      "loss": 0.1309,
      "step": 11221
    },
    {
      "epoch": 5.280941176470588,
      "grad_norm": 0.32713427521489763,
      "learning_rate": 1.4708289170581227e-05,
      "loss": 0.1328,
      "step": 11222
    },
    {
      "epoch": 5.281411764705882,
      "grad_norm": 0.38222381568253383,
      "learning_rate": 1.4689364231059436e-05,
      "loss": 0.3066,
      "step": 11223
    },
    {
      "epoch": 5.281882352941176,
      "grad_norm": 0.4236419828494949,
      "learning_rate": 1.4670450847729648e-05,
      "loss": 0.2178,
      "step": 11224
    },
    {
      "epoch": 5.2823529411764705,
      "grad_norm": 0.15249082157811683,
      "learning_rate": 1.465154902220717e-05,
      "loss": 0.0195,
      "step": 11225
    },
    {
      "epoch": 5.282823529411765,
      "grad_norm": 0.23599335547337807,
      "learning_rate": 1.4632658756106247e-05,
      "loss": 0.0586,
      "step": 11226
    },
    {
      "epoch": 5.283294117647059,
      "grad_norm": 0.355600195964496,
      "learning_rate": 1.4613780051040236e-05,
      "loss": 0.126,
      "step": 11227
    },
    {
      "epoch": 5.283764705882353,
      "grad_norm": 0.4956382561050533,
      "learning_rate": 1.4594912908621515e-05,
      "loss": 0.2812,
      "step": 11228
    },
    {
      "epoch": 5.284235294117647,
      "grad_norm": 0.4949629099775748,
      "learning_rate": 1.4576057330461361e-05,
      "loss": 0.2383,
      "step": 11229
    },
    {
      "epoch": 5.284705882352942,
      "grad_norm": 0.4437237961938785,
      "learning_rate": 1.4557213318170152e-05,
      "loss": 0.2695,
      "step": 11230
    },
    {
      "epoch": 5.285176470588235,
      "grad_norm": 0.35345249001585294,
      "learning_rate": 1.453838087335732e-05,
      "loss": 0.3574,
      "step": 11231
    },
    {
      "epoch": 5.285647058823529,
      "grad_norm": 0.18591163923207846,
      "learning_rate": 1.4519559997631142e-05,
      "loss": 0.0938,
      "step": 11232
    },
    {
      "epoch": 5.286117647058823,
      "grad_norm": 0.43419555031734064,
      "learning_rate": 1.45007506925991e-05,
      "loss": 0.4668,
      "step": 11233
    },
    {
      "epoch": 5.286588235294118,
      "grad_norm": 0.4346382327621478,
      "learning_rate": 1.448195295986756e-05,
      "loss": 0.2256,
      "step": 11234
    },
    {
      "epoch": 5.287058823529412,
      "grad_norm": 0.3790721850744393,
      "learning_rate": 1.4463166801041936e-05,
      "loss": 0.1084,
      "step": 11235
    },
    {
      "epoch": 5.287529411764706,
      "grad_norm": 0.17932602641723003,
      "learning_rate": 1.4444392217726697e-05,
      "loss": 0.0591,
      "step": 11236
    },
    {
      "epoch": 5.288,
      "grad_norm": 0.48159076348521,
      "learning_rate": 1.442562921152524e-05,
      "loss": 0.4395,
      "step": 11237
    },
    {
      "epoch": 5.2884705882352945,
      "grad_norm": 0.3442618909098726,
      "learning_rate": 1.4406877784040021e-05,
      "loss": 0.373,
      "step": 11238
    },
    {
      "epoch": 5.288941176470589,
      "grad_norm": 0.36504660465880717,
      "learning_rate": 1.4388137936872556e-05,
      "loss": 0.1416,
      "step": 11239
    },
    {
      "epoch": 5.289411764705882,
      "grad_norm": 0.4328732023651715,
      "learning_rate": 1.4369409671623267e-05,
      "loss": 0.4922,
      "step": 11240
    },
    {
      "epoch": 5.289882352941176,
      "grad_norm": 0.23468323173712669,
      "learning_rate": 1.4350692989891638e-05,
      "loss": 0.0815,
      "step": 11241
    },
    {
      "epoch": 5.2903529411764705,
      "grad_norm": 0.5208180462120098,
      "learning_rate": 1.4331987893276225e-05,
      "loss": 0.4102,
      "step": 11242
    },
    {
      "epoch": 5.290823529411765,
      "grad_norm": 0.36539892866155327,
      "learning_rate": 1.4313294383374468e-05,
      "loss": 0.0786,
      "step": 11243
    },
    {
      "epoch": 5.291294117647059,
      "grad_norm": 0.323565357572167,
      "learning_rate": 1.429461246178289e-05,
      "loss": 0.1172,
      "step": 11244
    },
    {
      "epoch": 5.291764705882353,
      "grad_norm": 0.48325348752977737,
      "learning_rate": 1.4275942130097096e-05,
      "loss": 0.3516,
      "step": 11245
    },
    {
      "epoch": 5.292235294117647,
      "grad_norm": 0.5940324441111505,
      "learning_rate": 1.4257283389911512e-05,
      "loss": 0.4219,
      "step": 11246
    },
    {
      "epoch": 5.292705882352941,
      "grad_norm": 0.2802413021187412,
      "learning_rate": 1.4238636242819761e-05,
      "loss": 0.1572,
      "step": 11247
    },
    {
      "epoch": 5.293176470588235,
      "grad_norm": 0.5178100692383333,
      "learning_rate": 1.422000069041439e-05,
      "loss": 0.3984,
      "step": 11248
    },
    {
      "epoch": 5.293647058823529,
      "grad_norm": 0.3068304814499946,
      "learning_rate": 1.420137673428694e-05,
      "loss": 0.2432,
      "step": 11249
    },
    {
      "epoch": 5.294117647058823,
      "grad_norm": 0.20817544884694766,
      "learning_rate": 1.4182764376028006e-05,
      "loss": 0.0396,
      "step": 11250
    },
    {
      "epoch": 5.294588235294118,
      "grad_norm": 0.520896583480689,
      "learning_rate": 1.4164163617227197e-05,
      "loss": 0.3652,
      "step": 11251
    },
    {
      "epoch": 5.295058823529412,
      "grad_norm": 0.3271377459604272,
      "learning_rate": 1.4145574459473064e-05,
      "loss": 0.0908,
      "step": 11252
    },
    {
      "epoch": 5.295529411764706,
      "grad_norm": 0.3230966711274056,
      "learning_rate": 1.4126996904353233e-05,
      "loss": 0.2285,
      "step": 11253
    },
    {
      "epoch": 5.296,
      "grad_norm": 0.4746826117602761,
      "learning_rate": 1.4108430953454352e-05,
      "loss": 0.4434,
      "step": 11254
    },
    {
      "epoch": 5.2964705882352945,
      "grad_norm": 0.33182731636991736,
      "learning_rate": 1.4089876608361971e-05,
      "loss": 0.2207,
      "step": 11255
    },
    {
      "epoch": 5.296941176470588,
      "grad_norm": 0.24942867692249593,
      "learning_rate": 1.4071333870660807e-05,
      "loss": 0.0369,
      "step": 11256
    },
    {
      "epoch": 5.297411764705882,
      "grad_norm": 0.2624555822702265,
      "learning_rate": 1.4052802741934404e-05,
      "loss": 0.0688,
      "step": 11257
    },
    {
      "epoch": 5.297882352941176,
      "grad_norm": 0.2833270268283933,
      "learning_rate": 1.4034283223765485e-05,
      "loss": 0.1885,
      "step": 11258
    },
    {
      "epoch": 5.2983529411764705,
      "grad_norm": 0.41160642568083466,
      "learning_rate": 1.4015775317735717e-05,
      "loss": 0.2168,
      "step": 11259
    },
    {
      "epoch": 5.298823529411765,
      "grad_norm": 0.6600156699271906,
      "learning_rate": 1.3997279025425701e-05,
      "loss": 0.5664,
      "step": 11260
    },
    {
      "epoch": 5.299294117647059,
      "grad_norm": 0.3463865065576745,
      "learning_rate": 1.397879434841514e-05,
      "loss": 0.3477,
      "step": 11261
    },
    {
      "epoch": 5.299764705882353,
      "grad_norm": 0.40387653724469164,
      "learning_rate": 1.3960321288282722e-05,
      "loss": 0.3203,
      "step": 11262
    },
    {
      "epoch": 5.300235294117647,
      "grad_norm": 0.27452520529028474,
      "learning_rate": 1.3941859846606152e-05,
      "loss": 0.1836,
      "step": 11263
    },
    {
      "epoch": 5.300705882352942,
      "grad_norm": 0.6472010921335013,
      "learning_rate": 1.3923410024962117e-05,
      "loss": 0.4531,
      "step": 11264
    },
    {
      "epoch": 5.301176470588235,
      "grad_norm": 0.6474006738071282,
      "learning_rate": 1.390497182492631e-05,
      "loss": 0.3242,
      "step": 11265
    },
    {
      "epoch": 5.301647058823529,
      "grad_norm": 0.2189384432645243,
      "learning_rate": 1.388654524807345e-05,
      "loss": 0.0366,
      "step": 11266
    },
    {
      "epoch": 5.302117647058823,
      "grad_norm": 0.35406228490384334,
      "learning_rate": 1.3868130295977282e-05,
      "loss": 0.3848,
      "step": 11267
    },
    {
      "epoch": 5.302588235294118,
      "grad_norm": 0.44194333588806983,
      "learning_rate": 1.3849726970210479e-05,
      "loss": 0.4062,
      "step": 11268
    },
    {
      "epoch": 5.303058823529412,
      "grad_norm": 0.18659594702909169,
      "learning_rate": 1.3831335272344801e-05,
      "loss": 0.0483,
      "step": 11269
    },
    {
      "epoch": 5.303529411764706,
      "grad_norm": 0.2564900467684339,
      "learning_rate": 1.3812955203951026e-05,
      "loss": 0.0596,
      "step": 11270
    },
    {
      "epoch": 5.304,
      "grad_norm": 0.5002464647164683,
      "learning_rate": 1.3794586766598848e-05,
      "loss": 0.3438,
      "step": 11271
    },
    {
      "epoch": 5.3044705882352945,
      "grad_norm": 0.7609476076061562,
      "learning_rate": 1.3776229961857043e-05,
      "loss": 0.4922,
      "step": 11272
    },
    {
      "epoch": 5.304941176470588,
      "grad_norm": 0.34521023122216515,
      "learning_rate": 1.3757884791293394e-05,
      "loss": 0.1328,
      "step": 11273
    },
    {
      "epoch": 5.305411764705882,
      "grad_norm": 0.44519982665704266,
      "learning_rate": 1.373955125647463e-05,
      "loss": 0.2363,
      "step": 11274
    },
    {
      "epoch": 5.305882352941176,
      "grad_norm": 0.5186388183448215,
      "learning_rate": 1.3721229358966545e-05,
      "loss": 0.4844,
      "step": 11275
    },
    {
      "epoch": 5.3063529411764705,
      "grad_norm": 0.5200257394143079,
      "learning_rate": 1.3702919100333925e-05,
      "loss": 0.3535,
      "step": 11276
    },
    {
      "epoch": 5.306823529411765,
      "grad_norm": 0.2787134859895664,
      "learning_rate": 1.3684620482140534e-05,
      "loss": 0.0698,
      "step": 11277
    },
    {
      "epoch": 5.307294117647059,
      "grad_norm": 0.41109292555867494,
      "learning_rate": 1.3666333505949173e-05,
      "loss": 0.5508,
      "step": 11278
    },
    {
      "epoch": 5.307764705882353,
      "grad_norm": 0.3577395188581212,
      "learning_rate": 1.3648058173321675e-05,
      "loss": 0.168,
      "step": 11279
    },
    {
      "epoch": 5.308235294117647,
      "grad_norm": 0.3655220484602116,
      "learning_rate": 1.3629794485818791e-05,
      "loss": 0.2217,
      "step": 11280
    },
    {
      "epoch": 5.308705882352941,
      "grad_norm": 0.3735573934714318,
      "learning_rate": 1.361154244500034e-05,
      "loss": 0.3594,
      "step": 11281
    },
    {
      "epoch": 5.309176470588235,
      "grad_norm": 0.34882564797382176,
      "learning_rate": 1.3593302052425159e-05,
      "loss": 0.1777,
      "step": 11282
    },
    {
      "epoch": 5.309647058823529,
      "grad_norm": 0.33737135613684466,
      "learning_rate": 1.3575073309651052e-05,
      "loss": 0.1084,
      "step": 11283
    },
    {
      "epoch": 5.310117647058823,
      "grad_norm": 0.35382152919861987,
      "learning_rate": 1.3556856218234858e-05,
      "loss": 0.2539,
      "step": 11284
    },
    {
      "epoch": 5.310588235294118,
      "grad_norm": 0.411267171207031,
      "learning_rate": 1.3538650779732362e-05,
      "loss": 0.1846,
      "step": 11285
    },
    {
      "epoch": 5.311058823529412,
      "grad_norm": 0.48610858902345083,
      "learning_rate": 1.3520456995698443e-05,
      "loss": 0.2656,
      "step": 11286
    },
    {
      "epoch": 5.311529411764706,
      "grad_norm": 0.4127828002797159,
      "learning_rate": 1.3502274867686936e-05,
      "loss": 0.4531,
      "step": 11287
    },
    {
      "epoch": 5.312,
      "grad_norm": 0.4818722075295411,
      "learning_rate": 1.348410439725065e-05,
      "loss": 0.3887,
      "step": 11288
    },
    {
      "epoch": 5.3124705882352945,
      "grad_norm": 0.5144254225852809,
      "learning_rate": 1.3465945585941462e-05,
      "loss": 0.5234,
      "step": 11289
    },
    {
      "epoch": 5.312941176470588,
      "grad_norm": 0.43256387344765984,
      "learning_rate": 1.3447798435310247e-05,
      "loss": 0.4512,
      "step": 11290
    },
    {
      "epoch": 5.313411764705882,
      "grad_norm": 0.3763571957074974,
      "learning_rate": 1.3429662946906799e-05,
      "loss": 0.1167,
      "step": 11291
    },
    {
      "epoch": 5.313882352941176,
      "grad_norm": 0.20430239407579193,
      "learning_rate": 1.3411539122279996e-05,
      "loss": 0.0325,
      "step": 11292
    },
    {
      "epoch": 5.3143529411764705,
      "grad_norm": 0.35629994333966764,
      "learning_rate": 1.3393426962977765e-05,
      "loss": 0.3887,
      "step": 11293
    },
    {
      "epoch": 5.314823529411765,
      "grad_norm": 0.25480951498101045,
      "learning_rate": 1.3375326470546872e-05,
      "loss": 0.1138,
      "step": 11294
    },
    {
      "epoch": 5.315294117647059,
      "grad_norm": 0.40751048044513627,
      "learning_rate": 1.3357237646533258e-05,
      "loss": 0.2891,
      "step": 11295
    },
    {
      "epoch": 5.315764705882353,
      "grad_norm": 0.27331470471737607,
      "learning_rate": 1.3339160492481793e-05,
      "loss": 0.0801,
      "step": 11296
    },
    {
      "epoch": 5.316235294117647,
      "grad_norm": 0.5225758733846134,
      "learning_rate": 1.3321095009936321e-05,
      "loss": 0.4629,
      "step": 11297
    },
    {
      "epoch": 5.316705882352942,
      "grad_norm": 1.0812092868646712,
      "learning_rate": 1.3303041200439729e-05,
      "loss": 0.418,
      "step": 11298
    },
    {
      "epoch": 5.317176470588235,
      "grad_norm": 0.4924044106563114,
      "learning_rate": 1.3284999065533946e-05,
      "loss": 0.3477,
      "step": 11299
    },
    {
      "epoch": 5.317647058823529,
      "grad_norm": 0.32842937167517994,
      "learning_rate": 1.326696860675981e-05,
      "loss": 0.0874,
      "step": 11300
    },
    {
      "epoch": 5.318117647058823,
      "grad_norm": 0.2835509183675477,
      "learning_rate": 1.3248949825657217e-05,
      "loss": 0.1147,
      "step": 11301
    },
    {
      "epoch": 5.318588235294118,
      "grad_norm": 0.5405566066188033,
      "learning_rate": 1.3230942723765092e-05,
      "loss": 0.2715,
      "step": 11302
    },
    {
      "epoch": 5.319058823529412,
      "grad_norm": 0.4005042592050412,
      "learning_rate": 1.3212947302621284e-05,
      "loss": 0.2695,
      "step": 11303
    },
    {
      "epoch": 5.319529411764706,
      "grad_norm": 0.19905244797751995,
      "learning_rate": 1.3194963563762718e-05,
      "loss": 0.03,
      "step": 11304
    },
    {
      "epoch": 5.32,
      "grad_norm": 0.43563418764915834,
      "learning_rate": 1.317699150872531e-05,
      "loss": 0.2676,
      "step": 11305
    },
    {
      "epoch": 5.3204705882352945,
      "grad_norm": 0.2108544965338929,
      "learning_rate": 1.3159031139043918e-05,
      "loss": 0.1523,
      "step": 11306
    },
    {
      "epoch": 5.320941176470588,
      "grad_norm": 0.29142322645355734,
      "learning_rate": 1.3141082456252467e-05,
      "loss": 0.2969,
      "step": 11307
    },
    {
      "epoch": 5.321411764705882,
      "grad_norm": 0.36861207576812344,
      "learning_rate": 1.312314546188386e-05,
      "loss": 0.3652,
      "step": 11308
    },
    {
      "epoch": 5.321882352941176,
      "grad_norm": 0.17077675428421898,
      "learning_rate": 1.3105220157470009e-05,
      "loss": 0.0317,
      "step": 11309
    },
    {
      "epoch": 5.3223529411764705,
      "grad_norm": 0.6835398150728804,
      "learning_rate": 1.3087306544541836e-05,
      "loss": 0.3125,
      "step": 11310
    },
    {
      "epoch": 5.322823529411765,
      "grad_norm": 0.28038582297889014,
      "learning_rate": 1.306940462462922e-05,
      "loss": 0.2217,
      "step": 11311
    },
    {
      "epoch": 5.323294117647059,
      "grad_norm": 0.4101003148502154,
      "learning_rate": 1.3051514399261104e-05,
      "loss": 0.293,
      "step": 11312
    },
    {
      "epoch": 5.323764705882353,
      "grad_norm": 0.6234125864972575,
      "learning_rate": 1.3033635869965398e-05,
      "loss": 0.5586,
      "step": 11313
    },
    {
      "epoch": 5.324235294117647,
      "grad_norm": 0.3260898390230209,
      "learning_rate": 1.3015769038268981e-05,
      "loss": 0.1963,
      "step": 11314
    },
    {
      "epoch": 5.324705882352941,
      "grad_norm": 0.5281667819036971,
      "learning_rate": 1.2997913905697816e-05,
      "loss": 0.5703,
      "step": 11315
    },
    {
      "epoch": 5.325176470588235,
      "grad_norm": 0.40831939379628607,
      "learning_rate": 1.2980070473776782e-05,
      "loss": 0.2891,
      "step": 11316
    },
    {
      "epoch": 5.325647058823529,
      "grad_norm": 0.16112588842017178,
      "learning_rate": 1.2962238744029795e-05,
      "loss": 0.0237,
      "step": 11317
    },
    {
      "epoch": 5.326117647058823,
      "grad_norm": 0.19671131373639117,
      "learning_rate": 1.2944418717979804e-05,
      "loss": 0.0276,
      "step": 11318
    },
    {
      "epoch": 5.326588235294118,
      "grad_norm": 0.5236456628517996,
      "learning_rate": 1.2926610397148707e-05,
      "loss": 0.4512,
      "step": 11319
    },
    {
      "epoch": 5.327058823529412,
      "grad_norm": 0.5995891184664324,
      "learning_rate": 1.2908813783057404e-05,
      "loss": 0.4707,
      "step": 11320
    },
    {
      "epoch": 5.327529411764706,
      "grad_norm": 0.27144567409999376,
      "learning_rate": 1.2891028877225863e-05,
      "loss": 0.0635,
      "step": 11321
    },
    {
      "epoch": 5.328,
      "grad_norm": 0.165199231454487,
      "learning_rate": 1.287325568117295e-05,
      "loss": 0.0203,
      "step": 11322
    },
    {
      "epoch": 5.3284705882352945,
      "grad_norm": 0.44426465539347515,
      "learning_rate": 1.2855494196416604e-05,
      "loss": 0.2949,
      "step": 11323
    },
    {
      "epoch": 5.328941176470588,
      "grad_norm": 0.30183775611275016,
      "learning_rate": 1.2837744424473761e-05,
      "loss": 0.1982,
      "step": 11324
    },
    {
      "epoch": 5.329411764705882,
      "grad_norm": 0.38029850719405345,
      "learning_rate": 1.2820006366860292e-05,
      "loss": 0.3027,
      "step": 11325
    },
    {
      "epoch": 5.329882352941176,
      "grad_norm": 0.819297774315989,
      "learning_rate": 1.2802280025091165e-05,
      "loss": 0.3008,
      "step": 11326
    },
    {
      "epoch": 5.3303529411764705,
      "grad_norm": 0.5461167454287917,
      "learning_rate": 1.2784565400680291e-05,
      "loss": 0.4121,
      "step": 11327
    },
    {
      "epoch": 5.330823529411765,
      "grad_norm": 0.29691737001033447,
      "learning_rate": 1.276686249514054e-05,
      "loss": 0.1709,
      "step": 11328
    },
    {
      "epoch": 5.331294117647059,
      "grad_norm": 0.1737083154311289,
      "learning_rate": 1.2749171309983852e-05,
      "loss": 0.0537,
      "step": 11329
    },
    {
      "epoch": 5.331764705882353,
      "grad_norm": 0.305714294503241,
      "learning_rate": 1.273149184672117e-05,
      "loss": 0.1514,
      "step": 11330
    },
    {
      "epoch": 5.332235294117647,
      "grad_norm": 0.1317231005627306,
      "learning_rate": 1.2713824106862369e-05,
      "loss": 0.0114,
      "step": 11331
    },
    {
      "epoch": 5.332705882352942,
      "grad_norm": 0.36219119818231044,
      "learning_rate": 1.2696168091916376e-05,
      "loss": 0.1079,
      "step": 11332
    },
    {
      "epoch": 5.333176470588235,
      "grad_norm": 0.43712728165530673,
      "learning_rate": 1.2678523803391116e-05,
      "loss": 0.5586,
      "step": 11333
    },
    {
      "epoch": 5.333647058823529,
      "grad_norm": 0.2055119127267538,
      "learning_rate": 1.2660891242793453e-05,
      "loss": 0.0396,
      "step": 11334
    },
    {
      "epoch": 5.334117647058823,
      "grad_norm": 0.2916670955342656,
      "learning_rate": 1.2643270411629346e-05,
      "loss": 0.0762,
      "step": 11335
    },
    {
      "epoch": 5.334588235294118,
      "grad_norm": 0.36744416294971455,
      "learning_rate": 1.2625661311403662e-05,
      "loss": 0.4668,
      "step": 11336
    },
    {
      "epoch": 5.335058823529412,
      "grad_norm": 0.4258656522555543,
      "learning_rate": 1.2608063943620312e-05,
      "loss": 0.373,
      "step": 11337
    },
    {
      "epoch": 5.335529411764706,
      "grad_norm": 0.3512611926528403,
      "learning_rate": 1.2590478309782243e-05,
      "loss": 0.1514,
      "step": 11338
    },
    {
      "epoch": 5.336,
      "grad_norm": 0.33329120703507426,
      "learning_rate": 1.2572904411391272e-05,
      "loss": 0.3945,
      "step": 11339
    },
    {
      "epoch": 5.3364705882352945,
      "grad_norm": 0.5305996403627774,
      "learning_rate": 1.2555342249948347e-05,
      "loss": 0.543,
      "step": 11340
    },
    {
      "epoch": 5.336941176470588,
      "grad_norm": 0.36068467614435645,
      "learning_rate": 1.2537791826953387e-05,
      "loss": 0.4863,
      "step": 11341
    },
    {
      "epoch": 5.337411764705882,
      "grad_norm": 0.5991304524472152,
      "learning_rate": 1.2520253143905223e-05,
      "loss": 0.2715,
      "step": 11342
    },
    {
      "epoch": 5.337882352941176,
      "grad_norm": 0.33687506985519194,
      "learning_rate": 1.2502726202301777e-05,
      "loss": 0.2676,
      "step": 11343
    },
    {
      "epoch": 5.3383529411764705,
      "grad_norm": 0.27263014272415215,
      "learning_rate": 1.2485211003639967e-05,
      "loss": 0.0811,
      "step": 11344
    },
    {
      "epoch": 5.338823529411765,
      "grad_norm": 0.3505890779156656,
      "learning_rate": 1.2467707549415612e-05,
      "loss": 0.3281,
      "step": 11345
    },
    {
      "epoch": 5.339294117647059,
      "grad_norm": 0.36157751367658714,
      "learning_rate": 1.2450215841123618e-05,
      "loss": 0.3086,
      "step": 11346
    },
    {
      "epoch": 5.339764705882353,
      "grad_norm": 0.38973185634667984,
      "learning_rate": 1.2432735880257904e-05,
      "loss": 0.1206,
      "step": 11347
    },
    {
      "epoch": 5.340235294117647,
      "grad_norm": 0.2784047322726768,
      "learning_rate": 1.2415267668311263e-05,
      "loss": 0.1416,
      "step": 11348
    },
    {
      "epoch": 5.340705882352941,
      "grad_norm": 0.3141650515520856,
      "learning_rate": 1.2397811206775632e-05,
      "loss": 0.2051,
      "step": 11349
    },
    {
      "epoch": 5.341176470588235,
      "grad_norm": 0.10733872150468973,
      "learning_rate": 1.2380366497141886e-05,
      "loss": 0.0109,
      "step": 11350
    },
    {
      "epoch": 5.341647058823529,
      "grad_norm": 0.3053727777440613,
      "learning_rate": 1.23629335408998e-05,
      "loss": 0.0938,
      "step": 11351
    },
    {
      "epoch": 5.342117647058823,
      "grad_norm": 0.6490920571025125,
      "learning_rate": 1.234551233953835e-05,
      "loss": 0.582,
      "step": 11352
    },
    {
      "epoch": 5.342588235294118,
      "grad_norm": 0.32493258227550925,
      "learning_rate": 1.232810289454531e-05,
      "loss": 0.0603,
      "step": 11353
    },
    {
      "epoch": 5.343058823529412,
      "grad_norm": 0.2769664717254873,
      "learning_rate": 1.2310705207407561e-05,
      "loss": 0.0894,
      "step": 11354
    },
    {
      "epoch": 5.343529411764706,
      "grad_norm": 0.4451599003265124,
      "learning_rate": 1.229331927961098e-05,
      "loss": 0.3867,
      "step": 11355
    },
    {
      "epoch": 5.344,
      "grad_norm": 0.3097739695938577,
      "learning_rate": 1.2275945112640363e-05,
      "loss": 0.1182,
      "step": 11356
    },
    {
      "epoch": 5.3444705882352945,
      "grad_norm": 0.3824164424653783,
      "learning_rate": 1.2258582707979554e-05,
      "loss": 0.3926,
      "step": 11357
    },
    {
      "epoch": 5.344941176470588,
      "grad_norm": 0.5141127688643485,
      "learning_rate": 1.2241232067111418e-05,
      "loss": 0.3086,
      "step": 11358
    },
    {
      "epoch": 5.345411764705882,
      "grad_norm": 0.36795607539684566,
      "learning_rate": 1.2223893191517753e-05,
      "loss": 0.1211,
      "step": 11359
    },
    {
      "epoch": 5.345882352941176,
      "grad_norm": 0.47767514493490415,
      "learning_rate": 1.2206566082679392e-05,
      "loss": 0.2441,
      "step": 11360
    },
    {
      "epoch": 5.3463529411764705,
      "grad_norm": 0.5208626901525342,
      "learning_rate": 1.2189250742076184e-05,
      "loss": 0.1777,
      "step": 11361
    },
    {
      "epoch": 5.346823529411765,
      "grad_norm": 0.36814757194875714,
      "learning_rate": 1.2171947171186897e-05,
      "loss": 0.4277,
      "step": 11362
    },
    {
      "epoch": 5.347294117647059,
      "grad_norm": 0.41098082474859415,
      "learning_rate": 1.2154655371489397e-05,
      "loss": 0.3008,
      "step": 11363
    },
    {
      "epoch": 5.347764705882353,
      "grad_norm": 0.404891242665525,
      "learning_rate": 1.213737534446042e-05,
      "loss": 0.3262,
      "step": 11364
    },
    {
      "epoch": 5.348235294117647,
      "grad_norm": 0.36339377991584093,
      "learning_rate": 1.2120107091575803e-05,
      "loss": 0.4004,
      "step": 11365
    },
    {
      "epoch": 5.348705882352942,
      "grad_norm": 0.5520338283705891,
      "learning_rate": 1.2102850614310366e-05,
      "loss": 0.2734,
      "step": 11366
    },
    {
      "epoch": 5.349176470588235,
      "grad_norm": 0.2760044050520635,
      "learning_rate": 1.2085605914137847e-05,
      "loss": 0.0962,
      "step": 11367
    },
    {
      "epoch": 5.349647058823529,
      "grad_norm": 0.49741118909644294,
      "learning_rate": 1.206837299253105e-05,
      "loss": 0.3457,
      "step": 11368
    },
    {
      "epoch": 5.350117647058823,
      "grad_norm": 0.4606704095776115,
      "learning_rate": 1.2051151850961766e-05,
      "loss": 0.3594,
      "step": 11369
    },
    {
      "epoch": 5.350588235294118,
      "grad_norm": 0.2886850319599355,
      "learning_rate": 1.2033942490900716e-05,
      "loss": 0.1211,
      "step": 11370
    },
    {
      "epoch": 5.351058823529412,
      "grad_norm": 0.47532281303607876,
      "learning_rate": 1.201674491381771e-05,
      "loss": 0.7383,
      "step": 11371
    },
    {
      "epoch": 5.351529411764706,
      "grad_norm": 0.3557582517875867,
      "learning_rate": 1.1999559121181523e-05,
      "loss": 0.3359,
      "step": 11372
    },
    {
      "epoch": 5.352,
      "grad_norm": 0.44171325844618575,
      "learning_rate": 1.1982385114459846e-05,
      "loss": 0.2773,
      "step": 11373
    },
    {
      "epoch": 5.352470588235294,
      "grad_norm": 0.4784227908822268,
      "learning_rate": 1.1965222895119442e-05,
      "loss": 0.3086,
      "step": 11374
    },
    {
      "epoch": 5.352941176470588,
      "grad_norm": 0.22220376600337144,
      "learning_rate": 1.1948072464626101e-05,
      "loss": 0.0398,
      "step": 11375
    },
    {
      "epoch": 5.353411764705882,
      "grad_norm": 0.6032375061956962,
      "learning_rate": 1.1930933824444472e-05,
      "loss": 0.3574,
      "step": 11376
    },
    {
      "epoch": 5.353882352941176,
      "grad_norm": 0.17623789483415575,
      "learning_rate": 1.1913806976038331e-05,
      "loss": 0.0586,
      "step": 11377
    },
    {
      "epoch": 5.3543529411764705,
      "grad_norm": 0.590321627112256,
      "learning_rate": 1.1896691920870394e-05,
      "loss": 0.2354,
      "step": 11378
    },
    {
      "epoch": 5.354823529411765,
      "grad_norm": 0.510116557637022,
      "learning_rate": 1.1879588660402356e-05,
      "loss": 0.5273,
      "step": 11379
    },
    {
      "epoch": 5.355294117647059,
      "grad_norm": 0.4213268158900576,
      "learning_rate": 1.18624971960949e-05,
      "loss": 0.375,
      "step": 11380
    },
    {
      "epoch": 5.355764705882353,
      "grad_norm": 0.2538099586962422,
      "learning_rate": 1.1845417529407791e-05,
      "loss": 0.1436,
      "step": 11381
    },
    {
      "epoch": 5.356235294117647,
      "grad_norm": 0.35002201693870516,
      "learning_rate": 1.1828349661799629e-05,
      "loss": 0.2891,
      "step": 11382
    },
    {
      "epoch": 5.356705882352941,
      "grad_norm": 0.477270067252586,
      "learning_rate": 1.1811293594728133e-05,
      "loss": 0.416,
      "step": 11383
    },
    {
      "epoch": 5.357176470588235,
      "grad_norm": 0.2300462900426497,
      "learning_rate": 1.179424932965002e-05,
      "loss": 0.0598,
      "step": 11384
    },
    {
      "epoch": 5.357647058823529,
      "grad_norm": 0.35049500883509077,
      "learning_rate": 1.1777216868020872e-05,
      "loss": 0.2314,
      "step": 11385
    },
    {
      "epoch": 5.358117647058823,
      "grad_norm": 0.19107865625640208,
      "learning_rate": 1.1760196211295398e-05,
      "loss": 0.0569,
      "step": 11386
    },
    {
      "epoch": 5.358588235294118,
      "grad_norm": 0.4694363551249489,
      "learning_rate": 1.1743187360927214e-05,
      "loss": 0.2539,
      "step": 11387
    },
    {
      "epoch": 5.359058823529412,
      "grad_norm": 0.5320632536619218,
      "learning_rate": 1.1726190318368977e-05,
      "loss": 0.2139,
      "step": 11388
    },
    {
      "epoch": 5.359529411764706,
      "grad_norm": 0.4253464854121676,
      "learning_rate": 1.1709205085072344e-05,
      "loss": 0.2598,
      "step": 11389
    },
    {
      "epoch": 5.36,
      "grad_norm": 0.26702822040715757,
      "learning_rate": 1.1692231662487866e-05,
      "loss": 0.0679,
      "step": 11390
    },
    {
      "epoch": 5.3604705882352945,
      "grad_norm": 0.48449297248789447,
      "learning_rate": 1.1675270052065222e-05,
      "loss": 0.2539,
      "step": 11391
    },
    {
      "epoch": 5.360941176470588,
      "grad_norm": 0.24000869679975195,
      "learning_rate": 1.1658320255253e-05,
      "loss": 0.0359,
      "step": 11392
    },
    {
      "epoch": 5.361411764705882,
      "grad_norm": 0.26643338638246816,
      "learning_rate": 1.1641382273498777e-05,
      "loss": 0.0674,
      "step": 11393
    },
    {
      "epoch": 5.361882352941176,
      "grad_norm": 0.5384228925220217,
      "learning_rate": 1.1624456108249164e-05,
      "loss": 0.4492,
      "step": 11394
    },
    {
      "epoch": 5.3623529411764705,
      "grad_norm": 0.2728776916789231,
      "learning_rate": 1.1607541760949735e-05,
      "loss": 0.1094,
      "step": 11395
    },
    {
      "epoch": 5.362823529411765,
      "grad_norm": 0.37058982324879275,
      "learning_rate": 1.1590639233045002e-05,
      "loss": 0.2773,
      "step": 11396
    },
    {
      "epoch": 5.363294117647059,
      "grad_norm": 0.2199014635568499,
      "learning_rate": 1.1573748525978626e-05,
      "loss": 0.0415,
      "step": 11397
    },
    {
      "epoch": 5.363764705882353,
      "grad_norm": 0.21354109268722055,
      "learning_rate": 1.155686964119309e-05,
      "loss": 0.0347,
      "step": 11398
    },
    {
      "epoch": 5.364235294117647,
      "grad_norm": 0.3964153991688905,
      "learning_rate": 1.154000258012992e-05,
      "loss": 0.1816,
      "step": 11399
    },
    {
      "epoch": 5.364705882352941,
      "grad_norm": 0.3520647363112602,
      "learning_rate": 1.1523147344229716e-05,
      "loss": 0.2773,
      "step": 11400
    },
    {
      "epoch": 5.365176470588235,
      "grad_norm": 0.45312876489511644,
      "learning_rate": 1.1506303934931926e-05,
      "loss": 0.6406,
      "step": 11401
    },
    {
      "epoch": 5.365647058823529,
      "grad_norm": 0.4181128033233732,
      "learning_rate": 1.1489472353675083e-05,
      "loss": 0.4023,
      "step": 11402
    },
    {
      "epoch": 5.366117647058823,
      "grad_norm": 0.3724198564772192,
      "learning_rate": 1.147265260189672e-05,
      "loss": 0.2354,
      "step": 11403
    },
    {
      "epoch": 5.366588235294118,
      "grad_norm": 0.46565075580646975,
      "learning_rate": 1.1455844681033272e-05,
      "loss": 0.2715,
      "step": 11404
    },
    {
      "epoch": 5.367058823529412,
      "grad_norm": 0.23672540018281749,
      "learning_rate": 1.1439048592520256e-05,
      "loss": 0.033,
      "step": 11405
    },
    {
      "epoch": 5.367529411764706,
      "grad_norm": 0.5055378592206383,
      "learning_rate": 1.1422264337792125e-05,
      "loss": 0.3848,
      "step": 11406
    },
    {
      "epoch": 5.368,
      "grad_norm": 0.49380203171137926,
      "learning_rate": 1.1405491918282333e-05,
      "loss": 0.3145,
      "step": 11407
    },
    {
      "epoch": 5.368470588235294,
      "grad_norm": 0.23288036985260083,
      "learning_rate": 1.1388731335423334e-05,
      "loss": 0.0447,
      "step": 11408
    },
    {
      "epoch": 5.368941176470588,
      "grad_norm": 0.45010514478540475,
      "learning_rate": 1.1371982590646584e-05,
      "loss": 0.5156,
      "step": 11409
    },
    {
      "epoch": 5.369411764705882,
      "grad_norm": 0.31908182330974616,
      "learning_rate": 1.135524568538247e-05,
      "loss": 0.2559,
      "step": 11410
    },
    {
      "epoch": 5.369882352941176,
      "grad_norm": 0.314308161158564,
      "learning_rate": 1.1338520621060415e-05,
      "loss": 0.2676,
      "step": 11411
    },
    {
      "epoch": 5.3703529411764706,
      "grad_norm": 0.29979831280157054,
      "learning_rate": 1.1321807399108845e-05,
      "loss": 0.1187,
      "step": 11412
    },
    {
      "epoch": 5.370823529411765,
      "grad_norm": 0.7614083807583614,
      "learning_rate": 1.1305106020955135e-05,
      "loss": 0.3828,
      "step": 11413
    },
    {
      "epoch": 5.371294117647059,
      "grad_norm": 0.3341126693717002,
      "learning_rate": 1.1288416488025675e-05,
      "loss": 0.1162,
      "step": 11414
    },
    {
      "epoch": 5.371764705882353,
      "grad_norm": 0.40561092317109965,
      "learning_rate": 1.1271738801745795e-05,
      "loss": 0.4121,
      "step": 11415
    },
    {
      "epoch": 5.372235294117647,
      "grad_norm": 0.43596484863933926,
      "learning_rate": 1.1255072963539885e-05,
      "loss": 0.2188,
      "step": 11416
    },
    {
      "epoch": 5.372705882352941,
      "grad_norm": 0.7761116047159136,
      "learning_rate": 1.123841897483131e-05,
      "loss": 0.3652,
      "step": 11417
    },
    {
      "epoch": 5.373176470588235,
      "grad_norm": 0.44683902273516024,
      "learning_rate": 1.1221776837042345e-05,
      "loss": 0.4043,
      "step": 11418
    },
    {
      "epoch": 5.373647058823529,
      "grad_norm": 0.2953704770188899,
      "learning_rate": 1.1205146551594357e-05,
      "loss": 0.1021,
      "step": 11419
    },
    {
      "epoch": 5.374117647058823,
      "grad_norm": 0.49996882270206133,
      "learning_rate": 1.1188528119907642e-05,
      "loss": 0.2578,
      "step": 11420
    },
    {
      "epoch": 5.374588235294118,
      "grad_norm": 0.28605545647375363,
      "learning_rate": 1.117192154340148e-05,
      "loss": 0.0898,
      "step": 11421
    },
    {
      "epoch": 5.375058823529412,
      "grad_norm": 0.446121639114466,
      "learning_rate": 1.115532682349417e-05,
      "loss": 0.2217,
      "step": 11422
    },
    {
      "epoch": 5.375529411764706,
      "grad_norm": 0.2992738060711146,
      "learning_rate": 1.1138743961602998e-05,
      "loss": 0.1582,
      "step": 11423
    },
    {
      "epoch": 5.376,
      "grad_norm": 0.6404759874325242,
      "learning_rate": 1.1122172959144177e-05,
      "loss": 0.6445,
      "step": 11424
    },
    {
      "epoch": 5.376470588235295,
      "grad_norm": 0.41672243217178107,
      "learning_rate": 1.1105613817532976e-05,
      "loss": 0.1484,
      "step": 11425
    },
    {
      "epoch": 5.376941176470588,
      "grad_norm": 0.10212043677253389,
      "learning_rate": 1.108906653818365e-05,
      "loss": 0.0108,
      "step": 11426
    },
    {
      "epoch": 5.377411764705882,
      "grad_norm": 0.4583827554316204,
      "learning_rate": 1.1072531122509381e-05,
      "loss": 0.2383,
      "step": 11427
    },
    {
      "epoch": 5.377882352941176,
      "grad_norm": 0.9491452399553824,
      "learning_rate": 1.1056007571922377e-05,
      "loss": 0.957,
      "step": 11428
    },
    {
      "epoch": 5.378352941176471,
      "grad_norm": 0.29364502193497766,
      "learning_rate": 1.1039495887833872e-05,
      "loss": 0.126,
      "step": 11429
    },
    {
      "epoch": 5.378823529411765,
      "grad_norm": 0.4303639730905759,
      "learning_rate": 1.1022996071653988e-05,
      "loss": 0.4043,
      "step": 11430
    },
    {
      "epoch": 5.379294117647059,
      "grad_norm": 0.1450493761326574,
      "learning_rate": 1.1006508124791935e-05,
      "loss": 0.0447,
      "step": 11431
    },
    {
      "epoch": 5.379764705882353,
      "grad_norm": 0.5015153450235245,
      "learning_rate": 1.099003204865585e-05,
      "loss": 0.2305,
      "step": 11432
    },
    {
      "epoch": 5.3802352941176474,
      "grad_norm": 0.43114549876436675,
      "learning_rate": 1.0973567844652858e-05,
      "loss": 0.1836,
      "step": 11433
    },
    {
      "epoch": 5.380705882352941,
      "grad_norm": 0.3011263835379001,
      "learning_rate": 1.0957115514189086e-05,
      "loss": 0.1982,
      "step": 11434
    },
    {
      "epoch": 5.381176470588235,
      "grad_norm": 0.2738458669978962,
      "learning_rate": 1.0940675058669673e-05,
      "loss": 0.1865,
      "step": 11435
    },
    {
      "epoch": 5.381647058823529,
      "grad_norm": 0.15306313767767024,
      "learning_rate": 1.0924246479498682e-05,
      "loss": 0.0284,
      "step": 11436
    },
    {
      "epoch": 5.3821176470588235,
      "grad_norm": 0.556195073441866,
      "learning_rate": 1.0907829778079207e-05,
      "loss": 0.3711,
      "step": 11437
    },
    {
      "epoch": 5.382588235294118,
      "grad_norm": 0.3526980039704095,
      "learning_rate": 1.089142495581331e-05,
      "loss": 0.1348,
      "step": 11438
    },
    {
      "epoch": 5.383058823529412,
      "grad_norm": 0.49175561612835605,
      "learning_rate": 1.0875032014102036e-05,
      "loss": 0.249,
      "step": 11439
    },
    {
      "epoch": 5.383529411764706,
      "grad_norm": 0.3439242821107739,
      "learning_rate": 1.0858650954345433e-05,
      "loss": 0.293,
      "step": 11440
    },
    {
      "epoch": 5.384,
      "grad_norm": 0.2209193076942894,
      "learning_rate": 1.0842281777942514e-05,
      "loss": 0.0415,
      "step": 11441
    },
    {
      "epoch": 5.384470588235294,
      "grad_norm": 0.5078155342526595,
      "learning_rate": 1.082592448629131e-05,
      "loss": 0.2471,
      "step": 11442
    },
    {
      "epoch": 5.384941176470588,
      "grad_norm": 0.3801611223272425,
      "learning_rate": 1.0809579080788805e-05,
      "loss": 0.103,
      "step": 11443
    },
    {
      "epoch": 5.385411764705882,
      "grad_norm": 0.1969498651452857,
      "learning_rate": 1.0793245562830966e-05,
      "loss": 0.1025,
      "step": 11444
    },
    {
      "epoch": 5.385882352941176,
      "grad_norm": 0.600401201723486,
      "learning_rate": 1.0776923933812775e-05,
      "loss": 0.4199,
      "step": 11445
    },
    {
      "epoch": 5.386352941176471,
      "grad_norm": 0.39904906370162724,
      "learning_rate": 1.0760614195128153e-05,
      "loss": 0.2559,
      "step": 11446
    },
    {
      "epoch": 5.386823529411765,
      "grad_norm": 0.4164666301861593,
      "learning_rate": 1.0744316348170035e-05,
      "loss": 0.2061,
      "step": 11447
    },
    {
      "epoch": 5.387294117647059,
      "grad_norm": 0.18146411639731347,
      "learning_rate": 1.0728030394330373e-05,
      "loss": 0.0422,
      "step": 11448
    },
    {
      "epoch": 5.387764705882353,
      "grad_norm": 0.3055096130024368,
      "learning_rate": 1.0711756335000005e-05,
      "loss": 0.2051,
      "step": 11449
    },
    {
      "epoch": 5.3882352941176475,
      "grad_norm": 1.0063729198971851,
      "learning_rate": 1.069549417156887e-05,
      "loss": 0.9531,
      "step": 11450
    },
    {
      "epoch": 5.388705882352941,
      "grad_norm": 0.3867486715898068,
      "learning_rate": 1.0679243905425822e-05,
      "loss": 0.1235,
      "step": 11451
    },
    {
      "epoch": 5.389176470588235,
      "grad_norm": 0.35769111725118846,
      "learning_rate": 1.0663005537958702e-05,
      "loss": 0.103,
      "step": 11452
    },
    {
      "epoch": 5.389647058823529,
      "grad_norm": 0.5576600819276822,
      "learning_rate": 1.0646779070554335e-05,
      "loss": 0.3379,
      "step": 11453
    },
    {
      "epoch": 5.3901176470588235,
      "grad_norm": 0.4768059679829414,
      "learning_rate": 1.0630564504598593e-05,
      "loss": 0.168,
      "step": 11454
    },
    {
      "epoch": 5.390588235294118,
      "grad_norm": 0.4954008184360589,
      "learning_rate": 1.0614361841476238e-05,
      "loss": 0.2617,
      "step": 11455
    },
    {
      "epoch": 5.391058823529412,
      "grad_norm": 0.3551674464888567,
      "learning_rate": 1.0598171082571044e-05,
      "loss": 0.1523,
      "step": 11456
    },
    {
      "epoch": 5.391529411764706,
      "grad_norm": 0.26201233303400506,
      "learning_rate": 1.0581992229265823e-05,
      "loss": 0.1299,
      "step": 11457
    },
    {
      "epoch": 5.392,
      "grad_norm": 0.8029073048049113,
      "learning_rate": 1.0565825282942303e-05,
      "loss": 0.3418,
      "step": 11458
    },
    {
      "epoch": 5.392470588235295,
      "grad_norm": 0.23452259318144422,
      "learning_rate": 1.0549670244981211e-05,
      "loss": 0.0364,
      "step": 11459
    },
    {
      "epoch": 5.392941176470588,
      "grad_norm": 0.3285738961872649,
      "learning_rate": 1.0533527116762296e-05,
      "loss": 0.0552,
      "step": 11460
    },
    {
      "epoch": 5.393411764705882,
      "grad_norm": 0.40252010117504244,
      "learning_rate": 1.0517395899664239e-05,
      "loss": 0.2617,
      "step": 11461
    },
    {
      "epoch": 5.393882352941176,
      "grad_norm": 0.45477687967593405,
      "learning_rate": 1.0501276595064716e-05,
      "loss": 0.416,
      "step": 11462
    },
    {
      "epoch": 5.394352941176471,
      "grad_norm": 0.3628030731008278,
      "learning_rate": 1.0485169204340432e-05,
      "loss": 0.3574,
      "step": 11463
    },
    {
      "epoch": 5.394823529411765,
      "grad_norm": 0.41837661072484844,
      "learning_rate": 1.0469073728867e-05,
      "loss": 0.2217,
      "step": 11464
    },
    {
      "epoch": 5.395294117647059,
      "grad_norm": 0.23045616491599935,
      "learning_rate": 1.0452990170019088e-05,
      "loss": 0.0579,
      "step": 11465
    },
    {
      "epoch": 5.395764705882353,
      "grad_norm": 0.34767224384655904,
      "learning_rate": 1.0436918529170264e-05,
      "loss": 0.3672,
      "step": 11466
    },
    {
      "epoch": 5.3962352941176475,
      "grad_norm": 0.46188498288313146,
      "learning_rate": 1.0420858807693162e-05,
      "loss": 0.2617,
      "step": 11467
    },
    {
      "epoch": 5.396705882352941,
      "grad_norm": 0.2746667263988407,
      "learning_rate": 1.0404811006959352e-05,
      "loss": 0.1396,
      "step": 11468
    },
    {
      "epoch": 5.397176470588235,
      "grad_norm": 0.5710682676611132,
      "learning_rate": 1.0388775128339389e-05,
      "loss": 0.3379,
      "step": 11469
    },
    {
      "epoch": 5.397647058823529,
      "grad_norm": 0.2445527791929684,
      "learning_rate": 1.037275117320281e-05,
      "loss": 0.1216,
      "step": 11470
    },
    {
      "epoch": 5.3981176470588235,
      "grad_norm": 0.5119147995107477,
      "learning_rate": 1.0356739142918169e-05,
      "loss": 0.2637,
      "step": 11471
    },
    {
      "epoch": 5.398588235294118,
      "grad_norm": 0.45148884478836676,
      "learning_rate": 1.0340739038852924e-05,
      "loss": 0.3066,
      "step": 11472
    },
    {
      "epoch": 5.399058823529412,
      "grad_norm": 0.49337309479277647,
      "learning_rate": 1.0324750862373598e-05,
      "loss": 0.3496,
      "step": 11473
    },
    {
      "epoch": 5.399529411764706,
      "grad_norm": 0.4094922757843426,
      "learning_rate": 1.0308774614845683e-05,
      "loss": 0.2373,
      "step": 11474
    },
    {
      "epoch": 5.4,
      "grad_norm": 0.4741279588801564,
      "learning_rate": 1.0292810297633569e-05,
      "loss": 0.3711,
      "step": 11475
    },
    {
      "epoch": 5.400470588235294,
      "grad_norm": 0.447662114616631,
      "learning_rate": 1.02768579121007e-05,
      "loss": 0.3867,
      "step": 11476
    },
    {
      "epoch": 5.400941176470588,
      "grad_norm": 0.35316579412389676,
      "learning_rate": 1.0260917459609535e-05,
      "loss": 0.1943,
      "step": 11477
    },
    {
      "epoch": 5.401411764705882,
      "grad_norm": 0.43416713428927445,
      "learning_rate": 1.024498894152142e-05,
      "loss": 0.2178,
      "step": 11478
    },
    {
      "epoch": 5.401882352941176,
      "grad_norm": 0.29649128830091687,
      "learning_rate": 1.0229072359196749e-05,
      "loss": 0.25,
      "step": 11479
    },
    {
      "epoch": 5.402352941176471,
      "grad_norm": 0.3106079788439805,
      "learning_rate": 1.0213167713994885e-05,
      "loss": 0.2324,
      "step": 11480
    },
    {
      "epoch": 5.402823529411765,
      "grad_norm": 0.4647073207561836,
      "learning_rate": 1.0197275007274126e-05,
      "loss": 0.5156,
      "step": 11481
    },
    {
      "epoch": 5.403294117647059,
      "grad_norm": 0.35928071663471567,
      "learning_rate": 1.0181394240391798e-05,
      "loss": 0.4531,
      "step": 11482
    },
    {
      "epoch": 5.403764705882353,
      "grad_norm": 0.37607733095780654,
      "learning_rate": 1.016552541470424e-05,
      "loss": 0.2178,
      "step": 11483
    },
    {
      "epoch": 5.4042352941176475,
      "grad_norm": 0.6567994179065229,
      "learning_rate": 1.0149668531566662e-05,
      "loss": 0.4336,
      "step": 11484
    },
    {
      "epoch": 5.404705882352941,
      "grad_norm": 0.4484683576218808,
      "learning_rate": 1.0133823592333385e-05,
      "loss": 0.0942,
      "step": 11485
    },
    {
      "epoch": 5.405176470588235,
      "grad_norm": 0.18004495485976765,
      "learning_rate": 1.0117990598357589e-05,
      "loss": 0.0325,
      "step": 11486
    },
    {
      "epoch": 5.405647058823529,
      "grad_norm": 0.2704303200883759,
      "learning_rate": 1.0102169550991529e-05,
      "loss": 0.0903,
      "step": 11487
    },
    {
      "epoch": 5.4061176470588235,
      "grad_norm": 0.26818219878158367,
      "learning_rate": 1.008636045158639e-05,
      "loss": 0.126,
      "step": 11488
    },
    {
      "epoch": 5.406588235294118,
      "grad_norm": 0.5198813512846089,
      "learning_rate": 1.0070563301492323e-05,
      "loss": 0.416,
      "step": 11489
    },
    {
      "epoch": 5.407058823529412,
      "grad_norm": 0.41615669800567895,
      "learning_rate": 1.0054778102058502e-05,
      "loss": 0.3223,
      "step": 11490
    },
    {
      "epoch": 5.407529411764706,
      "grad_norm": 0.5591609136747304,
      "learning_rate": 1.0039004854633082e-05,
      "loss": 0.4297,
      "step": 11491
    },
    {
      "epoch": 5.408,
      "grad_norm": 0.4027218779070535,
      "learning_rate": 1.0023243560563132e-05,
      "loss": 0.1143,
      "step": 11492
    },
    {
      "epoch": 5.408470588235295,
      "grad_norm": 0.5321571497340792,
      "learning_rate": 1.0007494221194762e-05,
      "loss": 0.2002,
      "step": 11493
    },
    {
      "epoch": 5.408941176470588,
      "grad_norm": 0.47188176752494754,
      "learning_rate": 9.991756837873061e-06,
      "loss": 0.2441,
      "step": 11494
    },
    {
      "epoch": 5.409411764705882,
      "grad_norm": 0.23842413066750542,
      "learning_rate": 9.976031411942054e-06,
      "loss": 0.0664,
      "step": 11495
    },
    {
      "epoch": 5.409882352941176,
      "grad_norm": 0.5307196717082754,
      "learning_rate": 9.960317944744783e-06,
      "loss": 0.2598,
      "step": 11496
    },
    {
      "epoch": 5.410352941176471,
      "grad_norm": 0.3785001071343024,
      "learning_rate": 9.944616437623243e-06,
      "loss": 0.3477,
      "step": 11497
    },
    {
      "epoch": 5.410823529411765,
      "grad_norm": 0.1394848801957179,
      "learning_rate": 9.928926891918426e-06,
      "loss": 0.024,
      "step": 11498
    },
    {
      "epoch": 5.411294117647059,
      "grad_norm": 0.5100325399327749,
      "learning_rate": 9.913249308970312e-06,
      "loss": 0.1289,
      "step": 11499
    },
    {
      "epoch": 5.411764705882353,
      "grad_norm": 0.1902202387136826,
      "learning_rate": 9.89758369011781e-06,
      "loss": 0.0215,
      "step": 11500
    },
    {
      "epoch": 5.412235294117647,
      "grad_norm": 2.148449597048725,
      "learning_rate": 9.881930036698854e-06,
      "loss": 1.5781,
      "step": 11501
    },
    {
      "epoch": 5.412705882352941,
      "grad_norm": 0.6883688214477386,
      "learning_rate": 9.866288350050372e-06,
      "loss": 0.4629,
      "step": 11502
    },
    {
      "epoch": 5.413176470588235,
      "grad_norm": 0.30858601637353733,
      "learning_rate": 9.850658631508196e-06,
      "loss": 0.0649,
      "step": 11503
    },
    {
      "epoch": 5.413647058823529,
      "grad_norm": 0.35460917339866765,
      "learning_rate": 9.835040882407207e-06,
      "loss": 0.1797,
      "step": 11504
    },
    {
      "epoch": 5.4141176470588235,
      "grad_norm": 0.43182062989566533,
      "learning_rate": 9.819435104081242e-06,
      "loss": 0.4375,
      "step": 11505
    },
    {
      "epoch": 5.414588235294118,
      "grad_norm": 0.38696761956409537,
      "learning_rate": 9.803841297863096e-06,
      "loss": 0.334,
      "step": 11506
    },
    {
      "epoch": 5.415058823529412,
      "grad_norm": 0.3161181159146492,
      "learning_rate": 9.788259465084541e-06,
      "loss": 0.1465,
      "step": 11507
    },
    {
      "epoch": 5.415529411764706,
      "grad_norm": 0.3685920928132292,
      "learning_rate": 9.772689607076394e-06,
      "loss": 0.1377,
      "step": 11508
    },
    {
      "epoch": 5.416,
      "grad_norm": 0.2009950317158226,
      "learning_rate": 9.757131725168343e-06,
      "loss": 0.0732,
      "step": 11509
    },
    {
      "epoch": 5.416470588235294,
      "grad_norm": 0.8558634393697134,
      "learning_rate": 9.74158582068914e-06,
      "loss": 0.4668,
      "step": 11510
    },
    {
      "epoch": 5.416941176470588,
      "grad_norm": 0.3754133056513377,
      "learning_rate": 9.726051894966492e-06,
      "loss": 0.2969,
      "step": 11511
    },
    {
      "epoch": 5.417411764705882,
      "grad_norm": 0.26147705085153194,
      "learning_rate": 9.710529949327018e-06,
      "loss": 0.0732,
      "step": 11512
    },
    {
      "epoch": 5.417882352941176,
      "grad_norm": 0.33829299906902205,
      "learning_rate": 9.695019985096409e-06,
      "loss": 0.3223,
      "step": 11513
    },
    {
      "epoch": 5.418352941176471,
      "grad_norm": 0.28485099594029795,
      "learning_rate": 9.679522003599305e-06,
      "loss": 0.0933,
      "step": 11514
    },
    {
      "epoch": 5.418823529411765,
      "grad_norm": 0.503711728137421,
      "learning_rate": 9.664036006159266e-06,
      "loss": 0.5586,
      "step": 11515
    },
    {
      "epoch": 5.419294117647059,
      "grad_norm": 0.22855712010344945,
      "learning_rate": 9.648561994098913e-06,
      "loss": 0.0674,
      "step": 11516
    },
    {
      "epoch": 5.419764705882353,
      "grad_norm": 0.38150388620627296,
      "learning_rate": 9.633099968739777e-06,
      "loss": 0.2734,
      "step": 11517
    },
    {
      "epoch": 5.4202352941176475,
      "grad_norm": 0.2870105391416525,
      "learning_rate": 9.617649931402399e-06,
      "loss": 0.1494,
      "step": 11518
    },
    {
      "epoch": 5.420705882352941,
      "grad_norm": 0.5078283674882528,
      "learning_rate": 9.602211883406308e-06,
      "loss": 0.5391,
      "step": 11519
    },
    {
      "epoch": 5.421176470588235,
      "grad_norm": 0.27399560850144633,
      "learning_rate": 9.586785826069932e-06,
      "loss": 0.2246,
      "step": 11520
    },
    {
      "epoch": 5.421647058823529,
      "grad_norm": 0.5436101176025282,
      "learning_rate": 9.571371760710783e-06,
      "loss": 0.75,
      "step": 11521
    },
    {
      "epoch": 5.4221176470588235,
      "grad_norm": 0.1367122894465647,
      "learning_rate": 9.55596968864531e-06,
      "loss": 0.0269,
      "step": 11522
    },
    {
      "epoch": 5.422588235294118,
      "grad_norm": 0.4721932338190693,
      "learning_rate": 9.540579611188875e-06,
      "loss": 0.2676,
      "step": 11523
    },
    {
      "epoch": 5.423058823529412,
      "grad_norm": 0.19369386293365548,
      "learning_rate": 9.52520152965588e-06,
      "loss": 0.0308,
      "step": 11524
    },
    {
      "epoch": 5.423529411764706,
      "grad_norm": 0.400089208732642,
      "learning_rate": 9.509835445359736e-06,
      "loss": 0.2852,
      "step": 11525
    },
    {
      "epoch": 5.424,
      "grad_norm": 0.4563813753435336,
      "learning_rate": 9.494481359612715e-06,
      "loss": 0.1387,
      "step": 11526
    },
    {
      "epoch": 5.424470588235294,
      "grad_norm": 0.38646858312430216,
      "learning_rate": 9.479139273726166e-06,
      "loss": 0.2471,
      "step": 11527
    },
    {
      "epoch": 5.424941176470588,
      "grad_norm": 0.3097187647419007,
      "learning_rate": 9.46380918901039e-06,
      "loss": 0.0991,
      "step": 11528
    },
    {
      "epoch": 5.425411764705882,
      "grad_norm": 0.2675529113078091,
      "learning_rate": 9.448491106774591e-06,
      "loss": 0.0659,
      "step": 11529
    },
    {
      "epoch": 5.425882352941176,
      "grad_norm": 0.4145868512555757,
      "learning_rate": 9.433185028327106e-06,
      "loss": 0.4551,
      "step": 11530
    },
    {
      "epoch": 5.426352941176471,
      "grad_norm": 0.2724743670629038,
      "learning_rate": 9.417890954975071e-06,
      "loss": 0.0674,
      "step": 11531
    },
    {
      "epoch": 5.426823529411765,
      "grad_norm": 0.22004883281183168,
      "learning_rate": 9.402608888024692e-06,
      "loss": 0.0525,
      "step": 11532
    },
    {
      "epoch": 5.427294117647059,
      "grad_norm": 0.29953388899964145,
      "learning_rate": 9.38733882878116e-06,
      "loss": 0.2109,
      "step": 11533
    },
    {
      "epoch": 5.427764705882353,
      "grad_norm": 0.2903290344854727,
      "learning_rate": 9.37208077854858e-06,
      "loss": 0.0864,
      "step": 11534
    },
    {
      "epoch": 5.428235294117647,
      "grad_norm": 0.4520226908530776,
      "learning_rate": 9.356834738630075e-06,
      "loss": 0.252,
      "step": 11535
    },
    {
      "epoch": 5.428705882352941,
      "grad_norm": 0.47968957796081124,
      "learning_rate": 9.341600710327773e-06,
      "loss": 0.7188,
      "step": 11536
    },
    {
      "epoch": 5.429176470588235,
      "grad_norm": 0.5651540690879873,
      "learning_rate": 9.32637869494265e-06,
      "loss": 0.416,
      "step": 11537
    },
    {
      "epoch": 5.429647058823529,
      "grad_norm": 0.4025554202991636,
      "learning_rate": 9.311168693774796e-06,
      "loss": 0.3574,
      "step": 11538
    },
    {
      "epoch": 5.4301176470588235,
      "grad_norm": 0.3162556929825003,
      "learning_rate": 9.295970708123223e-06,
      "loss": 0.1348,
      "step": 11539
    },
    {
      "epoch": 5.430588235294118,
      "grad_norm": 1.0455577433531433,
      "learning_rate": 9.280784739285896e-06,
      "loss": 0.5547,
      "step": 11540
    },
    {
      "epoch": 5.431058823529412,
      "grad_norm": 0.30151667674631183,
      "learning_rate": 9.265610788559758e-06,
      "loss": 0.1328,
      "step": 11541
    },
    {
      "epoch": 5.431529411764706,
      "grad_norm": 0.5232862425625403,
      "learning_rate": 9.250448857240789e-06,
      "loss": 0.2539,
      "step": 11542
    },
    {
      "epoch": 5.432,
      "grad_norm": 0.22518083474870118,
      "learning_rate": 9.235298946623838e-06,
      "loss": 0.0718,
      "step": 11543
    },
    {
      "epoch": 5.432470588235294,
      "grad_norm": 0.36196644335482314,
      "learning_rate": 9.220161058002834e-06,
      "loss": 0.2539,
      "step": 11544
    },
    {
      "epoch": 5.432941176470588,
      "grad_norm": 0.41349195227792374,
      "learning_rate": 9.205035192670563e-06,
      "loss": 0.0503,
      "step": 11545
    },
    {
      "epoch": 5.433411764705882,
      "grad_norm": 0.2340858339136981,
      "learning_rate": 9.189921351918888e-06,
      "loss": 0.083,
      "step": 11546
    },
    {
      "epoch": 5.433882352941176,
      "grad_norm": 0.4797471662681557,
      "learning_rate": 9.174819537038614e-06,
      "loss": 0.3555,
      "step": 11547
    },
    {
      "epoch": 5.434352941176471,
      "grad_norm": 0.3826941979985712,
      "learning_rate": 9.159729749319473e-06,
      "loss": 0.3594,
      "step": 11548
    },
    {
      "epoch": 5.434823529411765,
      "grad_norm": 0.2345719402616443,
      "learning_rate": 9.144651990050234e-06,
      "loss": 0.0898,
      "step": 11549
    },
    {
      "epoch": 5.435294117647059,
      "grad_norm": 0.2139196594379149,
      "learning_rate": 9.129586260518634e-06,
      "loss": 0.1006,
      "step": 11550
    },
    {
      "epoch": 5.435764705882353,
      "grad_norm": 0.4490483293122125,
      "learning_rate": 9.114532562011312e-06,
      "loss": 0.1934,
      "step": 11551
    },
    {
      "epoch": 5.4362352941176475,
      "grad_norm": 0.5220816009301131,
      "learning_rate": 9.099490895813938e-06,
      "loss": 0.1924,
      "step": 11552
    },
    {
      "epoch": 5.436705882352941,
      "grad_norm": 0.5520947768683253,
      "learning_rate": 9.084461263211168e-06,
      "loss": 0.5547,
      "step": 11553
    },
    {
      "epoch": 5.437176470588235,
      "grad_norm": 0.5273800078556579,
      "learning_rate": 9.069443665486593e-06,
      "loss": 0.457,
      "step": 11554
    },
    {
      "epoch": 5.437647058823529,
      "grad_norm": 0.3199092568769854,
      "learning_rate": 9.054438103922767e-06,
      "loss": 0.0923,
      "step": 11555
    },
    {
      "epoch": 5.4381176470588235,
      "grad_norm": 0.4354831781684606,
      "learning_rate": 9.039444579801302e-06,
      "loss": 0.3418,
      "step": 11556
    },
    {
      "epoch": 5.438588235294118,
      "grad_norm": 0.42136189839975075,
      "learning_rate": 9.024463094402656e-06,
      "loss": 0.4668,
      "step": 11557
    },
    {
      "epoch": 5.439058823529412,
      "grad_norm": 0.4965908404934829,
      "learning_rate": 9.009493649006322e-06,
      "loss": 0.5352,
      "step": 11558
    },
    {
      "epoch": 5.439529411764706,
      "grad_norm": 0.3062758518336207,
      "learning_rate": 8.994536244890827e-06,
      "loss": 0.0747,
      "step": 11559
    },
    {
      "epoch": 5.44,
      "grad_norm": 0.40961954901249137,
      "learning_rate": 8.979590883333532e-06,
      "loss": 0.4414,
      "step": 11560
    },
    {
      "epoch": 5.440470588235294,
      "grad_norm": 0.4234630687580191,
      "learning_rate": 8.964657565610884e-06,
      "loss": 0.3086,
      "step": 11561
    },
    {
      "epoch": 5.440941176470588,
      "grad_norm": 0.4265741466988509,
      "learning_rate": 8.949736292998277e-06,
      "loss": 0.3359,
      "step": 11562
    },
    {
      "epoch": 5.441411764705882,
      "grad_norm": 0.43421761214487437,
      "learning_rate": 8.934827066770012e-06,
      "loss": 0.25,
      "step": 11563
    },
    {
      "epoch": 5.441882352941176,
      "grad_norm": 0.3170934144154193,
      "learning_rate": 8.919929888199435e-06,
      "loss": 0.1475,
      "step": 11564
    },
    {
      "epoch": 5.442352941176471,
      "grad_norm": 0.41768079813475384,
      "learning_rate": 8.90504475855886e-06,
      "loss": 0.3887,
      "step": 11565
    },
    {
      "epoch": 5.442823529411765,
      "grad_norm": 0.43851424448912274,
      "learning_rate": 8.890171679119506e-06,
      "loss": 0.165,
      "step": 11566
    },
    {
      "epoch": 5.443294117647059,
      "grad_norm": 0.49460071089750524,
      "learning_rate": 8.875310651151657e-06,
      "loss": 0.707,
      "step": 11567
    },
    {
      "epoch": 5.443764705882353,
      "grad_norm": 0.47316642592181385,
      "learning_rate": 8.86046167592448e-06,
      "loss": 0.2812,
      "step": 11568
    },
    {
      "epoch": 5.444235294117647,
      "grad_norm": 0.3535476596023794,
      "learning_rate": 8.845624754706144e-06,
      "loss": 0.3145,
      "step": 11569
    },
    {
      "epoch": 5.444705882352941,
      "grad_norm": 0.3649808802233409,
      "learning_rate": 8.830799888763834e-06,
      "loss": 0.2539,
      "step": 11570
    },
    {
      "epoch": 5.445176470588235,
      "grad_norm": 0.5087544064363928,
      "learning_rate": 8.81598707936364e-06,
      "loss": 0.2422,
      "step": 11571
    },
    {
      "epoch": 5.445647058823529,
      "grad_norm": 0.44259292034846426,
      "learning_rate": 8.801186327770632e-06,
      "loss": 0.4863,
      "step": 11572
    },
    {
      "epoch": 5.4461176470588235,
      "grad_norm": 0.25981045372162453,
      "learning_rate": 8.7863976352489e-06,
      "loss": 0.0679,
      "step": 11573
    },
    {
      "epoch": 5.446588235294118,
      "grad_norm": 0.35531705330153374,
      "learning_rate": 8.771621003061463e-06,
      "loss": 0.3574,
      "step": 11574
    },
    {
      "epoch": 5.447058823529412,
      "grad_norm": 0.5339371889585145,
      "learning_rate": 8.756856432470315e-06,
      "loss": 0.457,
      "step": 11575
    },
    {
      "epoch": 5.447529411764706,
      "grad_norm": 0.4762421545540575,
      "learning_rate": 8.742103924736416e-06,
      "loss": 0.543,
      "step": 11576
    },
    {
      "epoch": 5.448,
      "grad_norm": 0.38408464563920675,
      "learning_rate": 8.727363481119704e-06,
      "loss": 0.1885,
      "step": 11577
    },
    {
      "epoch": 5.448470588235294,
      "grad_norm": 0.5809347994882325,
      "learning_rate": 8.712635102879124e-06,
      "loss": 0.5898,
      "step": 11578
    },
    {
      "epoch": 5.448941176470588,
      "grad_norm": 0.47430081346533315,
      "learning_rate": 8.697918791272485e-06,
      "loss": 0.1699,
      "step": 11579
    },
    {
      "epoch": 5.449411764705882,
      "grad_norm": 0.1841972134692871,
      "learning_rate": 8.68321454755665e-06,
      "loss": 0.041,
      "step": 11580
    },
    {
      "epoch": 5.449882352941176,
      "grad_norm": 0.4148515538989125,
      "learning_rate": 8.668522372987479e-06,
      "loss": 0.3535,
      "step": 11581
    },
    {
      "epoch": 5.450352941176471,
      "grad_norm": 0.6035944212517966,
      "learning_rate": 8.653842268819723e-06,
      "loss": 0.4453,
      "step": 11582
    },
    {
      "epoch": 5.450823529411765,
      "grad_norm": 0.2853227642686064,
      "learning_rate": 8.639174236307106e-06,
      "loss": 0.0967,
      "step": 11583
    },
    {
      "epoch": 5.451294117647059,
      "grad_norm": 0.49228818077871955,
      "learning_rate": 8.624518276702414e-06,
      "loss": 0.3477,
      "step": 11584
    },
    {
      "epoch": 5.451764705882353,
      "grad_norm": 0.2960520427401641,
      "learning_rate": 8.609874391257293e-06,
      "loss": 0.0483,
      "step": 11585
    },
    {
      "epoch": 5.4522352941176475,
      "grad_norm": 0.2211740038640452,
      "learning_rate": 8.595242581222394e-06,
      "loss": 0.0698,
      "step": 11586
    },
    {
      "epoch": 5.452705882352941,
      "grad_norm": 0.5414333134339916,
      "learning_rate": 8.580622847847402e-06,
      "loss": 0.375,
      "step": 11587
    },
    {
      "epoch": 5.453176470588235,
      "grad_norm": 0.6452255570935235,
      "learning_rate": 8.566015192380849e-06,
      "loss": 0.3262,
      "step": 11588
    },
    {
      "epoch": 5.453647058823529,
      "grad_norm": 0.18397814502959828,
      "learning_rate": 8.551419616070322e-06,
      "loss": 0.0322,
      "step": 11589
    },
    {
      "epoch": 5.4541176470588235,
      "grad_norm": 0.12915011723275832,
      "learning_rate": 8.536836120162389e-06,
      "loss": 0.0194,
      "step": 11590
    },
    {
      "epoch": 5.454588235294118,
      "grad_norm": 0.3390575368425607,
      "learning_rate": 8.522264705902504e-06,
      "loss": 0.3164,
      "step": 11591
    },
    {
      "epoch": 5.455058823529412,
      "grad_norm": 0.5346572352541707,
      "learning_rate": 8.507705374535157e-06,
      "loss": 0.2734,
      "step": 11592
    },
    {
      "epoch": 5.455529411764706,
      "grad_norm": 0.47865351343854556,
      "learning_rate": 8.4931581273038e-06,
      "loss": 0.4082,
      "step": 11593
    },
    {
      "epoch": 5.456,
      "grad_norm": 0.6172806686567103,
      "learning_rate": 8.47862296545081e-06,
      "loss": 0.4746,
      "step": 11594
    },
    {
      "epoch": 5.456470588235294,
      "grad_norm": 0.35991701167011897,
      "learning_rate": 8.464099890217607e-06,
      "loss": 0.2949,
      "step": 11595
    },
    {
      "epoch": 5.456941176470588,
      "grad_norm": 0.3299524145546912,
      "learning_rate": 8.449588902844484e-06,
      "loss": 0.1982,
      "step": 11596
    },
    {
      "epoch": 5.457411764705882,
      "grad_norm": 0.33175245020556404,
      "learning_rate": 8.43509000457075e-06,
      "loss": 0.1309,
      "step": 11597
    },
    {
      "epoch": 5.457882352941176,
      "grad_norm": 0.4203634973263005,
      "learning_rate": 8.420603196634745e-06,
      "loss": 0.2021,
      "step": 11598
    },
    {
      "epoch": 5.458352941176471,
      "grad_norm": 0.3871275836792604,
      "learning_rate": 8.406128480273633e-06,
      "loss": 0.249,
      "step": 11599
    },
    {
      "epoch": 5.458823529411765,
      "grad_norm": 0.2682147460405249,
      "learning_rate": 8.391665856723655e-06,
      "loss": 0.1973,
      "step": 11600
    },
    {
      "epoch": 5.459294117647059,
      "grad_norm": 0.13403640781808296,
      "learning_rate": 8.377215327220043e-06,
      "loss": 0.0198,
      "step": 11601
    },
    {
      "epoch": 5.459764705882353,
      "grad_norm": 0.9056838890340018,
      "learning_rate": 8.362776892996858e-06,
      "loss": 0.9453,
      "step": 11602
    },
    {
      "epoch": 5.460235294117647,
      "grad_norm": 0.36588722748444985,
      "learning_rate": 8.348350555287264e-06,
      "loss": 0.1182,
      "step": 11603
    },
    {
      "epoch": 5.460705882352941,
      "grad_norm": 0.49116034715548923,
      "learning_rate": 8.333936315323343e-06,
      "loss": 0.165,
      "step": 11604
    },
    {
      "epoch": 5.461176470588235,
      "grad_norm": 0.4896557042723789,
      "learning_rate": 8.319534174336106e-06,
      "loss": 0.5352,
      "step": 11605
    },
    {
      "epoch": 5.461647058823529,
      "grad_norm": 0.2899177831149316,
      "learning_rate": 8.305144133555608e-06,
      "loss": 0.1079,
      "step": 11606
    },
    {
      "epoch": 5.4621176470588235,
      "grad_norm": 0.5614837518550275,
      "learning_rate": 8.290766194210813e-06,
      "loss": 0.6133,
      "step": 11607
    },
    {
      "epoch": 5.462588235294118,
      "grad_norm": 0.27483188245472995,
      "learning_rate": 8.276400357529655e-06,
      "loss": 0.1592,
      "step": 11608
    },
    {
      "epoch": 5.463058823529412,
      "grad_norm": 0.6233256332988827,
      "learning_rate": 8.262046624739038e-06,
      "loss": 0.3027,
      "step": 11609
    },
    {
      "epoch": 5.463529411764706,
      "grad_norm": 0.4731546672599553,
      "learning_rate": 8.247704997064896e-06,
      "loss": 0.2334,
      "step": 11610
    },
    {
      "epoch": 5.464,
      "grad_norm": 0.38786590379906705,
      "learning_rate": 8.233375475732018e-06,
      "loss": 0.2637,
      "step": 11611
    },
    {
      "epoch": 5.464470588235294,
      "grad_norm": 0.3900040906034263,
      "learning_rate": 8.21905806196424e-06,
      "loss": 0.3223,
      "step": 11612
    },
    {
      "epoch": 5.464941176470588,
      "grad_norm": 0.16516011852365653,
      "learning_rate": 8.204752756984351e-06,
      "loss": 0.083,
      "step": 11613
    },
    {
      "epoch": 5.465411764705882,
      "grad_norm": 0.45219666971567335,
      "learning_rate": 8.19045956201404e-06,
      "loss": 0.2432,
      "step": 11614
    },
    {
      "epoch": 5.465882352941176,
      "grad_norm": 0.2539670266180984,
      "learning_rate": 8.176178478274082e-06,
      "loss": 0.0781,
      "step": 11615
    },
    {
      "epoch": 5.466352941176471,
      "grad_norm": 0.5380748605984598,
      "learning_rate": 8.161909506984132e-06,
      "loss": 0.2988,
      "step": 11616
    },
    {
      "epoch": 5.466823529411765,
      "grad_norm": 0.3878451455807487,
      "learning_rate": 8.147652649362802e-06,
      "loss": 0.1387,
      "step": 11617
    },
    {
      "epoch": 5.467294117647059,
      "grad_norm": 0.35128122279520485,
      "learning_rate": 8.133407906627715e-06,
      "loss": 0.2041,
      "step": 11618
    },
    {
      "epoch": 5.467764705882353,
      "grad_norm": 0.49292900142549556,
      "learning_rate": 8.119175279995448e-06,
      "loss": 0.291,
      "step": 11619
    },
    {
      "epoch": 5.4682352941176475,
      "grad_norm": 0.3080744040322835,
      "learning_rate": 8.104954770681532e-06,
      "loss": 0.1758,
      "step": 11620
    },
    {
      "epoch": 5.468705882352941,
      "grad_norm": 0.4651399378331303,
      "learning_rate": 8.090746379900475e-06,
      "loss": 0.4668,
      "step": 11621
    },
    {
      "epoch": 5.469176470588235,
      "grad_norm": 0.49236661688586747,
      "learning_rate": 8.076550108865743e-06,
      "loss": 0.4785,
      "step": 11622
    },
    {
      "epoch": 5.469647058823529,
      "grad_norm": 0.23761154287443537,
      "learning_rate": 8.062365958789746e-06,
      "loss": 0.0442,
      "step": 11623
    },
    {
      "epoch": 5.4701176470588235,
      "grad_norm": 0.26369725039449926,
      "learning_rate": 8.048193930883933e-06,
      "loss": 0.053,
      "step": 11624
    },
    {
      "epoch": 5.470588235294118,
      "grad_norm": 0.45364706217593337,
      "learning_rate": 8.034034026358587e-06,
      "loss": 0.3398,
      "step": 11625
    },
    {
      "epoch": 5.471058823529412,
      "grad_norm": 0.48895903954928616,
      "learning_rate": 8.019886246423107e-06,
      "loss": 0.5391,
      "step": 11626
    },
    {
      "epoch": 5.471529411764706,
      "grad_norm": 0.4055564071717596,
      "learning_rate": 8.005750592285725e-06,
      "loss": 0.291,
      "step": 11627
    },
    {
      "epoch": 5.4719999999999995,
      "grad_norm": 0.3455819795307333,
      "learning_rate": 7.991627065153727e-06,
      "loss": 0.1943,
      "step": 11628
    },
    {
      "epoch": 5.472470588235294,
      "grad_norm": 0.3147433970065289,
      "learning_rate": 7.977515666233347e-06,
      "loss": 0.2793,
      "step": 11629
    },
    {
      "epoch": 5.472941176470588,
      "grad_norm": 0.42871924677858103,
      "learning_rate": 7.963416396729722e-06,
      "loss": 0.1787,
      "step": 11630
    },
    {
      "epoch": 5.473411764705882,
      "grad_norm": 0.33556501693830726,
      "learning_rate": 7.94932925784702e-06,
      "loss": 0.3691,
      "step": 11631
    },
    {
      "epoch": 5.473882352941176,
      "grad_norm": 0.3617470034569276,
      "learning_rate": 7.935254250788364e-06,
      "loss": 0.0972,
      "step": 11632
    },
    {
      "epoch": 5.474352941176471,
      "grad_norm": 0.43694854455606513,
      "learning_rate": 7.921191376755808e-06,
      "loss": 0.2354,
      "step": 11633
    },
    {
      "epoch": 5.474823529411765,
      "grad_norm": 0.38283755867088526,
      "learning_rate": 7.90714063695041e-06,
      "loss": 0.1836,
      "step": 11634
    },
    {
      "epoch": 5.475294117647059,
      "grad_norm": 0.44405220010022567,
      "learning_rate": 7.893102032572174e-06,
      "loss": 0.1836,
      "step": 11635
    },
    {
      "epoch": 5.475764705882353,
      "grad_norm": 0.5001239805335508,
      "learning_rate": 7.879075564820025e-06,
      "loss": 0.2598,
      "step": 11636
    },
    {
      "epoch": 5.476235294117647,
      "grad_norm": 0.294447803626466,
      "learning_rate": 7.86506123489194e-06,
      "loss": 0.0659,
      "step": 11637
    },
    {
      "epoch": 5.476705882352941,
      "grad_norm": 0.2747628195477728,
      "learning_rate": 7.85105904398481e-06,
      "loss": 0.2207,
      "step": 11638
    },
    {
      "epoch": 5.477176470588235,
      "grad_norm": 0.2930774363578422,
      "learning_rate": 7.837068993294443e-06,
      "loss": 0.2295,
      "step": 11639
    },
    {
      "epoch": 5.477647058823529,
      "grad_norm": 0.29785263039944637,
      "learning_rate": 7.823091084015687e-06,
      "loss": 0.1138,
      "step": 11640
    },
    {
      "epoch": 5.4781176470588235,
      "grad_norm": 0.38606003069279415,
      "learning_rate": 7.809125317342369e-06,
      "loss": 0.3496,
      "step": 11641
    },
    {
      "epoch": 5.478588235294118,
      "grad_norm": 0.4082676830685059,
      "learning_rate": 7.79517169446715e-06,
      "loss": 0.1982,
      "step": 11642
    },
    {
      "epoch": 5.479058823529412,
      "grad_norm": 0.2449844257609908,
      "learning_rate": 7.781230216581779e-06,
      "loss": 0.0432,
      "step": 11643
    },
    {
      "epoch": 5.479529411764706,
      "grad_norm": 0.578791714417351,
      "learning_rate": 7.767300884876971e-06,
      "loss": 0.252,
      "step": 11644
    },
    {
      "epoch": 5.48,
      "grad_norm": 0.5506693587455088,
      "learning_rate": 7.753383700542271e-06,
      "loss": 0.3086,
      "step": 11645
    },
    {
      "epoch": 5.480470588235294,
      "grad_norm": 0.3554145339086195,
      "learning_rate": 7.739478664766347e-06,
      "loss": 0.2168,
      "step": 11646
    },
    {
      "epoch": 5.480941176470588,
      "grad_norm": 0.5305723780860346,
      "learning_rate": 7.725585778736698e-06,
      "loss": 0.377,
      "step": 11647
    },
    {
      "epoch": 5.481411764705882,
      "grad_norm": 0.45702802181908586,
      "learning_rate": 7.711705043639893e-06,
      "loss": 0.2754,
      "step": 11648
    },
    {
      "epoch": 5.481882352941176,
      "grad_norm": 0.19845176346944002,
      "learning_rate": 7.697836460661399e-06,
      "loss": 0.1201,
      "step": 11649
    },
    {
      "epoch": 5.482352941176471,
      "grad_norm": 0.5835499640335476,
      "learning_rate": 7.683980030985654e-06,
      "loss": 0.2812,
      "step": 11650
    },
    {
      "epoch": 5.482823529411765,
      "grad_norm": 0.1707542306654601,
      "learning_rate": 7.670135755796074e-06,
      "loss": 0.0264,
      "step": 11651
    },
    {
      "epoch": 5.483294117647059,
      "grad_norm": 0.3025080750029534,
      "learning_rate": 7.656303636275036e-06,
      "loss": 0.103,
      "step": 11652
    },
    {
      "epoch": 5.483764705882353,
      "grad_norm": 0.39372580108283867,
      "learning_rate": 7.64248367360384e-06,
      "loss": 0.4043,
      "step": 11653
    },
    {
      "epoch": 5.484235294117647,
      "grad_norm": 0.3073448261267174,
      "learning_rate": 7.6286758689627915e-06,
      "loss": 0.1035,
      "step": 11654
    },
    {
      "epoch": 5.484705882352941,
      "grad_norm": 0.28759802864109757,
      "learning_rate": 7.614880223531183e-06,
      "loss": 0.1162,
      "step": 11655
    },
    {
      "epoch": 5.485176470588235,
      "grad_norm": 0.3425304858160157,
      "learning_rate": 7.6010967384871725e-06,
      "loss": 0.2422,
      "step": 11656
    },
    {
      "epoch": 5.485647058823529,
      "grad_norm": 0.3906390672412982,
      "learning_rate": 7.587325415007967e-06,
      "loss": 0.3867,
      "step": 11657
    },
    {
      "epoch": 5.4861176470588235,
      "grad_norm": 0.27902674245493475,
      "learning_rate": 7.573566254269708e-06,
      "loss": 0.2275,
      "step": 11658
    },
    {
      "epoch": 5.486588235294118,
      "grad_norm": 0.35520000634576715,
      "learning_rate": 7.559819257447492e-06,
      "loss": 0.2021,
      "step": 11659
    },
    {
      "epoch": 5.487058823529412,
      "grad_norm": 0.46573996969028425,
      "learning_rate": 7.546084425715359e-06,
      "loss": 0.5273,
      "step": 11660
    },
    {
      "epoch": 5.487529411764706,
      "grad_norm": 0.4409848318847126,
      "learning_rate": 7.532361760246375e-06,
      "loss": 0.3164,
      "step": 11661
    },
    {
      "epoch": 5.4879999999999995,
      "grad_norm": 0.5732032036767993,
      "learning_rate": 7.5186512622124495e-06,
      "loss": 0.3203,
      "step": 11662
    },
    {
      "epoch": 5.488470588235294,
      "grad_norm": 0.37037089698652637,
      "learning_rate": 7.5049529327846135e-06,
      "loss": 0.4922,
      "step": 11663
    },
    {
      "epoch": 5.488941176470588,
      "grad_norm": 0.27069178923149856,
      "learning_rate": 7.491266773132715e-06,
      "loss": 0.085,
      "step": 11664
    },
    {
      "epoch": 5.489411764705882,
      "grad_norm": 0.2125963600172695,
      "learning_rate": 7.477592784425634e-06,
      "loss": 0.0835,
      "step": 11665
    },
    {
      "epoch": 5.489882352941176,
      "grad_norm": 0.35455625073458574,
      "learning_rate": 7.463930967831222e-06,
      "loss": 0.293,
      "step": 11666
    },
    {
      "epoch": 5.490352941176471,
      "grad_norm": 0.0861527360144261,
      "learning_rate": 7.450281324516228e-06,
      "loss": 0.0053,
      "step": 11667
    },
    {
      "epoch": 5.490823529411765,
      "grad_norm": 0.4997810402785268,
      "learning_rate": 7.436643855646402e-06,
      "loss": 0.2109,
      "step": 11668
    },
    {
      "epoch": 5.491294117647059,
      "grad_norm": 0.2945217820384707,
      "learning_rate": 7.423018562386479e-06,
      "loss": 0.1396,
      "step": 11669
    },
    {
      "epoch": 5.491764705882353,
      "grad_norm": 0.613117172558229,
      "learning_rate": 7.40940544590008e-06,
      "loss": 0.3574,
      "step": 11670
    },
    {
      "epoch": 5.492235294117647,
      "grad_norm": 0.3558972626717149,
      "learning_rate": 7.395804507349873e-06,
      "loss": 0.082,
      "step": 11671
    },
    {
      "epoch": 5.492705882352941,
      "grad_norm": 0.27815712562156586,
      "learning_rate": 7.3822157478974465e-06,
      "loss": 0.0771,
      "step": 11672
    },
    {
      "epoch": 5.493176470588235,
      "grad_norm": 0.5829918279448353,
      "learning_rate": 7.368639168703306e-06,
      "loss": 0.6602,
      "step": 11673
    },
    {
      "epoch": 5.493647058823529,
      "grad_norm": 0.29542349075079005,
      "learning_rate": 7.3550747709269885e-06,
      "loss": 0.1709,
      "step": 11674
    },
    {
      "epoch": 5.4941176470588236,
      "grad_norm": 0.3903089993137159,
      "learning_rate": 7.34152255572697e-06,
      "loss": 0.2139,
      "step": 11675
    },
    {
      "epoch": 5.494588235294118,
      "grad_norm": 0.45110508260631205,
      "learning_rate": 7.327982524260639e-06,
      "loss": 0.1216,
      "step": 11676
    },
    {
      "epoch": 5.495058823529412,
      "grad_norm": 0.5166267093474692,
      "learning_rate": 7.314454677684439e-06,
      "loss": 0.2578,
      "step": 11677
    },
    {
      "epoch": 5.495529411764706,
      "grad_norm": 0.5487921358239546,
      "learning_rate": 7.300939017153645e-06,
      "loss": 0.3906,
      "step": 11678
    },
    {
      "epoch": 5.496,
      "grad_norm": 0.4688909207446494,
      "learning_rate": 7.287435543822584e-06,
      "loss": 0.3047,
      "step": 11679
    },
    {
      "epoch": 5.496470588235294,
      "grad_norm": 0.5451263682379479,
      "learning_rate": 7.2739442588445666e-06,
      "loss": 0.3867,
      "step": 11680
    },
    {
      "epoch": 5.496941176470588,
      "grad_norm": 0.39029670598528765,
      "learning_rate": 7.260465163371754e-06,
      "loss": 0.4141,
      "step": 11681
    },
    {
      "epoch": 5.497411764705882,
      "grad_norm": 0.517412839096225,
      "learning_rate": 7.246998258555342e-06,
      "loss": 0.1748,
      "step": 11682
    },
    {
      "epoch": 5.497882352941176,
      "grad_norm": 0.3634222367311911,
      "learning_rate": 7.233543545545511e-06,
      "loss": 0.2617,
      "step": 11683
    },
    {
      "epoch": 5.498352941176471,
      "grad_norm": 0.41707859450590395,
      "learning_rate": 7.220101025491293e-06,
      "loss": 0.4062,
      "step": 11684
    },
    {
      "epoch": 5.498823529411765,
      "grad_norm": 0.455876445302377,
      "learning_rate": 7.2066706995407845e-06,
      "loss": 0.4551,
      "step": 11685
    },
    {
      "epoch": 5.499294117647059,
      "grad_norm": 0.29619254497537595,
      "learning_rate": 7.193252568841018e-06,
      "loss": 0.1182,
      "step": 11686
    },
    {
      "epoch": 5.499764705882353,
      "grad_norm": 0.22963800867988685,
      "learning_rate": 7.1798466345379115e-06,
      "loss": 0.0806,
      "step": 11687
    },
    {
      "epoch": 5.5002352941176476,
      "grad_norm": 0.3483102034436172,
      "learning_rate": 7.166452897776448e-06,
      "loss": 0.1904,
      "step": 11688
    },
    {
      "epoch": 5.500705882352941,
      "grad_norm": 0.3723782729290471,
      "learning_rate": 7.153071359700513e-06,
      "loss": 0.105,
      "step": 11689
    },
    {
      "epoch": 5.501176470588235,
      "grad_norm": 0.19051638698949344,
      "learning_rate": 7.1397020214529255e-06,
      "loss": 0.0405,
      "step": 11690
    },
    {
      "epoch": 5.501647058823529,
      "grad_norm": 0.1809318486269737,
      "learning_rate": 7.126344884175522e-06,
      "loss": 0.0265,
      "step": 11691
    },
    {
      "epoch": 5.502117647058824,
      "grad_norm": 0.28403824436349395,
      "learning_rate": 7.1129999490090735e-06,
      "loss": 0.3047,
      "step": 11692
    },
    {
      "epoch": 5.502588235294118,
      "grad_norm": 0.5057632785530825,
      "learning_rate": 7.099667217093252e-06,
      "loss": 0.2334,
      "step": 11693
    },
    {
      "epoch": 5.503058823529412,
      "grad_norm": 0.38202467056686923,
      "learning_rate": 7.08634668956678e-06,
      "loss": 0.3457,
      "step": 11694
    },
    {
      "epoch": 5.503529411764706,
      "grad_norm": 0.37189237804470265,
      "learning_rate": 7.073038367567313e-06,
      "loss": 0.1865,
      "step": 11695
    },
    {
      "epoch": 5.504,
      "grad_norm": 0.32508677725356655,
      "learning_rate": 7.059742252231393e-06,
      "loss": 0.1064,
      "step": 11696
    },
    {
      "epoch": 5.504470588235294,
      "grad_norm": 0.463594607242654,
      "learning_rate": 7.046458344694628e-06,
      "loss": 0.4648,
      "step": 11697
    },
    {
      "epoch": 5.504941176470588,
      "grad_norm": 0.28841649154981885,
      "learning_rate": 7.033186646091477e-06,
      "loss": 0.2275,
      "step": 11698
    },
    {
      "epoch": 5.505411764705882,
      "grad_norm": 0.44039957888467346,
      "learning_rate": 7.019927157555449e-06,
      "loss": 0.4004,
      "step": 11699
    },
    {
      "epoch": 5.5058823529411764,
      "grad_norm": 0.4725017855852475,
      "learning_rate": 7.006679880218974e-06,
      "loss": 0.4727,
      "step": 11700
    },
    {
      "epoch": 5.506352941176471,
      "grad_norm": 0.3652023479791296,
      "learning_rate": 6.993444815213378e-06,
      "loss": 0.1475,
      "step": 11701
    },
    {
      "epoch": 5.506823529411765,
      "grad_norm": 0.415285886522985,
      "learning_rate": 6.9802219636690585e-06,
      "loss": 0.1279,
      "step": 11702
    },
    {
      "epoch": 5.507294117647059,
      "grad_norm": 0.30606466064567234,
      "learning_rate": 6.967011326715294e-06,
      "loss": 0.1631,
      "step": 11703
    },
    {
      "epoch": 5.507764705882353,
      "grad_norm": 0.2330692876034343,
      "learning_rate": 6.953812905480316e-06,
      "loss": 0.0479,
      "step": 11704
    },
    {
      "epoch": 5.508235294117647,
      "grad_norm": 0.16770983427572433,
      "learning_rate": 6.940626701091356e-06,
      "loss": 0.0623,
      "step": 11705
    },
    {
      "epoch": 5.508705882352941,
      "grad_norm": 0.22409969611765077,
      "learning_rate": 6.927452714674614e-06,
      "loss": 0.0459,
      "step": 11706
    },
    {
      "epoch": 5.509176470588235,
      "grad_norm": 0.4556625794571588,
      "learning_rate": 6.914290947355122e-06,
      "loss": 0.3027,
      "step": 11707
    },
    {
      "epoch": 5.509647058823529,
      "grad_norm": 0.3981990632071252,
      "learning_rate": 6.901141400257048e-06,
      "loss": 0.2578,
      "step": 11708
    },
    {
      "epoch": 5.510117647058824,
      "grad_norm": 0.4437660859471274,
      "learning_rate": 6.888004074503395e-06,
      "loss": 0.5156,
      "step": 11709
    },
    {
      "epoch": 5.510588235294118,
      "grad_norm": 0.44088021111681647,
      "learning_rate": 6.87487897121613e-06,
      "loss": 0.2383,
      "step": 11710
    },
    {
      "epoch": 5.511058823529412,
      "grad_norm": 0.37463659788377285,
      "learning_rate": 6.861766091516274e-06,
      "loss": 0.1328,
      "step": 11711
    },
    {
      "epoch": 5.511529411764706,
      "grad_norm": 0.3497087613374636,
      "learning_rate": 6.848665436523631e-06,
      "loss": 0.2393,
      "step": 11712
    },
    {
      "epoch": 5.5120000000000005,
      "grad_norm": 0.6303810314713472,
      "learning_rate": 6.83557700735714e-06,
      "loss": 0.6758,
      "step": 11713
    },
    {
      "epoch": 5.512470588235294,
      "grad_norm": 0.499187092995651,
      "learning_rate": 6.82250080513459e-06,
      "loss": 0.6289,
      "step": 11714
    },
    {
      "epoch": 5.512941176470588,
      "grad_norm": 0.4215409286527045,
      "learning_rate": 6.809436830972753e-06,
      "loss": 0.1895,
      "step": 11715
    },
    {
      "epoch": 5.513411764705882,
      "grad_norm": 0.3304957983560556,
      "learning_rate": 6.796385085987338e-06,
      "loss": 0.3555,
      "step": 11716
    },
    {
      "epoch": 5.5138823529411765,
      "grad_norm": 0.3942075907820713,
      "learning_rate": 6.783345571293069e-06,
      "loss": 0.1089,
      "step": 11717
    },
    {
      "epoch": 5.514352941176471,
      "grad_norm": 0.28165252993861817,
      "learning_rate": 6.770318288003556e-06,
      "loss": 0.0864,
      "step": 11718
    },
    {
      "epoch": 5.514823529411765,
      "grad_norm": 0.37342045379010447,
      "learning_rate": 6.757303237231393e-06,
      "loss": 0.5,
      "step": 11719
    },
    {
      "epoch": 5.515294117647059,
      "grad_norm": 0.31732178641626246,
      "learning_rate": 6.744300420088156e-06,
      "loss": 0.1445,
      "step": 11720
    },
    {
      "epoch": 5.5157647058823525,
      "grad_norm": 0.4238202411060884,
      "learning_rate": 6.73130983768429e-06,
      "loss": 0.4492,
      "step": 11721
    },
    {
      "epoch": 5.516235294117647,
      "grad_norm": 0.24371292929253252,
      "learning_rate": 6.718331491129309e-06,
      "loss": 0.1582,
      "step": 11722
    },
    {
      "epoch": 5.516705882352941,
      "grad_norm": 0.4382342308436423,
      "learning_rate": 6.705365381531641e-06,
      "loss": 0.2812,
      "step": 11723
    },
    {
      "epoch": 5.517176470588235,
      "grad_norm": 0.3543436951993449,
      "learning_rate": 6.6924115099985856e-06,
      "loss": 0.1465,
      "step": 11724
    },
    {
      "epoch": 5.517647058823529,
      "grad_norm": 0.40577271194480335,
      "learning_rate": 6.679469877636523e-06,
      "loss": 0.1973,
      "step": 11725
    },
    {
      "epoch": 5.518117647058824,
      "grad_norm": 0.1947488459994186,
      "learning_rate": 6.6665404855507195e-06,
      "loss": 0.0332,
      "step": 11726
    },
    {
      "epoch": 5.518588235294118,
      "grad_norm": 0.2410285181891581,
      "learning_rate": 6.6536233348453925e-06,
      "loss": 0.1807,
      "step": 11727
    },
    {
      "epoch": 5.519058823529412,
      "grad_norm": 0.17505654361914785,
      "learning_rate": 6.640718426623759e-06,
      "loss": 0.027,
      "step": 11728
    },
    {
      "epoch": 5.519529411764706,
      "grad_norm": 0.31289671706032945,
      "learning_rate": 6.627825761987937e-06,
      "loss": 0.127,
      "step": 11729
    },
    {
      "epoch": 5.52,
      "grad_norm": 0.7073024044098456,
      "learning_rate": 6.61494534203903e-06,
      "loss": 0.373,
      "step": 11730
    },
    {
      "epoch": 5.520470588235294,
      "grad_norm": 0.3653065735376373,
      "learning_rate": 6.6020771678771066e-06,
      "loss": 0.1289,
      "step": 11731
    },
    {
      "epoch": 5.520941176470588,
      "grad_norm": 0.4679892308708975,
      "learning_rate": 6.589221240601123e-06,
      "loss": 0.3184,
      "step": 11732
    },
    {
      "epoch": 5.521411764705882,
      "grad_norm": 0.39231781442154423,
      "learning_rate": 6.576377561309082e-06,
      "loss": 0.3223,
      "step": 11733
    },
    {
      "epoch": 5.5218823529411765,
      "grad_norm": 0.3500461407702073,
      "learning_rate": 6.563546131097907e-06,
      "loss": 0.3301,
      "step": 11734
    },
    {
      "epoch": 5.522352941176471,
      "grad_norm": 0.38621323308101707,
      "learning_rate": 6.550726951063423e-06,
      "loss": 0.334,
      "step": 11735
    },
    {
      "epoch": 5.522823529411765,
      "grad_norm": 0.4097164763308152,
      "learning_rate": 6.5379200223004855e-06,
      "loss": 0.3125,
      "step": 11736
    },
    {
      "epoch": 5.523294117647059,
      "grad_norm": 0.5963356793962281,
      "learning_rate": 6.5251253459028545e-06,
      "loss": 0.625,
      "step": 11737
    },
    {
      "epoch": 5.523764705882353,
      "grad_norm": 0.2921304759521902,
      "learning_rate": 6.512342922963254e-06,
      "loss": 0.104,
      "step": 11738
    },
    {
      "epoch": 5.524235294117647,
      "grad_norm": 0.5754267299315243,
      "learning_rate": 6.499572754573362e-06,
      "loss": 0.4277,
      "step": 11739
    },
    {
      "epoch": 5.524705882352941,
      "grad_norm": 0.26608479014679504,
      "learning_rate": 6.48681484182384e-06,
      "loss": 0.2275,
      "step": 11740
    },
    {
      "epoch": 5.525176470588235,
      "grad_norm": 0.39668033885012866,
      "learning_rate": 6.474069185804248e-06,
      "loss": 0.1484,
      "step": 11741
    },
    {
      "epoch": 5.525647058823529,
      "grad_norm": 0.3249591538498643,
      "learning_rate": 6.461335787603134e-06,
      "loss": 0.2441,
      "step": 11742
    },
    {
      "epoch": 5.526117647058824,
      "grad_norm": 0.5544513821513543,
      "learning_rate": 6.448614648308009e-06,
      "loss": 0.4883,
      "step": 11743
    },
    {
      "epoch": 5.526588235294118,
      "grad_norm": 0.3861178679460654,
      "learning_rate": 6.435905769005306e-06,
      "loss": 0.1279,
      "step": 11744
    },
    {
      "epoch": 5.527058823529412,
      "grad_norm": 0.4788672630624457,
      "learning_rate": 6.423209150780423e-06,
      "loss": 0.4492,
      "step": 11745
    },
    {
      "epoch": 5.527529411764706,
      "grad_norm": 0.3778127384471701,
      "learning_rate": 6.4105247947177256e-06,
      "loss": 0.2617,
      "step": 11746
    },
    {
      "epoch": 5.5280000000000005,
      "grad_norm": 0.3006318146060473,
      "learning_rate": 6.3978527019004964e-06,
      "loss": 0.2988,
      "step": 11747
    },
    {
      "epoch": 5.528470588235294,
      "grad_norm": 0.23730331865089316,
      "learning_rate": 6.385192873411038e-06,
      "loss": 0.1748,
      "step": 11748
    },
    {
      "epoch": 5.528941176470588,
      "grad_norm": 0.4789755385515159,
      "learning_rate": 6.372545310330518e-06,
      "loss": 0.2275,
      "step": 11749
    },
    {
      "epoch": 5.529411764705882,
      "grad_norm": 0.24140321448888802,
      "learning_rate": 6.359910013739122e-06,
      "loss": 0.043,
      "step": 11750
    },
    {
      "epoch": 5.5298823529411765,
      "grad_norm": 0.40670820757553994,
      "learning_rate": 6.347286984715938e-06,
      "loss": 0.3926,
      "step": 11751
    },
    {
      "epoch": 5.530352941176471,
      "grad_norm": 0.49244257162964317,
      "learning_rate": 6.334676224339086e-06,
      "loss": 0.1904,
      "step": 11752
    },
    {
      "epoch": 5.530823529411765,
      "grad_norm": 0.2893168526769125,
      "learning_rate": 6.322077733685538e-06,
      "loss": 0.0674,
      "step": 11753
    },
    {
      "epoch": 5.531294117647059,
      "grad_norm": 0.34208054940650773,
      "learning_rate": 6.309491513831317e-06,
      "loss": 0.0977,
      "step": 11754
    },
    {
      "epoch": 5.5317647058823525,
      "grad_norm": 0.37806822545477525,
      "learning_rate": 6.296917565851295e-06,
      "loss": 0.063,
      "step": 11755
    },
    {
      "epoch": 5.532235294117647,
      "grad_norm": 0.44030690682008894,
      "learning_rate": 6.284355890819398e-06,
      "loss": 0.2832,
      "step": 11756
    },
    {
      "epoch": 5.532705882352941,
      "grad_norm": 0.6966145992072871,
      "learning_rate": 6.271806489808401e-06,
      "loss": 0.5859,
      "step": 11757
    },
    {
      "epoch": 5.533176470588235,
      "grad_norm": 0.4668557600415663,
      "learning_rate": 6.2592693638901125e-06,
      "loss": 0.6641,
      "step": 11758
    },
    {
      "epoch": 5.533647058823529,
      "grad_norm": 0.3548724342648851,
      "learning_rate": 6.246744514135293e-06,
      "loss": 0.2285,
      "step": 11759
    },
    {
      "epoch": 5.534117647058824,
      "grad_norm": 0.39604433008833856,
      "learning_rate": 6.234231941613571e-06,
      "loss": 0.2598,
      "step": 11760
    },
    {
      "epoch": 5.534588235294118,
      "grad_norm": 0.42823613676647004,
      "learning_rate": 6.221731647393607e-06,
      "loss": 0.1494,
      "step": 11761
    },
    {
      "epoch": 5.535058823529412,
      "grad_norm": 0.5411096126410393,
      "learning_rate": 6.209243632543015e-06,
      "loss": 0.2852,
      "step": 11762
    },
    {
      "epoch": 5.535529411764706,
      "grad_norm": 0.4206804389664036,
      "learning_rate": 6.196767898128291e-06,
      "loss": 0.3262,
      "step": 11763
    },
    {
      "epoch": 5.536,
      "grad_norm": 0.33210952477285555,
      "learning_rate": 6.184304445214932e-06,
      "loss": 0.1021,
      "step": 11764
    },
    {
      "epoch": 5.536470588235294,
      "grad_norm": 0.28620114526999235,
      "learning_rate": 6.171853274867406e-06,
      "loss": 0.0942,
      "step": 11765
    },
    {
      "epoch": 5.536941176470588,
      "grad_norm": 0.35881470831414713,
      "learning_rate": 6.159414388149059e-06,
      "loss": 0.2773,
      "step": 11766
    },
    {
      "epoch": 5.537411764705882,
      "grad_norm": 0.39414427207675157,
      "learning_rate": 6.146987786122259e-06,
      "loss": 0.1885,
      "step": 11767
    },
    {
      "epoch": 5.5378823529411765,
      "grad_norm": 0.4283742082905368,
      "learning_rate": 6.134573469848325e-06,
      "loss": 0.1357,
      "step": 11768
    },
    {
      "epoch": 5.538352941176471,
      "grad_norm": 0.6566847330067924,
      "learning_rate": 6.1221714403874415e-06,
      "loss": 0.5078,
      "step": 11769
    },
    {
      "epoch": 5.538823529411765,
      "grad_norm": 0.5221445163999253,
      "learning_rate": 6.109781698798843e-06,
      "loss": 0.7734,
      "step": 11770
    },
    {
      "epoch": 5.539294117647059,
      "grad_norm": 0.17911573299019923,
      "learning_rate": 6.097404246140669e-06,
      "loss": 0.0322,
      "step": 11771
    },
    {
      "epoch": 5.539764705882353,
      "grad_norm": 0.39325036485249726,
      "learning_rate": 6.085039083470006e-06,
      "loss": 0.3027,
      "step": 11772
    },
    {
      "epoch": 5.540235294117647,
      "grad_norm": 0.21119606031034555,
      "learning_rate": 6.072686211842909e-06,
      "loss": 0.0447,
      "step": 11773
    },
    {
      "epoch": 5.540705882352941,
      "grad_norm": 0.3063686948190594,
      "learning_rate": 6.060345632314384e-06,
      "loss": 0.2949,
      "step": 11774
    },
    {
      "epoch": 5.541176470588235,
      "grad_norm": 0.2788613296443602,
      "learning_rate": 6.048017345938339e-06,
      "loss": 0.293,
      "step": 11775
    },
    {
      "epoch": 5.541647058823529,
      "grad_norm": 0.6438968340278611,
      "learning_rate": 6.035701353767713e-06,
      "loss": 0.2754,
      "step": 11776
    },
    {
      "epoch": 5.542117647058824,
      "grad_norm": 0.4785435582538159,
      "learning_rate": 6.023397656854301e-06,
      "loss": 0.2715,
      "step": 11777
    },
    {
      "epoch": 5.542588235294118,
      "grad_norm": 0.43131308698225773,
      "learning_rate": 6.011106256248943e-06,
      "loss": 0.1279,
      "step": 11778
    },
    {
      "epoch": 5.543058823529412,
      "grad_norm": 0.3324997729336149,
      "learning_rate": 5.998827153001401e-06,
      "loss": 0.2793,
      "step": 11779
    },
    {
      "epoch": 5.543529411764706,
      "grad_norm": 0.25715601861183496,
      "learning_rate": 5.986560348160302e-06,
      "loss": 0.063,
      "step": 11780
    },
    {
      "epoch": 5.5440000000000005,
      "grad_norm": 0.3313652663090593,
      "learning_rate": 5.9743058427733414e-06,
      "loss": 0.2041,
      "step": 11781
    },
    {
      "epoch": 5.544470588235294,
      "grad_norm": 0.2791332898168402,
      "learning_rate": 5.9620636378871325e-06,
      "loss": 0.25,
      "step": 11782
    },
    {
      "epoch": 5.544941176470588,
      "grad_norm": 0.4660606730281734,
      "learning_rate": 5.9498337345471544e-06,
      "loss": 0.1318,
      "step": 11783
    },
    {
      "epoch": 5.545411764705882,
      "grad_norm": 0.5764545119748715,
      "learning_rate": 5.93761613379794e-06,
      "loss": 0.5938,
      "step": 11784
    },
    {
      "epoch": 5.5458823529411765,
      "grad_norm": 0.35325765195004205,
      "learning_rate": 5.925410836682953e-06,
      "loss": 0.2285,
      "step": 11785
    },
    {
      "epoch": 5.546352941176471,
      "grad_norm": 0.5678954224404853,
      "learning_rate": 5.913217844244545e-06,
      "loss": 0.3633,
      "step": 11786
    },
    {
      "epoch": 5.546823529411765,
      "grad_norm": 0.31441365636125357,
      "learning_rate": 5.90103715752408e-06,
      "loss": 0.2188,
      "step": 11787
    },
    {
      "epoch": 5.547294117647059,
      "grad_norm": 0.6849217876533463,
      "learning_rate": 5.888868777561862e-06,
      "loss": 0.4316,
      "step": 11788
    },
    {
      "epoch": 5.5477647058823525,
      "grad_norm": 0.5010329088333623,
      "learning_rate": 5.8767127053970745e-06,
      "loss": 0.3184,
      "step": 11789
    },
    {
      "epoch": 5.548235294117647,
      "grad_norm": 0.2566498580619434,
      "learning_rate": 5.864568942067954e-06,
      "loss": 0.064,
      "step": 11790
    },
    {
      "epoch": 5.548705882352941,
      "grad_norm": 0.34844253117201135,
      "learning_rate": 5.852437488611656e-06,
      "loss": 0.1074,
      "step": 11791
    },
    {
      "epoch": 5.549176470588235,
      "grad_norm": 0.42666744431977444,
      "learning_rate": 5.840318346064199e-06,
      "loss": 0.1641,
      "step": 11792
    },
    {
      "epoch": 5.549647058823529,
      "grad_norm": 0.36115728622020626,
      "learning_rate": 5.828211515460673e-06,
      "loss": 0.2334,
      "step": 11793
    },
    {
      "epoch": 5.550117647058824,
      "grad_norm": 0.24747911725051616,
      "learning_rate": 5.816116997835052e-06,
      "loss": 0.0618,
      "step": 11794
    },
    {
      "epoch": 5.550588235294118,
      "grad_norm": 0.36473225407737997,
      "learning_rate": 5.804034794220208e-06,
      "loss": 0.1641,
      "step": 11795
    },
    {
      "epoch": 5.551058823529412,
      "grad_norm": 0.26032520517994073,
      "learning_rate": 5.791964905648116e-06,
      "loss": 0.064,
      "step": 11796
    },
    {
      "epoch": 5.551529411764706,
      "grad_norm": 0.48877166871453814,
      "learning_rate": 5.779907333149553e-06,
      "loss": 0.2256,
      "step": 11797
    },
    {
      "epoch": 5.552,
      "grad_norm": 0.33654484902163123,
      "learning_rate": 5.767862077754276e-06,
      "loss": 0.0718,
      "step": 11798
    },
    {
      "epoch": 5.552470588235294,
      "grad_norm": 0.3698137315882187,
      "learning_rate": 5.755829140491064e-06,
      "loss": 0.1748,
      "step": 11799
    },
    {
      "epoch": 5.552941176470588,
      "grad_norm": 0.45115610047800064,
      "learning_rate": 5.743808522387544e-06,
      "loss": 0.4062,
      "step": 11800
    },
    {
      "epoch": 5.553411764705882,
      "grad_norm": 0.4999751968980209,
      "learning_rate": 5.7318002244703285e-06,
      "loss": 0.1865,
      "step": 11801
    },
    {
      "epoch": 5.5538823529411765,
      "grad_norm": 0.4963116627634152,
      "learning_rate": 5.7198042477650475e-06,
      "loss": 0.498,
      "step": 11802
    },
    {
      "epoch": 5.554352941176471,
      "grad_norm": 0.2776738125636946,
      "learning_rate": 5.7078205932961476e-06,
      "loss": 0.1182,
      "step": 11803
    },
    {
      "epoch": 5.554823529411765,
      "grad_norm": 0.48297494895138116,
      "learning_rate": 5.69584926208711e-06,
      "loss": 0.2715,
      "step": 11804
    },
    {
      "epoch": 5.555294117647059,
      "grad_norm": 0.3518504813551608,
      "learning_rate": 5.683890255160367e-06,
      "loss": 0.1338,
      "step": 11805
    },
    {
      "epoch": 5.555764705882353,
      "grad_norm": 0.34976941956114843,
      "learning_rate": 5.671943573537269e-06,
      "loss": 0.082,
      "step": 11806
    },
    {
      "epoch": 5.556235294117647,
      "grad_norm": 0.4941221789610682,
      "learning_rate": 5.6600092182381156e-06,
      "loss": 0.4785,
      "step": 11807
    },
    {
      "epoch": 5.556705882352941,
      "grad_norm": 0.45220228314901967,
      "learning_rate": 5.648087190282141e-06,
      "loss": 0.3086,
      "step": 11808
    },
    {
      "epoch": 5.557176470588235,
      "grad_norm": 0.4457027034089837,
      "learning_rate": 5.636177490687565e-06,
      "loss": 0.4082,
      "step": 11809
    },
    {
      "epoch": 5.557647058823529,
      "grad_norm": 0.46990898806357795,
      "learning_rate": 5.624280120471542e-06,
      "loss": 0.3672,
      "step": 11810
    },
    {
      "epoch": 5.558117647058824,
      "grad_norm": 0.3666805279621222,
      "learning_rate": 5.612395080650156e-06,
      "loss": 0.1357,
      "step": 11811
    },
    {
      "epoch": 5.558588235294118,
      "grad_norm": 0.3566260042670088,
      "learning_rate": 5.600522372238431e-06,
      "loss": 0.3184,
      "step": 11812
    },
    {
      "epoch": 5.559058823529412,
      "grad_norm": 0.4569940267315901,
      "learning_rate": 5.588661996250387e-06,
      "loss": 0.3535,
      "step": 11813
    },
    {
      "epoch": 5.559529411764705,
      "grad_norm": 0.5721478258047755,
      "learning_rate": 5.576813953698933e-06,
      "loss": 0.7617,
      "step": 11814
    },
    {
      "epoch": 5.5600000000000005,
      "grad_norm": 0.2302169964590412,
      "learning_rate": 5.564978245595958e-06,
      "loss": 0.0967,
      "step": 11815
    },
    {
      "epoch": 5.560470588235294,
      "grad_norm": 0.46967330666985596,
      "learning_rate": 5.553154872952303e-06,
      "loss": 0.2021,
      "step": 11816
    },
    {
      "epoch": 5.560941176470588,
      "grad_norm": 0.3019563875937552,
      "learning_rate": 5.541343836777729e-06,
      "loss": 0.0625,
      "step": 11817
    },
    {
      "epoch": 5.561411764705882,
      "grad_norm": 0.5479966099379325,
      "learning_rate": 5.529545138080943e-06,
      "loss": 0.3789,
      "step": 11818
    },
    {
      "epoch": 5.5618823529411765,
      "grad_norm": 0.5375943307660577,
      "learning_rate": 5.517758777869641e-06,
      "loss": 0.4648,
      "step": 11819
    },
    {
      "epoch": 5.562352941176471,
      "grad_norm": 0.38541834214117027,
      "learning_rate": 5.5059847571504e-06,
      "loss": 0.1309,
      "step": 11820
    },
    {
      "epoch": 5.562823529411765,
      "grad_norm": 0.3939753545239113,
      "learning_rate": 5.494223076928817e-06,
      "loss": 0.1689,
      "step": 11821
    },
    {
      "epoch": 5.563294117647059,
      "grad_norm": 0.4579224572581624,
      "learning_rate": 5.482473738209386e-06,
      "loss": 0.377,
      "step": 11822
    },
    {
      "epoch": 5.5637647058823525,
      "grad_norm": 0.44222826144522603,
      "learning_rate": 5.470736741995541e-06,
      "loss": 0.4238,
      "step": 11823
    },
    {
      "epoch": 5.564235294117647,
      "grad_norm": 0.3254453128086159,
      "learning_rate": 5.4590120892896775e-06,
      "loss": 0.1641,
      "step": 11824
    },
    {
      "epoch": 5.564705882352941,
      "grad_norm": 0.3919499020287127,
      "learning_rate": 5.447299781093178e-06,
      "loss": 0.2617,
      "step": 11825
    },
    {
      "epoch": 5.565176470588235,
      "grad_norm": 0.3547638418558188,
      "learning_rate": 5.435599818406278e-06,
      "loss": 0.3105,
      "step": 11826
    },
    {
      "epoch": 5.565647058823529,
      "grad_norm": 0.2784002333683158,
      "learning_rate": 5.423912202228275e-06,
      "loss": 0.0664,
      "step": 11827
    },
    {
      "epoch": 5.566117647058824,
      "grad_norm": 0.4624729705042466,
      "learning_rate": 5.412236933557274e-06,
      "loss": 0.2715,
      "step": 11828
    },
    {
      "epoch": 5.566588235294118,
      "grad_norm": 0.4425724862093522,
      "learning_rate": 5.400574013390441e-06,
      "loss": 0.2275,
      "step": 11829
    },
    {
      "epoch": 5.567058823529412,
      "grad_norm": 0.5523164115510325,
      "learning_rate": 5.388923442723864e-06,
      "loss": 0.2754,
      "step": 11830
    },
    {
      "epoch": 5.567529411764706,
      "grad_norm": 0.3767115259341111,
      "learning_rate": 5.37728522255253e-06,
      "loss": 0.2246,
      "step": 11831
    },
    {
      "epoch": 5.568,
      "grad_norm": 0.346746459645705,
      "learning_rate": 5.365659353870394e-06,
      "loss": 0.1396,
      "step": 11832
    },
    {
      "epoch": 5.568470588235294,
      "grad_norm": 0.5276270275395193,
      "learning_rate": 5.354045837670412e-06,
      "loss": 0.3848,
      "step": 11833
    },
    {
      "epoch": 5.568941176470588,
      "grad_norm": 0.4378910633948268,
      "learning_rate": 5.342444674944374e-06,
      "loss": 0.2852,
      "step": 11834
    },
    {
      "epoch": 5.569411764705882,
      "grad_norm": 0.27694120534535505,
      "learning_rate": 5.330855866683087e-06,
      "loss": 0.3008,
      "step": 11835
    },
    {
      "epoch": 5.5698823529411765,
      "grad_norm": 0.36879044815348094,
      "learning_rate": 5.319279413876343e-06,
      "loss": 0.0923,
      "step": 11836
    },
    {
      "epoch": 5.570352941176471,
      "grad_norm": 0.47157937516246545,
      "learning_rate": 5.307715317512767e-06,
      "loss": 0.252,
      "step": 11837
    },
    {
      "epoch": 5.570823529411765,
      "grad_norm": 0.36961194298002825,
      "learning_rate": 5.29616357858002e-06,
      "loss": 0.2988,
      "step": 11838
    },
    {
      "epoch": 5.571294117647059,
      "grad_norm": 0.3190416502368125,
      "learning_rate": 5.2846241980646795e-06,
      "loss": 0.2734,
      "step": 11839
    },
    {
      "epoch": 5.571764705882353,
      "grad_norm": 0.3553419632042993,
      "learning_rate": 5.273097176952224e-06,
      "loss": 0.1396,
      "step": 11840
    },
    {
      "epoch": 5.572235294117647,
      "grad_norm": 0.6465793479133032,
      "learning_rate": 5.261582516227197e-06,
      "loss": 0.3672,
      "step": 11841
    },
    {
      "epoch": 5.572705882352941,
      "grad_norm": 0.30846103682017206,
      "learning_rate": 5.250080216872932e-06,
      "loss": 0.1099,
      "step": 11842
    },
    {
      "epoch": 5.573176470588235,
      "grad_norm": 0.30102123409982057,
      "learning_rate": 5.238590279871824e-06,
      "loss": 0.0972,
      "step": 11843
    },
    {
      "epoch": 5.573647058823529,
      "grad_norm": 0.46808104748208285,
      "learning_rate": 5.227112706205155e-06,
      "loss": 0.3984,
      "step": 11844
    },
    {
      "epoch": 5.574117647058824,
      "grad_norm": 0.29162845240898105,
      "learning_rate": 5.2156474968531595e-06,
      "loss": 0.2793,
      "step": 11845
    },
    {
      "epoch": 5.574588235294118,
      "grad_norm": 0.2945292055579325,
      "learning_rate": 5.204194652795035e-06,
      "loss": 0.249,
      "step": 11846
    },
    {
      "epoch": 5.575058823529412,
      "grad_norm": 0.3924810028000014,
      "learning_rate": 5.1927541750089175e-06,
      "loss": 0.1279,
      "step": 11847
    },
    {
      "epoch": 5.575529411764705,
      "grad_norm": 0.3837689335484996,
      "learning_rate": 5.181326064471858e-06,
      "loss": 0.3105,
      "step": 11848
    },
    {
      "epoch": 5.576,
      "grad_norm": 0.3496642387594282,
      "learning_rate": 5.169910322159876e-06,
      "loss": 0.2754,
      "step": 11849
    },
    {
      "epoch": 5.576470588235294,
      "grad_norm": 0.429890621076331,
      "learning_rate": 5.158506949047975e-06,
      "loss": 0.2295,
      "step": 11850
    },
    {
      "epoch": 5.576941176470588,
      "grad_norm": 0.5081784918531359,
      "learning_rate": 5.147115946109992e-06,
      "loss": 0.5703,
      "step": 11851
    },
    {
      "epoch": 5.577411764705882,
      "grad_norm": 0.5215447030564119,
      "learning_rate": 5.1357373143188e-06,
      "loss": 0.3223,
      "step": 11852
    },
    {
      "epoch": 5.5778823529411765,
      "grad_norm": 0.26846968998630305,
      "learning_rate": 5.124371054646237e-06,
      "loss": 0.0496,
      "step": 11853
    },
    {
      "epoch": 5.578352941176471,
      "grad_norm": 0.4518611357482314,
      "learning_rate": 5.11301716806296e-06,
      "loss": 0.4766,
      "step": 11854
    },
    {
      "epoch": 5.578823529411765,
      "grad_norm": 0.5333449114668886,
      "learning_rate": 5.101675655538695e-06,
      "loss": 0.4883,
      "step": 11855
    },
    {
      "epoch": 5.579294117647059,
      "grad_norm": 0.39428488511748755,
      "learning_rate": 5.090346518042049e-06,
      "loss": 0.3066,
      "step": 11856
    },
    {
      "epoch": 5.5797647058823525,
      "grad_norm": 0.42706280493183846,
      "learning_rate": 5.079029756540598e-06,
      "loss": 0.5,
      "step": 11857
    },
    {
      "epoch": 5.580235294117647,
      "grad_norm": 0.27246504826546497,
      "learning_rate": 5.067725372000836e-06,
      "loss": 0.2949,
      "step": 11858
    },
    {
      "epoch": 5.580705882352941,
      "grad_norm": 0.22293077232122327,
      "learning_rate": 5.056433365388207e-06,
      "loss": 0.0967,
      "step": 11859
    },
    {
      "epoch": 5.581176470588235,
      "grad_norm": 0.4108421929072443,
      "learning_rate": 5.045153737667107e-06,
      "loss": 0.1885,
      "step": 11860
    },
    {
      "epoch": 5.581647058823529,
      "grad_norm": 0.3015036524369878,
      "learning_rate": 5.033886489800898e-06,
      "loss": 0.1118,
      "step": 11861
    },
    {
      "epoch": 5.582117647058824,
      "grad_norm": 0.32280855461256935,
      "learning_rate": 5.022631622751827e-06,
      "loss": 0.1738,
      "step": 11862
    },
    {
      "epoch": 5.582588235294118,
      "grad_norm": 0.6197963178164826,
      "learning_rate": 5.011389137481109e-06,
      "loss": 0.6289,
      "step": 11863
    },
    {
      "epoch": 5.583058823529412,
      "grad_norm": 0.5490827172095561,
      "learning_rate": 5.000159034948958e-06,
      "loss": 0.2354,
      "step": 11864
    },
    {
      "epoch": 5.583529411764706,
      "grad_norm": 0.2596922717950324,
      "learning_rate": 4.988941316114425e-06,
      "loss": 0.0771,
      "step": 11865
    },
    {
      "epoch": 5.584,
      "grad_norm": 0.49589970262638355,
      "learning_rate": 4.9777359819355765e-06,
      "loss": 0.2207,
      "step": 11866
    },
    {
      "epoch": 5.584470588235294,
      "grad_norm": 0.45301833907130734,
      "learning_rate": 4.9665430333694135e-06,
      "loss": 0.3086,
      "step": 11867
    },
    {
      "epoch": 5.584941176470588,
      "grad_norm": 0.04584157551724975,
      "learning_rate": 4.955362471371854e-06,
      "loss": 0.0067,
      "step": 11868
    },
    {
      "epoch": 5.585411764705882,
      "grad_norm": 0.3595171099708506,
      "learning_rate": 4.944194296897785e-06,
      "loss": 0.1562,
      "step": 11869
    },
    {
      "epoch": 5.5858823529411765,
      "grad_norm": 0.6716517088911871,
      "learning_rate": 4.933038510901027e-06,
      "loss": 0.5664,
      "step": 11870
    },
    {
      "epoch": 5.586352941176471,
      "grad_norm": 0.29266228512028464,
      "learning_rate": 4.921895114334301e-06,
      "loss": 0.1621,
      "step": 11871
    },
    {
      "epoch": 5.586823529411765,
      "grad_norm": 0.404505682187031,
      "learning_rate": 4.910764108149346e-06,
      "loss": 0.2119,
      "step": 11872
    },
    {
      "epoch": 5.587294117647059,
      "grad_norm": 0.33247441172628317,
      "learning_rate": 4.899645493296817e-06,
      "loss": 0.2656,
      "step": 11873
    },
    {
      "epoch": 5.587764705882353,
      "grad_norm": 0.39954673608195546,
      "learning_rate": 4.888539270726255e-06,
      "loss": 0.3145,
      "step": 11874
    },
    {
      "epoch": 5.588235294117647,
      "grad_norm": 0.5841560975473729,
      "learning_rate": 4.877445441386218e-06,
      "loss": 0.6992,
      "step": 11875
    },
    {
      "epoch": 5.588705882352941,
      "grad_norm": 0.32509365452541944,
      "learning_rate": 4.866364006224166e-06,
      "loss": 0.1089,
      "step": 11876
    },
    {
      "epoch": 5.589176470588235,
      "grad_norm": 0.4159550014597176,
      "learning_rate": 4.85529496618649e-06,
      "loss": 0.4316,
      "step": 11877
    },
    {
      "epoch": 5.589647058823529,
      "grad_norm": 0.6885701867943795,
      "learning_rate": 4.844238322218568e-06,
      "loss": 0.8789,
      "step": 11878
    },
    {
      "epoch": 5.590117647058824,
      "grad_norm": 0.5986665463170541,
      "learning_rate": 4.833194075264663e-06,
      "loss": 0.4043,
      "step": 11879
    },
    {
      "epoch": 5.590588235294118,
      "grad_norm": 0.4537927084232274,
      "learning_rate": 4.8221622262680336e-06,
      "loss": 0.5234,
      "step": 11880
    },
    {
      "epoch": 5.591058823529412,
      "grad_norm": 0.4615733416362113,
      "learning_rate": 4.811142776170845e-06,
      "loss": 0.1641,
      "step": 11881
    },
    {
      "epoch": 5.591529411764705,
      "grad_norm": 0.7647166261095448,
      "learning_rate": 4.800135725914195e-06,
      "loss": 0.5508,
      "step": 11882
    },
    {
      "epoch": 5.592,
      "grad_norm": 0.32193863734474937,
      "learning_rate": 4.789141076438163e-06,
      "loss": 0.1099,
      "step": 11883
    },
    {
      "epoch": 5.592470588235294,
      "grad_norm": 0.38474904318494874,
      "learning_rate": 4.778158828681733e-06,
      "loss": 0.1631,
      "step": 11884
    },
    {
      "epoch": 5.592941176470588,
      "grad_norm": 0.2450032445867262,
      "learning_rate": 4.767188983582854e-06,
      "loss": 0.1865,
      "step": 11885
    },
    {
      "epoch": 5.593411764705882,
      "grad_norm": 0.446305099090317,
      "learning_rate": 4.75623154207841e-06,
      "loss": 0.3613,
      "step": 11886
    },
    {
      "epoch": 5.5938823529411765,
      "grad_norm": 0.30893795273331254,
      "learning_rate": 4.745286505104201e-06,
      "loss": 0.2461,
      "step": 11887
    },
    {
      "epoch": 5.594352941176471,
      "grad_norm": 0.27302927671656435,
      "learning_rate": 4.734353873594998e-06,
      "loss": 0.0452,
      "step": 11888
    },
    {
      "epoch": 5.594823529411765,
      "grad_norm": 0.4283707463589848,
      "learning_rate": 4.723433648484503e-06,
      "loss": 0.2402,
      "step": 11889
    },
    {
      "epoch": 5.595294117647059,
      "grad_norm": 0.3414628507972623,
      "learning_rate": 4.712525830705338e-06,
      "loss": 0.3359,
      "step": 11890
    },
    {
      "epoch": 5.5957647058823525,
      "grad_norm": 0.5427969745374461,
      "learning_rate": 4.70163042118909e-06,
      "loss": 0.377,
      "step": 11891
    },
    {
      "epoch": 5.596235294117647,
      "grad_norm": 0.35494352893710096,
      "learning_rate": 4.690747420866314e-06,
      "loss": 0.2676,
      "step": 11892
    },
    {
      "epoch": 5.596705882352941,
      "grad_norm": 0.39594896537077373,
      "learning_rate": 4.679876830666434e-06,
      "loss": 0.1611,
      "step": 11893
    },
    {
      "epoch": 5.597176470588235,
      "grad_norm": 0.8048324992738916,
      "learning_rate": 4.669018651517859e-06,
      "loss": 0.6953,
      "step": 11894
    },
    {
      "epoch": 5.597647058823529,
      "grad_norm": 0.24152519230959954,
      "learning_rate": 4.658172884347943e-06,
      "loss": 0.0381,
      "step": 11895
    },
    {
      "epoch": 5.598117647058824,
      "grad_norm": 0.49231747999261116,
      "learning_rate": 4.6473395300829495e-06,
      "loss": 0.2109,
      "step": 11896
    },
    {
      "epoch": 5.598588235294118,
      "grad_norm": 0.3827040680716455,
      "learning_rate": 4.636518589648119e-06,
      "loss": 0.1465,
      "step": 11897
    },
    {
      "epoch": 5.599058823529412,
      "grad_norm": 0.3101946358876503,
      "learning_rate": 4.625710063967613e-06,
      "loss": 0.1709,
      "step": 11898
    },
    {
      "epoch": 5.599529411764706,
      "grad_norm": 0.400613974426702,
      "learning_rate": 4.614913953964511e-06,
      "loss": 0.2969,
      "step": 11899
    },
    {
      "epoch": 5.6,
      "grad_norm": 0.33341486599273223,
      "learning_rate": 4.604130260560873e-06,
      "loss": 0.3047,
      "step": 11900
    },
    {
      "epoch": 5.600470588235294,
      "grad_norm": 0.18244350006574153,
      "learning_rate": 4.593358984677681e-06,
      "loss": 0.0251,
      "step": 11901
    },
    {
      "epoch": 5.600941176470588,
      "grad_norm": 0.17724671882718493,
      "learning_rate": 4.5826001272348315e-06,
      "loss": 0.0244,
      "step": 11902
    },
    {
      "epoch": 5.601411764705882,
      "grad_norm": 0.23920146158575442,
      "learning_rate": 4.571853689151206e-06,
      "loss": 0.0417,
      "step": 11903
    },
    {
      "epoch": 5.6018823529411765,
      "grad_norm": 0.3621244884964851,
      "learning_rate": 4.5611196713446045e-06,
      "loss": 0.2217,
      "step": 11904
    },
    {
      "epoch": 5.602352941176471,
      "grad_norm": 0.30799242911830893,
      "learning_rate": 4.5503980747317425e-06,
      "loss": 0.2432,
      "step": 11905
    },
    {
      "epoch": 5.602823529411765,
      "grad_norm": 0.15400847587398117,
      "learning_rate": 4.539688900228322e-06,
      "loss": 0.0261,
      "step": 11906
    },
    {
      "epoch": 5.603294117647059,
      "grad_norm": 0.4614328484476601,
      "learning_rate": 4.528992148748944e-06,
      "loss": 0.2715,
      "step": 11907
    },
    {
      "epoch": 5.603764705882353,
      "grad_norm": 0.4398487978034609,
      "learning_rate": 4.518307821207162e-06,
      "loss": 0.2422,
      "step": 11908
    },
    {
      "epoch": 5.604235294117647,
      "grad_norm": 0.24878722829937297,
      "learning_rate": 4.507635918515496e-06,
      "loss": 0.062,
      "step": 11909
    },
    {
      "epoch": 5.604705882352941,
      "grad_norm": 0.31061812482276524,
      "learning_rate": 4.496976441585348e-06,
      "loss": 0.2363,
      "step": 11910
    },
    {
      "epoch": 5.605176470588235,
      "grad_norm": 0.43249456954792503,
      "learning_rate": 4.486329391327093e-06,
      "loss": 0.3848,
      "step": 11911
    },
    {
      "epoch": 5.605647058823529,
      "grad_norm": 0.3710181697049924,
      "learning_rate": 4.475694768650051e-06,
      "loss": 0.248,
      "step": 11912
    },
    {
      "epoch": 5.606117647058824,
      "grad_norm": 0.34706981265072584,
      "learning_rate": 4.4650725744624635e-06,
      "loss": 0.1074,
      "step": 11913
    },
    {
      "epoch": 5.606588235294118,
      "grad_norm": 0.31057740828782504,
      "learning_rate": 4.454462809671538e-06,
      "loss": 0.0752,
      "step": 11914
    },
    {
      "epoch": 5.607058823529412,
      "grad_norm": 0.273806291036883,
      "learning_rate": 4.443865475183384e-06,
      "loss": 0.0688,
      "step": 11915
    },
    {
      "epoch": 5.607529411764705,
      "grad_norm": 0.4759303860166497,
      "learning_rate": 4.4332805719030586e-06,
      "loss": 0.2354,
      "step": 11916
    },
    {
      "epoch": 5.608,
      "grad_norm": 0.3021980995899601,
      "learning_rate": 4.4227081007345565e-06,
      "loss": 0.125,
      "step": 11917
    },
    {
      "epoch": 5.608470588235294,
      "grad_norm": 0.5312939861103322,
      "learning_rate": 4.412148062580873e-06,
      "loss": 0.543,
      "step": 11918
    },
    {
      "epoch": 5.608941176470588,
      "grad_norm": 0.5249964047768085,
      "learning_rate": 4.401600458343818e-06,
      "loss": 0.4004,
      "step": 11919
    },
    {
      "epoch": 5.609411764705882,
      "grad_norm": 0.4173005005442682,
      "learning_rate": 4.39106528892424e-06,
      "loss": 0.3027,
      "step": 11920
    },
    {
      "epoch": 5.6098823529411765,
      "grad_norm": 0.42551303520288514,
      "learning_rate": 4.380542555221917e-06,
      "loss": 0.2539,
      "step": 11921
    },
    {
      "epoch": 5.610352941176471,
      "grad_norm": 0.19138674067828768,
      "learning_rate": 4.370032258135514e-06,
      "loss": 0.0496,
      "step": 11922
    },
    {
      "epoch": 5.610823529411765,
      "grad_norm": 0.44802590898324063,
      "learning_rate": 4.3595343985626476e-06,
      "loss": 0.2773,
      "step": 11923
    },
    {
      "epoch": 5.611294117647059,
      "grad_norm": 0.1993503210387218,
      "learning_rate": 4.349048977399933e-06,
      "loss": 0.0354,
      "step": 11924
    },
    {
      "epoch": 5.6117647058823525,
      "grad_norm": 0.5909337496933511,
      "learning_rate": 4.33857599554282e-06,
      "loss": 0.5508,
      "step": 11925
    },
    {
      "epoch": 5.612235294117647,
      "grad_norm": 0.5141767693986061,
      "learning_rate": 4.328115453885794e-06,
      "loss": 0.3418,
      "step": 11926
    },
    {
      "epoch": 5.612705882352941,
      "grad_norm": 0.5866075844418919,
      "learning_rate": 4.317667353322241e-06,
      "loss": 0.5078,
      "step": 11927
    },
    {
      "epoch": 5.613176470588235,
      "grad_norm": 0.44799488459424747,
      "learning_rate": 4.307231694744445e-06,
      "loss": 0.2832,
      "step": 11928
    },
    {
      "epoch": 5.613647058823529,
      "grad_norm": 0.282447406883562,
      "learning_rate": 4.296808479043678e-06,
      "loss": 0.083,
      "step": 11929
    },
    {
      "epoch": 5.614117647058824,
      "grad_norm": 0.38958561844076056,
      "learning_rate": 4.286397707110145e-06,
      "loss": 0.1348,
      "step": 11930
    },
    {
      "epoch": 5.614588235294118,
      "grad_norm": 0.44684373323829113,
      "learning_rate": 4.275999379832967e-06,
      "loss": 0.1553,
      "step": 11931
    },
    {
      "epoch": 5.615058823529412,
      "grad_norm": 0.28058463182877275,
      "learning_rate": 4.265613498100234e-06,
      "loss": 0.2402,
      "step": 11932
    },
    {
      "epoch": 5.615529411764706,
      "grad_norm": 0.5231942029934937,
      "learning_rate": 4.255240062798904e-06,
      "loss": 0.3105,
      "step": 11933
    },
    {
      "epoch": 5.616,
      "grad_norm": 0.28800300326752454,
      "learning_rate": 4.24487907481495e-06,
      "loss": 0.2617,
      "step": 11934
    },
    {
      "epoch": 5.616470588235294,
      "grad_norm": 0.46021910125244814,
      "learning_rate": 4.234530535033265e-06,
      "loss": 0.3477,
      "step": 11935
    },
    {
      "epoch": 5.616941176470588,
      "grad_norm": 0.35398201600419094,
      "learning_rate": 4.224194444337642e-06,
      "loss": 0.2715,
      "step": 11936
    },
    {
      "epoch": 5.617411764705882,
      "grad_norm": 0.23032304219387106,
      "learning_rate": 4.213870803610842e-06,
      "loss": 0.1494,
      "step": 11937
    },
    {
      "epoch": 5.6178823529411765,
      "grad_norm": 0.6027349225578855,
      "learning_rate": 4.203559613734542e-06,
      "loss": 0.5586,
      "step": 11938
    },
    {
      "epoch": 5.618352941176471,
      "grad_norm": 0.6162507290575518,
      "learning_rate": 4.193260875589371e-06,
      "loss": 0.3848,
      "step": 11939
    },
    {
      "epoch": 5.618823529411765,
      "grad_norm": 0.25565032479675154,
      "learning_rate": 4.1829745900549275e-06,
      "loss": 0.1157,
      "step": 11940
    },
    {
      "epoch": 5.619294117647058,
      "grad_norm": 0.39115811247306637,
      "learning_rate": 4.1727007580096565e-06,
      "loss": 0.3789,
      "step": 11941
    },
    {
      "epoch": 5.619764705882353,
      "grad_norm": 0.3387401378745163,
      "learning_rate": 4.162439380331023e-06,
      "loss": 0.126,
      "step": 11942
    },
    {
      "epoch": 5.620235294117647,
      "grad_norm": 0.39240020013544863,
      "learning_rate": 4.15219045789541e-06,
      "loss": 0.4258,
      "step": 11943
    },
    {
      "epoch": 5.620705882352941,
      "grad_norm": 0.3318609342151118,
      "learning_rate": 4.141953991578101e-06,
      "loss": 0.1367,
      "step": 11944
    },
    {
      "epoch": 5.621176470588235,
      "grad_norm": 0.39225366728216426,
      "learning_rate": 4.131729982253345e-06,
      "loss": 0.1699,
      "step": 11945
    },
    {
      "epoch": 5.621647058823529,
      "grad_norm": 0.4254168630010229,
      "learning_rate": 4.121518430794346e-06,
      "loss": 0.1826,
      "step": 11946
    },
    {
      "epoch": 5.622117647058824,
      "grad_norm": 0.2563546200629351,
      "learning_rate": 4.111319338073188e-06,
      "loss": 0.0542,
      "step": 11947
    },
    {
      "epoch": 5.622588235294118,
      "grad_norm": 0.41969453266858125,
      "learning_rate": 4.101132704960941e-06,
      "loss": 0.103,
      "step": 11948
    },
    {
      "epoch": 5.623058823529412,
      "grad_norm": 0.46053979493747826,
      "learning_rate": 4.09095853232761e-06,
      "loss": 0.2969,
      "step": 11949
    },
    {
      "epoch": 5.623529411764705,
      "grad_norm": 0.23734084809955425,
      "learning_rate": 4.080796821042082e-06,
      "loss": 0.1709,
      "step": 11950
    },
    {
      "epoch": 5.624,
      "grad_norm": 0.4727680400104765,
      "learning_rate": 4.07064757197223e-06,
      "loss": 0.4023,
      "step": 11951
    },
    {
      "epoch": 5.624470588235294,
      "grad_norm": 0.319105798658799,
      "learning_rate": 4.0605107859848604e-06,
      "loss": 0.0586,
      "step": 11952
    },
    {
      "epoch": 5.624941176470588,
      "grad_norm": 0.4545078550250721,
      "learning_rate": 4.050386463945699e-06,
      "loss": 0.3809,
      "step": 11953
    },
    {
      "epoch": 5.625411764705882,
      "grad_norm": 0.4660494881428432,
      "learning_rate": 4.040274606719418e-06,
      "loss": 0.3594,
      "step": 11954
    },
    {
      "epoch": 5.625882352941177,
      "grad_norm": 0.34845650377362297,
      "learning_rate": 4.030175215169612e-06,
      "loss": 0.125,
      "step": 11955
    },
    {
      "epoch": 5.626352941176471,
      "grad_norm": 0.3994374807033917,
      "learning_rate": 4.020088290158807e-06,
      "loss": 0.249,
      "step": 11956
    },
    {
      "epoch": 5.626823529411765,
      "grad_norm": 0.21321953122619575,
      "learning_rate": 4.010013832548514e-06,
      "loss": 0.0811,
      "step": 11957
    },
    {
      "epoch": 5.627294117647059,
      "grad_norm": 0.35155282652951847,
      "learning_rate": 3.999951843199095e-06,
      "loss": 0.103,
      "step": 11958
    },
    {
      "epoch": 5.627764705882353,
      "grad_norm": 0.3159594092438703,
      "learning_rate": 3.989902322969929e-06,
      "loss": 0.0674,
      "step": 11959
    },
    {
      "epoch": 5.628235294117647,
      "grad_norm": 0.4274422740949337,
      "learning_rate": 3.979865272719279e-06,
      "loss": 0.1885,
      "step": 11960
    },
    {
      "epoch": 5.628705882352941,
      "grad_norm": 0.2852142403733466,
      "learning_rate": 3.969840693304344e-06,
      "loss": 0.0737,
      "step": 11961
    },
    {
      "epoch": 5.629176470588235,
      "grad_norm": 0.3023402638278734,
      "learning_rate": 3.959828585581287e-06,
      "loss": 0.2969,
      "step": 11962
    },
    {
      "epoch": 5.6296470588235294,
      "grad_norm": 0.5017210040747833,
      "learning_rate": 3.949828950405226e-06,
      "loss": 0.3125,
      "step": 11963
    },
    {
      "epoch": 5.630117647058824,
      "grad_norm": 0.5167313404521562,
      "learning_rate": 3.939841788630111e-06,
      "loss": 0.5078,
      "step": 11964
    },
    {
      "epoch": 5.630588235294118,
      "grad_norm": 0.39082309906623613,
      "learning_rate": 3.9298671011089245e-06,
      "loss": 0.2139,
      "step": 11965
    },
    {
      "epoch": 5.631058823529412,
      "grad_norm": 0.3510403356270383,
      "learning_rate": 3.9199048886935865e-06,
      "loss": 0.1113,
      "step": 11966
    },
    {
      "epoch": 5.631529411764706,
      "grad_norm": 0.40731788791471674,
      "learning_rate": 3.909955152234867e-06,
      "loss": 0.1006,
      "step": 11967
    },
    {
      "epoch": 5.632,
      "grad_norm": 0.22707264414035302,
      "learning_rate": 3.900017892582552e-06,
      "loss": 0.1641,
      "step": 11968
    },
    {
      "epoch": 5.632470588235294,
      "grad_norm": 0.40558194046924423,
      "learning_rate": 3.8900931105853294e-06,
      "loss": 0.3613,
      "step": 11969
    },
    {
      "epoch": 5.632941176470588,
      "grad_norm": 0.5771707729954422,
      "learning_rate": 3.8801808070908215e-06,
      "loss": 0.5,
      "step": 11970
    },
    {
      "epoch": 5.633411764705882,
      "grad_norm": 0.4266328897351007,
      "learning_rate": 3.870280982945567e-06,
      "loss": 0.3672,
      "step": 11971
    },
    {
      "epoch": 5.633882352941177,
      "grad_norm": 0.4047416429934266,
      "learning_rate": 3.860393638995107e-06,
      "loss": 0.3301,
      "step": 11972
    },
    {
      "epoch": 5.634352941176471,
      "grad_norm": 0.3614797896159205,
      "learning_rate": 3.850518776083817e-06,
      "loss": 0.0938,
      "step": 11973
    },
    {
      "epoch": 5.634823529411765,
      "grad_norm": 0.21866913108445454,
      "learning_rate": 3.840656395055103e-06,
      "loss": 0.0588,
      "step": 11974
    },
    {
      "epoch": 5.635294117647058,
      "grad_norm": 0.36297509240697545,
      "learning_rate": 3.8308064967512095e-06,
      "loss": 0.2988,
      "step": 11975
    },
    {
      "epoch": 5.635764705882353,
      "grad_norm": 0.45268031076532667,
      "learning_rate": 3.820969082013414e-06,
      "loss": 0.334,
      "step": 11976
    },
    {
      "epoch": 5.636235294117647,
      "grad_norm": 0.5467524413891465,
      "learning_rate": 3.8111441516818776e-06,
      "loss": 0.3301,
      "step": 11977
    },
    {
      "epoch": 5.636705882352941,
      "grad_norm": 0.3246700651737955,
      "learning_rate": 3.8013317065956784e-06,
      "loss": 0.1133,
      "step": 11978
    },
    {
      "epoch": 5.637176470588235,
      "grad_norm": 0.2330912263243463,
      "learning_rate": 3.7915317475928296e-06,
      "loss": 0.1621,
      "step": 11979
    },
    {
      "epoch": 5.6376470588235295,
      "grad_norm": 0.42916901201000984,
      "learning_rate": 3.7817442755103457e-06,
      "loss": 0.4004,
      "step": 11980
    },
    {
      "epoch": 5.638117647058824,
      "grad_norm": 0.4291226950741453,
      "learning_rate": 3.7719692911840905e-06,
      "loss": 0.3652,
      "step": 11981
    },
    {
      "epoch": 5.638588235294118,
      "grad_norm": 0.44360273026988395,
      "learning_rate": 3.76220679544888e-06,
      "loss": 0.4004,
      "step": 11982
    },
    {
      "epoch": 5.639058823529412,
      "grad_norm": 0.3930426465887946,
      "learning_rate": 3.7524567891385305e-06,
      "loss": 0.2617,
      "step": 11983
    },
    {
      "epoch": 5.6395294117647055,
      "grad_norm": 0.2652941320744692,
      "learning_rate": 3.7427192730856927e-06,
      "loss": 0.0781,
      "step": 11984
    },
    {
      "epoch": 5.64,
      "grad_norm": 0.38322737852953837,
      "learning_rate": 3.7329942481220176e-06,
      "loss": 0.2246,
      "step": 11985
    },
    {
      "epoch": 5.640470588235294,
      "grad_norm": 0.37667390231047204,
      "learning_rate": 3.723281715078091e-06,
      "loss": 0.2969,
      "step": 11986
    },
    {
      "epoch": 5.640941176470588,
      "grad_norm": 0.3616132888514867,
      "learning_rate": 3.7135816747833667e-06,
      "loss": 0.1748,
      "step": 11987
    },
    {
      "epoch": 5.641411764705882,
      "grad_norm": 0.3755142737968915,
      "learning_rate": 3.7038941280663147e-06,
      "loss": 0.1709,
      "step": 11988
    },
    {
      "epoch": 5.641882352941177,
      "grad_norm": 0.3477408585910595,
      "learning_rate": 3.6942190757542735e-06,
      "loss": 0.0786,
      "step": 11989
    },
    {
      "epoch": 5.642352941176471,
      "grad_norm": 0.4952282188125007,
      "learning_rate": 3.684556518673548e-06,
      "loss": 0.2734,
      "step": 11990
    },
    {
      "epoch": 5.642823529411765,
      "grad_norm": 0.35592407697963774,
      "learning_rate": 3.6749064576493793e-06,
      "loss": 0.1172,
      "step": 11991
    },
    {
      "epoch": 5.643294117647059,
      "grad_norm": 0.4374657240706044,
      "learning_rate": 3.6652688935058904e-06,
      "loss": 0.4688,
      "step": 11992
    },
    {
      "epoch": 5.643764705882353,
      "grad_norm": 0.6006120356125971,
      "learning_rate": 3.6556438270662236e-06,
      "loss": 0.2578,
      "step": 11993
    },
    {
      "epoch": 5.644235294117647,
      "grad_norm": 0.21402001675498758,
      "learning_rate": 3.6460312591524044e-06,
      "loss": 0.0522,
      "step": 11994
    },
    {
      "epoch": 5.644705882352941,
      "grad_norm": 0.2703409825004869,
      "learning_rate": 3.6364311905853594e-06,
      "loss": 0.0552,
      "step": 11995
    },
    {
      "epoch": 5.645176470588235,
      "grad_norm": 0.4203021267881019,
      "learning_rate": 3.6268436221849994e-06,
      "loss": 0.1514,
      "step": 11996
    },
    {
      "epoch": 5.6456470588235295,
      "grad_norm": 0.41536992643534376,
      "learning_rate": 3.6172685547701695e-06,
      "loss": 0.3535,
      "step": 11997
    },
    {
      "epoch": 5.646117647058824,
      "grad_norm": 0.7483697921224272,
      "learning_rate": 3.6077059891585826e-06,
      "loss": 0.2871,
      "step": 11998
    },
    {
      "epoch": 5.646588235294118,
      "grad_norm": 0.25727938075723084,
      "learning_rate": 3.598155926166968e-06,
      "loss": 0.0703,
      "step": 11999
    },
    {
      "epoch": 5.647058823529412,
      "grad_norm": 0.24327771226301934,
      "learning_rate": 3.5886183666109405e-06,
      "loss": 0.0352,
      "step": 12000
    },
    {
      "epoch": 5.647529411764706,
      "grad_norm": 0.3525532033801126,
      "learning_rate": 3.579093311305031e-06,
      "loss": 0.4082,
      "step": 12001
    },
    {
      "epoch": 5.648,
      "grad_norm": 0.5308763568500937,
      "learning_rate": 3.5695807610627402e-06,
      "loss": 0.1914,
      "step": 12002
    },
    {
      "epoch": 5.648470588235294,
      "grad_norm": 0.2665766334538837,
      "learning_rate": 3.5600807166965173e-06,
      "loss": 0.1445,
      "step": 12003
    },
    {
      "epoch": 5.648941176470588,
      "grad_norm": 0.40214674910712767,
      "learning_rate": 3.5505931790176635e-06,
      "loss": 0.4492,
      "step": 12004
    },
    {
      "epoch": 5.649411764705882,
      "grad_norm": 0.3977933993589181,
      "learning_rate": 3.5411181488364802e-06,
      "loss": 0.5273,
      "step": 12005
    },
    {
      "epoch": 5.649882352941177,
      "grad_norm": 0.29939134385390537,
      "learning_rate": 3.531655626962204e-06,
      "loss": 0.3125,
      "step": 12006
    },
    {
      "epoch": 5.650352941176471,
      "grad_norm": 0.6006470798049292,
      "learning_rate": 3.5222056142029387e-06,
      "loss": 0.5117,
      "step": 12007
    },
    {
      "epoch": 5.650823529411765,
      "grad_norm": 0.44169510423676855,
      "learning_rate": 3.5127681113658044e-06,
      "loss": 0.1611,
      "step": 12008
    },
    {
      "epoch": 5.651294117647058,
      "grad_norm": 0.3954373455130933,
      "learning_rate": 3.503343119256774e-06,
      "loss": 0.2969,
      "step": 12009
    },
    {
      "epoch": 5.651764705882353,
      "grad_norm": 0.37884235331282223,
      "learning_rate": 3.493930638680803e-06,
      "loss": 0.2695,
      "step": 12010
    },
    {
      "epoch": 5.652235294117647,
      "grad_norm": 0.42346019963187886,
      "learning_rate": 3.484530670441782e-06,
      "loss": 0.3574,
      "step": 12011
    },
    {
      "epoch": 5.652705882352941,
      "grad_norm": 0.5287874392970315,
      "learning_rate": 3.4751432153424687e-06,
      "loss": 0.166,
      "step": 12012
    },
    {
      "epoch": 5.653176470588235,
      "grad_norm": 0.263068241128108,
      "learning_rate": 3.465768274184638e-06,
      "loss": 0.1826,
      "step": 12013
    },
    {
      "epoch": 5.6536470588235295,
      "grad_norm": 0.39291803592960406,
      "learning_rate": 3.4564058477689503e-06,
      "loss": 0.4336,
      "step": 12014
    },
    {
      "epoch": 5.654117647058824,
      "grad_norm": 0.2952670879770033,
      "learning_rate": 3.4470559368949825e-06,
      "loss": 0.166,
      "step": 12015
    },
    {
      "epoch": 5.654588235294118,
      "grad_norm": 0.26847585094154575,
      "learning_rate": 3.4377185423612785e-06,
      "loss": 0.0498,
      "step": 12016
    },
    {
      "epoch": 5.655058823529412,
      "grad_norm": 0.39758145929312244,
      "learning_rate": 3.428393664965301e-06,
      "loss": 0.3262,
      "step": 12017
    },
    {
      "epoch": 5.6555294117647055,
      "grad_norm": 0.23134755704299434,
      "learning_rate": 3.4190813055034127e-06,
      "loss": 0.1396,
      "step": 12018
    },
    {
      "epoch": 5.656,
      "grad_norm": 0.35348455466807305,
      "learning_rate": 3.4097814647709775e-06,
      "loss": 0.2695,
      "step": 12019
    },
    {
      "epoch": 5.656470588235294,
      "grad_norm": 0.4207406697070535,
      "learning_rate": 3.400494143562227e-06,
      "loss": 0.3281,
      "step": 12020
    },
    {
      "epoch": 5.656941176470588,
      "grad_norm": 0.3470562269022183,
      "learning_rate": 3.391219342670326e-06,
      "loss": 0.3203,
      "step": 12021
    },
    {
      "epoch": 5.657411764705882,
      "grad_norm": 0.430765757055295,
      "learning_rate": 3.3819570628874414e-06,
      "loss": 0.5391,
      "step": 12022
    },
    {
      "epoch": 5.657882352941177,
      "grad_norm": 0.34778478296029663,
      "learning_rate": 3.3727073050045406e-06,
      "loss": 0.1885,
      "step": 12023
    },
    {
      "epoch": 5.658352941176471,
      "grad_norm": 0.3920189654904357,
      "learning_rate": 3.3634700698116577e-06,
      "loss": 0.1216,
      "step": 12024
    },
    {
      "epoch": 5.658823529411765,
      "grad_norm": 0.5363335259723596,
      "learning_rate": 3.354245358097679e-06,
      "loss": 0.3223,
      "step": 12025
    },
    {
      "epoch": 5.659294117647059,
      "grad_norm": 0.32049775848308854,
      "learning_rate": 3.345033170650441e-06,
      "loss": 0.1309,
      "step": 12026
    },
    {
      "epoch": 5.659764705882353,
      "grad_norm": 0.45830925470000944,
      "learning_rate": 3.3358335082566966e-06,
      "loss": 0.3281,
      "step": 12027
    },
    {
      "epoch": 5.660235294117647,
      "grad_norm": 0.3698228626422182,
      "learning_rate": 3.326646371702185e-06,
      "loss": 0.1572,
      "step": 12028
    },
    {
      "epoch": 5.660705882352941,
      "grad_norm": 0.34349583259837285,
      "learning_rate": 3.317471761771462e-06,
      "loss": 0.1064,
      "step": 12029
    },
    {
      "epoch": 5.661176470588235,
      "grad_norm": 0.2729188646906925,
      "learning_rate": 3.30830967924815e-06,
      "loss": 0.0845,
      "step": 12030
    },
    {
      "epoch": 5.6616470588235295,
      "grad_norm": 0.5172285767758495,
      "learning_rate": 3.299160124914707e-06,
      "loss": 0.4629,
      "step": 12031
    },
    {
      "epoch": 5.662117647058824,
      "grad_norm": 0.731306069187779,
      "learning_rate": 3.290023099552541e-06,
      "loss": 0.5156,
      "step": 12032
    },
    {
      "epoch": 5.662588235294118,
      "grad_norm": 0.1423001482066179,
      "learning_rate": 3.2808986039420115e-06,
      "loss": 0.0151,
      "step": 12033
    },
    {
      "epoch": 5.663058823529412,
      "grad_norm": 0.3877491784333437,
      "learning_rate": 3.2717866388624114e-06,
      "loss": 0.3086,
      "step": 12034
    },
    {
      "epoch": 5.663529411764706,
      "grad_norm": 0.3358974987702559,
      "learning_rate": 3.2626872050919185e-06,
      "loss": 0.0566,
      "step": 12035
    },
    {
      "epoch": 5.664,
      "grad_norm": 0.4835397708421852,
      "learning_rate": 3.2536003034076774e-06,
      "loss": 0.5938,
      "step": 12036
    },
    {
      "epoch": 5.664470588235294,
      "grad_norm": 0.2750451999778068,
      "learning_rate": 3.2445259345857676e-06,
      "loss": 0.0322,
      "step": 12037
    },
    {
      "epoch": 5.664941176470588,
      "grad_norm": 0.28525938394348266,
      "learning_rate": 3.2354640994011693e-06,
      "loss": 0.1006,
      "step": 12038
    },
    {
      "epoch": 5.665411764705882,
      "grad_norm": 0.35290265960378503,
      "learning_rate": 3.22641479862783e-06,
      "loss": 0.1738,
      "step": 12039
    },
    {
      "epoch": 5.665882352941177,
      "grad_norm": 0.2905429133611641,
      "learning_rate": 3.2173780330385647e-06,
      "loss": 0.1094,
      "step": 12040
    },
    {
      "epoch": 5.666352941176471,
      "grad_norm": 0.4442020204423696,
      "learning_rate": 3.2083538034051726e-06,
      "loss": 0.2559,
      "step": 12041
    },
    {
      "epoch": 5.666823529411765,
      "grad_norm": 0.3207480743959322,
      "learning_rate": 3.199342110498404e-06,
      "loss": 0.165,
      "step": 12042
    },
    {
      "epoch": 5.667294117647058,
      "grad_norm": 0.6358518769377794,
      "learning_rate": 3.1903429550878435e-06,
      "loss": 0.4492,
      "step": 12043
    },
    {
      "epoch": 5.667764705882353,
      "grad_norm": 0.47330611065386236,
      "learning_rate": 3.181356337942109e-06,
      "loss": 0.6328,
      "step": 12044
    },
    {
      "epoch": 5.668235294117647,
      "grad_norm": 0.32095850537032705,
      "learning_rate": 3.1723822598286875e-06,
      "loss": 0.3066,
      "step": 12045
    },
    {
      "epoch": 5.668705882352941,
      "grad_norm": 0.6839302456983508,
      "learning_rate": 3.163420721513982e-06,
      "loss": 0.6367,
      "step": 12046
    },
    {
      "epoch": 5.669176470588235,
      "grad_norm": 0.46604858520384007,
      "learning_rate": 3.1544717237633965e-06,
      "loss": 0.3652,
      "step": 12047
    },
    {
      "epoch": 5.6696470588235295,
      "grad_norm": 0.4119367577201157,
      "learning_rate": 3.145535267341204e-06,
      "loss": 0.4102,
      "step": 12048
    },
    {
      "epoch": 5.670117647058824,
      "grad_norm": 0.3240416390597848,
      "learning_rate": 3.1366113530106097e-06,
      "loss": 0.0791,
      "step": 12049
    },
    {
      "epoch": 5.670588235294118,
      "grad_norm": 0.3604461354018364,
      "learning_rate": 3.1276999815337544e-06,
      "loss": 0.3008,
      "step": 12050
    },
    {
      "epoch": 5.671058823529412,
      "grad_norm": 0.2598156243036814,
      "learning_rate": 3.118801153671746e-06,
      "loss": 0.0522,
      "step": 12051
    },
    {
      "epoch": 5.6715294117647055,
      "grad_norm": 0.28686883116401884,
      "learning_rate": 3.1099148701845435e-06,
      "loss": 0.1099,
      "step": 12052
    },
    {
      "epoch": 5.672,
      "grad_norm": 0.38908289342266866,
      "learning_rate": 3.1010411318311226e-06,
      "loss": 0.2852,
      "step": 12053
    },
    {
      "epoch": 5.672470588235294,
      "grad_norm": 0.34041699527973,
      "learning_rate": 3.092179939369327e-06,
      "loss": 0.106,
      "step": 12054
    },
    {
      "epoch": 5.672941176470588,
      "grad_norm": 0.9752349256744569,
      "learning_rate": 3.083331293555935e-06,
      "loss": 0.6797,
      "step": 12055
    },
    {
      "epoch": 5.673411764705882,
      "grad_norm": 0.29110382811358754,
      "learning_rate": 3.074495195146659e-06,
      "loss": 0.1069,
      "step": 12056
    },
    {
      "epoch": 5.673882352941177,
      "grad_norm": 0.39249390754351104,
      "learning_rate": 3.0656716448961784e-06,
      "loss": 0.4785,
      "step": 12057
    },
    {
      "epoch": 5.674352941176471,
      "grad_norm": 0.37053442317693897,
      "learning_rate": 3.05686064355804e-06,
      "loss": 0.3516,
      "step": 12058
    },
    {
      "epoch": 5.674823529411765,
      "grad_norm": 0.37066938732529214,
      "learning_rate": 3.048062191884759e-06,
      "loss": 0.4102,
      "step": 12059
    },
    {
      "epoch": 5.675294117647059,
      "grad_norm": 0.45730726436055846,
      "learning_rate": 3.039276290627768e-06,
      "loss": 0.375,
      "step": 12060
    },
    {
      "epoch": 5.675764705882353,
      "grad_norm": 0.20289106182562994,
      "learning_rate": 3.0305029405373993e-06,
      "loss": 0.0439,
      "step": 12061
    },
    {
      "epoch": 5.676235294117647,
      "grad_norm": 0.4946594898358051,
      "learning_rate": 3.0217421423629706e-06,
      "loss": 0.3008,
      "step": 12062
    },
    {
      "epoch": 5.676705882352941,
      "grad_norm": 0.38968057736771533,
      "learning_rate": 3.0129938968526836e-06,
      "loss": 0.0962,
      "step": 12063
    },
    {
      "epoch": 5.677176470588235,
      "grad_norm": 0.36692647542594653,
      "learning_rate": 3.00425820475369e-06,
      "loss": 0.3789,
      "step": 12064
    },
    {
      "epoch": 5.6776470588235295,
      "grad_norm": 0.4375098573677336,
      "learning_rate": 2.9955350668120605e-06,
      "loss": 0.2832,
      "step": 12065
    },
    {
      "epoch": 5.678117647058824,
      "grad_norm": 0.3937786075524737,
      "learning_rate": 2.9868244837727815e-06,
      "loss": 0.3418,
      "step": 12066
    },
    {
      "epoch": 5.678588235294118,
      "grad_norm": 0.33781745649378453,
      "learning_rate": 2.978126456379809e-06,
      "loss": 0.1035,
      "step": 12067
    },
    {
      "epoch": 5.679058823529411,
      "grad_norm": 0.47652022201059563,
      "learning_rate": 2.969440985375948e-06,
      "loss": 0.3867,
      "step": 12068
    },
    {
      "epoch": 5.679529411764706,
      "grad_norm": 0.5333524103178429,
      "learning_rate": 2.960768071503006e-06,
      "loss": 0.2266,
      "step": 12069
    },
    {
      "epoch": 5.68,
      "grad_norm": 0.4495898899985985,
      "learning_rate": 2.9521077155017224e-06,
      "loss": 0.4941,
      "step": 12070
    },
    {
      "epoch": 5.680470588235294,
      "grad_norm": 0.3570914274108706,
      "learning_rate": 2.9434599181116736e-06,
      "loss": 0.1348,
      "step": 12071
    },
    {
      "epoch": 5.680941176470588,
      "grad_norm": 0.5283223974932614,
      "learning_rate": 2.934824680071468e-06,
      "loss": 0.2207,
      "step": 12072
    },
    {
      "epoch": 5.681411764705882,
      "grad_norm": 0.42769160013875474,
      "learning_rate": 2.9262020021185993e-06,
      "loss": 0.2275,
      "step": 12073
    },
    {
      "epoch": 5.681882352941177,
      "grad_norm": 0.5745696290212329,
      "learning_rate": 2.917591884989462e-06,
      "loss": 0.4707,
      "step": 12074
    },
    {
      "epoch": 5.682352941176471,
      "grad_norm": 0.47784108491326777,
      "learning_rate": 2.908994329419401e-06,
      "loss": 0.3105,
      "step": 12075
    },
    {
      "epoch": 5.682823529411765,
      "grad_norm": 0.32438704415194947,
      "learning_rate": 2.9004093361427282e-06,
      "loss": 0.2246,
      "step": 12076
    },
    {
      "epoch": 5.683294117647058,
      "grad_norm": 0.17143440241481428,
      "learning_rate": 2.891836905892592e-06,
      "loss": 0.1094,
      "step": 12077
    },
    {
      "epoch": 5.683764705882353,
      "grad_norm": 0.508415774430236,
      "learning_rate": 2.883277039401155e-06,
      "loss": 0.4395,
      "step": 12078
    },
    {
      "epoch": 5.684235294117647,
      "grad_norm": 0.6059762094004434,
      "learning_rate": 2.874729737399467e-06,
      "loss": 0.2539,
      "step": 12079
    },
    {
      "epoch": 5.684705882352941,
      "grad_norm": 0.4022577614235142,
      "learning_rate": 2.866195000617494e-06,
      "loss": 0.2852,
      "step": 12080
    },
    {
      "epoch": 5.685176470588235,
      "grad_norm": 0.21028816301352724,
      "learning_rate": 2.857672829784169e-06,
      "loss": 0.0371,
      "step": 12081
    },
    {
      "epoch": 5.6856470588235295,
      "grad_norm": 0.31494144466451546,
      "learning_rate": 2.84916322562731e-06,
      "loss": 0.1904,
      "step": 12082
    },
    {
      "epoch": 5.686117647058824,
      "grad_norm": 0.44038546093841663,
      "learning_rate": 2.840666188873669e-06,
      "loss": 0.2461,
      "step": 12083
    },
    {
      "epoch": 5.686588235294118,
      "grad_norm": 0.5741595695491388,
      "learning_rate": 2.8321817202489493e-06,
      "loss": 0.2891,
      "step": 12084
    },
    {
      "epoch": 5.687058823529412,
      "grad_norm": 0.5822051663342588,
      "learning_rate": 2.8237098204777874e-06,
      "loss": 0.5391,
      "step": 12085
    },
    {
      "epoch": 5.6875294117647055,
      "grad_norm": 0.4419867130207652,
      "learning_rate": 2.815250490283688e-06,
      "loss": 0.2949,
      "step": 12086
    },
    {
      "epoch": 5.688,
      "grad_norm": 0.4195006299887657,
      "learning_rate": 2.8068037303891233e-06,
      "loss": 0.4219,
      "step": 12087
    },
    {
      "epoch": 5.688470588235294,
      "grad_norm": 0.34480787497484344,
      "learning_rate": 2.7983695415155162e-06,
      "loss": 0.0879,
      "step": 12088
    },
    {
      "epoch": 5.688941176470588,
      "grad_norm": 0.45589779951722365,
      "learning_rate": 2.789947924383157e-06,
      "loss": 0.3945,
      "step": 12089
    },
    {
      "epoch": 5.689411764705882,
      "grad_norm": 0.38788257317595703,
      "learning_rate": 2.78153887971132e-06,
      "loss": 0.3789,
      "step": 12090
    },
    {
      "epoch": 5.689882352941177,
      "grad_norm": 0.4143054127312535,
      "learning_rate": 2.773142408218165e-06,
      "loss": 0.2314,
      "step": 12091
    },
    {
      "epoch": 5.690352941176471,
      "grad_norm": 0.5593462202279692,
      "learning_rate": 2.7647585106207682e-06,
      "loss": 0.6133,
      "step": 12092
    },
    {
      "epoch": 5.690823529411765,
      "grad_norm": 0.2226530215613681,
      "learning_rate": 2.7563871876352063e-06,
      "loss": 0.1079,
      "step": 12093
    },
    {
      "epoch": 5.691294117647059,
      "grad_norm": 0.5348797581187599,
      "learning_rate": 2.7480284399763907e-06,
      "loss": 0.2988,
      "step": 12094
    },
    {
      "epoch": 5.691764705882353,
      "grad_norm": 0.3572855613119104,
      "learning_rate": 2.7396822683582177e-06,
      "loss": 0.124,
      "step": 12095
    },
    {
      "epoch": 5.692235294117647,
      "grad_norm": 0.3976618325473087,
      "learning_rate": 2.731348673493483e-06,
      "loss": 0.3125,
      "step": 12096
    },
    {
      "epoch": 5.692705882352941,
      "grad_norm": 0.5018503247904758,
      "learning_rate": 2.723027656093918e-06,
      "loss": 0.582,
      "step": 12097
    },
    {
      "epoch": 5.693176470588235,
      "grad_norm": 0.23520037218067213,
      "learning_rate": 2.714719216870187e-06,
      "loss": 0.032,
      "step": 12098
    },
    {
      "epoch": 5.6936470588235295,
      "grad_norm": 0.5606606428759041,
      "learning_rate": 2.706423356531873e-06,
      "loss": 0.5625,
      "step": 12099
    },
    {
      "epoch": 5.694117647058824,
      "grad_norm": 0.36786718967815507,
      "learning_rate": 2.6981400757874584e-06,
      "loss": 0.2988,
      "step": 12100
    },
    {
      "epoch": 5.694588235294118,
      "grad_norm": 0.4915399352188861,
      "learning_rate": 2.6898693753443945e-06,
      "loss": 0.3047,
      "step": 12101
    },
    {
      "epoch": 5.695058823529411,
      "grad_norm": 0.5122918289595685,
      "learning_rate": 2.68161125590905e-06,
      "loss": 0.3027,
      "step": 12102
    },
    {
      "epoch": 5.6955294117647055,
      "grad_norm": 0.44730274681498444,
      "learning_rate": 2.673365718186693e-06,
      "loss": 0.377,
      "step": 12103
    },
    {
      "epoch": 5.696,
      "grad_norm": 0.35889780906746205,
      "learning_rate": 2.6651327628815277e-06,
      "loss": 0.3906,
      "step": 12104
    },
    {
      "epoch": 5.696470588235294,
      "grad_norm": 0.3181012030508439,
      "learning_rate": 2.656912390696708e-06,
      "loss": 0.1924,
      "step": 12105
    },
    {
      "epoch": 5.696941176470588,
      "grad_norm": 0.28003629084539977,
      "learning_rate": 2.6487046023342563e-06,
      "loss": 0.1035,
      "step": 12106
    },
    {
      "epoch": 5.697411764705882,
      "grad_norm": 0.3754871650292195,
      "learning_rate": 2.6405093984952275e-06,
      "loss": 0.2754,
      "step": 12107
    },
    {
      "epoch": 5.697882352941177,
      "grad_norm": 0.3359441139044814,
      "learning_rate": 2.6323267798794625e-06,
      "loss": 0.3555,
      "step": 12108
    },
    {
      "epoch": 5.698352941176471,
      "grad_norm": 0.40311181426851744,
      "learning_rate": 2.624156747185835e-06,
      "loss": 0.2363,
      "step": 12109
    },
    {
      "epoch": 5.698823529411765,
      "grad_norm": 0.3010252107908311,
      "learning_rate": 2.615999301112104e-06,
      "loss": 0.1963,
      "step": 12110
    },
    {
      "epoch": 5.699294117647058,
      "grad_norm": 0.3414595201685379,
      "learning_rate": 2.6078544423549286e-06,
      "loss": 0.1953,
      "step": 12111
    },
    {
      "epoch": 5.699764705882353,
      "grad_norm": 0.4126204481678666,
      "learning_rate": 2.599722171609936e-06,
      "loss": 0.1924,
      "step": 12112
    },
    {
      "epoch": 5.700235294117647,
      "grad_norm": 0.38010517411252137,
      "learning_rate": 2.591602489571687e-06,
      "loss": 0.3379,
      "step": 12113
    },
    {
      "epoch": 5.700705882352941,
      "grad_norm": 0.5060169730262796,
      "learning_rate": 2.5834953969336105e-06,
      "loss": 0.5898,
      "step": 12114
    },
    {
      "epoch": 5.701176470588235,
      "grad_norm": 0.4805345002058108,
      "learning_rate": 2.5754008943880855e-06,
      "loss": 0.3672,
      "step": 12115
    },
    {
      "epoch": 5.7016470588235295,
      "grad_norm": 0.4599938938615191,
      "learning_rate": 2.5673189826264596e-06,
      "loss": 0.1631,
      "step": 12116
    },
    {
      "epoch": 5.702117647058824,
      "grad_norm": 0.24513122514022515,
      "learning_rate": 2.559249662338947e-06,
      "loss": 0.2246,
      "step": 12117
    },
    {
      "epoch": 5.702588235294118,
      "grad_norm": 0.38153333816983226,
      "learning_rate": 2.5511929342146964e-06,
      "loss": 0.1406,
      "step": 12118
    },
    {
      "epoch": 5.703058823529412,
      "grad_norm": 0.4062928543810252,
      "learning_rate": 2.543148798941791e-06,
      "loss": 0.2559,
      "step": 12119
    },
    {
      "epoch": 5.7035294117647055,
      "grad_norm": 0.4043214509666668,
      "learning_rate": 2.535117257207264e-06,
      "loss": 0.2393,
      "step": 12120
    },
    {
      "epoch": 5.704,
      "grad_norm": 0.26661907775238647,
      "learning_rate": 2.5270983096970344e-06,
      "loss": 0.1475,
      "step": 12121
    },
    {
      "epoch": 5.704470588235294,
      "grad_norm": 0.29957376011470416,
      "learning_rate": 2.5190919570959533e-06,
      "loss": 0.2236,
      "step": 12122
    },
    {
      "epoch": 5.704941176470588,
      "grad_norm": 0.317883957167185,
      "learning_rate": 2.511098200087808e-06,
      "loss": 0.0742,
      "step": 12123
    },
    {
      "epoch": 5.705411764705882,
      "grad_norm": 0.4162886894546482,
      "learning_rate": 2.503117039355318e-06,
      "loss": 0.3906,
      "step": 12124
    },
    {
      "epoch": 5.705882352941177,
      "grad_norm": 0.20159280125425663,
      "learning_rate": 2.4951484755800886e-06,
      "loss": 0.0403,
      "step": 12125
    },
    {
      "epoch": 5.706352941176471,
      "grad_norm": 0.20281896091140422,
      "learning_rate": 2.487192509442676e-06,
      "loss": 0.0435,
      "step": 12126
    },
    {
      "epoch": 5.706823529411765,
      "grad_norm": 0.2999823264159142,
      "learning_rate": 2.4792491416226024e-06,
      "loss": 0.168,
      "step": 12127
    },
    {
      "epoch": 5.707294117647059,
      "grad_norm": 0.5093834540701783,
      "learning_rate": 2.471318372798209e-06,
      "loss": 0.25,
      "step": 12128
    },
    {
      "epoch": 5.707764705882353,
      "grad_norm": 0.48870268945927736,
      "learning_rate": 2.4634002036468536e-06,
      "loss": 0.3984,
      "step": 12129
    },
    {
      "epoch": 5.708235294117647,
      "grad_norm": 0.19009994896072702,
      "learning_rate": 2.4554946348447958e-06,
      "loss": 0.0532,
      "step": 12130
    },
    {
      "epoch": 5.708705882352941,
      "grad_norm": 0.38946187035070007,
      "learning_rate": 2.447601667067195e-06,
      "loss": 0.2578,
      "step": 12131
    },
    {
      "epoch": 5.709176470588235,
      "grad_norm": 0.4324306938289942,
      "learning_rate": 2.4397213009881456e-06,
      "loss": 0.5117,
      "step": 12132
    },
    {
      "epoch": 5.7096470588235295,
      "grad_norm": 0.2579846998551344,
      "learning_rate": 2.4318535372806757e-06,
      "loss": 0.248,
      "step": 12133
    },
    {
      "epoch": 5.710117647058824,
      "grad_norm": 0.41468622077558054,
      "learning_rate": 2.4239983766167315e-06,
      "loss": 0.3164,
      "step": 12134
    },
    {
      "epoch": 5.710588235294118,
      "grad_norm": 0.36262902494529636,
      "learning_rate": 2.416155819667193e-06,
      "loss": 0.1011,
      "step": 12135
    },
    {
      "epoch": 5.711058823529411,
      "grad_norm": 0.29081954214247335,
      "learning_rate": 2.4083258671018403e-06,
      "loss": 0.0703,
      "step": 12136
    },
    {
      "epoch": 5.7115294117647055,
      "grad_norm": 0.19939855595721925,
      "learning_rate": 2.4005085195893725e-06,
      "loss": 0.127,
      "step": 12137
    },
    {
      "epoch": 5.712,
      "grad_norm": 0.35896978840514077,
      "learning_rate": 2.3927037777974722e-06,
      "loss": 0.1465,
      "step": 12138
    },
    {
      "epoch": 5.712470588235294,
      "grad_norm": 0.5364724025407716,
      "learning_rate": 2.384911642392673e-06,
      "loss": 0.4062,
      "step": 12139
    },
    {
      "epoch": 5.712941176470588,
      "grad_norm": 0.45978791405720276,
      "learning_rate": 2.377132114040459e-06,
      "loss": 0.2559,
      "step": 12140
    },
    {
      "epoch": 5.713411764705882,
      "grad_norm": 0.3707111203085778,
      "learning_rate": 2.369365193405248e-06,
      "loss": 0.2334,
      "step": 12141
    },
    {
      "epoch": 5.713882352941177,
      "grad_norm": 0.21553652148874256,
      "learning_rate": 2.3616108811503775e-06,
      "loss": 0.0388,
      "step": 12142
    },
    {
      "epoch": 5.714352941176471,
      "grad_norm": 0.3516131756706093,
      "learning_rate": 2.3538691779381003e-06,
      "loss": 0.3184,
      "step": 12143
    },
    {
      "epoch": 5.714823529411765,
      "grad_norm": 0.16597615080841827,
      "learning_rate": 2.346140084429604e-06,
      "loss": 0.0693,
      "step": 12144
    },
    {
      "epoch": 5.715294117647058,
      "grad_norm": 0.3845915823079946,
      "learning_rate": 2.3384236012849778e-06,
      "loss": 0.126,
      "step": 12145
    },
    {
      "epoch": 5.715764705882353,
      "grad_norm": 0.33857579150694916,
      "learning_rate": 2.330719729163244e-06,
      "loss": 0.0942,
      "step": 12146
    },
    {
      "epoch": 5.716235294117647,
      "grad_norm": 0.24648237245410648,
      "learning_rate": 2.32302846872236e-06,
      "loss": 0.1797,
      "step": 12147
    },
    {
      "epoch": 5.716705882352941,
      "grad_norm": 0.5092569543422902,
      "learning_rate": 2.3153498206192e-06,
      "loss": 0.3438,
      "step": 12148
    },
    {
      "epoch": 5.717176470588235,
      "grad_norm": 0.32411954873704046,
      "learning_rate": 2.3076837855095563e-06,
      "loss": 0.2139,
      "step": 12149
    },
    {
      "epoch": 5.7176470588235295,
      "grad_norm": 0.2097837770073487,
      "learning_rate": 2.3000303640481386e-06,
      "loss": 0.0864,
      "step": 12150
    },
    {
      "epoch": 5.718117647058824,
      "grad_norm": 0.8130170400130744,
      "learning_rate": 2.292389556888591e-06,
      "loss": 0.7305,
      "step": 12151
    },
    {
      "epoch": 5.718588235294118,
      "grad_norm": 0.4423854775070703,
      "learning_rate": 2.2847613646834905e-06,
      "loss": 0.248,
      "step": 12152
    },
    {
      "epoch": 5.719058823529412,
      "grad_norm": 0.14833537864199278,
      "learning_rate": 2.2771457880842834e-06,
      "loss": 0.0618,
      "step": 12153
    },
    {
      "epoch": 5.7195294117647055,
      "grad_norm": 0.543726432395289,
      "learning_rate": 2.2695428277414165e-06,
      "loss": 0.5859,
      "step": 12154
    },
    {
      "epoch": 5.72,
      "grad_norm": 0.2401549889958617,
      "learning_rate": 2.261952484304219e-06,
      "loss": 0.1787,
      "step": 12155
    },
    {
      "epoch": 5.720470588235294,
      "grad_norm": 0.6162759739555046,
      "learning_rate": 2.254374758420924e-06,
      "loss": 0.208,
      "step": 12156
    },
    {
      "epoch": 5.720941176470588,
      "grad_norm": 0.3637659722546839,
      "learning_rate": 2.2468096507386956e-06,
      "loss": 0.2695,
      "step": 12157
    },
    {
      "epoch": 5.721411764705882,
      "grad_norm": 0.3092971584214811,
      "learning_rate": 2.239257161903668e-06,
      "loss": 0.2656,
      "step": 12158
    },
    {
      "epoch": 5.721882352941177,
      "grad_norm": 0.2777883698850972,
      "learning_rate": 2.231717292560825e-06,
      "loss": 0.0542,
      "step": 12159
    },
    {
      "epoch": 5.722352941176471,
      "grad_norm": 0.20068140664447945,
      "learning_rate": 2.224190043354135e-06,
      "loss": 0.0237,
      "step": 12160
    },
    {
      "epoch": 5.722823529411765,
      "grad_norm": 0.31412417412594257,
      "learning_rate": 2.2166754149264667e-06,
      "loss": 0.3281,
      "step": 12161
    },
    {
      "epoch": 5.723294117647059,
      "grad_norm": 0.4061628905739279,
      "learning_rate": 2.2091734079195565e-06,
      "loss": 0.4746,
      "step": 12162
    },
    {
      "epoch": 5.723764705882353,
      "grad_norm": 0.2627349440463651,
      "learning_rate": 2.2016840229741593e-06,
      "loss": 0.0933,
      "step": 12163
    },
    {
      "epoch": 5.724235294117647,
      "grad_norm": 0.3382016463659913,
      "learning_rate": 2.1942072607299135e-06,
      "loss": 0.1465,
      "step": 12164
    },
    {
      "epoch": 5.724705882352941,
      "grad_norm": 0.21752687240020077,
      "learning_rate": 2.1867431218253406e-06,
      "loss": 0.0781,
      "step": 12165
    },
    {
      "epoch": 5.725176470588235,
      "grad_norm": 0.3881789869673996,
      "learning_rate": 2.179291606897915e-06,
      "loss": 0.1768,
      "step": 12166
    },
    {
      "epoch": 5.7256470588235295,
      "grad_norm": 0.3663854936731289,
      "learning_rate": 2.171852716584044e-06,
      "loss": 0.0981,
      "step": 12167
    },
    {
      "epoch": 5.726117647058824,
      "grad_norm": 0.28268097724450486,
      "learning_rate": 2.1644264515190524e-06,
      "loss": 0.1396,
      "step": 12168
    },
    {
      "epoch": 5.726588235294118,
      "grad_norm": 0.45056803039162696,
      "learning_rate": 2.1570128123371667e-06,
      "loss": 0.3672,
      "step": 12169
    },
    {
      "epoch": 5.727058823529411,
      "grad_norm": 0.5465890699888611,
      "learning_rate": 2.1496117996715636e-06,
      "loss": 0.4297,
      "step": 12170
    },
    {
      "epoch": 5.7275294117647055,
      "grad_norm": 0.39042902154585757,
      "learning_rate": 2.142223414154287e-06,
      "loss": 0.064,
      "step": 12171
    },
    {
      "epoch": 5.728,
      "grad_norm": 0.6373874693544522,
      "learning_rate": 2.1348476564163997e-06,
      "loss": 0.3555,
      "step": 12172
    },
    {
      "epoch": 5.728470588235294,
      "grad_norm": 0.7699032783790786,
      "learning_rate": 2.1274845270877804e-06,
      "loss": 0.4961,
      "step": 12173
    },
    {
      "epoch": 5.728941176470588,
      "grad_norm": 0.39735084933877096,
      "learning_rate": 2.1201340267972933e-06,
      "loss": 0.3418,
      "step": 12174
    },
    {
      "epoch": 5.729411764705882,
      "grad_norm": 0.19553749238329876,
      "learning_rate": 2.1127961561727193e-06,
      "loss": 0.0479,
      "step": 12175
    },
    {
      "epoch": 5.729882352941177,
      "grad_norm": 0.20267544609263757,
      "learning_rate": 2.1054709158407235e-06,
      "loss": 0.1108,
      "step": 12176
    },
    {
      "epoch": 5.730352941176471,
      "grad_norm": 0.4788059196741195,
      "learning_rate": 2.0981583064269226e-06,
      "loss": 0.5078,
      "step": 12177
    },
    {
      "epoch": 5.730823529411765,
      "grad_norm": 0.3216796492472182,
      "learning_rate": 2.0908583285558834e-06,
      "loss": 0.3125,
      "step": 12178
    },
    {
      "epoch": 5.731294117647058,
      "grad_norm": 0.380709640394494,
      "learning_rate": 2.083570982851007e-06,
      "loss": 0.2676,
      "step": 12179
    },
    {
      "epoch": 5.731764705882353,
      "grad_norm": 0.3366337899261408,
      "learning_rate": 2.0762962699346965e-06,
      "loss": 0.3105,
      "step": 12180
    },
    {
      "epoch": 5.732235294117647,
      "grad_norm": 0.41945014082211746,
      "learning_rate": 2.06903419042827e-06,
      "loss": 0.4746,
      "step": 12181
    },
    {
      "epoch": 5.732705882352941,
      "grad_norm": 0.40605850805069876,
      "learning_rate": 2.0617847449519e-06,
      "loss": 0.1621,
      "step": 12182
    },
    {
      "epoch": 5.733176470588235,
      "grad_norm": 1.0273587099966666,
      "learning_rate": 2.054547934124756e-06,
      "loss": 0.5742,
      "step": 12183
    },
    {
      "epoch": 5.7336470588235295,
      "grad_norm": 0.3300016057920191,
      "learning_rate": 2.047323758564878e-06,
      "loss": 0.3223,
      "step": 12184
    },
    {
      "epoch": 5.734117647058824,
      "grad_norm": 0.4188754793445056,
      "learning_rate": 2.040112218889256e-06,
      "loss": 0.4062,
      "step": 12185
    },
    {
      "epoch": 5.734588235294118,
      "grad_norm": 0.41207522086293813,
      "learning_rate": 2.0329133157137965e-06,
      "loss": 0.4551,
      "step": 12186
    },
    {
      "epoch": 5.735058823529412,
      "grad_norm": 0.3350481692824548,
      "learning_rate": 2.0257270496533086e-06,
      "loss": 0.3242,
      "step": 12187
    },
    {
      "epoch": 5.7355294117647055,
      "grad_norm": 0.3512804221856021,
      "learning_rate": 2.0185534213215506e-06,
      "loss": 0.375,
      "step": 12188
    },
    {
      "epoch": 5.736,
      "grad_norm": 0.3098187871257601,
      "learning_rate": 2.011392431331166e-06,
      "loss": 0.2832,
      "step": 12189
    },
    {
      "epoch": 5.736470588235294,
      "grad_norm": 0.3938345255214308,
      "learning_rate": 2.0042440802937485e-06,
      "loss": 0.3047,
      "step": 12190
    },
    {
      "epoch": 5.736941176470588,
      "grad_norm": 0.3524106531013195,
      "learning_rate": 1.9971083688197943e-06,
      "loss": 0.2949,
      "step": 12191
    },
    {
      "epoch": 5.737411764705882,
      "grad_norm": 0.498437654960051,
      "learning_rate": 1.989985297518748e-06,
      "loss": 0.2637,
      "step": 12192
    },
    {
      "epoch": 5.737882352941177,
      "grad_norm": 0.44163602411233027,
      "learning_rate": 1.9828748669989405e-06,
      "loss": 0.3828,
      "step": 12193
    },
    {
      "epoch": 5.738352941176471,
      "grad_norm": 0.5471705828067926,
      "learning_rate": 1.9757770778676183e-06,
      "loss": 0.3242,
      "step": 12194
    },
    {
      "epoch": 5.738823529411764,
      "grad_norm": 0.49831703196053667,
      "learning_rate": 1.9686919307310145e-06,
      "loss": 0.3945,
      "step": 12195
    },
    {
      "epoch": 5.739294117647058,
      "grad_norm": 0.36163190238989634,
      "learning_rate": 1.961619426194194e-06,
      "loss": 0.3574,
      "step": 12196
    },
    {
      "epoch": 5.739764705882353,
      "grad_norm": 0.378856195857876,
      "learning_rate": 1.9545595648612078e-06,
      "loss": 0.1113,
      "step": 12197
    },
    {
      "epoch": 5.740235294117647,
      "grad_norm": 0.4739041359333475,
      "learning_rate": 1.9475123473349897e-06,
      "loss": 0.3848,
      "step": 12198
    },
    {
      "epoch": 5.740705882352941,
      "grad_norm": 0.42747491316344577,
      "learning_rate": 1.9404777742174083e-06,
      "loss": 0.5859,
      "step": 12199
    },
    {
      "epoch": 5.741176470588235,
      "grad_norm": 0.4911900365243532,
      "learning_rate": 1.9334558461092663e-06,
      "loss": 0.2012,
      "step": 12200
    },
    {
      "epoch": 5.7416470588235295,
      "grad_norm": 0.2598738586645569,
      "learning_rate": 1.9264465636102342e-06,
      "loss": 0.0518,
      "step": 12201
    },
    {
      "epoch": 5.742117647058824,
      "grad_norm": 0.49957922275466765,
      "learning_rate": 1.9194499273189834e-06,
      "loss": 0.3379,
      "step": 12202
    },
    {
      "epoch": 5.742588235294118,
      "grad_norm": 0.373732374012145,
      "learning_rate": 1.912465937833035e-06,
      "loss": 0.2373,
      "step": 12203
    },
    {
      "epoch": 5.743058823529411,
      "grad_norm": 0.32315760346366745,
      "learning_rate": 1.9054945957488466e-06,
      "loss": 0.2832,
      "step": 12204
    },
    {
      "epoch": 5.7435294117647056,
      "grad_norm": 0.31929274931218615,
      "learning_rate": 1.898535901661824e-06,
      "loss": 0.2812,
      "step": 12205
    },
    {
      "epoch": 5.744,
      "grad_norm": 0.6443833562707041,
      "learning_rate": 1.8915898561662923e-06,
      "loss": 0.7695,
      "step": 12206
    },
    {
      "epoch": 5.744470588235294,
      "grad_norm": 0.3007699492934762,
      "learning_rate": 1.884656459855427e-06,
      "loss": 0.1953,
      "step": 12207
    },
    {
      "epoch": 5.744941176470588,
      "grad_norm": 0.3302797188796825,
      "learning_rate": 1.8777357133214044e-06,
      "loss": 0.1025,
      "step": 12208
    },
    {
      "epoch": 5.745411764705882,
      "grad_norm": 0.31421454983574854,
      "learning_rate": 1.8708276171553017e-06,
      "loss": 0.0986,
      "step": 12209
    },
    {
      "epoch": 5.745882352941177,
      "grad_norm": 0.4007161970364061,
      "learning_rate": 1.86393217194708e-06,
      "loss": 0.3906,
      "step": 12210
    },
    {
      "epoch": 5.746352941176471,
      "grad_norm": 0.2913676813058001,
      "learning_rate": 1.857049378285652e-06,
      "loss": 0.082,
      "step": 12211
    },
    {
      "epoch": 5.746823529411765,
      "grad_norm": 0.4572406168462812,
      "learning_rate": 1.8501792367588642e-06,
      "loss": 0.1865,
      "step": 12212
    },
    {
      "epoch": 5.747294117647058,
      "grad_norm": 0.6169235480241669,
      "learning_rate": 1.84332174795343e-06,
      "loss": 0.3184,
      "step": 12213
    },
    {
      "epoch": 5.747764705882353,
      "grad_norm": 0.2852863596856651,
      "learning_rate": 1.8364769124550149e-06,
      "loss": 0.0723,
      "step": 12214
    },
    {
      "epoch": 5.748235294117647,
      "grad_norm": 0.48161516788129777,
      "learning_rate": 1.8296447308482176e-06,
      "loss": 0.1533,
      "step": 12215
    },
    {
      "epoch": 5.748705882352941,
      "grad_norm": 0.16035126165818375,
      "learning_rate": 1.8228252037165215e-06,
      "loss": 0.0248,
      "step": 12216
    },
    {
      "epoch": 5.749176470588235,
      "grad_norm": 0.4849660398638689,
      "learning_rate": 1.8160183316423605e-06,
      "loss": 0.1875,
      "step": 12217
    },
    {
      "epoch": 5.7496470588235296,
      "grad_norm": 0.33971562403564076,
      "learning_rate": 1.8092241152070696e-06,
      "loss": 0.2598,
      "step": 12218
    },
    {
      "epoch": 5.750117647058824,
      "grad_norm": 0.3031047652040683,
      "learning_rate": 1.8024425549909183e-06,
      "loss": 0.106,
      "step": 12219
    },
    {
      "epoch": 5.750588235294118,
      "grad_norm": 0.39083541962526586,
      "learning_rate": 1.795673651573093e-06,
      "loss": 0.2969,
      "step": 12220
    },
    {
      "epoch": 5.751058823529412,
      "grad_norm": 0.7885215896026442,
      "learning_rate": 1.788917405531648e-06,
      "loss": 0.4902,
      "step": 12221
    },
    {
      "epoch": 5.751529411764706,
      "grad_norm": 0.3635712752936827,
      "learning_rate": 1.7821738174436384e-06,
      "loss": 0.2656,
      "step": 12222
    },
    {
      "epoch": 5.752,
      "grad_norm": 0.4136673618084806,
      "learning_rate": 1.7754428878849868e-06,
      "loss": 0.4902,
      "step": 12223
    },
    {
      "epoch": 5.752470588235294,
      "grad_norm": 0.44853450381244875,
      "learning_rate": 1.7687246174305502e-06,
      "loss": 0.5938,
      "step": 12224
    },
    {
      "epoch": 5.752941176470588,
      "grad_norm": 0.3200816191447924,
      "learning_rate": 1.762019006654103e-06,
      "loss": 0.1055,
      "step": 12225
    },
    {
      "epoch": 5.7534117647058824,
      "grad_norm": 0.4136093175464655,
      "learning_rate": 1.7553260561283533e-06,
      "loss": 0.1543,
      "step": 12226
    },
    {
      "epoch": 5.753882352941177,
      "grad_norm": 0.14386547375520978,
      "learning_rate": 1.748645766424861e-06,
      "loss": 0.0391,
      "step": 12227
    },
    {
      "epoch": 5.754352941176471,
      "grad_norm": 1.0171690130323958,
      "learning_rate": 1.7419781381142028e-06,
      "loss": 0.9492,
      "step": 12228
    },
    {
      "epoch": 5.754823529411764,
      "grad_norm": 0.44333735027966287,
      "learning_rate": 1.7353231717658233e-06,
      "loss": 0.1602,
      "step": 12229
    },
    {
      "epoch": 5.7552941176470584,
      "grad_norm": 0.4030815126351391,
      "learning_rate": 1.728680867948068e-06,
      "loss": 0.1328,
      "step": 12230
    },
    {
      "epoch": 5.755764705882353,
      "grad_norm": 0.37208128017655784,
      "learning_rate": 1.7220512272282328e-06,
      "loss": 0.4629,
      "step": 12231
    },
    {
      "epoch": 5.756235294117647,
      "grad_norm": 0.3396533572566466,
      "learning_rate": 1.7154342501725316e-06,
      "loss": 0.1445,
      "step": 12232
    },
    {
      "epoch": 5.756705882352941,
      "grad_norm": 0.7478951239094775,
      "learning_rate": 1.708829937346079e-06,
      "loss": 0.6641,
      "step": 12233
    },
    {
      "epoch": 5.757176470588235,
      "grad_norm": 0.35678247601234314,
      "learning_rate": 1.7022382893129072e-06,
      "loss": 0.4473,
      "step": 12234
    },
    {
      "epoch": 5.75764705882353,
      "grad_norm": 0.4841587584090861,
      "learning_rate": 1.6956593066359825e-06,
      "loss": 0.3945,
      "step": 12235
    },
    {
      "epoch": 5.758117647058824,
      "grad_norm": 0.5369146799137331,
      "learning_rate": 1.6890929898771722e-06,
      "loss": 0.3457,
      "step": 12236
    },
    {
      "epoch": 5.758588235294118,
      "grad_norm": 0.3537419751776198,
      "learning_rate": 1.6825393395972942e-06,
      "loss": 0.2471,
      "step": 12237
    },
    {
      "epoch": 5.759058823529411,
      "grad_norm": 0.36495442372300424,
      "learning_rate": 1.6759983563560508e-06,
      "loss": 0.3184,
      "step": 12238
    },
    {
      "epoch": 5.759529411764706,
      "grad_norm": 0.4271616155515375,
      "learning_rate": 1.669470040712062e-06,
      "loss": 0.3301,
      "step": 12239
    },
    {
      "epoch": 5.76,
      "grad_norm": 0.3885771027974353,
      "learning_rate": 1.6629543932228817e-06,
      "loss": 0.5156,
      "step": 12240
    },
    {
      "epoch": 5.760470588235294,
      "grad_norm": 0.47142181136917843,
      "learning_rate": 1.6564514144449982e-06,
      "loss": 0.5391,
      "step": 12241
    },
    {
      "epoch": 5.760941176470588,
      "grad_norm": 0.4844385133850369,
      "learning_rate": 1.6499611049337835e-06,
      "loss": 0.4648,
      "step": 12242
    },
    {
      "epoch": 5.7614117647058825,
      "grad_norm": 0.4050433909317441,
      "learning_rate": 1.6434834652435447e-06,
      "loss": 0.4336,
      "step": 12243
    },
    {
      "epoch": 5.761882352941177,
      "grad_norm": 0.3605331216922862,
      "learning_rate": 1.6370184959275058e-06,
      "loss": 0.2539,
      "step": 12244
    },
    {
      "epoch": 5.762352941176471,
      "grad_norm": 0.3560628943765806,
      "learning_rate": 1.6305661975377916e-06,
      "loss": 0.1064,
      "step": 12245
    },
    {
      "epoch": 5.762823529411765,
      "grad_norm": 0.3407392666785784,
      "learning_rate": 1.6241265706254947e-06,
      "loss": 0.1934,
      "step": 12246
    },
    {
      "epoch": 5.7632941176470585,
      "grad_norm": 0.3523106292576035,
      "learning_rate": 1.617699615740542e-06,
      "loss": 0.4766,
      "step": 12247
    },
    {
      "epoch": 5.763764705882353,
      "grad_norm": 0.1936341358363392,
      "learning_rate": 1.6112853334318775e-06,
      "loss": 0.0806,
      "step": 12248
    },
    {
      "epoch": 5.764235294117647,
      "grad_norm": 0.4035305671696893,
      "learning_rate": 1.6048837242472634e-06,
      "loss": 0.4531,
      "step": 12249
    },
    {
      "epoch": 5.764705882352941,
      "grad_norm": 0.2159613254674797,
      "learning_rate": 1.598494788733462e-06,
      "loss": 0.0481,
      "step": 12250
    },
    {
      "epoch": 5.765176470588235,
      "grad_norm": 0.40253946939566837,
      "learning_rate": 1.5921185274361203e-06,
      "loss": 0.3105,
      "step": 12251
    },
    {
      "epoch": 5.76564705882353,
      "grad_norm": 0.5042120703023111,
      "learning_rate": 1.585754940899786e-06,
      "loss": 0.2559,
      "step": 12252
    },
    {
      "epoch": 5.766117647058824,
      "grad_norm": 0.24367914063164078,
      "learning_rate": 1.5794040296679411e-06,
      "loss": 0.1992,
      "step": 12253
    },
    {
      "epoch": 5.766588235294118,
      "grad_norm": 0.4301268404045702,
      "learning_rate": 1.5730657942830183e-06,
      "loss": 0.3633,
      "step": 12254
    },
    {
      "epoch": 5.767058823529412,
      "grad_norm": 0.49395243219644736,
      "learning_rate": 1.5667402352862846e-06,
      "loss": 0.1846,
      "step": 12255
    },
    {
      "epoch": 5.767529411764706,
      "grad_norm": 0.465144488255997,
      "learning_rate": 1.5604273532179913e-06,
      "loss": 0.2363,
      "step": 12256
    },
    {
      "epoch": 5.768,
      "grad_norm": 0.3281902684357561,
      "learning_rate": 1.554127148617307e-06,
      "loss": 0.1816,
      "step": 12257
    },
    {
      "epoch": 5.768470588235294,
      "grad_norm": 0.5674721923982169,
      "learning_rate": 1.5478396220222678e-06,
      "loss": 0.5352,
      "step": 12258
    },
    {
      "epoch": 5.768941176470588,
      "grad_norm": 0.43481139634994764,
      "learning_rate": 1.5415647739698944e-06,
      "loss": 0.2158,
      "step": 12259
    },
    {
      "epoch": 5.7694117647058825,
      "grad_norm": 0.3119674756553729,
      "learning_rate": 1.5353026049960748e-06,
      "loss": 0.0884,
      "step": 12260
    },
    {
      "epoch": 5.769882352941177,
      "grad_norm": 0.28611919129098484,
      "learning_rate": 1.5290531156356146e-06,
      "loss": 0.0884,
      "step": 12261
    },
    {
      "epoch": 5.770352941176471,
      "grad_norm": 0.4862713859342212,
      "learning_rate": 1.5228163064222698e-06,
      "loss": 0.2266,
      "step": 12262
    },
    {
      "epoch": 5.770823529411764,
      "grad_norm": 0.3286803213172474,
      "learning_rate": 1.516592177888698e-06,
      "loss": 0.1177,
      "step": 12263
    },
    {
      "epoch": 5.7712941176470585,
      "grad_norm": 0.4028110360934377,
      "learning_rate": 1.5103807305664573e-06,
      "loss": 0.3359,
      "step": 12264
    },
    {
      "epoch": 5.771764705882353,
      "grad_norm": 0.3457378127371209,
      "learning_rate": 1.5041819649860399e-06,
      "loss": 0.0481,
      "step": 12265
    },
    {
      "epoch": 5.772235294117647,
      "grad_norm": 0.4529498477690175,
      "learning_rate": 1.4979958816768556e-06,
      "loss": 0.3594,
      "step": 12266
    },
    {
      "epoch": 5.772705882352941,
      "grad_norm": 0.44001164709959034,
      "learning_rate": 1.4918224811672153e-06,
      "loss": 0.249,
      "step": 12267
    },
    {
      "epoch": 5.773176470588235,
      "grad_norm": 0.388472541172232,
      "learning_rate": 1.485661763984347e-06,
      "loss": 0.4395,
      "step": 12268
    },
    {
      "epoch": 5.77364705882353,
      "grad_norm": 0.2524655026203406,
      "learning_rate": 1.479513730654447e-06,
      "loss": 0.2002,
      "step": 12269
    },
    {
      "epoch": 5.774117647058824,
      "grad_norm": 0.35439088527160223,
      "learning_rate": 1.4733783817025613e-06,
      "loss": 0.4434,
      "step": 12270
    },
    {
      "epoch": 5.774588235294118,
      "grad_norm": 0.4638491351361612,
      "learning_rate": 1.4672557176526712e-06,
      "loss": 0.2451,
      "step": 12271
    },
    {
      "epoch": 5.775058823529411,
      "grad_norm": 0.4135838117398234,
      "learning_rate": 1.4611457390276915e-06,
      "loss": 0.1436,
      "step": 12272
    },
    {
      "epoch": 5.775529411764706,
      "grad_norm": 0.353469134645129,
      "learning_rate": 1.4550484463494384e-06,
      "loss": 0.1611,
      "step": 12273
    },
    {
      "epoch": 5.776,
      "grad_norm": 0.3070358192297051,
      "learning_rate": 1.4489638401386783e-06,
      "loss": 0.0869,
      "step": 12274
    },
    {
      "epoch": 5.776470588235294,
      "grad_norm": 0.2247434997524207,
      "learning_rate": 1.442891920915029e-06,
      "loss": 0.0723,
      "step": 12275
    },
    {
      "epoch": 5.776941176470588,
      "grad_norm": 0.3614323240973825,
      "learning_rate": 1.436832689197076e-06,
      "loss": 0.252,
      "step": 12276
    },
    {
      "epoch": 5.7774117647058825,
      "grad_norm": 0.38069895732730347,
      "learning_rate": 1.4307861455023218e-06,
      "loss": 0.1953,
      "step": 12277
    },
    {
      "epoch": 5.777882352941177,
      "grad_norm": 0.36635702723772734,
      "learning_rate": 1.4247522903471531e-06,
      "loss": 0.3496,
      "step": 12278
    },
    {
      "epoch": 5.778352941176471,
      "grad_norm": 0.335364448191792,
      "learning_rate": 1.4187311242468914e-06,
      "loss": 0.1147,
      "step": 12279
    },
    {
      "epoch": 5.778823529411765,
      "grad_norm": 0.5055870464413398,
      "learning_rate": 1.4127226477157915e-06,
      "loss": 0.3359,
      "step": 12280
    },
    {
      "epoch": 5.7792941176470585,
      "grad_norm": 0.34630082833285214,
      "learning_rate": 1.4067268612669769e-06,
      "loss": 0.3398,
      "step": 12281
    },
    {
      "epoch": 5.779764705882353,
      "grad_norm": 0.5573276301359334,
      "learning_rate": 1.4007437654125375e-06,
      "loss": 0.5156,
      "step": 12282
    },
    {
      "epoch": 5.780235294117647,
      "grad_norm": 0.3920472482171083,
      "learning_rate": 1.3947733606634647e-06,
      "loss": 0.2891,
      "step": 12283
    },
    {
      "epoch": 5.780705882352941,
      "grad_norm": 0.15760848257106358,
      "learning_rate": 1.388815647529634e-06,
      "loss": 0.0309,
      "step": 12284
    },
    {
      "epoch": 5.781176470588235,
      "grad_norm": 0.16803250878381673,
      "learning_rate": 1.3828706265199051e-06,
      "loss": 0.0547,
      "step": 12285
    },
    {
      "epoch": 5.78164705882353,
      "grad_norm": 0.3649310228891641,
      "learning_rate": 1.376938298141972e-06,
      "loss": 0.207,
      "step": 12286
    },
    {
      "epoch": 5.782117647058824,
      "grad_norm": 0.42835290565930756,
      "learning_rate": 1.3710186629025121e-06,
      "loss": 0.1504,
      "step": 12287
    },
    {
      "epoch": 5.782588235294118,
      "grad_norm": 0.42238503762905605,
      "learning_rate": 1.3651117213070716e-06,
      "loss": 0.2246,
      "step": 12288
    },
    {
      "epoch": 5.783058823529412,
      "grad_norm": 0.3810485777484728,
      "learning_rate": 1.3592174738601302e-06,
      "loss": 0.4531,
      "step": 12289
    },
    {
      "epoch": 5.783529411764706,
      "grad_norm": 0.9900521073509708,
      "learning_rate": 1.3533359210651184e-06,
      "loss": 0.4922,
      "step": 12290
    },
    {
      "epoch": 5.784,
      "grad_norm": 0.47734474926823545,
      "learning_rate": 1.3474670634243178e-06,
      "loss": 0.5078,
      "step": 12291
    },
    {
      "epoch": 5.784470588235294,
      "grad_norm": 0.43789708394778337,
      "learning_rate": 1.341610901438961e-06,
      "loss": 0.2178,
      "step": 12292
    },
    {
      "epoch": 5.784941176470588,
      "grad_norm": 0.4230909732583513,
      "learning_rate": 1.3357674356091974e-06,
      "loss": 0.168,
      "step": 12293
    },
    {
      "epoch": 5.7854117647058825,
      "grad_norm": 0.3444847864065954,
      "learning_rate": 1.3299366664340948e-06,
      "loss": 0.2305,
      "step": 12294
    },
    {
      "epoch": 5.785882352941177,
      "grad_norm": 0.36726173205253154,
      "learning_rate": 1.3241185944116217e-06,
      "loss": 0.4297,
      "step": 12295
    },
    {
      "epoch": 5.786352941176471,
      "grad_norm": 0.3703084137254245,
      "learning_rate": 1.3183132200386637e-06,
      "loss": 0.1074,
      "step": 12296
    },
    {
      "epoch": 5.786823529411764,
      "grad_norm": 0.39088613776888587,
      "learning_rate": 1.3125205438110408e-06,
      "loss": 0.3457,
      "step": 12297
    },
    {
      "epoch": 5.7872941176470585,
      "grad_norm": 0.48843621999915265,
      "learning_rate": 1.3067405662234743e-06,
      "loss": 0.3262,
      "step": 12298
    },
    {
      "epoch": 5.787764705882353,
      "grad_norm": 0.4175709961588167,
      "learning_rate": 1.3009732877696022e-06,
      "loss": 0.1758,
      "step": 12299
    },
    {
      "epoch": 5.788235294117647,
      "grad_norm": 0.4741567194070717,
      "learning_rate": 1.2952187089419642e-06,
      "loss": 0.5273,
      "step": 12300
    },
    {
      "epoch": 5.788705882352941,
      "grad_norm": 0.3019269470688543,
      "learning_rate": 1.28947683023205e-06,
      "loss": 0.2246,
      "step": 12301
    },
    {
      "epoch": 5.789176470588235,
      "grad_norm": 0.2707186123358767,
      "learning_rate": 1.2837476521302514e-06,
      "loss": 0.1426,
      "step": 12302
    },
    {
      "epoch": 5.78964705882353,
      "grad_norm": 0.24728697777371875,
      "learning_rate": 1.2780311751258265e-06,
      "loss": 0.0654,
      "step": 12303
    },
    {
      "epoch": 5.790117647058824,
      "grad_norm": 0.531052628504151,
      "learning_rate": 1.2723273997070182e-06,
      "loss": 0.5039,
      "step": 12304
    },
    {
      "epoch": 5.790588235294118,
      "grad_norm": 0.4651245882061886,
      "learning_rate": 1.2666363263609702e-06,
      "loss": 0.4863,
      "step": 12305
    },
    {
      "epoch": 5.791058823529411,
      "grad_norm": 0.6735265139379684,
      "learning_rate": 1.2609579555737104e-06,
      "loss": 0.5391,
      "step": 12306
    },
    {
      "epoch": 5.791529411764706,
      "grad_norm": 0.5848904046828256,
      "learning_rate": 1.2552922878301841e-06,
      "loss": 0.3008,
      "step": 12307
    },
    {
      "epoch": 5.792,
      "grad_norm": 0.27270229692637465,
      "learning_rate": 1.2496393236143044e-06,
      "loss": 0.1328,
      "step": 12308
    },
    {
      "epoch": 5.792470588235294,
      "grad_norm": 0.11531237486576057,
      "learning_rate": 1.2439990634088348e-06,
      "loss": 0.0122,
      "step": 12309
    },
    {
      "epoch": 5.792941176470588,
      "grad_norm": 0.23831305146000153,
      "learning_rate": 1.2383715076954902e-06,
      "loss": 0.1147,
      "step": 12310
    },
    {
      "epoch": 5.7934117647058825,
      "grad_norm": 0.39224664466208836,
      "learning_rate": 1.232756656954903e-06,
      "loss": 0.1436,
      "step": 12311
    },
    {
      "epoch": 5.793882352941177,
      "grad_norm": 0.3626037203011964,
      "learning_rate": 1.2271545116665727e-06,
      "loss": 0.3125,
      "step": 12312
    },
    {
      "epoch": 5.794352941176471,
      "grad_norm": 0.3596176155310867,
      "learning_rate": 1.2215650723089832e-06,
      "loss": 0.2559,
      "step": 12313
    },
    {
      "epoch": 5.794823529411765,
      "grad_norm": 0.5023242825207779,
      "learning_rate": 1.2159883393595026e-06,
      "loss": 0.207,
      "step": 12314
    },
    {
      "epoch": 5.7952941176470585,
      "grad_norm": 0.2149702669274543,
      "learning_rate": 1.2104243132943836e-06,
      "loss": 0.1099,
      "step": 12315
    },
    {
      "epoch": 5.795764705882353,
      "grad_norm": 0.20154986528911853,
      "learning_rate": 1.2048729945888458e-06,
      "loss": 0.0264,
      "step": 12316
    },
    {
      "epoch": 5.796235294117647,
      "grad_norm": 0.2972117047831132,
      "learning_rate": 1.1993343837169933e-06,
      "loss": 0.1357,
      "step": 12317
    },
    {
      "epoch": 5.796705882352941,
      "grad_norm": 0.35547097111235026,
      "learning_rate": 1.1938084811518311e-06,
      "loss": 0.2656,
      "step": 12318
    },
    {
      "epoch": 5.797176470588235,
      "grad_norm": 0.7133933722472655,
      "learning_rate": 1.1882952873653317e-06,
      "loss": 0.249,
      "step": 12319
    },
    {
      "epoch": 5.79764705882353,
      "grad_norm": 0.19938341772749898,
      "learning_rate": 1.1827948028283352e-06,
      "loss": 0.0757,
      "step": 12320
    },
    {
      "epoch": 5.798117647058824,
      "grad_norm": 0.3822886377719729,
      "learning_rate": 1.1773070280105822e-06,
      "loss": 0.1924,
      "step": 12321
    },
    {
      "epoch": 5.798588235294117,
      "grad_norm": 0.2935382287007771,
      "learning_rate": 1.171831963380815e-06,
      "loss": 0.1074,
      "step": 12322
    },
    {
      "epoch": 5.799058823529411,
      "grad_norm": 0.367292745652526,
      "learning_rate": 1.1663696094065755e-06,
      "loss": 0.1895,
      "step": 12323
    },
    {
      "epoch": 5.799529411764706,
      "grad_norm": 0.3052324879074981,
      "learning_rate": 1.160919966554391e-06,
      "loss": 0.2539,
      "step": 12324
    },
    {
      "epoch": 5.8,
      "grad_norm": 0.31663029213474325,
      "learning_rate": 1.155483035289706e-06,
      "loss": 0.0825,
      "step": 12325
    },
    {
      "epoch": 5.800470588235294,
      "grad_norm": 0.4028587177972297,
      "learning_rate": 1.150058816076832e-06,
      "loss": 0.2471,
      "step": 12326
    },
    {
      "epoch": 5.800941176470588,
      "grad_norm": 0.5479281737432663,
      "learning_rate": 1.1446473093790487e-06,
      "loss": 0.2021,
      "step": 12327
    },
    {
      "epoch": 5.8014117647058825,
      "grad_norm": 0.3505901452563672,
      "learning_rate": 1.1392485156585196e-06,
      "loss": 0.2236,
      "step": 12328
    },
    {
      "epoch": 5.801882352941177,
      "grad_norm": 0.5945486038705329,
      "learning_rate": 1.1338624353763093e-06,
      "loss": 0.3145,
      "step": 12329
    },
    {
      "epoch": 5.802352941176471,
      "grad_norm": 0.24921405215486508,
      "learning_rate": 1.1284890689924497e-06,
      "loss": 0.0466,
      "step": 12330
    },
    {
      "epoch": 5.802823529411764,
      "grad_norm": 0.44619182837371885,
      "learning_rate": 1.1231284169658239e-06,
      "loss": 0.2852,
      "step": 12331
    },
    {
      "epoch": 5.8032941176470585,
      "grad_norm": 0.28032475451234323,
      "learning_rate": 1.1177804797542821e-06,
      "loss": 0.1592,
      "step": 12332
    },
    {
      "epoch": 5.803764705882353,
      "grad_norm": 0.2782043340901861,
      "learning_rate": 1.1124452578145425e-06,
      "loss": 0.0664,
      "step": 12333
    },
    {
      "epoch": 5.804235294117647,
      "grad_norm": 0.36640999609958486,
      "learning_rate": 1.1071227516022574e-06,
      "loss": 0.2656,
      "step": 12334
    },
    {
      "epoch": 5.804705882352941,
      "grad_norm": 0.30371293555619366,
      "learning_rate": 1.1018129615720129e-06,
      "loss": 0.1768,
      "step": 12335
    },
    {
      "epoch": 5.805176470588235,
      "grad_norm": 0.40497860689182896,
      "learning_rate": 1.0965158881772961e-06,
      "loss": 0.3203,
      "step": 12336
    },
    {
      "epoch": 5.80564705882353,
      "grad_norm": 0.4189394240465857,
      "learning_rate": 1.091231531870479e-06,
      "loss": 0.4102,
      "step": 12337
    },
    {
      "epoch": 5.806117647058824,
      "grad_norm": 0.22323771660016734,
      "learning_rate": 1.0859598931028668e-06,
      "loss": 0.1099,
      "step": 12338
    },
    {
      "epoch": 5.806588235294118,
      "grad_norm": 0.3244998117221702,
      "learning_rate": 1.0807009723247328e-06,
      "loss": 0.1104,
      "step": 12339
    },
    {
      "epoch": 5.807058823529411,
      "grad_norm": 0.2582421305098902,
      "learning_rate": 1.0754547699851513e-06,
      "loss": 0.0605,
      "step": 12340
    },
    {
      "epoch": 5.807529411764706,
      "grad_norm": 0.44562781737516505,
      "learning_rate": 1.0702212865322135e-06,
      "loss": 0.6758,
      "step": 12341
    },
    {
      "epoch": 5.808,
      "grad_norm": 0.38462172734983563,
      "learning_rate": 1.0650005224128788e-06,
      "loss": 0.3555,
      "step": 12342
    },
    {
      "epoch": 5.808470588235294,
      "grad_norm": 0.3356774683750225,
      "learning_rate": 1.059792478073007e-06,
      "loss": 0.1377,
      "step": 12343
    },
    {
      "epoch": 5.808941176470588,
      "grad_norm": 0.5493669878859679,
      "learning_rate": 1.0545971539574094e-06,
      "loss": 0.373,
      "step": 12344
    },
    {
      "epoch": 5.8094117647058825,
      "grad_norm": 0.3692312602236156,
      "learning_rate": 1.0494145505097973e-06,
      "loss": 0.252,
      "step": 12345
    },
    {
      "epoch": 5.809882352941177,
      "grad_norm": 0.27053012032793544,
      "learning_rate": 1.0442446681727834e-06,
      "loss": 0.0757,
      "step": 12346
    },
    {
      "epoch": 5.810352941176471,
      "grad_norm": 0.4378254363351816,
      "learning_rate": 1.0390875073878812e-06,
      "loss": 0.1797,
      "step": 12347
    },
    {
      "epoch": 5.810823529411765,
      "grad_norm": 0.3655270357336111,
      "learning_rate": 1.0339430685955719e-06,
      "loss": 0.2363,
      "step": 12348
    },
    {
      "epoch": 5.8112941176470585,
      "grad_norm": 0.2379355132333609,
      "learning_rate": 1.0288113522352038e-06,
      "loss": 0.0444,
      "step": 12349
    },
    {
      "epoch": 5.811764705882353,
      "grad_norm": 0.43975902450788357,
      "learning_rate": 1.0236923587450263e-06,
      "loss": 0.3203,
      "step": 12350
    },
    {
      "epoch": 5.812235294117647,
      "grad_norm": 0.2977076263271049,
      "learning_rate": 1.0185860885622565e-06,
      "loss": 0.1011,
      "step": 12351
    },
    {
      "epoch": 5.812705882352941,
      "grad_norm": 0.42540499458213604,
      "learning_rate": 1.013492542122979e-06,
      "loss": 0.2402,
      "step": 12352
    },
    {
      "epoch": 5.813176470588235,
      "grad_norm": 0.41492715890575615,
      "learning_rate": 1.0084117198622287e-06,
      "loss": 0.1709,
      "step": 12353
    },
    {
      "epoch": 5.81364705882353,
      "grad_norm": 0.6022787960854121,
      "learning_rate": 1.0033436222138924e-06,
      "loss": 0.3672,
      "step": 12354
    },
    {
      "epoch": 5.814117647058824,
      "grad_norm": 0.4315382616098601,
      "learning_rate": 9.98288249610857e-07,
      "loss": 0.2969,
      "step": 12355
    },
    {
      "epoch": 5.814588235294117,
      "grad_norm": 0.32370646470491127,
      "learning_rate": 9.932456024848435e-07,
      "loss": 0.0645,
      "step": 12356
    },
    {
      "epoch": 5.815058823529411,
      "grad_norm": 0.2319340288416996,
      "learning_rate": 9.882156812665244e-07,
      "loss": 0.0688,
      "step": 12357
    },
    {
      "epoch": 5.815529411764706,
      "grad_norm": 0.19959220037456024,
      "learning_rate": 9.831984863854891e-07,
      "loss": 0.0447,
      "step": 12358
    },
    {
      "epoch": 5.816,
      "grad_norm": 0.23519657434260546,
      "learning_rate": 9.781940182702451e-07,
      "loss": 0.1069,
      "step": 12359
    },
    {
      "epoch": 5.816470588235294,
      "grad_norm": 0.18306565084859852,
      "learning_rate": 9.732022773481506e-07,
      "loss": 0.0332,
      "step": 12360
    },
    {
      "epoch": 5.816941176470588,
      "grad_norm": 4.382447371222967,
      "learning_rate": 9.682232640455644e-07,
      "loss": 0.6602,
      "step": 12361
    },
    {
      "epoch": 5.8174117647058825,
      "grad_norm": 0.4705930517698247,
      "learning_rate": 9.632569787877298e-07,
      "loss": 0.3711,
      "step": 12362
    },
    {
      "epoch": 5.817882352941177,
      "grad_norm": 0.21623973596760113,
      "learning_rate": 9.583034219987406e-07,
      "loss": 0.0742,
      "step": 12363
    },
    {
      "epoch": 5.818352941176471,
      "grad_norm": 0.4653322509594326,
      "learning_rate": 9.53362594101692e-07,
      "loss": 0.4609,
      "step": 12364
    },
    {
      "epoch": 5.818823529411764,
      "grad_norm": 0.2400706449369197,
      "learning_rate": 9.48434495518563e-07,
      "loss": 0.0554,
      "step": 12365
    },
    {
      "epoch": 5.8192941176470585,
      "grad_norm": 0.44982607065286484,
      "learning_rate": 9.435191266702003e-07,
      "loss": 0.1836,
      "step": 12366
    },
    {
      "epoch": 5.819764705882353,
      "grad_norm": 0.20730225186831838,
      "learning_rate": 9.386164879764346e-07,
      "loss": 0.0791,
      "step": 12367
    },
    {
      "epoch": 5.820235294117647,
      "grad_norm": 0.49326980270423687,
      "learning_rate": 9.337265798559645e-07,
      "loss": 0.4453,
      "step": 12368
    },
    {
      "epoch": 5.820705882352941,
      "grad_norm": 0.40483982829415355,
      "learning_rate": 9.28849402726406e-07,
      "loss": 0.3281,
      "step": 12369
    },
    {
      "epoch": 5.821176470588235,
      "grad_norm": 0.250326405807143,
      "learning_rate": 9.239849570042924e-07,
      "loss": 0.0549,
      "step": 12370
    },
    {
      "epoch": 5.82164705882353,
      "grad_norm": 0.5638952435984346,
      "learning_rate": 9.191332431050747e-07,
      "loss": 0.4883,
      "step": 12371
    },
    {
      "epoch": 5.822117647058824,
      "grad_norm": 0.4035283301457708,
      "learning_rate": 9.142942614431215e-07,
      "loss": 0.3438,
      "step": 12372
    },
    {
      "epoch": 5.822588235294118,
      "grad_norm": 0.3708643528042941,
      "learning_rate": 9.094680124317022e-07,
      "loss": 0.3438,
      "step": 12373
    },
    {
      "epoch": 5.823058823529411,
      "grad_norm": 0.36881676322158397,
      "learning_rate": 9.046544964830038e-07,
      "loss": 0.4766,
      "step": 12374
    },
    {
      "epoch": 5.823529411764706,
      "grad_norm": 0.410785626184861,
      "learning_rate": 8.998537140081141e-07,
      "loss": 0.1367,
      "step": 12375
    },
    {
      "epoch": 5.824,
      "grad_norm": 0.3790146638993453,
      "learning_rate": 8.950656654170552e-07,
      "loss": 0.4219,
      "step": 12376
    },
    {
      "epoch": 5.824470588235294,
      "grad_norm": 0.33807096893554317,
      "learning_rate": 8.902903511187332e-07,
      "loss": 0.2393,
      "step": 12377
    },
    {
      "epoch": 5.824941176470588,
      "grad_norm": 0.3728356133362803,
      "learning_rate": 8.855277715210052e-07,
      "loss": 0.1289,
      "step": 12378
    },
    {
      "epoch": 5.8254117647058825,
      "grad_norm": 0.29895183812747167,
      "learning_rate": 8.807779270305959e-07,
      "loss": 0.1465,
      "step": 12379
    },
    {
      "epoch": 5.825882352941177,
      "grad_norm": 0.5240717609519623,
      "learning_rate": 8.760408180531809e-07,
      "loss": 0.4531,
      "step": 12380
    },
    {
      "epoch": 5.826352941176471,
      "grad_norm": 0.3066774205888695,
      "learning_rate": 8.713164449933364e-07,
      "loss": 0.0737,
      "step": 12381
    },
    {
      "epoch": 5.826823529411765,
      "grad_norm": 0.5073535949522452,
      "learning_rate": 8.666048082545396e-07,
      "loss": 0.5625,
      "step": 12382
    },
    {
      "epoch": 5.8272941176470585,
      "grad_norm": 0.8440514645379583,
      "learning_rate": 8.619059082391689e-07,
      "loss": 0.6328,
      "step": 12383
    },
    {
      "epoch": 5.827764705882353,
      "grad_norm": 0.404326170980725,
      "learning_rate": 8.572197453485696e-07,
      "loss": 0.543,
      "step": 12384
    },
    {
      "epoch": 5.828235294117647,
      "grad_norm": 0.2295495509127623,
      "learning_rate": 8.525463199829385e-07,
      "loss": 0.0535,
      "step": 12385
    },
    {
      "epoch": 5.828705882352941,
      "grad_norm": 0.35302951346286504,
      "learning_rate": 8.47885632541423e-07,
      "loss": 0.1094,
      "step": 12386
    },
    {
      "epoch": 5.829176470588235,
      "grad_norm": 0.15136951566776946,
      "learning_rate": 8.432376834220544e-07,
      "loss": 0.028,
      "step": 12387
    },
    {
      "epoch": 5.82964705882353,
      "grad_norm": 0.4788793670718509,
      "learning_rate": 8.386024730217988e-07,
      "loss": 0.2207,
      "step": 12388
    },
    {
      "epoch": 5.830117647058824,
      "grad_norm": 0.39466729735135203,
      "learning_rate": 8.339800017365228e-07,
      "loss": 0.1152,
      "step": 12389
    },
    {
      "epoch": 5.830588235294117,
      "grad_norm": 0.4820375259293311,
      "learning_rate": 8.293702699610272e-07,
      "loss": 0.1377,
      "step": 12390
    },
    {
      "epoch": 5.831058823529411,
      "grad_norm": 0.31849702562196486,
      "learning_rate": 8.247732780889637e-07,
      "loss": 0.2793,
      "step": 12391
    },
    {
      "epoch": 5.831529411764706,
      "grad_norm": 0.2414684233129698,
      "learning_rate": 8.201890265129684e-07,
      "loss": 0.0483,
      "step": 12392
    },
    {
      "epoch": 5.832,
      "grad_norm": 0.29896423807057665,
      "learning_rate": 8.156175156245614e-07,
      "loss": 0.1187,
      "step": 12393
    },
    {
      "epoch": 5.832470588235294,
      "grad_norm": 0.5613987791183384,
      "learning_rate": 8.110587458141637e-07,
      "loss": 0.2676,
      "step": 12394
    },
    {
      "epoch": 5.832941176470588,
      "grad_norm": 0.4904758188013777,
      "learning_rate": 8.065127174711139e-07,
      "loss": 0.4434,
      "step": 12395
    },
    {
      "epoch": 5.8334117647058825,
      "grad_norm": 0.23675282485671884,
      "learning_rate": 8.01979430983668e-07,
      "loss": 0.0688,
      "step": 12396
    },
    {
      "epoch": 5.833882352941177,
      "grad_norm": 0.2800956833600361,
      "learning_rate": 7.97458886738983e-07,
      "loss": 0.1445,
      "step": 12397
    },
    {
      "epoch": 5.834352941176471,
      "grad_norm": 0.4476237318861145,
      "learning_rate": 7.929510851231502e-07,
      "loss": 0.2236,
      "step": 12398
    },
    {
      "epoch": 5.834823529411764,
      "grad_norm": 0.3504853890796166,
      "learning_rate": 7.884560265211615e-07,
      "loss": 0.168,
      "step": 12399
    },
    {
      "epoch": 5.8352941176470585,
      "grad_norm": 0.25659183866510454,
      "learning_rate": 7.839737113168931e-07,
      "loss": 0.105,
      "step": 12400
    },
    {
      "epoch": 5.835764705882353,
      "grad_norm": 0.4874642492286901,
      "learning_rate": 7.795041398931723e-07,
      "loss": 0.3066,
      "step": 12401
    },
    {
      "epoch": 5.836235294117647,
      "grad_norm": 0.49729914002444003,
      "learning_rate": 7.750473126317269e-07,
      "loss": 0.4414,
      "step": 12402
    },
    {
      "epoch": 5.836705882352941,
      "grad_norm": 0.4084188698032202,
      "learning_rate": 7.706032299131692e-07,
      "loss": 0.2871,
      "step": 12403
    },
    {
      "epoch": 5.837176470588235,
      "grad_norm": 0.41990179676763845,
      "learning_rate": 7.66171892117079e-07,
      "loss": 0.2559,
      "step": 12404
    },
    {
      "epoch": 5.83764705882353,
      "grad_norm": 0.5426183109526584,
      "learning_rate": 7.617532996218867e-07,
      "loss": 0.457,
      "step": 12405
    },
    {
      "epoch": 5.838117647058824,
      "grad_norm": 0.4600656494403717,
      "learning_rate": 7.573474528049739e-07,
      "loss": 0.334,
      "step": 12406
    },
    {
      "epoch": 5.838588235294118,
      "grad_norm": 0.3284002505150242,
      "learning_rate": 7.529543520426228e-07,
      "loss": 0.0781,
      "step": 12407
    },
    {
      "epoch": 5.839058823529411,
      "grad_norm": 0.39792783546265864,
      "learning_rate": 7.485739977100336e-07,
      "loss": 0.1582,
      "step": 12408
    },
    {
      "epoch": 5.839529411764706,
      "grad_norm": 0.4485664157427572,
      "learning_rate": 7.442063901812734e-07,
      "loss": 0.3906,
      "step": 12409
    },
    {
      "epoch": 5.84,
      "grad_norm": 0.41303146911897237,
      "learning_rate": 7.398515298294106e-07,
      "loss": 0.332,
      "step": 12410
    },
    {
      "epoch": 5.840470588235294,
      "grad_norm": 0.333004023107966,
      "learning_rate": 7.355094170263143e-07,
      "loss": 0.1309,
      "step": 12411
    },
    {
      "epoch": 5.840941176470588,
      "grad_norm": 0.3524833084054678,
      "learning_rate": 7.311800521428712e-07,
      "loss": 0.2891,
      "step": 12412
    },
    {
      "epoch": 5.8414117647058825,
      "grad_norm": 0.4482633680111588,
      "learning_rate": 7.26863435548819e-07,
      "loss": 0.1846,
      "step": 12413
    },
    {
      "epoch": 5.841882352941177,
      "grad_norm": 0.33479868299086046,
      "learning_rate": 7.22559567612796e-07,
      "loss": 0.2119,
      "step": 12414
    },
    {
      "epoch": 5.842352941176471,
      "grad_norm": 0.4707952204297293,
      "learning_rate": 7.182684487023748e-07,
      "loss": 0.2031,
      "step": 12415
    },
    {
      "epoch": 5.842823529411765,
      "grad_norm": 0.6946269457901985,
      "learning_rate": 7.139900791840792e-07,
      "loss": 0.4375,
      "step": 12416
    },
    {
      "epoch": 5.8432941176470585,
      "grad_norm": 0.552162694991924,
      "learning_rate": 7.097244594232498e-07,
      "loss": 0.6133,
      "step": 12417
    },
    {
      "epoch": 5.843764705882353,
      "grad_norm": 0.5685693627782803,
      "learning_rate": 7.054715897842289e-07,
      "loss": 0.4766,
      "step": 12418
    },
    {
      "epoch": 5.844235294117647,
      "grad_norm": 0.4393053285975512,
      "learning_rate": 7.012314706302257e-07,
      "loss": 0.1943,
      "step": 12419
    },
    {
      "epoch": 5.844705882352941,
      "grad_norm": 0.40944228549925066,
      "learning_rate": 6.970041023233508e-07,
      "loss": 0.5,
      "step": 12420
    },
    {
      "epoch": 5.845176470588235,
      "grad_norm": 0.49447876523727496,
      "learning_rate": 6.927894852246485e-07,
      "loss": 0.4219,
      "step": 12421
    },
    {
      "epoch": 5.84564705882353,
      "grad_norm": 0.20877213321040883,
      "learning_rate": 6.88587619694081e-07,
      "loss": 0.0226,
      "step": 12422
    },
    {
      "epoch": 5.846117647058824,
      "grad_norm": 0.4743467045121629,
      "learning_rate": 6.843985060904944e-07,
      "loss": 0.5117,
      "step": 12423
    },
    {
      "epoch": 5.846588235294117,
      "grad_norm": 0.5963104933094433,
      "learning_rate": 6.802221447716694e-07,
      "loss": 0.416,
      "step": 12424
    },
    {
      "epoch": 5.847058823529411,
      "grad_norm": 0.38215335655080407,
      "learning_rate": 6.760585360942872e-07,
      "loss": 0.2021,
      "step": 12425
    },
    {
      "epoch": 5.847529411764706,
      "grad_norm": 0.5310217706846843,
      "learning_rate": 6.719076804139134e-07,
      "loss": 0.457,
      "step": 12426
    },
    {
      "epoch": 5.848,
      "grad_norm": 0.3461746356133464,
      "learning_rate": 6.677695780850978e-07,
      "loss": 0.2578,
      "step": 12427
    },
    {
      "epoch": 5.848470588235294,
      "grad_norm": 0.3965709250726416,
      "learning_rate": 6.636442294612243e-07,
      "loss": 0.1748,
      "step": 12428
    },
    {
      "epoch": 5.848941176470588,
      "grad_norm": 0.29893231681879495,
      "learning_rate": 6.595316348946278e-07,
      "loss": 0.2422,
      "step": 12429
    },
    {
      "epoch": 5.8494117647058825,
      "grad_norm": 0.3608399248169828,
      "learning_rate": 6.554317947365273e-07,
      "loss": 0.1143,
      "step": 12430
    },
    {
      "epoch": 5.849882352941177,
      "grad_norm": 0.3741252996941443,
      "learning_rate": 6.513447093370927e-07,
      "loss": 0.084,
      "step": 12431
    },
    {
      "epoch": 5.850352941176471,
      "grad_norm": 0.3281528974316796,
      "learning_rate": 6.472703790453781e-07,
      "loss": 0.3164,
      "step": 12432
    },
    {
      "epoch": 5.850823529411764,
      "grad_norm": 1.1096729598839399,
      "learning_rate": 6.43208804209322e-07,
      "loss": 0.5352,
      "step": 12433
    },
    {
      "epoch": 5.8512941176470585,
      "grad_norm": 0.40776834654019806,
      "learning_rate": 6.391599851758466e-07,
      "loss": 0.3809,
      "step": 12434
    },
    {
      "epoch": 5.851764705882353,
      "grad_norm": 0.3578303606196977,
      "learning_rate": 6.351239222907089e-07,
      "loss": 0.1592,
      "step": 12435
    },
    {
      "epoch": 5.852235294117647,
      "grad_norm": 0.3088499284510898,
      "learning_rate": 6.311006158986165e-07,
      "loss": 0.1299,
      "step": 12436
    },
    {
      "epoch": 5.852705882352941,
      "grad_norm": 0.558665428484306,
      "learning_rate": 6.270900663431777e-07,
      "loss": 0.4199,
      "step": 12437
    },
    {
      "epoch": 5.853176470588235,
      "grad_norm": 0.30295135600851403,
      "learning_rate": 6.230922739669353e-07,
      "loss": 0.0542,
      "step": 12438
    },
    {
      "epoch": 5.85364705882353,
      "grad_norm": 0.8330910506839578,
      "learning_rate": 6.191072391112828e-07,
      "loss": 0.7812,
      "step": 12439
    },
    {
      "epoch": 5.854117647058824,
      "grad_norm": 0.2350625323809111,
      "learning_rate": 6.151349621165979e-07,
      "loss": 0.1465,
      "step": 12440
    },
    {
      "epoch": 5.854588235294118,
      "grad_norm": 0.378765487080045,
      "learning_rate": 6.111754433221094e-07,
      "loss": 0.3418,
      "step": 12441
    },
    {
      "epoch": 5.855058823529411,
      "grad_norm": 0.4393197640593026,
      "learning_rate": 6.0722868306598e-07,
      "loss": 0.2305,
      "step": 12442
    },
    {
      "epoch": 5.855529411764706,
      "grad_norm": 0.4726024565058008,
      "learning_rate": 6.032946816852902e-07,
      "loss": 0.208,
      "step": 12443
    },
    {
      "epoch": 5.856,
      "grad_norm": 0.4806901727763943,
      "learning_rate": 5.99373439516021e-07,
      "loss": 0.2539,
      "step": 12444
    },
    {
      "epoch": 5.856470588235294,
      "grad_norm": 0.47555949125530705,
      "learning_rate": 5.954649568930714e-07,
      "loss": 0.2969,
      "step": 12445
    },
    {
      "epoch": 5.856941176470588,
      "grad_norm": 0.533499822787725,
      "learning_rate": 5.91569234150241e-07,
      "loss": 0.5664,
      "step": 12446
    },
    {
      "epoch": 5.8574117647058825,
      "grad_norm": 0.3574360178918285,
      "learning_rate": 5.876862716202302e-07,
      "loss": 0.291,
      "step": 12447
    },
    {
      "epoch": 5.857882352941177,
      "grad_norm": 0.39875186863448575,
      "learning_rate": 5.838160696346906e-07,
      "loss": 0.3145,
      "step": 12448
    },
    {
      "epoch": 5.85835294117647,
      "grad_norm": 0.5008583733156158,
      "learning_rate": 5.799586285241242e-07,
      "loss": 0.5117,
      "step": 12449
    },
    {
      "epoch": 5.858823529411764,
      "grad_norm": 0.4243966310862799,
      "learning_rate": 5.761139486180178e-07,
      "loss": 0.2256,
      "step": 12450
    },
    {
      "epoch": 5.8592941176470585,
      "grad_norm": 0.42997533798184434,
      "learning_rate": 5.722820302446751e-07,
      "loss": 0.4219,
      "step": 12451
    },
    {
      "epoch": 5.859764705882353,
      "grad_norm": 0.38135415566062353,
      "learning_rate": 5.684628737314012e-07,
      "loss": 0.2656,
      "step": 12452
    },
    {
      "epoch": 5.860235294117647,
      "grad_norm": 0.7600661207999124,
      "learning_rate": 5.646564794043518e-07,
      "loss": 0.2334,
      "step": 12453
    },
    {
      "epoch": 5.860705882352941,
      "grad_norm": 0.3883000663314469,
      "learning_rate": 5.608628475886001e-07,
      "loss": 0.1348,
      "step": 12454
    },
    {
      "epoch": 5.861176470588235,
      "grad_norm": 0.48977896551769623,
      "learning_rate": 5.570819786081704e-07,
      "loss": 0.209,
      "step": 12455
    },
    {
      "epoch": 5.86164705882353,
      "grad_norm": 0.34389396103545,
      "learning_rate": 5.533138727859543e-07,
      "loss": 0.0967,
      "step": 12456
    },
    {
      "epoch": 5.862117647058824,
      "grad_norm": 0.35598719992145617,
      "learning_rate": 5.495585304437611e-07,
      "loss": 0.1807,
      "step": 12457
    },
    {
      "epoch": 5.862588235294117,
      "grad_norm": 0.4867887918068567,
      "learning_rate": 5.458159519023175e-07,
      "loss": 0.3496,
      "step": 12458
    },
    {
      "epoch": 5.863058823529411,
      "grad_norm": 0.3186331491364269,
      "learning_rate": 5.42086137481268e-07,
      "loss": 0.2109,
      "step": 12459
    },
    {
      "epoch": 5.863529411764706,
      "grad_norm": 0.2995087311424379,
      "learning_rate": 5.383690874991408e-07,
      "loss": 0.3145,
      "step": 12460
    },
    {
      "epoch": 5.864,
      "grad_norm": 0.24306409734767345,
      "learning_rate": 5.34664802273399e-07,
      "loss": 0.051,
      "step": 12461
    },
    {
      "epoch": 5.864470588235294,
      "grad_norm": 0.47489744419697555,
      "learning_rate": 5.30973282120406e-07,
      "loss": 0.3711,
      "step": 12462
    },
    {
      "epoch": 5.864941176470588,
      "grad_norm": 0.6442600186657895,
      "learning_rate": 5.272945273554429e-07,
      "loss": 0.2461,
      "step": 12463
    },
    {
      "epoch": 5.8654117647058825,
      "grad_norm": 0.3279300566835399,
      "learning_rate": 5.236285382926752e-07,
      "loss": 0.2266,
      "step": 12464
    },
    {
      "epoch": 5.865882352941177,
      "grad_norm": 0.1475184370053564,
      "learning_rate": 5.19975315245219e-07,
      "loss": 0.0204,
      "step": 12465
    },
    {
      "epoch": 5.866352941176471,
      "grad_norm": 0.2751598269458572,
      "learning_rate": 5.163348585250748e-07,
      "loss": 0.0811,
      "step": 12466
    },
    {
      "epoch": 5.866823529411764,
      "grad_norm": 0.4024518468497638,
      "learning_rate": 5.127071684431272e-07,
      "loss": 0.3809,
      "step": 12467
    },
    {
      "epoch": 5.8672941176470585,
      "grad_norm": 0.19751949932126558,
      "learning_rate": 5.090922453092283e-07,
      "loss": 0.0723,
      "step": 12468
    },
    {
      "epoch": 5.867764705882353,
      "grad_norm": 0.4593898537931474,
      "learning_rate": 5.054900894320979e-07,
      "loss": 0.1309,
      "step": 12469
    },
    {
      "epoch": 5.868235294117647,
      "grad_norm": 0.2762797456147398,
      "learning_rate": 5.019007011193732e-07,
      "loss": 0.0708,
      "step": 12470
    },
    {
      "epoch": 5.868705882352941,
      "grad_norm": 0.3859329516099745,
      "learning_rate": 4.983240806776257e-07,
      "loss": 0.4238,
      "step": 12471
    },
    {
      "epoch": 5.869176470588235,
      "grad_norm": 0.5176590508619066,
      "learning_rate": 4.947602284122942e-07,
      "loss": 0.2754,
      "step": 12472
    },
    {
      "epoch": 5.86964705882353,
      "grad_norm": 0.31273502543106774,
      "learning_rate": 4.912091446277688e-07,
      "loss": 0.1387,
      "step": 12473
    },
    {
      "epoch": 5.870117647058824,
      "grad_norm": 0.46344564392319654,
      "learning_rate": 4.876708296273068e-07,
      "loss": 0.2285,
      "step": 12474
    },
    {
      "epoch": 5.870588235294118,
      "grad_norm": 0.5028986413625588,
      "learning_rate": 4.841452837131166e-07,
      "loss": 0.2354,
      "step": 12475
    },
    {
      "epoch": 5.871058823529411,
      "grad_norm": 0.20068249766527868,
      "learning_rate": 4.806325071862905e-07,
      "loss": 0.0361,
      "step": 12476
    },
    {
      "epoch": 5.871529411764706,
      "grad_norm": 0.24135120112329342,
      "learning_rate": 4.771325003468385e-07,
      "loss": 0.0574,
      "step": 12477
    },
    {
      "epoch": 5.872,
      "grad_norm": 0.4919012504441209,
      "learning_rate": 4.736452634936716e-07,
      "loss": 0.4668,
      "step": 12478
    },
    {
      "epoch": 5.872470588235294,
      "grad_norm": 0.38604520533352643,
      "learning_rate": 4.701707969246182e-07,
      "loss": 0.1602,
      "step": 12479
    },
    {
      "epoch": 5.872941176470588,
      "grad_norm": 0.2659180171716636,
      "learning_rate": 4.6670910093642435e-07,
      "loss": 0.0596,
      "step": 12480
    },
    {
      "epoch": 5.8734117647058826,
      "grad_norm": 0.3686916910306963,
      "learning_rate": 4.6326017582473674e-07,
      "loss": 0.1279,
      "step": 12481
    },
    {
      "epoch": 5.873882352941177,
      "grad_norm": 0.40018304383575365,
      "learning_rate": 4.5982402188408653e-07,
      "loss": 0.4336,
      "step": 12482
    },
    {
      "epoch": 5.87435294117647,
      "grad_norm": 0.2839051467963796,
      "learning_rate": 4.5640063940797223e-07,
      "loss": 0.0742,
      "step": 12483
    },
    {
      "epoch": 5.874823529411764,
      "grad_norm": 0.3892766910196741,
      "learning_rate": 4.529900286887433e-07,
      "loss": 0.2158,
      "step": 12484
    },
    {
      "epoch": 5.875294117647059,
      "grad_norm": 0.4663621498124428,
      "learning_rate": 4.4959219001768355e-07,
      "loss": 0.6641,
      "step": 12485
    },
    {
      "epoch": 5.875764705882353,
      "grad_norm": 0.34261273789417673,
      "learning_rate": 4.4620712368501065e-07,
      "loss": 0.3145,
      "step": 12486
    },
    {
      "epoch": 5.876235294117647,
      "grad_norm": 0.3567711376875327,
      "learning_rate": 4.428348299797768e-07,
      "loss": 0.1982,
      "step": 12487
    },
    {
      "epoch": 5.876705882352941,
      "grad_norm": 0.2961953771332154,
      "learning_rate": 4.394753091900516e-07,
      "loss": 0.2832,
      "step": 12488
    },
    {
      "epoch": 5.8771764705882354,
      "grad_norm": 0.46763942575217365,
      "learning_rate": 4.361285616027055e-07,
      "loss": 0.3789,
      "step": 12489
    },
    {
      "epoch": 5.87764705882353,
      "grad_norm": 0.46535004001792424,
      "learning_rate": 4.327945875035932e-07,
      "loss": 0.2432,
      "step": 12490
    },
    {
      "epoch": 5.878117647058824,
      "grad_norm": 0.375493993747661,
      "learning_rate": 4.2947338717743694e-07,
      "loss": 0.1206,
      "step": 12491
    },
    {
      "epoch": 5.878588235294117,
      "grad_norm": 0.4196036450287003,
      "learning_rate": 4.2616496090790983e-07,
      "loss": 0.4551,
      "step": 12492
    },
    {
      "epoch": 5.8790588235294114,
      "grad_norm": 0.3899979036849886,
      "learning_rate": 4.228693089775526e-07,
      "loss": 0.1094,
      "step": 12493
    },
    {
      "epoch": 5.879529411764706,
      "grad_norm": 0.43999826523490776,
      "learning_rate": 4.195864316678066e-07,
      "loss": 0.249,
      "step": 12494
    },
    {
      "epoch": 5.88,
      "grad_norm": 0.5338447239556386,
      "learning_rate": 4.16316329259081e-07,
      "loss": 0.3262,
      "step": 12495
    },
    {
      "epoch": 5.880470588235294,
      "grad_norm": 0.45393246914967134,
      "learning_rate": 4.130590020306523e-07,
      "loss": 0.2266,
      "step": 12496
    },
    {
      "epoch": 5.880941176470588,
      "grad_norm": 0.32039497885431956,
      "learning_rate": 4.0981445026068147e-07,
      "loss": 0.2109,
      "step": 12497
    },
    {
      "epoch": 5.881411764705883,
      "grad_norm": 0.32336206481861746,
      "learning_rate": 4.0658267422631343e-07,
      "loss": 0.1582,
      "step": 12498
    },
    {
      "epoch": 5.881882352941177,
      "grad_norm": 0.4340212123887081,
      "learning_rate": 4.033636742035273e-07,
      "loss": 0.3594,
      "step": 12499
    },
    {
      "epoch": 5.882352941176471,
      "grad_norm": 0.39209443432379204,
      "learning_rate": 4.0015745046725336e-07,
      "loss": 0.3652,
      "step": 12500
    },
    {
      "epoch": 5.882823529411764,
      "grad_norm": 0.47088779594766783,
      "learning_rate": 3.9696400329132236e-07,
      "loss": 0.3574,
      "step": 12501
    },
    {
      "epoch": 5.883294117647059,
      "grad_norm": 0.5033041091801176,
      "learning_rate": 3.9378333294844967e-07,
      "loss": 0.4375,
      "step": 12502
    },
    {
      "epoch": 5.883764705882353,
      "grad_norm": 0.9687004110271215,
      "learning_rate": 3.906154397103012e-07,
      "loss": 0.9062,
      "step": 12503
    },
    {
      "epoch": 5.884235294117647,
      "grad_norm": 0.2392030399949387,
      "learning_rate": 3.874603238474272e-07,
      "loss": 0.0557,
      "step": 12504
    },
    {
      "epoch": 5.884705882352941,
      "grad_norm": 0.42071450043971476,
      "learning_rate": 3.8431798562927887e-07,
      "loss": 0.3711,
      "step": 12505
    },
    {
      "epoch": 5.8851764705882355,
      "grad_norm": 0.29771125238151436,
      "learning_rate": 3.8118842532424145e-07,
      "loss": 0.0967,
      "step": 12506
    },
    {
      "epoch": 5.88564705882353,
      "grad_norm": 0.21832372004861933,
      "learning_rate": 3.780716431995845e-07,
      "loss": 0.0432,
      "step": 12507
    },
    {
      "epoch": 5.886117647058824,
      "grad_norm": 0.40639787806550876,
      "learning_rate": 3.7496763952151177e-07,
      "loss": 0.1553,
      "step": 12508
    },
    {
      "epoch": 5.886588235294118,
      "grad_norm": 0.4311420171054221,
      "learning_rate": 3.7187641455509455e-07,
      "loss": 0.1455,
      "step": 12509
    },
    {
      "epoch": 5.8870588235294115,
      "grad_norm": 0.3467001819390306,
      "learning_rate": 3.687979685643549e-07,
      "loss": 0.0884,
      "step": 12510
    },
    {
      "epoch": 5.887529411764706,
      "grad_norm": 0.3885847008932044,
      "learning_rate": 3.6573230181221604e-07,
      "loss": 0.3086,
      "step": 12511
    },
    {
      "epoch": 5.888,
      "grad_norm": 0.44035314914857426,
      "learning_rate": 3.62679414560485e-07,
      "loss": 0.1875,
      "step": 12512
    },
    {
      "epoch": 5.888470588235294,
      "grad_norm": 0.2597527130173412,
      "learning_rate": 3.596393070699033e-07,
      "loss": 0.1855,
      "step": 12513
    },
    {
      "epoch": 5.888941176470588,
      "grad_norm": 0.27442047686863513,
      "learning_rate": 3.566119796000966e-07,
      "loss": 0.2451,
      "step": 12514
    },
    {
      "epoch": 5.889411764705883,
      "grad_norm": 0.39251754368455655,
      "learning_rate": 3.535974324096413e-07,
      "loss": 0.3047,
      "step": 12515
    },
    {
      "epoch": 5.889882352941177,
      "grad_norm": 0.4220140231363834,
      "learning_rate": 3.5059566575596497e-07,
      "loss": 0.2178,
      "step": 12516
    },
    {
      "epoch": 5.89035294117647,
      "grad_norm": 0.3984156903742033,
      "learning_rate": 3.4760667989544575e-07,
      "loss": 0.3203,
      "step": 12517
    },
    {
      "epoch": 5.890823529411764,
      "grad_norm": 0.5305614035059478,
      "learning_rate": 3.446304750833462e-07,
      "loss": 0.6055,
      "step": 12518
    },
    {
      "epoch": 5.891294117647059,
      "grad_norm": 0.3440215845942434,
      "learning_rate": 3.416670515738629e-07,
      "loss": 0.2041,
      "step": 12519
    },
    {
      "epoch": 5.891764705882353,
      "grad_norm": 0.3779022919965688,
      "learning_rate": 3.387164096200934e-07,
      "loss": 0.1011,
      "step": 12520
    },
    {
      "epoch": 5.892235294117647,
      "grad_norm": 0.4690999456517834,
      "learning_rate": 3.3577854947403615e-07,
      "loss": 0.4805,
      "step": 12521
    },
    {
      "epoch": 5.892705882352941,
      "grad_norm": 0.5716809739663934,
      "learning_rate": 3.3285347138657383e-07,
      "loss": 0.6875,
      "step": 12522
    },
    {
      "epoch": 5.8931764705882355,
      "grad_norm": 0.447036246246882,
      "learning_rate": 3.2994117560753986e-07,
      "loss": 0.1865,
      "step": 12523
    },
    {
      "epoch": 5.89364705882353,
      "grad_norm": 0.4063525450487518,
      "learning_rate": 3.2704166238565197e-07,
      "loss": 0.2617,
      "step": 12524
    },
    {
      "epoch": 5.894117647058824,
      "grad_norm": 0.4278956464071173,
      "learning_rate": 3.241549319685621e-07,
      "loss": 0.4941,
      "step": 12525
    },
    {
      "epoch": 5.894588235294117,
      "grad_norm": 0.44547382012104453,
      "learning_rate": 3.212809846027897e-07,
      "loss": 0.1885,
      "step": 12526
    },
    {
      "epoch": 5.8950588235294115,
      "grad_norm": 0.3785415272246932,
      "learning_rate": 3.1841982053378846e-07,
      "loss": 0.1738,
      "step": 12527
    },
    {
      "epoch": 5.895529411764706,
      "grad_norm": 0.47851195629795945,
      "learning_rate": 3.155714400059128e-07,
      "loss": 0.1348,
      "step": 12528
    },
    {
      "epoch": 5.896,
      "grad_norm": 0.19742904979072065,
      "learning_rate": 3.127358432624516e-07,
      "loss": 0.0515,
      "step": 12529
    },
    {
      "epoch": 5.896470588235294,
      "grad_norm": 0.4678194179980324,
      "learning_rate": 3.099130305455444e-07,
      "loss": 0.1836,
      "step": 12530
    },
    {
      "epoch": 5.896941176470588,
      "grad_norm": 0.5393018761857785,
      "learning_rate": 3.0710300209629836e-07,
      "loss": 0.5469,
      "step": 12531
    },
    {
      "epoch": 5.897411764705883,
      "grad_norm": 0.39070737483650086,
      "learning_rate": 3.043057581547048e-07,
      "loss": 0.2324,
      "step": 12532
    },
    {
      "epoch": 5.897882352941177,
      "grad_norm": 0.22206401423965552,
      "learning_rate": 3.0152129895963937e-07,
      "loss": 0.0233,
      "step": 12533
    },
    {
      "epoch": 5.898352941176471,
      "grad_norm": 0.3132795293390787,
      "learning_rate": 2.987496247489285e-07,
      "loss": 0.0986,
      "step": 12534
    },
    {
      "epoch": 5.898823529411764,
      "grad_norm": 0.16279985963328533,
      "learning_rate": 2.959907357592661e-07,
      "loss": 0.0305,
      "step": 12535
    },
    {
      "epoch": 5.899294117647059,
      "grad_norm": 0.9100325644009393,
      "learning_rate": 2.9324463222629715e-07,
      "loss": 0.2832,
      "step": 12536
    },
    {
      "epoch": 5.899764705882353,
      "grad_norm": 0.30852847530601896,
      "learning_rate": 2.9051131438455055e-07,
      "loss": 0.1318,
      "step": 12537
    },
    {
      "epoch": 5.900235294117647,
      "grad_norm": 0.41575043030447467,
      "learning_rate": 2.8779078246745635e-07,
      "loss": 0.1201,
      "step": 12538
    },
    {
      "epoch": 5.900705882352941,
      "grad_norm": 0.5216327611029683,
      "learning_rate": 2.85083036707362e-07,
      "loss": 0.3398,
      "step": 12539
    },
    {
      "epoch": 5.9011764705882355,
      "grad_norm": 0.25079903556744565,
      "learning_rate": 2.8238807733551583e-07,
      "loss": 0.0378,
      "step": 12540
    },
    {
      "epoch": 5.90164705882353,
      "grad_norm": 0.5797399421154329,
      "learning_rate": 2.797059045821004e-07,
      "loss": 0.293,
      "step": 12541
    },
    {
      "epoch": 5.902117647058824,
      "grad_norm": 0.5900016715578413,
      "learning_rate": 2.7703651867616583e-07,
      "loss": 0.5352,
      "step": 12542
    },
    {
      "epoch": 5.902588235294118,
      "grad_norm": 0.27509213349267675,
      "learning_rate": 2.743799198457131e-07,
      "loss": 0.1904,
      "step": 12543
    },
    {
      "epoch": 5.9030588235294115,
      "grad_norm": 0.6023770466465433,
      "learning_rate": 2.71736108317594e-07,
      "loss": 0.3496,
      "step": 12544
    },
    {
      "epoch": 5.903529411764706,
      "grad_norm": 0.36979241571344706,
      "learning_rate": 2.6910508431764457e-07,
      "loss": 0.2734,
      "step": 12545
    },
    {
      "epoch": 5.904,
      "grad_norm": 0.5395132660632292,
      "learning_rate": 2.6648684807053513e-07,
      "loss": 0.377,
      "step": 12546
    },
    {
      "epoch": 5.904470588235294,
      "grad_norm": 0.5761311933616006,
      "learning_rate": 2.6388139979988677e-07,
      "loss": 0.5391,
      "step": 12547
    },
    {
      "epoch": 5.904941176470588,
      "grad_norm": 0.34813043826544765,
      "learning_rate": 2.612887397282215e-07,
      "loss": 0.0859,
      "step": 12548
    },
    {
      "epoch": 5.905411764705883,
      "grad_norm": 0.492463502227305,
      "learning_rate": 2.5870886807696225e-07,
      "loss": 0.4863,
      "step": 12549
    },
    {
      "epoch": 5.905882352941177,
      "grad_norm": 0.4767461299059936,
      "learning_rate": 2.5614178506644934e-07,
      "loss": 0.3652,
      "step": 12550
    },
    {
      "epoch": 5.90635294117647,
      "grad_norm": 0.6479618408541699,
      "learning_rate": 2.5358749091590746e-07,
      "loss": 0.4648,
      "step": 12551
    },
    {
      "epoch": 5.906823529411764,
      "grad_norm": 0.46492968984666294,
      "learning_rate": 2.5104598584349543e-07,
      "loss": 0.1689,
      "step": 12552
    },
    {
      "epoch": 5.907294117647059,
      "grad_norm": 0.45132301747111436,
      "learning_rate": 2.4851727006627297e-07,
      "loss": 0.459,
      "step": 12553
    },
    {
      "epoch": 5.907764705882353,
      "grad_norm": 0.5707155831972935,
      "learning_rate": 2.460013438002173e-07,
      "loss": 0.5742,
      "step": 12554
    },
    {
      "epoch": 5.908235294117647,
      "grad_norm": 0.4157608212019548,
      "learning_rate": 2.434982072601732e-07,
      "loss": 0.457,
      "step": 12555
    },
    {
      "epoch": 5.908705882352941,
      "grad_norm": 0.3630638738602507,
      "learning_rate": 2.4100786065991975e-07,
      "loss": 0.1943,
      "step": 12556
    },
    {
      "epoch": 5.9091764705882355,
      "grad_norm": 0.3697149570425375,
      "learning_rate": 2.3853030421218667e-07,
      "loss": 0.082,
      "step": 12557
    },
    {
      "epoch": 5.90964705882353,
      "grad_norm": 0.39765170953601303,
      "learning_rate": 2.3606553812852146e-07,
      "loss": 0.2578,
      "step": 12558
    },
    {
      "epoch": 5.910117647058824,
      "grad_norm": 0.3431504709598111,
      "learning_rate": 2.3361356261945575e-07,
      "loss": 0.3438,
      "step": 12559
    },
    {
      "epoch": 5.910588235294117,
      "grad_norm": 0.7732590900868231,
      "learning_rate": 2.3117437789438865e-07,
      "loss": 0.4297,
      "step": 12560
    },
    {
      "epoch": 5.9110588235294115,
      "grad_norm": 0.5093596844812741,
      "learning_rate": 2.2874798416163686e-07,
      "loss": 0.6211,
      "step": 12561
    },
    {
      "epoch": 5.911529411764706,
      "grad_norm": 0.5264190656141136,
      "learning_rate": 2.2633438162843465e-07,
      "loss": 0.4824,
      "step": 12562
    },
    {
      "epoch": 5.912,
      "grad_norm": 0.3517132350995784,
      "learning_rate": 2.239335705009171e-07,
      "loss": 0.2871,
      "step": 12563
    },
    {
      "epoch": 5.912470588235294,
      "grad_norm": 0.49523217422716487,
      "learning_rate": 2.215455509841202e-07,
      "loss": 0.3242,
      "step": 12564
    },
    {
      "epoch": 5.912941176470588,
      "grad_norm": 0.30252775229692125,
      "learning_rate": 2.1917032328199747e-07,
      "loss": 0.1045,
      "step": 12565
    },
    {
      "epoch": 5.913411764705883,
      "grad_norm": 0.37864848545213603,
      "learning_rate": 2.1680788759738664e-07,
      "loss": 0.3711,
      "step": 12566
    },
    {
      "epoch": 5.913882352941177,
      "grad_norm": 0.668395072612716,
      "learning_rate": 2.144582441320597e-07,
      "loss": 0.5156,
      "step": 12567
    },
    {
      "epoch": 5.914352941176471,
      "grad_norm": 0.30352976601839404,
      "learning_rate": 2.1212139308670607e-07,
      "loss": 0.2695,
      "step": 12568
    },
    {
      "epoch": 5.914823529411764,
      "grad_norm": 0.5007026211434827,
      "learning_rate": 2.0979733466088277e-07,
      "loss": 0.6523,
      "step": 12569
    },
    {
      "epoch": 5.915294117647059,
      "grad_norm": 0.24327641087310833,
      "learning_rate": 2.0748606905309774e-07,
      "loss": 0.0635,
      "step": 12570
    },
    {
      "epoch": 5.915764705882353,
      "grad_norm": 0.5208754820550472,
      "learning_rate": 2.051875964607097e-07,
      "loss": 0.332,
      "step": 12571
    },
    {
      "epoch": 5.916235294117647,
      "grad_norm": 0.30880312524064435,
      "learning_rate": 2.0290191708006165e-07,
      "loss": 0.1992,
      "step": 12572
    },
    {
      "epoch": 5.916705882352941,
      "grad_norm": 0.556957498912227,
      "learning_rate": 2.0062903110633077e-07,
      "loss": 0.4238,
      "step": 12573
    },
    {
      "epoch": 5.9171764705882355,
      "grad_norm": 0.12873067773649602,
      "learning_rate": 1.9836893873364513e-07,
      "loss": 0.0128,
      "step": 12574
    },
    {
      "epoch": 5.91764705882353,
      "grad_norm": 0.35661518076309345,
      "learning_rate": 1.9612164015501696e-07,
      "loss": 0.3281,
      "step": 12575
    },
    {
      "epoch": 5.918117647058823,
      "grad_norm": 0.4123429469591808,
      "learning_rate": 1.9388713556239277e-07,
      "loss": 0.3066,
      "step": 12576
    },
    {
      "epoch": 5.918588235294117,
      "grad_norm": 0.4328939331863637,
      "learning_rate": 1.9166542514658656e-07,
      "loss": 0.3633,
      "step": 12577
    },
    {
      "epoch": 5.9190588235294115,
      "grad_norm": 0.14557009655230368,
      "learning_rate": 1.8945650909737985e-07,
      "loss": 0.0216,
      "step": 12578
    },
    {
      "epoch": 5.919529411764706,
      "grad_norm": 0.5026836143302656,
      "learning_rate": 1.8726038760338847e-07,
      "loss": 0.3145,
      "step": 12579
    },
    {
      "epoch": 5.92,
      "grad_norm": 0.35825845395134587,
      "learning_rate": 1.8507706085219566e-07,
      "loss": 0.1167,
      "step": 12580
    },
    {
      "epoch": 5.920470588235294,
      "grad_norm": 0.17474107235824704,
      "learning_rate": 1.8290652903025228e-07,
      "loss": 0.0757,
      "step": 12581
    },
    {
      "epoch": 5.920941176470588,
      "grad_norm": 0.3274152433657619,
      "learning_rate": 1.8074879232296002e-07,
      "loss": 0.3262,
      "step": 12582
    },
    {
      "epoch": 5.921411764705883,
      "grad_norm": 0.2691237024440563,
      "learning_rate": 1.7860385091455488e-07,
      "loss": 0.0574,
      "step": 12583
    },
    {
      "epoch": 5.921882352941177,
      "grad_norm": 0.41694762334490937,
      "learning_rate": 1.764717049882569e-07,
      "loss": 0.2051,
      "step": 12584
    },
    {
      "epoch": 5.92235294117647,
      "grad_norm": 0.6835034534930577,
      "learning_rate": 1.743523547261705e-07,
      "loss": 0.6094,
      "step": 12585
    },
    {
      "epoch": 5.922823529411764,
      "grad_norm": 0.20708844185063396,
      "learning_rate": 1.722458003092675e-07,
      "loss": 0.1504,
      "step": 12586
    },
    {
      "epoch": 5.923294117647059,
      "grad_norm": 0.30401911557668676,
      "learning_rate": 1.701520419174873e-07,
      "loss": 0.1074,
      "step": 12587
    },
    {
      "epoch": 5.923764705882353,
      "grad_norm": 0.21718081512993467,
      "learning_rate": 1.680710797296203e-07,
      "loss": 0.0378,
      "step": 12588
    },
    {
      "epoch": 5.924235294117647,
      "grad_norm": 0.47198355192225455,
      "learning_rate": 1.660029139234076e-07,
      "loss": 0.167,
      "step": 12589
    },
    {
      "epoch": 5.924705882352941,
      "grad_norm": 0.2810035428236155,
      "learning_rate": 1.639475446754912e-07,
      "loss": 0.0791,
      "step": 12590
    },
    {
      "epoch": 5.9251764705882355,
      "grad_norm": 0.554122107317638,
      "learning_rate": 1.6190497216138076e-07,
      "loss": 0.4219,
      "step": 12591
    },
    {
      "epoch": 5.92564705882353,
      "grad_norm": 0.2950793714747512,
      "learning_rate": 1.598751965555367e-07,
      "loss": 0.167,
      "step": 12592
    },
    {
      "epoch": 5.926117647058824,
      "grad_norm": 0.28872546080451583,
      "learning_rate": 1.5785821803132037e-07,
      "loss": 0.1025,
      "step": 12593
    },
    {
      "epoch": 5.926588235294117,
      "grad_norm": 0.4124098291971203,
      "learning_rate": 1.558540367609773e-07,
      "loss": 0.125,
      "step": 12594
    },
    {
      "epoch": 5.9270588235294115,
      "grad_norm": 0.4943808538773532,
      "learning_rate": 1.5386265291568723e-07,
      "loss": 0.418,
      "step": 12595
    },
    {
      "epoch": 5.927529411764706,
      "grad_norm": 0.3180703880430452,
      "learning_rate": 1.5188406666551412e-07,
      "loss": 0.1875,
      "step": 12596
    },
    {
      "epoch": 5.928,
      "grad_norm": 0.37643471894421254,
      "learning_rate": 1.4991827817943947e-07,
      "loss": 0.1309,
      "step": 12597
    },
    {
      "epoch": 5.928470588235294,
      "grad_norm": 0.2828936647988978,
      "learning_rate": 1.479652876253623e-07,
      "loss": 0.1455,
      "step": 12598
    },
    {
      "epoch": 5.928941176470588,
      "grad_norm": 0.4342689274187255,
      "learning_rate": 1.4602509517006588e-07,
      "loss": 0.2256,
      "step": 12599
    },
    {
      "epoch": 5.929411764705883,
      "grad_norm": 0.41193497497175,
      "learning_rate": 1.4409770097926765e-07,
      "loss": 0.2227,
      "step": 12600
    },
    {
      "epoch": 5.929882352941177,
      "grad_norm": 0.2558282805602685,
      "learning_rate": 1.421831052175526e-07,
      "loss": 0.0361,
      "step": 12601
    },
    {
      "epoch": 5.930352941176471,
      "grad_norm": 0.4383899351447508,
      "learning_rate": 1.4028130804843995e-07,
      "loss": 0.3965,
      "step": 12602
    },
    {
      "epoch": 5.930823529411764,
      "grad_norm": 0.4820719163474788,
      "learning_rate": 1.3839230963436642e-07,
      "loss": 0.3672,
      "step": 12603
    },
    {
      "epoch": 5.931294117647059,
      "grad_norm": 0.3817760034270271,
      "learning_rate": 1.3651611013666964e-07,
      "loss": 0.3945,
      "step": 12604
    },
    {
      "epoch": 5.931764705882353,
      "grad_norm": 0.3786363854883346,
      "learning_rate": 1.346527097155714e-07,
      "loss": 0.2285,
      "step": 12605
    },
    {
      "epoch": 5.932235294117647,
      "grad_norm": 0.4856672470619509,
      "learning_rate": 1.3280210853019447e-07,
      "loss": 0.2314,
      "step": 12606
    },
    {
      "epoch": 5.932705882352941,
      "grad_norm": 0.43235141254500414,
      "learning_rate": 1.3096430673862901e-07,
      "loss": 0.3262,
      "step": 12607
    },
    {
      "epoch": 5.9331764705882355,
      "grad_norm": 0.4213867499090105,
      "learning_rate": 1.2913930449781619e-07,
      "loss": 0.2168,
      "step": 12608
    },
    {
      "epoch": 5.93364705882353,
      "grad_norm": 0.5398277659408365,
      "learning_rate": 1.2732710196361463e-07,
      "loss": 0.4434,
      "step": 12609
    },
    {
      "epoch": 5.934117647058823,
      "grad_norm": 0.5965009983420176,
      "learning_rate": 1.2552769929081717e-07,
      "loss": 0.4727,
      "step": 12610
    },
    {
      "epoch": 5.934588235294117,
      "grad_norm": 0.3726533773498442,
      "learning_rate": 1.2374109663306763e-07,
      "loss": 0.1602,
      "step": 12611
    },
    {
      "epoch": 5.9350588235294115,
      "grad_norm": 0.36468267856057457,
      "learning_rate": 1.2196729414296058e-07,
      "loss": 0.1348,
      "step": 12612
    },
    {
      "epoch": 5.935529411764706,
      "grad_norm": 0.369532677629149,
      "learning_rate": 1.2020629197200814e-07,
      "loss": 0.106,
      "step": 12613
    },
    {
      "epoch": 5.936,
      "grad_norm": 0.39243320642014723,
      "learning_rate": 1.1845809027059006e-07,
      "loss": 0.1787,
      "step": 12614
    },
    {
      "epoch": 5.936470588235294,
      "grad_norm": 0.2238503168393797,
      "learning_rate": 1.1672268918800353e-07,
      "loss": 0.063,
      "step": 12615
    },
    {
      "epoch": 5.936941176470588,
      "grad_norm": 0.2857285471845557,
      "learning_rate": 1.1500008887249668e-07,
      "loss": 0.1611,
      "step": 12616
    },
    {
      "epoch": 5.937411764705883,
      "grad_norm": 0.27368164648466914,
      "learning_rate": 1.1329028947115181e-07,
      "loss": 0.1445,
      "step": 12617
    },
    {
      "epoch": 5.937882352941177,
      "grad_norm": 0.7998664999251319,
      "learning_rate": 1.1159329113000215e-07,
      "loss": 0.2295,
      "step": 12618
    },
    {
      "epoch": 5.93835294117647,
      "grad_norm": 0.20680828155258965,
      "learning_rate": 1.0990909399398173e-07,
      "loss": 0.0374,
      "step": 12619
    },
    {
      "epoch": 5.938823529411764,
      "grad_norm": 0.4279709033590647,
      "learning_rate": 1.0823769820694218e-07,
      "loss": 0.3262,
      "step": 12620
    },
    {
      "epoch": 5.939294117647059,
      "grad_norm": 0.3902104487867319,
      "learning_rate": 1.0657910391161928e-07,
      "loss": 0.4062,
      "step": 12621
    },
    {
      "epoch": 5.939764705882353,
      "grad_norm": 0.5502350365252883,
      "learning_rate": 1.0493331124964976e-07,
      "loss": 0.3398,
      "step": 12622
    },
    {
      "epoch": 5.940235294117647,
      "grad_norm": 0.5123329323382808,
      "learning_rate": 1.0330032036162117e-07,
      "loss": 0.5195,
      "step": 12623
    },
    {
      "epoch": 5.940705882352941,
      "grad_norm": 0.4690779345462096,
      "learning_rate": 1.0168013138697196e-07,
      "loss": 0.543,
      "step": 12624
    },
    {
      "epoch": 5.9411764705882355,
      "grad_norm": 0.43332697316088453,
      "learning_rate": 1.0007274446409141e-07,
      "loss": 0.2168,
      "step": 12625
    },
    {
      "epoch": 5.94164705882353,
      "grad_norm": 0.4811278471126423,
      "learning_rate": 9.847815973025308e-08,
      "loss": 0.2129,
      "step": 12626
    },
    {
      "epoch": 5.942117647058824,
      "grad_norm": 0.4102615010684342,
      "learning_rate": 9.689637732164802e-08,
      "loss": 0.293,
      "step": 12627
    },
    {
      "epoch": 5.942588235294117,
      "grad_norm": 0.4129468832439842,
      "learning_rate": 9.532739737336814e-08,
      "loss": 0.5156,
      "step": 12628
    },
    {
      "epoch": 5.9430588235294115,
      "grad_norm": 0.3474333283684218,
      "learning_rate": 9.377122001938963e-08,
      "loss": 0.2422,
      "step": 12629
    },
    {
      "epoch": 5.943529411764706,
      "grad_norm": 0.1744990923314015,
      "learning_rate": 9.222784539263949e-08,
      "loss": 0.0098,
      "step": 12630
    },
    {
      "epoch": 5.944,
      "grad_norm": 0.41556859802501955,
      "learning_rate": 9.069727362492896e-08,
      "loss": 0.1611,
      "step": 12631
    },
    {
      "epoch": 5.944470588235294,
      "grad_norm": 0.4120706025046923,
      "learning_rate": 8.917950484697012e-08,
      "loss": 0.4043,
      "step": 12632
    },
    {
      "epoch": 5.944941176470588,
      "grad_norm": 0.30107474153169494,
      "learning_rate": 8.767453918839263e-08,
      "loss": 0.3535,
      "step": 12633
    },
    {
      "epoch": 5.945411764705883,
      "grad_norm": 0.19591115808727105,
      "learning_rate": 8.618237677772699e-08,
      "loss": 0.0481,
      "step": 12634
    },
    {
      "epoch": 5.945882352941177,
      "grad_norm": 0.07691118476152128,
      "learning_rate": 8.470301774240463e-08,
      "loss": 0.0118,
      "step": 12635
    },
    {
      "epoch": 5.946352941176471,
      "grad_norm": 0.27127198301246924,
      "learning_rate": 8.323646220877444e-08,
      "loss": 0.1953,
      "step": 12636
    },
    {
      "epoch": 5.946823529411764,
      "grad_norm": 0.43833522340944975,
      "learning_rate": 8.178271030210292e-08,
      "loss": 0.4316,
      "step": 12637
    },
    {
      "epoch": 5.947294117647059,
      "grad_norm": 0.46740511507837007,
      "learning_rate": 8.034176214650745e-08,
      "loss": 0.2188,
      "step": 12638
    },
    {
      "epoch": 5.947764705882353,
      "grad_norm": 0.3284815908175262,
      "learning_rate": 7.891361786508953e-08,
      "loss": 0.127,
      "step": 12639
    },
    {
      "epoch": 5.948235294117647,
      "grad_norm": 0.2674931561597028,
      "learning_rate": 7.749827757981831e-08,
      "loss": 0.053,
      "step": 12640
    },
    {
      "epoch": 5.948705882352941,
      "grad_norm": 0.5815443280142196,
      "learning_rate": 7.609574141154706e-08,
      "loss": 0.7109,
      "step": 12641
    },
    {
      "epoch": 5.9491764705882355,
      "grad_norm": 0.3752545425100831,
      "learning_rate": 7.470600948006333e-08,
      "loss": 0.0767,
      "step": 12642
    },
    {
      "epoch": 5.94964705882353,
      "grad_norm": 0.3967892155485872,
      "learning_rate": 7.332908190408882e-08,
      "loss": 0.5195,
      "step": 12643
    },
    {
      "epoch": 5.950117647058823,
      "grad_norm": 0.428132358388377,
      "learning_rate": 7.196495880117947e-08,
      "loss": 0.4004,
      "step": 12644
    },
    {
      "epoch": 5.950588235294117,
      "grad_norm": 0.320495600005413,
      "learning_rate": 7.061364028785876e-08,
      "loss": 0.1758,
      "step": 12645
    },
    {
      "epoch": 5.9510588235294115,
      "grad_norm": 0.24394664021037712,
      "learning_rate": 6.927512647953437e-08,
      "loss": 0.0452,
      "step": 12646
    },
    {
      "epoch": 5.951529411764706,
      "grad_norm": 0.4195933185585777,
      "learning_rate": 6.794941749053151e-08,
      "loss": 0.3086,
      "step": 12647
    },
    {
      "epoch": 5.952,
      "grad_norm": 0.35239351063366525,
      "learning_rate": 6.663651343404297e-08,
      "loss": 0.2891,
      "step": 12648
    },
    {
      "epoch": 5.952470588235294,
      "grad_norm": 0.4433074892153491,
      "learning_rate": 6.533641442222904e-08,
      "loss": 0.293,
      "step": 12649
    },
    {
      "epoch": 5.952941176470588,
      "grad_norm": 0.3987550495750609,
      "learning_rate": 6.40491205661009e-08,
      "loss": 0.1387,
      "step": 12650
    },
    {
      "epoch": 5.953411764705883,
      "grad_norm": 0.5683980924079457,
      "learning_rate": 6.277463197560395e-08,
      "loss": 0.3594,
      "step": 12651
    },
    {
      "epoch": 5.953882352941177,
      "grad_norm": 0.37775522536814626,
      "learning_rate": 6.151294875960111e-08,
      "loss": 0.2598,
      "step": 12652
    },
    {
      "epoch": 5.95435294117647,
      "grad_norm": 0.5433847867390804,
      "learning_rate": 6.026407102583952e-08,
      "loss": 0.373,
      "step": 12653
    },
    {
      "epoch": 5.954823529411764,
      "grad_norm": 0.2473702911127174,
      "learning_rate": 5.902799888096721e-08,
      "loss": 0.0723,
      "step": 12654
    },
    {
      "epoch": 5.955294117647059,
      "grad_norm": 0.44406743971436285,
      "learning_rate": 5.780473243056638e-08,
      "loss": 0.3086,
      "step": 12655
    },
    {
      "epoch": 5.955764705882353,
      "grad_norm": 0.35402515435927084,
      "learning_rate": 5.659427177910347e-08,
      "loss": 0.2246,
      "step": 12656
    },
    {
      "epoch": 5.956235294117647,
      "grad_norm": 0.358668527945859,
      "learning_rate": 5.539661702996245e-08,
      "loss": 0.0659,
      "step": 12657
    },
    {
      "epoch": 5.956705882352941,
      "grad_norm": 0.1988660883968954,
      "learning_rate": 5.421176828541152e-08,
      "loss": 0.0576,
      "step": 12658
    },
    {
      "epoch": 5.9571764705882355,
      "grad_norm": 0.44662087547496787,
      "learning_rate": 5.30397256466697e-08,
      "loss": 0.1562,
      "step": 12659
    },
    {
      "epoch": 5.95764705882353,
      "grad_norm": 0.40192781766790137,
      "learning_rate": 5.1880489213806944e-08,
      "loss": 0.0923,
      "step": 12660
    },
    {
      "epoch": 5.958117647058824,
      "grad_norm": 0.5515863842703649,
      "learning_rate": 5.0734059085860704e-08,
      "loss": 0.582,
      "step": 12661
    },
    {
      "epoch": 5.958588235294117,
      "grad_norm": 0.4548400834878496,
      "learning_rate": 4.9600435360702685e-08,
      "loss": 0.3613,
      "step": 12662
    },
    {
      "epoch": 5.9590588235294115,
      "grad_norm": 0.3037028353588882,
      "learning_rate": 4.847961813517209e-08,
      "loss": 0.2988,
      "step": 12663
    },
    {
      "epoch": 5.959529411764706,
      "grad_norm": 0.448900178943214,
      "learning_rate": 4.737160750500901e-08,
      "loss": 0.1787,
      "step": 12664
    },
    {
      "epoch": 5.96,
      "grad_norm": 0.5057813851187435,
      "learning_rate": 4.6276403564804444e-08,
      "loss": 0.418,
      "step": 12665
    },
    {
      "epoch": 5.960470588235294,
      "grad_norm": 0.5118032533060524,
      "learning_rate": 4.519400640811688e-08,
      "loss": 0.332,
      "step": 12666
    },
    {
      "epoch": 5.960941176470588,
      "grad_norm": 0.630852300562569,
      "learning_rate": 4.412441612738904e-08,
      "loss": 0.4863,
      "step": 12667
    },
    {
      "epoch": 5.961411764705883,
      "grad_norm": 0.4789237772886933,
      "learning_rate": 4.3067632813964524e-08,
      "loss": 0.1602,
      "step": 12668
    },
    {
      "epoch": 5.961882352941177,
      "grad_norm": 0.548448108827003,
      "learning_rate": 4.2023656558104466e-08,
      "loss": 0.2695,
      "step": 12669
    },
    {
      "epoch": 5.962352941176471,
      "grad_norm": 0.23878242483564263,
      "learning_rate": 4.099248744897088e-08,
      "loss": 0.0454,
      "step": 12670
    },
    {
      "epoch": 5.962823529411764,
      "grad_norm": 0.49833793672612686,
      "learning_rate": 3.9974125574609994e-08,
      "loss": 0.5391,
      "step": 12671
    },
    {
      "epoch": 5.963294117647059,
      "grad_norm": 0.3506428812887685,
      "learning_rate": 3.896857102200224e-08,
      "loss": 0.3672,
      "step": 12672
    },
    {
      "epoch": 5.963764705882353,
      "grad_norm": 0.3917531359585929,
      "learning_rate": 3.797582387704556e-08,
      "loss": 0.3359,
      "step": 12673
    },
    {
      "epoch": 5.964235294117647,
      "grad_norm": 0.47390409354076957,
      "learning_rate": 3.69958842245055e-08,
      "loss": 0.498,
      "step": 12674
    },
    {
      "epoch": 5.964705882352941,
      "grad_norm": 0.24682243420944808,
      "learning_rate": 3.6028752148081766e-08,
      "loss": 0.0569,
      "step": 12675
    },
    {
      "epoch": 5.9651764705882355,
      "grad_norm": 0.4049313908317211,
      "learning_rate": 3.5074427730374945e-08,
      "loss": 0.1885,
      "step": 12676
    },
    {
      "epoch": 5.96564705882353,
      "grad_norm": 0.1654172482342397,
      "learning_rate": 3.4132911052886514e-08,
      "loss": 0.0265,
      "step": 12677
    },
    {
      "epoch": 5.966117647058823,
      "grad_norm": 0.8169095431231039,
      "learning_rate": 3.320420219603548e-08,
      "loss": 0.6719,
      "step": 12678
    },
    {
      "epoch": 5.966588235294117,
      "grad_norm": 0.5025937487464397,
      "learning_rate": 3.228830123910842e-08,
      "loss": 0.1924,
      "step": 12679
    },
    {
      "epoch": 5.9670588235294115,
      "grad_norm": 0.48342835495879033,
      "learning_rate": 3.13852082603594e-08,
      "loss": 0.6055,
      "step": 12680
    },
    {
      "epoch": 5.967529411764706,
      "grad_norm": 0.32667707339586355,
      "learning_rate": 3.049492333691006e-08,
      "loss": 0.1001,
      "step": 12681
    },
    {
      "epoch": 5.968,
      "grad_norm": 0.5452468308131748,
      "learning_rate": 2.961744654478293e-08,
      "loss": 0.6328,
      "step": 12682
    },
    {
      "epoch": 5.968470588235294,
      "grad_norm": 0.05499050098914347,
      "learning_rate": 2.875277795891806e-08,
      "loss": 0.0089,
      "step": 12683
    },
    {
      "epoch": 5.968941176470588,
      "grad_norm": 0.39960203209558864,
      "learning_rate": 2.7900917653189692e-08,
      "loss": 0.3418,
      "step": 12684
    },
    {
      "epoch": 5.969411764705883,
      "grad_norm": 0.3038864638424287,
      "learning_rate": 2.706186570032298e-08,
      "loss": 0.0859,
      "step": 12685
    },
    {
      "epoch": 5.969882352941177,
      "grad_norm": 0.29578490932930757,
      "learning_rate": 2.6235622171977278e-08,
      "loss": 0.0498,
      "step": 12686
    },
    {
      "epoch": 5.97035294117647,
      "grad_norm": 0.2856837045321453,
      "learning_rate": 2.542218713874611e-08,
      "loss": 0.0684,
      "step": 12687
    },
    {
      "epoch": 5.970823529411764,
      "grad_norm": 0.38256548454408007,
      "learning_rate": 2.462156067005727e-08,
      "loss": 0.1807,
      "step": 12688
    },
    {
      "epoch": 5.971294117647059,
      "grad_norm": 0.2974383188041118,
      "learning_rate": 2.3833742834339365e-08,
      "loss": 0.0645,
      "step": 12689
    },
    {
      "epoch": 5.971764705882353,
      "grad_norm": 0.5537241981655541,
      "learning_rate": 2.305873369882194e-08,
      "loss": 0.2363,
      "step": 12690
    },
    {
      "epoch": 5.972235294117647,
      "grad_norm": 0.37215037177851534,
      "learning_rate": 2.2296533329735356e-08,
      "loss": 0.1318,
      "step": 12691
    },
    {
      "epoch": 5.972705882352941,
      "grad_norm": 0.55904190911107,
      "learning_rate": 2.1547141792160882e-08,
      "loss": 0.7656,
      "step": 12692
    },
    {
      "epoch": 5.9731764705882355,
      "grad_norm": 0.33976909024334145,
      "learning_rate": 2.0810559150097327e-08,
      "loss": 0.1299,
      "step": 12693
    },
    {
      "epoch": 5.97364705882353,
      "grad_norm": 0.33246095547227966,
      "learning_rate": 2.008678546646103e-08,
      "loss": 0.1904,
      "step": 12694
    },
    {
      "epoch": 5.974117647058824,
      "grad_norm": 0.2787538530166166,
      "learning_rate": 1.9375820803052555e-08,
      "loss": 0.0815,
      "step": 12695
    },
    {
      "epoch": 5.974588235294117,
      "grad_norm": 0.5414500597444862,
      "learning_rate": 1.8677665220606653e-08,
      "loss": 0.2471,
      "step": 12696
    },
    {
      "epoch": 5.9750588235294115,
      "grad_norm": 0.3822296584857085,
      "learning_rate": 1.7992318778742298e-08,
      "loss": 0.3477,
      "step": 12697
    },
    {
      "epoch": 5.975529411764706,
      "grad_norm": 0.4577262959774879,
      "learning_rate": 1.7319781535996002e-08,
      "loss": 0.2344,
      "step": 12698
    },
    {
      "epoch": 5.976,
      "grad_norm": 0.31375110693263175,
      "learning_rate": 1.6660053549788498e-08,
      "loss": 0.1289,
      "step": 12699
    },
    {
      "epoch": 5.976470588235294,
      "grad_norm": 0.7042317777092172,
      "learning_rate": 1.6013134876491362e-08,
      "loss": 0.4727,
      "step": 12700
    },
    {
      "epoch": 5.976941176470588,
      "grad_norm": 0.29650712081837133,
      "learning_rate": 1.5379025571327088e-08,
      "loss": 0.0806,
      "step": 12701
    },
    {
      "epoch": 5.977411764705883,
      "grad_norm": 0.5200743710927813,
      "learning_rate": 1.4757725688469002e-08,
      "loss": 0.6211,
      "step": 12702
    },
    {
      "epoch": 5.977882352941176,
      "grad_norm": 0.44465996617373466,
      "learning_rate": 1.414923528097467e-08,
      "loss": 0.2891,
      "step": 12703
    },
    {
      "epoch": 5.97835294117647,
      "grad_norm": 0.365604044861238,
      "learning_rate": 1.3553554400802524e-08,
      "loss": 0.2891,
      "step": 12704
    },
    {
      "epoch": 5.978823529411764,
      "grad_norm": 0.30551789577353233,
      "learning_rate": 1.2970683098845192e-08,
      "loss": 0.1191,
      "step": 12705
    },
    {
      "epoch": 5.979294117647059,
      "grad_norm": 0.5846016402253139,
      "learning_rate": 1.240062142487952e-08,
      "loss": 0.6836,
      "step": 12706
    },
    {
      "epoch": 5.979764705882353,
      "grad_norm": 0.5108942574995776,
      "learning_rate": 1.1843369427583238e-08,
      "loss": 0.2412,
      "step": 12707
    },
    {
      "epoch": 5.980235294117647,
      "grad_norm": 0.24437941238425034,
      "learning_rate": 1.1298927154551608e-08,
      "loss": 0.0835,
      "step": 12708
    },
    {
      "epoch": 5.980705882352941,
      "grad_norm": 0.36907875863826406,
      "learning_rate": 1.0767294652280767e-08,
      "loss": 0.207,
      "step": 12709
    },
    {
      "epoch": 5.9811764705882355,
      "grad_norm": 0.2505761811266164,
      "learning_rate": 1.0248471966167737e-08,
      "loss": 0.0991,
      "step": 12710
    },
    {
      "epoch": 5.98164705882353,
      "grad_norm": 0.35803852320799395,
      "learning_rate": 9.74245914052707e-09,
      "loss": 0.3477,
      "step": 12711
    },
    {
      "epoch": 5.982117647058823,
      "grad_norm": 0.3580642755851353,
      "learning_rate": 9.24925621859085e-09,
      "loss": 0.3086,
      "step": 12712
    },
    {
      "epoch": 5.982588235294117,
      "grad_norm": 0.5055734271491581,
      "learning_rate": 8.768863242475389e-09,
      "loss": 0.4766,
      "step": 12713
    },
    {
      "epoch": 5.9830588235294115,
      "grad_norm": 0.4529003398469516,
      "learning_rate": 8.30128025318122e-09,
      "loss": 0.4297,
      "step": 12714
    },
    {
      "epoch": 5.983529411764706,
      "grad_norm": 0.40713694290299396,
      "learning_rate": 7.846507290676373e-09,
      "loss": 0.167,
      "step": 12715
    },
    {
      "epoch": 5.984,
      "grad_norm": 0.36306707788580767,
      "learning_rate": 7.404544393779799e-09,
      "loss": 0.3125,
      "step": 12716
    },
    {
      "epoch": 5.984470588235294,
      "grad_norm": 0.38480095977538376,
      "learning_rate": 6.975391600261282e-09,
      "loss": 0.1016,
      "step": 12717
    },
    {
      "epoch": 5.984941176470588,
      "grad_norm": 0.6308085979854294,
      "learning_rate": 6.55904894674153e-09,
      "loss": 0.2891,
      "step": 12718
    },
    {
      "epoch": 5.985411764705883,
      "grad_norm": 0.2409340420613166,
      "learning_rate": 6.155516468792088e-09,
      "loss": 0.2021,
      "step": 12719
    },
    {
      "epoch": 5.985882352941177,
      "grad_norm": 0.5370213252035928,
      "learning_rate": 5.764794200885381e-09,
      "loss": 0.2148,
      "step": 12720
    },
    {
      "epoch": 5.98635294117647,
      "grad_norm": 0.4168481182658114,
      "learning_rate": 5.38688217637806e-09,
      "loss": 0.126,
      "step": 12721
    },
    {
      "epoch": 5.986823529411764,
      "grad_norm": 0.3675022133148893,
      "learning_rate": 5.0217804275443084e-09,
      "loss": 0.1914,
      "step": 12722
    },
    {
      "epoch": 5.987294117647059,
      "grad_norm": 0.5804217101656434,
      "learning_rate": 4.669488985575842e-09,
      "loss": 0.3574,
      "step": 12723
    },
    {
      "epoch": 5.987764705882353,
      "grad_norm": 0.7603824101303825,
      "learning_rate": 4.3300078805652564e-09,
      "loss": 0.7539,
      "step": 12724
    },
    {
      "epoch": 5.988235294117647,
      "grad_norm": 0.453222045974539,
      "learning_rate": 4.003337141489371e-09,
      "loss": 0.5078,
      "step": 12725
    },
    {
      "epoch": 5.988705882352941,
      "grad_norm": 0.3040238508298319,
      "learning_rate": 3.689476796242541e-09,
      "loss": 0.0708,
      "step": 12726
    },
    {
      "epoch": 5.9891764705882355,
      "grad_norm": 0.4656025116923193,
      "learning_rate": 3.388426871653305e-09,
      "loss": 0.3438,
      "step": 12727
    },
    {
      "epoch": 5.98964705882353,
      "grad_norm": 0.4741716058666356,
      "learning_rate": 3.100187393434428e-09,
      "loss": 0.2412,
      "step": 12728
    },
    {
      "epoch": 5.990117647058824,
      "grad_norm": 0.6039347027710018,
      "learning_rate": 2.8247583861662484e-09,
      "loss": 0.3965,
      "step": 12729
    },
    {
      "epoch": 5.990588235294117,
      "grad_norm": 0.27466492189389163,
      "learning_rate": 2.562139873413249e-09,
      "loss": 0.1021,
      "step": 12730
    },
    {
      "epoch": 5.9910588235294115,
      "grad_norm": 0.3568177986503285,
      "learning_rate": 2.3123318775741805e-09,
      "loss": 0.4004,
      "step": 12731
    },
    {
      "epoch": 5.991529411764706,
      "grad_norm": 0.30821665952577926,
      "learning_rate": 2.075334419998631e-09,
      "loss": 0.0723,
      "step": 12732
    },
    {
      "epoch": 5.992,
      "grad_norm": 0.2471283845662252,
      "learning_rate": 1.8511475209370685e-09,
      "loss": 0.2305,
      "step": 12733
    },
    {
      "epoch": 5.992470588235294,
      "grad_norm": 0.34200328598874385,
      "learning_rate": 1.6397711995241869e-09,
      "loss": 0.1689,
      "step": 12734
    },
    {
      "epoch": 5.992941176470588,
      "grad_norm": 0.4490241879149022,
      "learning_rate": 1.441205473795559e-09,
      "loss": 0.2041,
      "step": 12735
    },
    {
      "epoch": 5.993411764705883,
      "grad_norm": 0.33712430300418217,
      "learning_rate": 1.2554503607375976e-09,
      "loss": 0.1602,
      "step": 12736
    },
    {
      "epoch": 5.993882352941176,
      "grad_norm": 0.5104593676433786,
      "learning_rate": 1.082505876204287e-09,
      "loss": 0.6211,
      "step": 12737
    },
    {
      "epoch": 5.99435294117647,
      "grad_norm": 0.46379103800935234,
      "learning_rate": 9.223720349671449e-10,
      "loss": 0.3848,
      "step": 12738
    },
    {
      "epoch": 5.994823529411764,
      "grad_norm": 0.28591923004994124,
      "learning_rate": 7.750488506985674e-10,
      "loss": 0.0771,
      "step": 12739
    },
    {
      "epoch": 5.995294117647059,
      "grad_norm": 0.28763157110383336,
      "learning_rate": 6.405363359718308e-10,
      "loss": 0.1108,
      "step": 12740
    },
    {
      "epoch": 5.995764705882353,
      "grad_norm": 0.3332665597570713,
      "learning_rate": 5.188345022943963e-10,
      "loss": 0.2715,
      "step": 12741
    },
    {
      "epoch": 5.996235294117647,
      "grad_norm": 0.3382433670557744,
      "learning_rate": 4.099433600579516e-10,
      "loss": 0.0713,
      "step": 12742
    },
    {
      "epoch": 5.996705882352941,
      "grad_norm": 0.4219141747332892,
      "learning_rate": 3.138629185550634e-10,
      "loss": 0.3379,
      "step": 12743
    },
    {
      "epoch": 5.9971764705882356,
      "grad_norm": 0.32235086049386896,
      "learning_rate": 2.3059318597917763e-10,
      "loss": 0.1196,
      "step": 12744
    },
    {
      "epoch": 5.99764705882353,
      "grad_norm": 0.24084855988103904,
      "learning_rate": 1.6013416947457946e-10,
      "loss": 0.024,
      "step": 12745
    },
    {
      "epoch": 5.998117647058823,
      "grad_norm": 0.2841000778089264,
      "learning_rate": 1.0248587501981986e-10,
      "loss": 0.0415,
      "step": 12746
    },
    {
      "epoch": 5.998588235294117,
      "grad_norm": 0.6383570477697937,
      "learning_rate": 5.7648307577595756e-11,
      "loss": 0.457,
      "step": 12747
    },
    {
      "epoch": 5.9990588235294116,
      "grad_norm": 0.37496396742175225,
      "learning_rate": 2.5621470944869883e-11,
      "loss": 0.1367,
      "step": 12748
    },
    {
      "epoch": 5.999529411764706,
      "grad_norm": 0.5110242849343675,
      "learning_rate": 6.4053678694442335e-12,
      "loss": 0.334,
      "step": 12749
    },
    {
      "epoch": 6.0,
      "grad_norm": 0.48952729339981604,
      "learning_rate": 0.0,
      "loss": 0.3418,
      "step": 12750
    },
    {
      "epoch": 6.0,
      "eval_loss": 0.3348323702812195,
      "eval_runtime": 44.7494,
      "eval_samples_per_second": 16.76,
      "eval_steps_per_second": 8.38,
      "step": 12750
    },
    {
      "epoch": 6.0,
      "step": 12750,
      "total_flos": 20880812359680.0,
      "train_loss": 0.5600017993403416,
      "train_runtime": 347587.0033,
      "train_samples_per_second": 0.073,
      "train_steps_per_second": 0.037
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 12750,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 20880812359680.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}